I don’t agree. I was for a long time of the school that understanding is the only things that matters, and the details can always be looked up. Then during my post doc ( physics) I worked with a couple of Russian guys who had been forced to memorise a lot. What they gained was an enormous speed. They could try six different ideas in their mind while I still looked up the details necessary to try my first. So don’t underestimate root learning.

There’s a third category here, and that’s functional skill. It’s quite a separate thing from either understanding or rote learning. They all feed into each other, of course, but the point I tried (and apparently failed) to make is that it’s functional skill that’s useful in real-world contexts.

Sometimes the thing holding you back from improvement is a lack of facts, and sometimes it’s poor understanding of theory. In my experience, however, the fastest way to get better at doing something is almost always to practice doing that thing; most of the time a sufficient collection of facts and theoretical understanding will come along as a side-effect of that work.

How much of their prior training was memorizing specific facts, and how much of it was drilling the mechanics of solving typical problems? The former is what most people use spaced repetition for and what I believe is of limited utility; the latter is incredibly valuable and I never meant to imply otherwise.

Well, as a counterpoint, imagine how much time could be saved if you were able to recall from memory not just the most-used functions that you need for a problem, but the next level down of sometimes or rarely used ones. Or for patterns, or for other such things that can be simplified down into memorizable / recallable blocks.

Regularly looking things up is ok, but actively re-experiencing the thing you're trying to learn is a better way to make it stick.

In many cases less time that what I spend looking it up. I regularly look up things I haven't needed before, a few minutes of reading and I know it. Many of those things are something I expect to never need again in my lifetime. The few seconds to memorize all those things is greater than the time saved. Particularly since I don't know what I will need next week and so I'll be spending a lot of time learning things I turn out to never need.

> actively re-experiencing the thing you're trying to learn is a better way to make it stick

On this we agree, but to me this means practical application in various contexts, rather than call-and-response memory drills.

For programming, memorisation doesn't matter that much. Not knowing the argument order for a function isn't a big deal. You use an IDE or you get a good offline docs viewer (Dash, DevDocs etc.) or you learn to Google efficiently.

Language learning is an obvious use case.

Also: law and medicine. Having knowledge mentally 'to hand' is pretty important if you've got a patient under general anaesthetic, or a judge asking you a very difficult question.

> Language learning is an obvious use case

Memorizing word pairs can certainly be done with spaced repetition, but it’s unclear how much that translates to actual language ability. Second-language acquisition appears to be primarily dependent on reading (or listening to) the target language for content, and most words are learned via seeing them in context instead of being looked up in a dictionary.

I have no experience with law or medicine, but I expect the story is similar: practical knowledge is what you need to hand, and not book knowledge. Book knowledge is what gets you through the exams and into the practical part of your training.

That depends entirely on how you implement spaced repetition.

Word pair is the easiest way to do it in anki, but it's not the only way to implement it.

Beyond the very basics necessary to extract some meaning from a second language, I’d be shocked if time invested in spaced-repetition drills of any design had better returns than reading the target language for pleasure.

Anki has two benefits: memorization and understanding. Memorization is what comes from going over the cards, which is what you allude to. Understanding comes from the process of creating the cards, where you think hard about the atoms of information in the text, and their interrelationships, and cast them into simple questions. Then memorization gives you a much deeper understanding than if you hadn't ankied the text.

I concur, what you describe is a very effective use of spaced repetition. It is different from what ‘ProstetnicJeltz described, where cards should take less than a minute to write — that describes memorization without first putting in the effort to understand.

Writing the information should take less than a minute once you formulated what you want to write. Learning takes more time.