Hacker News new | comments | ask | show | jobs | submit login
Effective learning: Twenty rules of formulating knowledge (1999) (supermemo.com)
63 points by jasim 72 days ago | hide | past | web | favorite | 17 comments

I tried using Mnemosyne (a Linux program that uses the same algorithm) years ago when I was in school, and did not find it effective. It was a pain to enter the cards, and a lot of my material was mathematical. Also, you needed to be really consistent. Take 2 weeks off and the algorithm behaved poorly.

Looking back, I think the problem was the lack of documentation on what to expect. There are always questions of how many cards should you do in one session, and how you should rate them.

A few days ago I started again, this time using org-drill. As with everything org-mode related, the experience is much better this time round. When it presents you a card, it has all the LaTeX rendered, and all images showing. Let's see if I don't get sick of it this time round.

Using a spaced repetition system for math has never worked well for me. It's useful for memorizing mathematical facts (for example, the mean/variance of various statistical distributions), but over time my intuition will fade, and I'm left parroting facts, having lost the understanding of why they are true.

In my experience, spaced repetition systems excel when I can create cards that take 10 seconds or less to answer. So they work really well for something like language learning. But it's difficult to reduce mathematical arguments into 10 second chunks.

What you say is generally true for mathematics, but it can help for the basics: Definitions of terms, etc. Basically, my goal was/is to memorize the facts one really should know without needing to look it up. For example, I've learned statistics at least thrice in my life, and beyond the very basics, the material doesn't stick. I think it's good for retaining a minimum amount of knowledge that you can use in the field any time you need it. Things like the recursive rule for the Gamma function, or that pairwise independence does not imply mutual independence (although a better flashcard would ask you to recall an actual counterexample), or what each probability distribution is useful for (I don't think I'd ever want to memorize the mean/variance, though).

I agree with you that overly relying on flashcards will take away the intuition, which is critical in fields like mathematics. At the same time, one can use flashcards in addition to other methods. It's not as if using one approach impacts your ability to learn via other approaches. Also, I think one can be clever in how they use them for mathematics. A lot of theorems rely on a certain key trick or two, so I think it makes sense to have flashcards asking what the trick is. I think flashcards are great for memorizing counterexamples, as well. Or heck, even regular examples. Depending on the branch of mathematics, knowing these is immensely helpful (e.g. in analysis).

Having said all that, I'm not using it for math presently.

> I agree with you that overly relying on flashcards will take away the intuition, which is critical in fields like mathematics.

Can't you just write a card that specifically asks you to give the intuition behind some concept?

E.g. "What's the intuitive interpretation of the gradient of a scalar field, \nabla \phi?" Answer: "\nabla \phi is a vector that points in the direction of greatest increase of \phi."


The link works, but not if you follow it from here.


You're right; clicking it fails but copy-pasting works.

Weird, why would that be?

EDIT: Well, copy-pasting opens the site but not the article.

For some reason it got truncated


Another link to the same page that works: https://www.supermemo.com/articles/20rules.htm

Doesn't work for me.

This is an alternative link I've been using this week personally:


I just wish the mobile supermemo version let you export your cards. I left supermemo a year ago for anki because of that - don't want to get locked in when I have thousands of cards.

The windows version apparently does allow exports, but it seems separate from the web version and does not run on Linux.

"Do not learn if you do not understand"

Seems strangely backward to me. Understanding is my end goal when learning.

Learning is not the same as studying, but if English is not your mother tongue you can easily mistake one word for the other. e.g. in german it is the same word. I assume it was written by a polish native speaker. Apparently in polish they also only have on word for it - uczyć.

The material you're using to learn has two levels: the concepts it tries to explain, and the concepts used to explain them. For the material to be truly useful you already need to have an understanding of the latter.

Some learning, of course, is by example/exposure rather than by explanation, otherwise you'd never be able to even get started.

I see this as a warning against rote memorization.

Page isn't working for me:


Same here

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact