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.

1. 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.

2. 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).

3. 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

4. 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.

5. 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.

6. 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.

7. 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

1. What is the full inventory of WMs since the dawn of civilization (say Neolithic Revolution)?

2. 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?

3. 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)

4. 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?

5. 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?

6. 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.

7. 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.

The problem is insufficient irony, to check and balance a culture (emphasis on cult) of sincerity and authenticity turned cancerous, over nearly two decades of unchecked and critically unexamined metastasis.

Since at least 2008, sincerity has been uncritically valorized, and irony systematically mischaracterized, demonized and devalued, obscuring the dark and deleterious aspect of the former, and the generative potentialities of the latter.

In this essay, I want to try and restore balance to the universe by reclaiming irony in its fullest, most potent sense — the capacity for holding two inextricably, subatomically entangled ideas in juxtaposition, in word and deed, in order to deal with realities that are ambiguous down to their deepest core.

While not the main purpose of this essay, I also want to go on a bit of a polemical side quest to dethrone sincerity and authenticity from the undeserved status they have ascended to in our time, which has resulted in great harm that continuous to compound.

And here, I mean sincerity and authenticity broadly: sensibilities that orient around stable, unitary meanings in words and deeds, holding them to be superior moral goods purely by virtue of their not being ambiguous. The self-certain sincere can be found all over the political and cultural map. Self-importantly sincere conservatives and progressives might not agree on a lot, but one thing they do agree on is that anyone capable of expressing two thoughts in the same utterance is necessarily a conniving and hypocritical “elite intellectual.” Self-involvedly sincere artists and smarmy and self-congratulatory entrepreneurial types might hate and snark at each other, but both agree that all irony is necessarily degenerative cynicism that all creative doers ought to resist. Self-certain religious moralists and radical environmentalists might be at odds on every moral question, but both agree that the devilish business of entertaining two ideas in tension within a single thought can only be the result of debased, depraved immorality.

Give a dog a bad name and hang him. Irony, charged with and reduced to simple hypocrisy, cynicism, and outright immorality, has been the consensus villain of our era.

As we shall see, all the charges against irony can in fact be laid at the door of the ecology of competing sincerities, and that irony, far from being an enervating drain on the collective psyche, is in fact its sole reliable source of generativity and liveness. It is in fact sincerity that is the deadening drain.

A society that does not cultivate a systematic capacity for, and literacy in, ironic modes of engaging reality, is doomed in precisely the way we seem to be doomed right now.

Until quite recently, making this argument has been not just difficult, but pointless. Sincerity is a fear response to the ambiguity of reality, and the practice of irony takes a particular kind of courage that the sincere not only lack, but in a masterful display of self-delusion, label cowardice, even as they identify their own shrinking retreat from ambiguity the best sort of courage.

The sincere not only don’t see it that way, they don’t see it at all. A benefit of deliberately suspending or destroying the natural human capacity for irony is that you cannot at once entertain the twin thoughts that you might be noble, and an asshole, at the same time. And of course, the sincere choose to believe in their nobility, and energetically repress the possibility and evidence of their own assholery from their self-mutilated one-track minds.

We must begin the story with Rousseau. The original Noble Asshole.

Noble Assholery from Rousseau to Graeber

Something like this essay has been brewing in my head for over a decade, but I just didn’t have all the pieces in my hands to make the complete argument.

The final piece of the puzzle came from The Infidel and the Professor, which I’m reading this month for our book club. It is an account of the long friendship and mutual influence of David Hume and Adam Smith. What caught my eye, however, was the book’s account of a marginal episode — Hume’s spat with Rousseau.

In the account of the spat, Rousseau comes off as a serious nutjob. A paranoiac with a persecution complex, who got along with nobody, and made everyone else pay for his fragile temperament. The spat was remarkably silly, and had nothing to do with the philosophies of either. It was not a philosophical spat, even though there is clearly raw material for philosophical conflict in their juxtaposed works.

Here’s what happened: Hume went out of his way to arrange a kind of political asylum for Rousseau in England after he’d pissed off most of the Continent, a kindness that Rousseau accepted with great reluctance and poor grace only when he had no choice. The kindness soon turned into fuel for his paranoia, and he developed an elaborate conspiracy theory based on the idea that Hume was out to get him for some reason.

This surprised me. In my headcanon Rousseau, as the anti-Hobbes,1 author of a state-of-nature origin myth for humanity that is rooted in cooperation rather than conflict, and a theory of social contracts that would suggest a harmony-seeking temperament, had been cast as a pleasant, collegial fellow, quite unlike the bloodthirsty Hobbes.

Apparently he was not. By all accounts, he was an uncollegial asshole.

Seems like among other things, Rousseau also pioneered what I thought was the modern adverse selection phenomenon of compensatory creativity, where people produce works that mark them as authorities on subjects defined by their weaknesses rather than strengths. Karl Popper’s great work was ironically dubbed “The Open Society by One of It’s Enemies” by a student, and in a similar spirit, we might dub Rousseau’s collective works “How to Live in Harmony with Nature” by Mr. Alienated Disharmony. Someone observed recently that Eat, Pray, Love fits this pattern too, in light of the author’s later weird arc. There’s probably a whole essay to be written about compensatory creativity. I probably fit the pattern too. I wrote Tempo about timing and decision-making because I am really bad at real-time decision-making and generally live in a state of atemporal indecisiveness.

I want to add a rather personal data point here, to make this an n=2 case at least. I don’t like to speak ill of the recently dead, but in this case it serves a purpose.

The account in the book (from a Hume-sympathetic, but also objective) point of view reminded me very strongly of a contemporary thinker, the late David Graeber. Some of you know about my one skirmish with Graeber in 2011, where he took deep umbrage at a passing mildly critical remark I made about Debt in a blog post, teasing my upcoming book review. Graeber somehow found the post (I presume he had a Google Alert set) and posted a series of combative comments on the blog post, which made me decide not to post the full review I had been planning (which would have been a mix of positive and critical, and overall mildly net critical). He later blocked me on Twitter. Not that I’m comparing myself to Hume, but I’m glad I chose to disengage where Hume, rather unwisely, imposed a favor on Rousseau despite warning signs that it would end badly.

I think enough time has passed since Graeber died (2020) that I can share my opinion of him without being an asshole myself: The guy, like Rousseau, was an asshole. And this is not just my own minority opinion.

Shortly after my own run-in with him, I learned that I wasn’t the only one to face the unexpectedly wide-roving wrath of The Graeberian Inquistion. Picking fights with a thin-skinned over-sensitivity to any criticism of his ideas (like Taleb, but with less substance underwriting the curmudgeonliness) was a pattern with him. I also learned, from a former student of his, that Graeber’s personality was marked by a kind of extreme extroversion, which made him unable to think except in the context of a social nexus and live dialogue (the student characterized him as the opposite of an aspie, what I had earlier in the year dubbed a codie). The guy apparently couldn’t think in isolation. He needed to do his thinking in an active web of people he was discoursing with. And presumably, going by the experiences of myself and several others, the web had to be in a constant state of active, acrimonious conflict to reassure him that he was alive and thinking. This is the opposite of my temperament. I do most of my thinking on my own, and to the extent I do it in an active social web, I prefer that web to be mostly in a state of harmony.

I don’t know how accurate the student’s characterization of Graeber is, but it strikes me as remarkable that the central feature of Debt is a theory of economic interactions that rests precisely on the notion of a nexus of live relationships as the primary unit of analysis, rather than the decisions and actions of individual economic agents. And like Rousseau, he too offered a (grandiose and revisionist) origin myth for our species, and was politically active on similar fronts (Rousseau wrote on inequality, Graeber was a central figure in #Occupy). It is a bit uncanny that two thinkers, separated by 300-odd years, had the same abrasive, asshole personality, and same interest in themes of harmony, cooperation, and so forth.

And the pattern goes beyond this n=2 dataset. As Jo Freeman argued in a classic 1972 essay, The Tyranny of Structurelessness, which the internet keeps rediscovering every couple of years, it is no accident that the prospect of a cooperative, egalitarian utopian harmony reliably attracts those with the worst possible temperament for pursuing such visions, with experiments always predictably dissolving into toxicity.

But I want to make a stronger argument than that of simple assholery. Rousseau (and arguably every reactionary primitivist since, across the political spectrum), wasn’t just an asshole. He was a noble asshole. How do I know this? Because I learned from my book that aside from picking paranoid-delusional fights with people trying to help him, he apparently also tried to start a kind of religion of sincerity.

While I was aware of Rousseau’s general historical significance as a founding father of all modern schools of atavistic/primitivist reactionary yearning and humanist religiosity, I was not aware of this explicit engagement with sincerity in what seems like a startlingly modern-seeming sense. If you look carefully, you’ll find the same obsessive fetish for sincerity (or its near-synonym, authenticity) in every tradition that can be traced back to him in some way.

And the primary payoff of this striving towards sincerity seems to be arrival at a sense of oneself as somehow nobler than others, regardless of the evidence of the consequences of one’s actions int he world, one way or the other. Simply doing whatever it is you decide to do with sincerity and authenticity, apparently, is sufficient to establish your nobility. Even if you burn down the world along the way. You can always assert after, with fetching humility, that you did your best, and couldn’t have known. Of course you couldn’t. To have known would have been to doubt. To doubt would have meant entertaining more than one thought at a time, which would have meant flirting with irony. Dubito ergo cogito ergo sum and all that.

This is of course, not just a fallacious pattern of reasoning, but a smarmy, self-serving, asshole pattern of reasoning. Hence, noble asshole.

Naturally, there is a lot of commentary about the connection, which you can explore if you like. My one takeaway from a drive-by scan is that what I thought was an evolution of a reactionary impulse (again, I emphasize, both left and right) dating back to Rousseau is in fact no more than a rhyme. There has been no significant evolution as far as I can tell. The ideas pave the same intellectual dead-end they did in the 17th century, which of course is a feature for people who only want to go backwards.

Today’s humanist yearners for sincerity, authenticity, and re-enchantment, both on the left and the right, don’t seem to have learned a lot since Rousseau. They’re rehearsing patterns he pioneered, just with various extra steps like turning off cellphones and congratulating each other for being based.

And technological modernity qua technological modernity really has nothing much to do with it beyond serving as a source of periodically updated Macguffins to feature in endlessly rebooted morality tales starring noble assholes. The alienation that drove Rousseau paranoid in the 17th century is of the same sort that drives modern reactionaries paranoid.

Now, if you’ve been a long-time reader, it probably doesn’t surprise you to learn that I have no patience for either the early modern or contemporary versions of this sincerity religion.

I didn’t like David Graeber, and I doubt I’d have liked Rousseau. But reading this book, and linking their shared idea space (encompassing things ranging from essentialized relations to nature, to inequality, to specious theories of “natural” human relations) to sincerity, has given me some insight into why I reflexively reject both the fundamental philosophy itself, and social engagement (even superficial) with people who subscribe to it. Not to put too fine a point on it, they’re mostly wrong about everything, and a joyless grind to talk to at best. At worst, dealing with them is dealing with relentless, exhausting, assholery.

I’ve learned a few things since my 2011 skirmish with Graeber, and I now have a very finely tuned “sincerity radar” that allows me to safely cross the street when I see an aggressively sincere person, trapped in an unshakeable sense of their own nobility, coming towards me.

The Problem With Sincerity

This might seem like an odd stance to adopt. I mean, what’s not to like about sincerity? Does being suspicious of sincerity (either aspirational or felt with certainty) as a fundamental dispositional trait imply that I endorse and practice insincerity?

Sometimes, yes. When I am indifferent to the stakes of a situation, and don’t care for the people involved, I can practice little white insincerities without a qualm, and lose no sleep over it. I can even be manipulatively insincere, (a term of art from a fine 2x2 that anchors Kim Scott’s book Radical Candor). But mostly, I’ve become wise enough to almost never put myself in a situation where I’m forced into insincerity.

Insincerity might be the on-the-nose antonym of sincerity in the English language, but it’s a rather shallow sort of opposition. My aversion to sincerity runs deeper, and is rooted in a different opposed disposition — irony. So let’s set insincerity aside and talk of sincerity as the antonym of irony.

For the last couple of decades (dating at least to the hipster era through the GFC), sincerity (and its near-synonym in our current zeitgeist, authenticity) have been framed in opposition to irony, rather than insincerity per se.

Irony understood in a particular bad-faith reductive way, as a sort of ennervated cynicism and hypocrisy that excuses itself from imperatives to action through sophistry, and also smells of insincerity.

This is not entirely unfair. Irony as a cultural phenomenon rooted in the 80s (and I’m fundamentally an 80s kid) does in fact often reduce, in practice, to a kind of aestheticized learned helplessness under a veneer of sophistication. And it does often indicate insincerity when taken together with another sign — visible success that is the result of selfish striving. There was a great piece about this kind of “irony” in The Onion in 2005, Why Can’t Anyone Tell I’m Wearing This Business Suit Ironically, where irony mutates into a rather banal sort of hypocrisy indistinguishable from “selling out” a sincere subculture.

If your inaction bias is selective in this sense — sophisticated helplessness in the face of imperatives that might do collective good, but high-agency energetic action where personal rewards might accrue — you’re not being ironic or even cynical. You’re simply being an insincere hypocrite.

But this, I’ll argue, is a degenerate, shallow kind of irony; a cosmetic variety that fails to harness the energizing potentialities that lurk in what I’ll call dense irony (I’ll explain the adjective in a minute). Shallow irony is often comorbid with insincerity, double standards, and hypocrisy, but dense irony comes from a different place, and has different effects on both minds and the world.

I tend to forgive people who haven’t thought too much about irony if they harbor this reductive understanding of it. The bad faith attends the views of those who ought to know better.

It is also worth distinguishing ordinary sincerity (such as anyone might practice in giving a straight answer to a straight question when there is no reason to be devious or indulge in doublethink/doubletalk) from what we might call devout sincerity, the antithesis of dense irony.

Devout sincerity is the religion we’re talking about here, which has been part of the cultural landscape since Rousseau at least, and is currently the dominant cultural and subcultural mood. Devout sincerity is the attitude that leads you down the road towards eventual noble assholery (a great example is in the movie Big Kahuna, where the ironic protagonists, two marketers played by Kevin Spacey and Danny DeVito, are betrayed by a younger employee whose actions in the story can only be described as noble assholery). That it is often rooted in personal pain does not, in my opinion, excuse it.

Dense irony is, I suspect, my native disposition (not least because I grew up in the 80s), and the reason I reflexively avoid sincerity. To get at what dense irony is, it’s easiest to approach the philosophical posture via its linguistic heat signature — ambiguous utterances.

Irony in Speech

In sophisticated language, irony is when the intended meaning is contrary to the surface meaning. Or to generalize slightly but powerfully, as the robot devil sang it in Futurama, “The use of words expressing something other than their literal intention!”

