REACTION · MEMORY · ATTENTION
Everyone's got lag.
Eleven short tests, three ways your brain can be fast. Every test you finish adds an axis to your fingerprint. The shape below is yours and nobody else's.
Finish any test and the fingerprint starts. One score is enough to make it personal.
The numbers behind the shape
| METRIC | SCORE | PERCENTILE | WHEN |
|---|---|---|---|
| Reaction Time | — | — | — |
| Choice Reaction | — | — | — |
| Anticipation | — | — | — |
| Sequence Memory | — | — | — |
| Number Memory | — | — | — |
| Visual Memory | — | — | — |
| Chimp Test | — | — | — |
| Aim Trainer | — | — | — |
| Go / No-Go | — | — | — |
| Typing | — | — | — |
| Stroop | — | — | — |
Percentiles are derived from published norms per test; the origin is declared on each test page. They switch to humanlag's own curves as first-party runs accumulate.
Reaction
How fast the signal crosses you.
Memory
How much sticks, and for how long.
Attention
Staying locked on when it counts.
What humanlag measures
Everything here measures one gap: the distance between the world doing something and you answering it. The reaction family times that gap directly, from a signal to your finger. The memory family measures how much of the world you can keep with you once it stops being shown. The attention family measures what happens when the signal fights back: when it is buried among decoys, when it lies to you, or when the right answer is to do nothing at all, which is the hardest instruction there is. Eleven tests, one number each, and the fingerprint above is the shape of all of them together.
The machines are built to measure honestly. Clocks start on the frame your screen actually paints, not when the software asked for it; presses count as the key goes down; every timed test averages enough attempts that one lucky run cannot flatter you, and a press faster than a human nervous system is refused as the guess it is. What a browser cannot see, the site declares instead of hiding: your screen and input device add real milliseconds, so the fairest comparison here is always you against yourself, on the same setup.
Every percentile is read off a declared curve: a starting assumption from published data, plus one run for every person who has actually played here, everyone in one pool. The Method page keeps the whole accounting, test by test, and the guides go deeper into the science: what a reaction time is made of, what actually improves it, and how much a head can hold. None of it is a diagnosis. It is a mirror with numbers on it, and the numbers get truer the more people look.
Your arenas
Challenge links created on this device. Each one opens the arena: the ranking of everyone who played through the link.