Is it worth learning to type faster? If – like me – basically what you do is type, this seems likely to be a pretty clear win, if you have any interventions that would improve it at all. Ryan Carey suggested a painful sounding intervention which improves maximum typing speed a lot, but said that since he usually types substantially below his maximum typing speed, this would not help. His model seems to be that typing speed is basically either bottlenecked by physical typing ability or something else (like thinking speed), and it is not much worth trying to speed up the one that is not bottlenecking the process. This sounds pretty reasonable, but did not match my intuitions, and seemed extremely cheap to test, so I decided to test it.

I tried a number of ways of reducing my typing speed, and chose three that were reasonably spaced across the spectrum (~90wpm, ~60wpm, ~30wpm) on a typing speed test. These were (respectively) Dvorak keyboard layout, Dvorak with my left pinky finger tied up with a rubber band or tape, and Qwerty keyboard layout. I measured each one three times on that test, and three times on longer (3-5m) journaling activities, mostly writing about issues in my life that I wanted to think about anyway. These journaling bouts tended to be faster than I would usually casually write I think, so this does not really test how much peak typing speed improves combination writing/staring into space speed. But they were slow enough to be real journaling, with some real insights, and were substantially slower than peak typing speed.

My results are below. They are a bit dubious, but I think are good enough for the purpose. Moving from the middle method to the top method improved my real speed in proportion to my peak speed. Between the bottom two, it made little difference. Further details are more confusing – that there is no difference between peak and real speed for Qwerty suggests that physical typing is a big bottleneck there, however improving the typing method to handicapped Dvorak – which has a higher peak speed – doesn’t improve real speed much either, suggesting inconsistently that thinking is a huge bottleneck, which also seems implausible if thinking is not such a big bottleneck for higher speeds (implied by the fact that real speeds get a lot higher with better typing methods). But if I wanted to think more about these things, I should probably just do some more tests. I’m not convinced this is worth it, but if anyone else does any, I’m curious to see.

Incidentally I suspect Qwerty gets a boost in journaling relative to typing tests. This is because I have to look at my hands a fair bit to do it, which is harder when you also have to look at the screen sometimes too.

I’m more inclined to trust the patterns in faster speeds, which I say is due to them being much closer to my real typing speed (from which I might improve), but could obviously be because it supports my prior intuitions.

Incidentally, a few plausible-to-me models that would fit either thinking or typing faster increasing real typing speed:

• Thinking speed increases linearly with exogenously increased typing speed, and is usually the bottleneck – then increasing physical typing ability always increases thinking speed, as does increasing thinking speed itself.
• You do bursts of thinking and typing, basically one at a time – then your speed is something like a weighted mean of the speeds.
• You either type or think all the time (this is the bottlenecking activity), and do the other activity part of the time, however when you do the faster one it slows down the bottlenecking activity, so speeding up either activity speeds the entire process.

Mean speeds

All data

Some graphs