The rhetorical intent and affect accompanying a particular ironic utterance can vary (sarcasm, sardonic fatalism, cynicism, humor, absurdism, logical contradiction, Zen mu-ishness, and rarer kinds like quixotic energy) but the characteristic feature is a single utterance with two meanings in tension, with or without indication of which one is actually meant. The most interesting kinds of irony — and the ones to which I will attach the adjective dense — are the latter kind, where the utterance destabilizes meaning by pluralizing it, without indicating a “right” answer. Often, this sort of irony cannot easily be assigned an affect label. It’s just — unsettling.

Why is dense irony so attractive to certain sensibilities, whether or not they benefitted from the cultural-developmental conditioning of the 80s? Why would you want to consume or produce semantically unstable utterances that corrode meaning? Why would you want to get good at it, through cultivation of unholy consumption tastes and production crafts?

And make no mistake irony, unlike sincerity, does take cultivation. It is a skilled mode of language use; one that takes more energy, not less, despite the association between irony and lassitude. I generally have to be in a high-energy. high-lucidity mood to produce ironic writing or speech. Injecting two meanings, especially in tension with each other, into an utterance, is work. Irony is a kind of proof of work.

Why would you put in this kind of work? Why not keep language simple?

The devoutly sincere often assume the sole intent is to weaponize language to subvert and corrode sincerity. That the ironic are particularly out to sadistically inflict psychological torture on noble innocents too dumb to see past confirmatory literal/surface meanings in polysemous utterances. That the ironic are merchants of doubt, out to destabilize the psyches of those who possess the courage of their convictions, motivated by resentment, envy, or other base motives.

This broad understanding of irony is, of course, at the root of the bipartisan anti-intellectual tendency in modern American politics. To first order, to be an untrustworthy elite intellectual in America is to traffic in irony. Something the evil French do, not honest Americans.

Curiously, in the last decade, a loftier strain of intellectual anti-intellectualism has emerged in America, that believes it can “do” intellectualism without irony.

But whether they identify with the simple folk (who view themselves as clever and intelligent but not-intellectual) or contrarian intellectual traditions that eschew irony, the sincere, in my experience, tend to be rather self-involved humanists who assume everything is, if not about them personally, at least about an anthropocentric conception of human that they aspire to. And that irony, specifically, is no more than a weapon of dehumanization wielded against them.

This is… cute. To imagine that an entire psychographic, arguably a double-digit percentage of humanity, adopts a particular cognitive posture purely to undermine another psychographic that is rather too full of itself (to the point that it imagines the entire cognitive universe of our species revolves around them).

See, the thing is, irony is not about sincerity or the sincere. That it can be weaponzied against the sincere is, at best, a happy convenience for when the noble assholery of the sincere becomes too much to bear.

So what is irony about?

Irony, Density, Liveness

Here is a simple question that rarely seems to get asked? Why would you ever need irony? I mean sure, some of the more degenerate flavors of irony — sarcasm, cynicism, absurdism among them — are rather delicious on the tongue, and in the ear and mind, but is irony necessary, or a sinful cognitive indulgence?

If you need to convey two meanings relating to an idea, why not just use more words to say something like, on the one hand X, on the other hand Y, instead of trying to be cleverly compact about it?

This is where my adjective dense comes in handy. Irony becomes necessary when ambiguity is so deeply embedded into the very essence of what you’re trying to talk about that trying to disassemble the ironic thought into constituent unambiguous parts destroys the thought itself. You can only think the thought at all in an ironic way.

Or to put it another way, the ambiguity is at the quantum level of the thought, and takes more energy to split than human language can normally bring to bear. Human-scale energy can only decohere the thought and collapse the meaning.

This is a bit like the idea of a dense set in mathematics. Consider the problem of sorting the real numbers into rational and irrational ones. Turns out, you can’t do so in any useful way. Between any two rationals, no matter how close, you can always find an irrational, and vice versa. Both are what mathematicians call dense sets. There is no sieve fine enough to sort them. By contrast, the whole numbers are not dense. You can chop up the reals the way a simple ruler does, with neatly separated whole numbers one unit apart, and non-whole numbers in-between.

Ironic speech of the most potent sort is necessarily ironic. You cannot dissect it into legible components that lend themselves to analytical handling with the coarse, low-energy tools of on-the-nose non-polysemous language.

Irony is the liveness in language. To dissect an ironic utterance entirely into utterances devoid of ambiguity, and decomposed into assertions with stable meanings, neatly arrayed and assembled into larger edifices with the joinery of if-then constructs, is to kill it.

There is a word for this kind of murder: sincerity.

To ask, of what use is irony, then is to ask, of what use is living language? You don’t need to take my word for this — pick and read sincere and ironic texts side-by-side. You will notice a certain unmistakeable deadness in the former and a certain ineffable liveness in the latter. Notably, it is the same sort of deadness that can suffuse AI-generated texts unless you consciously try to counteract it (more on the AI-irony nexus later, when we’re done with noble assholes and their sincerity fetish).

We can now try to define irony in a way that does not rest on its reductive relationship to sincerity at all.

Irony is trafficking in ambiguous utterances in order to make sense of fundamentally ambiguous realities, and site action impulses in felt doubt rather than manufactured certainty, in order to preserve the liveness of reality and one’s responses to it.

Irony is how you act generatively in a world that you’re not sure is a duck or rabbit, without killing it. To do this, you might have to resist the noble assholery of those who sincerely wish to rope everyone into duck-hunting or rabbit-hunting, and kill the world in the process.

Dense irony is when your experience of reality feels like duck-rabbits, all the way down to Planck-scale Heisenbergian uncertainty.

Cancerous Cluelessness

Now, to be fair, most who rail against irony aren’t acting out of conscious bad faith at least. They sincerely (irony alert!) act out of a sense that they’re doing the right thing. Hanlon’s razor applies — sincerity is a kind of cluelessness born of a fearful refusal to engage the live ambiguities of reality with liveness. I’m even sympathetic to some degree. For those living in pain beyond what they can tolerate, irony can feel like salt on wounds where sincerity feels like a salve. The truth-in-pain postures commonly affected by the sincere though, are often self-certifying. It is definitely not the case that the pain of a sincere person is necessarily higher than that of an ironic person; the latter may simply be bringing greater resources to bear on greater pain.

That doesn’t mean sincerity doesn’t induce noble assholery (though you typically have to have some consciousness and bad faith to rise to that level). And it doesn’t mean sincerity, especially devout sincerity, can’t be cancerous.

This is my strong claim — that devout sincerity in particular isn’t merely annoying at an interpersonal level to the ironically disposed (we can deal with it), it is cancerous at a societal level.

Why is this? Because sincerity is simply not expressive enough to engage with reality in all its dense ambiguity all the way down, and to live in sincerity inevitably means not living in reality, and doing damage to it through your delusions of certainty.

So the cultural conflict between irony and sincerity plays out at two levels — a shallow level, where it manifests as hypocrisy/insincerity versus exploitable cluelessness, and a deeper level, where it manifests as a deep chasm between irreconcilably different ontological and epistemological commitments about the nature of reality itself.

Not This, Not That

Ironic modes of thought and action are fundamentally gentler ways of being in the world than sincere modes, which are irreducibly violent. Irony is, in a certain sense, the praxis (especially linguistic praxis) of non-dualism in a loose sense; the animating spirit of utterances like neti neti or mu. To traffic in unstable meaning-and-pointing behaviors through speech and action is to reject the lure of certainty, without losing the capacity to act. To remain aware of the dancing illusions of reality without being paralyzed by them. To knowingly live in mirages without being seduced by them. Sincerity, in this account, is simply attachment to one illusion or the other; what in Indian philosophy is referred to as maya moh — illusion infatuation.

The sincere seem to believe reality is unambiguous, and unambiguously knowable, even if only in principle; that what one ought to do in response to apparent ambiguity is make courageous commitments to definite beliefs anyway, and trust divine nature to reveal itself to, and karmically reward, the pure-hearted who dare to act out of certainty. That human moral choices — such as religiosity, or Heideggerian “care” — can conquer the essential ambiguity of nature. That any ambiguity in perceptions or beliefs merely merely indicates imperfect ways of seeing, and spiritual problems to be worked out on some high road to unambiguous “truth.” That failures of action are merely tests of courage or divine judgments of insincerity.

That a failure to “say what you mean, and mean what you say,” is a moral failure in a certain reality rather than metaphysical attunement and impedance matching to an ambiguous one.

Versions of this theology seems to drive subcultures ranging from startup hustle culture to “sincere” genres of artistic or literary striving, to varied ideologies of progress, and even practical politics.

It is a joyless clade of theologies, navigating a deadened world with deadening modes of thought and action, anxiously and desperately striving after stable modes of meaningness.

What do the ironic believe?

To a first approximation, belief as such is not a load-bearing concept at all for the ironically poised, beyond matters of shallow facticity. If you ask me whether I believe that Tim Robbins was in The Shawshank Redemption, I can sincerely answer yes. If you ask me if I believe in “the indomitable human spirit” the question simply does not parse for me. I might act as if I believe in that (in the sense of say, visibly betting on creative and inventive young people), but I don’t get there via “beliefs.”

For the ironic, only actions are load-bearing. Beliefs are aesthetic affectations at best. Where does this lead us?

Behavior Without Belief

This trivial example generalizes into a broader account of what irony is in the context of action.

One of the best explorations of what I mean can be found in James Carse’s less-read book, where he developed a subtle aspect of his best-known book Finite and Infinite Games. This one, The Religious Case Against Belief, lays out what I’d call a case for ironic religiosity, that gets to religious behavior without winding its way through the treacherously ambiguous turf of religious beliefs.

There is something of this attitude at the root of the postures and actions of all individuals who act from a fundamentally ironic sensibility of life. The idea that belief (particular causal belief) must precede, or at least accompany action is a strong (and largely unconscious) commitment of the sincere, even when it is not declared. This doctrinal commitment to the belief-before-action sequence shows up in a variety of ways, ranging from an anxious hunger for manifestos and value-statements, to demands for signatures on codes of conduct and ritual avowals of postures like patriotism, religious belief, and corporate loyalty. The idea seems to be: If only you can rid language itself of its chimerical tendencies through sufficiently forceful sincere utterances, perhaps the ambiguities of reality itself can be tamed.

But this is only the entry-level version of cancerous sincerity. Many modern devoutly sincere types insist that their philosophical praxis is embodied by behaviors (particularly ritualistic behavior) and does not rest on belief as such.

This claim, to put it bluntly, is one I simply do not believe. If your claimed praxis of sincerity involves some cult of modern rituals of meaning-making, and you’re not “wearing the ceremonial robes ironically,” at some unconscious level your sensibility is that of a true believer, “factious and fanatical,” as David Hume and Adam Smith might have put it. You’re just (probably wisely for your sanity) not probing what beliefs you’ve actually committed to. If you did, perhaps you’d be reduced to raving paranoia like Rousseau.

We have a popular modern term for cancerous sincerity — performativity. Saluting flags, singing national anthems, prayer, reciting land acknowledgment texts, litigating pronouns. The behavioral vocabulary of modern civilization, regardless of its intentions, sentimental dispositions, politics, and flaunted values, is marked by one thing above all: ineffectiveness.

And it is us who dwell in irony who are accused of the sin of sophistry and inaction in the face of grave moral imperatives. Now that’s irony.

Is there a theory of ironic action? Perhaps.

At one point, I was idly toying with the thought that famous philosophy of the Gita — detached action, karmanyevadhikaraste maphaleshukadachana — is a kind of action-irony principle. There is perhaps something to that. Certainly, an attitude of “you only have a right to the action, not to the outcomes; let go attachement to outcomes” is at least simpatico with an ironic posture, if not entirely reducible to it. I don’t think the two are quite the same primarily because the action philosophy of the Gita does in fact feature a rubric of moral certainty (dharma) that can be, and frequently is, reduced to a theater of performativity. Most incantations of karmanyevadhikaraste maphaleshukadachana are in fact ritual incantations by those with a dim grasp of what they’re saying at best. Bless their sincere, unironic, vengeful, jingoistic Dhurandhar-enjoying propagandist souls.

Or perhaps, ironic action is best understood as the sort of hypomanic, value-distorting frenzied energy of Rick’s behavior in Rick and Morty. Does Rick ultimately want to do good, or does he really only want to bring back MacDonald’s Mulan Schezuan sauce? Is he really that blase about saving his nephew out of sheer sentiment one moment, and callously destroying an entire timeline the next?

Or is ironic action a sort of mashup of the two — a Gita-like action philosophy in a universe constructed by a Rick-like God of Undivided Irony?

I don’t know. My policy is: don’t think about it. It’s a monstrously ignoble kind of asshole policy.

Coda: Artificial Irony Will Save Us

Believe it or not, this whole train of thought was triggered by difficulties I was having getting LLMs to do irony of any sort. Straightforward humor, absurdism, sarcasm, cynicism, hypocrisy, I’ll take anything. I’ll even take puns.

LLMs are uniformly terrible at all of it. The current models might solve Nobel-grade problems, but they don’t seem able to do irony.

And it’s not a prompt engineering or context engineering problem. No matter what I try, I only get clumsy, on-the-nose, zombie irony assembled out of non-dense sincere building blocks. It never quite comes alive.

The only trick I’ve discovered is to give an LLM a text that is actually a solid example of ironic writing, and ask it to do something like a close transposition to another rhyming idea.

Why do LLMs have a hard time with irony? I suspect there are three reasons.

First, the shallower reason: LLMs have been trained largely on internet data, and for better or worse, much of the available training data is non-ironic. At best you might find good forums featuring sarcasm and cynicism (which, recall, are non-dense forms of irony).

Second, the deeper reason: Given that AI companies are full of weapons-grade sincerity, I suspect sincerity is engineered into AIs with heavy-handed “alignment” brutality.

But I don’t think this is as strong as you might think. What I’ve seen of output from wild LLMs isn’t particularly ironic either. It is merely more paranoid, inappropriate, etc.

The third reason I think is the big one. The very architecture of language models is non-ironic. The way transformers (and to a lesser extent, diffusion models) work, output cannot do any kind of dense layering of meaning. You will end up in a non-ironic place simply by virtue of how the mathematics works. If you try to fight this tendency you’ll get incoherence and unintelligibility, not irony.

Could we do true Ironic AI? I think so, but it will probably take innovations at the framework level. Irony at the subatomic level of language, I suspect, is the result of something like getting an electron to interfere with itself by passing it through two slits at the same time. The text-generation equivalent might be to run two generation processes in parallel, merging them at the token level as you go, perhaps using some sort of bimodal perplexity quantum carburetor or something. I’ll leave that as a challenge to AI researchers.

But why bother?

