Tony Onodi's Blog

Books I Read and Listened to in 2023

Here are all the books I read or listened to in 2023. In rough order of how good I thought they were and with some accompanying thoughts.

Harry Potter and the Methods of Rationality - Eliezer Yudkowsky (ebook)

When I first heard about this book, a roughly 2000 page discourse on rationality dressed up as Harry Potter fan fiction, I really really didn't think I'd ever want to read it. But I thought Yudkowsky was interesting, and came across enough gushing praise for HPMOR online that I caved, and downloaded the ebook, which is free because people other than J.K. Rowling are very much not allowed to sell Harry Potter books.

It hooked me pretty quickly, and then took over my life for a month, culminating in a final day, when I spent 8 hours reading. Yudkowsky is a fantastic writer, but don't take my word for it, read his alternative Council of Elrond, which, for no reason I can see, appears at the start of chapter 64, and doesn't contain any spoilers—or almost any relation at all—for the rest of the book.

I expected HPMOR to be far more of a textbook than it is. The genius of it is that it weaves into the story a rich model of the scientific method, and how to think about the world, but the story itself doesn't only stand on its own two feet, it shines. It has hyper competent, hyper rational characters, who are walking citation machines, but they're richly realised, believably human citation machines. And then, in the middle of all the fun is a chapter that moved me more than anything else I've ever read. If you've already read HPMOR and want to compare notes, or if you want to read it on its own you're really really sure you won't read the book—and have taken into account that 6 months ago, I was also really really sure I didn't want to read it—then I'm talking this chapter. But if you do read it as a standalone, I can't imagine it'll have nearly as much impact on you.

Mort - Terry Pratchett (ebook)

When I was a young teenager, I heard about Terry Pratchett books, and for some reason, probably partly because the author's name was Terry, decided they sounded lame, and put them in the same bin as Doctor Who. Unlike Doctor Who, Terry Pratchett got a reprieve. The final author's note of Harry Potter and the Methods of Rationality says any reader who wants more where that came from should read the Discworld novels, starting with Mort. So I did.

Of course this isn't more where that came from, the portal to that dimension has been closed for the safety of this dimension, but Mort still didn't disappoint. It's the only other thing I've read this year that's gripped me as hard as HPMOR. Thirteen year old Tony was wrong about a lot, but about Terry Pratchett especially. Go and read him!

Unsong - Scott Alexander (ebook)

This is the most original, and imaginative book I've read since There Is No Antimemetics Division, which was the most original, and imaginative book I'd read in a long time. Unfortunately, many parts made me feel—in a non-trivial way—like I was losing my mind. I think this was intentional. The mind-eroding parts are all Kabbalistic, and integral to the world of the story. And Kabbalah, and similar religious phenomena, are probably linked with schizotypal disorder. So the Kabbalah tracts have kind of succeeded. But I still didn't enjoy reading them.

As annoying as this was, I have to admit the kabbalistic tracts are impressive. They pull together interesting bits of trivia from every corner of human knowledge and feel a bit like Scott Alexander is showing off the hoard of arcana he's gathered from a lifetime of reading and producing words on the Internet. Personally, I think he's earned that right.

This part's spoilery, so read on at your own risk: one of the main characters in the book, Ana, is obsessed with theodicy, the study of why there's suffering in the world given that God exists (as He does, in the book). At the end of the book she gets to ask God's representative on earth why suffering exists in the world, and he tells her. I'm assuming Scott Alexander came up with the answer himself, that's no easy task, but in the context of the book world, and if you're willing to bite the bullet on a certain kind of utilitarianism, he kind of nails it. He's an atheist and he came up with the most plausible answer to this question I've ever heard. I'm also an atheist, so maybe it shouldn't be so surprising that it sounded plausible to me.

A Wizard of Earthsea - Ursual K. Le Guin (ebook)

I enjoyed this a lot. A Wizard of Earthsea is very fast paced, bordering on terse, I haven't come across that in a fantasy book before and I appreciated it.

