Beat Keep

A disc pulses at a steady tempo and you tap along with it. After four beats the pulse stops, and the rest of the round is yours alone: four more beats to hold, then five, then six, all the way to nine, with nothing to follow but the tempo you were just given. Every tap is measured against the tick the beat should have landed on, and at the end the game tells you the one thing you cannot feel while you are doing it — whether you ran fast or slow, and by how many milliseconds.

It runs entirely on your device: 0 network requests, works offline. No ads, no account, nothing to buy, and no betting, money or prizes of any kind.

1. Focus the board — click it, or press Tab until it is focused.
2. Watch the count-in. The disc fills on each of the first four beats of a round and empties between them. The header above it tells you the tempo in beats a minute and how wide the hit window is in milliseconds.
3. Tap with the pulse. Press Space or Enter on each beat. On a touchscreen, use the Tap button — it goes through exactly the same path as the key.
4. Keep going when the pulse goes dark. After the fourth beat the disc stops filling and the caption changes from "count-in" to "keep it". Nothing on screen moves from then on, on purpose. Hold the tempo yourself.
5. Read the pips. The row under the disc is one square per beat of the round: dashed squares are the count-in, solid ones the beats that score. A judged beat shows a tick if you were dead on, a left chevron if you were early, a right chevron if you were late, and a cross if the beat went by untouched.
6. Mind the four lives. In the unpaced half, a beat you never tapped costs a life, and so does a tap that lands outside the window. Nothing in the count-in costs anything — it is a free rehearsal.
7. Read the drift line at the end. It names your direction, gives your average distance from the beat in milliseconds, and tells you how many of your taps landed on that same side of it.

Why do people speed up when they keep a beat without a metronome?

Because once the metronome stops there is nothing left to correct against. While a pulse is playing you are running a control loop: each tap produces an error against the next click, and you nudge the following interval to cancel it. That loop is doing most of the work, and it hides how good your internal clock actually is. Take the pulse away and the loop has no input, so whatever small bias sits in your timekeeper is no longer being cancelled — it is being added up, beat after beat, and after eight or nine of them a bias far too small to notice on any single tap has become a visible shift. Wing and Kristofferson's 1973 model of this splits each interval into a central timekeeper and a motor delay, and the useful consequence is that unpaced timing errors behave like a random walk: without an external reference the phase error has no ceiling. On top of that drift there is a systematic head start. When people tap along with a metronome most of them tap slightly before each click rather than on it — commonly a few tens of milliseconds, and more for people without musical training. That lead is invisible while the clicks are there to be led, and it tends to carry into the unpaced half as a genuinely shorter interval. The net effect across people is that speeding up is the more common direction, particularly at slower tempos, but the honest finding is not about the crowd: it is that each individual is remarkably consistent. Whatever direction you drift, you will probably drift it again next round, which is why this game bothers to hand you the number instead of telling you a general fact.

What counts as being on the beat here?

Every beat has an exact tick it belongs to, and a window either side of it. A tap inside that window is a kept beat; a tap outside every window costs a life, as does a beat you never tapped at all. The window is not a number somebody picked: it is computed from the round's own tempo, roughly a third of the gap between beats and capped at 300 milliseconds, and the board prints the figure for the round you are on. Inside the window there is a tighter band, about a third of it again, where the tick symbol appears instead of a chevron and the score pays the most. The reason the window is derived rather than fixed is that a fixed grace period quietly changes meaning as the tempo moves: 150 milliseconds is a small slice of a slow beat and most of a fast one, so a constant window would make the quick rounds easier in relative terms exactly where they should be harder.

Is there any money, betting or in-app purchase in this?

None, and there never will be. There is no currency, nothing to buy, no lives to refill for a fee, and nothing to win or lose but the round in front of you. Six rounds and four lives is the whole economy.

Why is the count-in free, and why only four beats?

Free because it is a rehearsal, not a test. The four paced beats exist to put a tempo in your head, and charging you for fumbling them would measure your reaction to a pulse you had not settled into yet — which is a different skill from the one the round is about. Four because that is the count-in every musician already knows, and because it is the smallest number that gives you three complete intervals to average, which is what you actually need in order to have a tempo rather than a guess. Everything you do in the count-in still shows on the pips and the verdict line, so you can see how well you had locked on before the pulse left.

Does it get harder, and how exactly?

Three ways at once, and all of them follow from one another rather than being stacked on by hand. The unpaced stretch grows by one beat a round, so round one asks you to hold four beats and round six asks for nine — and drift compounds, so the ninth beat of a long stretch is much harder than the fourth of a short one. The tempo quickens by a fixed step each round, with a small seeded wobble that is deliberately too small to reorder the ramp, so every round really is faster than the one before it on every seed. And because the window is computed from the tempo, a faster round is automatically a tighter one, in milliseconds as well as in ticks. There is no separate difficulty dial.

Can a round ever be dealt that is impossible to play?

No, and that is guaranteed by how a round is built rather than checked afterwards. The generator writes the beat grid first — beat one at one period in, beat two at two periods in, and so on — which means the perfect run already exists before anything else is decided: tap on each of those ticks. The judgement windows are then derived from that grid and clamped so that two neighbouring windows can never touch, which is provable arithmetic rather than a tuning exercise. So a tap always belongs to at most one beat, a tap on the beat is always accepted, and the tests drive that reference run through the real game on thirty different seeds and require a clean sweep every time.

Can I play it with a keyboard or a screen reader?

The keyboard is the primary control — Space and Enter both tap, and there is nothing a pointer can do here that a key cannot. For a screen reader the round is announced in words: the tempo in beats a minute, the width of the window in milliseconds, whether the pulse is showing or off, and every beat as it is judged, as "on the beat", "early", "late" or "missed". Nothing anywhere in the game is distinguished by colour alone. Be aware of one real limitation, though: while the pulse is playing it is carried by sight and by the click from the sound bank, so the count-in needs one of those two senses. The live region deliberately does not count down to the next beat — telling you when it is about to arrive would hand back the exact thing the unpaced half takes away.

Does it work offline?

Yes. Once the page has loaded once, the whole game is on your device — every round is generated locally from a seed, timing is counted in simulation ticks rather than read from any clock but your own device's frame loop, and no server is involved in play at any point. Nothing about your session leaves the device.

Six rounds, four lives, thirty-nine scored beats in a complete session. That is a small sample for a timing measurement: a laboratory version of this task would take dozens of intervals at each tempo, so treat one session as something interesting about your afternoon rather than a fact about you. The simulation runs at sixty ticks a second, so every number is quantised to about seventeen milliseconds — a reading of "62 ms fast" is really "somewhere near 62". Absolute figures also carry your hardware: keyboard scan rate, screen refresh and touch latency all shift the whole distribution, and a touchscreen tap is measurably slower to register than a key press, so the two are not comparable. The direction of your drift survives all of that far better than its size, which is why the end card leads with the direction. And this game needs either sight or sound for its count-in; it is fully playable and fully announced with a screen reader, but the paced half will be harder without one of those channels.

Side Step is the other game here that measures something about you and then hands you the measurement instead of keeping it — there, the cost of a conflicting cue in your own reaction times. Tap Order times a visual search instead of a tempo, and Quick Count asks what you can take in without counting at all. Sequence Echo is the memory neighbour of this one: it asks you to reproduce a pattern rather than an interval.

Bookmark this page (Ctrl+D, or ⌘D on Mac) or install the app — it works offline the next time you need it.