Effective Spaced Repetition • Hacker News Board

Re: Effective Spaced Repetition

#251

Post by jbstack » Tue, Apr 11, 2023, 3:29 PM UTC

typon wrote: post #85

ekkeke wrote: post #13

Big fan of spaced repitition, especially for language learning. Unfortunately I feel like it fares worse for topics that require more application instead of memorisation, like mathematics or electrical engineering. Would love know if there was some super effective way to learn these similar to spaced repitiion. So far, the only thing that really works for me is solving lots of problems until I have the technique mast…

Is Duolingo basically spaced repetition for language learning?

Duolingo's spaced repetition is poor to non-existent. The whole point of spaced repetition is to prompted to remember information at the right time, but Duolingo relies on you to decide what to do and when. In the new Duolingo layout there is some pseudo-SRS in that the lessons are ordered such that concepts will be presented a few times with increasing gaps between them, but it's still on you to decide whether and when to do new lessons, and once you've done all the repetitions of the lesson you will not see it again unless you decide to go back and revisit it (in which case there is no help to decide when to go back to it).

Personally, I dislike having every app implement (or not) it's own version of SRS, so I combine language apps with Anki. For each lesson in an app, I make an Anki card which simply tells me to revisit that lesson. I then put those cards in a special deck with customised settings with larger review gaps so that I'm not overwhelmed with time-consuming lessons.

Re: Effective Spaced Repetition

#252

Post by jbstack » Tue, Apr 11, 2023, 3:41 PM UTC

michaelcampbell wrote: post #241

> Rule: Understand First Good in theory. Sometimes time crunches don't allow that, and to be honest after having used Anki for many years there are quite a lot of times where I rote-memorized a card and as I learned more about the subject the understanding came later. Usually this is when there are several cards hitting different aspects of a given topic; as I gain understanding in one area, it comes in another. So t…

I've found a similar logic applies to language learning. There's plenty of advice out there dismissing book / flashcard learning and saying you just need to get out and practice speaking with natives. In reality, you can't understand someone or form a sentence in reply if you don't know any of the words. Memorising thousands of words / phrases might not make you fluent but it's an important first step. When you're confronted by a real situation you'll now be able, with a lot of struggling, to put together a vaguely correct sentence. Eventually it will come naturally.

Re: Effective Spaced Repetition

#253

Post by jbstack » Tue, Apr 11, 2023, 4:06 PM UTC

swinglyj wrote: post #213

throwaway675309 wrote: post #191

"This is the most important thing. By far the worst failure mode is to put too much in a flashcard." He's wrong on this one. The problem with making extremely atomic flashcards is that many times it's impossible to prevent slight amounts of overlap between the cards. What this means is that answering card A might contain information that allows you to answer some subsequent card B. As a result you end up effectively…

Overlap isn't actually an issue. Overtime Anki will spread the cards out more so you don't see them together.

You can also make the related cards as siblings and use the bury feature so that you can only ever see one on a given day. I frequently use this to learn lists e.g.:

Card 1 front: Countries in the UK: 1. [...] 2. 3. 4. Card 1 back: Countries in the UK: 1. England 2. Scotland 3. Wales 4. Northern Ireland

Card 2 front: Countries in the UK: 1. England 2. [...] 3. 4. Card 2 back: Countries in the UK: 1. England 2. Scotland 3. Wales 4. Northern Ireland

Card 3 front: Countries in the UK: 1. 2. Scotland 3. [...] 4. Card 3 back: Countries in the UK: 1. England 2. Scotland 3. Wales 4. Northern Ireland

Card 4 front: Countries in the UK: 1. 2. 3. Wales 4. [...] Card 4 back: Countries in the UK: 1. England 2. Scotland 3. Wales 4. Northern Ireland

Because the cards are siblings, I'll only ever be tested on one in a given day. Each item in the list (aside from the first) is prompted by the preceding item, so that your brain learns to generate the entire list from memory sequentially. The same technique can be used to memorise a long quote by breaking it down into a list of chunks.

I also have a note type that consists of multiple cloze fields so that I can capture related information as siblings.

Re: Effective Spaced Repetition

#254

Post by s-video » Tue, Apr 11, 2023, 4:13 PM UTC

zetalyrae wrote: post #35

I wrote this. Ask me anything!

1. Have you read Andy Matuschak's guide to prompt writing? Your post reminded me of it. https://andymatuschak.org/prompts/

2. Do you have any sort of guide or principles for note-taking? I'm always debating whether or not it's worth taking notes, and when I do take notes I'm debating what the best way to do it is. (Hierarchal/bulleted information like in your post, or summarizing things in paragraphs, or what) A lot of times it's unclear to me what information is worth writing and it frustrates me.

Re: Effective Spaced Repetition

#255

Post by CobrastanJorji » Tue, Apr 11, 2023, 4:55 PM UTC

brigandish wrote: post #230

bsder wrote: post #228

Earlier quoted context omitted.

Agreed. I pulled the top 1000 Japanese words and then entered those along with the specific kanji readings. Learning wonky kanji readings is for those masochists who want to pass JLPT Level N1. You want to memorize in usage order as you are, nominally, an adult. You understand complex concepts and have developed motor skills. There is little point in learning in the order that a child would.

