July 31, 2026

Books to Read While the Algae Grow in Your Fur, July 2026

Attention conservation notice: I have no taste, and only the memory of qualification to opine on thermodynamics. Also, most of my reading this month was done at odd hours and/or while chasing after a pre-schooler, so I am probably less reliable and more grumpy than usual.

S. R. de Groot and P. Mazur, Non-Equilibrium Thermodynamics
This is a very comprehensive treatise on the state of the subject as of the 1962. (As discussed below, that's fine). The theory considered is basically about small departures from the equilibrium state. The intensive variables / thermodynamic forces (temperature, pressure, chemical potential, electrical potential, etc.) are near their equilibrium values, and the departures from equilibrium lead to fluxes of the extensive variables (flows of heat, of volume displacement, of chemical reactions, of electrical current) which are proportional in magnitude to the distance from equilibrium, the constants of proportionality being "phenomenological coefficients", in symbols (eq. IV.1) \[ J_i = \sum_{k}{L_{ik} X_k} \] with the \( J \)'s being the fluxes (extensive variables per unit time per unit volume); the \( X \)'s being the thermodynamic forces (intensive variables minus their equilibrium values); and the \( L \)'s being the phenomenological coefficients. The entropy production (change in entropy per unit time per unit volume) is bi-linear in the fluxes and forces (eq. III.21 and ch. IV.1): \[ \sigma = \sum_{i}{J_i X_i} = \sum_{i,k}{L_{ik} X_i X_k} \] The second law of thermodynamics now amounts to \( \sigma \geq 0 \), forcing the matrix of phenomenological coefficients to be non-negative definite. Various symmetry arguments put further restrictions on the \( L \) matrix. The authors then work out an extremely comprehensive theory of the resulting dynamics (which can be non-linear for a variety of reasons), including Gaussian stochastic fluctuations, and apply it to a huge range of physical problems, of increasing complexity. Comparisons to experimental data are mostly verbal and/or confined to footnotes, but they are made.
The book mostly aims, successfully, to stay at the level of pure thermodynamics. It does, however, invoke statistical-mechanical considerations, about an underlying microscopic structure to matter, in two places. One, which is extremely important to the whole development, is the "Onsager reciprocity relations" (ch. IV.3 and VIII): if, say, a gradient of chemical potential induces a heat flux (which it may), then a temperature gradient will also induce a chemical-reaction flux, and the two phenomenological coefficients will be equal, \( L_{ik} = L_{ki} \). (I'm glossing over the complications induced [you should pardon the expression] by external magnetic fields, but the book handles all that in detail.) The equality of the coefficients actually follows from the time-reversibility of the microscopic physics, from which the thermodynamic quantities emerge. The other important place where statistical-mechanical considerations show up is in the chapter (IX) on the kinetic theory of gases, showing how, at least sometimes, the phenomenological coefficients can actually be theoretically predicted, and not just experimentally measured.
There are some parts of the sciences where I get the impression that they are just done, not because they are now boring or useless, but because everything important has been worked out and received a settled expression. Linear irreversible thermodynamics strikes me as almost this way: the generation or two from Onsager through Prigogine to Mazur said almost everything worth saying, within their domain. The biggest modification a really modern treatment of these phenomena would make would be to add symmetry breaking and order parameters --- so "really modern" dates to 1975. I realize that my saying this makes it (forgive the expression) thermodynamically inevitable that one reader will inform me of all the groundbreaking work done in linear irreversible thermodynamics in recent years, while another reader explains how the theory is entirely refuted. In the meanwhile, I very much enjoyed revisiting thinking seriously about physics. §
(I browsed this in grad school, and sold my copy during one of my moves. Years later, I got stuck on a project until I came up with a trick I was briefly very proud of; then I realized that "my" trick was actually something I'd read about here, the "progress variables" from chapter X on chemical reactions. On the occasion of actually submitting the blessed paper, about which I hope to say more soon, I read / re-read this whole book from scratch, and I am glad I did, though I will confess I skipped the exercises. One thing I had forgotten since the 1990s was that the authors use a system of notation for matrices and tensors I have literally never seen anywhere else. --- I bought my current copy in 2023 directly from Dover, which at the time threw in an electronic copy for no extra cost. [The e-book is an EPUB with decent but not perfect OCR and no hyperlinks.] I can't tell from their website if they are still doing that.)
Emily Croy Barker, The Thinking Woman's Guide to Real Magic
Mind candy fantasy: in which an early-21st-century American young woman, whose English-lit thesis and love life are going equally poorly, does a good deed; gets rewarded with three wishes, without realizing it; and, in consequence, wanders off, first into Faery and then into a Fantasyland which is convincingly poor, sexist and dirty, but also magical. Drama ensues. §
Ilona Andrews, This Kingdom Will Not Kill Me
Mind candy fantasy: in which an early-21st-century American young woman wakes up in the world of her favorite fantasy series, which is not Game of Thrones, but is also obviously inspired by Martin, and does her best to thrive using her prodigious knowledge of the books. Inevitably, this is something of a commentary on why anyone would want to read such books. §

Scientifiction and Fantastica; Physics

Posted at July 31, 2026 23:59 | permanent link

July 07, 2026

Algal Programming Note

Attention conservation notice: Petty details of a desultory instance of a moribund medium.

Because there are half-a-dozen actual research projects I should be working on, to say nothing of a grant report to write and fall classes to design, I decided to procrastinate, i.e., to go to my drafts folder and dust off some old pieces of book-chat:

While I was at it, I switched the buy-this-book links from Powell's to Bookshop.org. Powell's is still my favorite (surviving) bookstore in the world, but it no longer has an affiliate program, and a lot of the old links were just 404'd. Links for books with a 13-digit ISBN should be OK now, but I am sure there are still some links which don't resolve; since they didn't work before, either, you have nothing to complain about.

Tomorrow I may feel up to rolling 1d6 to decide what to edit.

Update, 10 July: My urge to procrastinate was strong.

Self-Centered; Books to Read While the Algae Grow in Your Fur

Posted at July 07, 2026 16:15 | permanent link

June 30, 2026

Books to Read While the Algae Grow in Your Fur, June 2026

Attention conservation notice: I have no taste, and no qualifications to opine on socialist theory.

