I’m open to the idea and I probably haven’t thought about it as much as you, but I’m skeptical about the way you discuss going about it in your post and also that the work of the experts that seem to have inspired you is impactful.
I suspect the techniques you’ve discussed will greatly improve your memory, but I’d guess that it’s often not worth the time to memorize something. Based on my experience working as a software engineer, my attitude has been you can’t learn everything . At least in software engineering, you need to adapt to new languages and frameworks frequently and I’d imagine other relatively new fields are similar. In software engineering (and maybe these other fields) the most (if not the only) important thing to remember is what you need to google.
Additionally, I’d also guess that a lot of the truly most impactful work comes from learning about new domains . This also can’t be learned through memorization. Even googling the right questions often won’t give you a quick answer. Haseeb Qureshi wrote a post about this type of learning, which he calls “unstructured learning” that I think is worth reading https://haseebq.com/the-hard-thing-about-learning-hard-things/
I’m aware that you said you mentioned you’re not targeting superstars, but as far as I know there’s no reason superstars would learn differently than anyone else. I’d also guess anyone can get a lot more out of themselves if they do their best.
I’d also guess that it would be hard to make your idea be much more than an incremental improvement over the Learning How To Learn course. I admittedly haven’t taken that course so take that with a big grain of salt.
Maybe it could be worthwhile to try to create a course or program that focuses on teaching unstructured learning. I don’t know how you’d go about it and I think there’s a high chance the course would be crap, but maybe its worth the risk?
Cheers for the reply! Some thoughts from your comment:
Target audience/ sectors where this would be most useful
I definitely agree that in general what I have in mind is academia/research-type fields as the sectors where this system would be especially useful, particularly in committing to memory new ideas from fields, research papers etc. Whilst I’ve had some success using flashcards to learn Python and some other comp sci-adjacent things, it’s definitely the case that in programming your learn primarily by doing. I think the flashcards still help a great deal in i.e. ensuring I remember the essentials of a particular Python library despite not having used it for a long time, but I’d definitely agree overall that it’s less useful in programming. I’ve also ran into the bad habit of making coding flashcards on things I a) haven’t fully understood or b) haven’t really needed, which has wasted time—these are some of the things to avoid that I’d definitely make sure to cover.
Unstructured learning
Unstructured learning is a new concept to me so I’m excited to check it out! This sounds like it could be something I’ve totally missed from my workflow that could introduce some really big gains.
Intellectual superstars
Re: not targetting superstars, I was more saying that I’ve found myself struggling with motivation when I consider the career path & potentially early opportunities or many of the most well-known EAs / 80,000 hours interviewees, and I think this has the utility of empowering more people who come to EA and being productive late (for example I squandered a large part of my teenage years playing video games) to make up for lost time. Superstar academics could also benefit from the techniques, but they’d be (naturally) less likely to need them due to their potentially early start in academia (I’m thinking of a Yudkowsky here) & incredible IQs.
Competing with/ differentiating from Learning How To Learn
Re: “Learning How To Learn”—I love that course but found it mainly focused on theory rather than a detailed step-by-step guide on what tools to use, use patterns and antipatterns etc. My aim here would be to collect everything useful in one place (whilst keeping it concise), whereas I think LHTL lays a great foundation but a) is relatively long (being a video course) and incomplete.
Utility of memorisation
A final point of the utility of memorisation: I’ve found that it’s not simply about memorising the correct answers to things, it’s about increasing the speed at which you can recall potentially relevant information (referred to as “fluency” in the literature), allowing you to think of potential solutions faster, or use some cross-disciplinary data etc. In the case of coding, this means reducing Googling-time, which may be a marginal gain as Googling doesn’t take long, but I found made coding more enjoyable. But this idea of fluency has far more profound implications in research, where absorbing and internalising information from textbooks and seminal papers allows you to draw more elaborate connections between ideas and fields, and engage more deeply with further research, including developing original research questions (a lot of this I’m getting from the Michael Nielsen piece linked in my post rather than my own experience but something I 100% stand behind). A big focus for me in this website would be getting across the diverse & powerful utility & profundity of memorisation i.e. disabusing people of the notion that it’s simply about rote learning, something Nielsen does really well.
”With a few days work I’d gone from knowing nothing about deep reinforcement learning to a durable understanding of a key paper in the field, a paper that made use of many techniques that were used across the entire field “
”for creative work and for problem-solving there is something special about having an internalized understanding. It enables speed in associative thought, an ability to rapidly try out many combinations of ideas, and to intuit patterns, in ways not possible if you need to keep laboriously looking up information”
I’m open to the idea and I probably haven’t thought about it as much as you, but I’m skeptical about the way you discuss going about it in your post and also that the work of the experts that seem to have inspired you is impactful.
