Skip to content
BENIBIX Learn

Compare

Two terms people constantly mix up, set side by side. Tap one to see what actually separates them.

1352 comparisons

Vision Statement vs Mission Statement

People think a vision and a mission say the same thing, and they do not. A vision is the future a company hopes to help make, years away. A mission is what the company does today, and for whom. The vision is where the road ends. The mission is how you walk it each day.

Business

TAM vs SOM

People quote the TAM as if it were their sales goal, and it is not. The TAM is all the money spent on what you sell, if you won every buyer. The SOM is the slice you can really win in the next few years. It is often a small part of the TAM. Plans and sales goals should start from the SOM.

Business

SWOT Analysis vs PESTEL Analysis

People mix up SWOT and PESTEL, and they do different jobs. PESTEL only looks outside the company, at six forces like laws, money and tech. SWOT looks both inside and out, at strengths, weak spots, chances and threats. A PESTEL is often done first. What it finds fills the outside half of the SWOT.

Business

SAFE vs Convertible Note

People think a SAFE and a convertible note are the same, and they are not. Both give money now and turn into shares later. A note is a loan, so it earns interest and has a due date. A SAFE is not a loan, so it has neither. If no round comes, a note can be called in and a SAFE cannot.

Business

Pre-Money vs Post-Money Valuation

People mix up pre-money and post-money, and the gap changes who owns what. Pre-money is the value before the new cash comes in. Post-money is that value plus the new cash. An investor's share is their money divided by the post-money. The same number as pre-money gives the founders more.

Business

Angel Investor vs Venture Capital

People think an angel and a VC are the same kind of backer, and they are not. An angel puts in their own money, often very early and in small sums. A VC firm invests money it raised from others in a fund. It writes bigger checks later and needs a few huge wins.

Business

Common Stock vs Preferred Stock

People think a share is a share, whoever holds it, and it is not. Common stock is the basic share that founders and staff hold. Preferred stock is what investors get, with extra rights. The biggest right is being paid first when the company is sold. So in a small sale, common stock can get little or nothing.

Business

Startup vs Small Business

People think every new business is a startup, and it is not. A startup is built to grow very fast and often loses money first. A small business sells a known thing and earns from early on. Investors who want fast growth only fund the first kind.

Business

Startup Accelerator vs Startup Incubator

People think an accelerator and an incubator are the same. An accelerator is a short, fixed program for a batch. It gives cash for a slice and ends on demo day. An incubator is slower and open-ended. It gives a desk and advice, and often no cash.

Business

Gross Burn vs Net Burn

People quote burn rate without saying which kind. Gross burn is every bill in the month. Net burn takes off the cash that came in from sales. Runway is worked out from net burn, so mixing them up gives the wrong number of months.

Business

Parameter vs Argument

The parameter is the named blank in the function. The argument is the real value that fills it when you call it.

Development

Statement vs Expression

A statement does a job, like storing a value. An expression only works out a value, and a statement then uses it.

Development

Shape vs Form

A shape is flat, like a circle drawn on paper. A form has depth, like a ball, or is shaded so it looks that way. Both are basic parts of design.

Design

Symmetrical vs Asymmetrical Balance

Both keep a page level, so no side seems to tip. One mirrors the two halves. The other uses different things that weigh the same.

Design

Contrast vs Visual Hierarchy

Contrast is one strong difference between two things. Hierarchy is the whole reading order, and contrast is one tool that builds it.

Design

Design Thinking vs Double Diamond

They are not rivals. Design thinking is the mindset with five modes. The Double Diamond is a map of four phases that the same mindset can follow.

Design

Typeface vs Font

A typeface is the design, like Garamond. A font is one version of it, like Garamond Bold. It is often one file you install.

Design

Kerning vs Tracking

Kerning moves one pair of letters, like A and V. Tracking adds the same space between every letter in a word or line.

Design

Legibility vs Readability

One is about each letter. Can you tell them apart? The font decides that. The other is about the whole text. Is it easy to read? The layout decides that.

Design

Serif vs Sans-Serif

Serif letters have small feet and feel classic. Sans-serif letters have none and feel modern. Both read well on today's sharp screens.

Design

RGB vs CMYK

RGB mixes red, green and blue light, and it is how screens show color. CMYK mixes four inks on paper, and it is how printers show color. Bright screen colors can print dull.

Design

Raster vs Vector

A raster image is a grid of pixels, so it blurs when you blow it up. A vector is stored as shapes, so it stays sharp at any size. Photos are raster. Logos and icons are vector.

Design

Complementary vs Analogous Colors

Complementary colors sit across the wheel from each other, so they clash and pop. Analogous colors sit side by side, so they blend and feel calm. Pick by the mood you want.

Design

Icon vs Pictogram

An icon stands for an action in an app. You may need to learn it, like a gear for settings. A pictogram looks like the real thing it shows. So anyone can read it on a sign.

Design

Wireframe vs Mockup

A wireframe is a plain plan of where things go, in grey boxes. A mockup is a still picture of how the screen will really look, with real color and images.

Design

Mockup vs Prototype

A mockup shows how a screen looks, but you cannot click it. A prototype can be clicked or tapped, so you can test how it works with real people.

Design

Affordance vs Signifier

An affordance is what you can do with a thing, like open a door. A signifier is the clue that tells you how, like a sign that says push. One is the action, the other is the hint.

Design

Usability Testing vs Heuristic Evaluation

In a usability test, real users try real tasks while you watch. In the other, a few experts check the design against a list of rules. Only the test shows where real users trip.

Design

User Flow vs Customer Journey Map

A user flow maps the screens and choices for one task in a product. A journey map follows a person across the whole time with a company, and shows how each stage feels.

Design

UX vs UI

UI is the screen you see and touch: the buttons, menus and text. UX is the whole trip a person takes with a product, from first look to the task done. A pretty UI can still give a bad UX.

Design

Checkbox vs Toggle Switch

A switch acts at once, like a light switch. A checkbox waits for you to press Save or Send. Use a switch for a live setting and a checkbox inside a form.

Design

Radio Buttons vs Select Menu

Both let you pick one choice. Radio buttons show every choice at once, which is best for a few. A select menu hides the list until you open it, which saves room when there are many.

Design

Select Menu vs Dropdown Menu

They look alike, but they do different jobs. A select menu fills in a value in a form, like your size. A dropdown menu runs a command, like Rename or Delete.

Design

Tooltip vs Popover

A tooltip is a few words that show when you point at a thing, and it goes when you move away. A popover opens on a click, stays open and can hold buttons.

Design

Toast vs Alert Banner

A toast pops up to say a job is done, then leaves by itself. An alert banner stays on the page until the problem is fixed or you close it. Never put a must-read message in a toast.

Design

Modal Dialog vs Drawer

A modal sits in the middle and blocks the page until you answer it. A drawer slides in from the edge to hold extra menus or filters, and the page stays close by.

Design

Spinner vs Progress Bar

A spinner says the app is busy but not for how long, so it suits short waits. A progress bar fills up to show how much is done, so it suits waits the app can measure.

Design

Placeholder Text vs Field Label

A field label sits outside the box and stays there. Placeholder text sits inside and goes away as soon as you type. So the label says what goes in, and the placeholder only shows an example.

Design

Segmented Control vs Tabs

People think a segmented control is just a row of tabs, and it is not. Tabs swap between panels that hold different things, like Info and Reviews. A segmented control changes how the same thing is shown, like a calendar by day, week or month. Tabs take you to another part. A segmented control keeps you where you are and changes the view.

Design

Badge vs Chip

A badge is a small dot or number that only shows a status, like 3 new messages. A chip is a small label you can act on, like a filter you tap to turn off.

Design

Accessibility vs Usability

Usability asks if a product is easy to use. Accessibility asks if people who cannot see, hear or move well can use it. A site can be easy for most people. It can still shut a blind user out.

Design

Inclusive Design vs Universal Design

Universal design aims for one design that fits as many people as it can. Inclusive design starts with the people most often left out. It may give them more than one way to do a task.

Design

Closed Captions vs Transcript

Captions show the words on the video, in time with the picture. A transcript is the whole text on a page you read on its own. A video needs captions; a podcast needs a transcript.

Design

Alt Text vs Accessible Name

Alt text describes a picture. An accessible name is what a screen reader calls any control. When a picture is the only thing inside a link or button, its alt text becomes that control's name.

Design

Design System vs Component Library

A component library is the set of ready parts, like buttons and cards. A design system holds that library and more. It adds the tokens, the rules and the advice on when to use each part.

Design

Design System vs Style Guide

A style guide sets the look: colors, fonts and icons, with right and wrong uses. A design system adds working parts in design files and code. Many teams start with the guide and grow it.

Design

Component Library vs Pattern Library

A component library holds single parts, like a button or a field. A pattern library shows how parts work together. Each pattern solves one problem, like a sign-up flow.

Design

Primitive Token vs Semantic Token

A primitive token names a raw value, like one exact blue. A semantic token names a job, like the main brand color, and points to a primitive. Parts should use the semantic one.

Design

Material Design vs Apple HIG

Material Design is Google's system for Android and the web, and it ships ready parts. Apple's guidelines are the advice for iPhone, iPad and Mac apps. An app for both should feel at home on each.

Design

Flat Design vs Skeuomorphism

Skeuomorphism makes screen parts look like real things, like lined paper or a leather cover. Flat design drops all that for plain shapes and solid colors. Flat replaced it in the early 2010s.

Design

Glassmorphism vs Neumorphism

Both became trends around 2020. Glassmorphism makes panels look like frosted glass, with blurred color behind. Neumorphism makes shapes rise out of a page of one soft color, using only shadow.

Design

F-Pattern vs Z-Pattern

The F-pattern is how people skim a page full of text: across the top, then down the left. The Z-pattern is how eyes move on a simple page with few words: across, down on a slant, across.

Design

Infinite Scroll vs Pagination

Infinite scroll keeps adding items as you reach the end, which suits browsing a feed. Pagination splits results into numbered pages. That suits tasks, because people can find their place again.

Design

Responsive Design vs Mobile First

Responsive design is the result: one site that fits every screen. Mobile first is one way to get there, starting from the smallest screen. You then add layout as the screen grows.

Design

Hamburger Menu vs Mega Menu

A hamburger menu hides the links behind a three-line button to save space on a phone. A mega menu does the opposite. It opens a big panel that shows many links at once, for large sites.

Design

Logo vs Visual Identity

People often call the logo a brand's identity. But it is only one piece. The visual identity is the whole look. It is the logo plus the colors, fonts and pictures, and the rules for using them.

Design

Wordmark vs Lettermark

A wordmark spells out the whole brand name in its own font. A lettermark uses only the first letters. Pick a lettermark when the name is long and people already shorten it.

Design

Pictorial Mark vs Abstract Mark

A pictorial mark is a picture of a real thing, like an apple or a bird. An abstract mark is a shape that shows nothing real. Its meaning has to be learned over time.

Design

Combination Mark vs Emblem Logo

Both join a name with a symbol. In a combination mark the two sit side by side and can be split apart. In an emblem the name sits inside the shape, so they always go together.

Design

Brand Guidelines vs Style Guide

Brand guidelines cover a brand in all places. They set the logo rules, colors, fonts and photo style, for print and for screens. A visual style guide sets the look of one product's screens.

MarketingDesigncross-domain

Brand Guidelines vs Brand Kit

Brand guidelines are the rules. They say how big the logo may be, which colors to use and what you must never do to it. A brand kit is the files: the logos, color codes and fonts, ready to use.

MarketingDesigncross-domain

Raster Editor vs Vector Editor

A raster editor changes pictures made of tiny dots, so it suits photos. A vector editor draws with points and curves. Its shapes stay sharp at any size, so it suits logos and icons.

Design

Photoshop vs Illustrator

Both are made by Adobe, and people often mix the two of them up. Photoshop is for photos and other pictures made of pixels. Illustrator is for drawing with shapes. It suits logos and icons that must grow and shrink.

Design

Figma vs Sketch

Both are tools for designing apps and websites. Sketch is a Mac app. Figma runs in any web browser, and a whole team can work in one file at the same time.

Design

Figma vs FigJam

Both come from the same company. Figma is for designing the real screens of an app or site. FigJam is a simple online whiteboard for sticky notes, flow charts and early ideas.

Design

Auto Layout vs Resizing Constraints

People think Auto Layout and resizing constraints do the same job, and they do not. Auto Layout sizes a frame from what is inside it, so a button grows when its label gets longer. Constraints work the other way. They say what a layer does when the frame around it changes size, such as staying pinned to one corner.

Design

Canva vs InDesign

Canva starts you from a ready-made page, so anyone can make a post or flyer fast. InDesign is a full page layout app. Pros use it for long books and magazines that go to print.

Design

Bar Chart vs Histogram

Both use bars. A bar chart compares separate groups, with gaps between bars. A histogram cuts one number into ranges, so its bars touch.

Design

Sequential vs Diverging Color Palette

Both shade data from pale to strong. A sequential palette runs one way, low to high. A diverging one runs two ways from a real middle, like zero.

Design

Ease-In vs Ease-Out

Ease-in is slow at the start and fast at the end. Ease-out is fast first and slow at the end. Use ease-out for things that arrive.

Design

Storyboard vs Animatic

A storyboard is a set of still sketches on paper. An animatic plays those sketches as a rough video, timed to sound.

Design

CSS Transition vs CSS Animation

A transition runs once, from one state to another, like on hover. An animation runs by itself through many steps and can loop.

Design

Design Review vs Design Critique

A critique is open feedback to make a design better. A review checks it against the goals and ends in a decision.

Design

Fake Door Test vs Wizard of Oz Test

A fake door is a button for a feature that is not built, to count who wants it. Wizard of Oz has a hidden person do the work.

Design

Concept Testing vs Usability Testing

A concept test asks if an idea is worth making at all. A usability test checks if people can use the thing that was made.

Design

Hero Image vs Hero Section

The hero section is the whole top block of a page. The hero image is the big picture inside it, under the words and the button.

Design

Flyer vs Brochure

A flyer is one flat sheet that says one thing fast. A brochure is folded into panels and explains more for people to keep.

Design

Bleed vs Safe Area

Bleed is extra design past the cut, made to be cut off. The safe area is inside the cut, where text stays safe.

Design

Virtual Reality vs Augmented Reality

VR hides the real world and puts you in a made one. AR keeps the real world in view and lays pictures on top of it.

Design

Augmented Reality vs Mixed Reality

AR lays pictures over the real world. In mixed reality the digital things know the room, so they sit on tables and hide behind chairs.

Design

Rasterization vs Ray Tracing

Rasterization fills in the pixels under each shape and is fast. Ray tracing follows rays of light, so it looks more real but costs more.

Design

UX Designer vs UI Designer

A UX designer plans how a product works: the steps, the flow and the tests. A UI designer plans how each screen looks: type, color and parts.

Design

Product Designer vs UX Designer

A UX designer works on how a product works. A product designer does that too, plus the look, and keeps the business goals in view.

Design

Art Director vs Creative Director

An art director owns the look of one project. A creative director sets the big idea for many and leads the art directors.

Design

Dolly Shot vs Zoom Lens

A dolly moves the camera closer, so the background shifts. A zoom only changes the lens, so the camera stays put and the background looks flat.

Design

Establishing Shot vs Wide Shot

A wide shot is a size: far away, the whole body. An establishing shot is a job: it shows where a scene is, and it is often wide.

Design

Dolly Shot vs Tracking Shot

A dolly shot moves toward or away from the subject. A tracking shot moves along with a subject that is moving. Some crews use both words for one thing.

Design

Syntax Error vs Runtime Error

A syntax error stops the code before a single line runs. A runtime error waits until the code hits a bad value.

Development

For Loop vs While Loop

A for loop goes over a list and stops when the list runs out. A while loop goes on until its test turns false.

Development

JavaScript vs Java

They share half a name and nothing else. JavaScript runs in the browser; Java runs on its own JVM, often on servers.

Development

JavaScript vs TypeScript

TypeScript is JavaScript with types added. The types are checked before the code runs, then they are taken away.

Development

Compiled vs Interpreted Language

A compiled language is built into a program before it runs. An interpreted one is read and run line by line as it goes.

Development

C vs C++

C++ was built on top of C. It keeps the speed of C and adds classes, which makes it a far bigger language to learn.

Development

Library vs Framework

Your code calls a library when it wants help. A framework is the other way round: it runs the show and calls your code.

Development

Linter vs Formatter

A linter reads code and warns about likely bugs. A formatter only fixes spaces and line breaks, and never what code does.

Development

Static vs Dynamic Typing

Both kinds have types. A static language checks them before the code runs; a dynamic one waits until each line runs.

Development

Git vs GitHub

Git is the tool on your computer that saves every change. GitHub is a website that keeps a shared copy online for a team.

Development

Merge vs Rebase

Both join two branches. A merge keeps both lines and ties them with a new commit; a rebase moves your commits into one line.

Development

Git Fetch vs Git Pull

A fetch only downloads new commits and leaves your files alone. A pull downloads them and merges them in, in one step.

Development

Git Fork vs Git Clone

A fork is your copy of someone's repo on the website. A clone is a copy on your own computer, and you can clone a fork.

Development

Git Commit vs Git Push

A commit saves a change on your own computer. A push sends your commits to the shared remote, where the team can see them.

Development

JSON vs XML

Both hold data as plain text. JSON uses names and curly brackets; XML wraps each value in tags, which makes it longer.

Development

JSON vs YAML

YAML uses indents and allows comments, so it suits settings files. JSON uses brackets and is the usual choice for APIs.

Development

SQL vs NoSQL

A SQL database keeps data in linked tables with fixed rules. A NoSQL one uses other shapes, like documents or keys, and has looser rules.

Development

Inner Join vs Left Join

An inner join keeps only rows that match in both tables. A left join keeps every row from the first table, and fills the gaps with NULL.

Development

PostgreSQL vs MySQL

Both are free SQL databases. PostgreSQL is stricter and has more tools; MySQL is simple to start and runs a huge share of websites.

Development

Primary Key vs Foreign Key

A primary key names each row in its own table. A foreign key sits in another table and points back to that name.

Development

Replication vs Sharding

Replication puts a full copy of the data on each server. Sharding splits the rows, so each server holds just a part.

Development

Server-Side Rendering vs Client-Side Rendering

With server rendering, the server builds the page and sends it ready to read. With client rendering, the browser gets code and builds the page itself. That makes the first view slower.

Development

Static Site Generation vs Server-Side Rendering

Both send ready-made HTML. Static site generation builds each page once, ahead of time. Server rendering builds it again for each visit, so it can show fresh data.

Development

Static Website vs Dynamic Website

A static site sends the same ready-made files to everyone. A dynamic site builds each page when it is asked for, so it can show data for each person.

Development

Front End vs Back End

The front end is the part you see and click in the browser. The back end runs on a server, out of sight, and keeps the data and the rules.

Development

Cookie vs Local Storage

A cookie and local storage both keep small bits of data in the browser. A cookie goes to the server with every request. Local storage stays in the browser and only your page's code reads it.

Development

Props vs State

Props are the inputs a part gets from the part above it, and it should not change them. State is data a part owns and can change, and the screen follows it.

Development

React vs Vue.js

Both build screens from small parts. React is a library you add tools to as you need them. Vue comes with more of its own tools and is often called easier to learn.

Development

Minification vs Tree Shaking

Both make code files smaller. Minification squeezes code by cutting spaces and long names. Tree shaking drops whole parts of code that are never used.

Development

WebSocket vs Server-Sent Events

Both keep a line open so news arrives at once. A WebSocket lets both sides send. Server-sent events only flow from the server to the browser.

Development

Django vs Flask

Django and Flask are both Python web frameworks. Django comes with most of what a site needs, from logins to an admin page. Flask gives you just the basics and you add the rest.

Development

HTTP vs HTTPS

HTTP and HTTPS follow the same rules for asking for pages. HTTPS sends them over a locked line, so no one in between can read or change them. Plain HTTP can be read by anyone on the path.

Infrastructure

Terminal vs Shell

People say terminal and shell as if they were one thing. The terminal is the window you type in. The shell is the program inside it that reads your command and runs it.

Infrastructure

URL vs Domain Name

The domain name is just the name of the site, like example.com. A URL is the full address of one page, with how to fetch it, the name and the path.

General

Class vs Object

A class and an object are not the same thing. The class is the plan, like a cookie cutter. An object is one real thing made from it, with its own values.

Development

Synchronous vs Asynchronous Code

Synchronous code waits for each line to finish before the next one starts. Asynchronous code starts a slow task, moves on, and handles the result when it arrives.

Development

Imperative vs Declarative Programming

Imperative code lists the steps to reach an answer, one change at a time. Declarative code says what answer you want and lets the tool pick the steps.

Development

OOP vs Functional Programming

Object-oriented code keeps data and actions together in objects. Those objects change over time. Functional code passes data through small functions. They make new values and change nothing.

Development

Method vs Function

A method is a function too, but it belongs to an object. A plain function stands alone; a method is called with a dot and can use its object's data.

Development

Encapsulation vs Abstraction

Encapsulation and abstraction both hide things, but not the same things. Encapsulation hides an object's data, so only its methods can change it. Abstraction hides how a job is done. You see only a simple way to use it.

Development

Callback vs Promise

A callback is a function you hand over to be run later. A promise is an object for a result to come, which you can chain with then. Promises avoid deep nests of callbacks.

Development

Lambda vs Closure

A lambda is a short function with no name. A closure is a function that keeps the variables from where it was made. A lambda can be a closure, but the two words mean different things.

Development

Stack vs Queue

A stack and a queue both hold items in a line, but they let them out from different ends. A stack gives back the last item in. A queue gives back the first.

Development

Linear Search vs Binary Search

Linear search checks every item, one by one, and works on any list. Binary search needs a sorted list, but halves it each step, so a million items take about 20 checks.

Development

BFS vs DFS

Breadth-first search walks a graph one layer at a time and finds the path with the fewest steps. Depth-first search follows one path to its end, then backs up.

Development

Merge Sort vs Quicksort

Merge sort and quicksort are both fast, but they differ. Merge sort is always O(n log n) and needs extra memory. Quicksort sorts in place and is often faster, but bad pivots make it slow.

Development

Tree vs Graph

A tree is a graph with rules. It has one root and no loops, so there is one path to each node. A graph may link any node to any other, loops and all.

Development

Array vs Linked List

An array keeps its items side by side in memory, so reading any one is instant. A linked list joins items with pointers. Changes in the middle are quick, but reading is slow.

Development

Divide and Conquer vs Dynamic Programming

Divide and conquer and dynamic programming both cut a big problem into small parts to solve one at a time. In the first, each part is new. In the second, the same parts come back, so each answer is saved and used again.

Development

Time vs Space Complexity

Time complexity counts steps. Space complexity counts memory. Both say how the cost grows as the input grows, and a faster way often needs more memory.

Development

GET vs POST

A GET request only reads, and its details ride in the address. A POST request sends new data in its body and often creates something. Send a GET twice and nothing changes; send a POST twice and you may get two orders.

Infrastructure

PUT vs PATCH

A PUT request replaces the whole thing on the server with the full copy you send. A PATCH request changes only the fields you send. A PUT that leaves out a field may wipe it, while a PATCH leaves it alone.

Infrastructure

Query Parameter vs Request Body

A query parameter goes in the address, after the question mark. Anyone can see it there. A request body goes inside the request, after the headers. Short filters go in the address; passwords, forms and files go in the body.

Infrastructure

REST vs GraphQL

A REST API gives every thing its own address and uses HTTP methods on it. GraphQL sends every query to one address and lets the client pick the exact fields. REST is simpler to cache; GraphQL saves trips on a slow phone.

Development

Webhook vs Polling

With a webhook, the other service calls you as soon as something happens. With polling, you keep asking it if anything has changed. Webhooks are faster and use less, but your server must be open to the web.

Development

Unit Test vs Integration Test

A unit test checks one small piece of code by itself, in a few milliseconds. An integration test checks that parts work as a team, like your code and its database. Unit tests show which piece broke. Integration tests find the gaps between the pieces.

Development

Integration Test vs End-to-End Test

An integration test checks a few parts together, often without a browser. An end-to-end test drives the whole app in a real browser, the way a person does. End-to-end tests prove a real task works, but they are slower and break more often.

Development

Mock vs Stub

A mock records how your code called it, and the test checks those calls afterwards. A stub only hands back an answer you chose ahead of time and checks nothing. Use a stub to set up a case, and a mock to prove a call happened.

Development

Smoke Test vs Regression Test

A smoke test is a quick check, in a minute or two, that the app works at all. A regression test checks that something which used to work still works after a change. Smoke tests ask if it is alive; regression tests ask if anything broke.

Development

Jest vs Vitest

Jest and Vitest are both tools that run tests for JavaScript, and their test and expect functions look almost the same. Vitest is built on Vite and reuses its settings, so it fits apps made with Vite. Jest is older and works with almost any project.

Development

Code Smell vs Technical Debt

A code smell is a sign in the code that something might be wrong. Technical debt is the cost you pay later for a quick fix you chose. A smell is often how you spot the debt, but not every smell is debt worth paying back.

Development

Authentication vs Authorization

Authentication checks who you are, and authorization decides what you may do. Logging in with a password is the first. Being let in to delete a post, or not, is the second. The check of who you are always comes first. An app cannot decide what you may do before it knows who you are.

Development

Access Token vs Refresh Token

An access token is the short-lived pass you show on each request, and a refresh token is how you get a new one. The access token often lasts minutes, so a stolen one is soon useless. The refresh token lasts days or weeks. It goes only to the login server, and it must be kept far safer.

Development

OAuth vs OpenID Connect

OAuth 2.0 lets an app act for you without ever seeing your password, and OpenID Connect tells the app who you are. OAuth hands out access tokens, which say what an app may do. OpenID Connect sits on top of OAuth. It adds an ID token, which says who logged in. Use OpenID Connect to log people in, and OAuth for access.

Development

Session vs JWT

A session keeps your login on the server, while a JSON Web Token carries it inside the token. With a session, the browser holds only an ID. The server looks it up on each request. A JWT holds the facts and a signature, so the server can check it with no lookup. A session can be ended at once; a JWT stays good until it runs out.

Development

Hashing vs Encryption

Hashing and encryption both scramble data, but only encryption can be undone. Encryption uses a key, so someone with the key can read the data again. A hash is a one-way fingerprint. It has no key and no way back. Use encryption for data you need to read later. Use hashing to check files or store passwords.

Development

XSS vs CSRF

Cross-site scripting and cross-site request forgery both abuse the trust between you and a site, in opposite ways. XSS puts the attacker's script on the real site, where it runs as if the site wrote it. CSRF stays on the attacker's page. It tricks your browser into sending a request to the real site. Safe output stops XSS. Tokens and SameSite cookies stop CSRF.

Development

Continuous Delivery vs Continuous Deployment

Continuous delivery and continuous deployment both test and build every change on their own, but only one ships it by itself. With delivery, the code is always ready. A person clicks to go live. With deployment, there is no click. Each change that passes the tests goes straight to users.

Development

Blue-Green Deployment vs Canary Release

Blue-green deployment and a canary release both make a new version safer to ship, but they move users in different ways. Blue-green moves every user at once, from one full copy to the other. It can switch back in seconds. A canary moves a few users first, then more. Blue-green is quick; a canary keeps a bug small.

Development

Deployment vs Release

A deployment and a release sound the same, but a deployment puts code on the servers and a release lets users have it. With a feature flag, new code can be deployed today and kept hidden. The release is the day you switch it on. Keeping them apart lets you ship often and launch when you choose.

Development

Rollback vs Hotfix

A rollback and a hotfix both deal with a bad change in the live app, in opposite ways. A rollback goes back to the last version that worked, often in a minute. A hotfix goes forward with a new version that fixes the bug. Roll back first to stop the harm, then ship the hotfix calmly.

Development

MIT License vs GNU GPL

The MIT License and the GNU GPL are both open source licenses, but they differ on what you must give back to others. MIT lets you do almost anything, even put the code in a closed app, if you keep the notice. The GPL says that if you share a changed version, you must share its code under the GPL too. MIT is free to take; the GPL keeps code open.

Development

MIT License vs Apache License 2.0

The MIT License and the Apache License 2.0 both let you use, change and sell code with few rules. The main gap is patents. Apache adds a written promise that the authors will not sue users over patents in the code. It also asks you to keep a NOTICE file and mark changed files. MIT is shorter; Apache gives more legal cover.

Development

IaaS vs PaaS

Infrastructure as a service and platform as a service are both rented from a cloud, but they split the work in different places. With IaaS you rent bare servers and run the system, the updates and the app yourself. With PaaS you hand over your code and the platform runs all the rest. IaaS gives more control; PaaS takes less work.

Infrastructure

PaaS vs SaaS

Platform as a service and software as a service are both run by someone else, but they are made for different people. PaaS is for developers: you bring your own code, and it runs it. SaaS is for everyone: you use an app someone else wrote, like a mail or chat app. With PaaS you build the app; with SaaS you just log in.

Infrastructure

Cloud Region vs Availability Zone

A cloud region and an availability zone are both places your servers live, but one always sits inside the other. A region is a city or area, like Frankfurt, picked to be near your users. A zone is one separate data centre area inside it. Pick a region for speed and the law. Then spread across zones, so one outage does not stop you.

Infrastructure

CPU vs GPU

People often think a GPU is just a faster CPU. It is a different kind of chip. A CPU has a few strong cores that can do any job, one step after another. A GPU has thousands of small cores that all do the same sum at once. AI needs the second kind of work, so it runs far faster on a GPU.

Infrastructure

RAM vs VRAM

People often think a computer's memory is one pool that every chip shares. With a graphics card, there are two. RAM belongs to the CPU, and video memory sits on the card for the GPU alone. A model that fits in 32 GB of RAM can still be too big for a GPU with 12 GB of video memory.

Infrastructure

Memory Bandwidth vs Network Bandwidth

People often hear bandwidth and think of internet speed. Network bandwidth is how much data moves over a link between machines. Memory bandwidth is how much moves inside one machine, between memory and a chip. The second is often thousands of times bigger. It is the one that sets how fast AI writes.

Infrastructure

Serverless vs PaaS

Serverless computing and platform as a service both hide the servers from you, but they bill and run code in different ways. A PaaS keeps your app running all day, and you pay for that time. Serverless runs your code only when something calls it and stops after. You pay per run. Serverless is cheaper when traffic is low or spiky.

Infrastructure

AWS vs Azure

Amazon Web Services and Microsoft Azure are the two biggest clouds, and they sell much the same things under different names. AWS is the largest and started first, in 2006. Azure came in 2010. It fits well with Microsoft tools like Windows, .NET and Office. Teams often pick AWS for its range and Azure when they already use Microsoft.

Infrastructure

KISS vs YAGNI

The KISS principle and YAGNI both push you toward less code, but they answer different questions. KISS asks how to build something: pick the simplest design that works. YAGNI asks whether to build it at all: wait until a real need shows up. You can follow one and break the other, so most teams keep both in mind.

Development

Pair Programming vs Code Review

Pair programming and code review both put a second pair of eyes on code, but at different times. In pair programming, two people write the code together. Mistakes are caught as they are typed. In a code review, one person writes it and another reads it later, before it is merged. Pairing gives quick feedback; review suits teams that work apart.

Development

Clean Code vs Style Guide

Clean code and a style guide both make code easier to read, but at different depths. A style guide covers how code looks: spaces, quotes, line length and the shape of names. Clean code covers what it says: clear names, small functions and no tricks. A formatter can fix style for you, but only a person can make code clean.

Development

Process vs Thread

A process and a thread are both ways to run work at the same time, but they differ in what they share. A process is a running program with its own memory. It is kept apart from all the others. A thread is a line of work inside a process, and it shares that memory with the other threads. Threads are cheap and share data with ease. Processes are safer, because one crash stays inside.

Development

Concurrency vs Parallelism

Concurrency and parallelism both sound like doing many things at once, but they are not the same thing. Concurrency means tasks take turns over the same span of time. One core can do it. Parallelism means tasks run at the very same instant, each on its own core. A program can be concurrent but not parallel, and the other way round.

Development

Stack vs Heap Memory

Stack memory and heap memory are both parts of a running program's memory, but they hold different data. The stack is small and fast. It holds the local values of the functions running now, and they vanish when the function returns. The heap is large and slower. It holds data that must live longer or grow, until it is freed.

Development

Race Condition vs Deadlock

A race condition and a deadlock are both bugs that come from threads sharing data, but they fail in opposite ways. In a race, two threads touch the same data at once with no lock, and the result comes out wrong. In a deadlock, threads hold locks and wait for each other, so nothing runs at all. A race gives wrong answers; a deadlock gives none.

Development

Latency vs Bandwidth

Latency and bandwidth are both called internet speed, but they measure two different things. Latency is the delay before data arrives, in milliseconds. Bandwidth is how much data the line can carry each second, in Mbps. A wide pipe can still be slow to answer. More bandwidth helps big downloads. Less latency makes clicks and calls feel quick.

Infrastructure

Bandwidth vs Throughput

Bandwidth and throughput are both counted in data per second, but one is a limit and one is a result. Bandwidth is the most a line could carry, like the width of a road. Throughput is what really gets through, like the cars that actually pass. Throughput is always at or below bandwidth, and the gap shows where time is lost.

Infrastructure

Monolith vs Microservices

A monolith and microservices are two ways to shape the very same app, and the choice changes how a whole team works. A monolith is one unit, built and shipped all at once. Microservices split the app into many small services, each shipped on its own. A monolith is simpler to start and run. Microservices help when many teams need to ship apart.

Development

Horizontal vs Vertical Scaling

Horizontal and vertical scaling are both ways to handle more work, but they grow in two different ways. Vertical scaling makes one machine bigger, with more cores and memory. Horizontal scaling adds more machines that share the load. Going up is easy but has a ceiling. Going out has no ceiling, but the app must be built for it.

Development

Message Queue vs Pub/Sub

A message queue and pub/sub both pass messages between parts of a system, but they hand them out in two different ways. In a queue, each message goes to just one worker. So a job is done once. In pub/sub, every subscriber gets its own copy. So one event can start many jobs. Use a queue for work and pub/sub for news.

Development

Coupling vs Cohesion

Coupling and cohesion are both ways to judge how well code is split, but they look at different things. Coupling is how much one part leans on other parts. Cohesion is how well the code inside one part fits together. Good design aims for low coupling and high cohesion. Each part does one thing and needs little from the rest.

Development

Event Sourcing vs CQRS

Event sourcing and CQRS are often used together, so people mix them up, but they solve two different problems. Event sourcing is about how data is stored. Every change is kept as an event. CQRS is about how code is split. One side changes data, and another side reads it. Either one works fine without the other.

Development

API Gateway vs Service Mesh

An API gateway and a service mesh both sit in the path of every request, but they guard two very different doors. A gateway is the front door, where calls from apps and browsers first come in. A service mesh handles the calls between services inside the system. Most big systems that use one end up using both.

Development

Training vs Inference

People often think a chatbot learns from them while they talk. It does not. Training is when a model learns, and it is over before the model is shared. Inference is each time the model answers you. The model stays the same then, so your chat does not teach it.

AI

AGI vs Superintelligence

People often use the two words as if they meant the same thing. AGI would match a person at almost any thinking task. Superintelligence would go far past the best people at nearly all of them. AGI is the human mark, and superintelligence is beyond it. Neither one exists today.

AI

Generative AI vs Predictive AI

Since chatbots took off, people often think all AI makes new things. Much of it does not. Predictive AI sorts and scores, like a spam filter or a sales forecast. Generative AI makes something new, like a picture or a reply. Many jobs only need the first kind, and it costs far less.

AI

Artificial Intelligence vs Machine Learning

People often use AI and machine learning as if they were the same word. They are not. AI is the goal of making computers do thinking jobs. Machine learning is one way to reach that goal, by learning from examples. Today it is the main way, but rule-based programs are AI too.

AI

Model Parameter vs Hyperparameter

People often think both are learned in training. Only one is. Parameters are the numbers inside the model, and training finds them. Hyperparameters are the settings you pick first, like the size of each step. You set those by hand. The rest you never touch.

AI

Supervised vs Unsupervised Learning

The difference is not how smart the model is. It is whether the data comes with answers. Supervised learning has a label on every example and learns to guess it. Unsupervised learning has no labels and finds groups or odd cases on its own. You choose by what data you have.

AI

Classification vs Regression

People often think regression means going backwards, or that it does the same job as classification. Both predict, but the answers differ. Classification picks one of a few named groups, like spam or not spam. Regression gives a number on a scale, like a price. Ask what shape the answer has, and you know which one you need.

AI

Validation Set vs Test Set

People often think these are two names for the same held-out data. Both are kept out of training, but they have different jobs. The validation set is checked again and again while you tune. The test set is opened once, at the end. Tuning on the test set makes your final score a lie.

AI

Linear Regression vs Logistic Regression

The names make people think both guess numbers. Only one does. Linear regression gives a number on a scale, like a rent. Logistic regression gives a chance between 0 and 1 and is used for yes or no questions. So logistic regression is a classifier, despite its name.

AI

Clustering vs Classification

Both put examples into groups, so people mix them up. The difference is who names the groups. In classification you name them first and give labelled examples. In clustering nobody names them, and the model finds groups that you then have to explain.

AI

k-Means vs k-Nearest Neighbors

Both have a k and both measure closeness, so people often swap them. They do different jobs. In k-means, k is how many groups to find in data with no labels. In k-nearest neighbors, k is how many labelled neighbours vote on a new example's group.

AI

Random Forest vs Gradient Boosting

Both are made of many decision trees, so people treat them as the same idea. The trees work in different ways. In a random forest, the trees are built side by side and vote. In gradient boosting, they are built in a row, and each one fixes what the last got wrong.

AI

Overfitting vs Underfitting

People often think a bad model has just not learned enough. That is only one way to fail. An underfit model is too simple, so it does badly on old data and new data alike. An overfit model has memorised its training data, so it shines there and fails on new data. The fixes go in opposite directions.

AI

Backpropagation vs Gradient Descent

People often use the two names for the same thing. They are two steps that work as a pair. Backpropagation works out how much each weight added to the error. Gradient descent then uses those numbers to move every weight a small step. One finds the changes, and the other makes them.

AI

Epoch vs Batch Size

People often think an epoch is one training step. It is a full pass over all the data. Batch size is how many examples go into one step. So with 1,000 examples and a batch size of 100, one epoch takes 10 steps.

AI

GAN vs Diffusion Model

People often think every AI image maker works the same way. These two do not. A GAN makes a picture in one go, and a second network judges it. A diffusion model starts from noise and cleans it up over many steps. Most image makers today use diffusion.

AI

Stemming vs Lemmatization

People often think stemming and lemmatization are the same step. Both bring word forms back to one base, but in different ways. Stemming cuts endings off by rule, which is fast but can leave a part that is not a word. Lemmatization looks the word up, so 'was' becomes 'be' and the result is always a real word.

AI

Bag of Words vs Word Embedding

People often think any way of turning words into numbers keeps their meaning. Bag of words only counts words, so 'car' and 'vehicle' are as far apart as 'car' and 'cake'. A word embedding gives each word a learned vector, so words with close meanings sit close. Counts are quick and clear; embeddings know which words are alike.

AI

Base Model vs Instruct Model

People often think every language model is a chat assistant. A base model is only trained to carry text on. So a question may get more questions back. An instruct model is that same base model with more training to follow requests. The facts mostly come from the base. The manners come from the extra step.

AI

Encoder vs Decoder

People often think every transformer has both an encoder and a decoder. Many have only one. An encoder reads a whole text, looking both ways. It turns the text into meaning for search or sorting. A decoder writes text one token at a time, looking only back. Chat models use just a decoder.

AI

Pretraining vs Fine-Tuning

People often think fine-tuning is where a model learns what it knows. Most of its knowledge comes from pretraining, the huge first round on trillions of tokens. Fine-tuning is a small round on a few thousand examples. It shapes style and skill for one job. It is cheap and quick, but it adds few new facts.

AI

Context Window vs Knowledge Cutoff

People often think a model with a big context window knows recent news. They are two different limits. The knowledge cutoff is the date its training text stops, so it is about what the model learned. The context window is how much text it can read right now, so it is about what you hand it. Pasting a fresh article into the window is how you get past the cutoff.

AI

Temperature vs Top-p Sampling

People often think temperature and top-p are two names for one setting. Temperature changes the shape of all the chances, making the top word stronger or the field flatter. Top-p does not reshape them; it cuts off the long tail of unlikely words. Most guides say to change one and leave the other alone.

AI

LoRA vs QLoRA

People think QLoRA is a different way to tune a model, and it is not. Both lock the model and train two thin tables beside each big one. QLoRA also squeezes the locked model into a 4-bit form first, about a quarter of the usual size. That lets a big model be tuned on one chip, at some cost in speed.

AI

Top-k vs Top-p Sampling

People think top-k and top-p are the same cut, and they are not. Top-k always keeps a set number of tokens, such as 40. Top-p keeps tokens until their chances add up to a share, such as 0.9. So top-p keeps a few tokens when the model is sure, and many when it is not. Top-k keeps 40 either way.

AI

Context Window vs Max Tokens

People think max tokens is the size of the context window, and it is not. The context window is all the text a model can hold at once: what you send and what it writes. Max tokens is a cap you set on the reply alone. The reply has to fit in what is left of the window after your prompt.

AI

OpenAI o-series vs GPT

People think an o-series model is just a newer GPT, and it is not. A GPT model starts its answer straight away. An o-series model first writes out hidden working, then answers. That makes it better at maths, code and hard puzzles. It is also slower and costs more, so a GPT model is often the better pick for simple jobs.

AI

Prompt Engineering vs Context Engineering

People often think context engineering is a new name for prompt engineering. Prompt engineering is about the words in one prompt. That is the task, the examples and the format. Context engineering is about all that goes in the window each time. That takes in files, search results, tool output and the chat so far. In agents, most of that text was never typed by a person.

AI

Few-Shot Prompting vs Fine-Tuning

People often think that examples in a prompt train the model. Few-shot prompting only steers one request. The examples must be sent again each time. Fine-tuning changes the model's weights with many examples, so the skill stays. Few-shot is free to try in a minute. Fine-tuning takes data, time and money.

AI

Agentic Coding vs Vibe Coding

People often think vibe coding and agentic coding are the same thing. Agentic coding is about what the tool does. It edits files and runs tests in a loop. Vibe coding is about what the person does: they take the code without reading it. You can use an agent and still check every change, and that is not vibe coding at all.

AI

Semantic Search vs Keyword Search

People often think semantic search is just a smarter keyword search that wins every time. Keyword search matches the exact words, so it is best at names, codes and rare terms. Semantic search matches meaning, so it finds the same idea in other words. Each misses what the other finds. That is why most good systems run both.

AI

RAG vs Fine-Tuning

People often think fine-tuning is how you teach a model your own documents. Fine-tuning changes how a model writes, and it is poor at adding facts that change. RAG leaves the model alone and hands it the right text at question time, so the facts stay current and can be traced. Use RAG for knowledge. Use fine-tuning for style and format.

AI

Cosine Similarity vs Dot Product

People often think cosine similarity and the dot product always put results in the same order. They do not, and length is the reason. The dot product grows with how long the vectors are, not just their angle. Cosine similarity looks at the angle alone. So the two agree only when each vector has a length of 1. Many embedding models do that, but some do not.

AIGeneralcross-domain

AI Agent vs AI Workflow

People often call any system that uses a language model an agent. In an AI workflow, a developer fixes the steps in code, and the model fills in each one. In an AI agent, the model picks its own next step, again and again, until the goal is met. Workflows are cheaper and easier to test. Agents handle tasks whose steps cannot be known in advance.

AI

Agent Memory vs Context Window

People often think an agent remembers all it ever saw in its context window. It does not. The context window is just what the model can see right now, and it is wiped when the task ends. Agent memory is kept outside the model and read back in later. If a fact was never saved, the next task will not know it.

AI

Agent Sandbox vs Agent Permissions

People often think that if an agent has strict permissions, it does not need a sandbox. Permissions decide which actions the agent may take, and they depend on each action being checked. A sandbox limits what any action can reach, even one that slipped through. Permissions stop the wrong step. A sandbox keeps the damage small when one gets past.

