Inspired by: On learning difficult things

In his recent post, user So8res says his number one piece of advice for learning something difficult is to have study partner to learn with you. 

Since there is a decent amount of interest here in going through the MIRI course list, it might be worth finding other people here to learn and study this with, and to form pairs or groups. 

So here is a space for finding and organizing such partnerships!

Of course, part of the reason I wrote this is because I am interested in learning these books with people. My background: I'm currently a second year Ph.D. student in mathematics (number theory). I'm still pretty new to the type of math emphasized here. I have Probabilistic Graphical ModelsCategory Theory for Computer Scientists and The Logic of Provability (by George Boolos -- not on the course list, but good to get background for the Robust Cooperation paper and for understanding Loeb's theorem) all lying around. I'm also taking a class on numerical analysis. Part of my problem is that I start lots of projects and then end up fizzling out on them, and I hope having a partner will help with this. 

I've already been going through MIRI's publications with a friend from the local LW community, which has been really nice. I'm still interested in finding more partners <insert poly joke here> for going through books on the course list specifically. I'm also willing to explain things I understand, or let someone explain things to me (I've found that explaining things to someone else is a very good way of solidifying your understanding of something) when I have time. 

Some things to consider:

  • What is the best online format for doing this? I've been doing this sort of thing with Workflowy + Mathflowy but there is probably something better. 
  • Does a pair dynamic, or a group dynamic seem more likely to work? I'm hoping that there can be a collection of pairs all centered in a common community, or something like that.
  • If a central community seems like a good idea, how should it be centralized?
  • Probably some other issues/meta stuff.



Mentioned in
New Comment
22 comments, sorted by Click to highlight new comments since: Today at 12:15 PM


I am, of course, in the market for a study partner. Please don't be intimidated by the fact that I've been at this for a couple months already: my formal education included bachelors degrees in computer science and economics, so I still feel behind all you math majors and graduate students out there.

I'm not sure how best to make this work, but here are a few of my thoughts:

I recommend against limiting studying to times when you are together. I tend to do a fair bit of studying on bus rides and over meals, and I feel that only studying together would be somewhat limiting. It might be useful to have an IRC / video hangout going in the background or something, but for me, at least, the actual reading of the text is a more private endeavor.

I think that most of the gains from pairing up will come from social incentives (agreements to read a certain amount each day/week, checkups, etc) and the ability to have conversations along these lines:

So8res > Hey, ping me when you've read past the proof at the end of p480
Partner > Already past it.
So8res > How is [step] justified? Isn't that invalid because [misunderstanding]?
Partner > Nope. [reason]

The ability to have these conversations would have increased my velocity pretty significantly. There doesn't even need to be real-time communication: the ability to email someone with heavily contextualized questions, knowing that they've been working on the same content recently, would be incredibly valuable.

I also expect there's benefit to be had by meeting up at the end of each chapter and doing the following:

1) Motivate the major concepts of each chapter to each other. There's a gap between "I understand what was said" and "I understand why that was important to say", and explaining concepts really helps me cross that gap. I'd suggest either

  • Each member drafts a brief overview of these concepts in, say, an email; you're not allowed to read the other person's email until you've sent your own.
  • You divy up the theorems and each of you explain half to the other.

The first is probably more interesting (I'd enjoy contrasting the interpretations), but the latter requires half the effort.

2) Do the problems together. As long as both parties are participating, this may speed things up significantly. You have to be careful to avoid the scenario where one party watches while the other solves problems, though. At the very least, there's benefit to be had by doing problems and comparing answers.

These are my best guesses for how to make studying easier. Keep in mind that this probably varies from person to person, and that these views are likely to change in actual practice. That said, if this sounds desirable to you, let me know and we can get something started.

The ability to have these conversations would have increased my velocity pretty significantly. There doesn't even need to be real-time communication: the ability to email someone with heavily contextualized questions, knowing that they've been working on the same content recently, would be incredibly valuable.

This is especially important IMO. Used to be when I studied more actively, whenever I'd get stuck and someone I knew happened to be online and also studying the subject (or had mastered it), I'd just ping them up on chat.


I've got a copy of Conceptual Mathematics: A First Introduction to Categories, and I'm more actively working through Benjamin Pierce's Software Foundations (which covers a range of material from the Functional Programming and Automated Program Verification slots). I've got a nice torrent full of textbooks for various branches of math, machine learning, and AI (extending from undergrad Intro to AI up through Hutter's Universal AI).

I can also firmly say that statistics class (which I'm taking officially this semester) is a bitch due to the sheer amount of extra symbology to learn and, of course, loads and loads of integrals.

  • Name: Eli

  • Race: Dwarf

  • Class: Spiral Warrior (also second-year research grad-student)

  • Level: two papers in the works, one under submission to a conference

  • Optimization Power: I haven't precisely counted, but my life is very improbable

I'm currently a second year Ph.D. student in mathematics (number theory).