"William B. Fuckley", Glasslands
Mind candy: a thriller in the southwestern desert, and the Republic of California, some decades after the American collapse. The author does a good job of suggesting a plausible-sounding future, without tedious exposition (which is too often what is meant by "world-building"). This is clearly an amateur effort (there was a whole repeated paragraph), but as such it's very impressive and I am happy to have paid money for it. §
Kim Stanley Robinson, Forty Signs of Rain
Mind candy, at once near-future science fiction, scientist fiction, and climate fiction. I loved this, but partly this was the pleasant shock of recognition: much of the action revolves around the National Science Foundation and, physically, its old headquarters in Arlington, which Robinson captures extremely well (even the atmosphere of review panels). Two of the main characters are a federal scientist and her policy-wonk husband living in Bethesda, and having grown up there with such parents, he nails that too. (The only other work of art which has given me such a feeling of finally getting some cultural representation for my people is The Americans.) Of course, all of this has been deliberately wrecked by barbarians, so these details now carry an air of melancholy they didn't have when written. As for the idea of NSF scientists deliberately leading a charge to do something useful about climate change, words fail me. §
This is the first of a trilogy; I will be devouring the rest.
Moira J. Moore, Resenting the Hero
Mind candy. Comparing this to The Fifth Season, it is striking how different two fantasy novels about worlds over-supplied with natural disasters, and the people with the psychic gifts to stop those disasters with their minds, can be in tone and execution. This is immensely fluffier, but it has no pretensions to be anything except candy, and succeeds as such. §
(My off-line reading log says I read some later books in this series in 2010 while flying long-distance, but I retain no memory of that and [because] I didn't blog about them.)
Mike Beggs, Ben Burgis, and Bhaskar Sunkara, The Blueprint: How Socialism Can Work in the Real World
This is the best --- the most accessible and the most forceful --- case for market socialism since Alec Nove in the 1980s. I will not rehearse my own arguments for either socialism, or for market socialism over planning. (This book argues both points, in some detail. While not making a big deal of it, they are quite clear that Marxist political economy is no help at all at designing a functioning socialist economy.) Instead, I will take it for granted both that some form of socialism is desirable, and that market socialism is the only one which has any chance of working well.
The main move here is to replace private ownership of the means of production with representative democracy within each firm. People and firms still buy and sell at whatever prices they can negotiate, firms can go bankrupt, etc., but firms are owned by their workers, so workers get both a fixed wage and a fluctuating profit share. (Though they do not emphasize it, the authors are aware that this way of re-distributing surplus value allocating profit means workers would have less diversification in their income than they would under something like Roemer's stock-market socialism.) Some workers have more decision-making authority within the firm; all workers get to vote on who those people are. Employee-managed firms have, empirically and theoretically, a reluctance to hire new workers, but the authors have thought carefully about how to ameliorate this.
There is a system of competing publicly-owned banks, instead of a stock market and the rest of the current finance system, plus the central bank handling monetary policy and the payments system. There are some suggestions here about replacing simple loans from the public banks with more risk-sharing arrangements. I don't think this amounts to taking an equity position by another name, but I'd want someone better at finance than I to go over that. In any event, some (much?) of the profits of the public banks get passed back to the state budget. I think the authors would probably regard allocating part of the income stream of the public banks to a citizen's dividend as a friendly amendment (cf.).
There are some places where I have my hesitations. Three of these have to do with the boundaries of the firms, with the banks, and with working for civil society organizations.
  1. With workplace democracy, just which activities are carried on within a firm, and which by market transactions between firms, may become much more contentious than they are now. (I have long had a thing about this.) The authors are aware of this, but I think they might be under-rating it. Let me make up an example: there will, presumably, be chemical engineering design firms in the socialist commonwealth. (We may presume the actual chemical plants will be heavily regulated for their environmental impacts, if not mandatorily practicing green chemistry, that appropriate pollution charges will be applied, etc.) Those firms will need janitorial, book-keeping, etc., services which are not themselves chemical engineering. It is not at all clear whether the janitors, book-keepers, and the rest should be voting members of the firm, or whether instead the engineering firm should hire the services of janitorial, book-keeping, etc., firms, which will themselves be workers' collectives. On the one hand, provided everyone has safe, dignified working conditions, a democratic say in how their job is done, and enough pay to live a decent life, there doesn't seem to be much of an ethical issue about whether the janitors are part of the chemical firm or of their own firm. On the other hand, even in a socialist commonwealth, there are likely to be differences in profit rates between chemical engineering and janitorial services! (The human and material costs to entering the janitorial-services market are probably much lower than those of entering the chemical-engineering market, hence more competition and thinner margins in the former than the latter.) Of course in our world such boundary questions are set by a mixture of convenience to stock-holders, regulatory arbitrage, and transaction costs, which may be very bad for many people, but the socialist commonwealth will need some way to fix these boundaries, too.
  2. The public banks play a very big role in this scheme. They finance the formation and expansion of worker-owned enterprises. In fact, they are suppose to catalyze the formation of new enterprises. The authors realize all this means that the work of the bankers will need to be rewarded, but they are (understandably) not very specific about how, exactly, to do that. It's only too easy to imagine the work being done poorly ("they pretend to pay us, and we pretend to invest"), or the bankers in fact becoming a very rich and influential class, or indeed both at once.
  3. Turning to civil society: it's an important part of the scheme that all enterprises be worker-run collectives. (They allow for a small number of freelancers/sole-proprietors.) But there is a bit of a tension here with having a vibrant civil society. A bunch of citizens can get together as a society for the prevention of cruelty to animals (or androids, same difference), and volunteer their own time and resources: not a problem. But when the SPCA gets big enough that its administrative needs strain volunteerism, then what? It would seem that it can't just hire some clerical help, because those would be employees and not worker-owners. The best I can come up with is that the staff of the SPCA would have to form a collective firm with one client, viz., the SPCA. (Of course that firm might grow and come to serve multiple civil-society groups, which gets us back to some of the stuff about the boundaries of organizations.)
I want to make it clear that I do not regard any of these issues as fatal flaws.
The authors are extremely experienced at writing for popular audiences, and this shows (*). They have clearly read, and wrestled with, a lot of the academic literature, but this is not at all an academic book, and is a lot stronger for it. (Roemer or Stiglitz, let alone Kornai, this is not.) They do a good job of positively arguing for their vision, and of anticipating and countering objections. The one set of potential objections they are categorically silent on is "Well, how do we get there?", i.e., how we might transition from actually-existing capitalism to their form of market socialism. But I think it is enough that they have described, thoughtfully and accessibly, an attractive future which we might want to get to. §
Disclaimers: One of the authors sent me an advance copy; I was very impressed and ended up writing a blurb. (We'll see if it gets used.)
*: Burgis and Sunkara, in particular, are skilled participants in the online attention economy, whose voluminous ephemeral writings are marked by the usual adaptive vices of that world. All this is, mercifully, not present is this book. ^

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; The Progressive Forces; The Dismal Science

Posted at June 30, 2026 23:59 | permanent link

June 18, 2026

Data Over Space and Time

Collecting posts related to this course (36-3467/36-667).

Posted at June 18, 2026 22:42 | permanent link

June 03, 2026

Family Amusements

I realize my book-blogging has, like all my blogging of any kind, fallen off drastically. Fortunately, my younger brother has decided to take up book-blogging, so I will, in the meanwhile, refer you to the more suave and sociable Shalizi brother. He is also, as our mother once put it to us, "a real scientist", i.e., an experimental biologist like her, and not a mere theoretician. This shows, to the reader's great advantage.