Engines of Creation - K. Eric Drexler (paper)

In 1959 Richard Feynman gave a talk called There's Plenty of Room at the Bottom: An Invitation to Enter a New Field of Physics, in which he proposed nanotechnology, machinery that was small enough to manipulate matter one atom at a time. He suggested these could be built by making machinery that could build a scaled down replication of itself. That replication could then build a scaled down replication of the replication, and so on, until you arrive at the atomic scale.

Drexler picked these ideas up in the 80s, expanded on them, and eventually produced Engines of Creation (later followed by a more detailed book, which I haven't read yet), and that's kind of the history of nanotechnology to date. It remains an unrealised end state of the technology tree. Unlike Feynman, Drexler proposes a bottom-up approach, where we start with something like protein folding, and end with nanomachines. Though he's somewhat vague about how that would work in this book.

There's a lot of non-nanotech futurism in the book too, and it's really good. There's a chapter on "hypertext", which pretty accurately predicts the web, it starts with a metaphor that explains the web to a 1980s reader. That's not an easy thing to do, and I think Drexler does it well. Though now that everyone knows what it is, it's charmingly quaint.

The discussion of AI is also prescient. Drexler isn't a naive techno-optimist, and gives a better account of the dangers of human-level AI than most are able to, even now, as its arrival seems alarmingly close. Likewise, Drexler is the inventor of the term "gray goo", and his nanotechnology proposals include an ingenious miniature nanotech laboratory—designed to destroy its contents rather than allow them to leak into the world.

Engines of Creation spent less time, and gave less detail on nanotech than I'd expected, which was a little disappointing as it's what I read the book for. On the other hand, the rest of the subject matter was a pleasant surprise. Hopefully his following book, Nanosystems, will have what I came for.

Inadequate Equilibria - Eliezer Yudkowsky (ebook)

Scott Aaronson and Scott Alexander (Scott A(aronson|lexander), if you will) have both written better reviews of this book than I could hope to, so I'll aim for concision instead.

Inadequate Equilibria is an idea in book form. The idea is that, when deciding what to believe, you should sometimes ask questions at the meta level, and sometimes at the object level. If you want to know what an accurate valuation of a company's share price is, you ask a meta question: what does the stock market think? Because unless you work for a hedge fund, the market will almost certainly have a better idea than you. So there won't be any money left lying on the ground. For other questions, you're sometimes better off not deferring to authority, you might benefit from thinking about them at the object level instead. This can even be true if the question is quite important, lots of people would really benefit from knowing the correct answer, and the authority that's meant to know the answer seems very authoritative. This tends to happen when that authority isn't primarily motivated by finding the correct answer to the questions it's meant to be answering, but is instead motivated by something else, like seeming respectable, or not taking risks.

Inadequate Equilibria contains a lot of excrutiating examples of authorities doing obviously wrong and bad things at huge costs. It explains the reason this goes on even though the cost to society is so huge, and guides you through how to think about this sort of problem without turning into a deranged conspiracy theorist.

It's very good, and only 167 pages. I recommend reading it.

The Scout Mindset - Julia Galef (ebook)

If you want a quick summary of what The Scout Mindset is about, then I recommend reading How to change your mind by Sam Freedman. Even if you feel like you get the idea, I'd still really recommend The Scout Mindset. It has a lot of tips for how to help you notice when you're falling into the traps of the "soldier mindset", and how to overcome that tendancy when you do. It's also full of interesting stories of cognitive biases, and of people fighting the good fight and changing their minds, even when it was painful for them.

Valuable Humans in Transit and Other Stories - QNTM (paper)

Have you read Lena? No? OK, go and do that. Now go and read the rest of QNTM's back catalogue. Then read this book, which is mostly stories you've already read, now you've read all the online stories, plus three extras that are worth your while.

Play It As It Lays - Joan Didion (ebook)

Well written, intelligent, chique, I'd not read a Didion book before, and this one lived up to the hype.

Land is a Big Deal - Lars Doucet (ebook)

Lars Doucet won Scott Alexander's 2021 book review contest with his review of Progress and Poverty by Henry George. He then expanded that essay into a book, and this is that book. Poverty and Progress puts forward a theory about land, and land value taxes, that took off so successfully that it became its own "ism": Georgism. Georgism is sometimes posited as a third-way alternative to communism and capitalism. But, in most proposed forms, it just seems to me like capitalism with different taxes.

