Spacebar CPS Test

CPS means independent presses divided by elapsed seconds. This analysis tool goes beyond one total: it reports average CPS, a trailing three-second rolling CPS, the strongest sliding one-second window, the slowest complete one-second bucket, first-half and second-half speed, percentage change, and a consistency score. Keyboard and touch modes are labeled separately because they use different input mechanics and should not be treated as interchangeable evidence.

Average CPS equals valid independent presses divided by the selected active duration. Best one-second CPS uses a sliding 1,000 millisecond window, so a burst that crosses a wall-clock second boundary is still measured. Per-second bars use fixed buckets from the test start. The slowest metric considers only full seconds. Rolling CPS counts presses in the final three seconds, or the whole run when it is shorter.

Consistency starts at 100 and subtracts the coefficient of variation of full-second bucket counts, expressed as a percentage, then clamps the result between 0 and 100. It is an interpretable stability indicator rather than an AI score. CPS change compares second-half CPS with first-half CPS: a positive value means the later half was faster, while a negative value means output declined. A zero first half produces a neutral change instead of dividing by zero.

Five seconds emphasizes a short burst; ten seconds gives a familiar compact baseline; fifteen seconds adds pacing without creating a separate thin page; thirty seconds exposes more variation. Use the same keyboard, browser, hand position, duration, and input mode when comparing attempts. A median across several clean runs is often more representative than one personal best because it reduces the effect of an unusually lucky or disrupted round.

The keyboard test counts a Space keydown only when it is not a repeat and the key was previously released. Editable controls and buttons are excluded. Touch mode accepts pointer presses only inside its dedicated pad. A run is invalidated when the window loses focus or the document becomes hidden, because background scheduling breaks the visible continuous-test condition. Interrupted runs are not written into local history.

At most twenty aggregate results are stored in this browser with schema version, date, duration, input type, and calculated metrics. The history can be filtered, cleared, or exported to a UTF-8 CSV with stable column names. It does not store text, a continuous key sequence, or a public identity, and it is never uploaded automatically. Clearing site storage also removes the history. Device and browser timing still affect every result.

Use the CSV as a personal comparison log rather than evidence of a universal ranking. Keep keyboard and touch rows separate, filter to one duration, and compare average, best-second, and consistency together. One higher burst with a much lower consistency score may represent a different pacing pattern rather than a clear improvement. When sharing a CSV, review it first because dates and device-use choices can still be personal context even though raw key events are absent.