Kith and Kin

Posted at June 03, 2026 10:15 | permanent link

May 31, 2026

Books to Read While the Algae Grow in Your Fur, May 2026

Attention conservation notice: I have no taste, and no qualifications to opine on history or poetry.

Madeleine Robins, The Doxies Penalty
Mind candy: continuing Robins's series of historical mysteries set in a very-slightly-alternate Regency London. (I re-read the last one, Sleeping Partner, before turning to this.) Robins continues to be great and I hope that many more of these will flow from her pen. (She has a fine eye for how to offer her protagonists poisoned gifts.) §
Dan Abnett, Eisenhorn
Mind candy, calling for some comment.
I vaguely remember Warhammer 40,000 as a role-playing game from around 1990, i.e., when I was in high school. (This would have been the first edition.) I can't remember if my group ever actually got around to playing it *. I am also aware that it has become something of a nerd-cultural phenomenon, but have genuinely not kept up with it. Now, I am the kind of nerd who has a certain disdain for novels written in licensed intellectual properties, but I readily acknowledge that it can, in fact, be great. (John M. Ford's Star Trek books, for instance.)
When this collection fell across my path cheap, I picked it up out of curiosity. It's pretty decently adequate, as mind candy. There are places where Abnett shows flashes of real talent. (To avoid spoilers, I'll just say: The first alien planet; the opening scenes with the cranky biologist.) There are other aspects of the novels which are pretty unsatisfying. (I don't feel mean enough, today, to elaborate.) It'd be incomprehensible without some background familiarity with the setting. But I don't see anything special. My curiosity being satisfied, I am pretty unlikely to peruse any other fiction from this brand, unless someone reliable points to a particular instance as something extraordinary. (This is not an invitation to send me pointers.)
--- The question arises of why this particular brand/game/setting has become so popular; as this will let me take some hobby-horses out for a spin, I will pursue it. Why are there endless licensed books, to say nothing of fan-produced material and casual allusions, for Warhammer rather than any other role-playing game from around 1990? One possibility: if you are going to actually play a game, you need to coordinate with other players on which game. (Also, more notoriously, on a time and place...) This creates a role for focal points, and focal points are subject to increasing returns. Therefore, we should expect some games to be vastly more popular than others, even if all games were equally good, spoke equally to our moment in history, etc. This argument is somewhat undercut by the fact that some people are perverse enough to buy and read rule-books for RPGs they don't actually play. (I am one of them.) It also doesn't explain the tie-in material. But even when there are no focal points, we know that cultural trends can be self-reinforcing (Salganik, Dodds and Watts 2006; Salganik and Watts 2008). The important ingredient for such phenomena is that your odds of encountering artifacts from the trend must go up, the more of your peers are participating in the trend. Social media, and before that search engines and recommendation systems, deliver this in spades, but even a best-seller list will do (that's essentially what Salganik et al. used). I decline to seek an adaptive explanation for Warhammer's success, or to interpret it as meaningful symptom that Really Says Something About Our Society, unless someone shows that it's a lot more successful, or differently successful, than we'd expect from the nigh-inevitable tendency of cultural interaction to amplify, and so lock in, random accidents. (In other words: there are very high-amplitude branches of the wave-function where all the memes are about Traveller, or Vampire the Masquerade.) This concludes today's edition of A Series of Footnotes to Herbert A. Simon. §
*: We mostly did D&D and Call of Cthulhu, but I definitely remember Stormbringer, Beyond the Supernatural, Traveller and Gamma World. ^
Marie Boas, The Scientific Renaissance, 1450--1630
An old (1962) but still interesting history of the earlier part of the scientific revolution. In every field, Boas traces a basically-humanist impulse to discard medieval ideas and accomplishments, in favor of reviving antiquity, the more antique the better, which only gradually turned towards striving to go beyond Greco-Roman science to something actually new and unprecedented. She offers sympathetic (but unvarnished) portraits of the greats (Harvey, Gilbert, Vesalius, Tycho, Kepler, Galileo, Bacon, etc., etc.), along with attentive readings of their major works, interspersed with sketches of the broader environment and trends in learning and its application. It's not as deep, conceptually, as something like Wootton, but it's readable and humane, and covers most of what's important. §
Joseph Brodsky, Selected Poems, 1968--1996
I am a Philistine, because I thought this was just OK. There were some good poems, quite a few which I found boring, and absolutely nothing that has stuck in my memory even a few weeks later. I realize that the contemporary poets I enjoy have, in fact, a very different judgment of Brodsky, but I did warn you that I have no taste. §

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; The Commonwealth of Letters; Writing for Antiquity; The Great Transformation; Tales of Our Ancestors; Portraits of Detection, Pleasures of Crime

Posted at May 31, 2026 23:59 | permanent link

April 30, 2026

Books to Read While the Algae Grow in Your Fur, April 2026

Attention conservation notice: I have no taste, and no qualifications to opine on poetry or international political economy.

Henry Farrell and Abraham Newman, Underground Empire: How America Weaponized the World Economy
Regular readers (if I have any left) know that Henry is a long-time co-author and, even more, a friend; it'd be absurd to pretend I could write an objective review of one of his books.
With that disclaimer out of the way, this is a great book, which simultaneously illuminates some very, very important phenomena, and is very well-written. The phenomena are all about the ways in which globalization has turned out to create a series of "chokepoints", central nodes in the various networks of finance and communication, and to have placed those chokepoints under the control of the United States, without anyone really intending or even expecting this. Over-simplifying greatly: because the US was already the economic and technological leader of the world (and had been for some time), it made sense --- it was cheaper or more efficient --- to route connections through America, often through very specific parts of America. (Such as office parks in suburban northern Virginia.) Having created these connections, and the very material infrastructure which implements them, it then made more sense for further connections to route through those American nodes, rather than going through the expense of creating parallel, and perhaps little-used, routes. (There is a reason I finally posted about search and increasing returns while finishing this book.) Equally accidentally, and largely as a consequence of the War on Terror, the US government woke up to the fact that it could control those chokepoints, i.e., deny adversaries access to those networks; what was almost as good, it could very credibly threaten to do so. Hence "weaponized interdependence". Having become conscious of these powers (not least because of Henry and Newman's own article), the US government has proceeded to use them with abandon, provoking nigh-inevitable responses from other countries and other actors.
This is both a sophisticated work of social science (I can see where specific works on network theory or path dependence have shaped their thinking, and I daresay there are non-complex-systems influences I miss), and one which conveys a lot of detail about specific institutions and technologies. It is nonetheless not ponderously learned or monographic, but rather readable, even exciting, and something which can, and should, be read very widely. (Henry has a nice blog post about they ways they deliberately learned from thrillers and science fiction about how to narrate complicated systems.) It is, to repeat, a great book, and I am proud to call one of its authors a friend.
(Long ago, I made myself a promise that I would never recommend or review a book unless I finished it cover-to-cover. So it is that I only this month got to reading the last chapter, about how these powers could be restrained and/or turned towards good ends, such as combating climate change. This is painful reading in light of everything that has happened since the book came out in 2023, but there might still be some value to it, once the Republic emerges from its Babylonian captivity.) §
Jen Williams, The Bitter Twins
Mind candy fantasy: sequel to The Ninth Rain (in my backlog; suffice to say: good), and continuing the high standard of that book. I will say there was a bit in the middle where I set it down for a couple months, before I made myself press through it, but that was because I could tell that horrible things were about to happen to all of the separated bands of heroes, and I cared about the characters too much. (Horrible things did happen, but it was worth it.) I look forward to reading the sequel. §
Ludovico Ariosto (trans. David R. Slavitt), Orlando Furioso: A New Verse Translation [doi:10.4159/9780674053519]
At the border of "mind candy" and "classics of the Western canon" (hence the DOI). I shall emphasize the former. This is a big fat (600+ pp.) epic fantasy which follows an over-lapping set of characters, switching between viewpoints, as they go on quests, meet in single combat and in vast battles, fall in and out of love, perform daring rescues and even more daring feats of skull-duggery, listen to and enact prophecies, retrieve magic artifacts and tame or fight mythic beasts, go mad (hence the title), etc., etc. (It's even a sequel to a popular series, but continued by another author.) It is, in short, very much a recognizable genre work, both in content and form. It's also an epic poem, composed and translated in verse, and very, very much a product of a Renaissance world-view, projected back to the time of Charlemagne.
My grasp of Italian is that of a pre-kindergartener. (This despite the best efforts of my mother, and of Signora Anna and the other teachers at Silver Spring Bilingual Montessori.) I can have no opinion about whether the translation is accurate, but it is pretty good English verse. §

