Links & What I've Been Reading Q3 2022
Links
Machine Learning/AI
1. Language Models Can Teach Themselves to Program Better
2. Ajeya Cotra update on AI timelines (shorter, of course).
3.The Library of Babel, stable diffusion edition. I love this bit from the Borges story:
When it was announced that the Library contained all books, the first reaction was unbounded joy. All men felt themselves the possessors of an intact and secret treasure. The universe was justified; the universe suddenly became congruent with the unlimited width and breadth of humankind's hope.
The certitude that everything has been written negates us or turns us into phantoms.
4. On how various plans miss the hard bits of the alignment challenge.
5. Understanding Conjecture: Notes from Connor Leahy interview
We think that in order for things to go well, there needs to be some sort of miracle. But miracles do happen. When Newton was thinking about stuff, what were the odds that motion on earth was governed by the same forces that governed motion in the stars? And what were the odds that all of this could be interpreted by humans? Then you see calculus and the laws of motion and you’re like “ah yes, that just makes sense.
6. Inverse scaling prize winners!
Forecasting
7. Five Questions for Michael Story: "Nearly all forecasters are paid more by their day jobs to do something other than forecasting. The market message is “don’t forecast”!"
8. On training experts to be forecasters. Lots of good points in this one, especially on the softer social aspects of forecasting.
Metascience
10. Status bias in peer review. Would be curious to see an attempt at estimating how much of this is actually justified. After all, research quality follows a power law, and past results are certainly indicative of future performance. Perhaps there is not enough status bias in peer review!
The Rest
11. All the cool kids are listening to The Lunar Society. José Luis Ricón says Dwarkesh "is probably the best podcaster there is right now". Tyler Cowen says "highly rated but still underrated!". The Stephen Hsu episode is my favorite, but do check out the other ones too.
12. From the robot: the map is of the territory. "I am affirming that you have write access to the realm of the Gods."
13. From the banana, on the efficacy of depression treatments and more.
14. Dysgenics by the Numbers. In my view probably overstates the rate of loss within societies a bit. But overall completely right. Probably doesn't matter though.
15. Scraping training data for your mind. “But Karl Ove”, Renberg says about his writing, “there is… nothing _there_”.
16. A Future History of Biomedical Progress
Progress in tools has created the potential for a radically different research ethos that will end biomedical stagnation. But to understand this new research ethos, we must first understand the telos of the mechanistic mind and why it is at odds with the biomedical problem setting.
17. Good interview with Vitalik. "The kinds of communities you get when low taxes are the primary reason to come are just really boring and lame".
18. Evaluating Longtermist Institutional Reform. Public choice, counterfactuals, long-range forecasting.
Audio-Visual
What I've Been Reading
Nostromo by Joseph Konrad. Tangled, fragmented, unclear, conflicting, and circular narratives/motivations/goals/priorities. A chopped-up story from various points of view, taking a look at a world filled with great characters surrounding the titular Nostromo. Politics, heroism, revolt, the worth of social status, reputations, perceptions, allegiances, and material vs idealistic interests. Betrayals of all kinds. Private and public vindications and redemptions. Great stuff.
The Glass Bead Game by Hermann Hesse. Fascinating Borgesian novel about a futuristic game that combines all arts and sciences into some sort of grand unified plaything. It's about music, duty, the lifecycle of organizations, transcendence, the life of the mind, and probably much more on top of that. Highly recommended.
Imperial Twilight: The Opium War and the End of China's Last Golden Age by Stephen R. Platt. Kind of weirdly structured, it mainly takes a look at the era from the point of view of various minor players, mostly traders, "supercargos", and so on. The big politics don't get much attention. Somewhat revisionist I guess? It's fine.
Break-Out from the Crystal Palace: The Anarcho-Psychological Critique; Stirner, Nietzsche, Dostoevsky by John Carroll. A fairly shallow exegesis, written in an indefensible style. Just go straight to the primary sources.
The Twilight World by Werner Herzog. Another Herzog book! Nowhere near as brilliant as Conquest of the Useless, unfortunately, but still not bad. Concerns of those Japanese soldiers who kept up the guerilla war for decades after the end of WW2, refusing to surrender and refusing to face reality. Very Herzogian with the jungle and everything. Some wonderful metaphors.
