The World Machines Project
Yes, we're doing this nonfiction extended universe dammit
The loose World Machines framework I developed to guide the readings selection for the Contraptions Book Club has proved surprisingly popular and fertile, and people besides me are starting to use it to scaffold their thinking and writing. Besides me, Aneesh Sathe, Florian Lohse, Ivo Velitchkov, Chor Pharn, Kyle Mathews and a few others on here have been employing the World Machines frame to varying degrees, in their own writing and thinking. Which is hereby retroactively open-sourced or something. The framework is less an idea than a sort of opinionated cognitive coworking space I think.
In fact, I think I’ve accidentally started a collaborative World Machines Project out of a subset of members of the book club. Some of us have been batting around an idea of doing a kind of collaborative World Machines book (in addition to our individual threads of inquiry, with reuse of writings/materials). This goal of the WMP is to write that book. Or some suitably unholy LLMified monstrosity that only looks like a book. Minimum viable scaffolding, aggressively obnoxious use of LLMs at any and all stages, and rough consensus and running code as the guiding principle.
Being part of the book club (ie, having read a reasonable fraction of the books from the last 15 months) is necessary but not sufficient for membership. If you’ve written at least one essay referencing the World Machines frame, you are eligible to self-select into this set. To opt-in, simply join this chat and put at least one link to a World Machines framed essay in the Google Sheet linked there.
I want to put the lightest possible scaffolding around this, separate it somewhat from the book club, and see where it goes. My initial thought is a shared git repo set up as a shared Claude Code project. Maybe a DFOS space. Let’s discuss all that in the chat.
What’s a World Machine?
For those who came in late, the basic idea is that the world can be understood through the lens of long-lived “world machines” that take about 400 years to build, operate stably for 400 years, and then decline/collapse relatively rapidly. The connection to our book club is that each year, the book club studies one of these machines (“configurancies” would be a more accurate term, but let’s stick with “machines” as the more evocative one). Last year, we studied the Modernity Machine, and this year we’re studying the Divergence Machine. Next year, the plan is to study what I’ve tentatively dubbed the Liveness Machine.
At any given time, there are 3 world machines operating in parallel — a growing one, a mature one, and a dying/recently dead one. We can refer to them as the Dawn Machine, Day Machine, and Dusk Machine, following the scheme of the Cleons genetic dynasty on the Foundation TV show. We’re doing a kind of psychohistory after all.
Here’s a convenient table:
If you’ve been participating in the book club and this project interests you, just write an essay exploring some aspect of the idea, add it to the spreadsheet, and you’re in. If/how your contribution actually gets synthesized into the collective thing is a tbd question. There will be quality control and consensus mechanisms eventually, but for the moment I’ll be the BDFL of this thing. We can diverge individually, but converge ironically together.
If you haven’t been participating in the book club, dive in anywhere you like, by reading some reasonable sampling of the picks from the last 15 months (I’d say 3-4 is the bare minimum) and then write something.
And of course, you don’t have to participate in the WMP. You can just do the book club.
In Media Res Starter Notes
For those who are already in the flow of this thing, some starter notes that may help you reorient what you’re already doing a little to prepare to collaborate.
These notes may or may not make sense to people who haven’t been following this thread of the newsletter closely, but read them like an in media res introduction to a TV show episode or movie, where you’re dropped into the middle of the action with no explanation.
The book I’m currently finishing, The Infidel and the Professor, unlocked a key question for me: How the Dawn and Day machines relate when both are strong enough that neither can entirely dominate. In the 1740s-90s period when David Hume and Adam Smith were working with close mutual influence, they were both heretics (real heretics, not Thielean ersatz heretics) within the Modernity Machine and founding figures of the Divergence Machine, but didn’t have to pay much of a cost for their heresies. A key “tell” from the book is that both took religion and theology entirely out of their intellectual work; Hume openly and combatively, with extreme prejudice, Smith more circumspectly and diplomatically. This really captures the “generational war” aspect of WMs, making the Dawn/Day/Dusk typology very useful.
