Given how you set yourself up at the beginning of this post, I’m actually pretty impressed by how far you’ve come. You should feel proud, too. If you continue to self-improve and try stuff out at the rate you have been in the past year, you ought to be in a pretty good position in a few years.
If something isn’t working for you (e.g. mindfulness), you might want to try something else instead of continuing to try to make it work. For example, have you experimented with an exercise regime? (This post has some recommendations that I’ve been following.) I think I remember reading that CFAR found that people who tried out more things tended to see more things stick.
It seems like you’ve had a lot of success with measuring your time usage and making your goal of spending your time better in to an urge. Could you create more sophisticated measurements for your productivity? For example, exercises completed in Artificial Intelligence: A Modern Approach. Give yourself “points” according to the difficulty of material and the amount of it that you read. Or some kind of rubric that describes how productively you spent the past hour. This would both get you to spend your time more effectively (since you’d be motivated to actually accomplish things rather than log hours) and also maybe once you replaced your motivation system with a more effective one, you’d feel comfortable letting off on the time-tracking and letting your breaks better rejuvenate you.
I agree with the book The Power of Full Engagement that energy management matters more than time management and that highly rejuvenating breaks are critical. I think your idea of trying to maximize breaks for rejuvenation per minute is a good one. You might want to experiment with lots of different break activities and take notes on which ones work. (For example, an acquaintance swears by lying down, closing one’s eyes, and daydreaming about things that aren’t work.) Based on the results of productivity studies, I’m inclined to agree with this Quora answer that a focused 40 hours a week is what you should aim for.
In terms of studying, one idea is to aim to pass specific AP exams. This will get you out of taking college classes, so you won’t have to learn the same material twice. It’s pretty reasonable to be wondering what you should study. I think it would be a good idea to start taking notes related to all the topics you should study, pros, cons, etc. and maybe ask some questions on topic-specific message boards as well. This thinking up front will probably be worthwhile and may help you stick to particular topics. You might want to try to identify one topic to self-study in (e.g. programming is good since computers give you honest feedback) and really concentrate on that, then major in something else in college (e.g. math). For more in-depth advice, you might want to contact Cognito Mentoring… it seems like you’re exactly the sort of person they want to advise/help.
I think in general, achievement-oriented studying might be better than learning-oriented study. Instead of trying to learn electrical engineering just from a textbook, build a pocket amplifier and aim to really understand how it works (note on circuits in particular: my old physics teacher used to say that students who only had practical knowledge of circuits tended to do poorly when asked to reason about them theoretically. So you do want to understand what you are doing on a theoretical level and a textbook may be helpful there.) Instead of trying to read an AI textbook, write an AI to play a specific game (reading the chapters of the AI textbook as they are relevant, maybe skimming them at the beginning so you know where to look as you run in to problems). This has a few benefits: it’s learning by doing, achievements are better for your resume than books you’ve read, and you know that you’re learning useful skills and wrestling with reality. MOOCs are also a good way to go since they get you certificates, a community of other learners, deadlines, and graded assignments so you can learn by doing.
Another autodidact hack: I’m not much of an autodidact really, but I found that I preferred lecture notes to textbooks when it comes to teaching myself. Textbook authors have an incentive to make their textbooks long so they will cost more money. Lecturers have no such incentive. And lectures will often be an informal style that seeks to teach rather than impress. Something really cool you can do with lecture notes is: if you know that the material in a given set of notes was presented over the course of 1 hour, you can set a timer for 1 hour and then aim to get through the entire PDF of the lecture notes in just an hour, taking notes in your own terms on a sheet of paper for comprehension. This gives you a great source of motivation and allows you to go at your own speed rather than the lecturer’s speed (skipping things you already understand; soaking up concepts that seem tricky to make sure you really get them). For the sake of learning, I think it’s probably better to spread your study of a given topic out over many days so it will stick in your long-term memory. So you probably want to do something kind of like what they do in college and be studying many things gradually in parallel. Remember that you get knowledge points for every lecture you consume. To find lecture notes, look on the course websites for courses teaching what you want to learn.
My quick and dirty intuition is leaning towards the Stanford OHS. It’d be good for your college application (I assume) and help you overcome your autodidact issues. My impression was that you can retake the SAT and just send your best score but correct me if I’m wrong.
Hey, thanks for your comment; it’s provided me with some very helpful insights. I’ll now be responding to each of your points in turn.
I appreciate the optimism. Often I feel proud of myself when looking back over the past year, but I can never stop stressing over current issues. I still predict that, even if I never further optimize my habits, I will indeed be in a good position in a few years.
As I said in the discussion, I’ve tried a lot of things; the things that worked are my current habits. I’ve had a sort of exercise regime from the beginning (early on, I was mostly dedicated to improving my health) that has evolved over time. Based on the article you linked, it could certainly use some optimizations, however; I’ll be changing some things around.
