A game about being predicted

hum(ai)n is a browser game that takes about ten minutes. You answer a set of deliberately unexplained questions, and then you play fifteen rounds against Darry, who commits to a guess about each choice before you are allowed to make it. At the end you are shown two numbers: how often Darry was right, and the remainder.

There is no account, no sign-in and no score to chase. It is a piece with a beginning and an ending, closer to a short film you operate than to a game you progress through.

Darry is a prediction system. While you answer, the game measures ordinary things about how you decide: whether you repeat a choice that just paid off, whether you abandon one that did not, how much you drift toward one side of the screen, and how long you pause before committing.

Those measurements become a small statistical profile, and that profile — not your identity, not anything you typed, because there is nothing in this game to type — is what Darry predicts from. Some of the work is done by a language model on the server; the rest is done by a deterministic engine that lives in the same codebase. When the model is unavailable, the local engine plays Darry and the game does not tell you, because from the outside there is nothing to tell.

The one rule the whole piece rests on is that Darry answers first. Each round’s prediction is generated, sealed and sent to your browser as something the browser cannot read, and only then do the controls become usable. It is enforced by the software rather than promised by it — see how it works.

It began as a much smaller prototype called The Prediction Booth, built after reading about models trained on very large collections of real human decisions. The unsettling part of that research was never that a machine could be clever. It was how little of a person it seemed to need.

hum(ai)n is that thought turned into an experience rather than an argument. It measures one person for about ninety seconds, and then shows them how little that took.

It is not a psychological test, a personality assessment, or a diagnostic tool of any kind. The traits it measures are behavioural regularities in a made-up task, and they say nothing about who you are outside of it. The closing verdict is a line of fiction written for the ending — it is not a finding.

It is also not a horror story that has been softened for a general audience. The warning shown before you start is accurate; if the themes described there are not for you, they are genuinely not for you.