5 lessons from type-racing and sight-reading

Sometimes I race Internet strangers in typing competitions. I was also trained to sight-read as a piano player. The following 5 “in-the-moment” factors account for ~95% of variance in my performance on both tasks for passages of comparable difficulty. Obviously mean performance is determined mostly by practice, but relative variability is high enough for me to care, i.e. (standard deviation)/(mean) is maybe 0.15.

Type-racing and sight-reading are short bursts of effort so I get lots of feedback from them; it’s very easy to observe the effects of shifts in mentality/strategy. The pointers below may transfer more or less well to other tasks, and moreover are fairly standard advice, but I think I’ve somehow internalized their general importance through loads of empirical data.

1. Warm up

Cold hands are truly the bane of my existence. I’ll always do jumping jacks or some Hanon exercises before seriously attempting to sight-read. There’s also something to be said for mental warm-ups: running through a few typing/sight-reading passages before doing the real thing helps me recall the relevant variables as well as adapt to the current setting, e.g. a lower-than-usual desk.

2. Focus

This one is self-evident and basically synonymous with “trying hard” for me. You want to eject all thoughts from your brain except for the pertinent ones.

I think the surprising thing is that by default I don’t have focus. One reason for this is I can often survive without it. My fingers will eventually guide themselves to the right places on my keyboard (either keyboard!) with what feels like basically 0 effort. So not focusing is cost-effective in some sense. But autopilot isn’t a great operating mode - of course you’ll be faster and more accurate if you use more mental resources. Details that require attention over longer time horizons suffer in particular from autopilot, such as crescendos.

I tend to model focus as a binary state. A simple acknowledgement of the task at hand seems to be sufficient to avoid that dangerous autopilot-iness. Before a race or exercise I think something like, “Okay, it’s time to do the thing - but actually though.” This is ridiculously helpful.

3. Lower the stakes

Type-racing and sight-reading both impose time constraints. Races have winners and losers and often people listen to your sight-reading and judge you.

Stress seems entirely negative unless it has the indirect effect of improving focus, but I’m in favor of going ahead and improving that focus directly. To reduce stress, I like to simulate context. I pretend that nobody is watching me, or that in fact I am already the fastest typer in the world and am only there to show off.

During an activity you might receive feedback in real-time, e.g. you catch a glimpse of other racers pulling ahead or a judge taking notes. I try to ignore these signals since they rarely contain information relevant to my own actions and they break simulated context.

4. Look ahead

My old piano teacher would recommend that I read the next measure as I play the current one. When I type-race I look at upcoming words as I’m typing.

I don’t know why this strategy works so well. One model is that the brain requires some minimum time to load and transform notes/words into necessary muscle movements. These muscle movements can be stored in working memory and queried rapidly, so advance processing makes sense. Another explanation is that data is better processed in chunks. For example, the fingering I use for one chord should be informed by the fingering required by the next chord so as to avoid awkward transitions.

5. Strike a speed-accuracy balance

In a race it’s tempting to type as quickly as possible. Doing so creates lots of typos that you lose precious time correcting (you’re graded on accuracy as well as speed). On the other end of the spectrum, my piano teacher would say that I should only depress keys once I was sure my fingers were all lined up over the correct ones.

Essentially you want to experimentally determine the Pareto front of speed and accuracy, and then pick a solution according to success criteria. For instance, I generally choose to follow my piano teacher’s sight-reading advice because playing the wrong notes is very high-cost and rhythm is decently flexible.

Often I struggle with a bias towards perfectionism. I’m disturbed by angry red lines under mistyped words.

I have to override the compulsion to retype every time, even though it almost always increases WPM to simply move on because backspacing is costly.