My interpretation of your LW username just changed abruptly.

Non-mathematician here. Just curious... What are you alluding to? (“Adele_L” sounds like a given name plus an initial to me.)

It is in fact my name + initial.

But since I am a number theorist with this name, I know that gjm is thinking adele rings and L-functions.

Yup. (And of course I'd previously assumed, correctly as it turns out, that it was given name + initial.) What a nice coincidence!

The prospect of being formally in a study pair/group makes me anxious in case I'm a flake and feel like I've betrayed the other participant(s) by being akratic or being unable to keep up and then I will forever after be known as That Flake Who Couldn't Hack Model Theory That Everybody Should Laugh At etc. etc. I should probably work on that anxiety, but in the interim, as a more passive option, I've just created this Facebook group. Has the benefit that anybody who stumbles across it or this comment can join and dip in at their leisure.

I don't really know what to expect from the group and I'm fairly content at this point to let its direction be driven by whoever joins, but I would say that if you're unsure and hesitating whether to join or post a question or whatever, please Just Do It, rather than hovering, timing out, and giving up. Even if you're just curious or think you might want to join the group in future to comment but don't right now, feel free to join now and turn off notifications from the group to eliminate the Trivial Inconvenience for your future self.

Also, please do feel free to join if you're not actively studying FAI but want to help others!

I've heard of other LessWrongers using Google Hangouts for this and various personal projects, with all kinds of rules or variants fished from various GTD-related strategies.

Personally, I'd be interested in setting up an IRC chatroom, google hangout or other semi-persistent thinghy for pair/group-studying of... well, I'm not sure what yet. I'm looking at the courses in the MIT Challenge, reading through the first book of Feynman's Lectures (since it's finally available for free online here, which is the only place I've ever found it for less than 200$ /w shipping), other FAI/MIRI course, or whatever else ends up on top of the list after my current round of priority-realignment is over.


I second the vote for an IRC room. I spend enough time on IRC already that it would be convenient.

I would be very much interested in this!

Was this ever created?

Personally, I'd be interested in setting up an IRC chatroom, google hangout or other semi-persistent thinghy for pair/group

That would be great!

What's the current status of this? I'm looking to get started on the course list and would love a study partner.

From my experience doing group study for classes, there don't seem to be any major advantages or disadvantages for pairs vs small groups. The most relevant factor is how many eyeballs looking at something, but even that isn't a huge effect. Both are more effective than working alone (as the article concludes).

For a lot of things, getting together IRL looks like it would work best, but the logistics there can be difficult. For people who have Lesswrong meetups nearby, those are an obvious way to potentially coordinate meatspace study groups.

I'm interested in studying the latter chapters of Probabilistic Graphical Models, especially the parts on structure learning, learning Markov networks, and decision making.

I've got Categories for the Working Mathematician by Mac Lane; I will be going through this because I will be giving some talks on category theory to the math club here at my university. I pretty much don't have any logic and I want logic. I have Enderton's A Mathematical introduction to logic which is ok, though I think I want to find a new book. I also have Probability: The Logic of Science that I want to work through. I also want to go through MIRI papers. I am a math undergrad.

I would like to be a part of a study pair or a study group. There seems to be enough people that we can group together. I would like to learn from people, and teach people what I know (mostly pure math: category theory/abstract algebra/algebraic topology and basic calculus/real analysis).

I am not intending on exhaustively going through the MIRI course list, but there is significant crossover with my study plans. Other comments in this post have suggested some sort of forum / list / hangout around these endeavours. While I can't commit to studying alongside a partner, I would be enthusiastic about participating in such a group.

My copy of Discrete Mathematics and Its Applications by Kenneth Rosen (which is also the text used for the Discrete Maths course required for CS students at my university) is scheduled to arrive later today. I plan on doing an exhaustive read-through (including all or most of the exercises). I anticipate being able to spend 3 or more hours per week specifically on this text. If anyone wants to partner up, I'm game.

I am looking for a study partner who wants to Machine learning concepts with me and perhaps even work on some mini projects together. It should be someone who is serious about developing Data Science and ML engineering skills. At the moment, I'm going through Jason Brownlee's excellent ("advanced") primer on LSTMs as well as reading Kaggle threads on LSTM classifiers that ranked highly in the Kaggle competition. What I'm missing is fellow learners willing to hash through the material to discuss and clarify ideas. Want to join me? Feel free to contact me at: (john dot strong at etherpros dot com)(website:

I'm looking for a partner to read, study and do the exercises of the manual "Machine Learning in Action" by Peter Harrington in approximately 8 weeks.

I'm going through the PGM coursera class (It's one of the classes in the MIRI course-list). I'm definitely going to finish it because I'm doing it as an independent study at my University.

Message me if you'd like to join me. I have a few friends at school who read LW who said they'll probably join me. The more the merrier.

New to LessWrong?