CABINET / 01
CROWD CONTROL
A maze. An exit. A crowd that has some very confident directions.
MOVE / DECIDEWASD + F
A midnight arcade for unfinished humans // est. 2026
INSERTCOIN.QUESTIONREALITY.
Five cabinets. Five small worlds with rules that look suspiciously like life. Win imaginary points, lose imaginary certainty, and discover what the scoreboard forgot to count.
NOT A PSYCHOLOGICAL TEST. Not a personality diagnostic. Not a trap for your email. Your games and scores remain on this device.
Level select / five working machines
CHOOSE YOUR
ILLUSION.
Every cabinet has three difficulties, touch and keyboard controls, an honest score and a debrief. Pick your poison. The coin is fictional.
CABINET / 02
LUCKY YOU
Safe money or shiny danger? The house prints the odds this time.
RISK / REWARD1 / 2
CABINET / 03
TAB TRAP
Color? Shape? Time is running. The rule has changed again.
ATTENTION / SWITCH← / →
CABINET / 04
TRUST ISSUES
Work together. Or don't. Either way, the other player remembers.
COOPERATION1 / 2
CABINET / 05
MAYBE MACHINE
A signal arrives. Your beliefs need software updates.
EVIDENCE / BAYESSLIDER
NOW PLAYING / 01
CROWD CONTROL
The exit will not agree with everyone.
Select a difficulty, read the controls, and press START.
Intermission / the editor appears
THE GAME
IS NOT THE WORLD.
Some useful ideas begin as metaphors. Trouble begins when a metaphor puts on a lab coat and refuses questions.
A CROWD CAN BE INFORMATION. OR NOISE.
A group pointing in the same direction may know about a fire, a shortcut or a social norm. It may also be copying one confident stranger. Following is not always weakness; refusing is not always courage. Context determines whether independence is a virtue or just an expensive detour.
In the maze, crowd arrows are scripted and sometimes wrong. In life, other people often know things you do not. The serious question is not “Did I conform?” but “What evidence made their advice trustworthy?”
AN OUTCOME DOESN'T GRADE A DECISION.
A lousy gamble can win. A sensible risk can lose. The random payout does not retroactively make a choice brilliant or foolish. Information available beforehand, possible harms, values and your ability to absorb loss matter more than the victory jingle.
Nor is maximising expected coins a universal commandment. A certain small gain can matter greatly to someone whose bills are due. Our arcade knows no one’s obligations; real finance and life certainly do.
03 / 05
THE MACHINE KNOWS YOUR BUTTONS. NOT YOUR CHARACTER.
An attention game can introduce the feeling of switching mental rules. It cannot measure ADHD, intelligence, willpower or how attentive you are at work. A cooperation game can demonstrate how incentives shape behaviour. It cannot establish whether you are a generous friend.
The belief machine asks for probabilities under deliberately supplied numbers. In genuine inquiry, we often do not know the prior, the quality of the evidence or whether our model is wrong. Updating well also means questioning the measurement system.
The five games share one discipline: do not confuse a clear score with a complete account of what happened. A narrow feedback loop can teach one skill while hiding everything outside its frame. Good questions escape the cabinet.
THE SCOREBOARD IS A PARTICULAR KIND OF STORY.
Scores compress. Compression helps us notice differences, but it discards motives, luck and alternative goals. Even a fair leaderboard cannot answer who became kinder while playing. The highest-scoring strategy may fail under unfamiliar rules, or succeed only because the system rewards it.
Whenever a school, workplace or algorithm prints a number beside a human being, ask what it counts, what it omits and who had the authority to choose the formula.
REPLAY IS NOT REPENTANCE. IT IS REVISION.
On the next run you may choose differently, not because you were a bad person but because you encountered a new consequence. Trial-and-error is easier inside a game: the stakes are artificial, the rules visible, the restart button guaranteed.
Outside the arcade, other people bear some of our experiments. Be curious enough to revise; responsible enough to notice who pays the price. The monkey wants unlimited lives. The fox reads the terms and conditions.
Local-only / your small noisy record
HALL OF
MAYBE-FAME.
Numbers are from your browser only, not from strangers or a global ranking. Each difficulty has its own wall. Clear everything whenever you want.
THIS BROWSER / COMPLETED RUNS
00
THIS TAB / PLAYED
00
Scores persist in local browser storage when available. No login, telemetry or server-side leaderboard.
| CABINET | RUNS | PERSONAL BEST |
|---|
Not comparable across games: the score formulas intentionally differ. This wall remembers the game, not your worth.
Receipts / where the curiosity came from
SOURCES &
FURTHER READING.
Research informs the questions, not the validity of these original games as tests. Some linked publications may require access.
Solomon Asch (1955), “Opinions and Social Pressure” ↗
Historic conformity research. Our crowd maze is fictional, not a replication of an experiment.
Kahneman & Tversky (1979), “Prospect Theory” ↗
Foundational descriptive theory on gains, losses and uncertain outcomes. The betting cabinet is not a validated measure of risk preferences.
American Psychological Association, “Multitasking: Switching costs” ↗
Overview of evidence on task-switching costs and task demands, not a clinical benchmark for our game.
Stanford Encyclopedia of Philosophy, “Prisoner’s Dilemma” ↗
Defines strategic payoffs and contrasts individual incentives with collective outcomes.
Stanford Encyclopedia of Philosophy, “Formal Epistemology” ↗
Introduces conditional probabilities and Bayes’ theorem, used in the simplified belief game.
The Mind Mixtape, Sahil Tayal ↗
A sibling experiment in interactive storytelling. Inspiration for ambition only: this arcade has its own visual world and code.
Method note: Scores, timers, “crowds,” opposing strategies and Bayesian sensors are game mechanics designed for exploration, not observed participant research. Any psychological insight is an invitation to investigate, never an assessment of the player. There is no advertising, analytics, backend or tracking code.