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; Commit a Social Science; Networks; The Continuing Crises; The Dismal Science; The Beloved Republic; Kith and Kin; The Commonwealth of Letters

Posted at April 30, 2026 23:59 | permanent link

March 31, 2026

Books to Read While the Algae Grow in Your Fur, March 2026

Attention conservation notice: I have no taste.

Rachel Neumeier, This Hour, Our Vigil
Mind candy fantasy; A sequel to the Death's Lady books (in my backlog), where good things happen to our favorite characters, without any special drama. If I say it reads a bit like fanfic for Neumeier's own work, that sounds like I'm damning it, but I actually enjoyed it, because, I wanted good things to happen for those characters, who'd had drama enough and to spare. Utterly pointless if you have not read the earlier books. §
Doris Piserchia, A Billion Days of Earth
Mind candy science fiction, from 1976. This is an especially weird member of the Dying Earth genre. It's a billion days (= a few million years) in the future, and the descendants of human beings are known as "gods", while evolved (possibly uplifted) rats think of themselves as Homo sapiens. One of those wakes Something which I think is an old psycho-bio-weapon that preys on unconscious desires and assimilates its victims into further bits of its silvery, amorphous self. Apocalypse ensues, as viewed by an unusually level-headed rat-man, among others. It's not pleasant, but it is skillful, and strange, and makes me interested in Piserchia's other books. §
(I picked this up on 26 July 2001, in a bookstore in Flagstaff which no longer exists...)

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica

Posted at March 31, 2026 23:59 | permanent link

March 20, 2026

"Aware of All Internet Traditions: Generative AI as Information Retrieval and Synthesis" (Verbatim Remarks at the Cultural AI Workshop)

Attention conservation notice: Deep Thoughts about Generative AI and intellectual tradition, from a statistician who has not made a contribution to either area of scholarship.

The talk I gave last week at the NYU Cultural AI workshop was an expansion of some remarks from a blog post a year ago, and some stuff I said in e-mail to Henry Farrell. (Many thanks to Leif Wetherby and Tyler Shoemaker for the invitation and organization.) I was thinking of turning the talk into a post of its own. But Ben Recht has just scooped me, in the most generous way possible. So, if you're interested, read Ben, and, for the real gluttons for punishment, my slides.

(I am meditating on Leif's suggestion that I am trying to revive structuralism.)

Manual trackback (of a sort): Notes from a Small Press; Brad DeLong

Self-Centered; Minds, Brains and Neurons; Automata and Calculating Machines; Enigmas of Chance; The Commonwealth of Letters; The Collective Use and Evolution of Concepts

Posted at March 20, 2026 11:00 | permanent link

March 06, 2026

Search and Increasing Returns, or, No One Makes You Push to Github

Attention conservation notice: An economistic argument that computer networks are nigh-doomed to effective centralization, by someone who is neither an economist nor a computer scientist. Arcane, speculative, and not actionable even if correct. You would be better off spending your time reading a book.
Drafted in 2018, and deliberately not much updated. Posted now because I found myself re-using the joke of the subtitle in an e-mail.

Twitter is awful for many reasons, but not the least of them is the way it makes its users feel forced to keep using it. (One such complaint among many.) Early visions of how it could contribute to a beneficient or at least harmless ecosystem, such Steven Berlin Johnson's (*), presumed that it would be much less sticky, a more old-fashioned website people could leave. Now, more recently Johnson has written a spectacularly wrong-headed piece hailing blockchain as a blow for re-decentralizing the Web.

Of course, the primary working example of a block chain is Bitcoin, which is heavily centralized in holdings, in processing power, and in exchanges. And there are economies-of-scale reasons to expect this would be true of anything using either proof of work or proof of stake. But suppose the problem with Bitcoin is just that it's not got a very compelling use:

imagine if keeping your car idling 24/7 produced solved Sudokus you could trade for heroin --- @Theophite, 16 August 2018
and that nobody has a use for blockchains, not really.

Git repositories, on the other hand, are the part of the blockchain idea that's actually good for something: a way of tracking changes, with many authors, where it is very, very hard to go back and alter history without being caught. And git repositories are things that are in-principle easy to copy and move around, and could be hosted on any machine running HTTP (or, heck, FTP). So why does Github exist, let alone hold the position it does?

The answer, I think, comes in two parts. The first is that focal points are subject to increasing returns via network effects. The second is that search engines create and/or amplify focal points. Put these two together, and you get a very strong tendency for any particular line of activity --- be it trading solved sudokus for heroin or open-source software development --- to concentrate in just one, or at most a few, online locations.

Incidentally: This doesn't necessarily lock in first-mover advantage, because other forces can overcome this (revulsion, technical superiority), but it does mean that it will be very hard to avoid having at most a few dominant locations at any one time. It also means that the transitions between dominant locations will be brief.

If we really wanted to re-decentralize, we'd have to (1) get rid of focal points, or (2) get rid of search engines as we've known them, or (3) somehow make search conduct users to a distributed, decentralized focal "non-point" (focal blob? focal rhizome?). I therefore strongly suspect we are not going to re-decentralize. But then I didn't expect the Web would centralize as much as it has, despite having literally learned my Brian Arthur and Paul David at my father's knee, so what do I know?