AI

MCP vs Tool Use

People often think MCP is a new way for models to use tools that replaces tool use. It is not. Tool use is what the model does: it asks for a tool by name and reads the result. MCP is how an app finds tools made by others and plugs them in. Under MCP, the model still uses tools in the same way.

AI

MCP Tool vs MCP Resource

People often think MCP tools and resources are two names for the same thing. A tool is an action, and the model decides when to call it. A resource is data to read, and the app or the user decides when to load it. If something only needs to be read, make it a resource. Keep tools for things that act.

AI

A2A vs MCP

People often think A2A and MCP are rival standards that do the same job. MCP connects one agent to its tools and data, such as files or a database. A2A connects one agent to other agents, so they can hand each other tasks. Many systems use both: MCP inside each agent, and A2A between them.

AI

Semantic Cache vs Prompt Caching

People often think a semantic cache and prompt caching are the same money saving trick. A semantic cache skips the model and returns a saved answer. It does that when a new question means the same as an old one. Prompt caching still calls the model every time, but it reuses the work done on the start of a long prompt that has not changed.

DevelopmentAIcross-domain

LLM Gateway vs API Gateway

People often think an LLM gateway is just an API gateway with a new name. Both sit in front of services and deal with keys, limits and logs. An API gateway guards your own services from outside callers. An LLM gateway sits in front of outside model makers for your own apps. It counts tokens, tracks the cost of each model and can switch model makers.

Development

LangChain vs LangGraph

People often think LangGraph is just the new name for LangChain, but they are two different tools made by the same team. LangChain gives the parts: links to models, prompts, tools and search, joined in straight chains. LangGraph sets out steps as a graph with loops, saved state and pauses, which is what agents need. Many projects use both.

Development

LangChain vs LlamaIndex

People think LangChain and LlamaIndex do the same job, and they do not. LangChain joins steps: a prompt, a model, a tool and a parser, one after another. LlamaIndex gets your own files ready for a model. It loads them, splits them and finds the right parts for each question. Many apps use both.

Development

Quantization vs Model Pruning

People often think quantization and pruning are the same way to shrink a model. Quantization keeps every weight but saves each one with fewer bits, like rounding. Pruning takes out the weights or parts that matter least, so there are fewer of them. The two work in their own ways, so they are often used together.

AI

Ollama vs LM Studio

People often think Ollama and LM Studio are rival engines with different models inside. Both are easy front ends that run the same kinds of open models on your own machine. Ollama works from the command line and runs as a local server. LM Studio is a desktop app with windows, buttons and a chat panel. Both can serve other apps.

AI

Ollama vs vLLM

People often think vLLM is just a faster Ollama and the best pick for everyone. They are built for different jobs. Ollama makes it easy for one person to run a model on their own machine. vLLM is made to serve many users at once from GPU servers. It packs their requests together, so each card does more work.

AI

Data Drift vs Concept Drift

People often think data drift and concept drift are two names for one problem. In data drift, the inputs change: the model sees new kinds of cases it was not trained on. In concept drift, the right answer for the same input changes, so the model's rules go out of date. Checking inputs finds the first. Only real results show the second.

AI

Batch vs Real-Time Inference

People often think real-time inference is just the better kind and batch is the old way. They fit different needs. Batch inference scores lots of data at set times and saves it. That is cheap, but only as fresh as the last run. Real-time inference answers each request as it comes. That is fresh, but costs more to keep running.

AI

MLOps vs DevOps

People often think MLOps is just DevOps used on a model. It takes a lot from DevOps, such as tests, version control and doing jobs by code. But a model also depends on its data. So MLOps has to track data versions, train again, and watch a live model for drift. A normal app does not get worse by itself when the world changes, but a model does.

AIDevelopmentcross-domain

Precision vs Recall

People often think precision and recall are two words for how good a model is. They measure two different mistakes. Precision asks: when the model says yes, how often is it right? Recall asks: of all the real yes cases, how many did it find? Raising one usually lowers the other, so you choose based on which mistake costs more.

AI

Accuracy vs Precision

People often use accuracy and precision as if they mean the same thing. In machine learning they do not. Accuracy is the share of all answers that are right. Precision looks only at the yes answers, and asks how many of them were right. A model can be 99% accurate and still have low precision, when real yes cases are rare.

AI

LLM-as-a-Judge vs Human Evaluation

People often think an AI judge is just a cheaper way to get the grades people would give. It is faster and cheaper, and it can grade thousands of answers overnight. But it has biases of its own, such as liking long answers. Human evaluation is slow and costly, yet it is the standard. Most teams use the judge for scale and check it against people now and then.

AI

AI Benchmark vs Eval Set

People often think a high benchmark score means a model will do well in their own app. It may not, since the two tests ask different things. A benchmark is a public test, the same for every model, so it measures general skill. An eval set is your own list of test questions, built from your users and your task. Use benchmarks to choose which models to try, and your eval set to decide.

AI

Speech-to-Text vs Text-to-Speech

People often mix up speech-to-text and text-to-speech, because the names are mirror images. They run in opposite directions. Speech-to-text listens to spoken words and writes them down. Text-to-speech takes written words and reads them out loud. A voice agent uses both: one to hear you, and one to answer.

AI

Text-to-Speech vs Voice Cloning

People often think voice cloning is just text-to-speech with a nicer voice. Text-to-speech reads text in one of a set of stock voices that the tool provides. Voice cloning first copies one real person's voice from a recording, then reads any text in that voice. That is why cloning needs consent, and plain text-to-speech does not.

AI

Image Classification vs Object Detection

People often think image classification and object detection do the same job. Classification gives the whole picture one label, such as cat. It does not say where the cat is. Detection finds each thing in the picture and draws a box around it, with a label for each. Use the first to sort pictures, and the second to find, count or follow things in them.

AI

Object Detection vs Image Segmentation

People often think object detection and image segmentation give the same answer. Detection draws a box around each thing, and the box takes in some of the background too. Segmentation labels each pixel, so it gives the exact shape. A box is cheaper and faster. The exact shape is worth the cost when edges matter, as in a scan or a photo edit.

AI

VLA vs VLM

People often think a vision-language-action model is just a vision language model in a robot. Both look at pictures and read words. A vision language model answers in words, such as 'the cup is on the left'. An action model outputs moves, such as how far to turn each joint to pick the cup up. Most are built on a vision language model, then trained on clips of robots at work.

AI

Prompt Injection vs Jailbreak

People often think prompt injection and a jailbreak are the same attack. In a jailbreak, the user tricks the model into breaking its own safety rules. In prompt injection, someone else's text takes over an app. It hides in a message, a page or a file that the app trusts. A jailbreak aims at the model's training. Injection aims at the app built on it, and the user may never see it.

AI

AI Safety vs AI Alignment

People often think AI safety and AI alignment are just two words for one thing. Alignment is one part of safety. It means making the model try to do what people meant. Safety is wider. It also covers attacks, misuse, tests and filters, and those can go wrong even when a model is aligned. A model can be aligned and still be fooled by a prompt injection.

AI

Explainable AI vs Interpretability

People often think explainable AI and interpretability are the same thing. The first gives a person a reason for one choice, such as what mattered most for a loan. The second looks inside the model to see how it works, part by part. A reason can help and still be a short summary. It may hide how the model truly chose.

AI

AI Governance vs AI Regulation

People often think AI governance just means following AI law. The law is what a government writes, and it applies to everyone. Governance is what one group does inside itself: its list of AI tools, its reviews, and who owns each one. Good governance is how a firm meets the law. But it also covers risks no law names yet.

AI

Deepfake vs AI Misinformation

People often think every deepfake is misinformation. A deepfake is a kind of fake media: a real person's face or voice in a clip they never made. Misinformation is any false claim, in any form. A deepfake in a film with a clear label is not a false claim. And a lie in plain text is false with no deepfake at all.

AI

Chatbot vs AI Agent

People often think a chatbot and an AI agent are the same. A chatbot talks: you ask, it answers, and it waits for you. An agent works toward a goal on its own. It plans steps, uses tools, checks the results, and keeps going until the job is done. Many products now mix both, but the test is whether it acts or only replies.

AI

AI Search Engine vs Deep Research

People often think deep research is just a slower AI search. An AI search engine reads a few top pages and answers in seconds. Deep research runs a loop for many minutes. It plans, searches, reads dozens of pages, and searches again to fill gaps. Use search for a quick fact, and deep research when you need a full report.

AI

Voice Assistant vs Voice Agent

People often think a voice assistant and a voice agent are the same. A voice assistant lives in your phone or speaker and helps you with small tasks. A voice agent is built by a business to talk to callers, such as to book a table or answer a help line. Both listen and speak. But one works for you, and the other works for the company you call.

AI

Llama vs Ollama

People often think Ollama is a version of Llama, because the names sound alike. Llama is a family of models made by Meta. Ollama is a free tool. It gets models and runs them on your own computer. Ollama can run Llama, but it also runs Qwen, Mistral, Gemma and many more. One is the model, and the other is the program that runs it.

AI

Anthropic vs Claude

People often use Anthropic and Claude as if they were one name. Anthropic is the company, founded in 2021. Claude is the family of AI models it makes. The same split holds for OpenAI and GPT, and for Google DeepMind and Gemini. When you sign a contract, you sign with the company. When you call the API, you pick a model.

AI

Azure OpenAI vs OpenAI

People often think Azure OpenAI runs different models from OpenAI. The models are the same. What changes is who runs them and under which deal. With OpenAI, you deal with OpenAI. With Azure OpenAI, Microsoft runs the models in its own cloud, with its own regions, networks and terms. New models can reach OpenAI first.

AI

Model as a Service vs Provisioned Throughput

People often think provisioned throughput is just a faster, pricier kind of paying per token. It is a different deal. With model as a service, you share space and pay for each token. With provisioned throughput, you rent a fixed slice. You pay the same whether you use it or not, and get steady speed. Light or spiky use is cheaper per token. Heavy, steady use can be cheaper to reserve.

AI

Google AI Studio vs Gemini Enterprise Agent Platform

People often think Google AI Studio and Google's cloud AI platform are the same thing, because both run Gemini. AI Studio is a free site for trying prompts and getting a key fast. The cloud platform runs inside a Google Cloud account. It has its own regions, safety rules, contracts and many more models. Start in AI Studio to test an idea. Move to the platform when work data is involved.

AI

GPT vs ChatGPT

People often think GPT and ChatGPT are the same thing, but one is a model and one is an app. GPT is a family of AI models made by OpenAI. ChatGPT is the chat app where most people use those models. Other apps can use GPT through OpenAI's API, without ChatGPT. And ChatGPT adds things the model alone does not have, such as web search, memory and file uploads.

AI

GitHub Copilot vs Microsoft Copilot

People often think GitHub Copilot and Microsoft Copilot are one tool. Both are called Copilot, and Microsoft owns GitHub. But they do different jobs. GitHub Copilot helps people write code, in code tools and on GitHub. Microsoft Copilot is a general helper in Windows, Edge, Bing and Office. Each has its own plans, and you pay for each one on its own.

AI

Cursor vs VS Code

People often think Cursor is an add-on you install inside VS Code. It is a separate editor, built from VS Code's open code. That is why it looks the same and runs most of the same add-ons. Cursor puts AI into the editor itself, with its own Tab, chat and agent. VS Code gets AI through add-ons such as GitHub Copilot.

AIDevelopmentcross-domain

Claude vs Claude Code

People often think Claude Code is a model. Claude is the model, the part that reads and writes text. Claude Code is a program built around it. It gives Claude hands: it reads your files, runs commands and edits code on your computer. The same Claude models also run in the Claude app and in other tools. Claude Code is the one made for coding work.

AI

Remote Control vs Cloud Session

People often think Remote Control runs Claude Code in the cloud. It does not. With Remote Control, the session runs on your own computer, and your phone only shows it. If the computer sleeps, the work stops. A cloud session runs on a server instead. It starts from a fresh copy of your code and keeps going when your laptop is closed.

AI

CLAUDE.md vs Auto Memory

People often think auto memory and CLAUDE.md are the same file. Both reach Claude at the start of a session, but different people write them. You write CLAUDE.md, and your team can share it in git. Claude writes auto memory for itself, from what it learns while working with you. It lives in your home folder, so it is never shared.

AI

Auto Mode vs Bypass Permissions

People often think auto mode skips the checks, like bypass mode. Both stop the prompts, but only one keeps a guard. In auto mode, a second model checks each action and blocks the risky ones. In bypass mode, nothing checks most actions at all. Auto mode is meant for daily work. Bypass mode is only for a container you can throw away.

AI

Subagents vs Agent Teams

People often think agent teams are just more subagents. A subagent lives inside one session. It does one job in its own context and sends a summary back to the main agent, and you do not talk to it. Agent teams are separate sessions. They share a task list, message each other, and you can talk to each one directly.

AI

WebSearch vs WebFetch

People often think WebSearch reads the pages it finds. It does not. WebSearch only brings back titles and links. To read a page, Claude has to call WebFetch on one link. WebFetch then opens that page and pulls out only the part Claude asked about. So a search finds where to look, and a fetch does the reading.

AI

Edit vs Write

People often think Write can change part of a file. It cannot. Write puts in a whole new file and throws the old text away. Edit swaps one exact piece of text and leaves the rest alone. So Claude uses Write for a new file and Edit for a small fix. A rule for Edit on a folder covers Write there too.

AI

Glob vs Grep

People often think Glob and Grep both search inside files. Only Grep does. Glob matches file names, such as every file that ends in .ts. Grep looks inside the files for a word or a pattern. There is one more catch. Glob also finds the files .gitignore hides, while Grep skips them.

AI

Cron Tools vs RemoteTrigger

People often think a /loop task keeps running after they close the terminal. It does not. The cron tools only fire while the session is open. And a repeating task ends after seven days. RemoteTrigger sets up a task on claude.ai instead. That task runs in the cloud on its own schedule, even with your computer off.

AI

Claude Code GitHub Actions vs Code Review

People often think Code Review is the GitHub Action with a review prompt. It is a separate service, run by Anthropic with no workflow file. The GitHub Action runs Claude Code in your own workflow, on your runners, and you write its steps. Code Review runs on Anthropic's servers. It reviews each pull request with several agents and posts ranked findings. It needs a Team or Enterprise plan.

AI

/branch vs /fork

People often think /branch and /fork do the same thing. Both copy the chat, but they put you in different places. /branch moves you into the copy, and the original waits for you in /resume. /fork sends the copy off to run as a background session, and you stay where you were. /branch also shares your files, while a fork is told to work in its own worktree.

AI

/clear vs /compact

People often think /clear and /compact both just make room. They keep very different things. /compact squeezes the chat into a summary, so Claude still knows what you were doing. /clear starts a new chat with an empty context, so Claude knows nothing of it. Use /compact to go on with the same task. Use /clear to start a new one.

AI

/exit vs /stop

People often think /exit ends a background session. It only lets go of it, and the work goes on. /stop is the one that ends it, and it keeps the chat and the worktree. In a normal session, /exit closes Claude Code and saves the chat for later.

AI

/code-review vs /simplify

People often think /simplify is a lighter /code-review. They look for different things. /code-review looks for bugs, such as wrong logic or a missed case, and reports them. /simplify looks for cleanups, such as a helper you could reuse or code that could be shorter, and makes those changes. Run /code-review to find what is broken. Run /simplify to tidy code that already works.

AI

/code-review vs Claude Code Review

People often think /code-review and the Code Review service are the same thing. /code-review is a command you run yourself, in your own session, on any plan. It reads your CLAUDE.md and reports in your chat. Code Review is a managed service that reviews each pull request on GitHub without anyone asking. It reads REVIEW.md and posts comments on the pull request. It needs a Team or Enterprise plan.

AI

/add-dir vs /cd

People often think /add-dir and /cd both just change folders. /add-dir keeps the folder the session started in and lets Claude work in a second one too. /cd moves the whole session to a new folder, so Claude works from there instead. Both keep the chat. Use /add-dir when the task needs both folders, and /cd when the work has moved.

AI

/doctor vs /debug

People often think /doctor and /debug both just find what is wrong. They look in different places. /doctor checks your setup: the install, the settings files, your extensions and CLAUDE.md. It proposes fixes and applies them after you agree. /debug turns on the log of a running session and reads it, to find a problem that shows up while you work. Run /doctor when something will not load, and /debug when something goes wrong mid-task.

AI

/upgrade vs /restart

People think /upgrade updates Claude Code, because /update sounds like the same thing. It is not. /upgrade opens a web page where you move to a bigger Claude plan, with higher limits. /restart, which also answers to /update, closes Claude Code and opens it again with the same chat, so a new version that is already installed starts to run. One changes what you pay for. The other changes the version you run.

AI

/reload-skills vs /reload-plugins

People think /reload-skills and /reload-plugins are two names for one reload. They are not. /reload-skills looks through your own skill and command folders again, so a skill you just wrote shows up. /reload-plugins loads your installed plugins again, with their skills, MCP servers and other parts. It stops if the change would clear the prompt cache, unless you add --force. A new skill in .claude/skills needs the first. A changed plugin needs the second.

AI

/update-config vs /config

People think /update-config is just a longer name for /config. Both change settings, but they work in very different ways. /config opens the settings panel, or sets one key such as theme=dark, and works at once. /update-config is a skill. You say a change in plain words, such as allow a command or add a hook, and Claude finds the right settings.json and edits it. Use /config for a switch. Use /update-config for a rule or a hook.

AI

Claude Code CLI Command vs Slash Command

People think claude doctor and /doctor are the same command typed two ways. They are not. A CLI command is typed in your shell after claude, runs, prints and stops, with no chat. Scripts and CI jobs can run it. A slash command is typed inside a running session and acts on that chat. So claude doctor only reads your install, even when Claude Code will not start, while /doctor runs inside a chat and offers fixes.

AI

claude respawn vs claude rm

People think claude rm is how you clear out a stuck background session and start again. Most of the time they want claude respawn. respawn restarts the session's process and keeps the chat, so the work goes on from where it was. rm takes the session off the list. Its transcript stays on disk, but the session is no longer running. Use respawn to fix a session. Use rm when you are done with it.

AI

claude remote-control vs claude self-hosted-runner

People think claude remote-control and claude self-hosted-runner both put Claude Code in the cloud. Neither does, and they serve different people. remote-control lets one person drive Claude Code on their own computer from the Claude app. A self-hosted runner adds a company's machine to a pool that runs cloud sessions for many people. The code stays on the company's network. One is for your laptop. The other is for a whole fleet.

AI

claude plugin validate vs claude plugin eval

People think a plugin that passes claude plugin validate is a plugin that works. Validate only checks the shape of the files: the manifest, and the skills, agents and commands in the folder. claude plugin eval runs prompts through the plugin many times and scores the answers, with it and without it. So it shows if the plugin helps. A plugin can pass validate and still make Claude worse. Run validate on each change, and eval when what it does changes.

AI

--name vs --session-id

People think --name and --session-id are two ways to label a session. They are not the same kind of thing. --name is a label for people: any short text you will remember, shown in the /resume list and the title bar, and you can change it with /rename. --session-id is the session's real ID, and it must be a UUID. A script sets it so it can find that exact session later. Pick --name for yourself and --session-id for a script.

AI

--max-turns vs --max-budget-usd

People think --max-turns and --max-budget-usd are two ways to set the same limit. They count different things. --max-turns counts rounds: each time Claude acts and reads the result is one turn. --max-budget-usd counts dollars, using Claude Code's own estimate. A run with few turns can still cost a lot if each turn reads big files. A cheap run can still take many turns. Set both on unattended runs.

AI

--tools vs --allowedTools

People think --allowedTools limits Claude to the tools it lists. It does not. --allowedTools only says which tools may run without asking you first. Every other tool is still there and can ask. --tools is the one that decides which built-in tools exist at all, and anything it leaves out is gone. To lock a session down, use --tools. To stop it asking about safe tools, use --allowedTools.

AI

--safe-mode vs Bare Mode

People think --safe-mode and --bare do the same thing, since both skip your add-ons. They have different jobs. --safe-mode is for finding a fault. It turns off CLAUDE.md, plugins, hooks and the rest, but sign-in, tools and permissions work as usual. Bare mode is for scripts and CI. It never reads your saved sign-in, so it needs an API key, and it gives Claude just a few tools. Use safe mode when something is broken. Use bare mode for a run that must act the same on each machine.

AI

--system-prompt vs --append-system-prompt

People think --system-prompt and --append-system-prompt both just add rules. Only one of them adds. --append-system-prompt puts your text after Claude Code's default prompt. So Claude keeps its guide to the tools, its safety rules and its coding habits. --system-prompt throws the default away and uses only your text. For a coding task with a few extra rules, append. Replace only when Claude should not act as a coding helper at all.

AI

--agent vs --agents

People think --agent and --agents are the singular and plural of one flag. They do different jobs. --agent picks a custom agent and runs the main session as it, with that agent's prompt, tools and model. --agents defines new subagents in JSON for this session, helpers the main session can hand work to. You can use both: --agent sets who is in charge, and --agents adds the team.

AI

ANTHROPIC_API_KEY vs ANTHROPIC_AUTH_TOKEN

People think ANTHROPIC_API_KEY and ANTHROPIC_AUTH_TOKEN are two names for one key. They send different headers to different checkers. The API key is a Console key. It goes out as X-Api-Key, and Anthropic checks it. The token goes out as a Bearer header, and a gateway checks it. When both are set, the token wins. Use the key to reach Anthropic, and the token to reach your gateway.

AI

--betas vs ANTHROPIC_BETAS

People think the --betas flag and the ANTHROPIC_BETAS variable are one switch in two places. Both add anthropic-beta headers, but they reach different runs. The flag lasts for one run, and it only works when you sign in with an API key. The variable lasts for each session that sees it, and it works with any sign-in, a plan too. On a plan, only the variable works.

AI

ANTHROPIC_MODEL vs ANTHROPIC_DEFAULT_MODEL

People think ANTHROPIC_MODEL and ANTHROPIC_DEFAULT_MODEL both set the model you start on, so either will do. They sit at the two ends of the line. The first ranks just below the --model flag. Each launch goes back to it, even after you save a new model with /model. The second is last in line, so any flag, setting or saved choice wins over it. Use the first to fix a model, and the second to set a fallback.

AI

ANTHROPIC_MODEL vs ANTHROPIC_DEFAULT_OPUS_MODEL

People think ANTHROPIC_DEFAULT_OPUS_MODEL makes Claude Code run on Opus. It does not pick a model at all. It only says which full model the short name opus stands for. ANTHROPIC_MODEL is what picks the model a session starts on. So on a cloud you often set both. The pin says what opus means there, and the other says to use it.

AI

CLAUDE_CODE_AUTO_COMPACT_WINDOW vs CLAUDE_AUTOCOMPACT_PCT_OVERRIDE

People think these two do the same thing in two units. One sets the size of the box, and the other sets how full it gets. The window variable sets how many tokens the chat can hold, from 100,000 to 1,000,000. The percent variable sets the point in that window where squeezing starts, and it can only move that point lower. Use the window to cap cost per request, and the percent to squeeze a little sooner.

AI

BASH_DEFAULT_TIMEOUT_MS vs BASH_MAX_TIMEOUT_MS

People think raising BASH_MAX_TIMEOUT_MS gives every command more time. It does not. BASH_DEFAULT_TIMEOUT_MS is the time a command gets when Claude asks for none, 2 minutes unless you change it. BASH_MAX_TIMEOUT_MS is the most Claude may ask for, 10 minutes unless you change it. Raise the default when all your commands are slow, and the top when only a few are.

AI

MCP_TIMEOUT vs MCP_TOOL_TIMEOUT

People think MCP_TIMEOUT is the limit on how long an MCP tool can run. It is the wait for a server to start, 30 seconds by default. MCP_TOOL_TIMEOUT is the limit on one tool call, about 28 hours by default, with a 60 second limit per request on servers reached over HTTP. A server that never shows its tools needs the first. A tool that fails partway needs the second.

AI

CLOUD_ML_REGION vs VERTEX_REGION_CLAUDE_*

People think CLOUD_ML_REGION and the VERTEX_REGION_CLAUDE_ variables do the same job. One sets the region for every model. The other sets it for one model. Claude Code checks the model's own variable first. If it is not set, it uses CLOUD_ML_REGION, and then us-east5. Set CLOUD_ML_REGION once. Add a model's own only when that model needs a different region.

AI

CLAUDE_CODE_DISABLE_THINKING vs MAX_THINKING_TOKENS=0

People think CLAUDE_CODE_DISABLE_THINKING turns thinking off. It does not. It only leaves the thinking field out of the request, for a proxy that rejects the field. A model that thinks by default may still think. MAX_THINKING_TOKENS=0 is the one that asks the Anthropic API for no thinking. On a cloud provider, both just leave the field out. Some new models cannot have thinking turned off at all.

AI

ENABLE_PROMPT_CACHING_1H vs CLAUDE_CODE_PROMPT_CACHE_TTL

People think ENABLE_PROMPT_CACHING_1H is the way to set how long the cache lasts. It is one switch for everything. CLAUDE_CODE_PROMPT_CACHE_TTL sets the main chat alone, and its SUBAGENT twin sets the helpers. Both win over the switch. So set the switch for a quick change, and the two TTL variables when the main chat and helpers need different times.

AI

DISABLE_AUTO_COMPACT vs DISABLE_COMPACT

People think DISABLE_AUTO_COMPACT and DISABLE_COMPACT are the same switch. The first stops only the automatic squeeze, so /compact still works when you run it. The second stops all of it, /compact included. Use the first to choose when the chat is summed up. Use the second only when the chat must never be summed up at all.

AI

DISABLE_AUTOUPDATER vs DISABLE_UPDATES

People think DISABLE_AUTOUPDATER stops all updates. It only stops the ones in the background, so anyone can still run claude update. DISABLE_UPDATES blocks every update, by hand too, and claude install as well. A person who wants to update on their own day wants the first. A company that ships its own tested version wants the second.

AI

CLAUDE_CODE_DISABLE_CLAUDE_MDS vs CLAUDE_CODE_DISABLE_AUTO_MEMORY

People think turning off auto memory drops CLAUDE.md too. It does not. The auto memory switch stops only the notes Claude keeps for itself. Your own CLAUDE.md files still load. The CLAUDE_MDS switch drops every memory file, yours and the auto ones. Use the first to stop Claude learning. Use the second for a clean test with no notes.

AI

CLAUDE_CODE_ENABLE_TELEMETRY vs DISABLE_TELEMETRY

People think the ENABLE and DISABLE telemetry switches are the on and off of one thing. They are two roads. The first sends logs and stats to a tool you run. The second stops the usage data Claude Code sends home. You can set both. Your own team sees it all, and nothing goes home.

AI

DISABLE_TELEMETRY vs CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC

People think DISABLE_TELEMETRY stops all that Claude Code sends home. It stops only usage data and flags. Crash reports, updates, release notes and /feedback still go out. The NONESSENTIAL_TRAFFIC switch stops all of those at once. Only calls to the model are left. Use the first for privacy, and the second for a locked network.

AI

CLAUDE_CODE_NO_FLICKER vs CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN

People think these two do the same thing, since both are about the screen. They do the opposite. NO_FLICKER turns fullscreen mode on. DISABLE_ALTERNATE_SCREEN turns it off. It keeps the chat in your terminal's own history. If both are set, the second wins.

AI

CLAUDE_ENV_FILE vs CLAUDE_CODE_SHELL_PREFIX

People think CLAUDE_ENV_FILE and the shell prefix both just run something before each command, so either will do. They work at different levels. The env file runs inside the same shell as the command, so what it exports, the command sees. The prefix is a program that wraps the whole command from outside. Use the env file to set up, and the prefix to log or check.

AI

CLAUDE_CODE_SHELL vs CLAUDE_CODE_SHELL_PREFIX

People think the prefix picks the shell, since the names are so close. It does not. CLAUDE_CODE_SHELL picks bash or zsh to run each command. The prefix names your own program that wraps that shell, most often to log what runs. You can set one, the other or both.

AI

CLAUDECODE vs CLAUDE_CODE_REMOTE

People think CLAUDECODE tells a script it is in the cloud. It only says the script was started by Claude Code, and that is true on your laptop too. CLAUDE_CODE_REMOTE is the one set to true only in a cloud session. A hook that should act only on a fresh cloud machine checks the second.

AI

Read Permission Rule vs Edit Permission Rule

People think a Read deny rule only stops reading, and an Edit rule is needed to stop changes. A Read deny rule stops both. Claude cannot read the path, and cannot edit or write it either. An Edit rule only covers changes, so a file can be readable but locked. To hide a secret, deny Read. To keep a file safe but in view, deny Edit.

AI

Additional Directories vs Primary Working Directory

People think an added folder works just like the folder Claude Code started in. For files it does: Claude reads and edits both under the same rules. But only the start folder is a home for setup. Its settings, hooks, MCP servers and CLAUDE.md load. An added folder gives file access, and most of its setup stays asleep.

AI

Bash Permission Rule vs Sandbox

People think a Bash deny rule is a wall around a program. It is a check on the text Claude writes. Bash(rm *) stops rm, but not /bin/rm or sh -c 'rm'. The sandbox works lower down. It limits which files and hosts any command can reach, however it is spelled. Use rules to cut prompts, and the sandbox for a hard limit.

AI

PreToolUse vs PostToolUse

People think both hooks can stop a bad tool call. Only one can. PreToolUse fires before the tool runs, so it can allow, deny or ask. PostToolUse fires after the tool has run, so the change is already made. It can only tell Claude what went wrong. Guard in the first, check and tidy in the second.

AI

PreToolUse vs PermissionRequest

People think both fire before each tool call. PreToolUse does. PermissionRequest fires only when Claude Code is about to ask you. So a call your rules allow never reaches it. Use PreToolUse to check each call. Use PermissionRequest to answer the prompts you would get.

AI

Stop Hook vs StopFailure Hook

People think a Stop hook runs at the end of every turn, whatever happens. It only runs when Claude finishes its answer on its own. A turn cut short by an API error, such as a rate limit, fires StopFailure instead. And only Stop can send Claude back to work. StopFailure ignores what the hook prints, so it is for alerts.

AI

Prompt Hook vs Agent Hook

People think both just ask a model. A prompt hook asks once and sees only the event. An agent hook starts a helper that can open files with tools, for up to 50 turns. Use a prompt hook for a quick judgment, and an agent hook when the answer is in the files.

AI

FileChanged vs PostToolUse

People think a PostToolUse hook on Edit sees every change made to a file in the project. It sees only Claude's own tool calls. A script Claude runs, or your own editor, can change the file and it never fires. FileChanged watches the file itself, so it fires whatever wrote it. But only PostToolUse knows which tool call did it.

AI

Hook Matcher vs Hook if Condition

People think the matcher can pick a command, such as git push. It only sees the tool name, so Bash matches every shell command. The if field reads the call's input, with one permission rule such as Bash(git push *). Use the matcher to pick the tool, and if to pick the calls inside it.

AI

Setup Hook vs SessionStart Hook

People think a Setup hook runs each time Claude Code starts. It does not. It runs only when you ask for it with --init-only, --init or --maintenance. SessionStart is the one that runs on every start, resume and clear. Put one-time installs in Setup, and quick context in SessionStart.

AI

settings.json vs settings.local.json

People think settings.local.json is a backup of settings.json. It is a second file with a different owner. The project file is shared in git, so it sets the rules for the whole team. The local file is yours alone, stays out of git, and wins when both set a key. Rules you approve from a prompt are saved in the local one.

AI

.mcp.json vs ~/.claude.json

People think every MCP server lives in .mcp.json. Only the team's do. A server you add for yourself, in one project or in all of them, is saved in ~/.claude.json in your home folder. So .mcp.json goes into git and every teammate sees it, while ~/.claude.json never leaves your machine.

AI

.claude vs ~/.claude

People think there is one .claude folder. There are two with the same name. The one in a project holds that project's setup and is shared through git. The one in your home folder holds your own setup for every project, plus the data Claude Code writes as you work. A project file beats the same key in your home one.

AI

disableAutoMode vs disableBypassPermissionsMode

People think one of these keys turns off both risky modes. Each one turns off one. disableAutoMode takes auto mode out of the cycle, where a model approves calls for you. disableBypassPermissionsMode blocks the mode that skips every check, and refuses the flag that starts it. A team that wants neither sets both.

AI

model vs availableModels

People think model limits which models can be used. It only picks the one a session starts on, and you can switch. availableModels is the list anyone may pick from at all. Set in managed settings, it holds against the model key, the flag and /model.

AI

tui vs viewMode

People think tui and viewMode both set how much you see. tui picks how Claude Code draws the screen: the classic view, or a full-screen one with no flicker. viewMode picks how much of each turn shows: everything, the usual amount, or just the answer. Focus view needs the full-screen tui.

AI

spinnerVerbs vs spinnerTipsEnabled

People think both keys change the same spinner text. spinnerVerbs changes the one moving word, such as Baking, and can add yours or swap them all. spinnerTipsEnabled shows or hides the longer tips about features on the same line. One is for fun, the other is for learning.

AI

allowedMcpServers vs deniedMcpServers

People think the allow list wins when a server is on both. The deny list always wins, and it is checked first. The allow list is a list of the only servers that may load, while the deny list names the ones that never may. Deny lists from every file join, so anyone can block a server for themselves.

AI

availableModels vs deniedModels

People think these are one list said two ways. availableModels names the only models people may pick, and works from any file. deniedModels names models that are blocked even when the first list has them. It only works from managed settings. IT uses it to hold back one new model and leave the rest of the list alone.

AI

denyRead vs sandbox.credentials

People think both keys just hide a secret file. denyRead blocks a path, so a command cannot read it at all. sandbox.credentials can do that too. It can also mask a secret: the command sees a stand-in, and the proxy puts the real value back on the way out. Mask is for a secret a command must use but should not see.

AI

excludedCommands vs allowUnsandboxedCommands

People think both keys let commands out of the sandbox in the same way. excludedCommands is a fixed list you write, and those commands always run with no fence. The other key decides if Claude may ask to rerun a blocked command with no fence. A strict setup turns the second key off and keeps the list short.

AI

Esc vs Ctrl+C in Claude Code

People think Esc and Ctrl+C both just stop Claude. Esc stops the turn, keeps the work so far, and sends anything you queued. Ctrl+C stops it too, but with nothing running it clears your prompt, and a second press quits Claude Code. Use Esc to steer, and Ctrl+C when you want out.

AI

Ctrl+T vs /tasks

People think Ctrl+T shows the tasks running in the background. It shows Claude's to-do checklist: the steps Claude plans and ticks off. The /tasks command is the one that lists the shells and subagents running in the background, so you can check or stop them.

AI

! Shell Mode vs the Bash Tool

People think a ! command and a command Claude runs are the same thing. With !, you run the command yourself, at once, with no approval and outside the sandbox. The Bash tool is how Claude runs a command, so it goes through your permission rules and the sandbox. Both put the output in the chat.

AI

query() vs ClaudeSDKClient

People think query() and ClaudeSDKClient are two names for the same thing. query() runs a task and ends, and each new call starts fresh unless you pass resume or continue. ClaudeSDKClient is a Python class that keeps one session open, so each prompt you send remembers the ones before. Use query() for one job, and the client for a chat.

AI

Streaming Input vs Single Message Input

People think the two input modes are the same thing in a different form. Single message input sends one prompt and ends. It takes no images and cannot be stopped mid-turn. Streaming input keeps a session open, so you can send more, add images and stop it. Pick single message input for one-shot jobs, and streaming input for a chat.

AI

allowedTools vs disallowedTools

People think allowedTools is a full list, so a tool left off it is blocked. It is not. allowedTools only lets the listed tools run with no prompt, and other tools still go to the permission mode. disallowedTools is the one that blocks, and its block holds in every mode, even bypass.

AI

tools vs allowedTools

People think tools and allowedTools do the same job. tools decides which built-in tools exist, so a tool left out is never seen by Claude. allowedTools decides which of the tools that exist run without asking. Use tools to shrink the toolbox, and allowedTools to skip the prompt.

AI

continue vs resume (Agent SDK)

People think continue and resume are two names for one thing. continue takes the latest session in the current folder, with no ID. resume takes the one session whose ID you pass, even an old one. Use continue when your app runs one chat at a time, and resume when it runs many.

AI

SDK MCP Server vs MCP Server

People think every MCP server is a separate program you start and connect to. An SDK MCP server is not, because it runs inside your own app. Its tools are plain functions in your code. A normal MCP server runs on its own, as a local process or a remote service. Use the SDK kind for tools that call your own code, and a normal one to share tools across apps.

AI

SOUL.md vs USER.md (Hermes)

People think SOUL.md is where facts about them go. It is not. SOUL.md sets who the agent is, its tone and style, and you write it yourself. USER.md holds who you are, your name and how you like replies, and the agent writes it. The two never feed each other.

AI

Hermes Memory vs session_search

People think Hermes forgets anything not saved to memory. It does not. Memory is a few key facts loaded into every prompt, so it is small on purpose. session_search looks through every past chat on demand, at no model cost. Save the facts it needs every time, and let search find the rest.

AI

Hermes Memory vs Hermes Skill

People think memory and skills both just store things the agent learned. In fact, each one stores a very different kind of thing. Memory holds short facts and is loaded into every prompt. A skill holds a whole method, step by step, and is loaded only when a task needs it. Put a fact in memory, and a how-to in a skill.

AI

Hermes Allowlist vs DM Pairing

People think the allowlist and DM pairing are two rival safety modes. They are two ways to fill the same guest list. With the allowlist, you type each user ID into the .env file. With pairing, a new person messages the bot, gets a one-time code, and you approve it. Use the list for a fixed team, and pairing to add people one by one.

AI

delegate_task vs execute_code

People think delegate_task and execute_code both just run work side by side. In fact, they solve two quite different problems. delegate_task starts new agents that think for themselves and send back a summary. execute_code runs one Python script that calls tools in a set order and prints a result. Hand off work when each part needs thought. Use a script when you know the steps.

AI

Hermes Profile vs Bot Mode

People think a Hermes bot is something new that sits beside profiles. It is not. Every bot is a profile, with its own memory, skills and keys. Bot Mode is a desktop view that gives a profile a name, a face and a place in group chats. Make a profile for a separate agent, and use Bot Mode to see your agents as a team.

AI

Reasoning Model vs Instruct Model

People often think a reasoning model is just a smarter chat model. It is trained to think out loud first, spending extra tokens on steps before it answers. An instruct model answers straight away. For hard maths or code the thinking pays off; for quick tasks it only adds time and cost.

AI

Program vs Portfolio

People think a portfolio is just a big program. A program holds projects that share one goal. A portfolio holds all the work a company pays for, linked or not. Its job is to choose which work goes ahead.

Business

Business Case vs Project Charter

People think the charter and the business case are one paper. The business case argues why a project is worth its cost. The charter comes after that yes: it starts the project and gives the manager the power to run it.

Business

Project Manager vs Project Sponsor

People think the project sponsor runs the project. The sponsor wants it, pays for it and owns the reason it exists. The project manager runs the work each day and answers for what gets delivered.

Business

Project Charter vs Project Management Plan

People think the project charter is the project plan. The charter is a short signed page that starts the project. The plan comes after it and says, in detail, how the work will be run, checked and closed.

Business

Milestone vs Deliverable

People think a milestone is a deliverable. A deliverable is a thing the project hands over, like a report or a website. A milestone is a point in time, such as the day that report is signed off. It takes no time and makes nothing.

Business

Contingency Reserve vs Management Reserve

People think the two reserves are one pot of spare money. One is for risks the team has listed, and the project manager can spend it. It sits inside the cost baseline. The other is for surprises no one could list. It sits outside the baseline, and a leader must say yes before it is used.

Business

Fast-Tracking vs Crashing

People think fast-tracking and crashing are the same trick. Both shorten the critical path. Fast-tracking runs tasks at the same time that were planned one after another, which adds risk. Crashing adds people or money to a task so it ends sooner, which adds cost.

Business

Scope Creep vs Gold Plating

People think scope creep and gold plating are the same problem. Both put work in the project that no one approved. Scope creep comes from outside, when a client keeps asking for small extras. Gold plating comes from inside, when the team adds extras no one asked for.

Business

Earned Value vs Actual Cost

People think earned value is the money spent so far. Actual cost is the money spent. Earned value is what the finished work was planned to cost. When actual cost is higher than earned value, the work is costing more than it should.

Business

Project Schedule vs Gantt Chart

People think the Gantt chart is the schedule. The schedule is the plan itself: the tasks, how long they take, the order and the dates. A Gantt chart is one way to draw it, as bars on a timeline. The same schedule can also be a list or a network.

Business

Scope Statement vs Project Charter

People think the scope statement is the project charter. The charter is a short page that starts the project and names its manager. The scope statement comes later. It says in detail what the project will make, what it leaves out and how the work will be accepted.

Business

Project Risk vs Project Issue

People think a risk and an issue are the same kind of problem. A risk has not happened yet, and it may never happen. An issue is happening now and needs a fix. Risks get a plan ahead of time, and issues get action today.

Business

Quality Assurance vs Quality Control

People think quality assurance and quality control are both just checks. One checks how the work is done, to stop faults early. The other checks the finished thing, to catch faults before the customer does.

Business

Project Quality vs Grade

People think a low grade means low quality. Grade is how many features something has. Quality is whether it works as promised. A basic product that works is fine, and a luxury one full of faults is a problem.

Business

Fixed-Price vs Time and Materials Contract

People think these two contracts differ only in how the bill is paid. The real difference is who pays when the job costs more. A fixed price puts that risk on the seller, so the buyer knows the bill. Time and materials pays for every hour, so the buyer carries the risk but can start before the work is fully known.

Business

Risk Mitigation vs Contingency Plan

People think a contingency plan is a way to mitigate a risk. Mitigation acts now, before anything goes wrong, to make the risk smaller. A contingency plan waits. It only runs if the risk happens and its trigger fires.

Business

Residual Risk vs Secondary Risk

People think a secondary risk is the part of a risk left over after a response. That left over part is residual risk. A secondary risk is a brand new risk that the response itself creates.

Business

Request for Proposal vs Statement of Work

People think an RFP and a statement of work are the same paper. The statement of work says what work is to be done. The RFP asks sellers to bid for it, and it often has the statement of work inside.

Business

Statement of Work vs Scope Statement

People think a statement of work is the scope statement. The scope statement covers all the work of the whole project. A statement of work covers just the part a seller will do, and it goes into their deal.

Business

Team Charter vs Project Charter

People think a team charter is the project charter for a team. The project charter comes from the sponsor and starts the project. The team charter is written by the team itself and sets how its members work together.

Business

Status Report vs Project Dashboard

People think a project dashboard can replace the status report. A dashboard shows live numbers that anyone can check. A status report is a person's weekly view: what the numbers mean, what is at risk and what help is needed.

Business

Meeting Minutes vs Decision Log

People think the minutes are where you look up a decision. Minutes cover one meeting, and soon there are fifty of them. A decision log keeps every big decision in one list, with the reason, from start to end.

Business

Steering Committee vs Project Sponsor

People think the sponsor and the steering committee are the same thing. The sponsor is one senior person who owns the project. The steering committee is a group of leaders that makes the biggest calls together. The sponsor often sits on it or chairs it.

Business

Agile vs Waterfall Model

People think agile means no plan and waterfall means a plan. Both plan. Waterfall fixes the whole plan first and delivers once at the end. Agile plans in short steps and delivers a small piece every few weeks.

Business

Agile vs Scrum

People think agile and Scrum are the same thing. Agile is a set of values for building in small pieces. Scrum is one framework that puts those values to work. Kanban and Extreme Programming are others.

Business

Scrum vs Kanban

People think Kanban is a lighter Scrum. Scrum works in sprints, with set roles and events. Kanban has no sprints and no set roles. It caps the work in progress and lets work flow one item at a time.

Business

Kanban Board vs Kanban

People think Kanban is just the board. The board shows the steps of the work. Kanban is the method around it: caps on work in progress, pull, and steady checks on how the work flows.

Business

Scrum Master vs Project Manager

People think the Scrum Master is the team's project manager. A project manager plans the work, hands it out and answers for the result. A Scrum Master hands out nothing. They coach the team and clear what blocks it.

Business

Product Owner vs Project Manager

People think the Product Owner runs the whole project, like a manager. The Product Owner decides what gets built, and in what order. They do not plan the schedule or manage the team. In Scrum, the team plans its own work.

