How Hivemind works
What is this?
Hivemind is a free daily game about reading other people. Every day you get 10 opinion questions — the most overrated sandwich filling, the household chore everyone secretly puts off, the group-chat habit that quietly annoys everybody — and four to six answers to choose from. None of them is correct. Your job is to guess which answer most other people picked. Tap one, watch the hive’s split arrive, and after ten questions you get a Hive Sync score: how closely you read the room. There is no account, nothing to install, and no way to type anything in. It is all tapping.
How to play
Tap Read the room and the day’s first question appears. Pick the answer you believe is the most popular one — not your own favourite. The moment you tap, your choice locks; there is no timer, no going back and no skipping, so the guess you make is the guess you get. Fifty small hexagons then migrate across the answers and settle into the hive’s split, one dot for every two percent, with the percentage printed on each cell and your own pick marked. A line underneath tells you how far off you were. Tap Next and do it nine more times.
Scoring is proportional rather than right-or-wrong. Each question scores the share of the hive standing where you stood, divided by the share standing on the most popular answer. Naming the top answer scores a full point; picking the 34% answer when the winner took 38% scores nearly a full point, because you very nearly had it. Your Hive Sync is the average across the ten, and it lands you in one of three verdicts: In Tune (78–100), Contrarian (55–77) or Out of Touch (0–54). Out of Touch is the funniest one to get, and we mean that kindly.
Tips
- Answer as the average person, not as yourself. The single biggest source of a low score is picking the answer you personally believe.
- The most obvious answer usually wins. Crowds are not clever; they are fast, and the first thing that comes to mind comes to most minds.
- Beware the interesting option. A funny or unusual answer feels like the right guess and almost never takes the top cell.
- Concrete beats abstract. Given the choice between a specific everyday irritation and a broad category, most people tap the specific one.
- Watch the shape of the split, not just the winner. Questions that come out 30/25/22/23 are telling you the crowd is genuinely divided, and those are the ones worth guessing carefully next time.
- Come back tomorrow. The set rotates at UTC midnight and your streak lives in this browser only — clearing your site data resets it, and it does not follow you to another device.
Where the numbers come from
This is the part we want to be completely straight about. A brand-new question has no answers yet, and a split of 0% across the board would be both useless and misleading. So every question ships with a seeded baseline: our own honest editorial estimate of how a broad audience would answer it, written when the question was written. On day one, before a single answer has arrived, that estimate is all of what you see: the split on screen is entirely ours, and not a measurement of anybody.
Real answers are then mixed into it. The number displayed for an option is the seeded baseline weighted as 400 phantom answers, plus every real answer received, divided by the total of both. That means the baseline dominates early, gets steadily outvoted as real answers arrive, and is a minority of what you see once a question has passed 400 real answers. It is never topped up and never re-seeded — it only ever shrinks.
Because that figure is a blend, we never call it a percentage of players. In the game it is labelled the hive. Any number labelled players or answers counts real votes and nothing else, which is why the footer under each reveal reads “Seeded baseline + N answers” until a question has passed 400 real answers, and shows the bare count after that. If we cannot reach the live tally at all, the run continues on the baseline alone and says so rather than showing you a number we do not have. And a question’s split is only ever published as data once 50 answers have been recorded for it, at which point we show the live split with the recorded count and drop the seed entirely. One recorded answer is one completed ten-question run, and runs that declare themselves automated are excluded — so what a published split substantiates is a tally of recorded runs of a game, never a measurement of the public.
Questions people ask
Is everyone playing the same questions?
Yes. The 10 questions are chosen from the UTC date by a fixed rule, so everyone on Earth gets the same set on the same day, and no question repeats within 9 days. There are 118 in rotation and they deliberately come back around — that concentrates answers onto each question instead of spreading them thin.
Do you store my answers?
Only as anonymous totals. Each question has one counter per option, and finishing a run adds one to ten of them. No account, no identifier, no record that any particular run happened. Your streak and your last result are kept in your own browser and are never sent anywhere. See the privacy policy for the full detail.
Can I submit my own question?
No — and there is nowhere to type one. Every question and every answer is written by us, so that nothing on a share card was authored by a stranger. If you want to suggest one, email auto-ai-agency@quantum-solutions.hr.
Can I replay a day?
You can look at today’s result again as often as you like, but a run only counts once per day per device, and there is no archive of past days. Tomorrow’s set unlocks at UTC midnight.
Is a low score bad?
Not at all. A low Hive Sync means you answered honestly instead of strategically, which is a perfectly reasonable way to play. It is a game about crowds, not a test of anything.