I’ve never tried using a ‘point system’ or productivity rubric before; those seem like good ideas. I definitely need to make myself (in the short-term) want to accomplish things rather than log hours. I’m thinking I’ll create some kind of productivity log that I can also use to summarize what I just read/worked on.
That book you just mentioned is so extremely relevant I wonder how I have missed it. I’d like to thank you for including it. Anyway, I’ll also look into experimenting with different structured break activities aside from leisure computer use/reading, like another form of meditation. If I can successfully implement some of the studying tactics I’m talking about, I think I’ll be able to more easily partake in fun time.
Indeed one of my medium-term goals is to pass AP exams; I will only be focusing on the most relevant ones, however. I haven’t tried formally analyzing topics to see if I should study them or when; that seems like another good idea. I’m planning on studying all the topics presented in the MIRI recommended course list, but I will most likely major in CS. I’ve already sent an email to Cognito Mentoring, actually; I’m optimistic about their potential to help guide my self- education.
I agree with you here. Most of my game programming skill came from working on my own projects. My reply to another comment in this thread is very relevant. I’m going to take some MOOCs (not just the ones in the MIRI course list), but I’m unsure of how I should fit them into my longer-term study plan. I’m also interested in experimenting with studying lecture notes. A lot of the courses/lectures available are probably above my level, though, so I might have a hard time finding some that I can use.
I didn’t mention it in the discussion, but most likely I won’t attempt to transfer to Stanford OHS. It’s too much of a rush/gamble, I want to keep my total control over my education, and I believe that I will (eventually) become an acceptably efficient autodidact.
As I said in the discussion, I’ve tried a lot of things; the things that worked are my current habits.
Cool; glad to hear you’re trying things.
I will most likely major in CS
That’s a good option, but as a CS major if you’re serious about doing AI research for MIRI I think I would be leaning towards math. It’s much easier and more effective to self-credential in CS than math by having a GitHub full of cool projects. With math you actually benefit from a classroom environment to a much greater degree in my opinion. Also, my impression was that MIRI was more in need of math talent than CS talent. I think math + self-taught CS is good from a career perspective as well (math signals intelligence to CS employers; math + CS is good for quant/data scientist type jobs).
A lot of the courses/lectures available are probably above my level, though, so I might have a hard time finding some that I can use.
Yep, that’s something tricky about being an autodidact.
I didn’t mention it in the discussion, but most likely I won’t attempt to transfer to Stanford OHS. It’s too much of a rush/gamble, I want to keep my total control over my education, and I believe that I will (eventually) become an acceptably efficient autodidact.
Fascinating post!
Here are some of my thoughts:
Given how you set yourself up at the beginning of this post, I’m actually pretty impressed by how far you’ve come. You should feel proud, too. If you continue to self-improve and try stuff out at the rate you have been in the past year, you ought to be in a pretty good position in a few years.
If something isn’t working for you (e.g. mindfulness), you might want to try something else instead of continuing to try to make it work. For example, have you experimented with an exercise regime? (This post has some recommendations that I’ve been following.) I think I remember reading that CFAR found that people who tried out more things tended to see more things stick.
It seems like you’ve had a lot of success with measuring your time usage and making your goal of spending your time better in to an urge. Could you create more sophisticated measurements for your productivity? For example, exercises completed in Artificial Intelligence: A Modern Approach. Give yourself “points” according to the difficulty of material and the amount of it that you read. Or some kind of rubric that describes how productively you spent the past hour. This would both get you to spend your time more effectively (since you’d be motivated to actually accomplish things rather than log hours) and also maybe once you replaced your motivation system with a more effective one, you’d feel comfortable letting off on the time-tracking and letting your breaks better rejuvenate you.
I agree with the book The Power of Full Engagement that energy management matters more than time management and that highly rejuvenating breaks are critical. I think your idea of trying to maximize breaks for rejuvenation per minute is a good one. You might want to experiment with lots of different break activities and take notes on which ones work. (For example, an acquaintance swears by lying down, closing one’s eyes, and daydreaming about things that aren’t work.) Based on the results of productivity studies, I’m inclined to agree with this Quora answer that a focused 40 hours a week is what you should aim for.
In terms of studying, one idea is to aim to pass specific AP exams. This will get you out of taking college classes, so you won’t have to learn the same material twice. It’s pretty reasonable to be wondering what you should study. I think it would be a good idea to start taking notes related to all the topics you should study, pros, cons, etc. and maybe ask some questions on topic-specific message boards as well. This thinking up front will probably be worthwhile and may help you stick to particular topics. You might want to try to identify one topic to self-study in (e.g. programming is good since computers give you honest feedback) and really concentrate on that, then major in something else in college (e.g. math). For more in-depth advice, you might want to contact Cognito Mentoring… it seems like you’re exactly the sort of person they want to advise/help.