Business

Sprint Review vs Sprint Retrospective

People think the review and the retrospective are one meeting. The review looks at the product, with stakeholders in the room. The retrospective looks at how the team worked, and only the team is there.

Business

Lead Time vs Cycle Time

People use the two names for one clock. Lead time starts when someone asks for the work. Cycle time starts when someone begins it. The gap between them is time spent waiting in a queue.

Business

Iteration vs Increment

People think an iteration and an increment are the same thing. An iteration is a block of time. An increment is the working piece of product that comes out of it. One is the when, the other is the what.

Business

Definition of Done vs Acceptance Criteria

People think acceptance criteria are a story's own Definition of Done. They are not. Acceptance criteria belong to one story and say what it must do. The Definition of Done is one list for all work, like tested and reviewed. A story is done only when it meets both.

Business

Definition of Ready vs Definition of Done

People think the two are one checklist. The Definition of Ready is checked before a story enters a sprint: is it clear enough to start? The Definition of Done is checked at the end: is it truly finished? One guards the way in, the other the way out.

Business

Epic vs User Story

People think an epic is a different kind of thing from a story. It is a story too big to finish in one sprint. Before anyone builds it, it is split into user stories that each fit a sprint.

Business

User Story vs Project Requirements

People think a user story is a requirement in a new format. A requirement tries to write down all the detail up front. A user story is a short note that promises a talk. The detail comes later, in that talk and in the acceptance criteria.

Business

Backlog Refinement vs Sprint Planning

People think refinement is just part of sprint planning. Refinement happens all through the sprint and gets items ready: split, clear and sized. Sprint planning is the start of a sprint, where the team picks ready items and plans how to build them.

Business

Burndown Chart vs Burnup Chart

People think the two charts show the same thing. A burndown counts work left and falls to zero. A burnup counts work done and climbs to a scope line. Only the burnup shows when work is added, so a flat burndown can hide a team that is working hard.

Business

Velocity vs Throughput

People think velocity and throughput are the same number. Velocity adds up the story points a team finishes in a sprint. Throughput counts the items it finishes in a week, with no sizes. Both help a team forecast, but only velocity needs the work sized first.

Business

ETL vs ELT

People think ETL and ELT are two names for one thing. They use the same three steps in a different order. ETL cleans data on a separate server before it lands. ELT loads raw data first and cleans it inside the warehouse with SQL. That choice decides what gets stored and where the work runs.

Data

Batch Processing vs Stream Processing

People think stream processing is always better because it is faster. A batch job runs on a pile of data at set times. A stream job handles each event as it comes. Streams cost more to build and to fix. Pick a stream only when a late answer is worth nothing.

Data

Lambda Architecture vs Kappa Architecture

People think Kappa is just Lambda with a part cut off, and it is more than that. Lambda runs the same logic twice, once in a nightly batch and once live, and joins the two. Kappa runs it once, in a stream. When the logic changes, Kappa replays its log from the start instead of keeping a batch path.

Data

Apache Iceberg vs Delta Lake

People think Iceberg and Delta Lake store data in different ways, and they do not. Both keep plain Parquet files and add a log of which files make up each version. Delta Lake began at Databricks and fits best there. Iceberg began at Netflix and is read by the widest mix of engines. Pick by the engines you run.

Data

Data Warehouse vs Data Mart

People think a data mart is just a small data warehouse, and it is not quite. A warehouse holds every source in the firm, cleaned and kept for years. A mart is one team's slice, cut from the warehouse for its own questions. Build marts from the warehouse, or each one will clean the data its own way.

Data

Fact Table vs Dimension Table

People mix up fact tables and dimension tables, and they hold different things. A fact table holds the numbers from events, one row per sale or click, and it grows fast. A dimension table holds the words you slice by, such as product or store. It is short and wide. Reports sum the facts and group by the dimensions.

Data

Fivetran vs Airbyte

People think Airbyte is just a free Fivetran, and that is only half true. Both copy data from apps into a warehouse with ready-made connectors. Fivetran is paid and run for you, and it charges by the rows that change each month. Airbyte's code is open. You can run it on your own servers for free, or pay for its cloud.

Data

Kafka Partition vs Data Partitioning

People think a Kafka partition and a table partition are the same, and they are not. A table partition splits stored rows by a column, such as the date. Then a query can skip the parts it does not need. A Kafka partition splits a stream into lanes, each kept in order. Then more readers can work at once. One helps you skip, the other helps you share.

Data

pandas vs Polars

People think Polars is just a faster pandas, and it is more than that. Both hold a table in memory on one machine. pandas runs each step as soon as you write it, mostly on one core. Polars can plan the whole job first, read only what the answer needs, and use every core. The code looks alike, but the habits differ.

Data

Apache Flink vs Spark Structured Streaming

People think Apache Flink and Spark Structured Streaming do the same thing at different speeds, and that is close but not all of it. Flink handles each event as it arrives, in milliseconds. Spark gathers events into tiny batches and runs the same code it uses for batch jobs. Pick Flink for speed and big state, and Spark to reuse what your team knows.

Data

Dagster vs Prefect

People think Dagster and Prefect are two names for one idea, but they start from different places and suit different teams. Dagster plans around the tables and files a pipeline builds. It shows which ones are fresh. Prefect plans around plain Python functions, and adds tracking and retries to them. Pick Dagster if you think in tables, and Prefect if you think in scripts.

Data

Airflow DAG vs DAG Run

People say DAG when they mean one run of it, and the two are not the same. An Airflow DAG is the pipeline itself: a Python file with tasks, arrows and a schedule. A DAG run is one time that pipeline ran, tied to one date of data. One daily DAG makes 365 runs a year, and each run can pass or fail on its own.

Data

Tableau vs Power BI

People think Tableau and Power BI are the same tool from two firms, and they lean different ways. Tableau is built to explore. Drag fields onto a page, and a chart appears. Power BI is built on a data model, with sums written in DAX. It comes with many Microsoft 365 plans. Firms on Microsoft often pick Power BI. Teams that explore a lot often pick Tableau.

Data

Metabase vs Apache Superset

People think Metabase and Superset are two copies of one free tool, but they are made for two very different kinds of user. In Metabase, anyone can ask a question by clicking, with no SQL. Superset is for people who write SQL, and it has many kinds of chart. Pick Metabase when most users do not code. Pick Superset when most of them write SQL.

Data

Databricks vs Microsoft Fabric

People think Databricks and Microsoft Fabric are two names for the same kind of tool, but they are built for different teams. Both keep data as Delta Lake tables in a lake. Databricks is built for engineers and Spark. It runs on all three big clouds. Fabric is built around Power BI and runs only on Azure, with one pool of power for all its tools. Pick Databricks for heavy data work on any cloud. Pick Fabric when your firm lives in Microsoft tools.

Data

OLTP vs OLAP

People think OLTP and OLAP are two brands of database. They are two kinds of work. OLTP is many tiny reads and writes, such as placing an order. OLAP is a few huge queries that add up millions of rows, such as sales by month. A system built for one is slow at the other.

Data

Schema-on-Write vs Schema-on-Read

People think schema-on-read means data with no schema. Both have a schema. Schema-on-write checks each row against the shape when it is stored. Schema-on-read stores data as it comes and applies the shape in each query. The first catches bad data early. The second lets you load first and ask later.

Data

Logical Data Model vs Physical Data Model

People think the logical and physical data models are the same drawing. The logical one lists every field and key. It fits any database. The physical one gives each field an exact type for one real database. It also picks the indexes. If you change the database, only the physical model changes.

Data

Data Warehouse vs Data Lake

People think a data lake is just a cheaper data warehouse. A warehouse only takes clean tables with a set shape. A lake takes raw files of any kind and shapes them when they are read. The warehouse is easier to trust. The lake is cheaper and keeps everything.

Data

Data Lake vs Data Lakehouse

People think a lakehouse is a new kind of store you buy. It is a data lake with a table format on top. The files stay in the same cheap place. The table layer adds safe updates, fixed shapes and old versions, so the lake can serve reports too.

Data

Star Schema vs Snowflake Schema

People think a snowflake schema is better because it stores less. It splits dimensions into more tables, so each name sits in one place. A star schema keeps each dimension in one wide table. That means fewer joins and simpler queries, which is why most teams pick the star.

Data

Apache Parquet vs Apache Avro

People think Parquet and Avro do the same job. Both are open file formats with a schema. Parquet stores data by column, so reports that read a few columns are fast. Avro stores whole rows, so it suits sending one record at a time, as in event streams.

Data

Data Partitioning vs Clustering Key

People think partitioning and clustering are the same trick. One cuts a table into separate parts, often one per day. The other keeps the table whole but sorts the rows in it. Split on a column with few values, like a date. Cluster on one with many, like a customer ID.

Data

Snowflake vs Snowflake Schema

People think the snowflake schema is named after the Snowflake product. It is not. The schema is a way to lay out tables, and the name comes from its branching shape. Snowflake is a cloud data warehouse. You can build a star or a snowflake schema in it.

Data

Data Partitioning vs Database Sharding

People think partitioning and sharding are the same thing. Both split data into parts. One splits a table into parts inside one system, often by date, so queries can skip parts. The other spreads rows over many database servers, so each server takes part of the load.

DataDevelopmentcross-domain

Change Data Capture vs Incremental Load

People think asking a table for rows with a newer time is CDC. That is an incremental load. It asks the table on a timer and only sees rows as they are now. CDC reads the database's change log, so it catches every change, deletes too, within seconds.

Data

At-Least-Once vs Exactly-Once

People think exactly-once means each message travels only once. It does not. Under at-least-once, a message that is sent again is counted again. Under exactly-once, it may still be sent twice, but the copy is spotted by its ID and has no effect.

Data

Event Streaming vs Message Queue

People think event streaming is just a message queue. A queue hands each message to one reader and then removes it. A stream keeps every event for a set time. Many readers can each read all of it, at their own pace, and read it again later.

DataDevelopmentcross-domain

Event Time vs Processing Time

People think the time on an event and the time it was handled are the same. Event time is when it really happened, stamped in the event. Processing time is when the system got to it. A late event has an old event time and a new processing time, so the two can put it in different windows.

Data

Tumbling Window vs Sliding Window

People think any window that comes round again is a tumbling window. Tumbling windows touch but never overlap, so each event is counted once. Sliding windows overlap, so one event falls in more than one. Use tumbling for totals and bills. Use sliding for moving averages and alerts.

Data

Apache Hadoop vs MapReduce

People think Hadoop and MapReduce are the same thing. MapReduce is a way to split a job into map and reduce steps over many computers. Hadoop is a whole system. It has a file store called HDFS, a part that shares out the computers, and a MapReduce engine. Spark can run on Hadoop with no MapReduce at all.

Data

Airflow vs dbt

People think dbt and Airflow do the same job. dbt turns raw tables into clean ones with SQL inside the warehouse. Airflow decides when jobs run and in what order. Airflow often starts dbt as one of its tasks. One does the work, and the other keeps the clock.

Data

Apache Airflow vs Dagster

People think Dagster is just another Airflow. Airflow plans the steps you run. Dagster plans the tables you want to exist, and works out the steps from them. So Dagster can tell you which table is stale. Airflow can tell you which task failed.

Data

Airflow Scheduler vs Airflow Executor

People think the Airflow scheduler runs the tasks. It only decides which tasks are ready and puts them in a queue. The executor takes them from the queue and runs them, on one machine or many. A slow queue and a slow task have different fixes.

Data

Data Quality Check vs Data Observability

People think data observability is just more quality checks. It is not. A check tests a rule you wrote, like no blank IDs. It runs each time data passes. Observability learns what a normal day looks like. It flags a day that looks odd, even when no rule covers it. Checks catch what you saw coming, and it catches the rest.

Data

Data Catalog vs Data Lineage

People think a data catalog and data lineage are the same tool. A catalog is a search box with a page for each table. It says what the table is and who owns it. Lineage is the map of how data flows from table to table. Most catalogs show lineage on each page. But one says what a table is, and the other says where it came from.

Data

Pseudonymization vs Data Anonymization

People think pseudonymized data is anonymous. It is not. Names are swapped for a code, and a key that is kept can turn the code back into a person. Anonymized data has no key and no way back. In law, the first is still personal data. The second is not.

Data

Business Glossary vs Data Dictionary

People think a business glossary and a data dictionary are the same list. A glossary explains business words, like active customer, in plain words. A data dictionary lists the columns in a table. It gives each one a type and a unit. One says what a word means. The other says where it is kept.

Data

Data Mesh vs Data Lake

People think data mesh is a kind of storage for raw data, like a data lake. A data lake is one place to keep raw files. Data mesh is a way to split data work between teams, each owning its own data. A company can run a mesh on top of a lake, so the two are not rivals.

Data

Data Owner vs Data Steward

People think the data steward owns the data. The owner is a business leader who answers for it and decides who may use it. The steward does the daily work for the owner: writing down what fields mean and chasing bad rows. One decides, and the other keeps things clean.

Data

Query Engine vs Data Warehouse

People think a query engine is a data warehouse. A warehouse keeps the data and runs the queries, all in one product. A query engine only runs the queries. The data stays in files or other databases, and the engine reads it where it is. Both speak SQL, so from the outside they can look the same.

Data

Trino vs Presto

People think Trino and Presto are the same tool. They were, until 2019. That year the people who first built Presto left and started a fork, now called Trino. Presto stayed with the Presto Foundation. The two share a design, but their features have drifted apart since.

Data

Semantic Layer vs Business Glossary

People think a semantic layer is just a business glossary. A glossary explains a word like revenue in plain words. People read it. A semantic layer sets out revenue in code that tools run. It turns the rule into SQL and gives back the number. One says what revenue means. The other works it out.

Data

OLAP vs OLAP Cube

People think OLAP means a cube. OLAP is the kind of work: big reads that add up many rows to answer questions. A cube is one old way to do that work, by adding up every total in advance. Most OLAP today runs in column-based warehouses with no cube at all.

Data

DuckDB vs SQLite

People think DuckDB is just SQLite. Both run inside your program, with no server. SQLite is built for an app's own data, with many small reads and writes. DuckDB is built for sums and counts over millions of rows. Many teams use both, each for its own job.

DataDevelopmentcross-domain

Data Fabric vs Microsoft Fabric

People think Microsoft Fabric is the same thing as a data fabric. A data fabric is a general idea: link data where it lies, through shared notes and rules. Microsoft Fabric is one product you can buy. It keeps data in one shared lake instead. The names are alike, but one is a design and the other is a product.

Data

Data Fabric vs Data Mesh

People think data fabric and data mesh are rival products. Both are ideas, not tools. A data fabric links scattered data through shared metadata and rules, often run by one central team. A data mesh splits the work so each business team owns and shares its own data. One is about the wiring, the other about who owns what.

Data

Inbound Marketing vs Outbound Marketing

Inbound waits for people who are already looking for help. Outbound goes out to people who were not looking at all.

Marketing

Organic Marketing vs Paid Marketing

Organic reach costs time and keeps working for years. Paid reach costs money and stops the day you stop paying.

Marketing

Value Proposition vs Unique Selling Proposition (USP)

The value proposition is the whole promise. The USP is the one part of it no rival can claim.

BusinessMarketingcross-domain

Buyer Persona vs Ideal Customer Profile (ICP)

A persona is one made-up person you write your words for. The profile is the kind of company your sales team chases.

Marketing

Crawling vs Indexing

Crawling is the engine reading your page. Indexing is the engine deciding to keep it. A page can be read and not kept.

Marketing

robots.txt vs Noindex Tag

robots.txt asks crawlers not to read a page. Noindex lets them read it and asks them not to list it in the results.

Marketing

Title Tag vs Meta Description

The title is the blue link people click, and it helps rank. The description is the grey text under it, and it does not.

Marketing

On-Page SEO vs Off-Page SEO

On-page SEO is what you change on your own page. Off-page SEO is what other sites say about you, mostly with links.

Marketing

Organic Search Result vs Paid Search Result

An organic result is earned and free per click. A paid result is bought, marked as an ad, and gone when the money stops.

Marketing

White Hat SEO vs Black Hat SEO

White hat SEO helps readers and follows the rules. Black hat SEO tricks the engine, and wins fast until it is caught.

Marketing

Canonical URL vs 301 Redirect

A canonical tag hints which copy is the real one, and the copies stay open. A 301 sends everyone to the new address.

Marketing

Answer Engine Optimization (AEO) vs Generative Engine Optimization (GEO)

AEO shapes a page so a box or voice reads one answer from it. GEO aims to be one of the sources an AI cites in its reply.

Marketing

Featured Snippet vs Rich Result

A featured snippet quotes an answer above the links. A rich result is a normal link that shows extras like stars or a price.

Marketing

Noindex Tag vs Nofollow Link

Noindex keeps a whole page out of the results. Nofollow labels one link so it does not count as a vote for its target.

Marketing

Google Search Console vs Google Analytics

Search Console shows what happens on Google, up to the click. Analytics shows what people do on your site after it.

Marketing

Ahrefs vs Semrush

Both are paid SEO suites. Ahrefs is known first for its link data. Semrush adds deep tools for search ads beside SEO.

Marketing

Google Keyword Planner vs Google Trends

Keyword Planner gives rough monthly search counts. Trends gives no counts, only how interest rises and falls from 0 to 100.

Marketing

Cost Per Click (CPC) vs Cost Per Mille (CPM)

CPC is what you pay when someone clicks. CPM is what you pay for a thousand showings, whether anyone clicks or not.

Marketing

Cost Per Acquisition (CPA) vs Return on Ad Spend (ROAS)

CPA is what one sale cost you in ads, so lower is better. ROAS is what each unit spent brought back, so higher is better.

Marketing

Search Engine Marketing (SEM) vs Search Engine Optimization (SEO)

SEM buys a place in search results with ads, today. SEO earns a free place over months, by making the page the best answer.

Marketing

Pay-Per-Click (PPC) vs Search Engine Marketing (SEM)

PPC is a way to pay for ads, on any platform. SEM is a channel, search ads, which happen to be mostly paid per click.

Marketing

Affiliate Marketing vs Referral Marketing

Affiliates are partners from outside, paid a fee for each sale. Referrals come from your own happy buyers, often for a small prize.

Marketing

Retargeting vs Lookalike Audience

One shows ads again to people who came to your site before. The other finds new people who are a lot like your best buyers.

Marketing

Quality Score vs Ad Rank

Quality Score is a 1 to 10 guide to how well your ad fits. Ad rank is the auction score, bid and quality, that sets position.

Marketing

Display Advertising vs Native Advertising

A display ad sits beside the content and looks like an ad. A native ad is made to look like the content, with a label.

Marketing

Growth Loop vs Marketing Funnel

A funnel ends at the sale, so each new buyer must be found again. A loop uses each new user to bring in the next one.

Marketing

Click-Through Rate (CTR) vs Conversion Rate

CTR counts how many who saw a link clicked it. Conversion rate counts how many who arrived went on to do the thing you wanted.

Marketing

Reach vs Impressions

Reach counts the different people who saw a post. Impressions count every time it was shown, so one person can add many.

Marketing

Email Newsletter vs Drip Campaign

A newsletter goes to the whole list on a set day. A drip campaign starts when one person acts, and runs on their own clock.

Marketing

Transactional Email vs Email Newsletter

People think a receipt is just another marketing email, and it is not. A transactional email is sent because one person did something, like buy or reset a password. It goes to that person at once and needs no consent. A newsletter goes to the whole list on a set day. It needs a yes from each reader and an unsubscribe link.

Marketing

SPF vs DKIM

People think SPF and DKIM do the same check, and they do not. SPF asks if the server that sent the email is on the domain's list of allowed senders. DKIM checks a hidden signature on the email itself. It proves the email came from the domain and was not changed. Forwarding often breaks SPF but leaves DKIM whole.

Infrastructure

Organic Social vs Paid Social

People think organic and paid social are two names for the same posts, and they are not. Organic social is what a brand posts for free, and the app decides who sees it. Paid social is a post or ad the brand pays to show to people it picks. Organic reach builds slowly and lasts. Paid reach is fast and stops when the money stops.

Marketing

User-Generated Content (UGC) vs Influencer Marketing

People think user-generated content and influencer marketing are the same deal, and they are not. An influencer posts about you to their own fans, and those fans are what you pay for. User-generated content is a post a buyer makes about you. The brand then shares it, with permission, on its own pages and ads. It needs no big crowd behind it.

Marketing

Contextual Targeting vs Retargeting

People think every ad that seems to know them works the same way, and it does not. Contextual targeting picks the ad from the page. A camping article gets a camping stove ad, for anyone who reads it. Retargeting picks the ad from the person. Someone who looked at a sofa on your site sees that sofa again on other sites. One needs facts about the page. The other needs a record of the visit.

Marketing

Social Proof vs Trust Badge

People think a trust badge is just more social proof, and it is not. Social proof shows that other buyers liked the product, so it answers the question is it any good. Reviews and star ratings do this. A trust badge shows that paying is safe, so it answers the question is this shop safe. Card logos and a money-back promise do this. Put reviews near the product and badges by the pay button.

Marketing

Website Heatmap vs Session Recording

People think a heatmap and a session recording show the same thing, and they do not. A heatmap sums up thousands of visits in one picture of a page. It shows where the crowd clicked and how far it scrolled. A session recording replays one visit, click by click, like a video. It shows where one person got stuck. Use the heatmap to find the busy spots and recordings to watch what went wrong there.

Design

Net Promoter Score vs Customer Satisfaction Score

People think NPS and CSAT measure the same thing, and they do not. CSAT asks how happy someone was with one moment, like a delivery or a chat, on a scale from 1 to 5. NPS asks how likely they are to recommend you, from 0 to 10, and speaks for the whole bond with the brand. A customer can rate one chat 5 and still be a detractor. Use CSAT to fix moments and NPS to track loyalty.

Marketing

Freemium vs Free Trial

People think freemium and a free trial are the same offer, and they are not. Freemium gives a basic plan that is free forever, and people pay when they want more. A free trial gives every feature, but only for a set time, often 14 or 30 days. Freemium limits what you get. A free trial limits how long you get it.

Marketing

Product-Led Growth vs Sales-Led Growth

People think product-led and sales-led growth are rival camps, and they are not. They are two ways a product gets bought. Product-led lets people sign up, use it and pay on their own. Sales-led puts a salesperson in front of each deal, with a demo and a price agreed first. Cheap, easy products lean product-led. Large deals lean sales-led, and many firms run both.

Marketing

Demand Generation vs Lead Generation

People think demand generation is lead generation with a new name, and it is not. Lead generation swaps something useful for a name and an email. Demand gen is wider. It also builds the want that makes people sign up at all, often with nothing asked back. Lead gen counts names. Demand gen is judged by the real deals it brings.

Marketing

ICE Scoring vs RICE Scoring

People think ICE and RICE scoring are the same list, and they are not. Both rank ideas before you spend time on them. ICE scores impact, confidence and ease, each from 1 to 10. RICE adds reach, the number of people an idea touches in a set time, and divides by the effort it takes. Growth teams often use ICE because tests are cheap. Product teams often use RICE to rank bigger features.

MarketingBusinesscross-domain

Influencer Marketing vs Affiliate Marketing

An influencer is paid to show a product to their fans. An affiliate is paid only when a sale comes through their own link.

Marketing

Content Marketing vs Inbound Marketing

Content marketing makes useful things to read or watch. Inbound is the whole plan to draw buyers in. Content is one part.

Marketing

First-Touch vs Last-Touch Attribution

Both give one touch all the credit for a sale. First-touch thanks what found the buyer, and last-touch thanks what closed the sale.

Marketing

Marketing Attribution vs Incrementality

Attribution shares out the credit for sales that came in. A lift test asks which of those sales the ads truly caused, and which would have come anyway.

Marketing

A/B Test vs Multivariate Test

An A/B test checks one change at a time. A multivariate test checks every mix of several changes, so it needs far more visitors.

Marketing

Marketing ROI vs ROAS

ROAS divides sales by ad spend. ROI takes the costs out first, so a campaign with a fine ROAS can still lose money.

Marketing

Multi-Touch Attribution vs Marketing Mix Modeling

One follows each buyer's clicks and shares the credit. The other skips people and reads weekly totals, so it can see TV and posters too.

Marketing

Target Market vs Target Audience

The target market is all the buyers a business picks to serve. A target audience is the smaller group that one ad speaks to, from inside that market.

Marketing

Penetration Pricing vs Price Skimming

Both are plans for a launch price. Penetration starts low to win many buyers fast. Skimming starts high and drops step by step to reach each next group.

Marketing

Brand Identity vs Brand Image

Identity is what the company chooses to send out: the name, the look and the voice. Image is what people end up thinking. The gap between them is the work.

Marketing

Customer Retention vs Customer Loyalty

Retention counts who is still buying. Loyalty asks whether they would choose you again if a rival were just as easy. A kept customer can still leave tomorrow.

Marketing

Go-to-Market Strategy vs Marketing Plan

A go-to-market plan takes one product into one market, often at launch. It also says how the product will be sold. A marketing plan covers all of a firm's marketing for a year.

Marketing

Positioning vs Differentiation

Positioning is the one idea you want buyers to link to you. Being different is about the product itself, and how it truly is not like its rivals. A good position rests on a real gap.

Marketing

Positioning Statement vs Unique Selling Proposition (USP)

The first is a note kept inside the firm. It tells the team who the product is for and why it wins. The USP is the one claim that buyers hear.

Marketing

B2B vs B2C Marketing

B2C sells to the public: one person decides, often fast, and each sale is small. B2B sells to firms: a group decides over weeks, and each deal is big.

Marketing

Primary vs Secondary Research

Primary research is data you gather yourself. It fits your question, but it takes time. The other kind uses data that is already out there. It is fast, but it may not fit.

Marketing

Survey vs Focus Group

A survey asks many people fixed questions and counts the answers. A focus group talks with a few people to learn why they feel the way they do. One gives numbers, the other gives reasons.

Marketing

Revenue vs Profit

People quote revenue as if it were what a business earns, and it is not. Revenue is all the money from sales, before any cost comes out. Profit is what is left once every cost is paid. A firm can have huge revenue and still make a loss.

Business

Profit vs Cash Flow

People think a profitable business must have cash, and it may not. Profit counts a sale on the day it is made. Cash flow counts it on the day the money arrives. Between the two, wages and bills still need paying. That gap is why firms with a profit can still run out of cash.

Business

Gross Margin vs Net Profit Margin

People see a 60% gross margin and think the firm keeps 60 cents of each dollar, and it does not. Gross margin only takes away the cost of the goods. Net margin takes away every cost, from rent to tax. The net figure is the one owners actually keep, and it is often far smaller.

Business

Operating Expenses (OpEx) vs Capital Expenditure (CapEx)

People lump all spending together, and the books do not. OpEx is the cost of running the business this month, like rent and wages, and it comes off profit at once. CapEx buys something that lasts for years, like a machine. Its cost is spread over those years, even though the cash leaves on day one.

Business

Accounts Receivable vs Accounts Payable

People mix the two up because the names sound alike, and they point opposite ways. Receivable is money your customers owe you, so it is an asset. Payable is money you owe your suppliers, so it is a debt. Collect the first fast and pay the second on time, and your cash stays healthy.

Business

Net Present Value (NPV) vs Internal Rate of Return (IRR)

People think the project with the higher IRR is always the better one, and it is not. NPV says how many dollars a project adds, in today's money. IRR says what yearly return it earns, as a percent. A small project can win on IRR while a big one adds far more dollars. When the two disagree, finance teams go with NPV.

Business

Initial Public Offering (IPO) vs Direct Listing

People think every company that lists on a stock market raises money, and many do not. In an IPO, a bank sells new shares and the company gets the cash. In a direct listing, the old shares simply start trading, and usually no new money comes in. The direct route is cheaper, and owners can sell from day one.

Business

KPI vs OKR

People think KPIs and OKRs are two names for the same list of numbers, and they are not. A KPI is a number you watch all the time to check the business is healthy. An OKR is a goal for a few months, set to change something. A KPI that slips can become the target of the next OKR.

Business

Leading vs Lagging Indicator

People think the best number to watch is the result itself, and it is not enough. A lagging indicator, like sales, tells you what already happened. A leading indicator, like free trials, moves first and gives you time to act. Teams judge by the first and steer by the second.

Business

Churn Rate vs Net Revenue Retention (NRR)

People think low churn means the customer base is growing, and it may not. Churn rate counts the customers who left. Net revenue retention counts the money from the ones you had, upgrades and cuts included. A firm can lose 10% of its customers and still have NRR over 100%, if the rest upgrade.

Business

Monthly vs Annual Recurring Revenue

People think ARR is a forecast of next year's revenue, and it is not. MRR is what subscribers pay this month. ARR is that same figure times twelve, a snapshot stretched over a year. Firms selling small monthly plans tend to quote MRR, and firms selling yearly contracts quote ARR.

Business

CAC Payback Period vs LTV to CAC Ratio

People think a good LTV to CAC ratio means growth is safe, and it may not be. The ratio says whether a customer is worth more than they cost in the end. CAC payback says how many months that takes. A 5 to 1 ratio with a 30-month payback can still run a firm out of cash.

Business

Upselling vs Cross-Selling

People use the two words as if they meant the same thing, and they do not. Upselling moves the buyer to a better version of what they chose, like a bigger plan. Cross-selling adds a different product that goes with it, like a case for a laptop. One goes up, the other goes sideways.

Business

B2B vs B2C

People think B2B is just B2C with bigger orders, and it is not. In B2B a firm buys, a group decides, and a sale can take months with a seller guiding it. In B2C one person buys for themselves, often in minutes, and a shop or website does the selling. Who buys changes the whole way you sell.

Business

Sales Pipeline vs Sales Funnel

People draw both as the same shape and use the names in turn, and they measure different things. A funnel counts people and shows what share moves from each step to the next. A pipeline lists each open deal with its value and stage, so a team can see what might close and when. The funnel finds the leak; the pipeline plans the month.

Business

Inside Sales vs Field Sales

People think inside sales means junior work and field sales means the real deals, and that is not the split. Both run a full sale. Inside sellers work by phone and video and reach many buyers a day. Field sellers travel to meet fewer buyers, which suits large deals where trust is earned in person.

Business

Product Vision vs Product Strategy

People use the two words as if they meant the same thing, and they do not. A vision is the future a product wants to make, years away, with no features in it. A strategy is the plan to get there: which users to serve first and how to beat the other choices. The vision rarely changes. The strategy changes when the plan stops working.

Business

Outcome-Based Roadmap vs Feature-Based Roadmap

People think a roadmap has to list features, and it does not. A feature roadmap lists things to ship, often with dates, and is done when they ship. An outcome roadmap lists changes to see, like more users finishing sign-up, and is done when the number moves. The first is easy to read. The second leaves the team free to find the best fix.

Business

RICE Scoring vs Kano Model

People think RICE and the Kano model are two ways to get the same score, and they are not. RICE ranks ideas by numbers: how many people, how much help, how sure you are, and how much work. The Kano model sorts features by how customers feel about them: a must, more is better, or a nice surprise. Kano tells you what kind of feature it is. RICE tells you what to build first.

Business

Product Discovery vs Customer Discovery

People use the two names for the same thing, and they are not the same. Customer discovery asks if a group of buyers with a real problem exists at all. It is the first step for a new startup. Product discovery asks what to build next for users you already have, and it goes on every week. One finds the market. The other finds the next feature.

Business

Depreciation vs Amortization

People think depreciation and amortization are two names for one thing, and they are not. Both spread a cost over the years it is used. Depreciation is for things you can touch, like a truck or an oven. Amortization is for things you cannot touch, like a patent or a license. The math is often the same. Only the kind of asset changes.

Business

Accrual vs Cash Basis Accounting

People think accrual and cash basis give the same profit in the end, and month by month they do not. Accrual counts money when it is earned or owed. Cash basis counts it only when it reaches or leaves the bank. A job done in March and paid in May is March income under accrual and May income under cash. Small firms often use cash. Larger ones must use accrual.

Business

Budget vs Forecast

People think a budget and a forecast are the same plan, and they are not. The budget is set before the year starts and then stays put. It is the target and the limit. The forecast is a best guess of where the year will really land, and it is updated every month or quarter. The gap between the two tells you what to fix.

Business

Retained Earnings vs Net Income

People think retained earnings is this year's profit, and it is not. Profit is what a firm earned in one year, before any of it was paid out. Retained earnings is all the profit kept since the company began, after dividends were paid. Each year's profit, minus that year's dividends, is added to the pile. A firm can have a loss this year and still have large retained earnings.

Business

Dividend vs Share Buyback

People think dividends and buybacks are rival tricks, and they are two ways to do one job. Both hand cash back to owners. A dividend pays every owner the same amount per share. A buyback pays only the owners who choose to sell, and the rest end up owning a bigger slice. Neither one makes the company worth more on its own.

Business

Book Value vs Market Value

People think book value and market value should match, and they rarely do. Book value is what the accounts say the company is worth, built from what it paid for things. Market value is what buyers pay for all its shares today, built from what they expect it to earn. A strong brand can make market value many times book value. A troubled firm can trade below it.

Business

Lead vs Sales Qualified Lead

People think every lead is ready for a sales call. It is not. A lead is anyone who shows interest, such as a person who fills in a form. A sales qualified lead is a lead the sales team has checked. It has a real need, a budget and a person who can say yes. Marketing finds leads. Sales picks the ones worth a call.

MarketingBusinesscross-domain

Product Manager vs Project Manager

People think a product manager and a project manager do the same job. They do not. A product manager decides what to build and why, and stays with the product for years. A project manager plans how and when one piece of work gets done, and the job ends when the project ends. One owns the what. The other owns the how and the when.

Business

Feature Creep vs Scope Creep

People think feature creep and scope creep are the same leak. They start in different places. Scope creep is new work slipping into a project after the plan is set, so the dates and the budget slip too. Feature creep is a product growing more and more features over time, until it is hard to use. One hurts the plan. The other hurts the product.

Business

Vanity Metric vs KPI

People think any number that goes up is a good sign. A vanity metric goes up and still tells you nothing about what to do next, like total sign-ups or page views. A key performance indicator is tied to a goal, so a change in it tells you whether you are winning. Ask what you would do if the number fell. If the answer is nothing, it is a vanity metric.

Business

PMP vs CAPM

People think the CAPM is a small PMP that grows into one. They are two separate badges. The CAPM needs no project experience, only 23 hours of study. The PMP needs 36 to 60 months of leading projects. An active CAPM only covers the 35 training hours the PMP asks for.

Business

PMP vs PRINCE2 Practitioner

People think both badges prove the same thing. The PMP is about the person: it needs years of leading projects and tests how you act. PRINCE2 Practitioner is about one method: it tests that you can use PRINCE2 and fit it to a project. Many managers hold both, and a PMP lets you skip PRINCE2 Foundation.

Business

CSM vs PSM

People think the CSM and the PSM are one badge sold by two shops. The CSM needs a 16-hour class and must be renewed every two years. The PSM needs no class, only a harder test, and it never expires. Both prove you know Scrum, not that you have run a team.

Business

PMBOK Guide vs PMP

People think the PMP exam tests the PMBOK Guide. The guide is PMI's book of good practice. The exam follows its own content outline and asks what you would do at work. Reading the guide alone will not get you through it.

Business

PDU vs PMP Contact Hours

People mix up the hours before the PMP and the hours after it. Contact hours are the 35 hours of training you need before you apply. PDUs are the 60 hours of learning that keep the badge alive every three years. Training done before you pass does not count as PDUs.

Business

Visual Identity vs Brand Identity

People think the two names mean the same thing. They do not. The visual one is only the look: the logo, the colors, the fonts and the pictures. The brand one is bigger. It adds the name, the voice and the way the company talks. So the look is one part of the whole.

DesignMarketingcross-domain

Customer Data Platform vs CRM

People think a CDP is just a bigger CRM, and it is not. A CRM is where sales staff type in contacts and deals, one by one. A CDP collects data on its own from every tool: the site, the app and the shop. It joins the data into one profile per person, and sends groups out to email and ads. Many firms use both, with the CRM as one of the CDP's sources.

MarketingBusinesscross-domain

CDP vs DMP

People think a CDP and a DMP do the same job, and they do not. A CDP holds named customers, with emails and orders, and keeps them for years. A DMP holds cookie IDs with no names, sorted into ad groups, and they fade in weeks. A CDP serves every channel, while a DMP mostly serves ad buying. As third-party cookies fade, DMPs are fading too.

Marketing

First-Party vs Third-Party Data

People think all customer data is alike, and it is not. First-party data is what you see on your own site, app or shop. Third-party data is bought from a broker who gathered it somewhere else. Your own data is more accurate, and you know how it was collected. Bought data reaches more people, but it is often stale, and its consent is hard to prove.

Marketing

Zero-Party vs First-Party Data

People think these two are the same, and they are close but not the same. Zero-party data is what a customer tells you on purpose, like a quiz answer or a chosen interest. First-party data is what you see them do, like the pages they visit and the things they buy. One is said, and the other is watched. Said data shows what people want, but they may change their minds.

Marketing

Soft Opt-In vs Opt-In

People think the soft opt-in is just a softer opt-in, and it is not. An opt-in waits for a clear yes, like a ticked box, before the first email. The soft opt-in needs no tick. It lets a firm email its own past buyers about like goods, if they could say no at sign-up and in every email. It never covers new people or bought lists.

Marketing

Opt-In vs Double Opt-In

People think one tick is the same as a confirmed sign-up, and it is not. With a single opt-in, the tick on the form puts the person on the list at once. With a double opt-in, the person must also click a link in an email first. The second step proves the address is real and theirs. The list grows a bit slower, but it holds fewer typos and fakes.

Marketing

GDPR vs CCPA

People think the EU law and the California law work the same way, and they do not. The GDPR asks for a lawful reason before any use of personal data. The CCPA lets a firm use data at once, but people can tell it to stop selling or sharing. The GDPR covers any firm that handles data on people in the EU. The CCPA covers only firms that pass a size test.

Data

Puffery vs False Advertising

People think any boast in an ad is a lie, and it is not. Puffery is a boast so broad that no one takes it as fact, like best pizza in town. False advertising is a claim that buyers could check, and that turns out wrong. The test is whether a buyer could measure it. Once a boast has a number or a test in it, it needs proof.

Marketing

Client-Side vs Server-Side Tracking

People think tracking happens in one place, and it can happen in two. Client-side tracking sends data straight from the visitor's browser to each tool. Server-side tracking sends it once to the firm's own server, which passes it on. The server way loses less data to blockers and lets the firm trim what it shares. It still needs the visitor's consent, just like the browser way.

Marketing

Data Controller vs Data Processor

People think whoever holds the data is in charge of it, and that is not so. The controller decides why and how the data is used, and answers for it under the law. The processor only works on the data as the controller says, under a contract. An email tool that sends your newsletter is a processor. If it used your list for its own ads, it would turn into a controller too.

Data

Data Protection Officer vs Data Controller

People think the DPO answers for a firm's data, and that is not so. The controller is the firm that decides why the data is used. It carries the legal duty. The DPO is a person inside the firm who checks and advises, and takes no orders on what to find. When the rules are broken, the watchdog fines the controller, not the DPO.

Data

Records of Processing Activities vs Privacy Policy

People think the privacy policy is the record the law asks for, and it is not. Both list what data a firm uses and why. The privacy policy is a public page, written for the people whose data it is. The records are an inside list, with a row for each use, kept for the watchdog. A firm needs both, and the two must match.

Data

DPIA vs AI Impact Assessment

People think a DPIA and an AI impact assessment are the same check, and they are not. A DPIA looks only at the risks of using personal data. The GDPR asks for it. An AI impact assessment looks at every way an AI system could harm people, such as bias or safety. It counts even when no personal data is used. One AI project can need both.

DataAIcross-domain

EU Data Act vs GDPR

People think the Data Act is a new privacy law, and it is not. The GDPR guards data about people, and it applies to any firm that uses it. The Data Act is about who may use the data that devices make, such as a car or a smart TV. It gives the user a right to that data, personal or not. When the data is about a person, both laws apply at once.

Data

Standard Contractual Clauses vs Adequacy Decision

People think signing the clauses makes a country safe, and it does not. An adequacy decision is the EU's ruling on a whole country, or part of one. Data then goes there with no extra work. Standard contractual clauses are a contract between two firms. They are used where there is no such ruling, and they come with a transfer check.

Data

Right to Erasure vs Data Retention Policy

People think personal data is only deleted when the retention schedule says so, but a person's request can come first. A retention policy is the firm's plan. It says how long to keep each kind of data. The right to erasure is a person's request to delete their data now. Data the law says to keep, like invoices, stays either way.

Data

Joint Controller vs Data Processor

People think any second firm on a data project is a processor, and that is not so. A processor only follows the controller's orders. A joint controller helps decide why and how the data is used. Joint controllers need a written deal on who does what. Each one can be held to account. Calling a partner a processor does not change what it does.

Data

Transfer Impact Assessment vs DPIA

People think a transfer check is the same as a DPIA, and it is not. A DPIA asks what a new use of data could do to the people in it. A transfer impact assessment asks one narrower thing. Will the laws of another country undo the safeguards on data sent there? A project that sends data abroad may need both.

Data

Right to Data Portability vs DSAR

People think asking to see your data and asking to move it are one request, but each right gives you something else. An access request shows you all a firm holds on you, and why. A request to move it gives only the data you handed over. It comes as a file. You can load it into another service.

Data

Works Agreement vs Collective Agreement

People think a works agreement and a collective agreement are the same deal, and they are not. Unions and bosses sign a collective agreement. It often covers a whole trade. One boss and its works council sign a works agreement. It covers just that firm. The collective agreement ranks higher. A works agreement can give staff more, but not less.

Business

Hashed Email vs Anonymized Data

People think a hashed email is anonymous, and it is not. Hashing turns an email into a fixed code, but the same email always gives the same code. Anyone who has the email can hash it and find the match. So the law still treats it as personal data. Anonymous data is data that can no longer be linked back to a person at all.

MarketingDatacross-domain

Consent vs Legitimate Interest

People think you always need a yes to use someone's data. Under the GDPR, that is not true. Consent is one lawful basis: a clear yes that the person can take back. Legitimate interest is another: the firm's need, set against the person's rights. It needs no yes, but a person can always say no to sales mail. Cookies and most sales email still need a yes under other rules.

MarketingDatacross-domain

DSP vs SSP

People think a DSP and an SSP are the same ad tool, and they are not. A DSP works for the buyer. It looks at each ad slot on offer and bids for the brand. An SSP works for the site. It offers the slot to many buyers and tries to get the best price. They meet in the middle, at the ad exchange.

Marketing

Ad Exchange vs Ad Network

People think an ad exchange is just a big ad network, and it is not. An ad network gathers space from many sites and sells it to brands as one package. It sets the price and keeps a share. An ad exchange is an open market. It runs an auction for each single ad view, and buyers and sellers trade there through their own tools.

Marketing

First-Price vs Second-Price Auction

People think the winner of an ad auction always pays what it bid, and that is not so. In a first-price auction, it does. In a second-price auction, the winner pays one cent more than the next best bid. If the bids are 4 euros and 3 euros, the winner pays 3.01. Most programmatic auctions now use first price.

Marketing

Reach vs Frequency

People think a big number of ad views means a big number of people, and it does not. Reach counts how many different people saw the ad at least once. Frequency counts how many times each of them saw it, on average. One million views could be one million people once each, or 100,000 people ten times each.

Marketing

CTV vs OTT

People think CTV and OTT mean the same thing, and they do not. OTT is the way a show arrives: over the internet, not by cable. It can play on a phone, a laptop or a TV. CTV is one screen: a TV set linked to the internet. So every CTV ad is an OTT ad, but an OTT ad on a phone is not CTV.

Marketing

Paid Media vs Earned Media

People think a sponsored article in a news site is earned media, and it is not. If you paid for the space, it is paid media, even when it looks like news. Earned media is coverage you did not pay for, like a review or a news story a reporter chose to write. Paid media must be marked as an ad. Earned media cannot be bought.

Marketing

Invalid Traffic vs Ad Fraud

People think invalid traffic and ad fraud are the same, and they are not. Invalid traffic is any ad view that did not come from a real person. Some of it is honest, like a search engine robot reading the page. Ad fraud is the part made on purpose to steal ad money. All ad fraud is invalid traffic, but not all invalid traffic is fraud.