If I remember correctly (ha), WaniKani is actually ordered by the simplicity of the kanji, as they also point out that children can only start with simple concepts but non-native adults have the concepts covered but will struggle with complex kanji .

You do. Most off the early levels are kanji that a first grader would know, and it very loosely stays organized by grade level, with a lot of allowances.

Re: Effective Spaced Repetition

#256

Post by kaba0 » Tue, Apr 11, 2023, 6:38 PM UTC

c7b wrote: post #146

tiagod wrote: post #91

>A common failure mode (and I did this more than once, before I got the hang of it) is to use Anki for two weeks, then drop it, and pick it back up six months later only to find you have 600 cards due for review. This is not encouraging, and it defeats the point of spaced repetition, which is to review the cards on the intervals the algorithm chooses. >I don’t have much advice in this area, except that if you have pe…

The SM algorithms, trying to predict the perfect time for reviewing a question, seem like overengineering to me (and also not very well supported by research afaik - we have evidence that repetitions help, but much less what the perfect spacing is). Just randomly sample questions with probabilities proportional to how 'urgently' you need to see them, seems like the most intuitive approach to me. No daily quotas, no r…

It may have been linked already from this thread, but ebisu seems to be just that (it has a really great website detailing the math behind, plus a Python implementation)

Re: Effective Spaced Repetition

#257

Post by dalmo3 » Tue, Apr 11, 2023, 7:44 PM UTC

Tried Anki for a few months for learning programming concepts and internalizing ideas from books. I created all flashcards based on my own notes, for stuff I understood, did everything by the book.

After a while I realised I was visually learning the flashcards, i.e. when reading the prompt side I would see a photographic memory of the answer and get it right, but not necessarily think about the meaning of the contents. After a 15-20 min session I would remember very little of what I had just studied. Then I quit.

It reminded me of how trained chess players visualise the game as patterns instead of a collection of pieces.

Anyone with a similar experience?

Re: Effective Spaced Repetition

#258

Post by xk_id » Wed, Apr 12, 2023, 2:06 AM UTC

kurthr wrote: post #224

xk_id wrote: post #139

Earlier quoted context omitted.

Consensus among Serious language learners (i.e aiming for a B2+ level) is that duolingo is marginally useful for learning some vocabulary. A thorough course or textbook are essential

I used it for most of a year, and from what I can tell they go out of their gamified way to not teach any grammar or even vocabulary. Not all the time but ~20% popping up sounds and spellings, but providing no way at all to know/check what the meaning is... Why? Why not have a dictionary of the words you've "learned" and will be tested on? Going to a fixed sequence of learning "the path" so that you're forced to cove…

I'd say the answer is because they don't measure success in terms of who becomes competent in the language, but in how many users they have. typical neoliberal "number go up" mentality, rejecting ideological definitions of "genuine" value

https://youtu.be/m8Mc-38C88g?t=176

Re: Effective Spaced Repetition

#259

Post by stavros_ » Wed, Apr 12, 2023, 8:45 AM UTC

zetalyrae wrote: post #172

LVB wrote: post #88

Earlier quoted context omitted.

Great article? I'm curious what your Anki settings are, if you've changed anything. I'm pretty new to the app, but when I hit things like "you'll need see this card in 5 years" I had to dive into settings and start tweaking stuff. I'm more concerned about definitely remembering it in say 12 months than having too high of a load. But there are many adjustments to achieve this, so any thoughts?

I use Mochi with all settings out of the box. Mochi doesn't even implement the SM-2 algorithm, it just applies a multiplier to the review interval on getting it right/forgetting.

Triggered.

I spent so much time and effort comparing the various spaced repetition software before finally, grudgingly, choosing Anki over Mochi.

My rationale was that, while Mochi's UX is amazing, the algorithm matters. SuperMemo seems to justify this - SM18+ is a highly refined algorithm and 'seems' to provide much, much better performance than SM2/Anki.

So Mochi's incredibly basic 'engine' seems, by this logic, to be a pretty significant downside.

In the end what I care most about is memorization, and it seems like the best 'engine' for that is (Supermemo if you're on Windows, otherwise...) Anki + FSRS.

I'd appreciate being convinced I'm wrong - Using Mochi was a vastly more pleasant experience than using Anki.

Re: Effective Spaced Repetition

#260

Post by rockdiesel » Wed, Apr 12, 2023, 1:24 PM UTC

dalmo3 wrote: post #257

Tried Anki for a few months for learning programming concepts and internalizing ideas from books. I created all flashcards based on my own notes, for stuff I understood, did everything by the book. After a while I realised I was visually learning the flashcards , i.e. when reading the prompt side I would see a photographic memory of the answer and get it right, but not necessarily think about the meaning of the conte…

Yes. I had a similar experience. The answer would come to me before I even finished reading the prompt, but I never felt like I was actually learning the info. Just memorizing the pattern.

I surmised that it was a result of my deck being too small. It may be different with a larger deck and a greater variety of cards, but I never got my deck to that point.