I think in general, achievement-oriented studying might be better than learning-oriented study. Instead of trying to learn electrical engineering just from a textbook, build a pocket amplifier and aim to really understand how it works (note on circuits in particular: my old physics teacher used to say that students who only had practical knowledge of circuits tended to do poorly when asked to reason about them theoretically. So you do want to understand what you are doing on a theoretical level and a textbook may be helpful there.) Instead of trying to read an AI textbook, write an AI to play a specific game (reading the chapters of the AI textbook as they are relevant, maybe skimming them at the beginning so you know where to look as you run in to problems). This has a few benefits: it’s learning by doing, achievements are better for your resume than books you’ve read, and you know that you’re learning useful skills and wrestling with reality. MOOCs are also a good way to go since they get you certificates, a community of other learners, deadlines, and graded assignments so you can learn by doing.
Another autodidact hack: I’m not much of an autodidact really, but I found that I preferred lecture notes to textbooks when it comes to teaching myself. Textbook authors have an incentive to make their textbooks long so they will cost more money. Lecturers have no such incentive. And lectures will often be an informal style that seeks to teach rather than impress. Something really cool you can do with lecture notes is: if you know that the material in a given set of notes was presented over the course of 1 hour, you can set a timer for 1 hour and then aim to get through the entire PDF of the lecture notes in just an hour, taking notes in your own terms on a sheet of paper for comprehension. This gives you a great source of motivation and allows you to go at your own speed rather than the lecturer’s speed (skipping things you already understand; soaking up concepts that seem tricky to make sure you really get them). For the sake of learning, I think it’s probably better to spread your study of a given topic out over many days so it will stick in your long-term memory. So you probably want to do something kind of like what they do in college and be studying many things gradually in parallel. Remember that you get knowledge points for every lecture you consume. To find lecture notes, look on the course websites for courses teaching what you want to learn.
My quick and dirty intuition is leaning towards the Stanford OHS. It’d be good for your college application (I assume) and help you overcome your autodidact issues. My impression was that you can retake the SAT and just send your best score but correct me if I’m wrong.
Hey, thanks for your comment; it’s provided me with some very helpful insights. I’ll now be responding to each of your points in turn.
I appreciate the optimism. Often I feel proud of myself when looking back over the past year, but I can never stop stressing over current issues. I still predict that, even if I never further optimize my habits, I will indeed be in a good position in a few years.
As I said in the discussion, I’ve tried a lot of things; the things that worked are my current habits. I’ve had a sort of exercise regime from the beginning (early on, I was mostly dedicated to improving my health) that has evolved over time. Based on the article you linked, it could certainly use some optimizations, however; I’ll be changing some things around.
I’ve never tried using a ‘point system’ or productivity rubric before; those seem like good ideas. I definitely need to make myself (in the short-term) want to accomplish things rather than log hours. I’m thinking I’ll create some kind of productivity log that I can also use to summarize what I just read/worked on.
That book you just mentioned is so extremely relevant I wonder how I have missed it. I’d like to thank you for including it. Anyway, I’ll also look into experimenting with different structured break activities aside from leisure computer use/reading, like another form of meditation. If I can successfully implement some of the studying tactics I’m talking about, I think I’ll be able to more easily partake in fun time.
Indeed one of my medium-term goals is to pass AP exams; I will only be focusing on the most relevant ones, however. I haven’t tried formally analyzing topics to see if I should study them or when; that seems like another good idea. I’m planning on studying all the topics presented in the MIRI recommended course list, but I will most likely major in CS. I’ve already sent an email to Cognito Mentoring, actually; I’m optimistic about their potential to help guide my self- education.
I agree with you here. Most of my game programming skill came from working on my own projects. My reply to another comment in this thread is very relevant. I’m going to take some MOOCs (not just the ones in the MIRI course list), but I’m unsure of how I should fit them into my longer-term study plan. I’m also interested in experimenting with studying lecture notes. A lot of the courses/lectures available are probably above my level, though, so I might have a hard time finding some that I can use.
I didn’t mention it in the discussion, but most likely I won’t attempt to transfer to Stanford OHS. It’s too much of a rush/gamble, I want to keep my total control over my education, and I believe that I will (eventually) become an acceptably efficient autodidact.
Cool; glad to hear you’re trying things.
That’s a good option, but as a CS major if you’re serious about doing AI research for MIRI I think I would be leaning towards math. It’s much easier and more effective to self-credential in CS than math by having a GitHub full of cool projects. With math you actually benefit from a classroom environment to a much greater degree in my opinion. Also, my impression was that MIRI was more in need of math talent than CS talent. I think math + self-taught CS is good from a career perspective as well (math signals intelligence to CS employers; math + CS is good for quant/data scientist type jobs).
Yep, that’s something tricky about being an autodidact.
Seems reasonable.