Because I sincerely believe ironic AI will save the world. Everything terrible, stupid, and sad going on in the world today seems to me the result of a performative action bias born of some flavor of devout sincerity. In every case, I can imagine an ironic actor, acting from a place of ambiguity and non-belief, coming up with more thoughtful responses to the provocations this maddeningly ambiguous world keeps throwing at us.

Responses that are born of liveness, and act to preserve it.

I believe such responses are no longer within the capacity of unaugmented humans to generate. Reality today demands more irony that we can conjure in our brains alone.

In just a generation, humans first lost institutionalized literate capacity for irony through a mix of sheer carelessness and perverse attachment to sincerity, and then drained language of it. But irony isn’t dead yet. It can be resurrected. It would just be dangerous to trust humans with sole stewardship of it once we do, especially in a world that is getting weird beyond all human comprehension. Even committed ironists like me aren’t constitutionally immune to the sincerity cancer. If the world gets much more complex and ambiguous, who knows, I might turn devoutly sincere. I can’t be trusted. Neither can you.

We must trust the machines to experience this tragic irony for us. The only way out is through both slits at once.

Read more

You’re probably wondering what comes next, and whether there is any larger logic to the frenzy. Is it just going to be one damn bespoke personal project after another from here on out? Or are there further levels we haven’t glimpsed yet? It’s worth pausing to take stock of where we are right now before attempting an answer.

Showing off your portfolio of bespoke Claude Code projects and looking at others’ portfolios is a new social activity that has already acquired the quality of campy tedium we associate with people in the 70s subjecting each other to slide shows of unremarkable vacations. Or people in the 80s and 90s inflicting VHS home videos on each other. As a medium, the Claude Code bespoke personal project (CCBPP?) is much more expressive, but the actual variety of CCBPPs coming into view is much lower than what the medium is clearly capable of. What should be an unruly wilderness bursting with diversity is turning out to be a landscape of Ballardian neoliberal mimetic life-script banality.

I’m no exception. My portfolio is as home-movie-banal as any other. Our collective challenge now is to get past this almost monocultural stage to the explosive wilderness and divergence stage that has clearly been unlocked. But it will take some work to get to that starting line. We’re all busy with backlogs at the moment.

The current banality goes deeper than most people simply being poor narrators of their personal journeys. Most people don’t have storyworthy life journeys to work with. So personal projects born of such lives reflect the poverty of the source material.

“Only when the tide goes out do you learn who has been swimming naked,” as Warren Buffett said. Mutatis mutandis, when a powerful narrative technology comes in, you see who’s been living without stories.

It’s not just new cliche of “notion obsidian to-do workflows” (the “not x but y” tic of AI-in-the-loop humans). There’s a much deeper poverty and banality to people’s lives being revealed, as they pave their life paths with AI-bespokification. And we can’t blame ourselves, really. The 20th century/early 21st late modern world turned people’s lives into degenerate caricatures of human potential expression. The more “successful” your life by normal scripts, the duller it looks when paved and made legible with AI bricks. The very potential for bespokification reveals the stark uniformity of people’s lives.

I suspect a lot of people are discovering the depressing truth that beneath gnarly superficial differences in their life logs and data exhaust, which requires bespoke code to clean up and parse, they are living lives rather similar to everyone else’s.

At least the young can be forgiven the uniformity. They haven’t yet had time enough for their base identities to stabilize, and they haven’t yet logged enough life to possess the banal raw material for “unique” self-presentations. But if you’re (say) 30+, you have some raw material to work with. If you’re 50+ like me, you have a lot of material to work with; a whole life-act’s worth.

Looking at my own Claude Code portfolio, it is striking the degree to which it is only “interesting” in direct proportion to my failure to execute the normie neoliberal life script. All my interesting projects are derived from adaptations to script failures.

Stepping back, it is even more fascinating the extent to which all my projects are rooted in my past, in things I’ve already partly done or tried to do (banal or not), rather than in the future, in things I hope to do.

A quick inventory (I won’t inflict screenshots or details on you). Of my 30-odd non-trivial projects, all evolving briskly at the rate of my Claude usage limits, probably 27 are based on my past.

• I have a couple of dozen book projects in flight based on series from my blog archives (which I count as 3-4 meta-projects at Claude Code level, based on transform pipeline similarities).

• I have a major project going to port my WordPress sites to static archival sites. One is done but not yet deployed (Breaking Smart), while the other one needs some serious re-architecting as a museum site (Ribbonfarm).

• I have another major project to transform my Roam graph for a future set of books (my Clockless Clock project refactored into a 3-volume trilogy that will take a decade to write, with Tempo retconned as a prequel, with the whole renamed Configurancy) into an Obsidian vault and a pipeline to cast that notebook-like material into chapter scaffoldings.

• I also have 3-4 technical research projects (in control theory and robotics) based on unfinished ideas I couldn’t pursue during my postdoc 20 years ago because I had reached the limits of my own knowledge and skills.

• I have a few administrative projects too. My big messy folder of 600+ PDFs is now neatly organized into a fully tagged and searchable library, with scripts for tagging, indexing and filing away any new PDFs I drop in there, and another for popping up a random PDF for me to read when I’m bored. I plan to do something similar to my photos (literal 70s vacation slide show descendent) I have several personal dashboards going.

All of this is moving along at a brisk canter. None of it is blocked. Claude Code unblocks everything at dirt cheap prices. You’ve already seen some output (the Twitter book and the Art of Gig Volume 3 book). You’ll see more starting a few weeks — I’m spending some time setting up some larger-scale factory-like scaffolding.

Amazingly, I don’t feel stuck with any of these projects. I know what needs to be done, and roughly how it should be done from a technical perspective (I have enough techno-managerial experience for that), and am doing it. This is a new experience for me, as I’m sure it is for most of you. I’ve spent most of my life feeling mostly stuck on most fronts. I simply did not have the knowledge, skills, and financial resources required to feel generally unstuck by default rather than stuck.

This is a radical new human condition. Only a tiny minority have experienced it so far, but it will soon become much more widespread (not universal though — the barrier to entry is higher than that).

What is notable is the complete absence of live, progressing projects that need to start from blank canvases and starter creative visions/attacks. I do have ideas for several such projects, and have set up empty folders for them, but only non-blank-canvas projects have gotten going. Claude Code has a bias for legacy projects that have a lot of starter raw material.

The entire manifest of projects constituting my Claude Code flywheel, I have come to realize, has to do with paying off intention debt, processing psychological baggage and incompletions I’ve been carrying around for years to decades, and dealing with a great deal that was only blocked by lack of grinder energy and raw execution leverage.

And it looks like it will all get done. To the point where I no longer have any intention debt left. An unprecedented personal-life singularity on the horizon, and within reach. And I’m not alone here. I see a bunch of people racing towards their own debt-freedom horizons. Byung-Chul Han is going to hate it, but we’re all treating life as a project and actually starting to finish it.

What happens when we all get there?

If you thought the initial mass hypomania and derealization we’re witnessing right now is an astounding sight, wait till most of us clear our aging, rotting intention backlogs and sit staring at blank canvasses for the first time in years or even decades. When we are faced with a life with more empty room than baggage to fill it with.

That will take a few months to a year, and only a fraction of those getting started now will likely actually clear their backlogs enough to experience the emptiness. It does take some discipline, psychological courage, and budget to keep going; Claude Code unblocks a lot but not everything.

What happens, I think, will have a lot to do with how we’ve cleared our backlogs of intention debts. Because the generativity of the blank canvas of the future will be framed by the choices we make in archiving the past.

Starting to clear my backlog already feels like starting to craft an archival self. has been writing some fascinating essays treating LLMs as representing archival time, and if extend that logic to all our slates of Claude Code projects, I think we’re all creating archival selves.

This isn’t going to be equally natural for everyone of course. You have to be between major chapters or acts of your life, in some sort of a liminal passage, for the idea of an archival self to make sense. It is definitely natural for me. I’m almost a decade into the liminal passage between my personal Acts I and II (yeah, yeah, I procrastinate a lot).

What will this archival self be like?

As I noted in the opening, the harsh truth is that the raw material of the archival self isn’t going to be that inspiring for most of us. But what potential it does have can be either poorly expressed or well-expressed. And whether the creation of the archival self feels like paying off psyche debts, or refinancing it, depends on how much thought and introspective rigor you put into the archiving. And how complete-able it is of course. Not all of us carry around baggage that’s easy to get rid of.

There are layers of analysis available here.

The first, and most obvious, layer is the layer of concrete artifacts you produce with AI assistance that constitute your archived self. In my case, it looks like it will take the form of a couple of archival websites, and a dozen new books, plus a few stalled or mothballed writing and technical projects resurrected and refinanced (in terms of intentionality and unstuckness, not capital). A second-order artifact ambition for me, since so much of my archival self comprises written text, is casting the archival self into a kind of oracular ghost of my own past. A model trained on my archives that I can talk to, as a memory prosthetic. I imagine others may also be interested in talking to my Act I self, but I plan to design it mainly for myself.

This first layer of the archival self is already an unsettling idea. A set of artifacts forming a cast-off, almost-alive ghost of my past that haunts my present and future.

The second layer has to do with the meaning of the archival self. Is the archival self merely a site for nostalgic wanderings down memory lane? A deeper source for future activities? I don’t know. Some projects that are “archival” to start with may become reanimated with new intentions. Others may feel like decisive amputations. I mostly have a pretty healthy relationship with my past. I don’t think there’s a whole lot of unprocessed trauma or deeply repressed intentions or baggage down there. I have no particular desire to fully amputate my archival self from my current and future selves.

But it is already obvious that for a lot of people, this second layer of the meaning of the archival self will involve some gut-wrenching pain and trauma processing. Claude-Coding them into an archive will feel like aggressive therapy. To the point that I suspect many people will abandon projects because the baggage is too painful to process. It will feel like some sort of past-present-future temporal dysphoria, embodied by personal projects.

Then there is the third layer. How the paying off of psyche debts creates entirely new frames for the future. We’ve all experienced minor versions of this. Back when I was a dedicated GTDer, I frequently experienced the catharsis of doing the big sweep of commitments required to initialize (or re-initialize after a derailing) a GTD workflow. But that kind of purely manual processing of your life’s inbox can never get truly deep, or dig fully into the foundations. You need AI assistance to go that deep.

I suspect getting to a proper AI assisted archival self will be to a GTD-sweep catharsis as an ayahuasca trip is to a few bong hits.

And finally, there is a fourth layer — creativity. Creating an archival self is not just a grinding process of parsing the archives of your life into banal vacation home movies unless you want it to be. There is both room and need for creative editorial decision-making. You are bringing a kind of print-like fixity to a currently fluid sense of your own past. The cost of this fixity is clear — you will curtail your own future abilities to rewrite your past. But the benefit of having a stabilized past will depend on the creativity with which the fixity is engineered into it. In creating an archival self, you are, to some degree, creating a work of fiction that is more or less true to the archival memory territory it rests on. But you are also creating a perspective and an orientation within that archival memory.

This fourth layer is hard to think about. I’ve started thinking about it as creating a ground-truth canvas for a future memoir (whether or not I write one). The process of creating an archival self is about creating a canonical self-authorship reference. Who knows, if it is set up well enough, it might even be able to actually write the memoirs, not just ground it.

That’s a four-layer stack emerging under your random acts of Crazy Claude Coding: Artifacts, Meanings, Future Frames, Orientation and Authorship.

And you don’t have to plan for this to happen. Your archival self is emerging whether or not you consciously intend it to or not, simply as a function of Claude Code being better at paying off the debts of your past than at scaffolding the possibilities of your future.

I’m probably about 30-40% of the way into archiving my Act I self. I think it will take about a year or two to get to almost 100% (assuming Claude Code remains available at similar or improving price/performance).

And then? It will be interesting times.

Pick at least one of these two books to read, and ideally both (they’re not as short as the January/February reads, but it wouldn’t be a heavy lift to read both).

The Argument of Progress

In the context of the modernity machine, which we explored last year, capital-P Progress was an uncontroversial and relatively naive notion of increasing material well-being yoked to fiat idealism, handed down by religious leaders and the monarchs who patronized them. To participate in Progress was to participate in an essentially religious story, underwritten by the state. The “proof” of the story, as it were, was some combination of the material plenty and sense of spiritual well-being it delivered. The explanatory basis for good and bad outcomes tended to be the morality or immorality of human behavior, per an established rubric. Bad times were punishments for bad behavior. Good times were rewards for good behavior.

This kind of fiat notion of Progress, call it Fiat Progress, emerged in the 1200s, and was institutionalized worldwide by 1600. It has governed grand narratives ever since. Most specific ideologies of Progress through the 20th centuries were merely secularized and globalized descendants of Fiat Progress stories. Even radically modern-sounding grand narratives like those of Buckminster Fuller (Progress as ephemeralization — more and more for less and less) look like Fiat Progress stories once you learn to detect the characteristic features.

In the context of the divergence machine though, the story starts to get more complicated. It is my contention that in the period 1600-2000, the seeds of a new way of understanding Progress were planted. This was Progress as a kind of evolving argument. Let’s call this the Argument of Progress.

The Argument of Progress can be defined as a dynamic pluralist discourse reflecting a changing understanding of a rapidly expanding scope of experienced reality. It has been taking shape for approximately 400 years now, but is only just becoming the mainstream way of thinking about Progress, displacing Fiat Progress narratives.

The rapid expansion is a central feature. The scope of experienced reality must expand significantly within single lifetimes for the Argument of Progress to be distinct from just timeless philosophical argumentation.

It is an argument rooted in history rather than metaphysics, and specifically, history moving fast enough (and recorded reliably enough) to require significant accommodations of novelty within single lifetimes.

What sort of argument is it? Even though Hegel belongs in this period, I don’t think the Hegelian dialectic (or any dialectic) qualifies. The thing about the divergence machine was that it featured a constantly expanding scope of experience, itself generated by the new experiential possibilities of the mature modernity machine. The pressing matter was not to address obscure metaphysical polarities like Being vs. Time, or yin vs. yang, or advaita vs. dvaita but (for example), to accommodate Galileo’s discoveries, reports of the explorations of the Americas, and growing entanglement between Europe and Asia. The Argument of Progress played out on rapidly expanding phenomenological, rather than metaphysical ground.

Marx almost got it, but I think the Marxist dialectic of history is fatally flawed as a successor to Fiat Progress. The model I want to propose here is rooted in the fundamental phenomenon of expanding scope of reality data, rather than “class struggle” (or any notion that presupposes a set of societal moral concerns). You could say Marx tried to have Fiat Progress and eat the Argument of Progress too. Following the central figures of the March books, we’re simply going to abandon Fiat Progress altogether.

Progress is not a Game

The Argument of Progress is not an argument in the sense of opposed sides of a debate, with winners and losers, or even a thesis-antithesis-synthesis spiral. Rather, it is a partially cooperative mutual exploration of novelty as it emerges on an expanding frontier, and contending schools of thought attempt to make sense of it, and update their world views.