I don't mean this as a criticism though, there's lots to like about capitalism, and lots to dislike, and the theory of Georgism says that taxing land should mitigate a lot of the dislikeable parts. It's a really compelling theory, but it's never really been tried (just like communism and capitalism, according to the hardcore adherents of each).

On Writing Well - William Zinsser (ebook)

Maciej Cegłowski is in the top tier of writers I enjoy reading, so I read On Writing Well as soon as I saw him recommend it on twitter. It's good. It's about writing nonficiton only, and includes a lot of advice I already consciously think about. Probably because I'd picked it up from reading Politics and the English Language, and Paul Graham's various essays on writing.

But I think it might have helped to internalise some of those things I already knew. There's a section on how to cut out words, that includes a page of the original manuscript for On Writing Well, with edits, that really helps to illustrate the sort of brutality writers should strive for.

On the other hand, if you do read it, or read the Orwell and Graham essays I linked to earlier, I'd recommend reading Dan Luu's essay on writing too. He comes to his writing with different goals, and that means he writes in a different style. His goals might not be your goals, but you should read his essay just to check.

A City on Mars - Kelly Weinersmith and Zach Weinersmith (half paper, half ebook)

A City on Mars is a rallying cry against colonising space, it dumps all over the ideas of the space colonising nerds, but does so from a place of love. This couldn't target my world view more precisely. I love hypothesising about future space colonies and the technology they'd need, but ultimately I find them kind of pointless for now.

When I look into the future I don't see a gradual ramp towards a grungy space-faring future, I see an alarming ramp towards human-level AI followed by a future that's hard to predict, in which humans may or may not play any part whatsoever. The Weinersmiths advocate for a "wait and go big" approach to space, where we take advantage of the communications and cosmology that space offers, but keep our powder dry when it comes to off-world colonies. They make a pretty solid case that "wait and go big" is not only a desirable approach but, due to technical limitations, the only viable approach. I tend to agree with this, but suspect that we'll have to wait, but whether we're there when something goes big seems far from clear to me.

The New Science of Strong Materials - J.E. Gordon (paper)

This book came out in 1968, so its science isn't very new any more, but it is interesting, and—although I don't really know enough to say—almost certainly still relevant. It's also really funny. It's not funny in an XKCD kind of way (and this is not at all a slight on XKCD) where the author talks about real science, but goes out of their way to include gags. Its writing is just suffused with charm and wit in a way that's hard to convey.

Elon Musk - Ashlee Vance (audiobook)

I can think of at least one story in this, the firing of Musk's long-time assistant Mary Beth Brown—in fact, any mention of Mary Beth Brown at all—that makes this book worth reading, even in light of the new Isaacson Musk bio. There are probably more. But, because it came later, the new bio is mostly a superset of this one, so if you can only read one, go for the Isaacson one. I don't regret reading it though, and you probably won't either if you're interested in Musk.

Elon Musk - Walter Isaacson (audiobook)

There are a lot of incredibly weird anecdotes from Musk's life in this book, and it's worth reading just to gawp at these.

Gwern summarises what I think might be the most important thing to have come from this book: Musk is very likely bipolar. He gets highly agitated when things are ticking along nicely, and so throws himself into a fit of activity. When he says difficult things will be done to impossibly short timelines that everyone else knows are nonsense, or otherwise says things that are plainly untrue, he probably believes what he's saying. This is true of lots of people saying untrue things, because people are great at lying to themselves, but in Musk's case some of the mad stuff does seem to come to fruition, which is odd.

Very often though, these things blow up in his face in exactly the way you'd expect them to. But overall, he's so far come out ahead. I often read people say something to the effect of "you shouldn't bet against Musk", and historically this has been correct. But a constant theme throughout Isaacson's book is how he seems to be frying his brain with sleep deprivation and stress, and I'd guess this might catch up with him eventually. If you wanted evidence for this theory then you could easily point to his purchase and management of twitter, which is littered with unforced errors that Musk himself admits to.

