DuelReflex
A reaction test tells you one number. That number hides almost everything about how you actually react.
Two people can average the same and be nothing alike: one is quick and undisciplined, the other slower and impossible to fool. So this is a battery — each test puts the same question into a different situation, and the shape of your results, not the height of any one of them, is the answer.
Start with Showdown →The tests
- Lights OutspeedThe cleanest number you will get about yourself. Most people land between 220 and 300 ms.
- ShowdownspeedWhoever draws second usually wins. A physicist guessed it, a lab confirmed it, and you can test it on a friend.
- Stroop SnapchoiceNinety years old and it still gets everyone. You will read the word. Try not to.
- Hold FirecontrolTen green targets teach your finger a habit. Five red ones find out whether you own it.
- Ghost BallpredictionThe ball rolls behind cover and something in your head keeps rolling it. Call where it is.
- Free FallchoiceWatching one place is easy. Watching eight costs you time by a rule with a name — and you can feel it.
- Arrow RushcontrolA wrong key does not cost points. It costs time, which turns out to be worse.
None of these takes more than a minute or two. Some are a handful of attempts; the ones where a single attempt would tell you nothing — the colour test, the arrows — run longer on purpose. All of them work on a phone: wherever there are keys, there are targets to tap instead.
Your Reflex Type
Finish them all and your results resolve onto four axes. The relationship between your highest and your lowest — not your best score — decides which type you get.
- Speed
- How fast you go once the signal is unambiguous.
- Control
- Whether you can stop yourself when the signal says don't.
- Choice
- How fast you stay right when there is more than one answer.
- Prediction
- How well you model where something will be.
The types
- The Gunslinger — Fastest hand here. Also the twitchiest.
- The Sniper — Never early, never wrong.
- The Goalkeeper — You are already where it is going.
- The Analyst — Slow to move, impossible to confuse.
- The Duelist — Fast and disciplined. Rare combination.
- The Sprinter — One signal, one answer, gone.
- The Watchmaker — Precision over pace.
- The Metronome — No spikes, no holes.
Duel anyone
Every Showdown run ends with a challenge link. Whoever opens it faces the identical sequence of waits in the identical order, so the only thing that differs is the hand. Without that, one player simply draws shorter delays and wins on luck.
The link carries the times inside the URL itself. There is no account, nothing to sign up for, and nothing about you is stored.
Where these come from
None of these were invented here. Each one is a paradigm somebody published, usually decades ago, stripped down to the part you can actually feel.
- Lights Out is a racing start: five lights, a held pause, and the cue to go is the lights vanishing rather than anything appearing.
- Showdown is the gunslinger effect. Niels Bohr thought whoever drew second would move faster; Welchman and colleagues sat pairs of people at a table in 2010 and found the hunch held up.
- Stroop Snap runs the Stroop effect, named for John Ridley Stroop in 1935. Reading happens whether you want it to or not, and it gets in the way of naming a colour.
- Hold Fire is a go/no-go task, the standard way response inhibition gets measured. The number that matters is not your speed — it is how often you fired at the one you were told to leave alone.
- Ghost Ball is coincidence anticipation timing, the thing a Bassin timer measures in a sports lab. You are not reacting to anything. You are predicting.
- Free Fallis Hick's law, from W. E. Hick in 1952: every doubling of the options adds roughly the same slice of time, and eight places is enough that you can feel it happen.
- Arrow Rush is a speed-accuracy tradeoff. Nothing stops you going faster except what being wrong costs you.
How it is measured
Timing comes from the browser's own event and frame timestamps, not from when our code happens to run. Absolute milliseconds still depend on your screen and input device, so treat these as comparisons under the same conditions, not laboratory measurements. Results are grouped by input device for that reason, and anything below the floor of human reaction is thrown out.
What we can and cannot measure →What this is not
These are not clinical instruments. Nothing here screens for anything, and no score should be read as a sign of a condition — go/no-go tasks turn up in that setting often enough that it is worth saying plainly.
Free Fall borrows the home row because eight targets need eight places, not because anything here is scored on how many characters you can produce. And a good number is not a claim about how you will do at anything else. It is this test, under your conditions, today.