Accuracy before speed, and the arithmetic that proves it
The advice is old and usually delivered as a moral. It is actually a calculation, and you can run it on your own numbers.
Every typing tutor ever written tells you to prioritise accuracy. Almost none of them show you why, so it lands as discipline-for-its-own-sake and most people quietly ignore it.
Here is the arithmetic instead.
What an error costs#
An error is not one keystroke. It is the wrong key, the moment you notice, at least one backspace, and the correct key — and the noticing is the expensive part, because it interrupts the rhythm that was carrying the rest of the sentence.
Measured across sessions, a single corrected error costs most typists somewhere between four and eight keystrokes of equivalent time. Call it six.
Run it on yourself#
Take a typist at 60 raw WPM with 94% accuracy. Sixty words is 300 characters a minute; 6% of those go wrong, which is 18 errors, each costing about six keystrokes of time. That is 108 characters of repair work — over a third of the minute spent undoing the other two thirds.
Now take the same person at 52 raw with 99% accuracy. That is 260 characters, 2.6 errors, about 16 characters of repair. They are typing more slowly and finishing sooner.
Above roughly 95% accuracy, speed compounds. Below it, speed mostly buys you more to correct.
The threshold, not the ideal#
None of this argues for perfectionism. Chasing 100% makes people type far below their capability and the returns vanish above about 98%. The useful target is a floor rather than a summit: get comfortably above 95% first, then push tempo, and drop back the moment accuracy falls through the floor again.
Typepace's adaptive drills use exactly this rule. When session accuracy sits below the floor the generated plan shortens and slows; when it is above, the plan lengthens and the text gets harder. You can watch it happen in the day's plan on the Learn page.