Dan Luu’s blog, where he dismantles lazy cached answers that people often believe in, argues that a bunch of things are actually pretty good, argues that a bunch of things are actually pretty bad, measures some things to an extent I wouldn’t bother to do, and sometimes just goes in-depth about things he was interested in. Good technical content.
Nikhil Suresh’s blog, where a guy vehemently vents his frustrations with a mostly non-functional tech industry, grows to pursue positive change in a practical way, and encourages others to do the same.
Gwern Branwen’s site. He does a very thorough job of explaining what his site is about himself. I would say that independently of the content, the care that has gone into collating and organizing the content and into the UX/design is inspiring.
ACX, formerly SlateStarCodex, which I won’t bother to explain.
AI blogs:
Beren Millidge’s blog. Largely discussion of deeply technical and philosophical AI topics. I appreciate how well he can communicate a nuanced technical distinction and hit on the key constraints/tensions in a problem without being painfully verbose (something that I want to improve at).
Alex Irpan’s blog. Probably best known for his post about the fundamental problems in RL, there are other nice technical posts on here.
People I like to follow on LW: (for myriad reasons, including for technical AI thought, navel-gazing, miscellaneous new and interesting ideas/topics, or just the way they express ideas)
Leo Gao (particularly for shortform)
Daniel Kokotajlo (regardless of what you believe about AI 2027, he’s been an extremely good forecaster and has good models)
For books in book-format, here’s a link to my GoodReads. It’s not at all comprehensive, but it’s where I’ve cared to write about books I’ve read. I will single out a couple of books below for having a disproportionate impact on me:
Ender’s Game, which was the first work I read where the characters seemed to be thinking realistically and intelligently in a situation with significant moral stakes, and which contained such raw human and emotional elements. It probably helped that the characters were systematically-underestimated child geniuses. The followup book Speaker for the Dead was also important to my development as a person as an example of radical empathy and responsibility, and the less-popular Xenocide helped me with a religious struggle when I first read it; recently I’ve learned that quite a few people have had similar experiences, sometimes even with the same book. These books filled for me a role that I have heard HPMoR did for others I’ve talked to.
Inadequate Equilibria: I spent a lot of my life suffering under various delusions of market efficiency, but I never had a good conversation that could break down the concept for me; arguably I didn’t try to ask anyone with an adequately nuanced notion of what market efficiency was in the first place. A lot of the pieces were already in place for me, e.g, I knew about Nash equilibiria and could observe that the world seemed to be much worse than I’d expect in some places, but I didn’t connect these ideas and had no model for where to expect suboptimality or what mechanistic analysis I could apply to it. Sometimes, a painstakingly thorough elaboration of an idea from premises to conclusions is actually exactly what someone needs. (To this day I still systemically don’t seem to appreciate just how much low-hanging fruit there is for a human being to try to pick in improving the world. It’s something I’m working on.)
The Years of Lyndon Johnson, the magnum opus of the legendary biographer Robert Caro. Caro has a famously painstaking research process, and also in his memoir describes a thoughtful view towards a) telling the story of a society through the lens of an individual; b) the mechanisms of political power and their relevance to human life generally; c) the necessity of both grounding things in facts and also weaving a compelling narrative through the facts. This book has given me a ton to think about, e.g “life just 100 years ago, even 100 miles from where I grew up, could be brutally scarce”, “some aspects of politics have been the same for longer than I thought” (to repeat a point made below, “if this problem has always existed, maybe we will make it through despite it”), and a lot of questions about “when the ends justify the means” (e.g, LBJ did at least use his dirtily-earned power to push through some positive civil rights bills, but perhaps he’s better known for destroying the commons on the way to that power? and, who’s to say of those who did analogously underhanded things but successfully hid (by definition, we wouldn’t know about them)– might our welfare rest on their achievements?)
The Inner Game of Tennis, to be elaborated on later.
Other bodies of work that I appreciate:
I’ve revised my opinions about what YouTube channels are best for self-PT.
- I think that Venus Gabby has some of the most actually-effective exercises for lifters, in terms of time to payoff and bias to good propriceptive feedback.
- Garrett Kuljian has some good insight into a few specific issues. I’m not as sure about the excellence of his particular exercises.
- MSK Neurology has the most anatomically precise and pedantic guides for exercises, particularly regarding shoulder mechanics.
- I am now not that confident that Conor Harris is actually right about a bunch of stuff. I think his general vibe about the importance of breathing mechanics is right, and I think the focus on the pelvis and ribcage are generally good. However, I’m not that sure that the particular exercises he recommends are that good for making meaningful progress for a large segment of the population, and that conventional strengthening of weak links is often still very important.
Conor Harris’ YouTube channel is probably the best “self-PT yourself” channel on YouTube for people who obsessively systematize things / want to have the “perfect” exercise for stuff / have esoteric issues or severe versions of normal issues that don’t present in an obvious way. I think they are very information-dense and logically presented and recently their production value has increased so there are some uniquely helpful visuals in here. Importantly, I think he is basically right about the “true root causes” of a bunch of common complaints and also the most effective way of fixing them. Of course as with stuff that’s very information dense sometimes it’s boring and unmotivating to pay attention, but I consider these an excellent reference for when you want to inquire about a specific issue. Also there is a certain amount of clickbaity titling but that’s just the game these days, I guess. - I wonder if the Alexander Technique people are onto anything, but I haven’t really thought about it.
A bunch of Richard Hamming’s lectures can be found on this channel, not just the most famous one about first-class research linked below.
Other single pieces of work that I have a special appreciation for:
How much do you believe your results? on LW: At the time of reading this, despite taking courses in mathematical statistics in university, I had a sense that I had no idea what statistics was really about. Not that this managed to teach me all about that, but it illustrates an elementary, counterintuitive but important, thing in that field very instructively, and I imagined that deeply understanding 10-30 things like this might make me feel that I knew what statistics was about.
The classic advice on research and doing great work by Richard Hamming and this lesser-known talk
This relatively obscure comment on the SlateStarCodex subreddit. It gave me hope about some form of goodness in human nature, even in the midst of apparent mind-viruses. More generally, if this problem has always existed, maybe we will make it through despite it. (Conversely, it’s probably very hard to be rid of the problem, though…)