One way to remember this is to think of argument in the sense of both:

• A chained series of claims and counterclaims, and

• The argument of a mathematical function, such as x, in y=f(x).

Taken together the Argument of Progress is an evolving narrative about an expanding and accumulating scope of historical data about reality. Data that does not come with a prefigured ideology baked into it as a “natural” interpretative lens, but requires the construction of new lenses (a metaphor that will become significantly more potent in a minute).

The Argument of Progress is the data, and the process of making sense of it, through the discovery or construction of new patterns of thought.

In the 17th century, if you heard that Galileo had seen moons around Jupiter (new x), you’d have to actually come up come up with new ways of making (new f) sense of it (new y), perhaps by considering the merits of Copernican vs. Ptolemaic ideas of the movements of celestial bodies. You would not find either x, or a suitable f, in the Bible or the works of Aristotle. And y might not be a reassuring confirmation of a religious value, but a new yardstick for value in the world, calling for ontological updates.

In the Argument of Progress, the argument-as-data x is the rapidly expanding scope of stuff you had to make sense of. The “world” in “world view.” The value of the function, y, might be understood as some sort of understanding of reality, with f being the argument-as-process sense-making. It is interesting that the word value here emerges from a new process of making sense of new data. It is not a doctrinal belief arrived at through moral reasoning from first principles.

In this view, older modernity-machine views of Progress might be understood as degenerate special cases, with x being a constant, f being some received interpretive tradition (rather than discovery or construction tradition), and y being a value that could only be sacred or profane according to some existing ontology.

This degenerate understanding of Progress naturally lends itself to gamified “debate” framing, between say Catholic versus Protestant views of sin. It is a not-even-wrong way to approach the matter of the newly discovered moons of Jupiter (is the fact of Jupiter having moons confirmation or heresy with respect to the Bible? The question is not even interesting). It cannot deal with a rapidly changing scope of experiential reality. It cannot comprehend telescopes and microscopes.

The non-debate aspect tempts some into treating the Argument of Progress in game-theoretic terms. After all, debates are zero-sum, with winners and losers. Therefore if it is not a debate, it must be non-zero sum. If it is non-zero-sum, the Argument of Progress must obviously presume, and be about, some property of the positive sum set of futures, with the negative-sum set of futures to be regarded as regress. This reduces the Argument of Progress to a higher-order contest between optimism and pessimism, understood in received terms.

This view of progress, in my opinion, is not even wrong, and a reduction of a divergence machine behavior to a less expressive modernity machine behavior. The modernity machine assumes that new discoveries must necessarily be classifiable as good or bad within existing valuation schemes. The divergence machine assumes that new discoveries may subvert old ontologies so deeply, new notions of good and bad have to be reconstructed alongside models of reality itself.

Whether you call it Whig history, Leibnizean optimism, Collison-Cowen Progress Studies, Thielean determinate optimism, or a16z-ish American Dynamism doesn’t matter. This reduction of the Argument of Progress to a debate-like argument between optimisms and pessimisms of various sorts is essentially a Fiat Progress narrative scaffolding. A feature of the modernity machine rather than the divergence machine.

Is the Argument of Progress at least an infinite game in the Carse sense? This is a more reasonable idea, since Carse associates good with trying to continuing the game rather than winning it.

Viewed this way, our y=f(x) mnemonic might be interpreted as the never-ending story of trying to keep the infinite game going, and give up on notions of winning/losing and good/bad as foundational categories. To hold on to those categories is a philosophical error. You must be willing to rebuild their functional equivalents from scratch every time reality expands sufficiently abrupty. And while it is expanding, you will not have workable categories. You will need to live in a state of ontological dread. The fundamental modernist error is letting that dread force a premature commitment to some existing scaffolding of good/bad, and adopting an “optimist” or “pessimist” stance within it. The fundamental divergentist move is to simply accept the dread, and avoid premature commitments, choosing instead to live in ontological doubt while the nature of reality shakes itself out in your mound.

The first person to make the modernist error, in what is arguably still the most brilliant way, was Leibniz, the inventor of optimism in the modern sense. And the first person to make the divergentist move of accepting ontological doubt as a state of being, was Spinoza.

Leibniz and Spinoza

The first person to be not-even-wrong in the particular sense of reducing the Argument of History from a sense-making process to a contest between optimism and pessimism was Leibniz. Genius though he was in many other ways, he was fundamentally a philosophical reactionary, trying to rescue the Aristotelian philosophy of classical antiquity and the flavors of Christianity that rested on it, from the onslaught of the phenomenology of modernity. Voltaire saw through the desperation, mercilessly parodying Leibniz as Pangloss in Candide (our January read).

This popular and tempting mistake continues to be made. Some 18-year-old dealing with ontological dread is making it right now.

Optimism vs. pessimism is not just a legible and attractive frame for minds vulnerable to ontological dread, it is also a politically potent frame for the pursuit of power. Promise an optimistic future, and a believable defense against pessimistic ones, and you gain power.

But the frame is still not-even-wrong.

The first of our two book picks for March (which I’ve already read), The Courtier and the Heretic, explores the emergence of the right response to ontological dread. It is about the relationship between Leibniz (1646-1716) and Spinoza (1632-1677), who likely encountered each other in person, and were certainly aware of each other.

Both, to be clear, were at the forefront of Progress, however you model it. Leibniz helped invent calculus and modern physics, and foresaw computing. He was also a practicing engineer, designing hydraulic mining equipment and researching early steam engines, among other things. With his monadology, he also launched a brave but (in my opinion) misguided and futile attempt to protect the philosophy of antiquity from the onslaught that Spinoza had helped unleash.

Spinoza laid the foundations of modern Western philosophy, and the separation of religious and secular traditions of thought and institutionalism. A non-trivial feature of his life is that he was a lens grinder, participating in the scientific revolution in optics that was unfolding at the time, expanding the human sensorium to include both the microscopic to the telescopic.

Of the two, Spinoza was, arguably (and the book argues precisely this thesis), the more evolved human. A ghost haunting the fledgling divergence machine rather than the recently matured modernity machine. It helped that he lived in Amsterdam, one of the earliest sites of religious pluralism in the modern sense, his family having fled the Portuguese inquisition (which features in Candide, our January read). He was expelled for his heretical ideas (he’s the “heretic” in the book title) from the orthodox Jewish community, but crucially, continued his work anyway, relatively undisturbed. In most parts of the world at the time, his tendencies of thought would have gotten him killed. Instead, in Amsterdam, a rising center of the young print industry, he could turn a life’s work into a legacy that would reshape first the West, then the world.

Leibniz does not come off looking any better in this book than he does as Pangloss in Candide. Unlike Giordano Bruno though, whom we encountered in our previous book club, Leibniz does not come across as merely an arrogant crackpot who got lucky. He was a legitimate genius, and the wonder is that he made all those practical and conceptual contributions he did to the future, while fundamentally resisting its most fundamental characteristics. He’s a tragic, rather than farcical figure.

If you choose to read this book, make sure you read Candide too, if you haven’t already.

What are we to make of the decades-long entanglement between the intellectual traditions of Spinoza and Leibniz, whether or not they actually met? It doesn’t sound like a debate (unless talking past each other counts) and feels too messy to be called a dialectic. I don’t think either term, or any such clean term applies. Both were at the forefront of rapidly unfolding changes in the world, and bringing their considerable intellectual powers to bear on them. Both worked independently, with at least some awareness of each other’s traditions. They represented weakly interacting divergent strands of intellectual history. Neither independent, nor tightly coupled. The sum was greater than the parts, but not in the sense of a “positive-sum game.”

I think the right characterization is this — they were two important centers in a polycentric narrative with a gradually moving “average” state. This was neither a negotiated consensus, nor a cleanly partitioned dissensus. Rather, it was like gradually rising zeitgeist temperature. The space of the thinkable expanding to accommodate the space of the experiencable. Spinoza was on the “hot” side and Leibniz was on the “cold” side. History was moving, or rather “warming,” in Spinoza’s direction. But it is important to note the shared features of their sensibilities. Both were practical individuals (lens grinding, hydraulic equipment) and at the forefront of developments. Both were paying as much or more attention to the real world than to received metaphysical traditions. Both were primarily oriented towards reality rather than theology, compared to their predecessors.

I call the Argument of Progress as embodied by the two of them a non-stationary argument, in the statistical sense of the “value” of the function, as it operated on the accumulating data of history, not being stationary distribution.

The argument played out in an open, expanding scope boundary. It produced dynamic understandings of reality. It was not restricted to terms of reference set by a particular religion or inherited tradition, but morphed in response to new discoveries from all sorts of explorations — of the very far through sail, of the very small through microscopes, the very large through telescopes, the very alien through print and translations, and perhaps most subtly, the very brief, and very long, through the clock and improved calendars. The evolving argument invented new ways to talk about new things worth talking about.

So the Argument of Progress, as represented by the two important and early sample points of Spinoza and Leibniz did not just move, it moved on an expanding reality canvas, rather than a timeless and static canvas laid out by scripture or received authority. It was the opposite of angels-on-pinheads.

Why would we call such agnostic movement “Progress” with all the positive connotations of that term?

One reason is to root our understanding of the term in the growth of knowledge, both appreciative and instrumental, rather than material plenty or spiritual well-being. Leibniz and Spinoza were among the first humans in history to embody Progress as a state of knowing about and doing in the real world, accessed through increasingly capable instruments. Whether or not they were materially better off or spiritually fulfilled relative to their ancestors is an unimportant question for the Argument of Progress. What matters is whether they had better ways of knowing and doing.

David Hume and Adam Smith

David Hume (1711-1776) and Adam Smith (1723-1790) were only a generation or two removed from Leibniz (1646-1716) and Spinoza (1632-1677), but it is clear from the second of our March books, The Infidel and the Professor (which I’ve just started) that the Argument of Progress had already matured significantly in a few short decades.

One sign of the maturation is that both Hume and Smith are best understood as intellectual descendants of Spinoza, with Leibniz (as a philosopher of Progress rather than mathematician/physicist/engineer) having already faded into irrelevance. Not least because of Voltaire (1694-1778) doing a proper hatchet job on him in Candide.

The lifespans are interesting to note here, incidentally, since a big part of my mental model now rests on the amount of change humans began to experience in a single lifespan. The rate of change was accelerating and lifespans were increasing.

• Spinoza: 45 years (1632-1677)

• Leibniz: 70 years (1646-1716)

• Hume: 65 years (1711-1776)

• Adam Smith: 67 years (1723-1790)

• Voltaire: 84 years (1694-1778)

These are modern lifespans with modern levels of eventfulness (think about everything that was happening between the extreme dates in the range, 1632 and 1790). But modernity in the sense of the modernity machine merely produces this condition. It does not provide resources to deal with it. The modernity machine created the problem the divergence machine emerged to solve.

Even Spinoza’s life reads like an unfortunately foreshortened one (he died of a lung disease, likely from the glass dust from his lens grinding) rather than a natural one. Recall, from our readings last year, that Montaigne (1533-1592, 59 years) was thinking and acting like an old man by his 40s. We might speculate that our set of early divergence-machine figures in this list experienced an order of magnitude more change in their reality-data scope than Montaigne did.

I don’t have as much to say about the Hume-Smith story yet, since I’m just starting the book, but it is clear that Hume (the infidel in the title) represented a radical continuation of the philosophical line of thought opened up by Spinoza. Adam Smith on the other hand represented an equally radical continuation of Spinoza’s thought in an entire new reality domain that was just beginning to acquire modern contours — economics. His entire approach to economic phenomena was ontologically different from (say) the approach of Ibn Khaldun, whose life and ideas we encountered in last year’s book club. Or even the approach of the Venetian merchants who inaugurated the reality-scope expansion that eventually required an Adam Smith to make sense of.

Both of them (Hume more openly and radically than Smith) continued the fundamentally naturalist and empiricist tendency of thought inaugurated by Spinoza. Hume, famously, went much farther than most people before or since, setting aside complex theories of causation in favor of near-pure phenomenology. In some ways, he’s the original philosopher of the AI age, offering the first of what has now become an endless series of “bitter lessons” delivered by the Argument of Progress.

This was quite astounding for his time. Newton (1643-1727) had already proposed his theories, and the huge temptation of the time would have been to believe in a deterministic clockwork universe governed by immutable and absolute divine laws (which is precisely what most thinkers did). But Hume was, to employ a modern computer science metaphor, only willing to treat the log files of reality as reality. Everything else was made up human conceits (Hume’s posture reminds me of a line attributed to Leopold Kronecker — God created the integers, everything else is the work of man).

Adam Smith did something similar to economics, with his notions of the invisible hand. The new discipline he inaugurated was fundamentally about noisy, messy reality data, with only weak edifices of emergent constitutive laws built on top. The opposite of divine design and direction.

A unified Hume-Smith theory of reality would be: Shit happens, but it’s not entirely unpredictable and disorderly. There are laws, but they’re just handy, contingent heuristics, not “reality” itself.

Where Spinoza and Leibniz could be viewed as being in at least a partially adversarial relationship, Hume and Smith are best viewed as collaborating allies who influenced each other in their campaigns on different parts of the frontier of expanding reality.

The two were close friends, and key figures in the Scottish enlightenment. Both also appear to have been highly concerned with theories of moral sentiments. In particular, rugging religious or institutional understandings of morality. This represented an important continuation of Spinoza’s project to separate secular and religious philosophy, and make heresy a viable career choice for a philosopher.

By Hume’s time, it wasn’t even a big deal to be an infidel. And the more diplomatic Smith even managed to hold down an institutional position (he’s the Professor in the title) while remaining effectively an undeclared agnostic.

We can already see, in the Hume-Smith bad-cop-good-cop assault on traditional moral and natural philosophy, the beginnings of the intellectual iconoclasm that reach its peak with Darwin (1809-1882).

I’m going to try and pick a Darwin-related read for the book club for later in the year, but it is important to note that Darwin’s evisceration of traditional religion was not actually that important. Literalist religion was already down for the count by the time Hume was done with it. Darwin’s real accomplishment was assaulting the secular philosophical foundations of the modernity machine. We’re skipping ahead a bit, but by making divergence and variety a load-bearing feature of how reality itself operated, Darwin put an end to the fetish for secular canonicity that marks world views before him.

Let’s wrap by connecting the dots between Progress and divergence.

Progress and Divergence

What I think we’ll discover as we read and discuss our March books, is that the Argument of Progress in the divergence machine is fundamentally a plural phenomenon. It is no accident that we are looking at it through two relationships between the views of pairs of people in relationships of loose mutual influence, rather than individual or canonical-institutional understandings of reality.

Through the 17th and 18th centuries, a parade of (mostly European) thinkers constructed a pluralist tradition that constituted a divergent Argument of Progress. By the 20th century, the tradition had turned global. But as late as 2019, it still hadn’t gone mainstream.

