Three-Toed Sloth

Books to Read While the Algae Grow in Your Fur, May 2023

Attention conservation notice: I have no taste, and no qualifications to opine on crime control policy. Also, most of my reading this month was done at odd hours and/or while bottle-feeding a baby, so I'm less reliable and more cranky than usual.

(Left incomplete in 2023, because I got interrupted, and finally posted in 2026, because I wanted to procrastinate about half-a-dozen research projects.)

Herbert A. Simon, The Sciences of the Artificial

I imprinted on this as a teenager. I revisited it this month for the sake of the shoggothim. Unlike some things I was very impressed by at that age (e.g., Joseph Campbell), it more than holds up. Even within AI proper, where one might most suspect a book first published in 1969 to be obsolete --- look, it's true that transformer-based large language models are not built along the lines Simon would have pursued, but they are very much Simon-ian "artifacts". (Max Raginsky has written out the argument so I don't have to.) And the chapters on human cognition, on social institutions, and on complexity remain brilliant and inspiring. §

Update, 2026: more and yet more stuff about Simon and shoggothim. Also, what people in 2026 are calling "agents" and "harnesses" look an awful lot like Good Old Fashioned AI / symbol systems, even "expert systems", which interact with the stochastic text generators. (Cf. Ben Recht on "agents".)

Thomas Abt, Bleeding Out: The Devastating Consequences of Urban Violence--and a Bold New Plan for Peace in the Streets

The central ideas here are:

1. America has a real problem with violence, especially but not exclusively gun violence, especially but far from exclusively in its cities;
2. this sort of crime, like most sorts of crime, is extremely concentrated both spatially and socially;
   (Also: the people who commit this sort of crime are also massively more likely than most people to be the victims of it in their own turn. In this violent crimes differ from, e.g., wire fraud.)
3. the extreme concentration implies that there are policies we could use to reduce the incidence and severity of such violence, without having to totally re-make American society;
4. many effective policies have already been demonstrated in the field, with large effects

I am entirely persuaded about (1) and (2). (To be clear, Abt does not present these as novel insights of his own.) (3) is at least plausible. (It is neutral on the question of whether thoroughly re-making our society would be a good thing; even if it is, it's going to take a while. and the problem exists right now.) The place where I start to get cold feet is step (4), because evaluating the effectiveness of policies is such a mess, and so many thing which work on a small scale, with resources, tacit knowledge and dedication, will be disappointing when generalized. But lots of the specific ideas Abt puts forward are ones which where it's at least hard to see how they could hurt. §

Jen Williams, The Silver Tide

I very much enjoyed Williams's two previous fantasy novels, but I guess they didn't sell well enough here, because this third book in the trilogy seems to have never had an American publisher and it took me a long time to track down a UK copy. Having done so, I re-read the first two volumes and then tore through this one. They're all mind candy, but of high quality, and this one even makes me swallow a time-travel plot without complaint. But it really does rely on having read the rest of the trilogy beforehand. §

Books to Read While the Algae Grow in Your Fur; Commit a Social Science; Automata and Mechanical Amusements; Minds, Brains, and Neurons; The Dismal Science; Philosophy; Complexity; Scientifiction and Fantastica; The Beloved Republic