Three timed conditions, 45 seconds each. The same fixed text for everyone, and one sentence at the end telling you which condition costs you the most speed.
Before you start
Set aside about three minutes without interruption. The protocol is built for an English physical keyboard with typing assistance switched off, and a run has to reach all three deadlines to count. Pasting, switching tabs, leaving the typing field or an input-method composition makes the run practice-only — not because those are cheating, but because the timing they produce cannot be interpreted.
The clock on each stage starts at your first keystroke, not when you press the button, so there is no penalty for settling in.
What each stage is for
Flow is lowercase common words with no punctuation or digits: it establishes the ceiling the other two are measured against. Natural text adds capitals, commas and apostrophes. Precision adds digits, currency, brackets, dates and identifiers. Only one thing changes at each step, which is what makes the drop from one to the next attributable to something specific.
What the result means, and what it does not
The report describes copying performance on these passages under this protocol. It is not a population rank, a credential, or a measure of writing ability, and repetition will make fixed text familiar enough to flatter a later run.
Every metric states the evidence it needs, and withholds itself below that. If nothing crosses a reporting threshold, the report says no single condition stands out instead of promoting the least-good number to a diagnosis.
Method and privacy
The methodology page publishes every formula, threshold and validity rule used here — read it before drawing conclusions from a result. Keystroke traces are processed in your browser and never uploaded; only bounded summaries are saved to local storage so a later run has something to compare against.