*: To be clear, I have been a fan of Johnson's books since Interface Culture and Emergence. I am picking on two of his essays, but that's because I think he is too ready to find encouraging signs for a certain vision of how the Internet could transform the world for the better. To be clear, I share the vision, but now hold out little hope for it.

(Tweakage, 2 June 2026: Small changes in the last paragraph, avoid re-using the word "point" so much.)

The Dismal Science; Networks; Linkage

Posted at March 06, 2026 22:50 | permanent link

Statistical Complexity of Link Prediction

Attention conservation notice: Link to a two-page fragment of a mathematical paper that was abandoned over two decades ago.
I had not one, but two mentors in graduate school who, when approached with an idea or a question, were apt to go to their filing cabinet and pull out notes, from years or even decades back, which addressed that very issue, or one very close to it, and which they had never gotten around to finishing. (G. and C. had never met, and otherwise had little in common.) As a juvenile scientist, I found this both humbling and intensely irritating: why didn't they publish? As a middle-aged professor with a big directory of partially-finished projects, I have more sympathy, but still want to do better myself. I am therefore going to try to make a point of posting a lot of my fragments, and just ask that if someone decides to build on one, they put me in the acknowledgments.

In 2004, because I was thinking a lot about the statistical complexity of predicting stochastic processes, and learning about networks from Mark, Cris and Aaron, I tried my hand at defining statistical complexity for link prediction. The resulting complexity measure seemed straightforward-in-principle but too hard to calculate for anything very interesting (except maybe exponential-family random graphs, where it ends up being the entropy of increments to the minimal sufficient statistics).

In the ensuing 22 years, I have done literally nothing with the idea, but something reminded me of it the other day, so here's the two-page fragment I abandoned in March 2004.

The one thing I would add to the fragment is to consider a stochastic block model, where each node $ i $ has a latent discrete random variable $ X_i $, IIDly across nodes; $ X_i $ says which "block" (or "community" or "module", etc.) node $ i $ lives in. The probability of an edge between nodes $ i $ and $ j $ is a function of $ X_i $ and $ X_j $ alone, independent of all other dyads or anything else. The pair $ (X_i, X_j) $ is thus the "state" which fixes the distribution of the dyad. Of course, as pair of latent variables, this is not a statistic, a function of the observable graph $ G $. But there are many circumstances where, as we see larger and larger graphs, we can infer all the $ X_i $ from $ G $, with the probability of making any errors tending to zero. (Ed McFowland and I tried to summarize those conditions in our paper, because we wanted to use them as tools for something else.) In these situations, the sufficient statistic for predicting $ G_{ij} $ from the rest of the graph will in fact tend towards $ (X_i, X_j) $ as $ n \rightarrow \infty $, and so the limiting statistical forecasting complexity will be at most $ 2 H[X_i] $ (since $ X_i $ and $ X_j $ are IID). I say "at most" because there could be situations where distinct pairs of blocks have the same edge probability; if that's ruled out, the asymptotic statistical complexity will indeed be twice the entropy of the block variable for one node.

The same argument could extend to any graphon, if the node variables can be asymptotically recovered from the observed graph.

Enigmas of Chance; Networks; Complexity

Posted at March 06, 2026 21:09 | permanent link

February 27, 2026

How Statistics and Machine Learning Came To Have Two Different Kinds of Kernel Methods

Attention conservation notice: Re-purposed teaching materials, about a confusing point of terminology in two arcane disciplines. This is amateur history of science, which will not help you learn or practice those disciplines, even if you wanted to (which you don't). Also, it was written for an advanced undergraduate class in one of those disciplines, and presumes some familiarity with the jargon and concepts.
I have written several versions of this over the years, for various classes; this one is from 18 February 2026, for a class which emphasized (Nadaraya-Watson) kernel smoothing, splines, and kernel density estimation. Posted now for lack of other material to refer to in the future, rather than re-writing yet again.

TL;DR: In statistics and machine learning, "kernel methods" refer to two different families of methods, one based on convolution, the other on hiding a basis expansion in the guise of a sum over data points. These use two different (but overlapping) sets of kernels. In both cases, the name "kernel" comes from some problems involving integrals in mathematical physics. (TL;DR of the TL;DR: Blame the mathematicians.)

"Kernel" is one of those terms which is used in many distinct-but-related senses across different areas of mathematics (like "normal"). The common metaphor is "the seed (in some sense) from which some larger object or structure grows (in some sense)". In particular, in physics, there are a lot of problems which involve integral operators \( \mathcal{I} \) that map one function, say \( f \), to a new function \( \mathcal{I} f \), by the relationship \[ (\mathcal{I}f)(x) = \int{f(z) K(x, z) dz} \] The inner function \( K(x,z) \) is called the kernel of the operator. (Or at least that's what it came to be called in English; a lot of this was first worked out in German, in the 1800s, and I don't know the original German technical term very early 1900s, and the German word was Kern. [See update below.])

A particularly important class of integral operators, both in physics and in a lot of other fields, take the form \[ \int{f(z) G(x-z) dz} \] That is, the kernel isn't a two-argument function, of \( x \) and \( z \), but another one-argument function, only involving the difference between \( x \) and \( z \), say \( u = x-z \). This came to be called (in English) the convolution of the two functions \( f \) and \( G \). (The German original was Faltung; as late as 1933, an American mathematician writing about this uses that term because "there is no good English word" (*). I am told one would ordinarily translate this as something like "folding".) It was recognized a long time ago that if \( G(u) \) is a probability density function (pdf), then \( \int{f(z) G(x-z) dz} \) gives a weighted average of all the values of \( f \), with the weight given to \( f(z) \) depending on how close \( x \) is to \( z \). This implies that \( \overline{f}(x) = \int{f(z) G(x-z) dz} \) is a new function, related to \( f(x) \), but smoother, because \( \overline{f} \) is averaging out the oscillations and extremes of \( f \).

In the 1950s, the leading statisticians were, overwhelmingly, trained as mathematicians, and a lot of that mathematics was stuff which had grown out of mathematical physics, so a great deal of this mathematics was very familiar to them. (They were creating the graduate programs which would train statisticians as statisticians.) In particular, one cluster of statisticians who were doing (what we'd now call) signal processing were very interested in estimating the power spectra of radio (and other) signals --- how much energy was transmitted at each frequency. (This turns out to be very important for prediction and control.) The raw power spectra were extremely noisy, and the statisticians realized that by convolving those raw spectra with Gaussian (or other) kernels, they could get something much smoother and more stable, in effect trading some bias for a very large variance reduction. It was then quickly realized that the same idea could be used to estimate probability densities, convolving the empirical distribution with smooth kernels. And a few years after that, in 1964, Nadaraya and Watson (independently) realized that you could use this to do regression.

