Ask HN: Recommendations for Spaced Repetition (SRS) Beginners?
Someone posted a link to a Spaced Repetition article a few days ago [0]. I'd never heard of the concept before, and it sent me down a rabbit hole of different introductory and literature review articles [1-5], more in-depth discussions on how to adapt the method for more complex topics like mathematics [6-8], some previous HN discussions [9-11], various mainstream tools [12-15], as well as some tools currently being built by HN community members [16-18] (sorry if I missed anything important there). I've also gotten started and created and practized my first few dozen flash cards on some new topics I recently learnt, and the first experience has left me wondering a few things:
1. Are there downsides to this? Most reviews seem highly appreciative mentioning practically no downsides besides the time commitment, which supposedly pays off many times over, which just always makes me skeptical. Downsides I could imagine include eg: recent neuroscience suggests that forgetting & memory rearrangement are important elements of brain hygiene, related to decision-making, avoiding things like PTSD and others [19]. How sure are we that we're not messing with those?
2. It seems that the most recommended approach is to turn things you want to remember into a set of atomic questions that take <10s to answer. Are you aware of approaches that integrate better with note-taking? I wouldn't reviewing an interesting essay or reproducing an elegant proof in full, not just small tidbits.
2a. + Any approaches that integrate well with journalling? I wouldn't mind being reminded of a nice weekend trip with some pictures from time to time in a purely 'here's a nice memory' fashion, not questions like 'Who did you go on a trip with on Feb 9 2020?'.
3. So the basic workflow I could much rather see myself adopting on a long-term basis is something like: take 10+10mins each day to write down what you learnt or did in the past 24h that seems worth remembering, and review things from previous days (with some spacing algorithm behind it, and an option to remove things entirely). Do you know if there is a name for that / any science to it / tools that support it (it seems the Ilse notebook might be going in that direction)?
[0] https://news.ycombinator.com/item?id=32381206
Introductions: [1] https://www.gwern.net/Spaced-repetition [2] https://andrewjudson.com/spaced-repitition/2022/06/03/spaced-repitition.html [3] http://augmentingcognition.com/ltm.html [4] https://andymatuschak.org/prompts/ [5] https://www.wired.com/2008/04/ff-wozniak/
SRS for math: [6] https://cognitivemedium.com/srs-mathematics [7] https://cronokirby.com/posts/2021/02/spaced-repetition-for-mathematics/ [8] https://blogs.ams.org/mathgradblog/2011/03/10/space-math-repetition-repetition/
Previously on HN: [9] https://news.ycombinator.com/item?id=18895613 [10] https://news.ycombinator.com/item?id=24857437 [11] https://news.ycombinator.com/item?id=9833581
Mainstream tools: [12] https://apps.ankiweb.net/ - seems to be the most popular [13] https://super-memo.com/ - the OG [14] https://www.supermemo.com/en - web-app alternative to super-memo [15] https://mnemosyne-proj.org/ - OS alternative to Anki
HN community tools: [16] https://github.com/ilse-langnar/notebook [17] https://www.freshcardsapp.com [18] https://getpolarized.io/
Contrarian: [19] https://time.com/6171190/new-science-of-forgetting/
I've researched learning/ behavioural science for well over a decade. I like the approach that you've outlined.
What you've described is building a habit of reflection - which is a proven approach to learn and develop, plus spaced learning, which is also a very proven approach.
The main point I'd call out is that rather than just thinking about 'spaced-repitition', its' better to think of 'spaced retrieval'. The retrieval is crucial, because it challenges illusions of competence (the distinction between recognising something and actually knowing it). Importantly, retrieval is not just about self testing, the 'retrieval' part might also involve sharing, application, and/or experimentation.
Finally, spaced retrieval tends to focus on memory/knowledge. That is different to building skills. For building skills you are best to adopt Deliberate Pratice techniques, ie. cycles of focused practice with 'expert' feedback to identify what good looks like.
There's more of all of this on this Learning Playbook: https://modelthinkers.com/playbook/future-proof-with-learnin...
cheers A
Looks like an amazing resource, thank you!
> Are there downsides to this?
Absolutely. Just to give an example, I was able to narrow down my personality dynamics to a set which predicted that memory-focused interests would be intriguing but risky.
I measured the risk as I studied for upcoming test (US amateur radio license) and the subjective intrigue factor seemed to be the worst factor by far. It compounded risk early and often. As I saw early payoffs from spaced repetition and other techniques like Roman rooms, the intrigue convinced me to up the ante again and again.