I suspect the techniques you’ve discussed will greatly improve your memory, but I’d guess that it’s often not worth the time to memorize something. Based on my experience working as a software engineer, my attitude has been you can’t learn everything . At least in software engineering, you need to adapt to new languages and frameworks frequently and I’d imagine other relatively new fields are similar. In software engineering (and maybe these other fields) the most (if not the only) important thing to remember is what you need to google.
Additionally, I’d also guess that a lot of the truly most impactful work comes from learning about new domains . This also can’t be learned through memorization. Even googling the right questions often won’t give you a quick answer. Haseeb Qureshi wrote a post about this type of learning, which he calls “unstructured learning” that I think is worth reading https://haseebq.com/the-hard-thing-about-learning-hard-things/
I’m aware that you said you mentioned you’re not targeting superstars, but as far as I know there’s no reason superstars would learn differently than anyone else. I’d also guess anyone can get a lot more out of themselves if they do their best.
I’d also guess that it would be hard to make your idea be much more than an incremental improvement over the Learning How To Learn course. I admittedly haven’t taken that course so take that with a big grain of salt.
Maybe it could be worthwhile to try to create a course or program that focuses on teaching unstructured learning. I don’t know how you’d go about it and I think there’s a high chance the course would be crap, but maybe its worth the risk?
Cheers for the reply! Some thoughts from your comment:
Target audience/ sectors where this would be most useful
I definitely agree that in general what I have in mind is academia/research-type fields as the sectors where this system would be especially useful, particularly in committing to memory new ideas from fields, research papers etc. Whilst I’ve had some success using flashcards to learn Python and some other comp sci-adjacent things, it’s definitely the case that in programming your learn primarily by doing. I think the flashcards still help a great deal in i.e. ensuring I remember the essentials of a particular Python library despite not having used it for a long time, but I’d definitely agree overall that it’s less useful in programming. I’ve also ran into the bad habit of making coding flashcards on things I a) haven’t fully understood or b) haven’t really needed, which has wasted time—these are some of the things to avoid that I’d definitely make sure to cover.
Unstructured learning
Unstructured learning is a new concept to me so I’m excited to check it out! This sounds like it could be something I’ve totally missed from my workflow that could introduce some really big gains.
Intellectual superstars
Re: not targetting superstars, I was more saying that I’ve found myself struggling with motivation when I consider the career path & potentially early opportunities or many of the most well-known EAs / 80,000 hours interviewees, and I think this has the utility of empowering more people who come to EA and being productive late (for example I squandered a large part of my teenage years playing video games) to make up for lost time. Superstar academics could also benefit from the techniques, but they’d be (naturally) less likely to need them due to their potentially early start in academia (I’m thinking of a Yudkowsky here) & incredible IQs.
Competing with/ differentiating from Learning How To Learn
Re: “Learning How To Learn”—I love that course but found it mainly focused on theory rather than a detailed step-by-step guide on what tools to use, use patterns and antipatterns etc. My aim here would be to collect everything useful in one place (whilst keeping it concise), whereas I think LHTL lays a great foundation but a) is relatively long (being a video course) and incomplete.
Utility of memorisation
A final point of the utility of memorisation: I’ve found that it’s not simply about memorising the correct answers to things, it’s about increasing the speed at which you can recall potentially relevant information (referred to as “fluency” in the literature), allowing you to think of potential solutions faster, or use some cross-disciplinary data etc. In the case of coding, this means reducing Googling-time, which may be a marginal gain as Googling doesn’t take long, but I found made coding more enjoyable. But this idea of fluency has far more profound implications in research, where absorbing and internalising information from textbooks and seminal papers allows you to draw more elaborate connections between ideas and fields, and engage more deeply with further research, including developing original research questions (a lot of this I’m getting from the Michael Nielsen piece linked in my post rather than my own experience but something I 100% stand behind). A big focus for me in this website would be getting across the diverse & powerful utility & profundity of memorisation i.e. disabusing people of the notion that it’s simply about rote learning, something Nielsen does really well.
A few relevant quotes from his piece here:
”With a few days work I’d gone from knowing nothing about deep reinforcement learning to a durable understanding of a key paper in the field, a paper that made use of many techniques that were used across the entire field “
”for creative work and for problem-solving there is something special about having an internalized understanding. It enables speed in associative thought, an ability to rapidly try out many combinations of ideas, and to intuit patterns, in ways not possible if you need to keep laboriously looking up information”