So the "kernel" in "kernel density estimation" and "kernel smoothing" is kernel in the sense of function-you-convolve-the-data-with.

There is another set of methods, also called "kernel methods", which look rather different. Go back to how I defined integral operators: \[ (\mathcal{I}f)(x) = \int{f(z) K(x,z) dz} \] This is a linear transformation on functions: \( \mathcal{I}(af + bg) = a\mathcal{I}f + b \mathcal{I}g \), for any scalars \( a, b \) and functions \( f, g \). Just as multiplying by a matrix is a linear transformation on vectors, integral operators are (one kind of) linear transformation of functions. Just as matrices have eigenvectors, integral operators have eigenfunctions, where \[ \mathcal{I}\phi = \lambda \phi \] for some scalar \( \lambda \), the eigenvalue. (Eigen- is German again; roughly "self-" or "own-".) For some integral operators, i.e., for some kernels \( K(x,z) \), the eigenfunctions \( \phi_1, \phi_2, \ldots \) actually form a basis, meaning that, for any (well-behaved) function \( f \) \[ f(x) = \sum_{i=1}^{\infty}{c_i \phi_i(x)} \] (Alternately, we can always define a space of functions as "everything we can get by taking linear combinations of these eigenfunctions".)

Writing arbitrary functions as weighted sums of basis functions is a very old trick in math. Doing regression in terms of some set of basis functions is almost as old as a trick in statistics. So the eigenfunctions of (nice) integral operators can be the basis functions we use for regression.

That doesn't single them out from any other set of basis functions, but, again in the 1950s, people realized that kernels which satisfy some properties (like symmetry and non-negativity) can themselves be expressed in terms of the eigenfunctions and eigenvalues: \[ K(x,z) = \sum_{i=1}^{\infty}{\lambda_i \phi_i(x) \phi_i(z)} \] The usual (and correct!) explanation is to think of mapping \( x \) to the vector of function values ("features") \( (\phi_1(x), \phi_2(x) , \ldots ) \); doing the same thing to \( z \); and then taking the (weighted) inner product between those vectors. As we say: "\( K(x,z) \) is an inner product in feature space". Any linear method you can write in terms of inner products thus has a "kernelized" equivalent. (For example.)

This also implies that a weighted sum of kernel functions is equivalent to a weighted sum of eigenfunctions: \[ \sum_{i=1}^{n}{a_i K(x, x_i)} = \sum_{j=1}^{\infty}{b_j \phi_j(x)} \] (EXERCISE: Find an expression for \( b_j \).)

So doing a weighted sum of kernel functions (in this sense) is equivalent to doing an infinite weighted sum of these eigenfunctions (which form a basis). In fact, a lot of the time we can show that optimal function of the form \( \sum_{j=1}^{\infty}{b_j \phi_j(x)} \) can in fact be written in the form \( \sum_{i=1}^{n}{a_i K(x, x_i)} \), so that we really only have a finite-dimensional optimization problem. (Such a result is called a "representer theorem".) This was important when people, like the statistician Grace Wahba, worked out the mathematical details of smoothing splines. (One can actually write out the kernel, in this sense, that's implicit in spline smoothing, but I do not find it very illuminating.) (**)

What came to be called "kernel methods" or "kernel machines", in the 1990s, were predictive models of the form \( \sum_{i=1}^{n}{a_i K(x, x_i)} \), where, again, \( K \) is one of those two-argument kernels which lead to nice eigenfunctions and eigenvalues when plugged into an integral operator. The goal wasn't really smoothing (as it was with the convolution methods), but to get the power of using a huge --- even an infinite! --- set of basis functions, without having to explicitly estimate a huge set of coefficients, or evaluate a huge set of functions when making predictions. People talked about "the kernel trick" as this way of using two-argument kernels to implicitly use vast function spaces, without explicitly calculating the functions.

There are some kernels, in this sense, which also work as kernels, in the smoothing sense (e.g., Gaussians). But the two sets of kernels are distinct, and the two sets of methods are really distinct. If you do kernel smoothing with a Gaussian kernel, there is just no way to write that as a sum of Gaussian kernels with fixed weights. In general, kernel smoothing regression give us predictions of the form \[ S_G(x) = \sum_{i=1}^{n}{\frac{G(x-x_i)}{\sum_{j=1}^{n}{G(x-x_j)}} y_i} \] when we've seen data points \( (x_1, y_1), \ldots (x_n, y_n) \). In contrast, kernel regression in the implicit-function-expansion sense gives us predictions of the form \[ R_K(x) = \sum_{i=1}^{n}{\alpha_i K(x, x_i)} \] It is an easy EXERCISE to show that if \( K(x, x_i) = G(x-x_i) \), and \( G(u) \) is a (non-degenerate) pdf, then there is no set of weights \( \alpha_1, \ldots \alpha_n \) which will make \( S_G(x) = R_G(x) \) for all \( x \). It is a harder EXERCISE to show that there is no combination of weights and two-argument kernel \( K \) which will make \( S_G(x) = R_K(x) \) for all \( x \).

If, back in the 1950s, the one line of work had talked about "convolutional smoothing" and the other "implicit basis function expansions" (or something like that), we would not have this confusion.

Update, 6 March 2026: On the other hand, at least the convolutional smoothing people did not give their technique a misleadingly psychological name...

Updates, 11 March 2025:

  • Consulting the German original of Courant and Hilbert shows that the German term was Kern --- or at least, that's what was by 1924. (I should, of course, have thought to check this source before, but I didn't realize it was online!) If I can trust some German-English dictionaries, the primary meaning is something like "the edible part inside a hard-shelled nut or the skin of a fruit", which is very similar to the most basic meaning of the English "kernel". Turning to their citations, Courant and Hilbert refer to a fundamental 1903 publication on integral equations (and so on integral transforms) by Fredholm. This paper does not (so far as I can tell) ever introduce a name for what we'd call the kernel, it's «la fonction $ f(x,y) $» throughout. (Certainly «noyau», the modern French term, does not appear in Fredholm's paper.) Courant and Hilbert also cite Maxime Bôcher's 1909 Introduction to the Theory of Integral Equations (written in English), which flatly states (p. 13) "$ K $ is called the kernel of these equations", and adds, in a footnote, that the term was "first employed by Hilbert" in 1904 (in German). So "kernel" would seem to have been fixed as the English translation of the German Kern (in this context) some time between 1904 and 1909.
  • The oldest appearance of "kernel" in the Annals of Mathematical Statistics is indeed Parzen (1961). By 1967, people proposing other modes of density estimation refer to the convolutional method as the "'kernel' method" or "'kernel' technique" (with the scare quotes), indicating that the term was in use but still novel.