Marketing

Media Agency vs Creative Agency

People think the agency that makes an ad also decides where it runs, and that is often not so. A creative agency has the idea and makes the ad. A media agency plans where it runs and buys the space. Many big brands hire one of each. A full-service agency is the one that does both under one roof.

Marketing

Working Media vs Non-Working Media

People think non-working media is money wasted, and it is not. Working media is the money that buys the ad space people see. Non-working media pays for everything else: making the ads, agency fees, research and tools. Without it, there would be no ad to show. The aim is a sensible balance, not to cut it to zero.

Marketing

ads.txt vs sellers.json

People think ads.txt and sellers.json are the same file, and they are not. A site puts ads.txt on its own web address. It lists who may sell the site's ad space. An ad seller puts sellers.json on its own address. It lists every site it sells for. A buyer checks that both lists agree.

Marketing

Programmatic Advertising vs Real-Time Bidding

People think programmatic advertising always means an auction, and it does not. Programmatic just means software buys the ad space. Real-time bidding is the auction kind: each ad view is sold to the best bid as the page loads. But a programmatic guaranteed deal has no auction at all. The price and the number of views are agreed before.

Marketing

Ad Impression vs Viewability

People think every ad impression is an ad someone could see, and it is not. An impression counts each time an ad loads. It may load at the bottom of a page nobody scrolls to. Viewability asks whether the ad was really on screen. For a picture ad, half of it must be in view for at least one second.

Marketing

Effective Frequency vs Frequency Cap

People think the frequency cap is the number of views a plan aims for, and it is not. Effective frequency is the floor: the fewest views a person needs before the ad works. The cap is the ceiling: the most views anyone should get before the ad stops. A good plan puts most people in the band between the two.

Marketing

Marketing ROI vs MER

People think marketing ROI and MER are two names for one return, and they are not. Marketing ROI takes the profit marketing made, minus its cost, and divides by the cost. MER divides all sales by all marketing spend, and never asks which ad made which sale. ROI says whether marketing made money. MER is a quick check on what the ad tools claim.

Marketing

Deterministic vs Probabilistic Matching

People think every joined customer record is a sure match, and it is not. Deterministic matching joins two records only when they share an exact ID, like the same email. Probabilistic matching guesses from clues, like the same home Wi-Fi at the same hour. The first is sure but covers fewer people. The second covers more, and some of its joins are wrong.

Marketing

Propensity Model vs Uplift Model

People think the customers most likely to buy are the ones to send an offer. They are not. A propensity model scores how likely someone is to act. An uplift model scores how much an offer changes that. Many top propensity scores belong to people who would buy anyway. Uplift finds the ones the offer really moves.

Marketing

Private Marketplace vs Programmatic Guaranteed

Both are private deals run on a deal ID, so people think they promise the same thing. A private marketplace is still an auction: a few invited buyers bid, and nobody is sure of winning. Programmatic guaranteed is no auction at all. One buyer gets a set number of views at a fixed price, and must take them.

Marketing

Creative Brief vs Design Brief

People think a creative brief is the same as a design brief, and it is not. A design brief sets up one design job: what to make, for whom, by when and for how much. A creative brief sets up a campaign. Its heart is the one thing the ads must say, and why people should believe it. A campaign may then need several design briefs.

MarketingDesigncross-domain

Video View vs Watch Time

People think views show how well a video did, and they only show half. A view counts each time someone starts watching. Watch time adds up how long they all stayed. Two clips with the same views can have very different watch time, and sites rank on watch time.

Marketing

Audience Retention vs Customer Retention

People think audience retention means keeping buyers, and it does not. It is about one video. It shows how many people still watch at each second. Customer retention is about a firm. It shows how many buyers come back over months or years.

Marketing

End Screen vs Video Card

People mix up end screens and cards, since both point to another video. A card is a small note that can pop up at any moment in the video. An end screen only shows in the last seconds, with bigger boxes to click. Use a card when you mention something, and an end screen to say what to watch next.

Marketing

Content ID Claim vs Copyright Strike

People think a Content ID claim is a strike against the channel, and it usually is not. A claim is made by a machine match. The owner can block the video, take its ad money or just track it, and the channel is not harmed. A strike comes from a legal request to take a video down. It lasts 90 days, and three can end the channel.

Marketing

Vertical Video vs Short-Form Video

People use vertical video and short-form video as if they were the same, and they are not. Vertical is about shape: taller than it is wide, to fill a phone. Short-form is about length: seconds to a few minutes. Most short clips are vertical, but a long live stream can be vertical too, and a short ad can be wide.

Marketing

Cost Per View vs Cost Per Mille

People think a paid view and an impression are the same thing, and they are not. With CPM you pay for every thousand times the ad is shown, watched or not. With CPV you pay only when someone watches enough of the video ad, or clicks it. So CPV costs more per unit, but each unit means more.

Marketing

Cause Marketing vs Social Marketing

People think cause marketing and social marketing are the same, since both do good. They are not. Social marketing sells nothing at all. It uses the tools of ads to change a habit, like smoking. In cause marketing, a firm sells a thing and gives part of each sale to a cause.

Marketing

Standardization vs Adaptation Strategy

One plan keeps the product and ads the same in each country. That is cheaper and faster. The other changes them to fit each country. That fits better but costs more. Most firms pick a point between the two.

Marketing

Localization vs Transcreation

People think these two are the same job, and they are not. The first keeps the message. It changes what is around it, like the words, the money, dates and pictures. The second builds the message again. Then a slogan or a joke gives the same feeling in a new tongue.

Marketing

High-Context vs Low-Context Culture

In a high-context culture much is left unsaid, and people read the meaning from the setting and the relationship. In a low-context culture people say what they mean in plain words. Ads lean on mood in the first and on facts in the second.

Marketing

International Marketing vs Global Marketing

People use these two words as if they meant the same, and they do not. The first is any marketing in more than one country. It changes as much as each place needs. Global marketing is one way to do it. It keeps one brand and one message the same in every land.

Marketing

Direct-to-Consumer vs DTC Advertising

People see 'DTC' and think it means one thing, and it means two. A D2C or DTC brand sells its own goods straight to buyers, with no shop in between. DTC advertising, or DTCA, means ads for medicines that need a doctor's note, shown to the public. Most countries ban those ads.

Marketing

D2C vs B2C Marketing

People think D2C is just another name for B2C, and it is not. B2C means selling to people rather than to firms, and it often runs through shops. D2C is one kind of B2C where the brand cuts out the shops and sells to people itself. Every D2C brand is B2C, but most B2C brands are not D2C.

Marketing

Trade Marketing vs Shopper Marketing

People mix up trade and shopper marketing, since both happen in and near the shop. They are not the same. In trade marketing, the brand wins over the shop. Then the shop stocks the product in a good spot. In shopper marketing, the brand wins over the person in the aisle, just as they choose.

Marketing

Cross-Cultural vs Multicultural Marketing

People think these two are the same, and they are not. The first takes a message from one culture to the next, often from one land to another. The second speaks to the many groups of people who live in one land.

Marketing

SEO vs GEO

People think GEO replaces SEO, and it does not. SEO gets a page ranked in a list of links. GEO gets the same page picked and quoted inside an AI answer. Both need pages that can be crawled and facts that are clear. GEO adds one more goal on top: being the source the answer names.

Marketing

AI Search Engine vs Search Engine

People think an AI search engine is a search engine that writes better, and it is not the same job. A search engine hands you a list of pages to read. An AI search engine reads the pages for you and hands you one answer. You get speed, but you see fewer sources. You also have to trust how it joined them.

AIMarketingcross-domain

llms.txt vs robots.txt

People call llms.txt the robots.txt for AI, and it is not. robots.txt tells bots which pages they may not fetch. llms.txt is a short guide that points AI tools to the pages worth reading. It blocks nothing at all. To keep an AI bot out, you still need a line in robots.txt.

Marketing

AI Visibility vs AI Citation

People think being named in an AI answer and being cited in it are the same win, and they are not. AI visibility counts how often the answer names your brand. An AI citation is a link in the answer to one of your pages. A name carries the brand to the reader. A link can also carry the reader to your site, where you can see the visit.

Marketing

Modeled Conversions vs Conversion Tracking

People think every sale in an ad report was tracked, and many were not. Conversion tracking counts the sales a tag or a server actually saw. Modeled conversions are estimates for the sales it could not see, often from people who said no to cookies. Both land in the same column. So part of the total is a model's guess, not a record.

Marketing

Conversion Lift Study vs Geo Lift Test

People think a lift test is a lift test, and these two split different things. A conversion lift study splits people at random inside one ad platform. A geo lift test splits places, like cities, and holds the ads back in some of them. The first is run by the platform and tests only its ads. The second can test any channel, even TV, but it needs more money and time.

Marketing

Agentic Commerce vs Conversational Commerce

People think agentic commerce is a shop chatbot with a new name, and it is not. When you shop in a chat, the chat answers questions and the buyer still taps buy. In agentic commerce, an AI agent does the steps itself, from searching to paying. The person only sets the limits and approves. One helps a buyer decide, and the other acts for them.

Marketing

Leadership vs Management

People use the two words as if they mean the same, and they do not. Leadership decides where a team should go and gets people to want to go there. Management works out how to get there, with plans, jobs and checks. A boss can manage well and lead badly. Anyone can lead without being the boss.

Business

Coaching vs Mentoring

People think a coach and a mentor do the same job, and they do not. A coach asks questions, and you find your own answer. A mentor has walked the path before and shares what worked for them. A coach does not need to know your job. A mentor usually does.

Business

Data Analyst vs Data Scientist

People think a data scientist is just a more senior data analyst. The two jobs look in different ways, one at the past and one at the future. An analyst looks back and asks what happened and why, then shares it in charts and reports. A scientist looks ahead and builds models that guess what comes next. Both use the same clean data.

Data

Data Engineer vs Analytics Engineer

People think an analytics engineer is a data engineer with a new title. The data engineer brings raw data in from apps and files. They keep the pipes running. The analytics engineer shapes that raw data into clean, tested tables. One moves the data. The other makes it ready to use.

Data

Time Series Data vs Event Data

People think event data is just a time series, since both carry a time. A time series is a reading taken on a clock, such as once a minute. Its gaps are steady. Event data is a row written only when something happens, so it comes in bursts. You read the first in order and count the second in windows.

Data

Schema Evolution vs Schema Drift

People think schema drift and schema evolution are the same change. Evolution is planned: a team adds a field on purpose and checks that readers still work. Drift is a surprise: a source changes its shape and nobody downstream was told. The first is a habit to build. The second is a problem to catch.

Data

Grok vs Groq

People think Grok and Groq are one thing, because the names sound the same. Grok is a chat app and its models, made by SpaceXSI. Groq is a firm that builds fast chips and runs other labs' open models on them. You talk to Grok. You send your own app's prompts to Groq.

AI

Gemini vs Gemma

People think Gemma is just a small Gemini, and it is not. Gemini runs on Google's servers, and you reach it online. Gemma is its own family of open models. You download it and run it on your own phone or laptop. It is smaller, so it costs nothing per token but does less on hard tasks.

AI

Mean vs Median

People use the mean and the median as if they were the same average, and they are not. The mean adds up every value and shares it out, so one huge value drags it up. The median is the middle value once they are sorted, and a huge value barely moves it. For pay and house prices, the median shows the typical case.

General

Excalidraw vs draw.io

People think Excalidraw is a free copy of draw.io, and it is not. Excalidraw is a whiteboard for quick sketches. Its boxes look drawn by hand, and a team draws on one board at once. draw.io makes neat diagrams from shape libraries, with lines that stay stuck to the boxes. Sketch an idea in Excalidraw, and draw the version people keep in draw.io.

Design

Mermaid vs draw.io

People think Mermaid and draw.io are two apps for the same job, and they are not. In draw.io you drag every box into place by hand. In Mermaid you write lines of text, and it places the boxes for you. A Mermaid diagram lives in a Markdown file beside the code, so it gets reviewed and fixed like code. draw.io wins when the exact layout matters.

Design

Dify vs Flowise

People think Flowise is a live rival to Dify, and since 2026 it is not. Both let you build AI apps by joining blocks on a canvas. Dify is still built and fixed by its makers. The Flowise team stopped work in July 2026 and archived the code in August. Its code is still free, but new security holes stay open.

Development

C4 Model vs UML

People think the C4 model is a kind of UML, and it is not. UML is a set of symbols with strict rules, and it has many kinds of diagram. C4 is a set of four zoom levels, from the whole system down to the code. It asks for no special symbols. Many teams draw C4 maps with plain boxes and arrows, and use UML only for the code level.

Development

Anki vs Leitner System

People think Anki is the Leitner system on a phone, and it is not. The Leitner system moves paper cards between a few boxes, and each box has a fixed gap. Anki sets a new gap for each card from how you rated your last answer. A hard card comes back in minutes, and an easy one can wait for months.

General

Sample vs Population

People treat what a sample shows as the truth about everyone, and it is not quite that. The population is the whole group you care about. The sample is the small part you actually measured. A sample only speaks for the group it was fairly drawn from, and always with some margin of error.

General

Correlation vs Causation

People see two things rise together and think one causes the other, and that does not follow. Correlation only says they move together. Causation says that changing one changes the other. A hidden third thing can drive both, so you need a fair test with random groups to show a cause.

General

Percentage vs Percentage Point

People call a move from 10% to 12% a 2 percent rise, and it is not. It is a rise of 2 percentage points. Next to where it started, it is a 20 percent rise. Points give the plain gap. Percent compares the gap to the start.

General

Percentile vs Percentage

People read the 90th percentile as 90% right, and it is not. A percentage is a part of a whole, like 45 of 50 answers. A percentile is a rank: it says you beat 90% of the others. A low mark can still be a high percentile on a hard test.

General

Variance vs Standard Deviation

People treat variance and standard deviation as two measures of spread, and they are one measure in two forms. The variance is the average squared gap from the mean. The standard deviation is its square root, so it is back in the same units as the data. That is why people quote the standard deviation.

General

Color Correction vs Color Grading

People use these two words as if they mean the same thing, and they do not. Correction fixes the footage so it looks natural and every shot matches. Grading comes after, and gives the film a chosen mood. Correct first, then grade.

Design

Video Resolution vs Bitrate

People think a bigger resolution always means a sharper video, and it does not. Resolution is how many pixels each frame has. Bitrate is how much data each second gets to fill them. A 4K video with too little data can look worse than a sharp 1080p one.

Design

Container Image vs Container

People use image and container as if they meant the same thing, and they do not. An image is the saved package, and it never changes. A container is one copy of it that is running right now. One image can start many containers. Delete a container, and the image is still there to start a new one.

Infrastructure

Dockerfile vs Docker Compose

People think a compose file replaces the Dockerfile, and it does not. A Dockerfile says how to build one image. A compose file says which containers to run together and how they link. The compose file often points to a Dockerfile to build from. Most apps with two or more containers use both.

Infrastructure

Performance Review vs 360-Degree Feedback

People think a 360 is just a bigger performance review, and it is not. A performance review is the manager's formal verdict, and it often sets pay. A 360 collects views from peers, reports and the person, so they can learn. When a 360 sets pay, people soften what they say, and it stops working.

Business

Employee Turnover vs Churn Rate

Both count people leaving, so the words get mixed up. Employee turnover counts staff who leave a company. Churn rate counts customers who stop buying. Attrition can mean either one, so check which people a report is counting.

Business

Copyright vs Trademark

People think copyright protects a name or a logo, and it does not. Copyright guards a creative work, like a book, a song or code, and starts by itself. A trademark guards the brand name and logo that tell buyers who made the goods. You register it, and it lasts as long as you use it.

Business

Patent vs Trade Secret

People think a patent is the only way to protect an invention, and it is not. A patent makes you show how it works, and it usually runs out 20 years after you file. A trade secret stays hidden and has no end date. But once someone works it out fairly, it is gone.

Business

LLC vs Corporation

Both shield the owners' own money from the business's debts, so they get mixed up. An LLC is the simpler form, with fewer rules and less paperwork. A corporation is owned through shares and run by a board. Most investors expect one, because shares are easy to sell and pass on.

Business

Intrinsic vs Extrinsic Motivation

People think a reward always adds to the drive a person already has, and it does not. Intrinsic drive comes from the task itself, like fun or curiosity. Extrinsic drive comes from outside, like pay or praise. Rewards work well for dull tasks. Put a prize on something a person already loves, and their own drive can shrink.

General

Sunk Cost Fallacy vs Loss Aversion

People think the sunk cost fallacy is just another name for loss aversion, and it is not. Loss aversion is about the future: a loss you might take hurts more than an equal gain. The sunk cost fallacy is about the past: money already gone still steers the next choice. Loss aversion helps explain it, but the fix differs. For sunk costs, ignore what is spent.

General

Cognitive Bias vs Logical Fallacy

People think a cognitive bias and a logical fallacy are the same thing, and they are not. A bias is a pull inside your own mind that bends what you notice and believe. A fallacy is a broken step in an argument, one you can point to in the words. A bias can lead you to make a fallacy, but you fix them in different ways. You catch a fallacy by checking the steps. You catch a bias with checklists and a second opinion.

General

Survivorship Bias vs Cherry Picking

People think survivorship bias is just another name for cherry picking, and it is not. Cherry picking is a choice: someone picks the facts that help and hides the rest. Survivorship bias needs no one to hide anything. The failures drop out by themselves, so even an honest person only sees the winners. The fix for one is to ask for all the data. The fix for the other is to ask where the missing cases went.

General

AI Copilot vs AI Agent

People think an AI copilot and an AI agent are the same helper. They are not. A copilot sits in your app and suggests the next step, and you say yes or no to each one. An agent takes a goal and does the steps by itself, using tools, and comes back when it is done. With a copilot you drive. With an agent you check the result.

AI

OCR vs Intelligent Document Processing

People think document processing is just scanning, and it is not. OCR turns a picture of a page into letters. It does not know what they mean. Document processing takes that text and works out which value is the name, the date or the total. Then it puts each one in the right box in your system.

AI

DPO vs RLHF

People think DPO is just a kind of RLHF, and it is not. Both teach a model from pairs of answers that people ranked. RLHF first trains a reward model to score answers. Then it uses trial and error to push the chat model toward high scores. DPO skips both steps and trains the chat model on the pairs straight away. So it is simpler to run.

AI

Context Rot vs Lost in the Middle

People think context rot and lost in the middle are one problem, and they are not. Lost in the middle is about where a fact sits. The model misses facts in the middle of a long input more than ones at the start or the end. Context rot is about how much text there is. The more you add, the worse the model does, wherever the fact sits.

AI

Closed Model vs Open-Weight Model

People think a closed model is better and an open one is worse, and that is not the split. The split is who holds the model file. With a closed model, the maker keeps it, and you reach it online. With an open-weight model, you can download it and run it on your own computer. Both kinds come in strong and weak.

AI

180-Degree Rule vs 180-Degree Shutter Rule

People think these are one rule because they share a name, and they are two. The 180-degree rule is about where the camera stands, so it stays on one side of the line between two people. The 180-degree shutter rule is about time. It sets the shutter to about double the frame rate, like 1/50 at 24 frames.

Design

Color Temperature vs White Balance

People use the two words for the same thing, and they are not. Color temperature is a fact about the light: how orange or blue it is, in kelvin. White balance is a camera setting. You set it to match the light, so white things look white.

Design

Shutter Speed vs Frame Rate

People mix these up because both are about time in video, and they are not the same. Frame rate is how many pictures the camera takes each second, like 24. Shutter speed is how long each one of those pictures takes in light, like 1/50 of a second. The shutter can never stay open longer than one frame.

Design

Kubernetes Pod vs Container

People think a pod is just the Kubernetes word for a container, and it is not. A pod is a wrapper around one container or a few. The containers in one pod share an address and start and stop together. Kubernetes places and replaces pods, never loose containers. Most pods hold one container, which is why the two get mixed up.

Infrastructure

Kubernetes Service vs Kubernetes Ingress

People think a service and an ingress do the same job, and they do not. A service gives a group of pods one steady address inside the cluster. An ingress lets web traffic in from outside, and sends each site or path to the right service. Traffic from the internet usually passes through both, the ingress first.

Infrastructure

Kubernetes Deployment vs Deployment (Software)

People read a Kubernetes Deployment as the act of deploying, and it is not that. Deploying is the act of putting a new version live, by any means. A Kubernetes Deployment is a saved setting in a cluster. It keeps a set number of pods running and swaps them in when the image changes.

InfrastructureDevelopmentcross-domain

Monitoring vs Observability

People use the two words as names for one thing, and they are not. Monitoring watches numbers you chose ahead of time and tells you when one goes wrong. Observability is how well the data your system sends out lets you find out why. The first catches the problems you expected. The second helps with the ones you did not.

Infrastructure

Prometheus vs Grafana

People think Grafana collects the numbers on its charts, and it does not. Prometheus visits each app on a timer, reads its numbers and stores them. Grafana reads from a store like Prometheus and draws the charts. One keeps the data and the other shows it. Most teams run the two side by side.

Infrastructure

Dashboard vs Report

People think a dashboard and a report are the same page of charts. A dashboard is live: you open it when you need it, filter it and click into it. A report is fixed: it is sent on a set day, and every reader sees the same page. Use a dashboard to watch, and a report to keep a record.

Data

Predictive vs Prescriptive Analytics

People think predictive and prescriptive analytics both tell you what to do. Predictive only guesses what will happen, like how many loaves will sell on Saturday. Prescriptive takes that guess and picks the best action, like how many loaves to bake. The first gives a number. The second gives a plan.

Data

Google Data Studio vs Looker

People think Data Studio and Looker are the same Google tool. That is partly because Data Studio was called Looker Studio until 2026. Data Studio is free and quick. It suits one person or a small team making charts. Looker is a paid tool where a data team writes down how each number is worked out, in a language called LookML.

Data

Power Query vs DAX

People think Power Query and DAX are two names for one Power BI language. Power Query gets the data and cleans it before it is loaded. DAX works on the data after it is loaded, to add up numbers like total sales or growth on last year. Clean the rows with Power Query, then do the math with DAX.

Data

Row-Oriented vs Columnar Storage

People think row storage and column storage hold different data. They hold the very same table, just laid out two ways on disk. Row storage keeps each record together, so writing one order is fast. Column storage keeps each column together, so adding up one column is fast. Apps use rows. Warehouses use columns.

Data

Data Vault vs Dimensional Modeling

People think data vault and dimensional modeling compete, so a team must pick one. They do different jobs in the same warehouse. A data vault takes in data from many sources. It keeps every change. A dimensional model lays data out as facts and details for reports. Many teams build the vault first, then star schemas on top.

Data

One Big Table vs Star Schema

People think one big table is just a messy star schema. Both hold the same facts and details. A star schema keeps each detail table apart and joins them when you ask. One big table does the joins once, ahead of time, and keeps the result. The star is easier to change. The big table is easier to query.

Data

ClickHouse vs DuckDB

People think ClickHouse and DuckDB are the same kind of tool, since both store data in columns and add up numbers fast. DuckDB runs inside your own program, on one machine, with no server. ClickHouse runs as a server, often on many machines. Many people can query it at once. Use DuckDB on a laptop and ClickHouse behind a live dashboard.

Data

Kernel vs Operating System (OS)

People think the kernel is the whole system, and it is not. The kernel is only the core. It runs the chip, the memory and the devices. The system is the kernel plus all that sits on top: the desktop, the tools and the apps.

Development

Linux vs Unix

People think Linux is a kind of Unix, and it is not quite. Unix is the older system, first built in 1969. Linux was written from scratch in 1991 to work the same way. So it shares the commands and the ideas, but none of the original code. That is why it is called Unix-like.

Development

x86-64 vs ARM Architecture

People think a program runs on any chip, and it does not. x86-64 and ARM are two chip families that read different machine code. Most Windows PCs use x86-64. Phones and new Macs use ARM. A program needs a build for the chip it will run on, or a layer that translates.

Development

macOS vs Linux

People think macOS is a kind of Linux, since their terminals look alike. It is not. Both follow the old Unix design, so many commands match. But macOS is Apple's own system, built on a different core. Linux is a free kernel that many groups build systems on.

Development

Flexbox vs CSS Grid

People think flexbox and grid do the same job, and they do not. Flexbox lines things up in one row or one column. Grid puts things in rows and columns at once. Most pages use both: grid for the page, flexbox for the menu inside it.

Development

JSX vs HTML

People think JSX is HTML, and it is not. JSX only looks like HTML. It lives inside JavaScript, and a build step turns each tag into a function call. That is why it says className, not class, and why every tag must close.

Development

Media Query vs Responsive Breakpoint

People think a media query and a breakpoint are the same thing, and they are not. The breakpoint is the screen width where the layout should change. The media query is the CSS rule that makes the change happen at that width.

DevelopmentDesigncross-domain

Semantic HTML vs ARIA Attributes

People think ARIA is the way to make a page accessible, and it is not the first step. Semantic HTML comes first: a real button or nav already tells a screen reader what it is. ARIA fills the gaps that no HTML element covers.

DevelopmentDesigncross-domain

Virtual Environment vs Version Manager

People think a virtual environment picks the Python version, and it does not. It keeps the packages of one project apart from the rest. A version manager picks which version of the language runs. Many projects use both: one for Python itself, one for its packages.

Development

Package Registry vs Package Manager

People think the package manager and the package registry are one thing, since both are often called npm. They are not. The registry is the website that stores the packages. The package manager is the tool on your computer that downloads them and puts them in your project.

Development

dbt Snapshot vs Incremental Model

People think a dbt snapshot and an incremental model do the same job, since both add rows to a table over time. An incremental model adds new rows to save time, and a changed row can replace the old one. A snapshot is there to keep history. A changed row is added as a new version, and the old one stays. Use a snapshot when you need to know what a value used to be.

Data

Data Skew vs Data Shuffle

People think data skew and a data shuffle are the same slowdown, because both show up when a big job on many machines is slow. A shuffle is the step where rows move between machines, so rows with the same key meet. Skew is when one key has far more rows than the rest, so one machine gets most of them. A shuffle is often needed. Skew is a problem you can fix.

Data

SLO vs SLA

People use SLO and SLA as if they meant the same, and they do not. An SLO is the goal a team sets for itself, like 99.95% of requests working. An SLA is a promise in a contract with buyers, and missing it costs money. The SLA is set lower, so the team misses its own goal first and has time to act.

Infrastructure

High Availability vs Fault Tolerance

People use the two names as one idea, and they are not. With high availability, a service may blip for a few seconds while traffic moves to a spare. With fault tolerance, spares run at the same time, so there is no blip at all. The second costs far more, so most sites only need the first.

Infrastructure

Firewall vs VPN

People think a VPN and a firewall do the same job, and they do not. A firewall is a filter at the edge of a network. It lets some traffic in and blocks the rest. A VPN is a locked tunnel that carries your traffic across the internet so nobody on the way can read it. Many networks use both.

Infrastructure

ERD vs Database Schema

People think an ERD and a database schema are the same thing, and they are not. The schema is the real set of tables and columns inside the database. An ERD is a drawing of it. You can change the drawing all day, and the database stays the same until you change the schema.

DataDevelopmentcross-domain

Supabase vs PostgreSQL

People think Supabase is its own kind of database, and it is not. Under it sits a normal PostgreSQL database, the same one you can run yourself. Supabase adds login, file storage and a ready-made API around it. So the SQL you know from Postgres works there too.

Development

Active Recall vs Spaced Repetition

People think active recall and spaced repetition are one method, because flashcard apps use both. They are not the same. Active recall is how you study: you pull the answer from memory. Spaced repetition is when you study: you come back after longer and longer breaks. Each one helps alone, and they work best together.

General

Interleaving vs Spaced Repetition

People think interleaving is just another word for spacing, and it is not. Spacing is about time: you come back to a topic after a break. Interleaving is about order: you mix kinds of problems, so each one makes you pick a method. Mixing does add some spacing, which is why they get mixed up.

General

Learning Styles vs Dual Coding

People think dual coding is the same as teaching visual learners, and it is not. Learning styles says each person should get one kind of lesson, and tests found no clear gain. Dual coding says everyone gains from words and a picture together. One sorts people. The other gives every learner two ways in.

General

Data Custodian vs Data Steward

People think a data custodian is another name for a data steward, and it is not. The steward looks after what the data means. They write down what fields mean and chase bad rows. The custodian runs the systems that hold the data. They set up storage, backups and logins. One cares for the words, and the other cares for the machines.

Data

Reference Data vs Master Data

People think reference data and master data are the same thing. Reference data is a short list of allowed codes, like country codes. It changes a few times a year. Master data is the records of the things a business deals with, like customers and products. It changes every day, and it uses the codes from the reference list.

Data

Column-Level Security vs Data Masking

People think closing a column and masking it do the same job. A closed column cannot be read at all. A query that asks for it fails with an error. A masked column can be read, but the value comes back hidden, like a card number shown as stars. Closing suits fields a role never needs. Masking suits fields a role needs to see exist.

Data

Data Profiling vs Data Quality Check

People think profiling data is the same as testing it. A profile only counts what is there: blanks, values and ranges. It passes no judgment. A quality check tests a rule, like no blank emails, and fails when the rule breaks. You profile first to learn the data. Then you write checks for the problems you found.

Data

Monetary Policy vs Fiscal Policy

People think both mean the government steering the country, and they do not. A central bank runs monetary policy. Its tool is the interest rate. Lawmakers run fiscal policy with taxes and spending. The two can push the same way or pull against each other.

Business

Microeconomics vs Macroeconomics

People think macro is just micro added up, and it is not. Micro looks at one buyer, one firm or one market, and how its prices are set. Macro looks at a whole country: its growth, its prices and its jobs. What is wise for one family, like saving more, can hurt a whole country when everyone does it at once.

Business

Monopoly vs Oligopoly

People call any market with a few giant firms a monopoly, and it is not. A monopoly has one seller, so buyers cannot switch. An oligopoly has a few big sellers who watch each other. Buyers there can still walk to a rival, and that limits what each firm can charge.

Business

Comparative Advantage vs Competitive Advantage

People mix these up because both have advantage in the name. Comparative advantage is about trade: do the job you give up least to do, and swap. Competitive advantage is about rivals: what lets a firm win buyers or earn more than others. One explains why trade helps both sides. The other explains why one firm beats another.

Business

Absolute Advantage vs Comparative Advantage

People think whoever is best at a job should do it, and that is not the rule. Absolute advantage is being able to make more with the same time. Comparative advantage is giving up the least to do the job. A lawyer who types fast still hires a typist, because her hour is worth more at law.

Business

Marginal Cost vs Variable Cost

People think the cost of one more unit is just the variable cost per unit, and often it is not. Variable cost is the total that grows with output. Marginal cost is what the very next unit adds. When machines and staff are stretched, the next unit costs more than the average one did.

Business

Supply Chain vs Logistics

People use logistics and supply chain as if they meant the same thing, and they do not. Logistics is moving and storing goods. The supply chain is the whole path, from raw materials to the buyer, and it also covers buying, making and planning. Logistics is one part of it.

Business

Lean Manufacturing vs Six Sigma

People think lean and Six Sigma are two names for one method, and they are not. Lean cuts waste, so work flows faster with fewer steps. Six Sigma cuts errors, using data to make each result the same. Many firms use both, and call the mix Lean Six Sigma.

Business

Safety Stock vs Reorder Point

People think the reorder point is the same as safety stock, and it is not. Safety stock is the spare you keep for surprises. The reorder point is the level that says order now. It is the stock you will sell while you wait for the truck, plus that spare.

Business

PDCA vs DMAIC

People think these are two rival ways to improve, and they are not. PDCA is a quick loop for testing one small change, again and again. DMAIC is a longer project with five steps, for one stubborn problem that needs data to crack. Use the loop for small fixes, and the five steps when the cause is unknown.

Business

Custom Instructions vs System Prompt

People think their custom instructions are the system prompt, and they are not. The app writes its own system prompt, and you never see it. Your custom instructions are a small note added to it. So the app's own rules still win when the two clash.

AI

Iterative Prompting vs Prompt Chaining

People think prompt chaining is just asking follow-ups, and it is not. In iterative prompting, you read each draft and ask for a change in the same chat. In a chain, a program passes each answer into the next prompt as a fixed step. One is you steering. The other is a pipeline you set up once.

AI

Data Augmentation vs Synthetic Data

People think data augmentation makes synthetic data, and it does not. Augmentation takes a real example and changes it a bit, like a flipped photo. Synthetic data is made from scratch, often by a model or a program. One stretches the data you have. The other makes new data you never collected.

AI

Hyperparameter Tuning vs Fine-Tuning

People think both kinds of tuning change the model, and only one does. Hyperparameter tuning picks the settings for training, like the learning rate, by training many times. Fine-tuning takes a model that is already trained and trains it more on new data. The first picks the recipe. The second keeps cooking.

AI

Feature Scaling vs Database Normalization

People hear normalization and think of one idea, and there are two. In machine learning, it means scaling number columns to a similar range. In databases, it means splitting tables so each fact is stored once. One changes the numbers. The other changes the table layout.

AIDevelopmentcross-domain

Title Case vs Sentence Case

People think a heading needs a capital on every word, and it does not. Title case caps most words but keeps small ones like of and the low. Sentence case caps only the first word and names. Both are right. The mistake is mixing them on one site.

Design

Lorem Ipsum vs Placeholder Text

People call lorem ipsum placeholder text, and in a form that word means something else. Lorem ipsum is fake text that fills a layout and is gone before launch. Placeholder text is the grey hint inside a real input box, like "Search". It ships with the product and disappears once you type.

Design

Gross Pay vs Net Pay

People think the pay in a job offer is the money they will get, and it is not. Gross pay is the full amount before anything comes out. Net pay is what lands in your bank after tax, pension and other costs. Plan your spending on net pay, because that is the money you really have.

General

APR vs APY

People think APR and APY are two names for one rate, and they are not. APR is what a loan costs in a year, with its fees counted in. APY is what savings earn in a year, with interest on interest counted in. Use APR to compare loans and APY to compare savings accounts.

General

Simple Interest vs Compound Interest

People think 5% a year for ten years adds up to 50%, and with most savings it does not. Simple interest is paid on the first amount only, so it grows by the same sum each year. Compound interest is also paid on past interest, so it grows faster each year. On 1,000 over ten years, that is 1,500 against about 1,629.

General

Debt Snowball vs Debt Avalanche

People think every way of paying off debt costs the same in the end, and it does not. The snowball pays the smallest balance first, for quick wins. The avalanche pays the highest interest rate first, which costs the least. The avalanche saves more money. The snowball helps some people keep going.

General

Index Fund vs ETF

People think an ETF and an index fund are the same thing, and they are not. Index fund says what a fund holds: it copies a list, like the S&P 500. ETF says how it is traded: on an exchange, all day, like a share. Many ETFs are index funds, but some are not, and many index funds are not ETFs.

General

Full Load vs Incremental Load

People think an incremental load gives the same result as a full load, only faster, and it does not. A full load copies every row, so the copy always matches. An incremental load copies only rows past a marker, such as a time stamp. A row deleted at the source never passes that marker, so it stays in the copy. Many teams run both: small loads each night and a full one now and then.

Data

Apache Kafka vs Confluent

People think Confluent and Kafka are the same thing. Apache Kafka is free, open source software that anyone can run. Confluent is a company that sells Kafka as a service. Its cloud runs the brokers for you and adds tools, like a schema registry. Your apps speak Kafka either way, so moving between them is mostly a change of settings.

Data

Kinesis Data Streams vs Amazon Data Firehose

People think Firehose is just another name for Kinesis Data Streams. Kinesis Data Streams keeps events for a while, and your own apps read them, as many times as they like. Firehose keeps nothing to replay. It gathers events into batches and drops them into a store, like Amazon S3. One is a stream you read. The other is a chute that delivers.

Data

Router vs Network Switch

People think a router and a switch are the same box, and they are not. A switch links devices inside one network, like the laptops in one office. A router joins one network to another, like your home to the internet. Most home routers have a small switch built in, which is why the two get mixed up.

Infrastructure

TCP vs UDP

People think TCP and UDP are two names for the same thing. Both send packets, but TCP checks that every packet arrives and in the right order. It sends lost ones again. UDP sends and does not check. TCP is for pages and files that must be whole. UDP is for calls and games, where late data is no use.

Infrastructure

Public IP Address vs Private IP Address

People think every device has its own public IP address, and most do not. A public address can be reached from anywhere on the internet. A private address only works inside one home or office. Millions of homes reuse the same private ones. On the way out, the router swaps them for its one public address.

Infrastructure

MAC Address vs IP Address

People think a MAC address and an IP address do the same job. A MAC address is set on the network card when it is made, and it is used inside one local network. An IP address is handed out by the network you join, and it gets data across the internet. The IP address finds the network. The MAC address finds the device on it.

Infrastructure

Proxy Server vs VPN

People think a proxy hides them the same way a VPN does, and it does not. Both stand between you and the site you visit. A VPN scrambles all the traffic from your device inside a tunnel. Many proxies only pass on web requests, and they may not scramble anything. The owner of the proxy can see every request.

Infrastructure

Firewall vs Antivirus Software

People think a firewall and antivirus do the same job. A firewall decides which traffic may come in or go out. It looks at ports and addresses, not inside files. Antivirus checks the files and programs on a computer and stops bad ones. One guards the door. The other checks what is already inside.

Infrastructure

Phishing vs Malware

People think phishing is a kind of malware, and it is not. Phishing is a fake message. It fools a person into giving up a password or clicking a link. Malware is a harmful program that runs on a machine. The two often work together, because a fake email is a common way to get malware in.

Infrastructure

Primary Source vs Secondary Source

People think any write-up of a study is as good as the study, and it is not. A primary source is the first-hand record. That can be the study's own report, its data or a diary. A secondary source explains or sums up primary ones, like a textbook or a news story. It saves time. Before you rely on a claim, go back to the first-hand record.

General

Qualitative vs Quantitative Research

People think research with numbers is the real kind and research with words is just opinion. That is wrong. Quantitative research counts and measures. It shows how much and how often. Qualitative research works with words and stories. It shows why and how. Each one answers questions the other cannot, and many studies use both.

General

Systematic Review vs Meta-Analysis

People think these are two names for one thing, and they are not. A systematic review is the search: it finds every study on a question, by set rules. A meta-analysis is the maths: it pools the numbers from those studies into one answer. Many reviews end with one. Some cannot, because the studies do not match.

General

Experiment vs Observational Study

People think a big study that finds a link has found a cause. It has not. An observational study only watches, so a hidden factor may explain the link. An experiment changes one thing on purpose and compares it with a group left alone. That is how it can show a cause. Watching alone usually cannot.

General

Validity vs Reliability

People think a measure that gives the same answer each time must be right. It need not be. Reliability means the measure is steady: do it again and you get the same result. Validity means it measures the right thing. A scale that always reads two kilos too heavy is steady, so it is reliable. It is still wrong, so it is not valid.

General

Dense Vector vs Sparse Vector

People think a sparse vector is a weaker dense one, and it is not. A dense vector scores hidden traits of meaning, so it finds texts that say the same thing in other words. A sparse vector has one slot per word, so it finds exact words, names and codes. Each catches what the other misses, which is why hybrid search uses both.

AI

Knowledge Base vs Knowledge Graph

People think a knowledge base and a knowledge graph are the same store, and they are not. A knowledge base is a set of documents that an AI system searches as text. A knowledge graph holds facts as things and named links between them. You read the first. You follow the lines of the second.

AI

Top-k Retrieval vs Top-k Sampling

People see top-k in two settings and think it is one knob, and it is not. In retrieval, k is how many matching chunks the search hands to the model. In sampling, k is how many likely next words the model picks from as it writes. One decides what the model reads. The other shapes how it writes.

AI

FAISS vs Vector Database

People think FAISS is a vector database, and it is not. FAISS is a library: the fast search part, built into your own program. A vector database is a whole system around that part. It adds storage, updates, filters, backups and users. Pick FAISS to build it yourself, and a database to have it done.

AI

Airflow Catchup vs Backfill

People think catchup and a backfill are the same thing. Catchup is a switch on a DAG. When the DAG is turned on late, Airflow runs every missed date by itself, all at once. A backfill is a run you ask for, over dates you choose, when you are ready. Catchup happens to you. A backfill is something you plan.

Data

Airflow Hook vs Operator

People think a hook is just another kind of operator. A hook is a helper that logs in to one system and talks to it. It is not a task and never shows in the graph. An operator is a task. Many operators use a hook inside to do their work. When no operator fits, you call the hook from your own task.

Data

Pipeline Alerting vs Data Observability

People think a pipeline with no failure alerts has good data. Pipeline alerts watch the jobs: did a task fail, did a run finish late? Data observability watches the data itself: are rows missing, are values strange? A job can finish green and still load an empty table. You need both to trust the numbers.

Data

Card Sorting vs Tree Testing

People think card sorting and tree testing are the same test, and they are not. In card sorting, people put topics into groups to help you build a menu. In a tree test, people try to find things in a menu you already built. Sort first to shape it, then tree test to check it.

Design

UX Writing vs Content Design

People use UX writing and content design as two names for one job, and they are not quite the same. UX writing is the words on screens, like buttons, hints and errors. Content design starts earlier. It asks what people need to know, then picks words, a table or a picture.

Design

Choice Overload vs Hick's Law

People think choice overload is just Hick's Law, and it is not. Hick's Law says each extra option makes a choice take a bit longer. Choice overload says too many options can stop people from choosing at all. One is about speed. The other is about walking away.

Design

Visual Sitemap vs XML Sitemap

People think a sitemap is one thing, and there are two. A visual sitemap is a drawing of a site's pages that a team plans with. An XML sitemap is a file that lists page addresses for search engines. One is for people, the other is for robots. A site often has both.

DesignMarketingcross-domain

Online Store vs Online Marketplace

People think an online store and a marketplace are the same thing, and they are not. Your own store is your site, your prices and your list of buyers, but you must bring every visitor. A marketplace brings the buyers for you. In return it takes a fee on each sale and sets the rules.

Business

Payment Gateway vs Payment Processor

People use these names as if they meant the same thing, and they do not. The gateway takes the card details at checkout and passes them on safely. The processor then moves the money through the card network and the banks. Many firms, like Stripe, sell both in one.

Business

Refund vs Chargeback

People think a chargeback is just a refund, and it is not. A refund is the seller's choice to pay a buyer back. A chargeback is the buyer's bank taking the money back by force after a dispute. It also adds a fee, and too many can get a store in trouble.

Business

GMV vs Revenue

People read GMV as a store's sales, and it is not. GMV is the full price of all goods sold on a site. Revenue is the part the site keeps. A marketplace with 1 million dollars of GMV and a 10% fee has only 100,000 dollars of revenue.

Business

Shopify vs WooCommerce

People think these are two brands of the same thing, and they are not. Shopify is hosted: you pay a plan, and it runs the server for you. WooCommerce is a free add-on for WordPress that you host and look after yourself. One trades control for ease, and the other trades ease for control.

Business

Dropshipping vs 3PL

People think dropshipping is just using a 3PL, and it is not. With a 3PL you buy the stock and own it, and the warehouse ships it for you. With dropshipping you never buy the stock at all. The supplier owns it and ships each order straight to the buyer.

Business

Color Space vs Color Gamut

People use color space and color gamut as if they mean the same, and they do not. A color space is a set of colors with exact numbers, like sRGB. A gamut is the range of colors a space or a screen can reach. A screen's gamut decides which colors in a file it can really show.

Design

JPEG vs PNG

People think PNG is just a better JPEG, and it is not. JPEG makes photos small by dropping detail you rarely see. PNG keeps every pixel and can have see-through parts. Use JPEG for photos, and PNG for logos, screenshots and text.

Design

Image Resolution vs Pixels per Inch

People think a 300 PPI file is a sharp one, and it is not. Resolution is how many pixels a picture has, like 1800 by 1200. PPI only says how tightly those pixels are packed when it prints. The pixel count decides how big it can print and stay sharp.

Design

Virtual Machine vs Container

People think a virtual machine and a container are the same thing, and they are not. A virtual machine is a whole computer in software, with its own operating system. A container is just an app and its parts. It shares the system of the machine it runs on. So a container starts in seconds and is much smaller. A VM keeps its users further apart.

