Deliberate Practice

The Orange Site linked to some article or the other on Deliberate Practice, so here are some thoughts on DP.

Cultures vary in depth, or even within a culture; some pour the tea into a cup, while elsewhere there is a whole ceremony. DP may be necessary to get halfway good at a tea ceremony in a reasonable amount of time. Others pour their coffee from a device that was cleaned last who knows, while elsewhere one will wait for the barista to try a few more pulls until they get it just right. Either method works if you only want some stims for the day. Either may be a bad habit if done too often.

Cutting out distractions—smartphones, the internet—may help build concentration. Without sustained concentration, what can you learn? There is a Goldilocks zone, maybe three to five hours of good, deep practice per day, split into sessions maybe an hour to 90 minutes long. Spending more time on the same task is pointless and may even regress. The quality of the practice—focused, intent, looking for the errors and how to improve—is more important than quantity of practice.

Note how a typical 40 hour work-week minus 25 hours still leaves 15 hours to be spent doing... something. Certainly nothing productive. Stapling TPS reports together? 72 hours per week or the so-called 996 system? Most of the work will be to warm a chair.

Some will blame external factors rather than asking the difficult "how did I screw up?" questions. With study, say of Yet Another Stupid Death in a roguelike, one will usually identify multiple mistakes that you have made. From that you might create a plan of how to avoid or reduce those mistakes. Nah, it was the RNG.

Lacking a coach one can use metacognition. This is basically watching yourself and being mindful of what is going on, what isn't working, what needs improvement. A coach or a buddy will help if you are in the dumps that day; the article points out that two people are rarely both down, so one can Enkidu the Gilgamesh, or the other way around. A record of a practice session may be necessary, or to interrupt the session sometimes to take account of how things are going, how enthusiastic you are, etc.

Avoid vanity "number go up" metrics. Such metrics are easy to game, may grant false enthusiasm and false progress, and may not help with the actual practice. Gamification is everywhere these days, so vanity metrics can be hard to avoid. Internal enthusiasm is necessary, but may be hard to generate: children may drop a task the instant there is no longer parental pressure, and while external rewards can be valuable they may not be enough to sustain a practice through the long silent years.

Spaced repetition one may already know from tools like Anki or memrise. If not, the gist is to repeat new things frequently, and to review known things less and less often—enough to keep the concept in mind, but not to forget it, nor burn out on too much of what is already known.

Rest is very important: enough sleep, good sleep, naps, time spent doing something mindless like washing the dishes, doodling, or staring out a window. (No, really, boss, this is to improve my productivity!) Let the mind wander, relax.

You may not want to space out while washing a large and sharp kitchen knife, but I have seen a sushi chef mindlessly shave a daikon while busy watching the Mariners on the tele. How did he get there? Years of practice.

In Practice: Technology

I am reminded that text pagers are somewhat clunky; they do not divide the text up into logical units—say, by paragraph—so instead you will have a mix of scanning around the window (even a small 80x24 terminal), moving the cursor within the view, and moving page view around the document. This can be awkward. Could improvements be made here? Perhaps to only show a single paragraph, and offer means to easy move up and down through those, and maybe some means to attach notes or thoughts to particular paragraphs? In view(1) a map of

is pretty close to jump down to maybe the next paragraph and to put it at the top of the screen, though long paragraphs larger than the display will still need to be scrolled through, and you would need a similar map to move backwards.

Thus I have a goal: try to improve text pagers for more efficient ingestion of long documents. This suggests a few tasks, one of which would be to research existing pagers, what libraries exist, etc, and another "how to actually write a text pager" which is at a sweet spot for me, as I don't know exactly how to write one, but have some of the necessary domain specific knowledge—unix, terminals, curses—so the task will be challenging, but not impossible. Someone else might be "yeah, I have libraries for that in a few languages"—not a challenge—while someone else might like be oh, a terminal, is that a FVWM command?—too much of a challenge.

Or consider a musical instrument, which is also technology: S.M. Stirling's book "In the Courts of the Crimson Kings" (2008) uses the term tembst, or "matter shaped by intent for utility". DP here would be to play only the highest note for a session—generally not long on account of the flute filling up with water, or embouchure failure—which can be done in many different ways: hitting it from a cold start, sustaining it, repeating it, trying for a better tone, while sitting, standing, etc. The highest note has always been trouble on the 2011 Coyote Oldman flute; from the latest session I noted that the end of the flute should be a bit lower than usual. A private practice place may help, woodshedding, especially if you are going to spend too long on that one note. Next steps might include linking that high note into usable phrases, getting better at not slapping your fingers down too hard, posture, etc. Longer sentences may be more valuable than working on individual words.

Too much deliberate practice might be bad if it takes you away from, I don't know, adequate socialization. You could be playing that roguelike for idle fun, not with DP, in which case your expectations of victory may need to be dialed down a bit. Or a lot.