Musk is often cruel, and hard to sympathise with. But I think SpaceX and Tesla are doing important things that are so far out of distribution that people think you're making it up if you try to tell them about it. It also seems that if Musk was hit by a self-driving car bus they might very quickly stop doing those things, so it really gives me no great pleasure to wonder whether he might be sliding into a decline that might make the people working there wish he had been hit by a bus. So, I guess, here's to more years of Musk doing well, and hopefully not ruining too many random bystanders' lives on the way.

Going Infinite - Michael Lewis (audiobook)

Sam Bankman-Fried can go toe-to-toe with Elon Musk for weirdness. His weirdness fills this book, and pushes out a lot of the more technical examination I'd usually expect from Michael Lewis, and in some ways I felt like I learnt more about the inner workings of Jane Street than of FTX. That shouldn't put you off though, the weirdness alone is enough.

Scott Aaronson's reflections on the whole saga are very worth reading if you don't want to read the book (or if you do) and if you don't want to read those then here's the highlight:

First, FTX was actually a profitable business till the end. It brought in hundreds of millions per year—meaning fees, not speculative investments—and could’ve continued doing so more-or-less indefinitely. That’s why even FTX’s executives were shocked when FTX became unable to honor customer withdrawals: FTX made plenty of money, so where the hell did it all go?

Second: we now have the answer to that mystery. John Ray, the grizzled CEO who managed FTX’s bankruptcy, has successfully recovered more than 90% of the customer funds that went missing in 2022! The recovery was complicated, enormously, by Ray’s refusal to accept help from former FTX executives, but ultimately the money was still there, stashed under the virtual equivalent of random sofa cushions.

Yes, the funds had been illegally stolen from FTX customer deposits—according to trial testimony, at SBF’s personal direction. Yes, the funds had then been invested in thousands of places—incredibly, with no one person or spreadsheet or anything really keeping track. Yes, in the crucial week, FTX was unable to locate the funds in time to cover customer withdrawals. But holy crap, the rockets’ red glare, the bombs bursting in air—the money was still there! Which means: if FTX had just had better accounting (!), the entire collapse might not have happened.


Where Is My Flying Car - J. Storrs Hall (ebook)

The title sounds like it's just a catchy hook, but it's not. Quite a bit of the start of the book is a history of flying cars, and discussion of where they failed. This is reasonably interesting, but not really what anyone should read WIMFC for, you should read it for batshit futurism, and indignant rants about regulation and modern attitudes to progress.

Flying cars are at the less batshit end of Hall's futurism. Further along the spectrum is the space pier, a 300km long rail gun elevated 100km above the earth's surface on towers of diamond—all built by nanomachines of course. There's also the weather machine, a blanket for earth—made (again) of nanomachines floating in earth's atmosphere—that gives us complete control of the earth's thermostat, while also being able to collect solar power, and beam it to—you've guessed it—flying cars.

But while we're working up to our weather machine we should, according to Hall, power our flying cars (and just about everything else) with nuclear power. I'm not sure he ever identifies himself as such in the book, but Hall appears to be a pretty devoted libertarian, and dreams of a future where one can build a home on a mountainside with no connection to the outside world. Just a radioisotope generator for heat and power, a spring, or condenser for water, and of course, a flying car for roadless access. I'm not much of a libertarian, and I'm pretty content to lead a life that's very dependent on the rest of the world, but hypothesising about how one could live independently has always been appealing to me, if only because I enjoy this kind of puzzle solving.

On the other hand, I'm nowhere near as excited about flying cars. There are some pretty interesting calculations in the book about how there's enough airspace above America for everyone to fly around about as much as they currently drive. I can buy that flying-car-dependence could end up being an improvement on good old fashioned car dependence, but I doubt anything other than human levitation, or teleportation can really beat transit-connected walkability and bikeability in dense urban areas, which are where most economic activity happens.

Towards the start of the book Hall endorses Tyler Cowen's Great Stagnation theory and produces his own chart that shows how technologies that need less energy (most notably computers) have seen the most progress since the 70s, while those that need more energy have been comparatively stagnant. This chart was a highlight for me, but then Hall uses it as a reason to lament the pro-stagnation culture the world has since adopted ad nauseam. I could write a pretty long review of all I disagreed with in this book, but that's a kind of compliment, I enjoyed reading it, and it made me think.