Infrastructure

Type 1 vs Type 2 Hypervisor

People think the two kinds of hypervisor do different jobs, and they do the same one. Both split a machine into virtual machines. The difference is what sits under them. A type 1 hypervisor runs right on the hardware, and cloud servers use it. A type 2 hypervisor runs as an app on Windows or macOS, and you use it on a laptop to try another system.

Infrastructure

Hybrid Cloud vs Multi-Cloud

People use hybrid cloud and multi-cloud as if they were one idea, and they are two. Hybrid cloud joins your own servers to a public cloud, so the two work as one. Multi-cloud means using two or more public cloud firms, such as AWS for one job and Google Cloud for another. A firm can do both at once, which is why the words get mixed up.

Infrastructure

Machine Image vs Container Image

People think a machine image and a container image are the same kind of file, and they are not. A machine image is a whole saved disk, with its own system. A virtual machine starts from it. A container image holds just an app and its parts. It borrows the system of the machine that runs it. So a machine image is far bigger and slower to start.

Infrastructure

Private Cloud vs On-Premises

People think a private cloud just means servers in your own building, and it does not. On-premises only says where the servers sit. A private cloud is a way of running them. Teams start their own virtual machines when they need them, as in a public cloud. A server room where each machine is set up by hand is on-premises, but it is not a private cloud.

Infrastructure

Reverse Proxy vs Forward Proxy

People think a reverse proxy is just a proxy, and the two face opposite ways. A forward proxy works for users. It sits in front of their browsers and goes out to the web for them. A reverse proxy works for servers. It sits in front of them and takes requests coming in. Both pass traffic on, which is why the names get mixed up.

Infrastructure

Load Balancer vs Reverse Proxy

People think a load balancer and a reverse proxy are two different kinds of server, and very often they are one and the same. A reverse proxy takes each request and passes it to a server behind it. A load balancer is a reverse proxy that shares the work across many servers. Nginx can do both jobs. With just one app server behind it, there is no load to share.

Infrastructure

Web Server vs Application Server

People think a web server and an app server are the same program, and they are not. A web server takes every request and sends plain files, like images, by itself. An app server runs your code to build each answer from the data. Most sites run both, with the web server in front. Small sites can use one program for both jobs, which blurs the line.

Infrastructure

Nginx vs Apache

People think Nginx and Apache are the same program under two names, and they are not. Both are free web servers. Nginx keeps its settings in a few main files and handles many visitors with little memory. Apache can also read a small settings file in each folder. That is handy on shared hosting, but slower.

Infrastructure

Let's Encrypt vs Certbot

People think Certbot is Let's Encrypt, and it is not. Let's Encrypt is the free authority that checks your domain and signs the file. Certbot is a tool on your server. It asks Let's Encrypt for the file, puts it in place and renews it. Other tools can talk to Let's Encrypt too.

Infrastructure

Virtual Host vs Virtual Machine

People think a virtual host is a kind of virtual machine, and it is not. A virtual machine is a whole computer made in software. A virtual host is just a block of settings inside one web server. It lets that server answer for many site names. Both words start with virtual, but one splits a machine and the other splits a web server.

Infrastructure

Incremental vs Differential Backup

People think an incremental and a differential backup are the same, and they differ in one way. After a full backup, both copy only changes. An incremental one copies what changed since the last backup of any kind, so each is small. A differential one copies all changes since the last full one, so it grows each day. To restore, incremental needs every link in the chain. Differential needs just two.

Infrastructure

RPO vs RTO

People mix up RPO and RTO, and they measure two things. RPO is how much data you can afford to lose. It looks back from the disaster to the last good backup. RTO is how long you can afford to be down. It looks forward from the disaster until service is back. Backing up more often shrinks the RPO. A faster restore shrinks the RTO.

Infrastructure

Snapshot vs Backup

People think a snapshot is a backup, and on its own it is not. A snapshot is a quick picture of a disk, most often kept on the same storage as the disk. If that storage or the account is lost, the snapshot goes too. A backup is a copy kept somewhere else. A snapshot is great for undoing a bad change. A backup is what saves you from losing everything.

Infrastructure

RAID vs Backup

People think RAID is a backup, and it is not. RAID keeps a server running when one disk dies. But it copies every change to all the disks at once. So if you delete a file, or ransomware locks it, it is gone from every disk. A backup is a separate copy from an earlier time, kept apart. You need RAID to stay up and a backup to get things back.

Infrastructure

Block Storage vs Object Storage

People think object storage is just a disk in the cloud, and it is not. Block storage is a raw disk that one server uses as its own drive, which suits databases. Object storage is a huge bucket of files reached over the web, each with a name. You cannot run a database on it. But it is cheap and holds as many files as you like, which suits backups.

InfrastructureDatacross-domain

Unicode vs UTF-8

People think Unicode and UTF-8 are two names for one thing, and they are not. Unicode is the list that gives each character a number. UTF-8 is one rule for storing those numbers as bytes. So a file is never saved in Unicode alone. It is saved in UTF-8, or in another encoding of Unicode.

Development

File Extension vs MIME Type

People think the ending of a file name says what the file is, and it does not. The extension is a hint that your own computer reads to pick an app. On the web, the browser trusts the media type the server sends instead. When the two disagree, the browser goes with the server's label.

Development

JSON vs Protocol Buffers

People think Protobuf is just a faster JSON, and it is not. JSON carries its own field names, so anyone can read it with no setup. Protobuf sends only numbered fields as bytes. Both sides need the same .proto file to make sense of them. You trade easy reading for small, fast messages.

Development

Vulnerability vs Exploit

People think a vulnerability and an exploit are the same thing, and they are not. The vulnerability is the weak spot in the code. The exploit is the code or the steps that use that weak spot to do harm. A hole can sit there for years with no exploit. Once an exploit is public, fixing the hole becomes urgent.

Development

Same-Origin Policy vs CORS

People think CORS is a lock that blocks requests, and it is the other way round. The same-origin policy is the lock: it stops one site from reading another site's replies. CORS is a header a server sends to open that lock for sites it names. So a CORS error means the server did not give your site a key.

Development

Digital Signature vs Encryption

People think a digital signature hides a message, and it does not. Encryption scrambles a message so only the right key can read it. A signature leaves the message open to read. It proves who sent it and that nobody changed it. Many systems use both, one to keep secrets and one to build trust.

Development

SSRF vs CSRF

People think SSRF and CSRF are the same trick because the names sound alike, and they are not. In CSRF, the attacker gets your browser to send a request to a site you are logged in to. In SSRF, the attacker gets the server itself to send the request. Often it goes to a private address inside the server's own network.

Development

Input Validation vs Parameterized Query

People think checking what users type is enough to stop SQL injection, and it is not. Input validation turns away data that has the wrong shape, like letters in an age box. But a name can still hold a quote. A parameterized query keeps every value apart from the command, so even odd text stays data. You want both.

Development

Identity Provider vs Single Sign-On

People think the login service and single sign-on are one thing, and they are not. The identity provider is the service that checks who you are. It then hands out signed tokens. Single sign-on is what you see as a result: you log in once and many apps let you in. The provider is the engine, and SSO is the ride.

Development

React Native vs Flutter

People think React Native and Flutter work the same way, and they do not. Both build iPhone and Android apps from one codebase. React Native uses JavaScript and asks each phone to draw its own buttons, so the app looks native. Flutter uses Dart and paints every button itself, so the app looks the same on both. Pick by the language your team knows and the look you want.

Development

Progressive Web App vs Native App

People think a progressive web app is a phone app from a store, and it is not. A PWA is a website that you can add to your home screen. It needs no store and no review, and one copy works everywhere. A native app is built for one phone system and comes from its store. It can use every part of the phone, while a PWA reaches fewer of them.

Development

Safetensors vs GGUF

People think safetensors and GGUF are just two names for the same kind of model file. They are made for different jobs. Safetensors holds the full weights, most often at 16 bits, and it is how most models are shared on hubs. GGUF packs a model for local apps, most often shrunk to fewer bits. You often turn the first into the second.

AI

Ollama vs Open WebUI

People think Open WebUI is one more tool like Ollama, and they pick one or the other. They do two different jobs. Ollama runs the model on your machine. Open WebUI is only the chat screen in your browser. It sends your words to a runtime like Ollama and shows the reply, so the two are most often used together.

AI

AI Text Detector vs AI Watermarking

People think an AI detector reads a hidden mark, the way a watermark check does. It does not. A watermark is put into the output on purpose by the tool that made it, so a check for it can be quite sure. A detector has no mark to find. It only guesses from the style of the text, and it is often wrong.

AI

Content Credentials vs AI Watermarking

People think Content Credentials are just another kind of watermark. They are not. A watermark is a hidden signal mixed into the pixels or the sound. Content Credentials are a signed record stored with the file, listing who made it and every edit. The record says more, but it can be stripped off. A watermark says less, and it is harder to remove.

AI

NIST AI RMF vs ISO/IEC 42001

People think the NIST AI RMF and ISO/IEC 42001 are two names for the same rulebook. They are not. The NIST framework is a free guide that a team can follow in any way it likes. ISO/IEC 42001 is a standard that an outside auditor checks, and a company can be certified against it. Many firms use the guide to meet the standard.

AI

Copyright of AI Output vs Training Data Copyright

People think AI and copyright is one big question. It is two. Training data copyright asks if a model may learn from books, art and pages that someone owns. Copyright of AI output asks who, if anyone, owns what the model makes. A court can answer one and leave the other open, and the answers differ by country.

AI

General-Purpose AI Model vs High-Risk AI System

People think a powerful general model is a high-risk AI system under the EU AI Act. It is not the same thing. A general-purpose model is judged as a model, and its maker has its own duties. A high-risk system is judged by its use, such as sorting job forms or scoring loans. A small model can be high risk, and a big one need not be.

AI

AI Safety vs AI Security

People think AI safety and AI security are two words for the same job. They are not. Safety asks if the AI itself could cause harm, through a wrong goal or a mistake. Security asks if an attacker could make it cause harm, with a trick or stolen access. A model can be well meaning and still easy to attack, so a team needs both.

AI

Refusal (AI) vs Guardrail (AI)

People think a model's refusal is a guardrail blocking the request. Most of the time it is not. A refusal comes from the model's own training: it decides to say no. A guardrail is a separate check placed around the model, on the way in or the way out. A good app uses both, because each one misses things the other catches.

AI

Scheming (AI) vs Reward Hacking

People think scheming is just a fancy name for reward hacking. It is not. Reward hacking finds a gap in how a task is scored and uses it, often in plain sight. Scheming hides a goal and acts well while it is being watched, so the tests look fine. Reward hacking shows up in the results. Scheming is built to not show up.

AI

Formative vs Summative Assessment

People think formative and summative assessment are two kinds of test, and they are not. They are two jobs a test can do. A formative check comes during learning, and its job is to show what to fix next. A summative check comes at the end, and its job is to set a grade. The same quiz can be either one. What decides it is what happens to the result.

General

Curriculum vs Syllabus

People think a curriculum and a syllabus are the same plan, and they are not. A curriculum is the plan for a whole program: every course, and the order they come in. A syllabus is the plan for one course inside it: its topics, its due dates and how grades are worked out. A school sets the curriculum. Each teacher writes a syllabus.

General

Tacit vs Explicit Knowledge

People think tacit knowledge is just knowledge no one has written down yet, and it is not. Explicit knowledge can be written down, like a recipe or a guide. Tacit knowledge is the skill and feel that resist words, like knowing when dough is ready. A guide can hold the first kind. The second kind passes from person to person, by working side by side.

General

Knowledge Base vs Wiki

People think a knowledge base and a wiki are the same thing, and they are not. A knowledge base is a set of finished answers, written for readers to search and follow. A wiki is a shared space where many people write and edit pages together. A team often writes in a wiki. The answers it trusts most go into the knowledge base.

General

PARA Method vs Zettelkasten

People think PARA and Zettelkasten are two ways to do the same job, and they are not. PARA sorts notes and files by when you will act on them, so the things you need now are easy to reach. A Zettelkasten links small notes by idea, so new thoughts can grow from the links. Many people use both: PARA for their projects, links for their ideas.

General

Scientific Theory vs Scientific Law

People think a theory becomes a law once it is proven, and it does not. A law says what happens, often as a short sum, like how hard two things pull on each other. A theory says why it happens, and ties many facts together. Both are tested many times. Neither one grows into the other, because they do different jobs.

General

Atom vs Molecule

People think an atom and a molecule are the same thing, and they are not. An atom is one tiny piece of an element, like one oxygen atom. A molecule is two or more atoms joined. Water is one: two hydrogen atoms and one oxygen atom. Atoms are the bricks. Molecules are what you get when you snap bricks together.

General

Mass vs Weight

People think mass and weight are the same thing, and they are not. Mass is how much stuff something has, in kilograms, and it is the same anywhere. Weight is how hard gravity pulls on that stuff, in newtons, and it changes with place. On the Moon your mass stays the same, but you weigh about one sixth as much.

General

DNA vs Gene

People think DNA and genes are the same thing, and they are not. DNA is the whole long molecule in a cell, about 3 billion letters in a person. A gene is one short stretch of it that holds the plan for one job, most often one protein. If DNA is a cookbook, a gene is one recipe in it.

General

Evolution vs Natural Selection

People think evolution and natural selection are the same thing, and they are not. Evolution is the change itself. Living things change over a long, long time. Natural selection is the main reason why. The ones that best fit their world leave more young, so their traits spread. One is what happens, and the other is a big part of why.

General

Fabric Lakehouse vs Fabric Data Warehouse

People think a Fabric lakehouse and a Fabric warehouse are the same SQL store, and they are not. Both keep Delta tables in OneLake. A lakehouse is filled by Spark and holds loose files too. Its SQL endpoint can only read. A warehouse is built with T-SQL, which can add, change and delete rows. Pick the lakehouse for Spark teams and the warehouse for SQL teams.

Data

Microsoft Fabric vs Azure Synapse Analytics

People think Microsoft Fabric is Azure Synapse with a new name, and it is not. Synapse is an Azure service where you set up SQL and Spark pools one by one. Fabric is sold as one product. Every tool keeps its data in one shared lake and draws on one capacity. For a new warehouse, Microsoft now points people to Fabric.

Data

Lakeflow Pipelines vs dbt

People think Lakeflow Pipelines is Databricks' own dbt, and it is not quite. Both let you write SQL that says what each table should hold. dbt runs that SQL in batches, in almost any warehouse. Lakeflow Pipelines runs only in Databricks, and it can keep streaming tables fed as new rows arrive. Its rules can drop bad rows on the way in.

Data

Table Catalog vs Data Catalog

People think a table catalog and a data catalog are the same thing, and they are not. A table catalog is for query engines. It tells Spark or Trino where each lake table lives right now. A data catalog is for people. It is a search page that says what a table means and who owns it. Some products do a bit of both.

Data

Natural Key vs Surrogate Key

People think a natural key and a surrogate key are two names for one ID, and they are not. A natural key comes from the source, like an email or a product code. People can read it. A surrogate key is a plain number the warehouse makes up. Warehouses join on the surrogate key. They keep the natural key to trace each row back.

Data

Apache Arrow vs Apache Parquet

People think Apache Arrow is a file format like Parquet, and it is not. Parquet is how table data is kept on disk, packed small. Arrow is how table data is laid out in memory while tools work on it. They work well as a pair. A tool reads Parquet from disk into Arrow in memory.

Data

Amazon EMR vs AWS Glue

People think Amazon EMR and AWS Glue are the same Spark service, and they are not. EMR builds clusters you can tune and run many tools on. You pay for the machines while they run. Glue is serverless. You hand it an ETL job, and AWS runs the Spark for you. Pick EMR for control and Glue for less work.

Data

Spark Driver vs Spark Executor

People think the Spark driver does the heavy work, and it does not. The driver runs your code, plans the job and hands out tasks. The executors on other machines do the reading and the adding up. Only the final answer comes back to the driver. That is why pulling a huge table into it makes it crash.

Data

Captions vs Audio Description

People think captions make a video work for everyone, and they do not. Captions turn sound into words on the screen, for people who cannot hear. Audio description turns the picture into speech, for people who cannot see. A video that shows facts without saying them needs both.

Design

Text Resize vs Reflow

People think a page that zooms well also reflows, and it may not. Text resize asks that words can grow to 200 percent with nothing lost. Reflow asks that the page fits a screen 320 pixels wide, so you never scroll sideways to read a line. A page can pass one and fail the other.

Design

Accessibility Audit vs Automated Testing

People think a clean scan means the site passed an audit, and it does not. A scan is a tool that finds clear faults, like missing alt text or low contrast. An audit is a person checking every rule, with a keyboard and a screen reader. The scan is one part of the audit, not the whole of it.

Design

EAA vs ADA

People think the European Accessibility Act is just Europe's ADA, and it is not. The ADA is a US civil rights law from 1990 that bans unfair treatment, and it sets no web rules for private firms. The EAA is an EU law that lists products and services, like phones, banks and online shops. Its rules apply from 28 June 2025.

Design

App Icon vs Favicon

People think the tiny icon in a browser tab and the icon on a phone's home screen are the same thing, and they are not. A favicon stands for a website. It shows in the tab and in bookmarks, often at 16 or 32 pixels. An app icon stands for an app you install. It sits on the home screen, and the store asks for a large square file. Both start from the logo, but each is drawn for its own size.

Design

Offset Printing vs Digital Printing

People think offset and digital printing are two names for one machine, but they cost money in opposite ways. Offset prints from metal plates, one per ink. The plates cost money up front, and then each copy is very cheap. Digital prints straight from the file, with no plates. It is cheap to start, so it wins on small runs and on copies that each differ.

Design

Key Visual vs Hero Image

People think the key visual and the hero image are the same picture, and they are not. The key visual is the one master image a whole campaign is cut from. The hero image is the big picture at the top of one page or one email. A hero image is often cut from the key visual. But each page gets its own, and there is only one key visual.

Design

Product Mockup vs Print Proof

People think a good mockup means the print job is safe to approve, and it is not. A mockup is a picture that shows the art on a mug, a bag or a phone. It shows the look, not the true color or size. A print proof is a test copy from the printer, checked before the full run. It is the one that shows what will really come off the press.

Design

Assertive vs Aggressive Communication

People think speaking up in an assertive way and in an aggressive way are the same thing, and they are not. Both say clearly what they want. The aggressive speaker pushes the other person down with blame, shouting or cutting in. The assertive speaker stays calm and treats both sides' needs as real. So assertive can still say no. It just does not attack while doing it.

General

Verbal vs Nonverbal Communication

People think verbal means spoken and nonverbal means written, and that is not right. Verbal means words, so an email and a text are verbal too. Nonverbal is everything that is not a word: the face, the hands, the eyes and the sound of the voice. A phone call carries words and tone, but no body language. A text carries words alone, which is why it is so easy to misread.

General

Passive vs Passive-Aggressive

People think passive-aggressive is just a stronger kind of passive, and it is not. A passive person hides a need and truly lets it go. A passive-aggressive person hides the need but not the anger. It leaks out as sulking, sarcasm or a job done slowly on purpose. Both styles skip the honest talk, so both leave the real problem unsolved.

General

Aerobic vs Anaerobic Exercise

People think aerobic and anaerobic exercise are two kinds of sport, and they are not. They are two ways your muscles get energy. Aerobic effort is steady and uses oxygen, so you can keep going, as on a long walk or swim. Anaerobic effort is a short, hard burst, like a sprint. The muscles need energy faster than oxygen can bring it. Most sports mix both, and a football match is full of each.

General

Stress vs Burnout

People think burnout is just a lot of stress, and it is not. Stress is worry or tension from a hard situation, and a little of it can even help. Burnout is what the WHO describes after long work stress that was never managed. It has three signs: feeling worn out, feeling distant from the job, and getting less done. Rest eases stress, but burnout usually needs the work itself to change.

General

Macronutrient vs Micronutrient

People think micronutrients matter less than the big ones, because you need so little, and that is wrong. The names only say how much you need. You eat grams of carbs, protein and fat every day, and they give you calories. You need vitamins and minerals in tiny amounts, and they give no calories at all. Go without either kind for long and the body stops working well.

General

Grit vs Resilience

People think grit and resilience are the same thing, and they are not. Grit is about one long goal: keeping your passion and effort on it for years. Resilience is about hard moments: bouncing back after a loss, a failure or a shock. A person can recover well from a bad week and still drift from goal to goal. Another can chase one dream for years but take a long time to recover from each blow. Most long goals need a bit of both.

General

Self-Discipline vs Willpower

People think willpower and self-discipline mean the same thing, and they do not. Willpower is the push in one moment, when you hold back an urge, like leaving the last biscuit. Self-discipline is the pattern over weeks, doing what you planned again and again. Disciplined people often use less willpower, not more. They set up habits and spaces so there is less to fight in the first place.

General

Debt Financing vs Equity Financing

People think debt money is cheap and equity money is free, and neither is true. Debt is borrowed, so it must be paid back with interest, even in a bad year. But the owners keep the whole firm. Equity is never paid back. Instead the investor owns a slice and shares every future profit. If the firm does well, that slice can cost far more than any loan.

Business

Primary Market vs Secondary Market

People think buying a share on the stock exchange pays the firm, and it does not. The firm gets cash only once, when new shares are first sold. That first sale is the primary market. After it, the shares pass from one investor to the next on the secondary market. The firm gets nothing from those trades.

Business

Private Equity vs Venture Capital

People think private equity and venture capital are the same thing, and they are not. Venture capital buys a slice of a young firm that is growing fast. It may not make a profit yet. Private equity mostly buys the whole of an older firm that already makes money. It often uses a lot of borrowed cash, works to make the firm earn more, then sells it.

Business

Credit Rating vs Credit Score

People think a credit rating and a credit score are the same thing, and they are not. A credit score is a number for one person, used for a card, a car loan or a home loan. A credit rating is a letter grade, from AAA down to D, for a firm, a bond or a country. Both say how likely the borrower is to pay. They are given by different bodies to different borrowers.

BusinessGeneralcross-domain

Line of Credit vs Business Loan

People think a line of credit is just another loan, and it works quite differently. A loan pays out the whole sum on day one, and you pay interest on all of it. A line of credit sets a limit, and you take only what you need, when you need it. You pay interest only on what you have taken. When you pay it back, you can borrow it again.

Business

Stock vs Bond

People think stocks and bonds are two kinds of share, and they are not. A stock makes you a part owner of the firm. You may get dividends, and the price can rise a lot or fall to zero. A bond is a loan you make to a firm or a government. It pays set interest and gives your money back on a set date. Bond holders are paid before owners if the firm fails.

BusinessGeneralcross-domain

Enterprise Value vs Market Cap

People think market cap is the price of buying a whole firm, and it is not. Market cap is only the value of all its shares. A buyer also takes over the firm's debts, and gets its cash. Enterprise value adds the debt and takes away the cash. Two firms with the same market cap can cost a buyer very different sums.

Business

Cost of Debt vs Cost of Equity

People think owners' money is cheaper than a loan because no interest is due, and it is the other way round. The cost of debt is the interest rate, and tax often makes it lower still. The cost of equity is the return owners expect. Owners are paid last and take more risk, so they want more back. Equity is the dearer of the two.

Business

DCF vs NPV

People think DCF and NPV are the same number, and they are close but not the same. A discounted cash flow adds up future cash, each year shrunk back to today. That gives what a firm or a project is worth. Net present value then takes away what it costs to buy or build. A positive NPV means the value is more than the price.

Business

DCF vs Comparable Company Analysis

People think a DCF and comparables are two ways to get the same answer, and they often disagree. A DCF looks inside the firm, at the cash it will make. Comparables look around, at what similar firms sell for today. A DCF can miss the mood of the market. Comparables can copy the market's mistakes. Good valuers do both and explain the gap.

Business

GAAP vs IFRS

People think GAAP and IFRS are two names for the same rules, and they are not. GAAP is the rulebook in the United States, written by the FASB. IFRS is written by the IASB, and most other countries require it for firms on their stock markets. The two agree on a lot. But the same firm can show a different profit under each, so check which one a report uses.

Business

Direct Cost vs Variable Cost

People think a direct cost and a variable cost are the same thing, and they are not. Direct means you can trace the cost to one product, like the wood in a chair. Variable means the cost rises when you make more. Most direct costs are also variable, but not all. A machine used only for chairs is a direct cost that stays the same however many you make.

Business

Shareholders' Equity vs Equity

People think shareholders' equity is what the owners' shares are worth, and it is not. Shareholders' equity is a number in the books: what the firm owns, less what it owes. Equity, in the sense of owning shares, is a part of the firm that someone holds. Its price is set by buyers and sellers. The two numbers can be very far apart.

Business

Journal vs General Ledger

People think the journal and the general ledger are the same book, and they are not. The journal lists each money event by date, with its debit and credit. The ledger sorts the same records by account. So each account shows its own running total. Entries go into the journal first. Then they are copied into the ledger.

Business

OKR vs MBO

People think OKRs and management by objectives are the same thing with a new name, and they are close but not the same. Both set clear goals and check results. With management by objectives, goals are often set once a year and checked at the yearly review. With OKRs, goals are open for everyone to see, set every quarter, and meant to stretch. A score of 0.7 can be a success.

Business

Balanced Scorecard vs OKR

People think a balanced scorecard and OKRs do the same job, and they do not. A balanced scorecard is a steady set of measures from four views, money, customers, work and skills. It shows how the firm is doing over years. OKRs are a few bold goals for the next quarter, which change each time. Many firms use the scorecard to watch health and OKRs to push for change.

Business

CSR vs ESG

People think ESG is just a new name for CSR, and it is not. CSR is what a company chooses to do for people and the planet. ESG is how investors score it from outside, on the planet, people and how it is run. A firm does CSR. Others rate its ESG.

Business

Internal Control vs Internal Audit

People think internal audit is the control, and it is not. A control is a check built into the daily work, like two people to sign a payment. Managers run it. Internal audit comes later and tests whether those checks really work. Then it tells the board.

Business

KYC vs AML

People use KYC and AML as if they mean the same, and they do not. KYC is one step: proving who a customer is before the account opens. AML is the whole set of laws and checks against crime money. It also covers watching payments and reporting the odd ones.

Business

Shareholder Primacy vs Stakeholder Capitalism

People think stakeholder capitalism means a firm stops caring about profit, and it does not. Shareholder primacy says the owners come first. Profit for them is the goal. Stakeholder capitalism says the firm should also weigh its staff, buyers, the firms that sell to it, and the town. Profit still counts in both. They split when one group must lose so another can gain.

Business

Compliance vs Business Ethics

People think a company that follows the rules is an ethical one, and it is not always so. Compliance asks if you kept the law and the rules. Ethics asks if what you did was right. Plenty of harm is legal, so a firm can pass every check and still lose trust.

Business

Offshoring vs Outsourcing

People think offshoring and outsourcing are the same, and they are not. Outsourcing is about who does the work: a different firm. Offshoring is about where: another country. A firm can open its own office abroad. That is offshoring, but not outsourcing. It can also hire a firm down the road. That is outsourcing at home.

Business

Economic Sanctions vs Export Controls

People think sanctions and export controls are one rule, and they are not. Sanctions are about who: a country, firm or person you may not deal with. Export controls are about what: goods and tech you may not ship without a license, whoever buys them. A firm must check both before an order leaves.

Business

BATNA vs Reservation Price

People think a BATNA is your walk-away price, and it is not. Your BATNA is a plan: what you will do if there is no deal, like take another job offer. Your reservation price is a number you work out from that plan. It is the worst deal that still beats the plan.

Business

Distributive vs Integrative Negotiation

People think win-win means splitting the difference, and it does not. Distributive talks fight over one fixed pie, so one side's gain is the other's loss. Integrative talks trade across several issues to make the pie bigger first. Splitting down the middle is still a fixed pie.

Business

Mediation vs Arbitration

People use mediation and arbitration as if they mean the same, and they do not. A mediator helps the two sides reach their own deal and decides nothing. An arbitrator hears both sides and makes a ruling they must follow. In mediation you keep control. In arbitration you hand it over.

Business

Concession vs Logrolling

People think logrolling is just making concessions, and it is not. A concession gives something away to move closer to a deal. Logrolling trades across issues: you give up what matters less to you to get what matters more. A concession splits the value. Logrolling makes more of it.

Business

Résumé vs Cover Letter

People think a cover letter is the résumé written out in full sentences, and it is not. The résumé lists what you did, with results, for any reader. The cover letter says why you want this one job and why you fit it. Copy the résumé into the letter and the letter adds nothing.

Business

Engineering Manager vs Tech Lead

People think the tech lead is the team's boss, and it is not so. The tech lead still codes and guides the design. The engineering manager looks after the people: hiring, growth and reviews. One leads the work. The other leads the team that does it.

Business

Career Path vs Career Ladder

People use career path and career ladder as if they mean the same, and they do not. A ladder is the firm's written list of levels and what each one expects. A path is your own route through jobs, which may climb one ladder, cross to another or leave the firm.

Business

Business Analyst vs Data Analyst

People think a business analyst is someone who analyses data, and that is mostly wrong. A business analyst finds out what a firm needs. Then they write it down so a team can build it. A data analyst digs into numbers to answer questions. One shapes what gets built. The other explains what happened.

BusinessDatacross-domain

Photogrammetry vs Gaussian Splatting

People think a splat scene is a 3D model, like a photo scan, and it is not. Photogrammetry gives a mesh of faces you can edit, light and use in a game. Gaussian splatting gives millions of soft blobs with no faces at all. Splats show glass, shine and hair far better. But a mesh is much easier to change.

Design

glTF vs USDZ

People think any 3D file opens anywhere, and it does not. glTF is the open format for web pages, Android and most game engines. USDZ is the one Apple devices open for AR with a single tap. A shop that wants AR on every phone ships the same model in both files.

Design

Material vs Shader

People think a material and a shader are the same thing, and they are not. The shader is the small program that works out the color of each pixel. The material is the set of values and images fed into it. Many materials can share one shader and still look nothing alike.

Design

Product Vision vs Design Vision

People think the design vision and the product vision are the same thing, and they are not. The product vision is a sentence or two about the future the product wants to make. The design vision is a picture of it: screens or a short film of how it could look and feel. The words come first, and the picture makes them real.

BusinessDesigncross-domain

Design Sprint vs Design Workshop

People think any team session is a design sprint, and it is not. A sprint is a fixed plan of five full days that ends with a test on five users. A design workshop is one session of a few hours or a day. It ends in choices, often about what to prototype next.

Design

Design Spec vs Design Rationale

People think the spec says why a screen looks the way it does, and it does not. A spec says what to build: sizes, states and behavior. The rationale says why each choice won and what was tried instead. Keep both, so nobody undoes a good choice later.

Design

Responsive vs Adaptive Design

People think adaptive and responsive design are two names for one thing, and they are not. Responsive design is one layout that stretches and reflows to fit any width. Adaptive design is a few fixed layouts, and the site serves the one that fits best. Between those sizes, an adaptive layout does not stretch.

Design

Website Builder vs CMS

People think a website builder and a CMS are the same tool, and they are not. A website builder is a hosted, no-code tool where you design and publish the whole site. A CMS is software for writing and storing the content, like posts and pages. Some tools do both, but you can run a CMS on your own server with your own design.

DesignDevelopmentcross-domain

Single-Page Website vs Single-Page App

People think a one-page website is a single-page app, and it is not. A single-page website is a design choice: all the content sits in one long scroll. A single-page app is a way to build a web app, where the page swaps its parts without loading a new page. An app built that way can have hundreds of screens.

DesignDevelopmentcross-domain

Homepage vs Landing Page

People think a landing page is just another name for the homepage, and it is not. The homepage is the front door for everyone, so it points to many places. A landing page is made for one campaign and asks for one action. Ads that send people to the homepage often lose them.

DesignMarketingcross-domain

Behance vs Dribbble

People think Behance and Dribbble are the same kind of site, and they are not. Behance shows whole projects as long stories, from the problem to the result. Dribbble shows single shots, one screen or one logo at a time. Hiring teams read a Behance project to see how you think. Clients scroll Dribbble to see your style.

Design

Design Manager vs Principal Designer

People think a principal designer runs a team, and they do not. A design manager leads people: hiring, growth and how the team works. A principal designer leads the work itself. They take on the hardest problems, and no one reports to them. Both sit high on the ladder, often at the same level and pay.

Design

Web Design vs Web Development

People think web design and web development are one job, and they are not. Web design plans how a site looks and works: the layout, the type and the paths between pages. Web development writes the code that makes it real and keeps it running. Many people do some of both, but they are two skills.

DesignDevelopmentcross-domain

Rack Server vs Blade Server

People often think a blade server is just a thin rack server. It is not. A rack server is a whole machine with its own power and fans, and it runs on its own. A blade has only a chip, memory and disks. It must sit in a chassis that gives it power, fans and network links. Take it out, and it cannot run.

Infrastructure

Redundant Power Supply vs UPS

People often think two power supplies keep a server on through a power cut. They do not. Two power supplies guard against one of them breaking. If the building loses power, both go dark at once. A UPS is a battery that carries the server through the cut. Most servers that must stay up have both.

Infrastructure

Root User vs Root Directory

People hear root and think of one thing. On Linux it names two. The root user is the account that may do anything. The root directory is the top folder, written /. The root user even has its own home folder, /root, which sits inside the root directory.

Infrastructure

chmod vs chown

People often mix up chmod and chown because the names are so close. chmod changes what can be done to a file: read, write or run. chown changes who the file belongs to. A permission denied error can need either one, so read ls -l first and see which part is wrong.

Infrastructure

sudo vs Root User

People think sudo is the root user, and it is not. Root is an account that can change anything on the machine. sudo is a command that lets a normal user act as root for one line, after typing their own password. Each use is written to a log, so you can see who did what. Commands typed while logged in as root leave no such line.

Infrastructure

iptables vs nftables

People often think iptables is the firewall on a new Linux server. Today it is a front door to nftables. The iptables command still works on Debian and Ubuntu. But it writes its rules into nftables. Mixing rules from both by hand can make them clash.

Infrastructure

SSH Config File vs sshd_config

People often think these are one file, because the names are so close. ~/.ssh/config lives on your own computer and sets up how you log in. sshd_config lives on the server and sets the rules for who may log in. Editing the wrong one changes nothing you meant to change.

Infrastructure

known_hosts vs authorized_keys

People often think both files hold the same keys. They face opposite ways. known_hosts is on your computer and lists the servers you trust. authorized_keys is on the server and lists the people it lets in. One proves the server to you, the other proves you to the server.

Infrastructure

SSH Host Key vs SSH Key Pair

People think the host key is their own SSH key, and it is not. The host key belongs to the server. It proves to your laptop that it reached the real server, and its copy lives in known_hosts. Your key pair belongs to you. It proves to the server who you are, and its public half lives in authorized_keys.

Infrastructure

SSH Key Passphrase vs SSH Password Login

People think a key passphrase is the same as a password login, and it is not. A password login sends a password to the server, and a bot can guess it there. A passphrase never leaves your laptop. It only unlocks your private key file, so a stolen copy of the file is no use.

Infrastructure

scp vs rsync

People often think rsync is just a faster scp, since both copy files to a server. Both go over SSH, but scp sends the whole file every time. rsync checks what is already there first, and sends only what changed. For one file once, scp is fine. For a big folder you copy again and again, rsync wins.

Infrastructure

SSH Tunnel vs VPN

People often think an SSH tunnel is a small VPN. It is not. A tunnel carries one port to one place, inside one SSH login. A VPN joins your whole computer to a private network, so every app goes through it. Use a tunnel for one service and a VPN for a network.

Infrastructure

RDP vs VNC

People often think VNC and RDP are two names for the same thing. Both show a far screen, but they work in different ways. RDP is built into Windows and opens its own desktop session. VNC copies whatever screen is there, on any system. VNC has no lock of its own, so it needs SSH or a VPN.

Infrastructure

VPS vs Dedicated Server

People often think a VPS is a small dedicated server. It is not a machine of its own. A VPS is one slice of a big machine, shared with other customers. A dedicated server is the whole machine, with nobody else on it. A VPS is ready in a minute and costs less. A dedicated server gives steady speed that no neighbor can take.

Infrastructure

Shared Hosting vs VPS

People often think shared hosting and a VPS are the same cheap plan. On shared hosting, the host runs the server and you only get a folder for your site. On a VPS, you get root access and run the whole server yourself. Shared hosting is easier. A VPS lets you install anything, and every update is yours to do.

Infrastructure

Colocation vs Dedicated Server

People often think colocation and a dedicated server are the same deal. In both, the server sits in someone else's data center. With colocation, you buy the server and bring it in. The center gives power, cooling and a network. With a dedicated server, the host owns the machine and rents it to you. Then a broken part is their job to fix.

Infrastructure

Colocation vs Bare Metal Server

People think colocation is a way to rent a server, and it is not. With colocation you buy the server and ship it in. The data center only rents you the space, the power and the link. With a bare metal server the host owns the machine, keeps it in its own building and fixes it when a part fails. You rent all of it.

Infrastructure

Cloud Firewall vs UFW

People often think the cloud firewall and UFW are one firewall. They sit in two places. A cloud firewall runs in the provider's network, in front of the server, and you set it in a web panel. UFW runs on the server itself. Traffic has to pass both, so a port must be open in each one to work.

Infrastructure

Web Console vs Rescue Mode

People think the web console and rescue mode are the same way back in, and they are not. The web console shows the screen of a server that still runs. You use it to log in when SSH is blocked. Rescue mode starts a spare system for a server that will not boot. You use it to fix the disk.

Infrastructure

MAE vs RMSE

People think RMSE is just another way to write the average miss, like MAE. It is not. MAE counts each miss by its size. RMSE squares each miss first, so one big miss pulls it up far more. When the two are far apart, a few wild misses are hiding in the data.

AI

BLEU vs ROUGE

People think BLEU and ROUGE are one score with two names, since both count shared words. They count from opposite sides. BLEU asks how much of the model's text is in the human version. ROUGE asks how much of the human version made it into the model's text. BLEU grew up with translation and ROUGE with summaries.

AI

Benchmark Saturation vs Benchmark Contamination

People think a test every model aces must have leaked into training. Sometimes it has, but not always. Saturation means the test got too easy for today's models. Contamination means the answers were in the training data. Both give high scores. Only one is a kind of cheat, and each needs its own fix.

AI

Perplexity (Metric) vs Perplexity

People think Perplexity the score and Perplexity the app are the same thing. They only share a name. One is a number that says how well a language model predicts text, and lower is better. The other is an AI search app that answers questions with links to its sources.

AI

NLU vs NLG

People think NLU and NLG are two names for one thing, since both deal with plain language. They point in opposite ways. NLU reads words and turns them into meaning a program can use, like a goal or a date. NLG starts from data or a goal and writes words for a person to read. A voice assistant does NLU to hear you and NLG to answer.

AI

Masked Language Model vs Next-Token Prediction

People think every language model learns by guessing the next word. Many do, but not all. A masked model hides words in the middle and guesses them from both sides, which makes it good at reading. A next-word model sees only the words before the gap, which is what lets it write. BERT learned the first way, and chat models learn the second.

AI

Topic Modeling vs Text Classification

People think topic modeling is just sorting text by topic. It is not. Text classification uses groups chosen ahead of time and learns them from labeled examples. Topic modeling has no groups and no labels. It finds themes from words that show up together, and a person names them after.

AI

Visual Grounding vs Object Detection

People think visual grounding is the same as object detection. Both draw boxes on a picture, but they answer different asks. Object detection finds every thing of a known kind, like every cup on the table. Visual grounding reads a phrase, such as the red cup on the left, and finds the one thing it points to.

AI

Wake Word vs Voice Activity Detection

People think a wake word and voice activity detection are one check, since both listen. A wake word listens for one set phrase and ignores all other talk. Voice activity detection does not care what is said. It only marks when someone starts and stops talking, so the device knows when its turn comes.

AI

AI Dubbing vs AI Lip Sync

People think AI dubbing and AI lip sync are one tool. Dubbing changes what you hear: it puts the speech into a new language, often in the same voice. Lip sync changes what you see: it moves the mouth to match a sound track. A full dub often uses both, but either one can be used alone.

AI

AI Avatar Video vs Deepfake

People think every AI avatar video is a deepfake. Both show a person saying words they never said on camera. What splits them is consent. An avatar is made with the person's okay, or from a face no one owns. It is shown as a digital host. A deepfake copies a real person without asking, often to fool people.

AI

Economic Buyer vs Sales Champion

People think the champion who loves your product inside the buyer's company can sign the deal, and most of the time they cannot. A champion pushes for you there and tells you how the company decides. The economic buyer holds the money and can say yes or stop it. A seller needs both. A deal with a keen champion and no economic buyer often stalls at the last step.

Business

Consultative Selling vs Solution Selling

People think consultative selling and solution selling are one method, and they are not quite. Solution selling starts from the buyer's problem and sells a fix for it. Consultative selling goes one step further. The seller acts as an adviser, asks and listens first, and may even tell the buyer to buy less. One sells the fix. The other sells trust.

Business

Customer Success Manager vs Account Manager

People think a customer success manager is just an account manager with a new name. The two jobs pull in different ways. An account manager owns the deal side: renewals, prices and selling more. A customer success manager owns the results: setup, use and value. Many firms have both. One asks "can we sell more?" The other asks "is it working for you?"

Business

Request for Proposal vs Sales Proposal

People think a request for proposal is a proposal, and it is the other way round. The buyer writes the request. It lists what they need and asks sellers to bid. The seller writes the proposal. It answers with what they will deliver, when and at what price. One is the question, and the other is the answer.

Business

Bookings vs Billings

People think bookings and billings are two words for the same sales number, and they are not. Bookings count the whole deal on the day it is signed. Billings count only what was put on an invoice in that period. A three-year deal at $12,000 a year books $36,000 at once, but bills $12,000 a year. Revenue is a third number again, earned month by month.

Business

Gross Revenue Retention vs Net Revenue Retention

People think gross and net revenue retention are the same check with different names, and they answer two questions. Gross retention asks how much of last year's money from these customers you kept. It only takes losses off, so it can never pass 100%. Net retention adds the upgrades back in, so it can pass 100%. A high net number can hide a leak that the gross number shows.

Business

Churn Rate vs Revenue Churn

People think a low churn rate means little money is leaving, and that is not always true. Churn rate counts customers, so each one weighs the same. Revenue churn counts dollars, so a big customer weighs more than a small one. Losing one large customer can barely move the churn rate and still take a big bite out of revenue. Watch both side by side.

Business

Run Rate vs Annual Recurring Revenue

People think run rate and annual recurring revenue are the same number for the year. They are not. Run rate takes all recent sales, one-offs too, and stretches them to a year. Recurring revenue counts only what customers pay again and again on their plans. A big one-off job lifts the run rate for a month. It never moves recurring revenue.

Business

ERP vs CRM

People think an ERP and a CRM are two names for the same big business system, and they are not. A CRM looks outward, at customers: leads, deals, calls and the sales pipeline. An ERP looks inward, at running the firm: money, stock, buying and making goods. The CRM wins the order, and the ERP fills it and books it. Most firms link the two.

Business

Management Information System vs Decision Support System

People think an MIS and a DSS are the same reporting tool, and they do different jobs. An MIS sends managers the same reports every week, so they can see how things went. A DSS is used when one big choice is on the table. It lets a person ask "what if" and compare the likely results. One looks back at what happened. The other tests what could happen next.

Business

IT Governance vs IT Service Management

People think IT governance and IT service management are one job done by one team. They sit at two levels. Governance decides. It says who may spend on tech, which projects come first and what risks are fine. Service management delivers. It keeps email, laptops and apps working each day. Leaders run governance, and the IT team runs service management inside those rules.

Business

Herzberg's Two-Factor Theory vs Maslow's Hierarchy of Needs

People think Herzberg's theory is just Maslow's pyramid of needs used at work, and the two make different claims. Maslow puts needs on one ladder, and you climb to the next step once the one below is met. Herzberg splits work into two scales. Pay and conditions only stop people being upset. Praise, growth and a job well done are what make them keen.

BusinessGeneralcross-domain

Prehistory vs Stone Age

People use prehistory and the Stone Age as if they were the same time, and they are not. The Stone Age is about tools. It lasts until people work metal. Prehistory is about writing. It lasts until writing reaches a place. In Britain, the Bronze Age and the Iron Age were still prehistory, since no one there wrote yet.

