Typing Test
Twenty-five words, as fast as you can type them. Words per minute is the score; a typo costs you the time it takes to fix.
The ladder
Bar widths show roughly how many people land in each band on the provisional curve. Five characters count as a word, spaces included, over the time from your first key to your last; a typo costs the time it takes to fix.
What moves it
The keyboard. A phone keyboard runs at a fraction of a desk keyboard's speed, and its suggestion bar is a different instrument again. Compare yourself to yourself on the same one; the curve does not split by device, everyone plays in one pool.
The language. Common words in the language you type every day come out of your fingers on their own; in another language you spell. Take the test in your own language, on a keyboard laid out for it.
Touch typing. Ten fingers that know where the keys are beat two that look. Hunt-and-peck tops out around 40 wpm; touch typists start there.
What this measures
Words per minute is the oldest benchmark on this site by a margin of a century: it was standardized when typewriters were office machinery, and the counting rule has not moved since. Five characters, spaces included, make one word, whatever the language, so a run in German and a run in English score on the same scale, and 40 wpm meant the same thing on a Remington as it does here.
You get twenty-five common words drawn fresh from your language's bag; the clock runs from your first key to your last and nowhere else, so reading ahead is free. A typo turns red and holds the passage until you backspace it, which converts every mistake into its true cost, seconds, instead of a deduction computed afterward. Paste, drop and autocomplete are refused: the run is your fingers or nothing.
The six languages' bags are matched on average word length, so the course is about the same distance everywhere. Type in the language your hands actually live in: common words in it leave your fingers whole, while a foreign word is spelled out letter by letter. One pool, phones included, declared with the curve on the Method page.
Questions people ask
The standard way: every five characters of the passage, spaces included, count as one word, divided by the time from your first key to your last. Twenty-five words is about 160 characters; typed in 40 seconds, that is 48 wpm. The word lists of the six languages are matched on word length, so a run is about the same length in every language. The words change from run to run, so the character count does a little too; the count on the card is the one that was scored.
They cost you time, not points. A wrong character turns red and the passage does not move on until you backspace it, so every typo is paid for in the seconds it takes to fix, and the accuracy on the card says how many keys went wrong. Nothing is subtracted afterwards: the number is simply how fast you produced the right text.
On the provisional curve, 45 wpm is the middle, 70 is quick and 110 is professional territory. Everyday typists on a desk keyboard land between 35 and 60; on a phone, expect about half. The curve is an assumption until humanlag has its own runs; everyone plays in one pool, phones included.
Because the average settles fast, and a long sample starts measuring something else. By 160-odd characters a wpm figure sits within a few points of what a three-minute test would say; past that, the test is increasingly about endurance and sustained concentration, which are real abilities but different ones. Twenty-five words keeps a run near half a minute, short enough to go again, and going again is where the honest number lives.
It starts as an assumption worth 1000 runs, shaped from published norms, and every run played here adds one to it, everyone in one pool. The Method page says which norms, how the percentile is read, and how many runs this test has measured so far.