The WMs framework feels like “Strauss-Howe for civilizations” with a cycle time of 1000 years instead of 80-100. But I’m very wary of cyclic history models (Kondratriev, Perez, Turchin, Sorokin, all the way back through Toynbee, Gibbon, Ibn Khaldun etc). The WMs framework is meant to be a clarifying and legibilizing scaffolding, not a “theory” of history. The WMP should put raw historical phenomenology first, rather than force-fitting it into the framework. There will be loose ends and that’s fine. The WMs framework is meant as sensemaking triage and a shared conceptual language, not as a Grand Unified Theory (GUT).
The Book Club —> Theorizing route seems very useful, so I think we should codify it as a requirement for participation. My suspicion is the WM framework will be exactly as useful as the number of relevant reads (books mainly) that precede a written piece. I think a good protocol for this is — at any given time, your book reference set should be at least 30% from our shared book club list. If it falls below, you’re essentially forking from our consensus headspace. Which is fine, but it means it will be less useful for the rest of us trying to synthesize. Otoh, only reading within the book club is probably bad. If you’re not bringing in ideas from stuff only you are reading
The current Dawn machine, which I’m calling the Liveness machine, starts with the cusp technology of generative AI, which is poised between a divergent non-living process, and a self-organized critical living process. Shoggoth-like basically. We’ll study it next year, so resist the temptation to jump the gun on it.
Random thought I’m trying to chase down now: The Modernity Machine was a pull machine, pulled along by a telos of Progress.™ The MM is convergent because the same small set of pull forces act on everybody. The Divergence Machine, MM otoh, is a push machine, driven by individual or small-scale push forces. This is why it diverges (think front-wheel drive, vs. rear-wheel drive with no steering… the latter is going to go off in random directions). One implication that divergent history is a much stronger function of “grounding” conditions.
A lot of people who are enjoying the WMs framework also enjoy cybernetics/system dynamics approaches to the underlying topics (eg. Maturana/Varela autopoiesis etc). I’ve said this before, but just to put it on the record for this project, I’m mildly hostile to these, and as BDFL, I’ll be adopting a kind of “disagree but commit” attitude towards contributors who explore threads based on those ideaspaces. It’s not that there’s no value there (there’s plenty), but the ideas come with more baggage and their own history/tradition than I want to deal with.
I think we’ll be doing a kind of psychohistory. Asimov tripped on chaos theory, which he tried to retcon into Prelude to Foundation, but I think our broad approach will be closer to long-range weather/climate forecasting. And instead of Seldon Vaults with our digital ghosts trying to nudge history in the future, our candidate influence mechanisms will look like terraforming or weather control tech. Except in events/time rather than space. And instead of a first/second foundation conceits, we’ll have some sort of blurry protocol that has high-n cardinal structure rather than ordinal structure.
Starter Questions
What is the full inventory of WMs since the dawn of civilization (say Neolithic Revolution)?
Can we retcon a WM onto any historical era or are there necessary/sufficient conditions? For eg: if planetary connectivity is too weak, is a WM meaningful. A good test case is the Bronze Age, where the tin trade was the primary “global” dynamic afaik. Is that enough to call it a WM, or should we treat that age as a set of river-valley civs that did some trading?
Assuming the 400 year time constant and 1000-1200 year full lifecycle of contemporary WMs, was it slower before? I’d imagine so. For eg. taking the Axial Age as a quasi-useful construct, that had a lifespan of about 1600 years (800 BCE to 800 CE)
What’s the micro-to-macro fractal structure of WMs? Is there necessarily one? Can there be “thin” WMs that are primarily at one or other scale?
How is the prevailing set of WMs understood in its own time? We are thinking about WMs from our location in 2026. How did people in 1776 understand MM and DM? Did they anticipate LM from that distance? Did they relate to the Medieval Machine differently from us, as an active shaper of history rather than a romanticized source of larps?
How is the prevailing set of WMs understood from different loci within it. Besides the obvious geographic diversity angle (American vs. European vs. Chinese understandings for eg), there are probably other interesting loci.
How can we map/visualize WMs well?
Join the chat to continue talking about this stuff. We’ll move to a better place than substack chat eventually, but let’s start there since we’re all lazy.