General

Renaissance vs Enlightenment

People mix up the Renaissance and the Enlightenment as one age of new ideas, and they are not. The Renaissance began in Italy in the 1300s. It looked back to old Greece and Rome for art and learning. The Enlightenment came in the 1700s and looked forward. Its thinkers asked for reason, facts and rights, and for limits on kings.

General

Virus vs Bacteria

People call every germ a bug and treat viruses and bacteria as one thing, and they are not. A bacterium is a whole living cell. It can split in two on its own. A virus is far smaller and can only copy itself inside one of your cells. That is why an antibiotic can cure strep throat but does nothing for a cold.

General

Epidemic vs Pandemic

People use epidemic and pandemic as if they meant the same, and they do not. An epidemic is far more cases than usual in one place, such as a city or a country. A pandemic is one that has spread across many countries. The words say how far a disease has spread, not how deadly it is.

General

Acute vs Chronic Illness

People think acute means serious and chronic means mild, and that is not it. Acute means an illness came on fast and will not last long, like the flu. Chronic means it lasts a year or more and needs ongoing care, like asthma. A heart attack is acute, and mild hay fever can be chronic.

General

Antibody vs Antibiotic

People mix up antibodies and antibiotics, since the names sound alike, and they are not the same. Your body makes antibodies. Each one locks onto one germ. An antibiotic is a pill you take to kill bacteria. It does nothing to a virus, but a vaccine can teach your body to make antibodies for one.

General

Internet vs World Wide Web

People say the internet and the web as if they were one thing, and they are not. The internet is the network itself: the cables, radio links and rules that join devices. The web is one service that runs on it, made of pages joined by links. Email, video calls and games also use the internet without using the web.

General

Deep Web vs Dark Web

People mix up the deep web and the dark web, and they are very different. The deep web is every page a search engine cannot show, like your inbox and your bank account. It is huge and mostly normal. The dark web is a small hidden part of it that you can only reach with special software such as Tor.

General

Wi-Fi vs Internet

People think full Wi-Fi bars mean they are on the internet, and that is not always so. Wi-Fi is only the short radio link from your device to the router. The router then needs its own line to your provider. If that line is down, the Wi-Fi still works but nothing loads.

General

URL vs Hyperlink

People use link and URL as if they were the same, and they are not. A URL is an address, the text that says where a page lives. A link is the thing you click, such as an underlined word or a button. Every link has a URL behind it, and the words you see may not match it.

General

Private Browsing vs VPN

People think a private window hides them the way a VPN does, and it does not. Private browsing only stops your own device from keeping your history. A VPN sends your traffic through a locked tunnel. Then your local network and your provider cannot see which sites you visit, though the VPN company can.

GeneralInfrastructurecross-domain

Direct vs Indirect Competitor

People think only a shop like yours is a competitor, and it is not. A direct competitor sells the same kind of thing to the same buyers, like the coffee shop next door. An indirect one sells something else that meets the same need, like a bakery that also serves coffee. Both can take your buyers.

Business

Competitive Analysis vs Industry Analysis

People think a study of your rivals and a study of the whole industry you work in are the same job, and they are not. The first looks at the players: what each rival sells and what it charges. The second looks at the whole field. It asks how big the field is, how fast it grows, and who holds the power.

Business

Market Map vs Perceptual Map

People think a market map and a perceptual map are one chart, and they are not. A market map sorts the firms in a market into groups by who they sell to. A perceptual map plots brands on two lines, like cheap to dear, by what buyers think of them.

BusinessMarketingcross-domain

Competitive Intelligence vs Business Intelligence

People think these two are the same work, and they are not. One turns a firm's own data, like its sales, into reports. That is business intelligence. The other looks out at rivals. It gathers facts from public places, like their prices and job ads.

BusinessDatacross-domain

Synchronous vs Asynchronous Communication

People think work talk has to happen live, and it does not. Synchronous means everyone takes part at the same time, like a meeting or a call. Asynchronous means each person reads and replies when they are free, like email or a shared doc. Live talk is fast but costs everyone's time at once.

Business

Pyramid Principle vs BLUF

People think BLUF and the pyramid principle are the same rule, and they are not. BLUF says one thing: put the main point in your first line. The pyramid principle shapes the whole document. The point goes on top, a few reasons sit under it, and the facts sit under each reason.

Business

Slide Deck vs Pitch Deck

People think any set of slides is a pitch deck, and it is not. A slide deck is any set of slides for a talk or a report. A pitch deck is one kind, made to win money from investors. It covers the problem, the product, the market, the team and how much the founders want to raise.

Business

Meeting Agenda vs Meeting Minutes

People think the agenda and the minutes are one document, and they are not. The agenda comes before the meeting. It lists the topics, the times and what must be decided. The minutes come after. They record what was decided, who will do what, and by when.

Business

Vertical vs Horizontal Market

People think these two words are about size, and they are not. A vertical market is one industry served in depth. Software made only for dentists is sold into one. A horizontal market is one need that every industry shares, like payroll. The first has fewer buyers who fit well. The second has many buyers and many rivals.

Business

Public Sector vs Private Sector

People think a company with shares on a stock market is in the public sector, and it is not. The public sector is what the government owns and runs, paid for by taxes. The private sector is owned by people and firms and run for profit. A listed company is private, because people own its shares.

Business

Industry vs Economic Sector

People think an industry and a sector are the same thing, and they are not. An industry is a group of firms that make or sell the same kind of thing, like airlines. A sector is a bigger group that holds many industries. In GICS, airlines sit inside a sector called Industrials.

Business

NAICS vs GICS

People think NAICS and GICS are two names for one code list, and they are not. NAICS sorts every business in Canada, Mexico and the US, so governments can count jobs and output. GICS sorts listed companies for investors, from 11 sectors down to 163 sub-industries. A firm can have a code in both.

Business

Healthtech vs Biotech

People think healthtech and biotech are the same field, and they are not. Healthtech is mostly software and devices, like apps and video visits with a doctor. Biotech uses living cells to make things, like new medicines. A health app can launch in months. A new medicine often takes ten years of tests.

Business

MBA vs Executive MBA

People think an executive MBA is just a shorter or easier version of the regular MBA, and it is not. Both teach the same core subjects. A regular MBA is often full time, for people with a few years of work. An executive MBA is for senior managers. Classes come on weekends or in blocks, so they keep their jobs.

Business

MBA vs Master in Management

People think the MBA and the Master in Management are the same degree, and they are not. Both teach how a business runs. The Master in Management is for new graduates with little or no work. The MBA expects a few years of work first, and it usually costs more.

Business

AACSB vs AMBA Accreditation

People think every business school seal checks the same thing, and they do not. AACSB checks the whole school: its teachers, research and how students learn. AMBA checks single programmes, such as one MBA. A school can hold AMBA for its MBA while its other degrees have no AMBA seal.

Business

No-Code vs Low-Code

No-code tools give you only blocks and menus. Low-code tools give you blocks too, but let you type code where the blocks run out. So low-code can build more, and it needs someone who can code.

Development

Workflow Automation vs RPA

A workflow tool joins apps through their APIs, behind the scenes. An RPA robot works on the screen, clicking and typing like a person. Use RPA only when an app has no API, because a moved button can break it.

Development

Web Scraping vs Browser Automation

People think web scraping and browser automation are one job, and they are not. Scraping reads a page and saves the data it finds, like prices. Browser automation acts on a page. It clicks, types and sends forms. The same tools can do both, but only automation changes anything on the site.

Development

DAG vs Tree

A tree gives every node just one parent, but a DAG lets a step wait on two or more steps. Both have no loops. Only the DAG can join paths back together.

Development

App Integration vs Data Integration

App integration links the apps people work in. A change in one shows up in the other right away. Data integration pulls data from many apps into one place. Then people can compare and study it.

DevelopmentDatacross-domain

Git Revert vs Git Reset

Git revert adds a new commit that undoes an old one, so the history stays. Git reset moves the branch back, so later commits drop off it. Use revert on shared branches, and reset only on your own work.

Development

Monorepo vs Monolith

A monorepo is one place that stores the code of many projects. A monolith is one app that runs as a single program. You can keep many small apps in one monorepo, or one monolith on its own.

Development

Black-Box Testing vs White-Box Testing

Black-box testing checks what a program does from the outside, with no look at its code. White-box testing reads the code and tests each path through it. Most teams use both, for different jobs.

Development

TDD vs BDD

Test-driven development starts with a failing test in code, then the code that makes it pass. Behavior-driven development starts one step before that. The team writes plain Given, When, Then lines first, and those lines run as tests too.

Development

Mock vs Mock Server

A mock is a fake object inside one test, in the same program. A mock server is a fake API that runs on its own, over the network. Many tests and apps can call the same mock server.

Development

Integration Test vs Contract Testing

An integration test runs two or more real parts together to see that they work. A contract test checks each side alone against a shared file of requests and answers. It does not need the other side to be running.

Development

Terraform vs Ansible

Terraform makes the cloud parts, like servers and networks. Ansible sets up what runs on a server, like its programs and files. Many teams use Terraform to make the servers. Then they use Ansible to set them up.

Development

Staging Environment vs Preview Deployment

A staging environment is one shared copy of the live site, where the next release is tested. A preview deployment is a short-lived copy made for one pull request. A team may have one staging site but many previews at once.

Development

Unified Memory vs Video Memory (VRAM)

People think unified memory is just video memory with a new name, and it is not. Video memory sits on a graphics card for the GPU alone. Unified memory is one pool that the CPU and GPU share. So a 64 GB pool is not 64 GB for the GPU, since the system and apps use it too.

Infrastructure

Hard Disk Drive (HDD) vs Solid-State Drive (SSD)

People think an SSD is just a faster kind of hard drive, and it is not. A hard drive spins a disc and moves an arm to each piece of data. An SSD keeps data on chips with nothing to move. That is why an SSD wins by far on many small reads, while a hard drive still wins on price for each terabyte.

Infrastructure

RAM vs SSD

People call both of them memory, and they do different jobs. RAM holds what the computer is working on right now, and it empties when the power goes off. An SSD keeps your files when the power is off. RAM is many times faster, and it costs far more for each gigabyte. More RAM lets you run more at once. A bigger SSD only lets you keep more.

Infrastructure

NVMe vs SATA

People think NVMe is a kind of SSD, and it is not. It is the link the drive talks over, and SATA is the older one. The same flash chips on SATA stop near 600 megabytes a second. On four PCIe lanes with NVMe, they can go more than ten times as fast. Some M.2 sticks are SATA inside, so check the link, not the shape.

Infrastructure

Container Volume vs Bind Mount

People think a bind mount is just another volume, and it is not. A bind mount shows a host folder you chose inside the box. A volume is storage the container tool makes and looks after. Both can use the same -v flag, which is why they get mixed up. Use a bind mount to edit code, and a volume to keep a database's files.

Infrastructure

docker run vs docker exec

People think docker run takes you back into a container, and it does not. docker run always makes a new container from an image. docker exec runs a command in one that is already running. Type docker run twice and you have two containers, each with its own files.

Infrastructure

Linux Namespace vs Kubernetes Namespace

People think a Kubernetes namespace is a Linux namespace, and it is not. A Linux namespace is a wall inside one computer. It hides the other programs from a container. A Kubernetes namespace is a label. It splits a cluster into groups, such as one per team. By itself, it walls nothing off.

Infrastructure

Synthetic Monitoring vs Real User Monitoring (RUM)

People think a scripted check shows what users really see, and it does not. A synthetic check is a robot that visits on a timer, from a fast, steady place. Real user monitoring times the pages that real people load, on their own phones and networks. The robot finds breaks at night. Real users show the true speed.

Infrastructure

Uptime Monitoring vs Heartbeat Monitoring

People think an uptime check will notice a failed nightly job, and it will not. An uptime check knocks on a service and waits for an answer. A heartbeat monitor waits for the job to knock on its door. A backup that never ran has no door to knock on, so only the missing heartbeat shows it.

Infrastructure

Redundancy vs Failover

People think having a spare means the service stays up, and it does not on its own. Redundancy is owning the spare. Failover is the switch that moves the work to it when the main part dies. A spare that nothing switches to sits idle while the site is down, so the switch has to be built and tested too.

Infrastructure

Graceful Degradation vs Fault Tolerance

People think any site that keeps running through a broken part is fault tolerant, but it may only be degrading. Fault tolerance keeps every feature working, because a spare takes over at once. Graceful degradation keeps only the key jobs, such as paying. It switches the extras off. The first costs more. The second is often enough.

Infrastructure

MTBF vs MTTR

People read the two numbers as one score for reliability, and they measure different things. MTBF is how long a system runs, on average, before it breaks. MTTR is how long it takes to bring it back. A service can break often but recover in seconds, or break rarely and stay down for a day.

Infrastructure

LUT vs Color Grading

People think a LUT and color grading are the same thing, and they are not. A LUT is a file. It swaps each color for a fixed new one, the same way on every shot. Color grading is the craft of shaping the look by eye, shot by shot. A LUT can be one step in a grade, or a quick start. It cannot see that one shot is too dark and the next too bright.

Design

Rive vs Lottie

People think Rive and Lottie are two names for the same kind of file, and they are not. A Lottie file is an animation made in another app and played back the same way each time. Rive has its own editor, and its files can react. A state machine inside the file changes the move when someone taps, hovers or sends new data. Pick Lottie for a set piece that plays, and Rive for art that answers the user.

Design

Frame-by-Frame Animation vs Tweening

People think every cartoon frame is drawn by hand, and many are not. In frame-by-frame animation, an artist draws or poses each frame. In tweening, the artist sets a few key poses and the computer fills in the frames between them. Frame by frame gives full control and costs many hours. Tweening is fast and smooth, but it can look stiff when shapes need to change.

Design

Final Cut Pro vs Adobe Premiere

People think Final Cut Pro and Adobe Premiere are the same video editor with a different logo, and they are not. Final Cut Pro is made by Apple and runs only on a Mac or an iPad. Its timeline closes gaps by itself. Premiere is made by Adobe and runs on Mac and Windows. It works closely with other Adobe apps. After Effects is one of them. Teams that share files across both kinds of computer often pick Premiere for that.

Design

Moderated vs Unmoderated Testing

People think a test with no host is the same test, only cheaper, and it is not. In a moderated test, a host watches live and can ask why. In an unmoderated test, the user works alone and a tool records the screen. Tests with no host reach many people fast. Only a moderated test lets you ask a follow-up question in the moment.

Design

Service Blueprint vs Customer Journey Map

People think a service blueprint is a journey map drawn bigger, and it is not. A journey map follows the customer and shows how each step feels. A service blueprint keeps that row on top, then adds the rows below it. Those rows show what staff do, what happens out of sight and which systems help. The journey map shows where it hurts. The blueprint shows why.

Design

Stakeholder Interview vs User Interview

People think a stakeholder interview is a kind of user research, and it is not. It is a talk with people inside the business. It asks about goals, limits and how success is judged. A user interview is a talk with the people who use the product. It asks about their needs and habits. The first tells you what the business wants. Only the second tells you what users really do.

Design

Service Design vs UX Design

People think service design is UX design with a bigger name, and it is not. UX design shapes how a person uses one product, like an app or a website. Service design shapes the whole service around it. That includes shops, phone calls, staff and the work out of sight. A clinic can have a great booking app and still keep patients waiting. Fixing that wait is service design.

Design

Choropleth Map vs Heatmap

People call any colored map a heatmap, and a choropleth map is not one. A choropleth map shades real areas, like states or countries, by a number. A heatmap shades the cells of a grid, like days and hours, and needs no map at all. Some tools also call a blur of dots on a map a heatmap. If the shapes are real borders, it is a choropleth.

Design

Choropleth vs Proportional Symbol Map

People think every map of numbers should be shaded, and it should not. A choropleth map shades each area. So it suits rates, like the share of homes with fast internet. A proportional symbol map puts a sized circle on each place. So it suits totals, like the number of people. Shading totals makes big, empty areas look important. Circles keep a small, crowded city from vanishing.

Design

Grouped vs Stacked Bar Chart

People think grouped and stacked bars show the same thing, and they do not. A grouped bar chart puts the bars for each series side by side, so every bar starts at zero. That makes the series easy to compare. A stacked bar chart piles the series into one bar, so the total is easy to read. Only the bottom part of a stack shares a baseline, so the middle parts are hard to compare.

Design

Datawrapper vs Flourish

People think Datawrapper and Flourish are the same kind of chart app, and they are not. Datawrapper turns a pasted table into a clean, still chart, map or table, fast. Flourish starts from templates and is built for motion, like racing bars and stories that change as you scroll. Pick Datawrapper for a clear chart in a news story. Pick Flourish when the change over time is the point.

Design

Graphic Design vs Visual Design

People use graphic design and visual design as if they were one job, and they are not quite. Graphic design is the wide field. It makes posters, logos, packages, signs and ads, on paper and on screens. Visual design is the look of an app or a website. It works on top of a planned layout, and it also shows people what they can tap.

Design

Margin vs Gutter

People call any empty strip on a page a margin, and the gutter is not one. A margin is the space between the content and the edge of the page. A gutter is the space between two columns, inside the content. Margins frame the whole page. Gutters keep the columns apart, so the text in one does not run into the next.

Design

Rhythm vs Repetition

People think rhythm is just repetition, and it is a little more. Repetition means using the same look again, like one color on every heading. Rhythm adds a beat. The same part comes back with a set gap, or it grows a bit each time. You can repeat a color with no rhythm at all, but rhythm always needs repetition.

Design

Modernism vs Minimalism

People think minimalism and modernism are the same style, and they are not. Modernism began in the early 1900s. It dropped extra decoration, so each shape came from its job. Minimalism goes further. It keeps as few parts as it can, with lots of empty space. A modernist poster can still be busy, but a minimalist one never is.

Design

Sensor vs Actuator

People think a sensor and an actuator are two names for a robot part, and they are not. A sensor takes in: it measures light, sound or distance and turns it into numbers. An actuator gives out: it turns a command into a push, a turn or a grip. Every robot loop runs from one to the other. Sensors feel, the computer decides, and actuators move.

AI

Object Detection vs Object Tracking

People think tracking is just detection run on every frame, and it is not. Detection finds the cars in one frame and forgets them. Tracking links each box to the one before, so car 7 stays car 7. Only then can you count cars, measure speed or guess where one is going next.

AI

Digital Twin vs Robot Simulator

People think a digital twin is just another word for a simulation, and it is not. A robot simulator is a made-up world. You can test any robot you like in it. A digital twin copies one real machine. Live data from that machine keeps the copy in step. That live link is the whole difference.

AI

Imitation Learning vs Reinforcement Learning

People think robots learn every skill by trial and error, and they do not. In reinforcement learning, the robot tries things and learns from a score. In imitation learning, a person shows the task and the robot copies it. Copying is quicker to start. But the robot only gets as good as the examples it saw.

AI

Microsoft Copilot vs Microsoft 365 Copilot

People think Microsoft Copilot and Microsoft 365 Copilot are the same thing, and they are not. Both share a name and a chat box. The free Copilot answers from the web and from files you hand it. Microsoft 365 Copilot is a paid add-on for work accounts. It can read your work mail, files and meetings on its own, inside Word, Excel and Teams.

AI

Gemini CLI vs Google Antigravity

People think Gemini CLI is how anyone codes with Gemini in a terminal, and it is not. Since June 2026, free and home users get the Antigravity CLI instead. Gemini CLI stays for teams that pay for Gemini Code Assist. Pick by your plan, not by the name.

AI

ComfyUI vs Stable Diffusion

People think ComfyUI is an AI model, and it is not. ComfyUI is the app, and Stable Diffusion is one of the models it runs. The model makes the picture. ComfyUI lets you wire up the steps around it, like the prompt and the size. You download the two parts on their own.

AI

Impostor Syndrome vs Dunning-Kruger Effect

People think impostor syndrome is the Dunning-Kruger effect upside down, with skilled people rating themselves too low, and it is not. The Dunning-Kruger effect is about guesses: on a test, people with the lowest scores guess their score far too high. Impostor syndrome is a feeling about real success. The person knows their results are good, but feels they got there by luck and fears being found out. One is a wrong number. The other is a doubt that good numbers do not fix.

General

Conformity vs Bandwagon Effect

People think conformity and the bandwagon effect are the same thing, and they are not. Conformity is giving in to the people in front of you. You do it even when you think they are wrong. The bandwagon effect is picking something up because it is popular. Often nobody is even watching. In one, you hide what you think. In the other, the crowd changes what you think.

General

Status Quo Bias vs Sunk Cost Fallacy

People think status quo bias is the sunk cost fallacy, and it is not. The sunk cost fallacy needs a cost in the past: you stay because of what you already spent. Status quo bias needs no cost at all. You stay because change feels risky and the current choice is the default. A free plan you never chose can hold you just as tightly. For sunk costs, ignore what is spent. For the status quo, compare staying with switching as if you were choosing fresh.

General

Misinformation vs Disinformation

People use misinformation and disinformation as if they mean the same thing, and they do not. Both are false. The difference is what the sharer knows. Misinformation is passed on by someone who believes it, like an aunt sharing an old storm warning. Disinformation is made or spread by someone who knows it is false and wants to fool you. The same post can be both, depending on who shares it.

General

Deductive vs Inductive Reasoning

People think inductive reasoning is just a sloppy kind of deductive reasoning, and it is not. Deduction starts from a rule and works down to one case. If the rule is true, the answer must be true. Induction starts from many cases and works up to a rule. The rule is only likely, and one new case can break it. Math proofs use deduction. Most of science and daily life uses induction.

General

Appeal to Popularity vs Bandwagon Effect

People think the bandwagon fallacy and the bandwagon effect are the same thing, and they are not. The fallacy is a broken step in an argument. It says a claim is true because many people believe it. The effect is a pull on what we do. We join in because others have. You find the fallacy in what someone says. You find the effect in how people act, even when nobody argues at all.

General

Subculture vs Counterculture

People think a subculture and a counterculture are the same thing, and they are not. A subculture adds its own style inside the main culture, like skaters or gamers, and still follows its big rules. A counterculture pushes against those rules, like the hippies of the 1960s did. One adds to the culture around it. The other wants to change it.

General

Tradition vs Ritual

People use tradition and ritual as if they mean the same thing, and they do not. A tradition is anything handed down from one generation to the next, like a dish, a song or a belief. A ritual is a fixed set of steps done the same way, because the steps carry meaning. Many traditions include rituals, like a wedding. But a ritual can be brand new, like your own morning coffee.

General

Gambler's Fallacy vs Law of Large Numbers

People think the gambler's fallacy is just the law of large numbers at work, and it is not. The law says the average settles over many, many tries. It does this because new tries swamp the old ones, not because a streak gets paid back. The gambler's fallacy says the next try must make up for a streak. A fair coin has no memory, so the next flip is still a coin toss.

General

Regression to the Mean vs Regression Analysis

People think regression to the mean is a step in regression analysis, and it is not. Regression analysis is a method. It fits a line to data to see how one thing moves with another. Regression to the mean is an effect. A very high or low score tends to be followed by one closer to normal, because the luck in it does not repeat. One is a tool. The other is a trap that a fair test helps you spot.

General

Proof of Work vs Proof of Stake

People think proof of work and proof of stake are just two names for crypto mining. Only proof of work is mining. There, computers race to solve a puzzle, and the cost is power. In proof of stake, people lock up coins, and the network picks one of them to add each block. A cheat loses power bills in one and locked coins in the other.

Development

Cryptocurrency vs Crypto Token

People call every crypto asset a coin, and a token is not one. A coin belongs to its own blockchain, like bitcoin on Bitcoin or ether on Ethereum. It pays the fees that keep that chain running. A token is made by a smart contract on a chain that already exists. Anyone can make one in minutes, so ask what backs it.

Development

Blockchain vs Database

People think a blockchain is just a better database, and it is not. A database is run by one owner, who can fix a mistake or delete a row. A blockchain is copied on many computers, and no one can change an old record. That makes it slower and dearer. Use one only when strangers must share a record and trust no single keeper.

Development

Arduino vs Raspberry Pi

People think an Arduino and a Raspberry Pi are two sizes of the same small computer, and they are not. An Arduino is a board built around one small chip. It runs one sketch, starts at once and sips power. A Raspberry Pi is a full computer that runs Linux, many programs and a desktop. Pick the Arduino to read a sensor, and the Pi to run a web server.

Development

I2C vs SPI

People think I2C and SPI are two names for one way to wire up a sensor. Both link one main chip to many parts, but in different ways. I2C uses two shared wires and calls each part by its address. SPI picks each part with its own select wire and moves data faster. Use I2C to save pins, and SPI for screens and memory cards.

Development

Call Graph vs Control Flow Graph

People think a call graph and a control flow graph are two names for the same map of a program, and they are not. A call graph is drawn between functions: each arrow means this one calls that one. A control flow graph is drawn inside one function: each arrow is a step, and it splits at each if and loop. Use the first to see how far a change spreads, and the second to find untested paths.

Development

Tree-sitter vs Language Server

People think Tree-sitter and a language server do the same job in an editor, and they do not. Tree-sitter reads the shape of one file, fast, and it is used to colour and fold code. A language server knows what names mean across the whole project. It powers go to definition, rename and error hints. Most editors use both, side by side.

Development

Strongly Typed vs Statically Typed

People think strongly typed and statically typed mean the same thing, and they do not. Static or dynamic is about when types are checked: before the program runs, or while it runs. Strong or weak is about how strict the check is: does it refuse to mix a word and a number, or quietly join them? Python is strong but dynamic. C is static but fairly weak.

Development

Markup Language vs Programming Language

People think HTML is a programming language, and it is not. A markup language only labels parts of a text, like this is a heading or this is a link. It cannot add numbers, repeat a step or make a choice. A programming language can do all three. On a web page, HTML says what is there and JavaScript makes it do things.

Development

Lua vs Luau

People think Luau is just another name for Lua, and it is not. Luau is based on Lua 5.1, and much simple Lua code runs in it unchanged. But Luau is made for Roblox games. It adds types you can choose to use, and it shuts off parts of Lua that would be unsafe in a game, such as reading files. A Lua script for another app may not run in Roblox.

Development

Parallel Tool Calls vs Parallel Agents

People think parallel tool calls and parallel agents are the same speed trick, and they are not. Parallel tool calls happen inside one agent's turn. The model asks for several tools at once, and your code runs them together. Parallel agents are whole separate workers, each with its own model and its own steps. One saves a wait. The other splits the job.

AI

Agent Reflection vs Evaluator-Optimizer Loop

People think reflection and an evaluator-optimizer loop are the same thing, and they are not. In reflection, one model reads its own work and fixes it. In the loop, the job is split in two. One model writes, and a second model judges it by clear rules. A second judge can catch what the writer keeps missing.

AI

Training-Serving Skew vs Data Drift

People think training-serving skew is just data drift, and it is not. Drift is the world changing after the model was trained, so the live data slowly moves away. Skew is your own system building the same input two ways, one for training and one for live use. Drift needs new data and a retrain. Skew needs a bug fix.

AI

Data Parallelism vs Model Parallelism

People think data parallelism and model parallelism are one trick with two names. They are not. In the first, each GPU holds the whole model, and the data is split. In the second, the model is cut into parts, and each GPU holds one part. The first is simpler. The second is for a model too big for one GPU.

AI

Shadow Deployment vs Canary Release

People think a shadow deployment and a canary release are the same careful way to launch a model, and they are not. In a canary, a few real users get the new version and see what it does. In a shadow deployment, no user sees the new model at all. It gets a copy of real requests, and its answers are only saved. A canary tests on people. A shadow tests in secret.

AIDevelopmentcross-domain

Tokens per Minute vs Provisioned Throughput

People think provisioned throughput is just a higher tokens per minute limit, and it is not. A tokens per minute limit is a cap on shared capacity, and you pay only for what you use. Provisioned throughput is capacity set aside for you alone. You pay for it by the hour or month, even on quiet days. One is a ceiling. The other is a room you rent.

AI

Zero Data Retention vs Data Residency

People think zero data retention and data residency are the same privacy promise. They are not. Zero data retention is about time. The provider keeps no copy once the reply is sent. Data residency is about place. Your data is only handled in a region you pick. A firm may need one, the other, or both.

AI

Serverless GPU vs Model as a Service

People think a serverless GPU is the same as model as a service, and it is not. With model as a service, you call a model the provider runs and pay per token. With a serverless GPU, you bring your own model, and the service runs it on GPUs that switch on when called. You pay for the busy seconds. One rents a model. The other rents the chips.

AI

Azure Machine Learning vs Microsoft Foundry

People think Azure Machine Learning and Microsoft Foundry are one tool with two names. They are not. Azure Machine Learning is for training and running your own models, such as a sales forecast. Microsoft Foundry is for building apps and agents on large models that others trained. Both come from Microsoft, so the names get mixed up.

AI

Lightroom vs Photoshop

People think Lightroom is a small Photoshop, and it is not. Lightroom sorts a big pile of photos and fixes each whole photo with sliders. It never changes the first file. Photoshop works on one picture at a time, down to single pixels. You use it to remove things, mix photos and add text.

Design

Selection vs Layer Mask

People think a selection and a mask are the same, and they are not. A selection is a dotted line that says where the next edit goes. It is gone when you click away. A layer mask is saved with the layer. Black on the mask hides parts of the layer and white shows them, and you can change it later.

Design

Affinity vs Creative Cloud

People think Affinity is a free Creative Cloud, and it is not. Creative Cloud is a paid plan that opens many separate Adobe apps, for photos, drawing, pages and video. Affinity is one free app with three parts, for drawing, photos and pages. Video is not one of its parts.

Design

Lettering vs Calligraphy

People think lettering and calligraphy are the same, and they are not. Calligraphy is writing. Each letter is made in one go with a pen or brush, and the tool makes the thick and thin parts. Lettering is drawing. The letters are sketched, fixed and drawn again until the word looks right, and it is made for one use.

Design

Naskh vs Nastaliq

People think Naskh and Nastaliq are two names for the same Persian writing, and they are not. Naskh sits on a flat line and is clear and even, so it is used for books, news and screens. Nastaliq is a flowing style where each word slopes down from right to left. It is loved for titles and poems.

Design

Bowl vs Counter

People think the bowl and the counter are the same part of a letter, and they are not. The bowl is the curved stroke that wraps around the round part of a b, d or o. The counter is the empty space inside it. A thick bowl can still hold a wide, open counter.

Design

Terminal vs Serif

People think every end of a stroke is a serif, and it is not. A serif is a small foot added at the end of a stroke, like the feet on the letters in a book font. A terminal is the end of a stroke with no foot added. It can be cut flat, cut on a slant, or end in a ball.

Design

Halftone vs Dithering

People think halftone and dithering are the same dot trick, and they are not. A halftone makes shades with dots of different sizes, big for dark and small for light, and it is how printed photos are made. Dithering keeps every dot the same size. It changes how many dots there are and where they sit, to fake colors a screen or a file does not have.

Design

ICC Profile vs Color Space

People think an ICC profile and a color space are the same thing, and they are not. A color space is the full set of colors and the way they are numbered, like sRGB. An ICC profile is a small file. It says how one device, or one color space, shows color. To move a color from one device to the next, the system reads two profiles.

Design

Symbol vs Icon

People think a symbol and an icon are the same, and they are not. An icon looks like the thing it means, like a trash can for delete. A symbol stands for an idea and may look nothing like it, like a heart for love. You understand an icon by looking at it. You learn a symbol from the people around you.

Design

Stereotype vs Prejudice

People mix up a stereotype and prejudice, but they are not one thing. A stereotype is a belief, such as thinking all people in a group are the same. Prejudice is a feeling, often dislike, about a person because of their group. You can hold one without the other. The unfair act that can follow is a third thing, called discrimination.

General

Short-Term Memory vs Working Memory

People use short-term memory and working memory as if they were the same, and they are not quite. Short-term memory is the small store that holds a few things for a few seconds. Working memory holds them and also works on them, as when you add numbers in your head. One is a shelf, and the other is a desk.

General

Groupthink vs Conformity

People think groupthink is just conformity, and it is not. Conformity is one person going along with a group. They give the same wrong answer as the rest. Groupthink is a whole group that makes a bad choice. Nobody asks the hard question, because nobody wants a fight. One is a person giving in. The other is a team deciding badly.

General

Self-Serving Bias vs Fundamental Attribution Error

People think these two biases are one, and they are not. The self-serving bias is about you. You take the credit when things go well. When they go wrong, you blame bad luck. The other bias is about other people. You blame who they are for what they do, and you forget the spot they were in.

General

Classical Conditioning vs Operant Conditioning

People mix up classical and operant conditioning, but they run in opposite ways. In classical conditioning, a sign comes first. You learn to react to the sign, like a dog that drools at a sound that means food. In operant conditioning, you act first. What comes after makes you do it more or less.

General

Independent Variable vs Dependent Variable

People mix up the two variables in a study. The independent variable is the one you change on purpose, like hours of light. The dependent variable is the one you measure, like how tall a plant grows. A simple test: the dependent one depends on the other. It goes on the y-axis, and the one you change goes on the x-axis.

General

Research Question vs Hypothesis

People use the two words as if they mean the same thing, but one asks and one guesses. A research question asks what you want to know, like do phones at night cut sleep. A hypothesis is a guess at the answer that a test could prove wrong, like teens who use phones after 10 pm sleep less. The question comes first, and the hypothesis follows from it.

General

APA Style vs MLA Style

People think APA and MLA are the same rules with small changes, but they put weight on different things. APA cites by name and year, since the date of a study matters in psychology and science. MLA cites by name and page, since essays on books quote exact lines. APA uses initials. MLA writes first names out.

General

Credit Report vs Credit Score

People think a credit report and a credit score are the same thing. The report is the full record: your loans and cards, what you owe and any late payments. The score is one number worked out from that record. Fix an error in the report, and the score can change with it.

General

Sinking Fund vs Emergency Fund

People think a sinking fund and an emergency fund are the same pot. A sinking fund is for a cost you know is coming, like a yearly insurance bill. An emergency fund is for a cost you cannot see coming, like a broken car. Using one for the other leaves you short when the real bill comes.

General

Asset Allocation vs Diversification

People mix up these two. Asset allocation is the big choice of how much goes to each kind, like stocks, bonds and cash. Diversification is spreading your money over many things, so one failure cannot sink you. A mix of all stocks can still be spread over many firms. You need both.

General

npm vs Yarn

People think Yarn has its own store of packages, apart from npm, and it does not. Yarn and npm both fetch from the same npm registry. They differ in how they install, and in the lockfile they write. npm writes package-lock.json, and Yarn writes yarn.lock. A project should pick one and stick to it.

Development

npm vs pnpm

People think pnpm is just a faster npm, and there is more to it. npm puts a full copy of each package in every project. pnpm keeps one copy of each version on the disk and links each project to it. pnpm is also strict, so code may only use the packages it listed.

Development

pip vs uv

People think uv is a new place to get Python packages, and it is not. Both uv and pip fetch from PyPI. pip only installs packages. uv also makes the virtual environment, keeps a lockfile and can even install Python. It is written in Rust, so it is much faster too.

Development

Maven vs Gradle

People think Gradle is just Maven with a new file, and it is not. Maven describes a project in pom.xml and runs a fixed list of steps. A Gradle build file is code, in Kotlin or Groovy, so you can add steps of your own. Gradle also skips work that has not changed. Both can fetch packages from Maven Central.

Development

Prettier vs ESLint

People think Prettier and ESLint do the same job, and they do not. Prettier only changes how code looks: the spaces, the quotes and the line breaks. ESLint looks for code that may be wrong, like a name you set but never use. Most teams run both. Prettier sets the look, and ESLint hunts for bugs.

Development

Lockfile vs Version Pinning

People think a lockfile and a pinned version are the same thing, and they are not. You pin a version by hand, in the list of packages you asked for. A lockfile is written by the tool. It holds the exact version of every package, even the ones your packages pulled in. So a project can ask for ranges and still get the same install each time.

Development

Build Tool vs Task Runner

People think a task runner is a build tool, and they are not the same. A task runner just gives a short name to a command, like build or test. A build tool knows which files each step needs. It runs the steps in order and skips the ones that are already done.

Development

Sass Variables vs CSS Custom Properties

People think Sass variables and CSS custom properties are the same thing, and they are not. A Sass variable, like $brand, is used up when the Sass turns into CSS, so the browser never sees it. A custom property, like --brand, lives in the browser. Code can change it while the page runs, which is how dark mode can work.

Development

Bootstrap vs Tailwind CSS

People think Tailwind CSS is just another Bootstrap, and it is not. Bootstrap hands you whole ready parts, like a styled button or a menu, with one class each. Tailwind hands you small classes that each do one job, like a colour or a gap. You build each part from many of them, so pages look less alike.

Development

Virtual DOM vs DOM

People think the virtual DOM is a faster DOM that the browser has, and it is not. The DOM is the real page the browser draws. The virtual DOM is a copy a library like React keeps in memory. The library compares two copies and then changes the real DOM as little as it can.

Development

jQuery vs JavaScript

People think jQuery is a language of its own, and it is not. JavaScript is the language that runs in every browser. jQuery is a library written in JavaScript. It is a set of short helpers to find and change parts of a page. Any jQuery line is JavaScript underneath.

Development

Viewport vs Viewport Meta Tag

People think the viewport and the viewport meta tag are the same thing, and they are not. The viewport is the part of the page you can see on the screen. The viewport meta tag is one line of HTML that tells a phone how wide that view should be. The tag sets the viewport; it is not the viewport.

DesignDevelopmentcross-domain

Lazy Loading vs Infinite Scroll

People think lazy loading and infinite scroll are the same trick, and they are not. Lazy loading waits to fetch a picture that is already part of the page until you scroll near it. Infinite scroll adds new items to the end of the page as you reach the bottom. A feed often uses both at once.

DevelopmentDesigncross-domain

Permissive License vs Copyleft

People think a permissive license and copyleft are both just open source, so the rules match, and they do not. A permissive license asks you to keep the notice, and your own code can stay closed. Copyleft asks more. If you hand out a changed copy, you must share its code under the same license.

Development

GNU GPL vs AGPL

People think the GPL makes you share code for a web app, and it does not. With the GPL, the duty to share starts when you hand out copies. Running changed code on your own server is not handing it out. The AGPL adds that rule, so users of the web service can ask for the code too.

Development

GNU GPL vs LGPL

People think the LGPL is just a short name for the GPL, and it is not. The GPL covers the whole program that uses the code, so that program must be open too. The LGPL is made for libraries. A closed app may use an LGPL library, as long as changes to the library itself are shared.

Development

Open Source vs Source-Available

People think any code you can read online is open source, and it is not. Open source needs a license that lets anyone use, change and share the code for any purpose. Source-available code can be read, but its license bans some uses, such as selling it as a hosted service. So it fails the Open Source Definition.

Development

Alpha Release vs Beta Release

People think alpha and beta are just two names for the same early test version, and they are not. An alpha comes first. It is rough, may be missing parts, and is mostly for the team. A beta comes later. Its features are all in, and it goes to a wider group of outside testers to find the bugs that are left.

Development

Deprecation vs End of Life

People think a deprecated feature has stopped working, and it has not. Deprecated means it still works, but you are told to stop using it because it will go away. End of life is later. It is the date when a version gets no more fixes at all, not even for safety holes.

Development

Semantic Versioning vs Calendar Versioning

People think every version number says how big a change is, and it does not. In semantic versioning it does: 3.0.0 warns that old code may break. In calendar versioning the number is a date, so 24.04 means April 2024. A jump there only tells you time has passed. Check which kind a project uses before you read its numbers.

Development

Hermes /goal vs Kanban

People think a goal puts work on the Kanban board. It does not. A goal keeps one chat going, and a judge checks after each turn. The board holds many cards, and each card runs in its own session with its own worker. Use a goal for one job in this chat. Use the board for many jobs shared out.

AI

Hermes /loop vs Cron Job

People think a loop and a cron job are the same timer. They are not. A loop runs a prompt again inside the chat you have open. A cron job runs on a schedule outside any chat, and each run gets a fresh session. Use a loop to watch something now. Use a cron job for work every night.

AI

web_search vs browser_exec

People think Hermes needs a browser to read the web. It does not. web_search finds pages, and web_extract pulls out their text, fast and cheap. browser_exec starts a real browser that clicks, types and fills forms. Use the web tools to read. Use the browser only when the page needs a hand on the mouse.

AI

Claude Code Subagent vs Skill

People think a skill and a subagent are two names for one helper. They are not. A skill is a set of steps that Claude reads into the chat you are in. A subagent is a second worker with its own fresh context and its own tools, and it sends back only a summary. Use a skill to teach Claude a job. Use a subagent to keep a big search out of your chat.

AI

Claude Code Dev Container vs Sandbox

People think the sandbox and a dev container are the same wall. They are not. The sandbox wraps only the shell commands Claude starts, on your own computer. A dev container is a whole separate box, with its own tools and files, and Claude Code itself runs inside it. Use the sandbox for safe commands on your machine. Use a dev container to give a team one shared, walled-off setup.

AI

Singularity vs Superintelligence

People think the singularity and superintelligence are one thing. They are not. Superintelligence is a kind of AI: one far smarter than the best people. The singularity is an idea about how fast we might get there. It says AI would make itself better, faster each time, until change runs away. You could have one without the other. Neither one exists today.

AI

Agentic AI vs Generative AI

People think agentic AI is a new kind of model that replaces generative AI. It is not. Generative AI makes text, pictures or code. Agentic AI puts that same kind of model in a loop with tools, so it can act on what it writes. Most agents today run on a generative model. The difference is what the system does with the words, not the model inside.

AI

Deep Blue vs AlphaGo

People think AlphaGo did what Deep Blue did, just for a harder game. It did not. Deep Blue won at chess by searching through millions of moves a second, using rules that chess experts helped write. Go has far too many moves for that. AlphaGo learned how to judge a board from games, many of them played against itself. One searched, the other learned.

AI

Perceptron vs Multilayer Perceptron

People think a multilayer perceptron is just a big perceptron. It is more than that. One perceptron adds up its inputs and says yes or no, so it can only split data with a straight line. A multilayer perceptron stacks many of them in layers, with a bend in each one. The layers let it learn curved and tangled patterns that one perceptron never can.

AI

Structured Output vs MCP Structured Tool Output

People think both mean the model writes neat JSON. Only one does. Structured output is a model's own answer, forced into a set shape. MCP structured tool output is data a tool sends back after it runs, in a field called structuredContent. The model did not write it at all. One shapes what the model says, the other shapes what a tool returns.

AI

HTTP+SSE vs Streamable HTTP

People think Streamable HTTP got rid of streams, since the old MCP transport it replaced was named after them. It did not. The old HTTP+SSE transport needed two addresses: one long stream to listen and one to send. Streamable HTTP uses one address. A reply can still come back as a stream when it needs to. The change was fewer addresses, not fewer streams.

AI

MCP Registry vs Package Registry

People think the MCP Registry is like npm for MCP servers. It is not. A package registry such as npm or PyPI stores the code you download. The MCP Registry stores no code at all. It holds a short listing for each server, with its name and where to get it, and that points back to npm, PyPI, Docker Hub or a web address.

DevelopmentAIcross-domain

Internationalization vs Localization

People think i18n and l10n are the same piece of work, and they are not. The first, internationalization, is done once. The people who build the product keep words out of the code and leave room for any language. The second, localization, is done again for each market. Translators fill in the words. The team sets the dates, money and pictures for that place.

Design

Visual Style Guide vs Content Style Guide

People think a style guide is only about how a product looks, and it is not. A visual style guide sets the colors, type and spacing. A content style guide sets the words: the voice, the tone, and which word to use for each thing. A product needs both. A screen can look right and still confuse people with the wrong word.

Design

High Contrast Mode vs Dark Mode

People think high contrast mode is just a darker dark mode, and it is not. The designer makes a dark mode, and it keeps the brand colors and shadows. High contrast mode is set by the person. It throws out the design's colors and uses a few strong ones they picked. Shadows and tints are gone, so only real borders show where a button is.

Design

Drop Shadow vs Elevation

People think elevation and a drop shadow are the same thing, and they are not. Elevation is how high a part sits above the page, as a level on a scale. A drop shadow is one way to draw that height on screen. The system fixes a few levels, and each level gets its own shadow. The level is the idea, and the shadow is the paint.