- City of Golden Shadow by Tad Williams. Gwern gave it 5 stars so I couldn't resist, but I didn't enjoy it at all. Absurdly overlong at 800 pages, it just ends with a cliffhanger (and there's more than one sequel). There's a series of parallel fantastical stories set in a virtual reality and they're all pointless and awful. A bit outdated in terms of how it imagines the internet, it does have a few interesting ideas but overall I don't think it's worth the effort.
Zero to One by Peter Thiel. I guess it's the best business book I've ever read. A bunch of concepts from it have penetrated the broader culture (definite vs indefinite optimism for example). It's a quick read so go for it
Philip Larkin: Poems selected by Martin Amis by Philip Larkin. His best poems are great, but be warned that they are also extraordinarily pathetic in a way that can really fuck up your mood (if not your soul).
HHhH by Laurent Binet. Split into 257 short chapters, it blends a straightfoward and minimally fictionalized retelling of Operation Anthropoid (the assassination of Reinhard Heydrich) with all sorts of metafictional elements, as Binet constantly comments on the issues with constructing a historical novel, compares his approach to other books and movies, and even brings his personal life into it. Irony, humor, self-consciousness (especially about the author's view of Heydrich), the tension between history and fiction, and a slow, horrific build-up that absolutely fills you with terror. Strange how powerful emotionally a book that is at the same time so detached can be. Quite good and very different.
The Man Who Loved Only Numbers: The Story of Paul Erdős and the Search for Mathematical Truth by Paul Hoffman. Fun and highly readable pop biography, I blasted through it in a day. "I doubt if he would have recognized my first name even though I worked with him for twenty years. The only person he called by his first name was Tom Trotter, whom he called Bill."
The Making of the Fittest: DNA and the Ultimate Forensic Record of Evolution by Sean B. Carroll. A curious artifact from a different era. Perfectly captures the zeitgeist of the peak of internet atheism and creationism debates. One could make statements about human evolution then which would be quite dangerous today. The stuff on DNA and evolution is pretty wide and not that deep. All of it has been covered better elsewhere. There's a chapter on EvoDevo for example, but it stays on the surface of things and I would recommend reading Endless Forms Most Beautiful (by the same author!) instead. On top of the evolution stuff you also have a random sprinkling of skeptic-related causes (dull and cringey rants about chiropractors), plus a very generic liberal environmentalism which basically ignores everything the author had written up to that point. Probably more interesting as a marker of a (short, but memorable) era than a book about DNA and evolution.
The Moon and Sixpence by Somerset W. Maugham. A fictionalized retelling of the life of Paul Gaugin, as a middle-aged English man abandons his family to go be a painter in Paris (and eventually Tahiti). I wasn't convinced by the central character, and there's nothing to this novel beyond him. The "egotistic, single-minded genius" trope has been done much better elsewhere, and the novel really strays very far from the actual life of Gaugin.
The Golden Bowl by Henry James. I made it about 50 pages in. Not for me.
The Beginning of Infinity: Explanations That Transform the World by David Deusch. An unstructured mishmash of warmed-over pop science and a cavalcade of bad arguments around abduction, philosophy of science, intelligence, infinity, qualia, etc. The arguments
about superhuman general intelligence not being possible because humans are universal Turing machines are utterly absurd and could be added verbatim to "On the Impossibility of Supersized Machines". One of the worst treatments of abduction in the history of philosophy, and that's really saying something. Deutsch's comments on heritability are downright idiotic, and it's clear that he didn't even bother spending 30 seconds reading the wikipedia page. He just makes stuff up (incorrectly). A lot of uppity commentary about shit he doesn't understand. And then it's just filled with a whole bunch of random shit, like a galaxybrained theory of why the UK has the best voting system, a terrible theory of aesthetics, etc.First Light by Geoffrey Wellum. Fairly conventional WW2 memoir from a British fighter pilot. Not bad, not great.
The Marsh Arabs by Wilfred Thesiger. A standard tale of a British explorer somehow making himself accepted and comfortable among the primitive natives (look up Thesiger's pics), and bemoaning the disappearance of their way of life. This particular one, among the pastoral tribes of the marshes of southern Iraq. Perhaps what makes it unique is that it is set not in the 19th century, but in the late 1950s. Comfy but unexceptional, ultimately the Madan are just not that interesting.