Down Under - Bill Bryson (audiobook)

I've enjoyed every Bill Bryson book I've read, it's hard not to, and Down Under was no exception. It also took my enthusiasm for the idea of visiting Australia from "not much" to "an alright amount".

The Dispossessed - Ursula K. Le Guin (audiobook)

An allegory that explores the contrast between communism and capitalism with all the subtlety of a bunch of students arguing about communism and capitalism in the student union. Still though, it's a good story, great writing, and I really recommend it.

Parfit: A Philosopher and His Mission to Save Morality - David Edmonds (audiobook)

Peter Singer is often called the father of the effective altruism movement, and—less often—Derek Parfit is named its grandfather. He was far less of a public intellectual than the philosophers associated with EA now, but Singer himself called Parfit the closest thing to a genius he'd ever met.

One of the anecdotes in this book that blew me away more than most wasn't about Parfit. It's the story of how Tyler Cowen, now the Internet's most online economist, read Reasons and Persons (Parfit's first book, and magnum opus), and was so infatuated by it that he got in touch with Parfit, and the two ended up co-authoring a paper that built on Parfit's work. Reasons and Persons is, I think, a graduate-level philosophy text. I made an attempt on the first chapter, and it didn't go well for me. It says a lot about Cowen that he was able read Reasons and Persons, and then make his own contributions to the field, despite it being quite different to his own.

Back to Parfit. He's a man who went to Oxford University for his undergrad, and then just stayed there for the rest of his seventy-four year life. Edmonds points out that he was quite literally cloistered for his entire adult life, yet his biography still ends up being interesting. It helps that moral philosophy is inherently interesting, and Parfit obviously made huge contributions to it. I ran headlong into some of his thought experiments, and there's at least one for which, months later, I couldn't propose anything vaguely resembling a consistent position.

Parfit also inhabited the role of eccentric intellectual perfectly. In his younger life this was very affable, and endearing. In his later life he took on an indifference to those close to him that seemed cruel at times, and it's kind of suggested that this might have been a result of his work on identity and morality. A person who is entirely rational about their morality would probably try to close the gap between how they feel about those close to them and how they feel about strangers. And so, it seems, Parfit did.

Snow Crash - Neal Stephenson (audiobook)

This is the book the word and concept of a "Metaverse" comes from. So you should read it if you want to be able to say you've read that book. But you should also read it because it's fun. The plot, and world, are over the top in a way that I hadn't expected, but it's good fun over all.

Solar Power Finance Without the Jargon - Jenny Chase (paper)

This is a good history of the financing that allowed solar power to roll so dramatically down a price curve. At times it meanders towards being a memoir of the author's career as a solar power analyst, which sounds dull, and probably would be were it not for the fact that it's kept quite brief, and does at least provide interesting context for the solar power revolution of the last thirty years.

Permutation City - Greg Eigan (audiobook)

The narration in this audiobook is bad. Except for the parts with non-native-English dialogue, which are excruciating. To add to my woes, this is one of the rare ficiton books whose subject is complex enough that the audiobook is hard to follow. I really should have given up on this one. But I didn't, I slogged through, and I regret it. I'll go back and read the book at some point, because it did seem like it was good. I still enjoyed it even. But I wasn't quite able to keep up, and I'm annoyed about it.

This is How You Lose the Time War - Amar El-Mohtar and Max Gladstone (ebook)

This book took a while to get going, but I warmed to it in the end. When I started reading it I thought I'd be reading a sci-fi book. And I was. But it's also listed as "#3 most gifted in Lesbian Romance" on Amazon, and I did not see that coming.

Feersum Endjinn - Iain M. Banks (ebook)

More than a quarter of this book is written from the point of view of a character who can only spell phonetically. If you search reddit for discussions of Feersum Endjinn, this is all anyone talks about. It's most of what I'm going to talk about. It was a terrible choice. The story is enjoyable enough, especially the ending, but it's very hard to forgive a book that makes you wade through so much mud.