*: Norbert Wiener, The Fourier Integral and Certain of Its Applications (Cambridge, England: Cambridge University Press, 1933), p. 45. ^

**: The oldest example I have run across of using kernel methods (in this sense) in statistics is Emanuel Parzen's work (1960/1963, 1961) on time series analysis, where it was motivated, in part, as a way around having to estimate power spectra; despite Parzen's eminence in statistics, this approach does not seem to have been much used, at least not then. (I'm sure it's no coincidence that Parzen was Wahba's doctoral adviser!) What makes this extra curious is that Parzen also wrote a 1962 paper where he (more-or-less) introduced convolutional kernel density estimation (in which he cites prior work on estimating power spectra), and that did take off. Someone with enough knowledge of mathematics and statistics, and access to his archives, could write a history-of-science paper on what led Parzen, in particular, to introduce both kinds of kernel methods into statistics. This would fascinate, oh, easily a dozen people other than myself. ^

Enigmas of Chance; Mathematics

Posted at February 27, 2026 11:46 | permanent link

October 31, 2025

Books to Read While the Algae Grow in Your Fur, October 2025

Attention conservation notice: I have no taste, and no qualifications to opine on historical genetics and the transmission of inequality. Also, most of my reading this month was done at odd hours and/or while chasing after a preschooler, so I'm less reliable and more cranky than usual.

(Left almost-finished in 2025, because I got interrupted, and posted in 2026, because I wanted to procrastinate about half-a-dozen research projects.)

Martha Wells, Queen Demon
Excellent fantasy mind-candy, continuing the story from Witch King. As in that book from my back-log, the story is told both in the the "past" time-line, in which Our Heroes rebel against Mysterious Evil Overlords, and the "present" time-line, in which Our Much Aged Heroes Have to Deal with Idiots Trying to Undo Everything They Built with Such Sacrifice in Their Youth, and hints of returning Evil Overlords. (Analogy to the current situation of the once-and-future free world is probably intended by the author, and anyway irresistible). I do not think this book could really be enjoyed without the previous book, but with it, it's lots of skillfully-written fun, and leaves me eager for more. §
Carles Lalueza Fox, Inequality: A Genetic History [doi:10.7551/mitpress/14145.001.0001]
There is a basic point here which is correct and important, and can be made much more simply and starkly than Lalueza Fox puts it. Throughout most of the life of our species, a large fraction children born died very young, far before they had any chance to have children themselves. (In many times and places, most children.) But this death was not entirely randomly distributed: having richer and/or more powerful parents increased a child's odds of surviving and reproducing, through means like "not starving to death", "not being so malnourished as to be picked off by disease", and "not being killed in a fight with the next valley over". (This is true even though what passed for medicine was useless.) This implies that social inequality mattered for whose genes got passed down. (It was not the only factor that mattered, but it was one of them.) Social inequality thus left traces in our gene pool. Getting information about past inequalities from the shape of the present gene pool, and from ancient genomes we recover, means solving a tricky inverse problem, but that's something we're getting increasingly good at doing.
I think a great book could be written on all this. (Among other things, it should be catnip to anyone who wants to call themselves a historical materialist.) This book, however, is merely OK. It combines accessible accounts of scientific work, by Lalueza Fox and others, in historical genetics with uninspired book reports about recent semi-popular social science on inequality, in about a 2:1 mixture. There's nothing actively wrong with the reportage, not that I could tell, but it's neither sure-footed nor illuminating. (You'd think, to read this, that social science on inequality began with Piketty.) I will probably end up adding this to the reading list for the statistics of inequality class, if I ever get to teach it again, but with a bit of a sigh. Someone should write that great book; in the meanwhile, this will have to do. §
(Over the last two hundred years or so, as societies have gone through the singularity demographic transition, infant and childhood mortality has gone way down, and fertility has fallen dramatically. This all contributes to mute the impact of social inequality on the gene pool. If 1/3 of all children born die before age 20, which children are in that third can carry a lot of information about social position; matters are very different if the mortality by age 20 is instead below 1%.)

Books to Read While the Algae Grow in Your Fur; Scientifiction and Fantastica; The Natural Science of the Human Species; Writing for Antiquity; Commit a Social Science; Teaching: Statistics of Inequality and Discrimination

Posted at October 31, 2025 23:59 | permanent link

September 30, 2025

Books to Read While the Algae Grow in Your Fur, September 2025

Attention conservation notice: I have no taste; and while I do have some qualifications to opine on statistical mechanics and on machine learning, these books are ones I began in 1999 (Dotsenko) and 2000 (Husmeier), so the works themselves are unlikely to be of interest to you. Also, most of my reading this month was done at odd hours and/or while chasing after a pre-schooler, so I am probably less reliable and more grumpy than usual.

I will presume the reader understands the terms "spin glass" and "neural network".

(Left almost-finished in 2025, because I got interrupted, and posted in 2026, because I wanted to procrastinate about half-a-dozen research projects.)

Viktor Dotsenko, An Introduction to the Theory of Spin Glasses and Neural Networks (1994; doi:10.1142/2460)
This is a short (~150 pp), brisk first book on the topic for physicists. Most of the book --- the first ten chapters --- are good introduction to common models of spin glasses. On the mechanistic side, the emphasis is on frustration, and how it leads to slow relaxation through multiple local minima. On the theoretical side, the emphasis is overwhelmingly on the replica trick and replica symmetry breaking *. I appreciated the detailed discussion of experimental results, but I would have appreciated still more clarity about distinctions between quantities which are functionals of the ensemble / probability distribution, and those which are functions of the realized microscopic state **. Something which would have whizzed right past me in graduate school is the role of distributions over matrices which are symmetric under permutation of rows and columns, i.e., of exchangeable arrays; now I want to read about connections between spin glasses and graphons.
The final chapters, about neural networks, especially the Hopfield model (which is perhaps back in fashion). These are OK, but if you want to see the statistical mechanics of neural networks (of that vintage), you're really better off with either "The Statistical Mechanics of Learning a Rule" (Watkin, Rau and Biehl, 1993), or Engel and Van de Broeck (2001).
The first ten chapters are still a perfectly good introduction to spin glasses and replica symmetry breaking for those with a command of statistical mechanics at the level of a first graduate course (e.g., Landau and Lifshitz). §
*: Dotsenko has a whole later book about the replica method, but I haven't read it. ^
**: In a lot of statistical mechanics, we can be sloppy about this distinction, because considerations of ergodicity and concentration of measure imply there's not much difference (cf.). But spin glasses are different! ^
---Presumably (?) a different Viktor Dotsenko.
Dirk Husmeier, Neural Networks for Conditional Probability Estimation: Forecasting Beyond Point Predictions (1999, based on a 1998 Ph.D. thesis; doi:10.1007/978-1-4471-0847-4
The problem here is to make distributional forecasts of a (scalar) random variable \( Y \) given a (vector) random variable \( Z \), in the form of training a neural network to calculate the conditional cumulative distribution function \( \mathbb{P}\left( Y \leq y|Z = z \right) \), with \( y \) and \( z \) as inputs. The particular version of this of most concern here is when we have a dynamical system or stochastic process \( X_t \), and \( Z = (X_t, X_{t-1}, \ldots X_{t-k} \) and \( Y = X_{t+1} \), i.e., predictive distributions for time series, especially in a state-space reconstruction setting.
Back in the 1990s, when we were all very impressed by the universal approximation results for neural networks, lots of people did stuff with three-layer neural networks (i.e., one hidden layer), and one actually had to make a bit of argument if one wanted to do more than three layers. Husmeier uses four layers, because he's worried about the following problem: Suppose \( Y = f(Z) + \epsilon \), with independent noise \( \epsilon \), and we let the variance of \( \epsilon \) shrink towards zero. The nonlinearities in the neural network will let us represent a CDF, sure enough. but the nonlinearity gets applied to a linear transformation of the input variables. In that situation, it'd seem like the only way to approach a deterministic limit is if the function \( f \) is linear in the input \( Z \). I am not entirely sure this is right, but it's certainly true that the extra layer Husmeier allowed himself made it a lot easier to get good results.
Much of the book is about ringing changes on this basic arrangement, emphasizing, as I said, the prediction of dynamical systems plus noise, with good results. Several chapters (13--16) concern ensemble methods ("network committees"), with various quasi-Bayesian or cross-validated schemes for weighting the members of the ensemble.
Something which whizzed by me back in the day (if I read those parts at all) is that, starting in chapter 7, Husmeier assigns random parameters to the lowest layer(s) of the network and then leaves them fixed, only optimizing the upper layers. This "random vector functional link net" approach did not originate with him, but it does vastly speed up fitting, without any real degradation in predictive performance compared to full back-propagation (or other optimization). This is a precursor to, or even a form of, what we now call random feature methods, and I am going to mine his references (starting here) to figure out where it came from and what happened to it.
I should have paid a hell of a lot more attention to this book back when it came out. §

