Practice and Mastery [CF Article]

4 Likes

from the linked article:

In other words, initially learn ideas using full conscious attention (doing that involves practice). Then make them automatic (that involves more practice, which still requires some attention directed at improvement) so they no longer require your focused attention (you’ve now achieved mastery), which lets your attention be available for more advanced ideas.

I’ve been learning things using the flashcard app Anki.

I hadn’t been learning the cards to mastery. Once I got cards right, I’d classify them as learned and the app would move me to gradually spacing out practice on them. Then I never learned them to mastery. I’d get some of them wrong after a while and then I’d have the app move me back to practicing them more frequently, but then again I’d move to spacing out the practice once I got the card right.

Now I’m trying a different strategy. I want to master a card by doing it over and over in the same session, until I can get the right answer quickly and automatically, not just get the right answer. I hope this kind of learning will stick better.

1 Like

Mastery is when you’re very good at something. You know all about it. That generally means you can do it quickly, easily, with low attention and with few errors.

I don’t think that this part fits: “You know all about it.”

E.g., I think that I would qualify as a master at walking - in all aspects except the know all about it-part. I don’t know all about walking.

One guess is that the “You know all about it.” part is about “That generally means you can do it quickly, easily, with low attention and with few errors.”. But then I don’t think it adds anything as that summary is enough by itself to me. I think it would be better without the “You know all about it.” part. I think that something like this would be better:

Mastery is when you’re very good at something. Having achieved mastery in something generally means you can do it quickly, easily, with low attention and with few errors.

Am I missing something?

Good question. A couple things. First, knowing all about something is pretty typical of mastery.

Second, you aren’t a master of giving a lecture on walking. You only have mastery of some parts of walking and talking about it isn’t one of those parts.

You can look at it this way: your knowledge of walking is very robust and thorough. You can deal with things going wrong. You can deal with many contexts and still walk well. You can adapt and be flexible with your walking. You know how to walk with shoes or barefoot. You know how to walk with a limp. You can hop on one foot. You can skip, run, jump, crawl. You could use crutches quickly even if you’ve never used them before. You can walk on stone, dirt, wood, rug and more. You can walk on flat ground or a 1 degree incline up or down, or a 2 degree incline, or 3 degrees, etc. You can walk in places with bumps, holes, dog poops, obstacles. You can walk on rough ground where you have to stop on obstacles like rocks. You can walk along the bottom of a pool (shallow enough for your head to be above water). You can walk in sand or a little ways into the ocean.

There are parts of walking you don’t know about; you aren’t a master of those.

“Walking” can be approximated as a single thing. We can talk about it that way and that’s usually good enough. But it isn’t a single thing. We could divide “walking” up into a bunch of skills that you know in a thorough, complete way (know all about them instead of half-finished learning) and some other walking skills that you don’t have mastery of.

1 Like

You were spot on on my mistake. I was very much looking at walking as a single thing. Explaining how one can have mastery of a part of walking, but not necessarily other parts, helped me understand this better. Thanks.

Another issue is you mastered walking a long time ago then forgot some of the conscious knowledge but kept the subconscious automatizations.

1 Like

Skills degrade significantly slower after you achieve mastery, especially if you keep using the skill. Once you know something well enough to do it in an automatic, intuitive way, it’s much harder to forget.

This seems true to me but I have a hard time understanding why this is the case. It seems weird that there is a discontinuity in forgetfulness once a certain level of skill is achieved. What caues the degredation of medium leveled skills? What prevents that degredation with mastered skills? What happens to intuitions that fall into disuse? Is forgetting proportional to the time spent on learning or something like the number of connections to other ideas?

Discontinuities/breakpoints are common in reality and actually very important things to look for. Now that you bring it up, I realize that having a strong intuition expecting continuity is common, widespread and wrong. I’ve talked about breakpoints in CF stuff a bunch but didn’t connect that idea with there being a widespread intuition against it.

CF’s basic position is that differences only matter when they cross some sort of breakpoint. Any kind of optimization of a quantity, which does not cross a breakpoint, is a waste of time. This is related to Goldratt on optimization but different. There are various reasons for this position. At the abstract level, for the claim to be universally true, it relies on the fact that success at a goal, as against failure at that goal, is a breakpoint. So any optimization must change the results from failure to success for some goal that you care about or else it doesn’t matter.

However, there is a widespread intuition that optimizing things – increasing the amounts of good traits – is generically good and provides continuous benefit (rather than discontinuous or breakpoint based benefit). This intuition is highly visible in e.g. the Less Wrong and Effective Altruism communities and MCDA papers.

I talked about breakpoints in the CF course on gumroad a bunch and they come up in some articles e.g. Breakpoints, Categories and Margins of Error

Re memory, some issues are that complete skills are higher value (more worth remembering) and make more sense (it’s easier to remember a good, complete explanation than some pieces that don’t you don’t know how to fit together).

1 Like

GOAL: To summarize and relate to my own understanding.

The best way I can think of to understand discontinuities in skill acquisition is through an analogy to building a car (just thought of this while reading your post). Skills acquistion is like building a car because a car that is missing pieces won’t really be a car, in the sense of fulfilling its purpose. A complete car will hold together for many years and many thousands of miles of driving. The completed car won’t needs tons of ongoing maintence to remain usable. An incomplete car will fall apart quickly if it works at all. There is a breakpoint in completing the car manufacturing process that gets the car over the finish line to being useful and durable.

So, skill acquisition has some of the same properties of car manufacturing. A completed skill (one that has been mastered) will be durable and usable with minimal ongoing maintenance. This is because the skill has crossed some breakpoint at which the ideas have been “fastened together” well enough that they should hold for a long time.

1 Like