Learn / Chapter 1
Accuracy before speed
By the end of this chapter: Hold 97% accuracy through a full session without slowing to a crawl, and know why that is the faster route.
Almost everyone trying to type faster tries to type faster. It is the wrong lever, and the evidence on this is unusually clear. Accuracy is not the cautious alternative to speed. It is the mechanism that produces it, and the reason is arithmetic rather than discipline: a mistake does not cost you one keystroke, it costs you several.
Why does typing accurately make you faster?
Because the time you lose is spent fixing, not typing. In the CHI 2018 keystroke dataset (Dhakal et al., 136 million keystrokes from 168,000 typists), typists who prioritised accuracy improved their speed roughly 19% faster than those who chased raw speed. Speed built on a shaky error rate is not speed, it is a larger cleanup bill arriving slightly sooner.
What does one uncorrected mistake actually cost?
Three to five extra keystrokes, not one. Soukoreff and MacKenzie (CHI 2003) put the cost of an error at an estimated three to five keystrokes once you account for noticing it, reaching for backspace, deleting, and retyping. A 95% accurate run is not 5% worse than a perfect one. It is closer to 15 to 25% slower once the corrections are paid for, which is why accuracy shows up in your net speed long before it shows up in your pride.
So how accurate should you actually be?
Aim to hold 97% or better, and treat anything below 94% as a signal to slow down. Below roughly 94% you are spending more time correcting than typing, and the practice stops building the motor pattern you want and starts rehearsing the recovery from a broken one. Above 97% you can safely push your pace, because there is headroom for the pace to cost you something.
Why not just type fast and clean it up later?
Because you are practising the mistake, not just making it. Errors typed at speed are motor patterns like any other, and they are estimated to take three to five times longer to unlearn than they took to acquire (Soukoreff & MacKenzie, CHI 2003). Every fast, sloppy run is a small deposit into a habit you will later pay to remove. This is the single most common way people spend months on a plateau while practising diligently.
Does slowing down mean staying slow?
No, and this is the part people get wrong. Accuracy-first is a training discipline, not a permanent speed limit. You hold accuracy high, your error rate falls, and then speed rises on its own because nothing is dragging on it. What you must not do is hold accuracy high by typing so cautiously that you never approach your limit at all, which trains nothing. The target is the fastest pace at which you can still hold 97%.
Practice
Make the trade impossible to cheat
Reading this changes nothing on its own. The drill below runs in precision mode, which freezes the input on a mistake until you press the right key. You cannot outrun an error, so the only way through is the one this chapter argues for: find the fastest pace at which you stay accurate, and sit there.
To finish the chapter, log 3 runs at 97% or better. Three rather than one, because the thing being trained is repeatability. One clean run is luck.
Start the chapter drill →What comes next
Chapter one is free, and it is the foundation the rest of the course is built on: every later chapter assumes you can hold accuracy while you work. The chapters after this one go into diagnosing where your speed actually goes, and they are part of Purist. The trainer itself stays free: every mode, every layout, always.
Related reading: how to type faster, how to learn a new keyboard layout, or how we measure accuracy and consistency.