Books to Read While the Algae Grow in Your Fur; Enigmas of Chance; Physics

Posted at September 30, 2025 23:59 | permanent link

September 28, 2025

Unsolicited Opinions

Attention conservation notice: Unsupported assertions, accumulated over many years in my drafts folder, last updated in early 2022, now more or less painful in retrospect.
In no particular order, except where they are. I will not elaborate or explain.
  • "The Tyranny of Structurelessness" explains a hell of a lot about modern life.
  • If the problem is that powerful people are not accountable for their actions, network forms of organization are the very opposite of solutions.
  • Increasing returns $\Rightarrow$ monopolistic competition $\Rightarrow$ market failure explains a hell of a lot about modern life.
  • Endogamy systematically confounds genetic and cultural inheritance. It would be astonishing if personal-level variables which are entirely the product of social structure, custom, tradition, and strategic action weren't genetically predictable.
  • During the 20th century, and in much of the world even today, genetic variation in resistance to lead poisoning during brain development would, psychometrically, look like a heritable general intelligence.
  • Multiculturalists who expect different cultural groups to have different values and standards of excellence should not expect those groups to be equally represented in all occupations and professions, especially if people are free to enter and leave different lines of work.
  • There is a branch of the wave-function where Margaret Thatcher (trained research chemist and honorary FRS) led a successful push by the international community to address global warming through carbon pricing. (This probably require postponing the collapse of the Soviet bloc by at least a few years, so that her major speech to the UN about the urgency of tackling global warming didn't happen right before the Berlin Wall came down.) In that timeline, pollution charges are regarded as one of the signature policies of neoliberalism.
  • Modern civilization may be doomed because (i) we need to act collectively and globally on climate change, (ii) the USA is the indispensable nation at the center of the network, (iii) the Senate and Electoral College give a veto over American policy to a combination of plutocracy, reactionary folk Christianity and paranoia, and (iv) people like me are unwilling to move to Wyoming, or even West Virginia, to counter this. (I realize there is some tension between this and the previous statement.)
  • A true Burkean conservative in America should work for the preservation and development of the institutions bequeathed us by our wise and heroic ancestors, serving us for half a century, a century, or more: the welfare state; the regulatory agencies; international organizations and treaties; the military-industrial complex; academia; affirmative action; Hollywood and the rest of corporate popular culture.
  • The quantitative social sciences would be in much better shape if the first method everyone learned was $k$-nearest-neighbors, or maybe classification and regression trees, followed by the bootstrap. Linear models and $t$-tests should be, for social scientists, the hyper-mathematical arcana at the back of the textbook which their methods class skipped because there wasn't time.
  • "Prophesying upon the eigenvectors" is more transparently about making stuff up than "interpreting the coefficients", and therefore preferable.
  • There is something very odd about looking at the distribution of node degrees within a network. It makes the network itself into an odd hybrid of a single interdependent object (a relation) and a whole population. Going "something something ergodicity mumble mumble sampling reveals a process grumble grumble shut up and calculate" makes this only a little less weird to me.
  • A good philosophy of probability should explain why all applied Bayesian statistics relies on Monte Carlo, the most frequentist method imaginable. Ideally, it would also explain why Monte Carlo works with deterministic pseudo-random number generators (so the algorithmic information content of the pseudo-random numbers is extremely low).
  • I have toyed for years with the phrases "the simulation interpretation of probability" and "the birth of Trumpism out of the spirit of fandom", but I cannot work out what should stand behind them. (I hope they are unrelated.)
  • Complexity science is a series of footnotes to Norbert Wiener and Herbert Simon.
  • The mythopoetic impulse never goes away, but in populations which have gone through the Flynn/Gellner transition to widespread clerk-ship, it expresses itself not as stories, but as theories which aren't actually explanatory. (It gives us the Oedipus complex, rather than Oedipus.) Every form of progress has its costs.
  • Believing that there was a unique and irreversible transition from viewing time as an eternal cycle to viewing time as unidrectional history is taking sides.
  • His ideas are flawed by a vicious circle; you are engaged in iterative approximation; I stand firm at a fixed point.
  • Dentists are not, in fact, talking their book when they recommend pre-emptive action for wisdom teeth.
  • In a well-functioning social institution, incentives, information and (internalized) norms ensure that it doesn't matter who fills various roles in the institution, it behaves the same way regardless. Institutions are always impersonal. It follows that institutions are information-hiding abstractions, if we're thinking like programmers. But if we're thinking like physicists: institutions are collective degrees of freedom.
  • Blackouts are humanity confronting its own work, and the effects of its own collective actions, as a hostile and alien force.
  • The parable of the Nazi bar is a folk version of the Schelling model.
  • No one should be allowed to opine about artificial intelligence unless they've at least spent an hour or two with ELIZA and then stepped through the code.
  • "Philip K. Dick and the Fake Humans" doesn't yet explain a hell of a lot about modern life, but it will.

Posted at September 28, 2025 12:32 | permanent link

Three-Toed Sloth