The short version

Vibe Check is a public-opinion instrument dressed as a game. Place where you land on a spectrum, lock it in, then see how the crowd answered and keep moving through a shuffled set of questions.

1. Tap or click the blank dial to place your needle. Tap again or drag to fine-tune it; arrow keys work too.
2. Use the separate button to lock in your answer.
3. Read the reveal, then choose Next question, swipe left on a phone, or exit to Explore. You can also skip an unanswered board.

The shuffled feed does not loop back through boards you already completed. Explore lets you browse the focused Main Set, the monthly AI Pulse, and a community pool that grows as visitors publish boards.

The monthly AI Pulse is a separate three-question set. Its wording stays fixed so changes from one month to the next can be read as movement in this crowd rather than movement in the question.

Three kinds of reveal

• The real figure: you guess a national statistic, then see your answer beside a published representative survey and the Vibe Check crowd.
• The other side: after giving your own view, you can guess where respondents on the opposite half of that same dial landed.
• The whole crowd: after giving your own view, you can guess where other people on this Vibe Check board landed.

The second guess is optional. Skipping it reveals the results without inventing a prediction; your original opinion is already safely recorded. Opposite-side comparisons stay hidden until at least ten people are available for that comparison.

Why results come second

Seeing a crowd number first can pull a later answer toward it—an effect called anchoring bias. That is why unanswered boards never preview the average.

You may choose View results without voting, but it is a trade: the board is permanently closed to voting in that browser before any result appears. This keeps a revealed number from becoming an anchored vote.

Each board accepts one answer from you during its current period. Some are one-time questions; others reopen on the cadence printed with their results.

Shared board cards are deliberately result-free too, so a friend can arrive without seeing the answer they are about to judge.

How to read the results

AVERAGE 35%

The mean position for the window printed beside the result, with repeat answers reduced to each person’s latest one. The red needle points at it.

The coloured rays

The shape of the spread. A tall narrow fan means people broadly agree; a wide flat one means they don’t.

The arc and its two numbers

Where the middle 80% of answers fall. About one in ten went below the lower number, and one in ten above the higher one.

The ring of ten blocks

The spectrum split into the ten bands the results are described in — the fuller a block, the more people landed in it. Point at one (or tap it on a phone) and the line under the dial tells you which band it is and how many people chose it. Yours is outlined in teal.

Everything drawn on the dial is a position on the spectrum, never a percentage of people. The how-many-people figures are always in the sentence underneath.

Main and community boards

The Main Set is the smaller research-led collection where responses are deliberately concentrated. The Community page is the larger browse pool: additional Vibe Check questions appear alongside public boards people make. Visitor-made items are marked as such, and their individual pages carry a separate disclosure.

Anyone can make a standard board, an “other side” board, or a “whole crowd” board. Visitor-made boards cannot claim a real-world benchmark without editorial source review. They begin unlisted, can be reported, and only receive prominent placement after moderation.

After voting on a visitor-made board, you may privately mark it useful or not for you. That signal is not a public score and is not part of the research data. It only nudges discovery; a negative mark also removes that board from your future community stream.

What’s collected

• Where you placed each mark, as a number between 0 and 100, and when. On predict-then-reveal boards, your opinion and your prediction are stored as two separate kinds of record.
• No account, no email, no name, no cookies, no tracking scripts.
• Monthly reminder signup happens on Substack, not inside Vibe Check. Substack keeps that email; it is never sent back here or joined to the random browser ID used for answers.
• The optional place question records only where you place yourself between rural and densely urban. No address, GPS location, city or ZIP code is requested.
• A random ID links your answers to each other. It’s 16 random bytes generated in your browser — nothing about you feeds into it, and it isn’t connected to your name, email, IP address or any account. It exists so the results can answer questions like “do people who call something addictive also call it harmful?”, which is impossible with unlinked answers. Clearing your browser data destroys it.
• Receipts for your own answers and predictions are kept in your browser’s local storage so the page can show you what you picked. Clearing browser data erases those local receipts; the anonymous research records remain.
• To stop one person stuffing a board, your IP address is scrambled into a one-way code and used to count how many answers came from you today. The address itself is never written down, the code can’t be turned back into it, only a running count is kept, and it’s deleted after a day. It is never stored with your answers.
• As with any website, the server that hosts this keeps ordinary request logs, which include IP addresses. Those are the host’s standard logs; this project doesn’t read them, and they’re never joined to votes.

What these results can mean

Vibe Check visitors are self-selected, not a representative sample of the country. Crowd averages describe the people who answered this board; they should not be relabeled as “what Americans believe.” Published benchmark boards show their outside source, field date and sample note so the two kinds of number stay distinct.

The project is strongest at showing distributions, perception gaps, changes over time, and relationships among answers from the same anonymous browser. Those patterns can be interesting without pretending the sample represents everyone.

What this is for

This is a public snapshot of where people actually land on questions that usually get argued about in absolutes. It’s here for anyone to answer, share, argue with, and use to see how they compare to everyone else. The results are open to all — there’s no members-only version.

Aggregate results — averages, spreads, response counts and how answers relate to each other — may also be written about publicly. Individual answers and predictions are kept out of the public export and are not sold. The site’s hosting and storage services process requests and records to operate the project, but there is no advertiser or data-broker feed.

Some of these questions are personal. Your answers are grouped together by a random ID so they can be compared with each other, but no name, email, account, precise location or IP address is attached to an answer. Because a linked answer history is still sensitive, individual records are kept private and only aggregate results are published.