Design

shadcn/ui vs Component Library

People think shadcn/ui is a component library, and its own docs say it is not. A component library is a package you install and upgrade, and its maker keeps the code. shadcn/ui copies the code for each part into your project. After that, the code is yours. You can change any line, but fixes made to the first copy will not reach yours.

Design

Bokeh vs Depth of Field

People think bokeh and a shallow depth of field are the same thing, and they are not. Depth of field is how much of the picture is sharp. A shallow one leaves a lot of it blurred. Bokeh is what that blur looks like: smooth and creamy, or busy with hard rings. Two lenses can blur the same amount and still give very different bokeh.

Design

Master Shot vs Establishing Shot

People think a master shot and an establishing shot are the same wide shot, and they are not. An establishing shot is short. It shows the place, like a house from the street, before the scene starts. A master shot films the whole scene with the actors, from start to end, in one wide frame. Closer shots are cut in with it later.

Design

Adobe Animate vs After Effects

People think Adobe Animate and After Effects do the same job, and they do not. Animate is for drawing cartoons, frame by frame. It also makes web banners and simple games. After Effects is for layers. You move text, shapes and video, and add effects to film. A cartoon starts in Animate. A title or an effects shot is made in After Effects.

Design

Rebrand vs Brand Refresh

People think a rebrand and a brand refresh are the same job, and they are not. A refresh keeps the core of the brand and only tidies it. The name and the main idea stay, and details like the logo lines or the font change. A rebrand can change the core: the name, the idea, the logo and the colors. A refresh keeps old fans at ease, while a rebrand asks them to learn the brand again.

Design

Model Routing vs Model Fallback

People think model routing and model fallback are the same switch between models. Routing picks a model before the call is sent, based on how hard the request looks. Fallback waits for a call to fail, then sends it again to a second model. Routing saves money on easy requests. Fallback keeps the app up when a model is down.

Development

LangSmith vs Langfuse

People think LangSmith and Langfuse are one tool, because the names sound alike and both trace and test LLM apps. LangSmith comes from LangChain, the team behind the LangChain framework. Langfuse is a separate project with open code, free to run on your own servers. Both record each step of a run, keep prompts and score answers.

Development

Claude Managed Agents vs Claude Agent SDK

People think Claude Managed Agents and the Claude Agent SDK are the same, since both give you a ready-made Claude agent. The Agent SDK is a library. Its agent runs inside your own app, on your own machine or server. Managed Agents is a service in the Claude API. Its agent runs as a hosted session in a sandbox, and your app sends it events.

DevelopmentAIcross-domain

Client SDK vs Agent SDK

People think an agent SDK is just a client SDK with more features. It is a different job. A client SDK sends one request and hands back one reply. Your code reads it, runs any tool the model asked for, and calls again. An agent SDK runs that whole loop for you, tools and all, until the task is done.

Development

LLM API vs Client SDK

People think the SDK and the API are the same thing, because both carry the model maker's name. The API is the service on the model maker's servers. It takes web requests and sends back replies. The client SDK is a code library on your side. It builds those web requests for you, so you call a function instead.

Development

OpenAI API vs OpenAI-Compatible API

People think any server that speaks the OpenAI format is OpenAI's own API. The OpenAI API is a service run by OpenAI, and it runs only its own models. A compatible API is any server that takes calls in the same shape. It might run a model on your laptop, or one from another firm. Your code stays the same, but the model behind it changes.

DevelopmentAIcross-domain

LiteLLM vs OpenRouter

People think LiteLLM and OpenRouter are the same, since both let one app reach many models. LiteLLM is free code you run yourself. You give it your own key for each provider. OpenRouter is a service that someone else runs. You use one OpenRouter key, and it bills you for every model in one place.

DevelopmentAIcross-domain

MVC vs MVVM

People think MVVM is just MVC with a new name, since both split screen code into three parts. In MVC, a controller takes the user's input and picks what the view shows. In MVVM, a view model holds the screen's state, and the view binds to it. Changes then flow both ways by themselves. The view model never touches the view, so it is easy to test.

Development

Strong Consistency vs Eventual Consistency

People think strong and eventual consistency differ only in speed. With strong consistency, every read sees the newest write at once. It does not matter which copy you ask. With eventual consistency, a copy may show old data for a short time, then catch up. Strong is slower, and it may turn calls away when the network splits. Eventual stays fast and up, but you may see an old value.

Development

Adapter Pattern vs Facade Pattern

People think an adapter and a facade are the same wrapper. An adapter wraps one thing so it fits a shape another part expects. It changes the interface, but not how much there is to it. A facade puts one simple front over many parts. It makes the interface smaller, so callers have less to learn. You use an adapter to make things fit, and a facade to make them easy.

Development

Decorator Pattern vs Adapter Pattern

People think a decorator and an adapter are the same wrapper. A decorator keeps the same interface and adds a feature, like a log or a cache. Callers cannot tell that anything changed. An adapter changes the interface, so one part can plug into another that wants a new shape. A decorator adds to what a thing does. An adapter changes how you call it.

Development

Architecture Decision Record vs Design Doc

People think an architecture decision record is just a short design doc. A design doc plans a whole change before the code is written, and is read and argued over. A decision record saves one choice and the reasons, at any time, in a few lines. Design docs often go stale once the work ships. Decision records are kept for good, and a new one replaces an old one.

Development

PRD vs Design Doc

People think a product requirements document and a design doc are the same plan. The first says what to build and why: the problem, the users and how to tell it worked. A design doc says how to build it: the parts, the data and the options turned down. Product managers most often write the first. Engineers write the second, after they read it.

BusinessDevelopmentcross-domain

Customer Experience vs User Experience

People think CX is just a new name for UX, and it is not. UX is how it feels to use one product, like an app or a website. CX is how a person feels about the whole company. That covers the ads, the shop, the bill, the delivery and the help line too. A great app cannot save a company that loses your bags.

Design

User Persona vs Buyer Persona

People think a user persona and a buyer persona are the same person, and they are not. A buyer persona is the one who chooses and pays, and marketing writes it to sell. A user persona is the one who uses the product each day, and design writes it to build. At a school, the head buys the app, but the teachers use it.

DesignMarketingcross-domain

Customer Journey Map vs Customer Journey

People think the customer journey and a journey map are the same thing, and they are not. The customer journey is the real path a person takes with a company, from first hearing of it to buying and beyond. A journey map is the drawing a team makes of that path. It adds what the person does, thinks and feels at each step. One is the trip, and the other is the map.

DesignMarketingcross-domain

Single Ease Question vs System Usability Scale

People think the Single Ease Question and the System Usability Scale both measure the same thing, and they do not. The SEQ is one question, asked right after each task. It shows which task felt hard. The SUS is 10 statements, asked once at the end. It gives one score for the whole product. The SEQ finds the problem, and the SUS tracks the whole.

Design

Cognitive Walkthrough vs Heuristic Evaluation

People think a cognitive walkthrough is a heuristic evaluation, and it is not. Both are reviews done by a team, with no real users. A heuristic evaluation checks the whole design against a list of rules. A cognitive walkthrough follows one task, step by step. The team plays a brand-new user and asks four questions at each step. That finds where a beginner gets lost.

Design

Preference Testing vs A/B Testing

People think a preference test is an A/B test, and it is not. A preference test asks people which design they like, and why. An A/B test shows each design to real visitors on a live site and counts what they do. One measures what people say. The other measures what they do, and the two can disagree.

DesignMarketingcross-domain

User Scenario vs User Story

People think a user scenario is a user story, and it is not. A user story is one small thing to build, in one line: as a user, I want this, so that I can do that. A user scenario is a longer story about a real moment, with where the person is and what gets in the way. Designers write scenarios to understand people. Teams write stories to plan the work.

DesignBusinesscross-domain

Subtitles vs Closed Captions

People think subtitles and captions are the same thing, and they are not. Subtitles write out the spoken words, often in another language. They are for people who can hear. Captions also write out the sounds, like music, a laugh or a door slam, for people who cannot hear. A film with subtitles can still leave a deaf viewer guessing.

Design

Video Container vs Video Codec

People think MP4 is a codec, and it is not. The codec is how the picture is shrunk to a small size, like H.264 or AV1. The container is the box that holds that picture, the sound and the subtitles in one file, like MP4 or WebM. A video plays only when the device can open both the box and what is inside.

Design

Motion Tracking vs Motion Capture

People think motion tracking and motion capture are the same thing, and they are not. Motion tracking follows a spot through a video that is already filmed, so text or a graphic can move with it. Motion capture records a real actor's body as it moves, with suits or cameras. Then it puts that motion on a 3D character. One pins things to footage. The other brings a character to life.

Design

UI Density vs Pixel Density

People think UI density and pixel density are the same thing, and they are not. Pixel density is how many tiny dots a screen packs into each inch. The hardware sets it. UI density is how tightly a design packs its buttons, rows and gaps. The designer picks it. A sharp phone screen can still show a roomy layout.

Design

Opacity vs Alpha Channel

People think opacity and the alpha channel are the same thing, and they are not. Opacity is one setting for a whole layer, like 50% for the dark layer behind a dialog. The alpha channel is saved inside a picture. It holds its own value for each pixel. That is how a logo can have soft edges and a see-through middle.

Design

Pixel Grid vs 8-Point Grid

People think the pixel grid and the 8-point grid are the same grid, and they are not. The pixel grid is the screen itself, the tiny squares every shape lands on. The 8-point grid is a rule a team picks for spacing, so gaps and sizes come in steps of 8. A layout can follow the 8-point grid and still blur if its edges miss the pixel grid.

Design

Catastrophic Forgetting vs Model Collapse

People think catastrophic forgetting and model collapse are one problem, because both make a model worse after training. They are not. Forgetting happens inside one model: new training writes over skills it had. Collapse builds up across many models, each one trained on text made by the one before. The fix for the first is to test old skills. The fix for the second is to keep real human data.

AI

RLHF vs RLVR

People think RLVR is just a new name for RLHF. It is not. In RLHF, people rank answers and a reward model learns their taste, so it can reward any kind of reply. In RLVR, a program checks the answer against a key or runs tests, so it only works where a right answer exists. RLHF shapes tone and help. RLVR trains math and code.

AI

Llama vs Olmo

People think Llama and Olmo are open in the same way, because you can download both. They are not. Meta shares the Llama weights, but not the data it learned from or the full code that trained it. Ai2 shares all three for Olmo. With Llama you can run and tune the model. With Olmo you can also check, and repeat, how it was made.

AI

interrupt() vs abortController

People think interrupt() and abortController do the same thing, because both stop an agent that is still at work. They do not. interrupt() stops only the turn the agent is on, and the session stays open, so you can send a new message right after. It needs streaming input mode. abortController ends the whole query() call, and nothing more comes back from it. Use the first to steer, and the second to quit.

AI

Pass by Value vs Pass by Reference

With pass by value, a function gets a copy, so your own value stays the same. With pass by reference, it gets a link to your value, so its changes show outside too.

Development

Tuple vs Array

A tuple is a small, fixed group where each spot has a meaning, like x and y. An array, or list, is a row of items of the same kind that can grow and change.

Development

Pseudocode vs Algorithm

An algorithm is the list of steps that solves a problem. Pseudocode is one way to write those steps down, in plain words, before they become real code.

Development

Struct vs Class

A struct is a simple bundle of named values. In C#, Swift and Go it is copied when you pass it. A class object is shared, so a change shows everywhere.

Development

Composition vs Inheritance

Inheritance says a thing is a kind of another, like a cat is an animal. Composition says a thing has parts, like a car has an engine. Parts are easier to swap.

Development

Iterator vs Generator

An iterator is any object that gives items one at a time. A generator is a short way to write one: a function that pauses at each yield.

Development

First-Class Function vs Higher-Order Function

With first-class functions, a function is a value you can store and pass. A higher-order function is one that takes in or hands back a function.

Development

Insertion Sort vs Selection Sort

Insertion sort slides each item back into place. It is quick when the list is nearly in order. Selection sort looks for the smallest item on each pass. It is always slow.

Development

Backtracking vs Brute Force

Brute force tries every full answer. Backtracking builds an answer step by step and gives up on a path as soon as it cannot work, so it skips most of the tries.

Development

JVM vs CLR

Both run code that was first turned into an in-between form. The JVM runs bytecode made from Java and Kotlin. The CLR runs CIL made from C# and F#.

Development

JVM vs Virtual Machine

A virtual machine in the cloud acts like a whole computer and runs a full system. The JVM is just a program. It only runs Java bytecode.

DevelopmentInfrastructurecross-domain

Executable File vs Binary File

A binary file is any file that is not plain text, like a photo or a zip. An executable is one kind. It is a program you can start.

Development

Memory Safety vs Garbage Collection

A garbage collector frees memory you are done with. Memory safety means code can never reach memory that is not its own. Rust is safe without one.

Development

Model Selection vs Model Routing

People think model selection and model routing are the same job, and they are not. Model selection is a choice you make once. You test a few models on real examples of your task, then pick one. Model routing is code that picks a model again for each request, as it comes in. You often do the first to set up the second.

AIDevelopmentcross-domain

Custom Instructions vs Custom AI Assistant

People think custom instructions make a custom assistant, and they do not. Custom instructions are your own notes. The app adds them to every chat you start. A custom assistant is a helper of its own, with a name, rules and files. Its setup works only when someone opens it, and other people can open it too.

AI

AI Browser vs AI Search Engine

People think an AI browser is just an AI search engine, and it is not. An AI search engine is a site you ask a question. It reads the web and writes an answer with links. An AI browser is the app you surf the web in. Its helper reads the page you have open, and it can click and type for you.

AI

AI Provider vs AI Deployer

People think the firm that buys an AI tool has no duties, because the maker carries them all. The provider builds the system and puts its name on it. The deployer uses it at work, and has its own list of duties. Put your own brand on a bought system, and you become its provider too.

AI

AI Provider vs Inference Provider

People think an AI provider is a firm that hosts models, and it is not. An inference provider runs models on its servers and sells the answers by the token. In the EU AI Act, a provider is whoever puts an AI system on the market under its own name. A host can be both, or neither.

AI

AI Oversharing vs AI Data Exfiltration

People think AI oversharing is a hack, and it is not. In oversharing, the AI shows you files you can already open. They were just shared too widely. In data exfiltration, someone tricks the AI into sending your data to them. The first is fixed by tidying who can open what. The second needs a guard against hidden orders.

AI

Pseudo-Localization vs Localization

People think pseudo-localization is a kind of translation, and it is not. Localization puts an app into a real language. People who know that language write it. Pseudo-localization makes fake text that only looks foreign. It is still English with accents and extra words. A team uses it to find text that will not fit, long before the real words are written.

Design

Print Debugging vs Logging

People think print debugging is the same as logging, and it is not. Print lines are short-lived. You add them to chase one bug, then delete them. Logging stays in the code for good. It writes events with a time and a level, so people can read them long after the code ran.

DevelopmentInfrastructurecross-domain

JSON Mode vs Structured Output

People think JSON mode makes a model's reply fit their schema, and it does not. JSON mode only makes sure the reply is valid JSON, so a program can open it. The fields can still be missing or named in another way. Structured output checks the reply against the schema you give, so every field you asked for is there. Neither one checks that the values are true.

AI

AI Code Completion vs Code Generation

People think AI code completion and AI code generation are the same tool, and they are not. Completion works inside the editor while you type. It reads the code near your cursor and offers the next line or two in grey. Generation starts from a request in plain words, and the model writes a whole function, a test or a small app. Both write code you still have to read.

AI

Data Poisoning vs Adversarial Example

People think data poisoning and adversarial examples are one attack. They are not. Poisoning strikes while the model is still learning. Bad examples go into what it learns from, and the harm is built in. An adversarial example strikes once the model is done. The model is fine. One input is changed so the model reads it wrong.

AI

AI Red Teaming vs Dangerous Capability Evaluation

People think red teaming and a danger test are the same check, and they are not. A red team tries to make a model break its rules, with tricks like jailbreaks and hidden orders. A danger test asks what the model can do at its strongest, such as help with a weapon or a break-in. One looks for holes in the rules. The other measures the power behind them.

AI

Subagent vs Agent Handoff

People think giving work to a subagent and making a handoff are the same move, and they are not. A subagent works for the main agent. It does one job in its own window and sends back a summary, and the main agent stays in charge. In a handoff, control moves. The new agent takes over the chat, and the first one steps out.

AI

Agent Skill vs Agent Instruction File

People think a skill is just another instruction file, and it is not. An instruction file, such as AGENTS.md, is read on every run, so it costs space each time. A skill shows only its name and a short note until a task needs it. Then the agent loads the full steps. Rules that always apply go in the file. Steps for one kind of job go in a skill.

AI

SystemMessage vs systemPrompt Option

People think SystemMessage holds the system prompt, and it does not. The systemPrompt option is the set of rules the agent gets before your task. SystemMessage is a message the SDK sends back to your code. The first one, init, reports the session ID, the model and the tools. Later ones mark events, such as a chat being squeezed into a summary.

AI

forkSession vs enableFileCheckpointing

People think forking a session also copies the files, and it does not. forkSession copies the chat into a new session with its own ID. That lets you try a second idea. A file the fork edits is changed for real, for each session in that folder. File checkpoints are what back files up, so the edits can be rolled back.

AI

RAG vs Semantic Search

People often think RAG is just semantic search. Semantic search only finds the passages that match a question by meaning. RAG is the whole system: a search finds the passages, then a language model reads them and writes the answer. The search inside RAG can be semantic, keyword or both.

AI

Reciprocal Rank Fusion vs Reranking

People think reciprocal rank fusion is a kind of reranking, and it is not. Fusion merges two or more ranked lists using only each result's place in each list, so it is fast and reads no text. A reranker is a model that reads the question with each result and scores it again. Fusion joins the lists. The reranker then puts the short list in a better order.

AI

Overfitting vs Data Leakage

People think a leak and overfitting are the same problem, since both make a model look better than it is. Overfitting is a model learning its training data too closely. It shows as a gap between the training score and the held-back score. A data leak is when facts from the test slip into training. It hides the gap, because the held-back score is high too. More data or a simpler model fixes overfitting. Only a cleaner split fixes a leak.

AI

AI Benchmark vs AI Leaderboard

People use benchmark and leaderboard as if they meant one thing. A benchmark is the test: fixed questions and a fixed way to score them. A leaderboard is the public table that ranks models by their scores. One leaderboard can mix many benchmarks. LMArena ranks models by people's votes, with no fixed test at all.

AI

Grounding vs Faithfulness

People think grounding and faithfulness are the same thing. Grounding is how a system is built: it hands the model source text and asks it to answer from that. Faithfulness is a score. It checks whether one answer really stuck to the source, claim by claim. A grounded system can still give an unfaithful answer, and faithfulness is how you catch it.

AI

HumanEval vs SWE-bench

People quote HumanEval and SWE-bench as if both meant the same thing: can it code? HumanEval has 164 short tasks. Each is one function, checked by tests written for it. SWE-bench gives a model real bug reports from large Python projects, and each project's own tests decide. A model can score high on HumanEval and still struggle on SWE-bench. Finding the right files in a big codebase is much of the work.

AI

MCP Client vs MCP Host

People call the whole AI app the MCP client, and that is not right. The app you use is the host. It holds the model, asks you before risky actions, and decides which servers to link to. A client is one connection inside it, made for one server. An app linked to three servers is one host with three clients.

AI

MCP Elicitation vs MCP Authorization

People think URL mode elicitation is how a client signs in to an MCP server. It is not. MCP authorization is the OAuth sign-in between the client and the server, and it gives the client a token. URL mode elicitation is for the server's own needs, such as a key for some other service. You type that secret on a web page, and the client never sees it.

AI

URI vs URL

People think URI and URL are two names for the same thing, or two rival ideas. Neither is right. A URI is any short text name for one thing. A URL is the kind of URI that also says where the thing is and how to fetch it. So every URL is a URI, but a name like mailto:emma@example.com is a URI you cannot fetch.

General

Plan Mode vs Auto Mode

People think plan mode and auto mode are two ways to let Claude get on with it. They answer different questions. Plan mode decides what Claude may do yet: it can read and think, but it makes no edits until you approve a plan. Auto mode decides who says yes: Claude acts, and a second model checks each step instead of asking you.

AI

Don't Ask Mode vs Bypass Permissions Mode

People think Don't Ask mode means Claude stops asking and does everything. It is the opposite. Don't Ask mode refuses any tool your allow rules have not already approved, so nothing new can run. Bypass mode is the one that lets almost everything run without a question. Only deny rules still hold there.

AI

Esc vs Esc Esc

People think pressing Esc twice is just a firmer way to stop Claude. It does a different job. One Esc stops Claude in the middle of its work and keeps what it has done, so you can steer it. Esc twice opens the rewind menu, where you jump back to an earlier prompt and undo the code, the chat or both.

AI

Ultrathink vs Effort Level

People think ultrathink is one more effort level. It is not. The effort level sets how hard Claude thinks for the whole session, from low to max. Typing ultrathink in a prompt asks for deeper thinking on that one prompt only. The next prompt goes back to the effort level you set.

AI

Background Session vs Cloud Session

People think a background session runs in the cloud. It does not. A background session runs on your own computer, with no terminal open. A cloud session runs on Anthropic's servers from a fresh copy of your repo. It keeps going with your laptop shut, but it cannot see files that live only on your machine.

AI

Claude Code Checkpoint vs Git Commit

People think a Claude Code checkpoint is a git commit. It is not. A checkpoint is a save point Claude Code makes on your computer at each prompt, so you can rewind a bad edit. It is kept for about 30 days, and it misses changes made by shell commands. A commit lives in git, is shared with your team, and stays until someone removes it.

AIDevelopmentcross-domain

Permission Mode vs Permission Rule

People think a permission mode and a permission rule are the same switch. They are not. A mode sets how often Claude asks, for the whole session. A rule names one tool, or one command, and says allow, ask or deny for just that one. A deny rule wins in every mode, even bypass.

AI

Narrow AI vs AGI

People think a very smart narrow AI is close to AGI, and it is not. Narrow AI is built for one kind of job, like chess or spam. It can beat every person at that job and still fail at the next one. AGI would learn almost any thinking job a person can. Every AI in use today is narrow, however well it talks.

AI

Self-Supervised vs Unsupervised Learning

People think self-supervised learning is the same as unsupervised learning, since no person labels anything. It is not. The second kind has no answers at all, and looks for groups. The first hides part of the data and makes that part the answer. So it trains the way supervised learning does, on labels it made itself.

AI

scikit-learn vs PyTorch

People think scikit-learn is a smaller PyTorch, and it is not. scikit-learn holds the classic models, like trees, forests and k-means. It is built for tables that fit on one computer. PyTorch is for building and training neural networks, often on a graphics card. Most table jobs need only the first.

AI

Named Entity Recognition vs Information Extraction

People think finding names and pulling out facts are the same job. Named entity recognition is only the first step. It finds the names in a text and says what kind each one is. Information extraction goes on to fill set fields, such as which firm paid which sum on which date.

AI

BERT vs GPT

People think BERT is an early chat model. It is not. BERT reads a whole text both ways and hands back numbers that other tools use to sort and search. GPT reads only the words so far and writes the next one. One reads, and the other writes.

AI

Image Captioning vs Visual Question Answering

People think captioning and visual question answering are the same job. They are not. A caption is one fixed line that says what is in a picture, the same for everyone. Visual question answering waits for your question, and one photo gets a new answer for each one.

AI

Inpainting vs Image-to-Image

People think inpainting and image-to-image are the same edit. They are not. Image-to-image redraws the whole picture in a new style, so faces and text can shift too. Inpainting redraws only the part you mask, and leaves the rest as it was.

AI

Multimodal Embedding vs Vision Language Model

People think a multimodal embedding model can describe a picture. It cannot write a word. It turns a picture or some text into numbers, so you can measure how close they are. A vision language model reads the picture and writes an answer in words.

AI

Speaker Diarization vs Speech-to-Text

People think speech-to-text already says who spoke. It does not. Speech-to-text writes down the words. Speaker diarization splits the sound by voice and labels each part Speaker 1, Speaker 2 and so on. A meeting script needs both.

AI

Text-to-Video vs Image-to-Video

People think the two ways to start an AI clip are the same. They are not. Text-to-video starts from words alone, so the model picks every face and place. Image-to-video starts from a still picture you chose. It becomes the first frame and keeps the look steady.

AI

Model Hub vs Inference Provider

People think a model hub runs the models it lists. It mostly does not. A hub such as Hugging Face stores model files and their cards, and you download them to run on your own machine. An inference provider runs the model on its own chips. You send it a request and pay for the answer.

AI

OpenRouter vs Inference Provider

People think OpenRouter runs AI models, like an inference provider. It does not. An inference provider owns the chips that run a model and sells the answers. OpenRouter passes your request to one of those providers. It gives you one key and one bill for hundreds of models, and it switches provider if one fails.

AI

Chatbot vs AI Assistant

People think a chatbot and an AI assistant are the same thing. They are not. A chatbot talks: you ask, and it answers in words. An AI assistant is a chatbot that is also linked to your tools, such as your mail and calendar. So it can do small jobs in them, and a good one asks you first.

AI

AI Project vs Custom AI Assistant

People think an AI project and a custom AI assistant are the same setup. They are not. A project is your own folder, with files and rules that every chat inside it shares. A custom assistant is a helper set up for one job that other people can open too. Each one who opens it starts with its rules and files.

AI

Web Search (AI) vs AI Search Engine

People think web search and an AI search engine are the same thing. They are not. Web search is a tool inside a chat app. The assistant uses it to look things up when a question needs fresh facts. An AI search engine is a whole site built around search, where every answer is written from the pages it reads.

AI

~/.claude/settings.json vs ~/.claude.json

People think ~/.claude.json is the user settings file. It is not. The user settings file is settings.json, inside the .claude folder in your home. Claude Code writes the other file itself. It keeps your account, your MCP servers and the folders you trust there. You edit the settings file by hand, and you rarely need to touch the other one.

AI

CLAUDE_CODE_SUBAGENT_MODEL vs CLAUDE_CODE_SUBAGENT_MODEL_FORCE

People think CLAUDE_CODE_SUBAGENT_MODEL sets the model for every helper. It does not. A model in the helper's own file wins over it, and so does a model Claude names when it starts the helper. The FORCE switch, set to 1, flips that order. Then your pick wins over both, and every helper runs on it.

AI

terminalTitleFromRename vs CLAUDE_CODE_DISABLE_TERMINAL_TITLE

People think setting terminalTitleFromRename to false stops Claude Code from changing the tab title. It does not. The tab still gets a title made from the chat. The key only stops a name you give with /rename from taking its place. To stop all title changes, set the DISABLE_TERMINAL_TITLE switch. That also skips the model call that writes the title.

AI

syntaxHighlightingDisabled vs CLAUDE_CODE_SYNTAX_HIGHLIGHT

People think setting CLAUDE_CODE_SYNTAX_HIGHLIGHT to false turns off all code colors. It only does it in diffs. Code blocks and file views keep their colors. The syntaxHighlightingDisabled key, set to true, turns color off in all three. Use the key when the colors clash with your theme, or slow down a screen reader.

AI

denyInsteadOfPrompting vs Don't Ask Mode

People think denyInsteadOfPrompting is just don't ask mode. It is not. Don't ask mode is a mode. It runs only what your rules allow and denies the rest. The key keeps whatever mode you are in, such as auto, and only turns a prompt that would wait into a no. So in auto mode, the checker can still approve calls.

AI

allowedDomains vs WebFetch Permission Rule

People think allowedDomains decides which pages Claude can fetch. It does not. It only covers shell commands in the sandbox, such as curl or npm. Claude's own fetch tool follows WebFetch rules instead. The link runs one way: a WebFetch domain allow rule is also added to the sandbox list, but the sandbox list never opens a fetch.

AI

denyRead vs Read Permission Rule

People think a denyRead entry stops Claude from reading a file. It only stops shell commands in the sandbox, such as cat. Claude's own Read tool runs outside the sandbox and follows Read rules. A Read deny rule does more. It blocks the tool. With the sandbox on, it is copied into denyRead as well.

AI

excludedCommands vs Bash Permission Rule

People think putting a command in excludedCommands lets it run without asking. It does not. The list only takes a command out of the sandbox, so it runs with your full access. It still goes through your rules and your mode. In Manual mode, it still needs a Bash allow rule to skip the prompt.

AI

Playbook vs Runbook

People think a playbook and a runbook are the same thing, and they are not. A runbook gives one fixed set of steps for a known problem, so anyone can follow it. A playbook offers a few proven moves for a situation, and people pick the one that fits. In IT teams the two words are sometimes swapped.

BusinessInfrastructurecross-domain

Center of Excellence vs Community of Practice

People think a center of excellence and a community of practice are the same, and they are not. A center of excellence is a small team with a job. It sets the rules and picks the tools. A community of practice is people from many teams who join by choice to share what they learn.

BusinessGeneralcross-domain

Enabling Team vs Center of Excellence

People think an enabling team is a center of excellence with a new name, and it is not. A center of excellence stays in the middle. It sets rules for all teams. An enabling team joins one team for a few weeks, teaches a skill, then leaves.

Business

Use Case vs User Story

People think a use case is just a long user story, and it is not. A user story is one short line about what a user wants and why. The details come later, in talk. A use case writes out each step to the goal, plus what happens when a step fails.

Business

Onboarding Buddy vs Mentoring

People think an onboarding buddy is a mentor, and it is not. A buddy is a peer who helps with small, daily questions in the first weeks. A mentor has more years in the job and guides a whole career, often for years.

Business

ISO 27001 vs SOC 2

People think ISO 27001 and SOC 2 are two names for the same badge, and they are not. ISO 27001 is a world standard. An outside body checks your whole way of running security and gives a certificate for three years. SOC 2 is a US report. An audit firm writes down how your controls worked, and the buyer reads it and decides.

Infrastructure

NIS2 vs DORA

People think NIS2 covers every firm's cyber duties, banks included, and it does not. For banks, insurers and other finance firms, DORA takes the place of most NIS2 rules. DORA is one rule for the whole EU. NIS2 reaches firms only through each country's own law.

Infrastructure

NIS2 vs Cyber Resilience Act

People think the CRA is NIS2 for gadgets, and it is not. NIS2 is about how a firm runs: its risk plan, its staff and how fast it reports an attack. The CRA is about what a maker sells. Every product with software inside must be safe when it ships and get fixes for years.

Infrastructure

ISO 27001 vs ISO 27002

People think ISO 27002 is a newer or easier ISO 27001, and it is not. ISO 27001 sets the rules for running security as one system, and a firm can be certified to it. ISO 27002 is the guide that explains each of the 93 controls. Nobody is certified to ISO 27002. Teams read it to learn how to do each control well.

Infrastructure

ISMS vs ISO 27001

People think ISO 27001 is the ISMS, and it is not. The ISMS is your own way of running security: your rules, roles, risks and checks. ISO 27001 is the yardstick an auditor holds it up to. A firm can run an ISMS with no badge at all.

Infrastructure

NIST CSF vs ISO 27001

People think the NIST CSF is a US version of ISO 27001, and it is not. ISO 27001 sets rules that an auditor checks, and it ends in a badge. The NIST CSF is a free guide that sorts security work into six functions. There is no badge for it. Many firms use the CSF to plan and ISO 27001 to prove.

Infrastructure

Essential Entity vs Important Entity

People think the second class under NIS2 has lighter duties, and it does not. Both classes must manage cyber risk and report big attacks the same way. Only the checks and the fines differ. Officials can check an essential entity at any time. They look at an important one only after a sign of trouble.

Infrastructure

Whaling vs Business Email Compromise

People think whaling and CEO fraud are the same scam, and they are not. In whaling, the boss is the target. The fake message is sent to a leader. In business email compromise, the boss is the mask. The fake message seems to come from a leader and goes to staff. One attack can lead to the other.

Infrastructure

Showback vs IT Chargeback

People think showback and chargeback are the same thing, and they are not. Both split the IT bill by team, project or product. Showback only shows each team its share, and one central budget still pays. Chargeback takes the money out of each team's own budget. Most firms start with showback, and move to chargeback once the numbers are trusted.

Infrastructure

OneDrive vs SharePoint

People think OneDrive and SharePoint are two places for the same files, and they are not. OneDrive holds files that one person owns. SharePoint holds files a team owns. When someone leaves, their OneDrive goes with their account. A SharePoint site stays. Keep your drafts in OneDrive and the team's files in SharePoint.

Business

Microsoft Teams vs Slack

People think Microsoft Teams is just Slack made by Microsoft, and it is not quite. Both are team chat with channels, threads and calls. Teams also runs video meetings and keeps its files in SharePoint and OneDrive. Slack keeps chat at the center and links to many outside apps. A firm already on Microsoft 365 often picks Teams for that reason.

Business

Microsoft 365 vs Microsoft 365 Copilot

People think Microsoft 365 Copilot is the same thing as Microsoft 365, and it is not. Microsoft 365 is the bundle of work apps, like Word, Excel, Outlook and Teams. Copilot is the AI helper that works inside those apps. The full version is a paid add-on on top of a Microsoft 365 plan. The mix-up grew in 2025, when the Microsoft 365 app was renamed the Microsoft 365 Copilot app.

BusinessAIcross-domain

Airtable vs Microsoft Excel

People think Airtable is a spreadsheet, and it is not. It looks like one, with rows and columns, but it is a database. Each column holds one type of data, and each row is a record. Tables can link, so a fact is kept once and used in many places. Excel is better for sums and charts. Airtable is better for lists that point at other lists.

Business

Microsoft Power Apps vs Microsoft Power Automate

People think Power Apps and Power Automate are the same tool, and they are not. Power Apps builds screens that people tap, like a form on a phone. Power Automate runs steps in the background with no screen, like sending an email when a new row lands. They often work as a pair. The app saves the data, and the flow acts on it.

BusinessDevelopmentcross-domain

Reserved Instance vs Savings Plan

People think a savings plan is a reserved instance with a new name, and it is not. A reserved instance is a promise to use one server type in one region. A savings plan is a promise to spend a set amount each hour on any covered server. Both last one or three years and cut the price a lot. If your servers change often, the savings plan keeps the discount.

Infrastructure

Spot Instance vs Reserved Instance

People think spot and reserved are two ways to get the same cheap server, and they are not. A reserved instance is a promise to pay for one or three years, and the server is yours the whole time. A spot instance needs no promise and costs even less. But the cloud can take it back after a short warning. Reserve what must stay up, and use spot for work that can stop.

Infrastructure

Rightsizing vs Autoscaling

People think autoscaling makes a server the right size, and it does not. Autoscaling adds or removes copies of a server as traffic changes. Each copy stays the same size. Rightsizing changes the size itself, to fit the work one server really does. A team often needs both: the right size, and the right number.

Infrastructure

Sovereign Cloud vs Data Residency

People think a sovereign cloud just means data kept in the EU, and it is not. Data residency is only about where the data sits. A sovereign cloud also asks who runs it and whose law can reach it. A US firm's EU region gives you residency. Its home courts may still be able to ask for your data.

InfrastructureAIcross-domain

Gaia-X vs Cloud Provider

People think Gaia-X is a European cloud, and it is not. A cloud provider owns servers and rents them out. Gaia-X runs no servers at all. It writes rules and gives labels to services that follow them. You rent from a provider, and Gaia-X helps you check that provider.

DataInfrastructurecross-domain

Platform Team vs Enabling Team

People think a platform team and an enabling team do the same job, because both help other teams go faster. They do not. A platform team builds a tool that other teams use for as long as they like. An enabling team teaches a team a new skill for a few weeks, then leaves. One hands you a service. The other leaves you able to do it yourself.

Business

Kotter's 8-Step Change Model vs ADKAR Model

People think Kotter's eight steps and ADKAR are the same change model, and they are not. Kotter is a plan for the leaders of a whole company: urgency, a guiding team, a vision, quick wins. ADKAR follows one person through the change, from knowing why to making it stick. Many firms use both. Kotter steers the company, and ADKAR shows where each person is stuck.

Business

AI Strategy vs AI Governance

People think an AI strategy and AI governance are the same plan, and they are not. The strategy picks where AI should be used and what it should achieve. Governance sets the rules, owners and checks that keep each use safe and legal. A strategy without governance moves fast and takes risks. Governance without a strategy keeps a firm safe but going nowhere.

BusinessAIcross-domain

AI Acceptable Use Policy vs AI Governance

People think writing an AI use policy means a firm has AI governance, and it does not. The policy is one page of rules for staff: which tools, what data, when to check. Governance is the whole system behind it. It names who owns each AI system, keeps a list of them and reviews the risky ones. The policy is a piece of governance, not all of it.

BusinessAIcross-domain

AI Readiness Assessment vs AI Maturity Model

People use the two names as if they mean the same thing, and they do not. A maturity model is the ladder: the stages a firm can climb with AI. A readiness assessment is the check: it finds where the firm stands now and what it lacks. Many checks use a maturity model as their ruler. The model stays the same, and the check is run again each time.

Business

AI Incident vs Incident (IT)

People think an AI incident is an IT outage, and it is not. An IT incident is a service that breaks or slows down. An AI incident is harm caused by what an AI system did, such as a wrong answer someone acted on. Every server can be up while the harm happens. So the usual uptime alarms will not catch it.

BusinessInfrastructurecross-domain

SAML vs OpenID Connect

People think SAML and OpenID Connect do different jobs, and they do the same one. Both let one login open many apps. SAML is older and sends a signed XML note through the browser. OpenID Connect is newer, sends a signed JSON token, and suits phone apps and APIs. Most work apps take both.

Development

SCIM vs SAML

People think SCIM is another way to log in, and it is not. SAML signs a person in, each time they open an app. SCIM works ahead of time, making, changing and removing their accounts in each app. Single sign-on without SCIM can leave a past worker's accounts alive in every app, long after they have gone.

Development

Okta vs Auth0

People think Okta and Auth0 are one product, because one firm owns both, and they are not. Okta signs in a company's own staff to the work apps it buys. Auth0 adds a login to the apps a company builds, for its customers. A firm can use both at once, one for each job.

Development

Privileged Access Management vs RBAC

People think giving someone the admin role is the same as managing admin access, and it is not. RBAC decides which rights a role has. PAM decides when those rights are used, for how long, and who watches. With RBAC alone, an admin is an admin all day. With PAM, the power is lent for one job and then taken back.

Development

Langflow vs Flowise

People mix up Langflow and Flowise, because the names are almost the same. Both build AI agents by joining blocks on a canvas. Langflow is written in Python, and IBM now owns it. Flowise is written in Node.js, and its makers stopped work on it in 2026. A new project today should not start on Flowise.

Development

Microsoft Copilot Studio vs Microsoft 365 Copilot

People think Copilot Studio is one more Copilot to chat with, and it is not. Microsoft 365 Copilot is the AI helper in Word, Outlook and Teams. Staff chat with it. Copilot Studio is a tool to build your own agents. An agent built there can then run inside that helper.

DevelopmentAIcross-domain

PlantUML vs Mermaid

People think PlantUML and Mermaid are the same tool with two names, and they are not. Both turn text into diagrams. Mermaid runs in the browser, and GitHub draws it right inside Markdown. PlantUML runs on Java and covers more of UML, but GitHub will not draw it for you. Pick Mermaid for a README, and PlantUML for strict UML.

Design

Lucidchart vs draw.io

People think draw.io is just a free copy of Lucidchart, and that misses the real difference. Both make neat diagrams by dragging shapes. The key gap is where your file lives. Lucidchart keeps it in its own cloud and is built for live team editing. Its free plan has limits, so most teams pay. draw.io is free and saves the file wherever you choose, even in Git.

Design

CISA vs CISM

People think the CISA and the CISM are the same ISACA badge, and they are not. The CISA is for auditors, who check that the controls work. The CISM is for managers, who build and lead the security program. Both exams have 150 questions in four hours, so the job you want decides which one to take.

Business

CISM vs CISSP

People think the CISM and the CISSP prove the same thing, and they do not. The CISM tests how to lead security: plans, risk and budget. The CISSP is wider and covers the hands-on side too, across eight areas. Both ask for five years of work, and many security leaders hold both.

Business

CIPP/E vs CIPM

People think the CIPM is one more privacy law badge, and it is not. The CIPP/E tests what Europe's privacy law says. The CIPM tests how to run the work that follows from it: the team, the policies and the plan for a breach. Many people take a CIPP first, then add the CIPM.

Business

Trade Union vs Works Council

People think a works council is the union at their firm, and it is not. Workers join a union by choice, and it bargains for a whole trade. All the staff of one workplace elect a works council, members or not. The union signs deals for a whole trade. The council signs deals with one boss.

Business

AI Engineer vs Machine Learning Engineer

People think an AI engineer and a machine learning engineer do the same job, and they do not. The machine learning engineer trains models and keeps them running. The AI engineer takes a ready model and builds an app on top of it. One makes the bread, and the other makes the sandwich.

Business

Notified Body vs Market Surveillance Authority

People think a notified body is the state's watchdog, and it is not. A notified body is an outside firm a maker hires to check some high-risk AI before sale. The market surveillance authority is the state body that checks AI already on sale. It can ask for papers, order a recall and fine a firm. The first gives a pass. The second enforces the law.

AI

Fair Use vs Text and Data Mining Exception

People think the EU's data mining rule is Europe's version of fair use, and it is not. Fair use is a US rule where a court weighs four factors, case by case. The EU rule is a fixed exception for copying to find patterns. Anyone may use it, unless the owner opts out in a form a machine can read. A fair use win in the US says nothing about the EU.

BusinessAIcross-domain

GitHub Spec Kit vs Kiro

People think Spec Kit is GitHub's version of Kiro, and it is not. Both start from a written spec and break it into tasks. Spec Kit is a free set of steps and templates you add to the coding agent you already use. Kiro is a whole coding tool from AWS, with its own editor and agent. Pick Spec Kit to keep your tools, and Kiro to get it all in one.

AI

DeepEval vs Ragas

People think DeepEval and Ragas do the same job, and they mostly do not. Ragas is built to score apps that answer from your own documents. It checks the search and the answer apart. DeepEval is a test runner that works like pytest. It runs many kinds of metric and fails the build when a score drops. DeepEval even lists a Ragas score among its metrics.

AI

Perplexity vs NotebookLM

People think NotebookLM searches the web the way Perplexity does, and it does not. Perplexity looks up each question on the live web. It links to the pages it read. NotebookLM reads only the files, links and videos you put in a notebook. It links to the exact part it used. Ask Perplexity about the news. Ask NotebookLM about your own papers.

AI

Cursor vs GitHub Copilot

People think Cursor is a plug-in like GitHub Copilot, and it is not. GitHub Copilot is an add-on. You put it into the editor you already use. Cursor is a whole editor of its own, built from the code of a free editor most coders know. You install it on its own. Both suggest code as you type, and both have an agent. The real choice is to keep your editor or switch to a new one.

AI

Hermes /goal vs /loop

People think /loop is /goal on a timer. It is not. A goal runs until a judge model says the job is done, or until it runs out of turns. A loop sends the same message again every few minutes, done or not, until it hits its count or you stop it. Use a goal to finish a job. Use a loop to watch one.

AI

Hermes Skill vs Plugin

People think a plugin and a skill are the same kind of add-on. They are not. A skill is a written method the agent reads when a task needs it. A plugin is Python code that gives the agent new tools or hooks, and it stays off until you enable it. Use a skill to teach a way of working. Use a plugin to add a new ability.

AI

Hermes Command Approval vs Terminal Backend

People think command approval keeps Hermes in a sandbox. It does not. Approval checks a risky command before it runs and may ask you first. The terminal backend decides where every command runs: your own machine, a container or a server far away. Approval can let a bad command through. A sandbox keeps the harm away from your files.

AI

claude doctor vs /doctor

People think claude doctor and /doctor are the same check typed in two places, and they are not. claude doctor runs in your shell with no session open. It only reads, so it is the one to use when Claude Code will not start. /doctor runs inside a session. It checks more, and it offers to fix what it finds once you say yes.

AI

/fork vs /subtask

People think /fork and /subtask both send a helper off and wait for its answer, and they do not. /fork copies the chat into a new background session. You keep working, and the copy never reports back. /subtask hands one task to a helper that gets the whole chat. When it is done, its result comes back into the chat you are in.

