Programmable Mutter: where it's been and where it's going.

With more to say about the former than the latter.

The beginning of a new year is a good time to think about what I’ve been up to with this newsletter, and where it goes next. It was supposed to be a completely idiosyncratic project, but has gathered far more readers than I ever expected. I write what I want to write, not what I think people want to read. Even so, a surprising number of people seem to want to read it.

I didn’t have much idea of what this newsletter would turn into when I began writing back in Summer 2023. I had a book coming out with Abe Newman, and my frank ambition was to promote it (which I still ought be doing - go buy!). That entailed writing about the financial, informational and production networks that weave the world economy together, and how they were being weaponized by the U.S. and its adversaries. I did end up writing about this - but nearly as much as I expected to.

But I was also beginning to get engaged in a new set of conversations about soi-disant “AI.” I’d somehow come across “AI Dungeon” - a very early application of GPT-2 - in early 2020, and was completely fascinated. It was clearly something new in the world. As people began to properly pay attention, I was talking with other interested scholars like Alison Gopnik, Cosma Shalizi, James Evans, Hugo Mercier and Marion Fourcade. So I started writing posts as a way of pulling these informal conversations together, and identifying common themes and differences across various disciplines and approaches. Some of this is turning into more traditional academic collaboration and writing - more on that as it develops.

Finally, I’ve a long standing interest in the relationship between democracy and technology. How do new technologies affect our notion of the public? How do they affect the ability of democratic and non-democratic governments to do their work? Why have many people in Silicon Valley moved away from democracy? And so on.

Not everything that I’ve written falls into one of these three categories. Quite a lot (including most of the writing that I’m happiest with) falls into two or more. The apparent complexity of this newsletter is generated by the unpredictable collisions between these varying individual obsessions.

If I had to guess, I suspect that I’ll be writing a little less about AI in the future (though there may be a short term bump of new activity in the next few months). I’ll be writing more about democracy and its problems for all the obvious reasons. And I’d like to be writing less about the weaponization of global networks, but I suspect that there is going to be so much crazy happening that it will be impossible to ignore.

The one thing I regret is that I don’t have much time to hang out in comments - writing this newsletter nearly every week, on top of a lot of other writing, speaking and teaching commitments, is as much of a stretch as I can manage right now. The good news is that it’s free, and barring something wildly unexpected, it is going to remain so.

So that is where I think this is going. And since I’ve acquired a lot more readers in the last year (base readership has quintupled), here are the pieces that I think are particularly worth reading if you haven’t read them already. NB that I have not strived to achieve a foolish consistency across all this writing: these pieces are slightly different versions of me thinking aloud, with some unfortunate repetition, rather than a sustained effort to create a dazzling coherent vision.

So if you want to keep on reading, keep on reading! And Happy New Year to all.

Weaponizing the world economy

My time in the Torment Nexus. Influencing policy isn’t what it’s cracked up to be.

If post-neoliberalism is in trouble, we're all in trouble. The world is not flat, and we need a new politics of the state.

"Small Yard, High Fence": These four words conceal a mess. Why the Biden administration’s policy of denying tech to China began to break down.

What will happen to U.S. economic power under Trump? Policy chaos is not a good recipe for continued global influence.

“AI”

Shoggoths amongst us. Riffing on a piece co-authored with Cosma Shalizi for the Economist: how LLMs are like markets, bureaucracy and democracy rather than individual human minds.

The political economy of Blurry JPEGs. A friendly response to Ted Chiang - lossy summarizations of available human knowledge can actually be pretty useful!

ChatGPT is an engine of cultural transmission. On Alison Gopnik and how LLMs are like libraries and languages.

Kevin Roose's Shoggoth. Why we treat LLMs as intelligent agents, even when they aren’t.

The Map is Eating the Territory: The Political Economy of AI. The battles over AI and intellectual property.

Vico's Singularity. What Kafka and Renaissance humanism have to tell us about AI.

Large Language Models are Uncanny. LLMs are haunted: voids that seem to speak.

There's a killer app for Large Language Models. LLMs are prayer wheels of organizational ritual. Riffing on another Economist piece - this one with Marion Fourcade.

After software eats the world, what comes out the other end? Gopnikism and Ted Chiang revisited: how LLMs affect cultural creativity. Also, my favorite of all the post titles I’ve come up with.

The Management Singularity. How LLMs will transform large organizations.

Democracy and tech

The cult of the founders. Elon Musk is a prophet trying to do a priest's job.

Silicon Valley is an aristocratic culture. Reading as a strategy for power.

Illiberalism is not the cure for neoliberalism. Danielle Allen’s democratic alternative.

Why did Silicon Valley turn right? Look at its political economy for answers.

The building blocks of state capacity liberalism. Trying to put the state capacity debate in order.

Not popularism. Not deliverism. Partyism. A different theory of what’s wrong with the Democratic party.

Unclassified

Patrick O'Brian is a Great Conservative Writer. Exactly what it says on the box.