I had little experience with electronics or radio and nonetheless decided to study for all three license levels simultaneously.
I ended up having to measure my success not by progress, but by frequency of extremely negative ahem, "ideations". IOW I would quit studies for the day/week when I wanted to end it all. I had the granularity figured down to how many minutes I likely had to study before I'd have to lose a day of all other productive work due to exhaustion. But the prize was so big that I couldn't just give up. And all of the memory techniques worked great.
I wasn't able to find a hyper-objective model for this but the mixed models were good enough for me. In my case the most relevant was Jungian cognitive functions' Si-as-recall model, combined with several other perspectives from the personality dynamics world, including theories by John Beebe.
If you are able to do some basic research and determine whether you are an intuitive personality, that would be where I'd start to draw a big warning line. Especially if you have strengths in conceptualization, contingency planning, imagination, reading science fiction, and so on. Not only are these strengths often found in recall-deficit personality groups, but moreso the more detailed the nature of the recall. In addition, recall perception blocks intuitive perception and vice versa. So by definition the expression of your gifts can become blocked, which logically means that you are blocking your ability to create positive outcomes and energy for yourself.
It's interesting stuff but please be careful and mind the subjective factors. Especially if you have any thoughts like, "I learn some really interesting stuff, so if only I could retain what I learn..." (I'd never be caught blanked out in a conversation, losing a debate, etc.)
> Especially if you have any thoughts like, "I learn some really interesting stuff, so if only I could retain what I learn..."
I'm not sure I follow, is that a risk factor or a positive indication in your opinion? My goal isn't to shine with this knowledge in conversations (apart from technical job interviews maybe).
There's a set of personality dynamics that works like this. Really common in tech:
- My unspoken (if even consciously known) mission is to not be caught dead not-knowing things.
- So I will always learn things. I keep buying books, I subscribe to learning-focused services, use my library card, etc.
- However, I have this really damn embarrassing tendency to forget stuff, just at the wrong time!
- I've already got the big picture down--that's boring, I could talk about the big picture all day (future, low detail) and I can generally see where things are going in the world.
- So, I will work on my memory, bc that's interesting. (past, high detail, can be overlaid on the future for some pattern extrapolation even)
To some, there's an additional performer-complication: "My memory will help me perform well, and a good performance is my unspoken mission in life." This could apply to debates, job interviews, who knows what.
This loose chain of various factors would, in my mind, point at someone who is likely to overidealize the impact of memory improvements in their life.
It also points at someone who should work on subjective knowledge organization before getting into any memory improvements. They should have an external system which for all practical purposes supports the most likely or common holes in their memory. This will also give them designer-builder energy that they usually also crave, it will solve memory problems in the big picture (using a perceptive view-gift of the future to solve a weakness).
So it's like a loose subjective risk factor based on some alignment or non-alignments with those things.
Otherwise, some amount of memory work is really handy. I mean you'll never forget a thing you really want to remember, ever again. It's a muscle that requires conscious work, but still...I mean I can remember grocery lists from over 30 years ago, critical film theorists from 20 years ago, and how could I ever forget seeing my favorite pair of lazy-day flip-flops looking up at me from within a circuit diagram? That kind of thing.
Yeah, the whole 'life-hack' mindset can be quite toxic. But frankly, my other procrastination habits (mostly browsing comment sites like HN) aren't much better (actually quite bad [0]), so if this one comes with a timer and a daily limit it may actually help reduce screen time. Funny enough, the very first thing I 'ankified' just before writing the post was something I picked up very recently as an alternative to browsing the phone on my commute, solving a Rubik's cube.
[0] https://www.nytimes.com/2019/04/24/well/mind/putting-down-yo...
The downside to me is prioritization and effort.
Is everything you learn worth remembering?
If it’s worth remembering, how much time should you contribute to this effort?
There are posts of people complaining that their review sessions take hours. Maybe worth it for an exam, but not a life style.
There is an adjacent concept called “incremental reading” as well.
For myself, I find it quite useful. It’s a relatively modern invention so the tooling feels slightly immature. There’s supermemo, the original srs system that only runs on windows and anki that is pretty good but missing a lot of functionality that supermemo offers.
I find it very useful. I remember the things that I want to remember but there’s definitely a cost associated with it.