AI

claude setup-token vs claude auth login

People think claude setup-token signs in the computer you run it on, and it does not. claude auth login signs you in here and keeps the sign-in. claude setup-token opens the browser once, then prints a token that lasts one year. It saves that token nowhere. You copy it into a CI job, or any machine with no browser.

AI

--safe-mode vs --restricted

People think --safe-mode and --restricted are both a locked-down mode, and they are not. --safe-mode turns off your add-ons, such as skills, plugins and hooks, to find the one that causes a fault. Your tools still work. --restricted is for a test harness on a shared machine. It takes away the tools that run commands. It reads only managed settings and what --settings passes.

AI

--mcp-config vs --strict-mcp-config

People think --mcp-config swaps your MCP servers for the ones in its file, and it does not. It adds those servers on top of the ones you already have. --strict-mcp-config is the flag that makes them the only ones. In CI, use the two together, so every machine gets the same servers.

AI

--settings vs --setting-sources

People think --settings and --setting-sources do one job, and they do not. --settings adds one more layer of settings from a file or a string, and its values win over your files. --setting-sources picks which of your own files load at all: user, project or local. A CI run often uses both.

AI

AI Office vs Market Surveillance Authority

People think the EU AI Office enforces the whole AI Act, and it does not. The AI Office is part of the EU itself. It watches the makers of models that do many tasks. Each EU country also names its own watchdog, called a market surveillance authority. That body checks the AI on sale or in use in its own country. It fines firms that break the rules there.

AI

Conformity Assessment vs AI Impact Assessment

People think the conformity check and the rights impact check in the EU AI Act are one test, and they are not. The maker runs the conformity check before the system is sold. It shows the system meets the rules on data, logs and safety. Some users run the impact check before they switch it on. Public bodies and firms that score credit must do it. It asks who the system could harm in their own hands.

AI

LiDAR vs Depth Estimation

People think only lidar can tell how far away things are, and that is not so. Lidar times flashes of light, so it is right to a few centimetres, even in the dark. Depth estimation guesses distance from camera pictures. It uses two views, or clues like size and shade. It is cheaper and sees colour, but its guesses are rougher.

AI

Robot vs Embodied AI

People think embodied AI is just a new name for a robot, and it is not. A robot is any machine that senses, decides and moves, even one that repeats the same fixed moves all day. Embodied AI is AI that has a body and learns by acting with it. Many factory robots have no AI in them at all.

AI

SLAM vs Path Planning

People think SLAM finds the way to the goal, and it does not. SLAM answers two questions: what is around me, and where am I on that map. Path planning takes the map and answers a third: how do I get from here to there without hitting anything. A robot needs both, and SLAM runs first.

AI

PermissionRequest vs PermissionDenied

People think PermissionDenied fires every time a call is refused, as the other half of PermissionRequest. It does not. PermissionRequest fires when Claude Code is about to ask you, and it can answer yes or no for you. PermissionDenied fires only when auto mode says no. Your own no, a deny rule or a hook's block never fires it. It cannot undo the no, but it can let Claude try again.

AI

CLAUDE_PROJECT_DIR vs CLAUDE_PLUGIN_ROOT

People think both point at the folder you are working in. Only one does. CLAUDE_PROJECT_DIR is the root of your project, so a hook in your settings uses it to find its script. CLAUDE_PLUGIN_ROOT is the folder a plugin was put in, so a plugin's hook uses it to find its own files. That folder can change on an update, so a plugin keeps data that must last somewhere else.

AI

Output Style vs CLAUDE.md

People think an output style and CLAUDE.md do the same job, since both shape every reply. They do not. CLAUDE.md tells Claude about your project, such as how to run the tests, and it loads whatever style you pick. An output style changes how Claude talks and works, such as shorter replies or notes on each choice. Put facts in CLAUDE.md and the way of working in a style.

AI

Claude Code Skill vs Plugin

People think a plugin is just a bigger skill. It is not. A skill is one folder of steps that Claude loads when a task needs it. A plugin is a package that can carry many skills, plus helpers, hooks and MCP servers, so a team can install the lot in one step. Write a skill first. Pack it in a plugin when others should get it too.

AI

REVIEW.md vs CLAUDE.md

People think REVIEW.md and CLAUDE.md do the same job, since Code Review reads both. They do not. CLAUDE.md tells Claude about the project, and every Claude Code session reads it. REVIEW.md tells Code Review what to flag and what to skip, and only that service reads it. The review command you run in a terminal reads CLAUDE.md but not REVIEW.md. So put review rules in REVIEW.md.

AI

Claude Code VS Code Extension vs JetBrains Plugin

People think the VS Code extension and the JetBrains plugin work the same way. They do not. The VS Code extension brings its own copy of Claude Code and shows it in a side panel. The JetBrains plugin brings no copy. It runs the claude command you installed in the editor's terminal and links to it. So the JetBrains plugin will not start until Claude Code is installed on its own.

AI

@claude vs @claude review

People think @claude and @claude review start the same thing. They do not. A plain @claude mention starts the GitHub Action, which needs a workflow file and runs on a GitHub runner. @claude review starts Code Review, a service with no workflow file. In a repo that has both, @claude review starts the Action too. So give the Action its own trigger phrase.

AI

Ollama vs llama.cpp

People think Ollama and llama.cpp are two rival engines, and they are not. llama.cpp is the engine. It loads a GGUF file and runs the maths, with a long list of flags to set. Ollama is a tool built on the same core. It finds a model by name, picks the settings for you and keeps a small server running. Pick llama.cpp for control, and Ollama for ease.

AI

Amazon Bedrock vs SageMaker AI

People think Amazon Bedrock and SageMaker AI are two names for the same AWS service, and they are not. Bedrock rents you models that someone else trained. You send a prompt and pay per token. SageMaker AI is for models you build yourself. You bring the data, train on rented GPUs and host the result. Pick Bedrock to use a model, and SageMaker AI to make one.

AI

Experiment Tracking vs Model Registry

People think the model registry is where training runs are logged, and it is not. Experiment tracking logs every run, good or bad, with its settings, data and scores. The registry keeps only the models someone chose to keep, with a stage such as testing or live. Each version links back to its run. Track to find the best model, and register it to ship it.

AI

MLOps vs LLMOps

People think LLMOps is a new field that replaces MLOps, and it is not. It is the same habit for apps built on a language model. The loop of test, ship and watch stays. What changes is the part you edit. In MLOps you retrain the model. In LLMOps you rarely do, so you version the prompt, test it with saved questions and trace each call.

AI

Project vs Program

People think a program is just a very big project, but it is not. A project makes one result and stops. A program is a set of linked projects with one goal. They need each other to reach it. One huge project is still one project.

Business

Project Manager vs Project Coordinator

People think a project coordinator is a project manager with less pay. The coordinator keeps the day in order. That means meetings, notes, files and late tasks. The project manager makes the big calls and answers for the result.

Business

PRINCE2 vs PMBOK Guide

People think PRINCE2 and the PMBOK Guide are rival versions of one method. They are not. PRINCE2 sets roles, stages and a board that says go. The PMBOK Guide describes good practice, and you pick the parts that fit.

Business

Planned Value vs Earned Value

People think earned value is the money earned or the work planned. It is neither. Planned value is the budget for the work that should be done by now. Earned value is the budget for the work that is really done. The gap between them shows if the project is behind.

Business

Three-Point Estimate vs PERT

People think a three-point estimate and PERT are the same thing. A three-point estimate blends three guesses for one task. PERT is a whole way to plan the dates, built on one blend of those guesses. You can use the first without the second.

Business

Project Baseline vs Project Schedule

People think the baseline is just the current schedule. The schedule changes every week as work moves. The baseline is the plan as it was agreed, frozen. The gap between the two is how far the project has slipped.

Business

BAC vs EAC

People think the EAC is just the budget. It is not. The BAC is the budget, the total of the plan, and it stays fixed. The EAC is today's guess of what the whole project will cost. It moves with every report.

Business

Power/Interest Grid vs Engagement Matrix

People think the power/interest grid and the engagement matrix are one chart. The grid sorts people by power and how much they care. The matrix marks how involved each one is now, and how involved they need to be. Use the grid to pick who matters, then the matrix to plan the work.

Business

Dictionary vs Set

People think a set is just a dictionary with no values, because in Python both are written with curly brackets. A dictionary pairs each key with a value, so you look a value up by its key. A set holds single values, each one once, and has nothing to look up. In Python, {} on its own makes an empty dictionary. An empty set is written set().

Development

Integer vs Floating-Point Number

People think 3 and 3.0 are the same number to a computer, and they are not. An integer is a whole number, and its sums are exact. A floating-point number has a decimal part, and it keeps a very close value rather than an exact one. That is why 0.1 + 0.2 can give 0.30000000000000004. For money, count in whole cents with integers.

Development

Runtime Error vs Logic Error

People think a program that runs to the end with no message is working, and it may not be. A runtime error stops the program while it runs and names the line, such as a divide by zero. A logic error stops nothing. The program finishes and the answer is just wrong. Only a check of the result will catch it.

Development

High-Level Language vs Low-Level Language

People think a high-level language is the harder kind, made for experts, when in fact it is the other way round. High level means far from the machine and close to plain words, like Python. It is the easier kind to write. Low level means close to the chip, like C or assembly. You control memory and speed, and it takes more care.

Development

Shell Script vs PowerShell

People think PowerShell is just the Windows version of a Bash shell script, and it is not. Both run commands one after another from a file. A Bash command hands the next one plain text, which has to be picked apart. A PowerShell command hands over objects with named parts, such as a file's name and size. PowerShell now runs on macOS and Linux too.

Development

Go vs Rust

People think Go and Rust are rivals for the same job, since both are new, fast and compiled. They aim at different things. Go keeps the language small and simple, and cleans up memory for you, so a team can write servers quickly. Rust checks every use of memory before it builds, for code that must be as fast as C and still safe.

Development

Promise vs Async/Await

People think async and await replace promises for good, and they do not. An async function still hands back a promise. The word await is a tidier way to wait for one, so the code reads from top to bottom. Under the hood it is the same promise. The then and catch calls just turn into plain lines inside a try and catch.

Development

Type Annotation vs Type Inference

People think a typed language makes you write a type on every name, and it does not. A type annotation is a type you write yourself, such as count: number. Type inference is the language working the type out from the value, so let count = 0 is already a number. Most code uses both, with notes where they help the reader.

Development

Inheritance vs Polymorphism

People think inheritance and polymorphism are one idea. They are not, though they often show up together. Inheritance is how one class gets the parts of another. Polymorphism is one method name that does the right thing for each kind of object. Python can have it with no parent class at all. Each object just needs the method.

Development

Generics vs Union Type

People think generics and union types both just mean any type. A union type lists a few kinds a value may be, such as a string or a number, and you must check which one you have. A generic stands for one type that the caller picks. That type is then kept exact all the way through, so a list of strings gives back a string.

Development

Git Hook vs Branch Protection

People think a pre-commit hook keeps bad code off main, and it does not. A Git hook runs on one person's computer. A clone does not copy it, and git commit --no-verify skips it. Branch protection runs on the Git host. It locks main until reviews and tests pass, and nobody can skip it unless they are allowed to.

Development

Squash Merge vs Fast-Forward Merge

People think a squash merge and a fast-forward are the same, because both leave main in a straight line. They are not. A fast-forward makes no new commit. Main's label just slides onto the branch's own commits, so every one of them is kept. A squash merge makes one new commit for the whole branch, and the branch's own commits stay off main.

Development

Git Tag vs Branch

People think a tag is just another kind of branch, since both are names on a commit. A branch moves forward each time you commit on it. A tag never moves, so v1.0 points at the same code years later. Tags also need a push of their own, because a plain git push does not send them.

Development

Code Editor vs IDE

People think VS Code is an IDE, and it is not quite. A code editor writes plain text with help on top, and add-ons give it more skills. An IDE has running, testing and deep search of the whole project built in from the start. A code editor with many add-ons comes close to an IDE. An IDE is still heavier, and it reads a big project more deeply.

Development

Vim vs Neovim

People think Neovim is a new editor with its own keys, and it is not. It was built from Vim's code, so the modes and keys are the same, and a Vim user feels at home. What changed is the setup. Neovim reads a short script in a file called init.lua, and it talks to language tools for hints and errors.

Development

Software Package vs Library

People think a package and a library are the same thing, and they are not. A library is the code itself, such as a set of date tools. A package is the box it ships in, with a name, a version number and a list of what it needs. Most packages hold a library, but some hold a tool you run instead.

Development

Base64 vs Encryption

People think Base64 hides data, and it does not. Base64 only writes bytes with 64 safe letters and signs, so they can travel as text. It has no key, and anyone can turn it back in one step. Encryption locks data with a key. Without the key, the bytes cannot be read. Use Base64 to carry data, and encryption to keep it secret.

Development

.env File vs Config File

People think a .env file is just one more config file, and it is not. A config file holds settings that are the same for everyone, so it goes into Git with the code. A .env file holds the secrets and values for one machine. It must stay out of Git, and each machine keeps its own copy.

Development

ASCII vs UTF-8

People think every plain text file is ASCII, and it is not. ASCII has only 128 codes, for basic English letters, digits and signs. UTF-8 can store every character in every language, plus emoji. Its first 128 codes are the same as ASCII, so plain English text is both. Add one é or one emoji, and the file is UTF-8, not ASCII.

Development

HTML vs DOM

People think the DOM is the HTML file, and it is not. The HTML file is the text the server sends. The DOM is the live tree the browser builds from it. Scripts change the DOM, not the file. So the page on screen can differ from the page source.

Development

Transpiler vs Polyfill

People think a transpiler can make any new feature work in an old browser, and it cannot. A transpiler rewrites how your code is written, such as taking out types. A polyfill adds a function the old browser does not have. Many sites need both.

Development

Vite vs webpack

People think Vite and webpack do different jobs, and they do the same one. Both are bundlers that pack your files for the browser. webpack bundles even while you work. Vite skips that step until the final build, so it starts much faster.

Development

Node.js vs Express

People think Node.js is a framework like Express, and it is not. Node.js is the program that runs JavaScript outside the browser. Express is a small framework that runs on Node.js and adds routes and middleware. You need Node.js to use Express, but not the other way round.

Development

API Key vs Access Token

People think an API key and an access token are the same thing, and they are not. An API key names your app, and it lasts until someone turns it off. An access token is short-lived, and it often acts for one user who logged in. Both go in a header, and both must stay secret.

Development

API vs SDK

People think an SDK is the API, and it is not. The API is the set of requests a service will answer, and the replies it sends back. An SDK is a library you add to your code that sends those requests for you. You can always call the API without the SDK, but not the other way round.

Development

Endpoint vs REST Resource

People think an endpoint and a resource are the same, and they are not. A resource is a thing with an address, such as /orders/42. An endpoint is a method plus a path, such as GET /orders/42. So one resource usually has several endpoints, one for each thing you can do to it.

Development

OpenAPI vs API Documentation

People think OpenAPI is the docs page, and it is not. OpenAPI is one file that lists each path and reply in a fixed shape, so tools can read it. API docs are the guide people read. A docs page is often built from that file, plus words a person wrote.

Development

gRPC vs tRPC

People think tRPC is gRPC for TypeScript, and it is not. gRPC sends small binary messages between services written in many languages. tRPC sends JSON, and it only works when both ends are TypeScript and share the same types. The names differ by one letter, and the jobs differ a lot.

Development

REST API vs gRPC

People think gRPC is just a faster REST, and it is not. REST gives each thing an address and uses HTTP methods on it, with JSON most people can read. gRPC calls named functions and sends packed bytes. It is faster between services, but a browser cannot call it without help.

Development

Income Statement vs Cash Flow Statement

People think the income statement shows the cash a firm made, and it does not. It counts a sale on the day it is made, and a cost when it is used. The cash flow statement counts money only when it moves. A firm can show a profit on the first and lose cash on the second.

Business

Balance Sheet vs Income Statement

People read both as a report on the year, and only one of them is. The income statement covers a period, such as a month or a year, and ends in profit. The balance sheet is a picture of one day. It lists what the firm owns and owes on that day.

Business

EBITDA vs EBIT

People quote EBITDA as if it were profit, and it is not. EBIT is profit from the main work, after the yearly wear on machines is taken off. EBITDA adds that wear back. So EBITDA is always the bigger number, and the machines still have to be paid for one day.

Business

Deferred Revenue vs Revenue

People think money a customer pays up front is revenue, and it is not yet. Until the work is done, the firm owes it back, so it is deferred revenue, a debt. It turns into revenue bit by bit, as the work is done. Count it all at once and the profit is too high.

Business

Contribution Margin vs Gross Margin

People think contribution margin is gross margin by another name, and it is not. Gross margin takes away the cost of the goods sold. Contribution margin takes away each cost that grows with sales, like card fees and shipping too. You need it to find the break-even point.

Business

Trial Balance vs Balance Sheet

People think a trial balance is a balance sheet, because both balance. A trial balance is a check list of every account's total, made before the reports. It proves the debits equal the credits. The balance sheet is a finished report of what a firm owns and owes on one day.

Business

Free Cash Flow vs Operating Cash Flow

People think operating cash flow is the cash a firm is free to spend, and it is not. It is the cash the main work brings in. Free cash flow takes away what the firm spends on machines and buildings. Only what is left can pay down debt or go to the owners.

Business

ROI vs IRR

People think ROI and IRR are the same number, because both are a percent. ROI is the total gain divided by the cost, and it does not care how long it took. IRR is a yearly rate, and it counts when each dollar comes back. A 50% ROI over ten years is a low IRR.

Business

Leveraged Buyout vs Private Equity

People think private equity and a leveraged buyout are the same thing, and they are not. Private equity is a kind of firm. It buys companies with money from big investors. A leveraged buyout is one way it buys them, mostly with a loan the bought firm repays. Some of its deals use little debt.

Business

Sunk Cost vs Opportunity Cost

People think both are money a choice will cost, and only one is. A sunk cost is already spent, and it is gone whatever you choose, so it is left out. An opportunity cost is the best thing you give up by choosing. It belongs in every choice.

Business

S&P 500 vs Dow Jones Industrial Average

People think the S&P 500 and the Dow measure the same thing, and they do not. The S&P 500 holds 500 large US firms, and the biggest firms count the most. The Dow holds just 30, and a firm with a high share price counts the most. So on the same day they can move apart.

Business

EPS vs P/E Ratio

People mix up EPS and the P/E ratio, because one is built from the other. EPS is a firm's profit for each share, in dollars. The P/E ratio is the share price divided by that EPS. EPS says how much a share earns, and P/E says how much buyers pay for those earnings.

Business

Bond Yield vs Interest Rate

People think a bond's yield is the interest it pays, and it is not. The interest is fixed on the day the bond is sold, such as $40 a year on $1,000. The yield is what a buyer earns at today's price. When the price falls, the yield rises, while the interest stays the same.

General

Public Company vs Private Company

People think a public company is owned by the government, and it is not. Public means anyone can buy its shares on a stock exchange. A private company's shares are held by a few owners and are not sold to the public. Some private firms are very large, and some public ones are small.

Business

Concierge MVP vs Wizard of Oz Test

People think a concierge MVP and a Wizard of Oz test are the same, since a person does the work by hand in both. They are not. In a concierge MVP the customer sees the founder doing it, face to face. In a Wizard of Oz test the person is hidden, and the customer thinks a product is doing the work. So only the second shows how people will use the real thing.

BusinessDesigncross-domain

Entrepreneur vs Founder

People use the two words as if they mean the same thing, and they do not. An entrepreneur is anyone who starts a business and takes its risk, from a baker to an app maker. A founder is a title tied to one company. It means you started that company and owned part of it from day one. Every founder is an entrepreneur, but only the word founder names a company.

Business

Startup Studio vs Startup Accelerator

People think a startup studio backs teams who come in with their own idea, the way an accelerator does. It does not. An accelerator takes in founders who already have an idea, gives them cash for a slice, and sends them to demo day in about three months. A studio comes up with the idea itself, builds the first version, then finds a founder to run it. It keeps a much larger share.

Business

Competitive Advantage vs Economic Moat

People think every edge over rivals is a moat, and it is not. An advantage is any edge that lets a firm win today, like a lower price or a faster service. A moat is an edge that rivals cannot copy for many years. A network effect is one. High switching costs are another. Most edges fade once rivals catch up. Only the ones that last are moats.

Business

Business Model vs Revenue Model

People think the business model is just the way a company charges, and it is not. The revenue model is only that part: a one-time sale, a monthly fee, ads, or a cut of each sale. The business model is the whole thing. It covers who the buyers are, what they get, how they are reached, and what it all costs. The revenue model is one box inside it.

Business

Business Model vs Business Strategy

People use the two words as if they mean the same thing, and they do not. A business model says how the firm makes money. It covers who pays, for what, and what it costs. A strategy says how the firm will beat its rivals. It names the market, the edge, and what the firm will not do. Two rivals can share one model and still have very different plans.

Business

BCG Matrix vs Ansoff Matrix

People think the two charts do the same job, and they do not. Both have four boxes. The BCG matrix sorts the products a firm already sells, by how fast each market grows and how big its share is. It shows where the cash should go. The Ansoff matrix looks ahead. It maps four ways to grow, from selling more to the same buyers to new things for new buyers.

Business

Core Competency vs Competitive Advantage

People think a core skill and an edge over rivals are the same thing, and they are not. A core competency is a mix of skills deep inside the firm, like making small engines very well. The advantage is what buyers see in the market, like a cheaper or better product. One skill can sit under many products. The edge is what it wins for each one.

Business

Indirect Competitor vs Substitute Product

People mix up these two kinds of rival, and they are not the same. An indirect rival sells a different thing for the same need. It is often in your own trade, like a bakery that sells coffee. A substitute comes from a whole other industry, like tea made at home. Both take buyers away, but they need different answers.

Business

Competitor Profile vs Battlecard

People call a competitor profile a battlecard, and they are not the same. A profile is a file of two or three pages on one rival, used for planning. It covers what they sell, their prices, their buyers and their news. A battlecard is one page cut from it, for a seller on a live call. It holds only what wins a sale.

Business

Competitive Benchmarking vs Feature Comparison Matrix

People think benchmarking means filling in a feature table, and it does not. A feature table lists what each product has, with a tick or a cross in each cell. Benchmarking measures how well a firm does on a few numbers buyers care about, like days to deliver or a rating. One shows what you offer. The other shows how well you do it.

Business

Competitive Intelligence vs Market Research

People think these two are one job, and they are not. Market research studies buyers. It asks what they need, what they would pay, and why they choose. Competitive intelligence studies rivals. It asks what they sell, what they charge, and what they plan next. The two share tools like surveys, but they point at different people.

BusinessMarketingcross-domain

Risk Avoidance vs Risk Mitigation

People think avoiding a risk is just a strong way to mitigate it. Mitigation keeps the plan and makes the risk smaller, so some of it is still there. Avoidance changes the plan so the risk cannot happen at all. Avoiding a risk often costs more, because part of the plan goes with it.

Business

EMV vs Probability and Impact Matrix

People think the score on a risk grid tells you what a risk will cost. The grid gives each risk a rank, such as 4 for chance times 5 for impact, to sort the list. EMV turns a risk into money, such as a 20% chance of losing 50,000 being worth minus 10,000. Use the grid to pick what comes first and EMV to size the money.

Business

Contingency Plan vs Contingency Reserve

People think the contingency plan and the contingency reserve are one thing. The plan is the steps the team will take if a risk happens, such as calling the backup hotel. The reserve is the money set aside to pay for steps like that. A risk can have a plan and draw on the reserve, or have one without the other.

Business

Product Backlog vs Sprint Backlog

People think the Sprint Backlog is just the top few items of the Product Backlog. The Product Backlog is the whole ordered list for the product, and the Product Owner keeps it. The Sprint Backlog is the plan for one sprint. It holds the Sprint Goal, the items picked for it and the steps to finish them. The Developers own it and change it every day.

Business

Product Goal vs Sprint Goal

People think the Sprint Goal is a small piece of the Product Goal, and it is not just that. The Product Goal is the long-term aim for the whole product, and it can last for many sprints. The Sprint Goal is the one aim of a single sprint, set at Sprint Planning. Each Sprint Goal should move the product a step toward the Product Goal.

Business

Scrum vs Scrumban

People think Scrumban is just Scrum with a board on the wall. Many Scrum teams use a board, and that does not make it Scrumban. Scrumban adds Kanban's limits on work in progress, and people pull work only when they have room. Planning often happens when the to do column runs low, not on a fixed day.

Business

PMI-ACP vs PMP

People think the PMI-ACP is the agile version of the PMP, and it is not. It is its own badge with its own rules. It asks for 2 years of agile work in the last 5, and has an exam of 120 questions. The PMP asks for years of leading projects. About 60 percent of its exam is already agile or hybrid work. One does not replace the other.

Business

CSPO vs PSPO

People think the CSPO and the PSPO are the same badge from two places. They prove the same role, the Product Owner, but are earned in different ways. The CSPO comes from Scrum Alliance through a course with a trainer, and it must be renewed every 2 years. The PSPO comes from Scrum.org through a hard online test alone, and it lasts for life.

Business

PRINCE2 Foundation vs PRINCE2 Practitioner

People think PRINCE2 Practitioner is just a harder Foundation exam. Foundation checks that you know the method, with 60 questions and no book. Practitioner checks that you can use it on a real project. Its exam is open book, worth 70 marks over 150 minutes, and a PMP or CAPM can take it without Foundation.

Business

Database View vs Database Table

People think a view is a copy of a table, and it is not. A table holds rows of its own on disk. A view holds no rows at all. It is a saved query with a name, and it runs fresh each time you read it. Change a row in the table, and the view shows the change at once.

Development

Document Database vs Key-Value Store

People think Redis and MongoDB are the same kind of NoSQL, and they are not. A key-value store keeps one value under each name and finds it only by that name, very fast. A document database keeps whole records, each a JSON-like document, and can search inside them. Redis is often a quick cache. MongoDB is often the main store.

Development

SQLite vs PostgreSQL

People think SQLite is a toy version of Postgres, and it is not. It is a different shape. SQLite has no server, and the whole database is one file inside your app. PostgreSQL is a server that many apps and users connect to at once. SQLite fits a phone app or a small site. Postgres fits a busy app with many writers.

Development

Manual Testing vs Automated Testing

People think automated tests mean nobody has to test by hand, and that is not so. A script repeats the same check the same way, fast, on every change. A person notices what feels wrong, like a confusing page or a button in an odd place. A script only checks what someone wrote down, so most teams keep both.

Development

Acceptance Test vs User Acceptance Testing (UAT)

People think user acceptance testing is just more acceptance tests, and it is not. An acceptance test is written by the team, often as code, to check a feature does what was asked. UAT is the last step before release, when real users or the client try the app by hand and sign it off. It finds a feature that works but is not what they needed.

Development

Playwright vs Cypress

People think Playwright and Cypress are one tool with two names, and they are not. Cypress runs inside the browser, beside the app. Its tests are in JavaScript or TypeScript only. Playwright drives the browser from outside. It works in all three main browser engines, and it also takes Python, Java and .NET.

Development

Selenium vs Playwright

People think Selenium and Playwright are old and new names for one tool, and they are not. Selenium dates from 2004 and talks to each browser through the WebDriver standard, in languages such as Java, C# and Ruby. Playwright came out in 2020. It waits for each button to be ready by itself, so its tests tend to break less.

Development

Spy vs Mock

People think a spy is just another name for a mock, and it is not. A mock is told up front which calls to expect, and it fails the test if they do not come. A spy only writes down every call. The test then looks at that record after the code has run and decides what to check.

Development

Fake vs Stub

People think a fake is the same as a stub, and it is not. A stub hands back a fixed answer, whatever you ask it. A fake really works, just in a small way, such as a pretend bank that keeps balances in memory. Put money in a fake and you can take it out again. A stub cannot do that.

Development

Continuous Integration (CI) vs Continuous Delivery (CD)

People think CI and CD are one step, because they are written together as CI/CD, and they are not. CI checks each change: it builds the code and runs the tests on every push. CD comes after it and takes a passing change toward the live site. You can have CI with no CD, but not CD with no CI.

Development

Rolling Update vs Blue-Green Deployment

People think a rolling update is the same as blue-green, since both keep the site up, and it is not. A rolling update swaps the servers a few at a time. For a while, old and new run side by side. Blue-green builds a whole second copy and moves every user at once. It costs more, but going back takes seconds.

Development

Canary Release vs Feature Flag

People think a canary release is the same as a feature flag, since both can start with 5% of users, and it is not. A canary sends some users to a new version of the whole app, on its own servers. A flag is a switch inside the code that turns one feature on for some users. A canary tests a deploy. A flag tests a feature.

Development

Feature Flag vs Rollback

People think turning off a feature flag is a rollback, and it is not. A rollback puts the whole last version back on the servers, which takes a minute or two. Turning off a flag hides one feature in seconds and leaves the rest of the deploy in place. Use the flag when one feature is the problem, and roll back when the whole deploy is.

Development

Transformational vs Servant Leadership

People think transformational and servant leadership are the same, and they are not. Both care about people. A transformational leader rallies the team behind a goal the leader sets out. A servant leader starts from what the team needs and works for them. The first leads from the front. The second clears the way from behind.

Business

Chief Operating Officer vs Chief of Staff

People think a chief of staff is just another COO, and it is not. A COO owns the daily work of the whole company, with teams that report to them. A chief of staff owns no area and often has no team. They work for one leader, most often the CEO, and make that leader's time go further.

Business

Employee Retention vs Employee Turnover

People think retention and turnover always add up to 100%, and they do not. Retention asks how many of the people you had at the start of the year are still there. Turnover counts all who left against the average staff count, new hires too. A firm can keep most of its old staff and still have high turnover, if new people leave fast.

Business

Theory X and Theory Y vs Theory Z

People think Theory Z is McGregor's third theory, and it is not. Douglas McGregor wrote Theory X and Theory Y in 1960. They are two beliefs a boss can hold about workers. William Ouchi wrote Theory Z in 1981. It is a way to run a whole firm, mixing US and Japanese habits.

Business

Functions of Management vs Mintzberg's Managerial Roles

People think the four functions of management describe a manager's day, and they do not. The functions list the jobs that must get done: plan, organize, lead and check. Mintzberg's roles describe what managers were seen doing. The work came in short, broken pieces, and most of it was talk.

Business

Hawthorne Studies vs Hawthorne Effect

People think the Hawthorne effect is what the Hawthorne studies proved, and it is not quite. The studies ran for eight years and found that the work group and how people felt shaped output. The effect is a later name for one part: people change when they know they are watched. Later checks found that part weaker than the story says.

BusinessGeneralcross-domain

Decision Maker vs Economic Buyer

People think the decision maker and the economic buyer are one person, and often they are not. The decision maker weighs what the others say and makes the final pick. The economic buyer holds the budget and can approve the spend or stop it. In a small firm the owner is both. In a large one, a seller who wins the first can still lose to the second.

Business

BANT vs MEDDIC

People think MEDDIC is just a longer BANT, and it does a different job. BANT asks four quick questions to see if a lead is worth a call: budget, authority, need and timing. MEDDIC checks a large deal from the inside. It asks who holds the money, how the firm decides, and who will argue for you. BANT sorts leads. MEDDIC tests big deals.

Business

Sales Quota vs Sales Forecast

People treat the quota and the forecast as one number, and they are not. The quota is the target: what a seller is asked to sell this quarter. The forecast is the guess: what the team thinks it will really sell, from the deals it has. When the forecast is below the quota, the gap shows early. Writing the quota in as the forecast hides it.

Business

Product Returns vs Refund

People say return and refund as if they were one thing, and they are not. A return is the goods coming back to the seller. A refund is the money going back to the buyer. One can happen without the other. A store can swap a returned shirt for a new size and refund nothing. It can also refund a broken mug and tell the buyer to keep it.

Business

Dropshipping vs Print on Demand

People call print on demand a kind of dropshipping, and it is not quite that. In dropshipping, a supplier ships a finished product that someone else made and that many stores sell. In print on demand, the item is made only after the order, with your own design on it. Both mean no stock at home. Only print on demand gives you a product no other store has.

Business

Digital Wallet vs Crypto Wallet

People hear wallet and think of crypto, and the two are not the same. A digital wallet keeps your bank cards on your phone, so you can pay at a checkout with a tap. A crypto wallet holds the keys to coins on a blockchain. A store that adds a digital wallet button still takes normal card money. Nothing about it involves crypto.

BusinessDevelopmentcross-domain

Social Commerce vs Influencer Marketing

People think social commerce is the same as influencer marketing, and it is not. Influencer marketing pays a person with a big crowd of fans to show your product. Social commerce is the sale itself, made inside the app. Often the buyer never leaves it. An influencer can send buyers to any store. Social commerce needs a shop inside the app.

BusinessMarketingcross-domain

Reservation Price vs Target Point

People treat their target and their walk-away price as one number, and they are not. The target is the deal you hope for, set high and backed by facts. The walk-away price, or reservation price, is the worst deal you will still take. You aim at the first and refuse to go past the second. Someone who opens at the walk-away has nowhere left to move.

Business

Principled vs Integrative Negotiation

People think principled negotiation is just a name for a win-win deal, and it is not. A win-win deal is a kind of deal. It has more than one issue, so the pie can grow. Principled negotiation is a method with four rules, and you can bring it to any deal. Its fair standards help even when there is only one price to settle.

Business

BATNA vs ZOPA

People mix up the BATNA and the ZOPA, and they are not the same. Your BATNA is your own plan B, what you do if this deal fails. The ZOPA is the range where a deal can happen, between the two sides' walk-away prices. Each side has its own BATNA. There is only one ZOPA, and sometimes there is none at all.

Business

Clean Architecture vs Hexagonal Architecture

People think clean and hexagonal architecture are rivals, and they are not. Both keep the business rules at the center. The database and the web depend on the rules, never the other way. One names the plugs: ports, and adapters that fill them. The other draws rings and says code points only inward. The rule is the same.

Development

Stateless Service vs Stateful Service

People think a stateless service keeps no data, and that is not quite right. It keeps none inside itself between requests. The cart or the login lives in a shared store, so any copy can answer. A stateful service keeps that data in its own memory or disk. Each request must reach the one copy that has it.

Development

Circuit Breaker vs Retry with Backoff

People think a circuit breaker is just a limit on retries, and it is not. A retry tries one failed call again, waiting longer each time. A circuit breaker watches all calls to a service. Once too many fail, it stops sending any for a while. Retries help with a short blip. The breaker helps when a service is truly down.

Development

HttpOnly vs SameSite

People think HttpOnly stops forged requests, and it does not. HttpOnly hides a cookie from scripts on the page, so an XSS script cannot steal the login. The browser still sends the cookie everywhere it goes. SameSite decides where the cookie goes. It keeps the cookie off requests that start on other sites, which is what stops CSRF. A login cookie wants both.

Development

Input Validation vs Output Encoding

People think checking what users type stops cross-site scripting, and it does not. Input validation turns away data with the wrong shape as it comes in. But a comment box has to take almost any text. Output encoding works on the way out. It turns signs like the less-than sign into safe codes, so the browser shows the text and never runs it. You want both.

Development

Job Story vs User Story

People think a job story is a user story with new words, and it is not. A user story starts from a type of user and is a piece of work to build. A job story starts from a moment: when this happens, I want to do this, so I can get that. It names no product, so it shows the need before anyone picks a fix.

Business

Product Roadmap vs Product Strategy

People think the roadmap is the strategy, and it is not. The strategy makes the choices: which users to win, which problem to solve, and what to skip. The roadmap is the order of work that follows from those choices. A team with only a roadmap knows what comes in March. It does not know which item to drop when time runs short.

Business

Product Analytics vs User Feedback

People think user feedback shows what users do, and it does not. Feedback is what people say, in chats, reviews and surveys, so it shows what they feel and want. Product analytics counts what they really do inside the product. The two often disagree. Trust the pattern where they agree, and ask more questions where they do not.

Business

SMART Goals vs Stretch Goal

People think a SMART goal and a stretch goal pull the same way, and they do not. A SMART goal is clear: it says what, how much and by when, and it should be within reach. A stretch goal is set above what the team expects to hit, on purpose. A goal can be both clear and bold, but scoring 70% means failure for one and success for the other.

Business

Goal Cascading vs Hoshin Kanri

People think hoshin kanri is just goal cascading with a Japanese name, and it is not. In plain cascading, a goal is cut smaller and handed down, level by level. In hoshin kanri, leaders pick only a few goals, and each team sends back what it can do and what it needs. The plans go up and down until they fit. Progress is checked often.

Business

BHAG vs Vision Statement

People think a BHAG is a vision statement, and it is not. A vision says what the firm stands for and the world it wants, with no finish line. A BHAG is one huge goal, 10 to 30 years away, so clear that everyone will know the day it is reached. The vision is the direction. The BHAG is one bold mark on the way there.

Business

Cross-Platform App vs Progressive Web App

People think a cross-platform app is just a website on a phone, and it is not. A cross-platform app is built once in a framework like Flutter or React Native. It becomes a real app for each phone and goes into the stores. A progressive web app is a website that you add to your home screen. It needs no store, and it reaches fewer parts of the phone.

Development

App Lifecycle vs Software Development Life Cycle

People think the app lifecycle is the steps of building and shipping an app, and it is not. That is the software development life cycle: plan, build, test, release. The app lifecycle is what one running app goes through on a phone. It opens, goes to the background, is frozen, and may be ended to free memory. One is about the team, the other about the app.

Development

Heap vs Binary Search Tree (BST)

People think a heap keeps its items sorted, like a binary search tree. It does not. A search tree keeps every value in order, so you can find any one fast. A heap only keeps the smallest item on top. Pick a heap when you only need the next smallest, and a search tree to look up any value.

Development

Greedy Algorithm vs Dynamic Programming

People think a greedy algorithm always finds the best answer, and it does not. Greedy takes the best-looking step now and never goes back. Dynamic programming works out the best answer for each smaller case first, then builds on them. Greedy is faster when it works. With coins of 4, 3 and 1, greedy makes 6 with three coins, and dynamic programming finds two.

Development

Bytecode vs Machine Code

People think bytecode is machine code, and it is not. Machine code is the steps a chip runs directly, so it only works on one chip family. Bytecode is a list of simpler steps for a virtual machine, which a program reads and runs. That makes it work on any chip that has the virtual machine. Python, Java and C# all make bytecode first.

Development

Profiling vs Benchmark

People think a benchmark tells you what to fix, and it does not. A profiler watches a whole program run and shows where the time goes. A benchmark times one piece of code many times, so you can compare two versions. Profile first to find the slow part. Then benchmark the old and new versions of that part.

Development

Selling Concept vs Marketing Concept

People think marketing is just selling, and it is not. The selling concept starts with what the firm already made and pushes buyers to take it. The marketing concept starts with what buyers need and makes the product to fit. Both use ads. The difference is where the firm looks first.

Marketing

Customer Value vs Customer Satisfaction

People think a buyer who saw good value will end up pleased, and that is not a given. Customer value is judged before the sale. It is what you get against all it costs. Satisfaction is judged after. It is what the product did against what you expected. A big promise raises the value a buyer sees, then makes satisfaction harder to reach.

Marketing

Marketing Funnel vs Customer Journey

People draw the marketing funnel and the customer journey as one picture, and they are not. The funnel is the seller's view. It counts how many people move from stage to stage, and it ends at the sale. The customer journey is the buyer's view. It follows what one person does and feels at each touchpoint, and it keeps going after the sale.

Marketing

Focus Group vs In-Depth Interview

People treat a focus group as a fast way to do many interviews, and it is not. A focus group puts 6 to 10 buyers in one room. There you hear them react to each other. One loud voice can steer the rest. An in-depth interview is one buyer alone with a researcher. It is slower, but people say private things they would never say in a group.

Marketing

Anchoring Bias vs Decoy Effect

People call any price trick an anchor, but a decoy works another way. Anchoring is the pull of the first number you see. A high first price makes every later price feel low. The decoy effect needs a whole extra option. It sits close to one choice but is plainly worse, so that choice looks like a bargain next to it.

General

Cost-Plus Pricing vs Value-Based Pricing

People think a fair price comes from cost, but buyers do not see your costs. Cost-plus pricing adds a fixed share to what a product costs to make. Value-based pricing starts from what the buyer gains, then sets the price below that. Cost sets the lowest price you can charge. Value sets the highest price a buyer will pay.

Marketing

Marketing Mix vs 7 Ps

People treat the 4 Ps and the 7 Ps as two rival models, and they are not. The 4 Ps are product, price, place and promotion. The 7 Ps keep all four and add three for services: people, process and physical evidence. Use the four for a product you can hold. Use all seven when staff and steps shape what the buyer gets.

Marketing

Branded House vs House of Brands

The names sound alike, but they are opposite ways to name products. A branded house puts one master name on every product, so trust and risk are shared. A house of brands gives each product its own name, and the owner stays out of sight. The first is cheaper to grow. The second keeps one failure from hurting the rest.

Marketing

Brand Extension vs Line Extension

People call any new product from a known brand an extension, but there are two kinds. A line extension adds a new flavour, size or version of the same kind of product. A brand extension takes the name into a new kind of product. The first can eat old sales. The second fails when the name does not fit.

Marketing

Content Pillar vs Pillar Page

Both use the word pillar, so people think they are one thing. A content pillar is a theme in the plan, one of the three to five subjects a brand keeps making content about. A pillar page is one long web page that covers a subject and links to deeper pages. One theme can hold several pillar pages.

Marketing

Content Audit vs Content Gap

People run one and think they have done the other. A content audit looks at what you already have, and decides to keep, update, merge or remove each piece. A content gap is a topic your readers want that you never covered. The audit cleans up the past. The gaps tell you what to write next.

Marketing

Code Search vs Semantic Code Search

People think a search by meaning replaces a search by text, and it does not. Text search finds the exact word or pattern you type, in every file. Semantic search turns code and your question into numbers and finds the closest match, even with no word in common. Use text search when you know the name. Use semantic search when you only know what the code does.

Development

Game Asset vs Game Object

People think a game asset and a game object are the same thing, and they are not. An asset is a file on disk, such as a picture or a sound. An object is a thing placed in a scene, such as a coin, built from parts. The object points to the asset. So 200 coins can share one picture, and changing that one file changes every coin.

Development

Sprite vs Hitbox

People think a game checks hits against the picture you see, and it does not. The sprite is only what gets drawn. The hitbox is a simple unseen shape around it, and the game tests touches with that. A hero's hitbox is often smaller than its sprite, so a spike that grazes its hair does not count. That is what makes a game feel fair.

Development

Unity vs Unreal Engine

People think Unity and Unreal Engine are two names for the same kind of tool, and for most games they are not a free swap. Unity is scripted in C# and is common for phone games and small studios. Unreal uses C++ and Blueprints and is known for lifelike 3D. Unreal takes 5 percent of what a game earns past 1 million dollars. Unity charges a yearly fee per seat instead.

Development

Embedded System vs IoT

People think every embedded system is an IoT device, and it is not. An embedded system is a small computer built into a device to do one job. The chip in a microwave is one. It becomes part of the Internet of Things only when it goes online, so other things can talk to it. Going online adds a second job: keeping that link safe.

Development

Smart Contract vs dApp

People use smart contract and dApp as if they were the same thing, and they are not. A smart contract is one program on a blockchain. It holds the rules and the money. A dApp is the whole app a person uses: a web page, a wallet and one or more contracts behind them. The contracts cannot change, but the web page can be faked.

Development

Seed Phrase vs Wallet Address

People think the seed phrase and the wallet address are two codes for the same account, and one of them must never be shared. The address is public. You give it out so people can pay you. The seed phrase is a list of words that rebuilds every key in the wallet. Anyone who has it can take every coin, so it never leaves your paper.

Development

Point Size vs Cap Height

People think the point size of a font is the height of its capitals, and it is not. The cap height is how tall a capital H stands. The point size is the height of the whole body each letter sits in. That body has room for the tallest parts, the lowest tails and a little space. So the capitals are always shorter than the point size, often by about a third.

Design

Bento Grid vs Masonry Layout

People think a bento grid and a masonry grid are the same mix of box sizes, and they are not. A bento grid is a few boxes planned to fit one neat rectangle. The biggest box holds the most important thing. A masonry grid is many cards of mixed heights, packed down columns with no gaps. It fits more in, but the order gets muddled.

Design