Guides

A display can only change what it shows once per refresh, so part of every browser reaction time belongs to the monitor. Here is how much, worked out from the refresh rate rather than asserted.

Every reaction time site tells you what percentage of people you beat. Almost none can say which people, on what hardware, filtered how — and the one study that measured its own equipment shows why that matters.

One slow trial in five drags an average by tens of milliseconds and barely moves the middle one. Here is the arithmetic, and why every trial is shown rather than folded into a single number.

A percentile needs a crowd to compare you against. A chance level needs nothing but the rules of the test itself — the exact odds a guess would have gotten your score by accident.

Words per minute survives moving from a keyboard to a touchscreen as a unit. What it means doesn't — two published studies put a 15 WPM gap between them, and it's why this site tests mobile first rather than porting a desktop test down.

The same published study reports two mean digit spans a full digit apart, from two samples of different sizes. Span figures move with sample, modality and stopping rule — 'average is seven' is folklore standing in for a number that was never that stable.

A chimpanzee beat nine humans at a memory task in a headline that ran for years — until two humans were given practice and matched the chimp. What that says about retaking any test on this site, and why we still won't sell you a way to improve.

Caffeine, one bad night and the hour on the clock all move the same number the reaction time test reports, by amounts three separate literatures have actually measured. None of the three are a claim about who you are.

Aim-trainer scores are reported in bits per second rather than raw milliseconds because throughput stays stable across hardware while raw times don't. Here's the math behind that stability, and why the effective-width correction was deliberately left out.