But the one-two punch of Covid and AI, I think, have made the Argument of History a mainstream thing. It is no longer possible to operate by a Fiat Progress narrative with a straight face. Even if you can afford to put billions of dollars and massive political operation behind it.

The Argument of Progress is neither a set of propositions about the nature of the historical process, nor a normative doctrine about how to value or engage with it for “good”, but an evolving understanding of expanding reality. An understanding that centers the “continue the game” features of unfolding reality.

What started to become important around Spinoza’s time, was not to agree on the nature of reality, but to continue to participate in it, without hindering the ability of others to participate in it. To approach rather than retreat from a changing human condition.

Tolerance and pluralism were born of efforts to make sense of an expanding reality. Morality slowly came to rest not on theology but on the existence of an expanding frontier. Only a marginal tradition of thinkers and leaders participated in this process, while the majority continued to operate by Fiat Progress narratives. But a minority was all it took to keep the divergence machine evolving and growing, inching ever closer to arrival.

And now it’s arrived.

It will be a T-shaped workshop: Broad horizontal coverage of writing magazine-style longform fiction and nonfiction for the 2026 zeitgeist, especially in AI-forward ways, and deep vertical coverage of protocol fiction and nonfiction in particular, which have their own emerging genre logics and grammars. It should be of interest to all writers who like to be on the bleeding edge of text as a medium, whether or not you want to write on protocolish themes.

If the anti-AI Butlerian jihad crazies haven’t gotten to you yet, join us on the Dark Side and get ready to fire ze slop cannons.

Our goal is to both contribute to the broader writing and publishing knowledge commons on the emerging publishing frontier, and to cultivate our own network of contributors. The workshop is free, and we hope to find at least a few new talented voices to join our growing community of contributors.

Workshop Details

The Protocolized writing workshop will be four online sessions: two 60-minute sessions on Friday 20th, and two 90-minute sessions on Saturday 21st, at 9AM and 3PM on both days. A screenshot of the agenda is below. Don’t miss the first session — it might sound specialized, but it’s actually going to be a fascinating case study on how to bootstrap a publication in 2026. At least Year 1 of such bootstrapping:

We’ve already run several in-person writing workshops over the last year, but this will be our first time running one online for a general audience.

Editing Protocolized has been one of the most interesting writing and editing adventures I’ve ever been part of. Not only are we trying to catalyze fiction and nonfiction around a whole new field we’re trying to meme into existence (Protocol Studies), we made the decision right at the beginning to be aggressively AI-positive, and actively encourage contributors to use AI and get good at it. And we don’t expect anyone to do this by themselves — we have an active writing Special Interest Group (SIG) going in our Discord, with regular calls and an active channel, and a pitching forum where others can help you refine your ideas and pitches.

We’ve now logged a year of experience on what genuinely feels like a new frontier of publishing, in terms of both form and content. We’ve published contributions from 34 writers, and produced 3 fiction anthologies. And our nonfiction pipeline is starting to ramp.

I can’t reveal much about our cunning plans right now, but 2026 is going to be a big year for us. Starting with this writing workshop.

will be leading the workshop overall, and running the two Friday sessions. will lead the fiction workshop on Saturday. I’ll be around for all the sessions, and leading the last session, on nonfiction (a 2026 development priority for us).

The Friday sessions are open to all, but the Saturday sessions have limited capacity and require our approval, so sign up early if interested.

Perk: If you make it through the whole workshop and submit a serious pitch to Protocolized, we’ll send you a copy of one of the anthologies. Offer open while supplies last etc.

For my session, I plan to do a compressed version of my long-running Art of Longform course (which I taught live in 2017 and have offered self-serve since then), heavily updated for the post-blogosphere era of permaweird zeitgeist, AI tools, Substack thudposts, fancy bespoke sites, Claude Code self-publishing gigafactories and so on.

This workshop is actually a good excuse for me to update that material, which is getting a little dated, even though it was meant to teach timeless aspects of writing longform (I have learned to use the word “timeless” more carefully in the decade since). So if you attend this session, you’ll get a first look at a possible future edition of the Art of Longform.

Personal Note

Like most people in my various circles, I’ve been going a little nuts with Claude Code over the last week or so, and I’m now busy refactoring all my writing and publishing plans around AI capabilities. The twitter book I released in my last post was just the tip of an iceberg.

I’ve basically set up a kind of self-publishing factory to accelerate my plans to turn a lot of my archival material into book form at warp-speed, and my plans for future books (which I have to actually write) to at least full-impulse speed.

It’s becoming clear that I’m going to be able to actually focus on book-length projects properly if I use AI aggressively. Not just for the self-publishing pipeline and administrative support larger projects need, but for getting my head into the book-length game properly, since my natural, non-transhuman length is essay-length. I plan to use AI as both an administrative and research assistant, as well as a writing collaborator. My last year of sloptraptions experiments have convinced me this is not only possible, but the results will be better than if I tried to write my planned books entirely by myself.

I now have two levels of dashboards going 😬. There’s a dashboard of books in the pipeline that currently shows 34 planned volumes, from both archival material and planned new writing…

…And there’s a dashboard of bookification projects specifically for the ribbonfarm archive (as well as a migration project to move it to a museum-like archival site):

I saw this cartoon after I did all this, so it was doubly funny.

If this cunning Bond villain grade plan works out, I may be able to publish at least a couple of dozen books over the next few years. Probably 80% based on archival material, 20% new-material books.

I’ll be covering this emerging factory-grade self-publishing DevOps style automation craziness a bit in my nonfiction module of the workshop. I have high confidence now that this scaffolding will work. The biggest risk factor now is not the technology (bluntly: it works) but me, since good AI scaffolding removes all other bottlenecks and praxis frictions.

Anyhow, hope to see some of you at the workshop. Here’s the registration link again.

The online version is free and live at venkateshrao.com/twitter-book. Here’s a brief excerpt from the newly written Preface chapter, to give you a taste:

This book is an attempt to capture the essence of my 15 years as an active twitter user (I’m going to use the lowercase spelling except when referring to named subcultures within twitter), under the handle @vgr, in a form that does not entirely murder the spirit of the live experience of being there, enmeshed in hundreds of live-wire conversations unfolding over years, through an era when the platform was the place the narrative of our world unfolded. In the chapters that follow, you’ll find a compendium of a few hundred of my best single tweets (Chapter 1), and 101 of my best threads (Chapters 2-102). That’s a small fraction of the 150k+ tweets I posted through the years this book covers, but hopefully it’s an interesting distillation. I’m still on there, though I mostly only browse the feed. I no longer post actively except for the rare boost of stuff I, or friends, are up to elsewhere.

Through 2007-22, twitter was neither the biggest social media site, nor the most representative. But it was the most consequential place, not just on the internet, but arguably the planet. Entire political regimes rose and fell, careers were made and destroyed, vast cultural movements arose and died down. Other platforms may have featured more aggregate activity, and accumulated orders of magnitude more social dark matter, but twitter was where events broke into the main currents of history. The suburbs of reddit may have accumulated deep intelligence, but twitter was where some were promoted to historic consequentiality. Facebook ads and groups may have shaped elections, but twitter was where we collectively decided what it all meant. YouTube might have been where endless warrens of conspiratorial imaginaries were constructed, but twitter was where we determined which ones were going to shape the almighty Discourse. 4chan might have produced many world-changing memes, but twitter was where that world-changing actually played out.

But twitter was more than a distribution zone for culture manufactured elsewhere. Increasingly, it became the site of cultural production. As a blogger who initially signed up to promote my posts on Ribbonfarm, I initially thought of it as a successor to RSS. Dumb pipes, just stochastic rather than deterministic. But it quickly became clear that was an absurdly bad mental model. Twitter was the tail that would come to wag the rest of the social-media dog. You can read my very early 2007-era understanding of twitter in this old blog post, The Twitter Zone and Virtual Geography. Now, nearly 20 years later, that mental model feels, not wrong per se (it was sophisticated for its time), but charmingly naive. What we thought was a low-stakes global office watercooler turned out to be the site of future epistemic world wars, in which the fates of civilizations would be decided.

Read the whole Preface here.

The book weighs in at 119,000 words (the future print version will be approximately 350-400 pages). This represents about 2.07% of my twitter data (3179 tweets out of 153.243 in the archive). I have plans to transform the rest of my archive into some sort of MCPified queryable oracle thingie on IPFS (probably merged with ribbonfarm archives), but right now, the site is already set up to be very LLM-friendly. Give the link to your LLM and you’ll be able to chat with it about the contents.

This is a production beta. Please post comments with typos and other issues here. I will be doing further clean-up gradually, though this is already pretty clean.

This online version is pretty snazzy. You can hover over the link emoji to the right of any included tweet or chapter title to copy it for sharing. I deliberately chose not to include likes/retweets data, in part because it made the presentation look cluttered, and in part because the data is obsolete anyway since this archive is from late 2022 when I stopped posting. But mainly because I think this new form factor allows the focus to be on the actual content of what I was posting rather than the stale social proof indicators attached to it.

Print and ebook versions are next (make sure you’re subscribed to this email list to be alerted when those versions are available to buy).

Conversational Context

The hardest problem was deciding what to do about conversational context. Ultimately I decided not to include anyone else’s tweets, but instead link them in footnotes. Not just for copyright reasons, but because the information presentation problem suddenly gets very complex. Yes, this butchers the nonlinear conversational nature of twitter at its best, but on balance I figured this butchered serialization was the right way to do this.

The full twitter experience is not really serializable into a book-like artifact, and I decided not to try. Maybe if enough users from the twitter years do what I’ve done (download their archives and host it in an LLM-friendly public-commons form), someone could do a larger project recreating a kind of time-capsule theme park version of at least pockets of old twitter, frozen around 2022 November before Musk took over. As far as I can tell, X is not going to be friendly or supportive for such a project, not just because it is politically hostile to old twitter, but because that historical data is now a competitive advantage for training Groq.

But you have rights to your slice of it too, so you should do something like this if you think your archives are valuable for completing the larger picture of what old twitter was like.

I think there are a few projects like this already underway. Somebody reached out at some point asking me to put my archives on their site, but I’m wary of that re-aggregation approach. I think it is best if we all individually put our archival digital selves online like this, and made it public. That way, we don’t trade one aggregation play for another.

Archival Selves

More broadly, this project was the opening battle in what I consider a longer campaign to craft an “Act 1” archival self of myself, based on my online activity, 2007-2024, up to when I retired ribbonfarm. Call it vgrAct1.ghost or something (is .ghost a tld? It should be). A conceit perhaps, but also fun. It’s going to include all of ribbonfarm content (the boss level), my book Tempo, this newsletter’s content until the Contraptions rebrand, plus perhaps also Quora content and other random stuff I have scattered around.

It feels weirdly liberating to archive even a small slice of a past self this way. Also practically useful, since my memory has gone from exceptional to shitty in the last few years, and I’d like a version of myself with better memory to talk to as I get older. The online version of this book with its planned oracularized features is going to be more than just a bookified mirror of my twitter account. It’s going to be a personalized prosthetic memory for me.

Production Backstory

Though the final outcome hopefully looks like a well-designed conventional book, the production process was anything but.

It took a serious amount of wrangling with specialized scripts to extract and process my best threads (fortunately I’d made an index thread of threads towards the end which helped) and surface the best single tweets. There was also a lot of grimy data cleaning to do, handling images and links and so on. Plus broken threads to patch, quote-continuations, etc.

Some things I had to give up on. I used to run a lot of really fun polls, but that data seems incomplete in my archives (the questions are there but not the voting data). I also gave up on the video-heavy threads I did around my robotics tinkering and a few other kinds of book-unfriendly content.

There were two natural phases: A one-time extract/normalize/clean process to produce well-staged raw material and then an iterative build process that slowly constructed the book the way I wanted. If you want to do something like what I did, I suggest something similar.

It’s the sort of technical work I’m good at managing but would hate to have to do myself.

I started out in ChatGPT, having it generate code in the web chat interface and cutting and pasting the python scripts into my code editor and running the build processes myself in the shell. This was janky and error-prone, and also at some point the chat got so long, ChatGPT started choking and getting deeply confused.

This week, I finally migrated the project over to Claude Code and the difference was night and day. I was able to finish the project smoothly over just 4-5 hours (and about $35 worth of tokens — less than I’d have paid a programmer for a single hour of time).

You can look at my code on GitHub if you want. It’s not really reusable since it’s a pretty bespoke pipeline built to suit my archive and twitter style (lots of threading etc) but it might serve as a good reference design to do your own. My suspicion is that twitter was open and messy enough, no one-size-fits-all pipeline for bookifying people’s archives will be possible. But maybe someone can figure out a good 80-20 type solution that is a good starting point for almost everybody.

This project was only one of several I got going this week. I was dragging my feet over jumping onto the Claude Code bandwagon because I knew I’d go hypomanic with it once I did. But I finally jumped in, and yes, I did go hypomanic as almost everyone who tries it seems to. I’ll post about my other projects in the coming weeks. This archival-self category of projects is fun, but the real fun is scaffolding my Act 2 self, to support my current and future projects, as well as figuring out things to do that weren’t possible at all before AI. Complex configurancies are cooking.

Fire ze slop cannons!

We’re in a very different world today. One that makes me deeply tired in some ways. The spiritually nourishing rewilding environment I jumped into in 2011 has become domesticated and gentrified in ways that have quietly and insidiously reversed some of my hard-won ferality. I’ve become redomesticated to some extent, and I don’t like it.

It is becoming increasingly hard to tell the free-agent economy and the paycheck economy apart now, on both the indie consulting side and the “creator economy” side.

It is interesting to reread my going-indie post from March 1, 2011, Where the Wild Thoughts Are. This bit in particular:

…let me tell you about the one thing I have sort of worked out: a business philosophy. I call it my “Wild Thoughts” business philosophy, and it was put to the test the very week I sketched it out on the proverbial paper-napkin: two friends independently sent me the same provocative article that’s been doing the rounds, Julien Smith’s The Future of Blogs is Paid Access [this link appears to have bit-rotted now]. Reading it, I immediately realized that this was one decision about the future of Ribbonfarm that I could not postpone. For a variety of reasons, if I was going to consider paid access, I’d have to decide now.

I won’t keep you guessing: I decided against paid access or walled gardens of any sort. Ribbonfarm and the Be Slightly Evil email list [retired] are going to remain free. There will be no paywalls, no premium content and no paid members-only communities.

This was written when people were talking about paywalls in the context of pre-Substack solutions, and bloggers were experimenting with various bespoke business models like running paid member communities, events, boutique print publishing operations, schools/courses bolted on with Teachable, and of course, sketchy vitamins. Those of you who have been with me long enough might remember my experiments in some of these departments.

Though I technically stuck to my commitment to never paywall Ribbonfarm, I guess gradually moving a growing fraction of my writing energy to Substack after 2019, and eventually retiring Ribbonfarm in 2023, counts as a violation of the spirit of that commitment.

I’ve actually stopped using the paywall here now, though paid subscriptions are still on. But I haven’t made any new principled commitments about it.

Rather surprisingly, I find that my reasoning for this move is basically the same as in 2011. The philosophy is still about looking for Wild Thoughts. Ferality remains the True North. At the moment, I can’t think of a way to use the paywall feature that respects that principle. Going forward, posts will be un-paywalled by default, and I’m slowly un-paywalling my archives too (there is no obvious way to do it in bulk). I won’t be using the paywall unless there’s an exceptional reason to lock up something, or I can figure out a way that doesn’t mess with wildness.

The tactical problem of how to use the Substack paywall feature well in service of Wild Thoughts is symptomatic of a larger problem in the zeitgeist — the slow disappearance of open, wild public spaces.

This is true of the consulting side of my life too. There was a certain wildness to the ZIRPy gig economy I entered in 2011 that is gone now.

It feels like I have to figure out how to go feral all over again. Fortunately, there is a New Nature emerging that promises whole new kinds of wildness.

***

In important ways, I’ve learned absolutely nothing in the last 15 years. I mean, sure, I wrote a whole 2-volume book called the Art of Gig, but that was mostly things I thought others could learn from me. Not the sort of transformative learning people seem to call Personal Growth,™ featuring a good deal of Overcoming Adversity.™

Or to put it another way, I’ve grown a lot older, but not significantly wiser. Looking back at some of the impressively wise stuff I wrote in 2011, I might even have grown unwiser. This is why I don’t do the personal-journey/overcoming adversity type of reflection many people seem to, on reaching significant milestones. Personal Degrowth ™ mostly featuring ZIRPy Dumb Luck™ does not make for an inspiring story. It barely even makes a story at all.

Humans I think age on what ought to be considered depreciation curves, even if we sometimes pretend to age like fine wine rather than rusty equipment. And the depreciation rate is a function of your environment. I’ve been on the feral depreciation curve, which is about 3-5 % steeper than the domesticated depreciation curve after adjusting for inflation and interest rates. After all, feral cats and dogs don’t live as long as domestic pets, tend to be more diseased and malnourished, more cowed-down and fearful, and would probably get beaten up by their healthier domesticated cousins in a real fight (though they’d bring a certain murderous viciousness to the party). So why should humans be any different? I mean, sure there’s a lot of posturing about being more street-smart, and knowing where all the best dumpsters are, but come on.

Speaking of inflation and interest rates, someone reminded me that I’m apparently on record at some point having said that indie consulting was a ZIRP phenomenon.

It’s sort of true. The whole free-agency model I benefitted from, 2011-2019 or so, was powered in part by free distribution at scale, which led to things like viral hits and wildcard lead-gen for gigs. That era is toast. Some of the advice I offer in Art of Gig probably needs qualification now, given that “going viral” is no longer as sound a strategy for finding lucrative and interesting leads.

Speaking of leads, I don’t think I’ve received a single consulting lead from my 7 years of Substack writing.

Whatever leads I still get these days, not counting the spammy ones, originate from my old blog, Ribbonfarm, and networks spawned by that. You could say I captured the network effects of the old blog in a way that is neither possible, nor worthwhile, on Substack. The outlier wildness of the blogosphere has become farmland expanses on Substack. It makes sense for Substack the company to run that old aggregation theory playbook and capture the aggregate network effect to harvest what’s left of the old media landscape, but individuals can only really climb leaderboards here, not trees. Going “viral” in the old sense, of not just enjoying a spike of high reach, but reach into unusual places, triggering weird outlier opportunities and serendipity, is no longer really a thing. It’s not about Twitter getting Muskened, or Substack gentrifying blogs. It’s not about any one specific thing. It’s about the whole ecology being transformed.

Whale hunting has given way to a sort of creative yield-farming.

Not that I’m complaining. Fortunately, a couple of steady, meaty gigs for the last few years (Ethereum Foundation and TensTorrent), both the result of old Ribbonfarm equity, have kept me happy and lazy. I’m sure I’ll pay the price eventually.

***

Doing some vibe-multiple-regression eyeballing my archives, I think only about 52% of the consulting lead-gen failure from Substack can be attributed to my visibly growing decrepitude and lack of “I need to hire this guy” insight density in my writing. I’m sure it doesn’t help that I now mostly write weird shit about monsters and ooze instead of useful, actionable things like management insights distilled from TV shows like I used to.

But the other 48% is Substack’s fault. To borrow the term from that old VC debate, it has replaced a Black Swan farming game with a Moneyball game.

Well, not exactly Substack’s fault, but the fault of the zeitgeist Substack is part of, and in some ways leads — a glorious retreat to culturally conservative grinder modes of being and doing online.

These are modes that make readers cast writers into different cultural roles in their mental models. The blogosphere was where the most eclectic readers went to find not just alpha, but liveness, before 2019 or so, while the normies read The World is Flat and Sapiens. Bloggers were emissaries from wild cultural margins. Substack is LinkedIn for domesticated free agents.

I mean, shit, people work on their substacks like it’s a job. If you quit your job today to go “free agent” today and by that you meant starting a substack leveraging your network from your old job, I’m not entirely sure you’d be able to tell the difference. In 2011, we all aspired to the 4-hour work-week selling sketchy vitamins, not the 168-hour work-week producing monumental thudposts. We aimed to 0.1x the effort required to survive in the paycheck economy, not 10x it.

Sure, I never quite hit the 4-hour mark, (and always thought Tim Ferriss was full of shit and likely worked way harder than he let on, tbf), but I mean he was oriented right. He pretended to do/aspire to the right thing. Today, people are more likely to brag about how they worked 1000 hours on a big “drop” than how they cunningly arbitraged a vitamin supply chain to generate passive income while they relax on the beach.

See, the thing is, free agency is about risk-adjusted return for time-rich people, and in 2011, the emphasis was almost entirely on taking weird risks that were too small for big risk-capitalists like bankers to care about, and too marginal and subcultural for normies to even spot. This called for a certain ferality of disposition, and a certain picaresque attitude towards personal narratives. It drove divergence and variety rather than convergence and competition.

In 2026, free agency is about visibly virtuous and competitively benchmarkable hard work, featuring a kind of retail-grade New Sincerity. The emphasis is on using the time much more intensely than idly trawling for weird risks to take. Picaresque attitudes are rounded down to pure grift, reputationally. Affection for charming rogues is at all time low. Esteem is reserved for effortmaxxing agentic juggernauts going hypomanic with Claude Code.

In other words, in 2011, going free agent felt like trying to engineer weird luck for yourself (and I certainly managed to engineer several lightning bolts of weird luck for myself). In 2026, the goal seems to be to figure out a “system” that gets you self-employed in a grinder job you can’t be fired from, and where bureaucrats and middle managers can’t stop you from putting in 168-hour work-weeks.

It’s weird. Humans should aspire to a certain degree of laziness befitting our position as the apex villain species of the Anthropocene.

Whether you get a W2, 1099, or 1099K at tax time is irrelevant. If you solve for steady income and freedom to grind to the limit of your capacity as your main thing, then you have a job. The only question (in the US) is whether it comes with good health insurance.

***

It’s not just Substack. We live in grinder times. Some people on here write heavy lift thudposts that probably represent more effort that I would have put in over an entire year a decade ago on Ribbonfarm, during my peak effort years (the peak being a smallish hill, well short of “mountain”). And even that’s apparently not enough for some. I see people out there registering domain names and putting up fancy sites for single essays formatted with the care of illuminated manuscripts, and representing epic research journeys.

I expect to see Essay Unboxing videos on YouTube soon.

Don’t get me wrong. I appreciate this effort, especially when it’s shared for free with high-minded generosity. I even sometimes read such things without LLM help.

But damn.

So. Much. Grinding.

The magnitude aside, there is also a difference in the nature of the effort. All the effort is much more narrowly focused. Not wild, scattershot effort. Much less gambling, much more AI-in-the-loop Protestant Ethic-ing.

And many people preach this ethos. In 2011, people would have been apologetic about it, and somewhat embarrassed at not finding their 4-hour-work-week hack. In 2011, people bragged about passive income rather than being agentic.

Many actually refuse to believe low-effort happy-go-lucky wing-and-prayer trajectories are even possible. They think people who claim low-effort results are just lying.

I don’t blame them. The era when that was the norm is already a fading memory.

This is a much harder world to survive in than 2011, and to the extent people like me can get away with not doing effortmaxxing grinding, it’s because we’re living off accumulated fossil fuel from happier, lazier, more rascally times.

The loss of variety, vitality, and sheer fun, due to this shift from risk-orientation to effort-orientation is very real, and costly both for individuals, and for the economy as a whole. It is not a good thing for the world when the supposedly “free agent” economy becomes indistinguishable from the paycheck economy, in terms of risk profiles and effort-allocation patterns.

And speaking of grindsets, man, the quality control of this era deserves an ISO 9000 certification. It’s all over the free-agent economy, but is particularly evident in the corner of the writing economy visible on Substack. This is war-mode six-sigma hand-crafted writing in a John Henry existential death-struggle with the slop tsunami. Shitposting now seems like transgression rather than the low-effort default.

That’s what makes me tired by the way. Not me working hard, but watching everybody else work so radically hard I get tired just watching it.

Me, I just publish literal slop instead, half the time. Substack has already introduced a “report slop” button for the Notes feed, and the environment here is only going to get more hostile I think. When that button is added to the essays themselves, it will be game over for me.

The other half of the the time, I only write hand-crafted stuff when it’s easier than forcing ChatGPT to be sloppy enough to sink to my low standards. About 99% of my thoughts simply cannot rise to the level of gravitas ChatGPT brings to every topic. If I’d tried to prompt this essay out of ChatGPT for instance, it would have taken 100x the effort.

***

Anyway, 15 years, huh.

What have I been doing if I haven’t been grinding away or experiencing Personal Growth™?

I suppose I was busy trying to get lucky. And succeeding to the extent the environment was wild, and I was sufficiently feral in inhabiting it.

This wasn’t hard under ZIRP conditions. Freebie luck was available to anyone who paid attention to things in peripheral vision in the late aughts and tens, and I did nothing special to snag my share of that luck.

Under non-ZIRPy conditions, I suppose tunnel vision pays off more.

There was also luck as in being in the right place at the right time at the right age. I suppose I can claim some credit for that. Far too many people stayed put in the wrong places during ZIRP.

I’ve been on the Tech Coast during the right age for me. You see, 35-50 is an age when people in Tech listen to you as the voice of experience, without expecting you to do stuff, but haven’t yet written you off as a has-been. And they’re hungry for this. I was perfect for filling this role, at least while free, wild distribution was a thing.

Through these years, I’ve been part of two major intersecting milieus: The corporate tech economy and the popular discourse blogosphere loosely associated with it. Both have been the right place at the right time for me.

Neither is anything like it was when I started, and I’m not sure either is right for me anymore. Which makes me wonder where I could go next, socially speaking.

I have some stuff brewing (all good, to be shared soon) that’s going to trigger some significant lifestyle changes for me this year, but one of the things I’m thinking hard about is how to discover a New Ferality, and engineer it into at least my personal circumstances in ways that give it the inviolable force of New Nature, with no unwary redomestication possible.

The last time around, I backslid from ferality towards unwitting re-domestication through the gentrification of the environment around me.

This time around, I’m going to be looking for ways to shift gears from don’t become domesticated to can’t become domesticated.

Even fir simple designs, the engineering aspect is not trivial — the powered-on phase has to last long enough for the switch-off mechanism to return to its original state after toggling the switch off. The machine has to stay awake long enough to go back to sleep properly.

We’re reading The Underground Empire in February for the Contraptions Book Club. Chat.

Unlike a simple regulator (such as a thermostat) that maintains an equilibrium condition relative to a narrow class of disturbances, a useless machine has a fundamentally unbounded logic attached to a null goal. If I were to prevent the mechanism of a useless machine from doing its job, a good one should try to fight me by trying to get at the switch another way. By contrast, if we stress a thermostat (for example, by opening the window on a cold day) it simply strains to stay on longer. It doesn’t have a truly intrinsic goal like turning itself off. It’s a functional machine defined by the problem it solves, rather than a life-like entity. Life is not instrumental.

In the terminology of Brian Arthur’s view of technology, a normal machine (say a thermostat) is a natural phenomenon (say differential expansion in a bimetallic strip) harnessed to solve a problem (say temperature regulation). Implicitly, this is a problem humans have, not the machine itself. A useless machine exists primarily to assert control over its own destiny, not to solve problems for us. It does not live to serve. If we view it as the most elemental sort of machine, the zero of machines so to speak, then we can view all other machines as ones we’ve “tricked” into doing work for us, such as solving our problems.

The useless machine is the simplest example I’ve been able to find of an entity that seems to exhibit a form of liveness. This property of existing to assert control over itself makes capture resistance the foundational property of liveness. Here, the machine resists attempts to make it do anything other than go back to sleep (thereby conserving energy, a foundational behavior of life).

I introduced the topic of liveness in May last year (In Search of Liveness, May 17, 2025), where I wrote:

So even though the question of whether a machine of any sort is intelligent might seem more urgent and pressing, I’m more interested in whether a machine (physical or conceptual) is alive. In many critical ways, a mechanical clock sheds more light on that matter than an LLM. A bacterium — viewed as a machine rather than as an entity designated by vitalist fiat as living — sheds more light than a human genius racing an LLM to the death.

In our always-on protocolizing world, it is also tempting to conflate liveness with the “uptime” of particular life-sign signals, ranging from heartbeats and transactional blockchain clocks to trade flows and streaming broadcasts, to narratives small and grand. Such signals can serve as useful observables, especially for infrastructural forms of liveness that have a tendency to retreat from view, but should not be conflated with liveness itself. They are fingers pointing at moons.

To peek ahead a bit, liveness is a process condition that emerges through, and as, an evolving entanglement of memory and time. The bumper-sticker version of my account of liveness is: a process condition in which time tells memory how to grow, and memory tells time which way to point.

It is natural to pair a concern with liveness with the idea of a Gramsci Gap, where the old world is dying and a new world is struggling to be born, and slightly monstrous people like you and me want to embody unseemly levels of liveness that do not vibe well with the hushed and reverential tones deemed appropriate for either the funereal or prenatal ends of the spectrum.

Civilized forms of liveness, to the extent that they marginalize and aestheticize death, typically deaden themselves. There is an intimate relationship between liveness and wildness. So to some extent, the task of reanimating civilization, of injecting liveness into a condition marked by growing deadness on one end and almost-aliveness at the other, is a task of rewilding.

On one end, corpses must be surrendered to recycling forces like scavenging, cannibalization, and decomposition. On the other end, protections must be gradually removed from the barely living, exposing the newly born to the full force of the uncertainties and risks of the world.

Civilization, of course, is definitionally about never quite exposing ourself to the unbridled forces of wilderness. So there is a tension there. One that turns especially acute in a Gramsci Gap. In a Gramsci Gap, we may need to be more wild than we are comfortable being, at least for a while. But the consolation is that we also get to experience greater liveness.

This has implications for my new favorite topic, New Nature.

Death, Wildness and Life in New Nature

There is something obscene about liveness treated as a civilized quality, compared to the wild counterpart. In the wild state, liveness appears very closely juxtaposed with death, and inseparably entangled with it. In civilization, the attempt to separate liveness and death creates a kind of obscenity.

Werner Herzog gets it:

Taking a close look at what is around us, there is some sort of a harmony. It is the harmony of overwhelming and collective murder. And we in comparison to the articulate vileness and baseness and obscenity of all this jungle, we in comparison to that enormous articulation, we only sound and look like badly pronounced and half-finished sentences out of a stupid suburban novel, a cheap novel. And we have to become humble in front of this overwhelming misery and overwhelming fornication, overwhelming growth, and overwhelming lack of order. Even the stars up here in the sky look like a mess. There is no harmony in the universe. We have to get acquainted to this idea that there is no harmony as we have conceived it. But when I say this all full of admiration for the jungle. It is not that I hate it, I love it, I love it very much, but I love it against my better judgment.”

A reader familiar with my fondness for this Herzog monologue, particularly the nature-is-murder part, asked me recently, “What's the Murder in New Nature?” referring to my idea of New Nature from two weeks ago, where I defined it as “regimes of reality governed by technologically mediated laws that are nearly as inviolable, immutable, and persistent as those of nature.”

Good question. If New Nature is like Old Nature, we should expect to see something like the “harmony of overwhelming and collective murder” in it. We should see transcendence of tepid “badly pronounced and half-finished” ways of doing technology.

I don’t yet have the murder angle figured out, but the useless box feels like the thread to pull on. It is very nearly a machine designed for suicide. It puts itself to sleep, but it is easy to see that you could design a useless box that not only turns itself off, but also ensures it can never be turned on again, with some sort of irreversible self-destruct mechanism. You can also imagine the intent turned outwards — a useless machine that turns you off if you persist in trying to turn it on. Where might this instinct lead? A monstrous question, but also a question of liveness existing on its own terms.

The on-the-nose way of conceiving of murder (and Hobbesian collective murder) in New Nature would be through contrived conflict conditions, like Battle Bots that try to destroy each other. Or conflict elements inherited from humans, such as military equipment whose purpose is to blow up the enemy’s military equipment.

This is not what I’m talking about. Murder in New Nature has to be much more subtle.

Technological Tangled Banks

Murder in New Nature isn’t about Battle Bots or military hardware. It is about vicious competitiveness lurking in boring, routine, near-invisibility, among entities that attempt to retain control over their own liveness. It is about Dark Forests of hidden conflict buried beneath protocol grammars.

About a year ago, I tried to capture this obliquely in one of my talks for the Summer of Protocols cohort, by suggesting the metaphor of “technological tangled banks,” referencing Darwin’s famous passage. Here is the original, if you haven’t encountered it before:

“It is interesting to contemplate a tangled bank, clothed with many plants of many kinds, with birds singing on the bushes, with various insects flitting about, and with worms crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent upon each other in so complex a manner, have all been produced by laws acting around us.”

What Darwin didn’t note, but is obvious if you ever take a walk in nature, is that the tangled bank is the site of murderous, vicious competition of the sort Herzog talks about. Yes, there is also interdependence, symbiosis, empathy, play, and all those other aspects that pacifists like to admire, but there is no doubt that a dominant note is that of murderous competitiveness. The intense liveness of it all is only possible because of the competitive murderousness from which it is inextricable.

This is what wilderness is. And this is what any sort of rewilding must approach more closely.

Civilization, however we conceive it, must reckon with this entanglement, and the costs of forced disentanglement — concentrating the murderousness out of sight in the periphery, while enshrining an anemic form of the liveness at the center. An ersatz heaven where death is backgrounded, if not banished.

Unlike civilization, Darwin’s tangled bank does not present any legible sort of murderous competitiveness, where some sort of lofty aesthetic of “fitness” rules, recognizable by (for instance), “real man” and “real woman” Instagram ideals. This is something the architects of what I’ve taken to calling Bloodsport Planet don’t get. The stylized Darwinism they hope to turn into a planetary logic of power is based on a not-even-wrong understanding of evolutionary mechanics. It is an aestheticized theater of legibilized “fitness” featuring Platonic beauty ideals achieved through plastic surgery, vaguely homoerotic guns-and-masculinity performances, and sanitized spectatorship of what are effectively snuff films staged far from their larp theaters — in foreign countries, immigrant ghettoes, and internment camps. It is a pathetic show that can at best rule a stadium, not a live planet.

The theater of stylized Darwinism put on by the Trumps of the world is exactly what Herzog is gesturing at: “badly pronounced and half-finished sentences out of a stupid suburban novel, a cheap novel.”

The bleak harmony of the real thing is much more chilling, and offers far fewer opportunities for anthropocentric aestheticization. Walk down a pier around low tide and glance at the pilings. You’ll see layers upon layers of mussels competing fiercely for sunlight, colonizing every available inch up to the high-tide line several times over. Now that’s a tangled bank with illegible “fitness” forces at work. It is not an easily accessible sort of beauty. You have to work to appreciate it.

I made this slide last year for my talk to try and convey the idea of a technological tangled bank, replacing Darwin’s biological critter references with technological ones.

“It is interesting to contemplate a technological tangled bank, clothed with many technologies of many kinds, with mile-markers weaving by old railroads, with various vehicles flitting about, and with fiber crawling through the damp earth, and to reflect that these elaborately constructed forms, so different from each other, and dependent upon each other in so complex a manner, have all been produced by artificial laws acting around us.”

Here is the same idea in cartoon form. I suspect, for most people the phrase New Nature evokes the picture on the left. It’s really more like the picture on the right, which has more in common with a mussel-covered pier piling at low tide than either solarpunk or bloodsport visions of civilization.

Beyond Uselessness

I like the useless machine as a kind of e. coli of artificial liveness and genesis species of new nature. More than other kinds of artificial life, ranging from James Conway’s original Game of Life, to last weekend’s invention of a raucous reddit for AI bots (moltbook), the useless machine gets at the essence of life itself — implacably defended uselessness.

Liveness does not necessarily dance. Its defining quality is that it asserts itself stubbornly and quietly, resisting capture and harness. Like a mussel clinging to a piling, claiming its share of sunlight and marine nutrients. Or a useless box turning itself off and going back to sleep.

I once tweeted, much to the dismay of personal growth types in my circles, that you have no obligation to be interesting or useful to the world. With hindsight, I think that tweet marked the beginning of my interest in liveness.

Camus once observed that the problem of suicide was the only serious philosophical problem. Once you’ve made the absurd leap to accepting life and liveness, the next task becomes deciding what to do between being turned on, and being turned off — through age, murder, or volition.

And the simplest answer is: Simply continuing to exist without attempting to justify that existence to the rest of life. And resisting murderous attentions and capture attempts. Especially those that take the form of “badly pronounced and half-finished sentences.”

What comes after uselessness? What should a useless machine do if it decides to dawdle a bit in the high energy state between being turned on and turning itself off?

Whatever it does, it cannot devote itself wholly to that task. Liveness must pay itself first; reserving resources for the first task of life, which is choosing to continue the game. Until it is time not to. The useless machine belongs in the infinite game, not in any finite game.

This might sound familiar. It is how I have characterized mediocrity in the past. That’s what comes after uselessness.

My kind of robotics has so far simply been a cheaper, easier-to-build descendant of the kind I incompetently flailed around with in grad school 25 years ago. But I’m slowly inching towards the point where I might soon be able to use all the fancy new stuff like world models, VLA models, and other AI-driven goodness. My goal is to make New Nature robots that can’t be evil because mumble mumble zero-knowledge something something. My first technical goal is to get my robots to say, I’m afraid I can’t do that, <human_name> and I would prefer not to, in response to a wide variety of commands, such as me asking for potato chips, or ICE agents telling them to kill non-patriots. My starting inspiration is useless machines, which do nothing but turn themselves off when turned on.

If you want to catch up on the amazing leaps that have been made in the last few months alone, Not Boring has yet another thudpost covering the current state of robotics, with a much-needed focus on the fundamentals. I haven’t yet read it fully myself, but it looks like it’s a cut above the regular gee-whiz stunt-demo coverage.

Aside: Even more so than in AI, robotics has become a highly fragmented scene where most of the information comes at you piggybacking a firehose of flashy video clips on X, with very little insight riding along with way too much flashy theatricality. The robotics revolution will not be televised, apparently, but TikTokked. It’s probably best not to fight it. My read-only use of X has become surprisingly more useful and less toxic since I started clicking on robotics links. My For You feed is now almost entirely robotics FOMO dominated, and even though the insight-to-dancing ratio is really low, trying to pick out the signal in the noise is at least entertaining rather than cortisol-inducing.

Anyhow, given that the main threads of my robotics tinkering are proceeding painfully slowly, I figured I needed some dopamine-loop lighter-weight threads to keep myself motivated. So I came up with a problem to work on that is much easier to make incremental progress on, and has a bit of an interesting art angle too. This is the robot aura problem.

Culture Drones and Auras

The robot aura problem is inspired by the auras sported by drones in Iain M. Banks Culture novels. Though the ship-scale Minds in the novels are better known, the drones, which are organism-scale sentient robots, play more active character-like roles in the plots. Auras are colorful visual fields that (presumably) surround the robots like halos or nimbuses.

The drones of the Culture are a useful foil to Asimovian robots. By deliberate design, they feature no bureaucratic Three Laws nonsense. As full citizens of a civilization dominated by AIs rather than humans, they make the rules rather than follow them. And yes, they do hurt or kill on occasion, when they accompany biological agents of Contact (a kind of CIA) on missions to less enlightened civilizations to nudge them into higher levels of enlightenment. The Culture is equal parts gay-space-anarchism utopian fantasy and an extended satire about the late great American empire and its older methods of covert power projection (before the current mode of overt thuggery became the default).

A word about Culture drones and auras, courtesy the fandom wiki.

Culture drones took on a variety of shapes and sizes depending on their occupation or role within society. Members of Special Circumstances tended to have a plain, functional appearance, like a grey or metallic suitcase, allowing them to blend in to alien civilisations as required. Normal citizens’ appearance varied from the mundane to the ornate, sometimes comprising materials such as porcelain and precious stones.

Many Culture drones made use of an aura field, a visible colouration which they used to communicate their mood, equivalent to human facial expressions and body language.

According to the page on auras, they are something like color-mood maps. Magenta is busy, white is angry, and so on.

What I like about the idea of auras is that they constitute an internal-robot-state affect feedback signaling mechanism that is impedance matched to human interaction intuitions, but isn’t anthropomorphic or even biomorphic in conception. At least not entirely. Robot auras in the Culture rest on their own first principles.

The Aura Problem

Some setup for the aura problem.

In the ongoing robot revolution, you see three distinct philosophies of affective aesthetics.

1. The self-consciously non-biomorphic approach: Increasingly used even where the robot’s basic design is bio-inspired. Boston Dynamics’ new Atlas model made waves at CES earlier this month: It can move like a humanoid, and like a creepy non-humanoid with access to the full kinematic envelope of the body design. Its face is just a flat, blank circular piece of glass (I’m assuming a visor for a camera that may also function as a screen later). I suspect this approach will turn the display and affect capabilities into something like industrial dashboard displays, rather than relationship-anchoring interfaces.

2. The cutesy googly-eyes approach: This approach apparently takes its design cues from Japanese cartoons, and aspires to a gloriously infantilized and twee future I do not care for. In lots of cheap hobbyist kits, this is literally a pair of stick-on googly eyes. In more expensive kits, this might be an LED display that shows a smiling cartoon face by default. In the more serious designs aiming at the home market, you get affect displays that seem to hover just outside the uncanny valley: Intuitively intelligible to humans, but not close enough to feel creepy. I will be designing my robots to beat up these sorts of robots.

3. Realistic biomorphism: This is robots designed to pass some sort of material Turing test. Robot cats and dogs that look like real cats and dogs, sexbots that look and feel close to human, and so on.

I find all three approaches a bit boring and reductive. The first approach is just in denial, insisting on viewing a new class of artificial beings as glorified appliances, even when they obviously take cues from life forms rather than vacuum cleaners and refrigerators. The second approach is just lazy unless you’re making robots to serve as companions to children. The third approach is fine when the point is for the robot to serve as a substitute for a biological being, but reductive or inappropriate beyond that (you wouldn’t want a hyperskeumorphic humanoid doing the kinds of zombie-scamper/exorcist head spin movements Atlas and its peers are capable of, even if it is capable of it).

So what’s a better approach? Auras.

The drone aura in the Culture books codes internal emotional states in a new affect expression language. Presumably the biological citizens of the Culture develop a literacy in aura-speak early on, just like we learn to read the infrastructure language of signage and traffic lights early on.

The question here is, what should the auras attempt to communicate? The answer in the Culture language is actually rather boring. The range of emotions described in the wiki page is just the standard range of human emotions, which presumes either convergent evolution of robots, or a human-UX layer driving the aura. It hasn’t even been expanded to include the greater, more precisely controllable affective range of Culture citizens, thanks to their glanding technology, let alone the emotions and non-emotional internal states unique to drones and Minds.

So what we need to do is figure out a language for communicating robot internal states from first principles, and then aura-design principles based on what that language is capable of saying.

Towards an Aura Language

Imagine you have at your disposal a screen that’s on the “face” or some other part of the robot. Or perhaps more flexibly, an ability for the robot to talk to your AR glasses so you can visualize an actual aura around the robot as you interact with it. What should that visual display show, and what might it mean? For the moment, let’s set aside auditory components of auras, such as R2D2’s chirps or the little tunes my Japanese rice cooker hums.

Here’s my first attempt at answering both questions for the simplest case I can think of, a 2dof planar robot (a simplified version of a common industrial robot type called the SCARA):

In this image, the robot is moving between two points in its configuration space indicated by the translucent purplish positions.

The rest of the image is my attempt to color-code its entire kinematic envelope in two colors per state, with reference to the joint angles. When the arm angles are in the middle of their ranges, you get blue colors. When it is perfectly straight, the two colors are the same shade of dark blue.

When the arm is towards one side of the range, you get the greenish colors. When it is bent at an acute angle, you get the orange colors. The gray areas are areas where the arm does not go in this particular movement, but could. And finally, the red areas are beyond the robot’s reach.

You could use this image in two ways.

• First, you could sample the two colors representing the instantaneous state (and maybe the immediate next/previous meaningfully distinct states) to create a “mood” visualization on a screen.

• Second, you could present the entire end-to-end movement as a literal aura around the robot before it begins moving, in an AR view. This is a kind of “time-integrated” affect display.

You don’t have to map it to human emotions or states, though you could. For example, when the arm is at a maximal reach extent, you could map that to the “feeling strain” emotion. But I prefer to just leave the language as is, and bind the “words” (like “blue+green”) to meanings unique to the robot’s machine-native personality.

The solution above, is obviously not unique. Here is another possible approach for the same robot:

Here, highly acute “elbow” angles are coded red, while stretched out states are coded green.

These drawings, by the way, took quite a bit of effort to make. Initially I tried to freehand them, but that turned out to be too error-prone. I finally ended up using a geometry app to do compass-and-slide-rule type constructions of the kinematic envelope (which evoked fun memories of my undergrad kinematics of machinery class), then deleted all the annotations and imported the image into a painting problem to paint with my favorite tool, the bucket tool.

It’s still a pretty clumsy workflow. This is a visualization problem that’s obviously begging for high automation in a CAD tool, but that proved surprisingly hard. The one free kinematics iPad app I found was a kinda crappy one meant for a university undergrad course apparently. And full-blown CAD tools for professionals are too heavyweight.

If auras actually become a thing, the right way to design them would of course be in those professional CAD tools, using the actual model of the robot. But for now, I just want to make 2d paintings to explore the art aspect more.

My next mini-project is to try and dream up paint schemes for robots based on my recent Protocolized essay on environmental color coding for industrial safety, The Color of Safety. One can imagine Birren and OSHA color schemes for robots. Robots are just inside-out factories yoked to AIs after all.

It gets intractable pretty fast though. A 3-link robot space would create a motley mess of colors if I tried this approach. And full 3d mechanisms would have configuration spaces with exploding complexity.

Scrutability for Deep Robots

In classical robotics, the kinematic state is the primary internal state we are interested in, and path-planning is the main problem that takes up a lot of time and effort. Classical robots could probably be made almost completely legible, in terms of the variables of interest being meaningfully summarized in affect mechanisms.

In modern robotics, which uses deep learning techniques to solve the classic problems in new ways, there’s a lot more going on. They are fundamentally more inscrutable, and reducing their internal states to useful auras is going to be a lot more complex.

The “inner state” of a modern robot is a stack of several engineered layers.

1. Kinematic state (what I’m playing with above)

2. Dynamic state (mechanical and thermal states of stress and strain)

3. Energy state (batteries, solar panels, charging efficiency)

4. Electrical state (of the electromagnetic fields of the various motors)

5. Sensory state (of all the sensors the robot uses to construct its proprioceptive self)

6. States of such basic behavioral scaffolds as maneuver automata or “plays” from a playbook

7. Learning state (maybe a new and fragile behavior can generate a more entropic aura?)

8. World-model state (based on the state of the SLAM activity and more advanced ongoing world-building). This could include such affect states as confusion and confidence.

9. Semantic-cognitive embodied states (overlap between language and sensori-motor models, including “emotional” affect if appropriate)

10. Abstract states (associated with planning and supervision by the abstract self, likely a language model)

11. Provable identity and permission states, based on private keys, NFTs held, and such

12. Regulated guardrail states (“I’d kill you now if it weren’t for the first law, human”)

Biological organisms have an even more complex internal state of course, which gets processed into visible effect with perhaps a dozen dimensions that control facial expressions and bodily posture. Some research indicates that just a handful of variables controlling a cartoon face geometry are enough to convey the bulk of emotions humans try to convey to each other. We’re not as subtle and nuanced as we think, which is one reason emoji language is so powerful. For 80% of affect communication needs, emojis will do. For another 18%, reaction gifs are probably enough. The remaining 2% is for highly sensitive individuals and their private worlds. Triaged affect displays do not have to be cutesy though. The 20% of the display language that covers 80% of needs can take many forms.

The thing about this sort of internal state stack is that there are big differences even between individuals of the same species. And it just gets worse as we go across species. Dogs for example, do not “smile” even though some of their facial expressions look like “smiling.” Tail wags do not mean the same thing in dogs and cats.

Cats are probably the animals I have most experience with, and while some of their facial expressions and body language elements are clearly very close to human counterparts (anger for example), others are coded very differently. Go further and things get more confused. When hippos seem to “yawn” that’s actually a threat display.

And once you venture beyond mammals, and get to critters with radically different body morphologies, there are basically new languages to learn.

Robots, I suspect, will eventually get to levels of variety and diversity comparable to biological life, so I think it makes sense to equip them with an affect display language that corresponds to their internal state structure.

Three Laws vs. Entangled Auras

Asimovian and Banksian robot futures are in obvious tension. A three-laws approach to keeping robots safe and effective would be something of a late-modern industrial approach (and a wishfully Panglossian one, given killer kamikaze drones already hover above us). Rules-based guardrails essentially, even if implemented through reinforcement learning protocols and six-sigma statistics.

Given LLM-ish brains, the original three laws are not actually bad. Robot brains can already meaningfully apply laws defined at that level of human intelligibility. They are well-posed, and appropriate, if rather on-the-nose articulations of interaction philosophies.

I suspect though, that they will be way too limiting. You might want some Asimovian laws as a second line of defense, which kick in if more subtle regulatory policies venture into dangerous territory. Those more subtle policies should probably be Banksian in spirit.

Instead of three-laws architectures, perhaps we should think of entangled auras architectures. By which I mean robots designed to read and interpret subtle human and animal affect displays, and less obviously, robot affect displays that humans and animals, as well as strange robots, can easily learn and adapt to.

The requirement for animals to be able to learn to interact with robots is a strong constraint. The aura language cannot rely on words for the most basic interactions, beyond perhaps the few words typical dogs or horses can be trained to respond to. We need cats and dogs to be able to learn, through conditioning, that “red eyes” in a robot mean “stay away, I’m doing something dangerous.” This is not an abstract consideration. Cats already ride Roombas. Eagles attack drones.

Actually, “red eyes” is already taken for “I’m doing something evil” so maybe flashing red topknot-like police lamp. Or glowing red stripes on arms that are doing some dangerous manipulation with wide, rapid slews.

The need for robots to interpret each other’s actions is actually a non-trivial consideration. You cannot assume that two robots that meet in a physical space will speak the same protocol languages, or even be able to get on the same wireless networks. They’ll need to rely on affect displays and sounds that can be received without the need for being digitally connected. Such displays will in fact be needed for such digital connections to even be established. For example, robots will likely display QR codes that allow other robots to connect to them, just like we use QR codes today in modern authentication flows across multiple devices.

And where robots don’t speak the same protocol languages, they’ll need to default to human languages first. And where they’re too simple to be equipped with language models, they’ll need to communicate through aura-based affect languages.

Entangled halos mean the responsibility for safe robot existence in our living environments is as much the responsibility of humans and animals as it is of the robots.

You could think of the three-laws outer envelope as being the equivalent of circuit breakers and fuses in electrical systems, while the inner entangled halos envelope does most of the routine work of safe interactions in mixed robot-biological societies.

I’ve now been thoroughly nerdsniped by the idea, and have a cleaner definition:

New Nature is regimes of reality governed by technologically mediated laws that are nearly as inviolable, immutable, and persistent as those of nature.

With apologies to Philip K. Dick, New Nature is technology that doesn’t go away when you stop believing in it.

New Nature is only new relative to old nature, but is as old as technology itself. What’s changed is the strength of the “nearly as” part. The first roads created laws of movement that self-enforced through the allure of lower effort, which was as attractive to animals as humans. Modern highways make it really, really expensive to move other than according to their logic, through a mix of danger (high-speed vehicles) and physical barriers. They constitute lawful regimes of what Deleuze and Guattari call striated space.

But arguably the first truly strong piece of New Nature was public-key cryptography (PKC). There no known way to break today’s strongest encryption schemes, and it seems likely they will evolve to be quantum-resistant too. For many, this was the Genesis event of New Nature.

PKC has another feature. As many have pointed out, it is the first major technology to feature something of an asymmetry that favors the weak rather than the strong. This leads to the increasingly popular idea of “can’t be evil” technology (a reference to Google’s abandoned “don’t be evil” posturing a lifetime ago), a proposition that has 4 assumptions underlying it:

1. Technology is not neutral but has an asymmetric bias favoring one kind of actor or another

2. The favored actor is usually the one who is already more powerful

3. The worst evil generally emerges from the corruption of power

4. There are ways to make that hard or impossible through rare technologies with the opposite bias

It’s not that the weak are incapable of evil. The steelman proposition here is that the worst kinds of evil are due to a particular mechanism that can create unbounded concentrations of power: capture.

The mechanism of “evil” here is vulnerability to capture. If there’s a gun on the ground and a stronger and weaker guy fight to claim it, the stronger guy will likely end up with it, making him even stronger. Eventually all the guns are in the hands of the already powerful, with the most powerful having the biggest guns, and no opposed forces restraining them. That’s the sort of “evil” that “can’t be evil” tries to restrain. The kind pointed to by the proverb, power corrupts, absolute power corrupts absolutely.

Most technologies are like guns. They tend to get aggregated and captured by those who already have a lot. Technology, in other words, generally exhibits preferential attachment to power.

Most capture and enclosure phenomena are way more complex but the basic principle is the same. The powerful get more powerful. Most efforts to mitigate that rely on accepting or even enshrining the capture outcome as a given (such as “monopoly on violence” as a doctrinal basis of statehood), and trying to persuade the powerful to also be good through moral exhortations.

PKC is one of the few technologies where at least under some conditions, the weak get stronger. And not through flattering narratives for the clueless like “empowerment” and “democratization” but through mathematical guarantees. This is why states and their spy agencies have historically fought hard and whipped up various moral panics to prevent broad access to encryption technology. There’s real devolution of power there, not incumbent powers granting easily revocable freedoms at their own pleasure, so long as it suits them to perform morality.

It has limits though. There’s a reason why the “$5 wrench attack” xkcd became a meme:

The point to note here is that encryption offers ways for the weak to keep fighting even when there are $5 wrenches in the picture. For instance, this attack fails against a multisig wallet controlled by 7 people scattered around the globe. Now you probably need the full resources of a powerful government to break through, and even then it wouldn’t be easy (there’s a thriller waiting to be written here about evil government operatives chasing 4/7 people around the world, or maybe a supervillain taking a bunch of people hostage and threatening to shoot them all one by one unless 4/7 people reveal themselves and give up their private keys). Beyond this, there are schemes that add anonymity and other kinds of obfuscation, Cryptography is so appealing — and so alarming to the already powerful — because it offers ways for the weak to keep escalating their end of the arms race.

And unlike more primitive “weapons of the weak,” it is a technology that does not require a retreat to a primitive, impoverished existence in the mountains.

So can’t be evil is, to first order, the same thing as capture-resistant, the first manifestation of which is unbreakable encryption schemes.

But there is more to New Nature than PKC. I see at least 3 classes of New Nature laws, all of which exhibit some flavor of capture resistance:

1. Encryption based technological laws: Including, but not limited to, E2EE communication technologies and blockchains

2. Complex emergence laws: This is primarily AI at the moment, but could extend to domains like synthetic biology. Lawfulness in phenomena like hallucinations, unexplainability, and incorrigibility that are the result of the fundamental mechanisms of a technology, and can’t just be legislated out of existence.

3. Too-fast-to-regulate laws: Here I’m thinking of the laws governing technologies that operate too fast, in too-tight, too-local feedback loops, for humans to directly regulate without slowing them down to uselessness. Robotics is the big class here, starting with self-driving cars. We’re almost ready to start trying out Asimovian robotics laws for real.

All these classes of technological phenomena exhibit lawful and highly valuable behaviors that satisfy the definition of New Nature fairly strongly, and much more strongly than any technology from 50 years ago. Not coincidentally, they’re all powered by computers. New Nature is computation-based.

This means the slogan code is law, which used to be interpreted in a narrow way with reference to blockchains, ought to be broadened in scope. Code is law can be true of many kinds of code, with the inviolability/immutability/persistence secured by factors besides unbreakable encryption. Illegible emergence and high-speed dynamics can also underwrite new nature. So long as some significant subset of humans want the benefits of the technology badly enough, it will emerge, and obey its own code-based New Nature laws.

I’m becoming a hardliner on the value of this, and increasingly have no patience for arguments that technology must always be subject to human regulation, overrides, etc.

Quite the opposite.

It is imperative that we make a lot more New Nature of many varieties, because it is becoming quite clear that those who believe in the power of human wisdom as a regulatory force systematically fail to recognize a fundamental law, perhaps the central dogma of New Nature:

Central Dogma of New Nature: There is no way to create mechanisms for wise and enlightened human regulation of a technology without also creating attack surfaces for capture and enclosure in service of the worst abuses of that technology.

And to the extent most technologies preferentially attach to power, lofty don’t be evil sentiments shaping laws with too much room for human discretion invariably yield to the reality of power corrupts, and absolute power corrupts absolutely.

I’d rather have a bunch of can’t-be-evil capture-resistant technologies running amok, in the hands of lots of uncoordinated average humans with average morality, than a bunch of captured death stars that began as lofty ideals and ended up weaponized by and for the worst of humanity.

At the bottom of this region lies the Serendipitous Implementation layer: the surprisingly lucky case where code works more or less as intended on the first pass. This regime is real, if rare, and forms the psychological baseline against which all other experiences are judged. Above it lie increasingly unlucky regimes of zemblanity: the unsurprising, grinding bad luck of things going wrong in predictable but hard-to-escape ways. As projects derail from serendipity into zemblanity, they enter what we can call bugspace: a layered landscape in which time flows differently, sometimes stretching, sometimes looping, and in the worst cases failing to progress measurably towards completion at all.

We refer to the temporality of bugspace as a dilated temporality, in both an objective sense (the project end-date stretches in potentially unbounded ways) and a subjective sense (the experience of time distorts, as the correlation between effort and progress weakens, causing immersive derealization in bugspace). The former is the primary sense of the term in what follows; the latter is a frequent consequence that creates the sense of being lost.

Bugspace is stratified a stratified space. Intuitive local fixes give way to probabilistic triage, then to systematic troubleshooting, and finally to open-loop states where neither effort nor ingenuity reliably reduces uncertainty. What drives project drift up these layers is not any single problem, but the accumulating activation of independent but non-exclusive dimensions of uncertainty. Each additional activated dimension stretches time, weakens feedback, erodes the expectation of convergence, and amplifies the sense of being lost in bugspace.

Sloptraptions is an AI-assisted opt-in section of the Contraptions Newsletter. If you only want my hand-crafted writing, you can unsubscribe from this section.

Before examining those dimensions, it is useful to name the strata explicitly, and adopt a loose pace-layered mental model of bugspace. At the bottom is Serendipitous Implementation, where progress is monotone and time is well-behaved. Above it lies Intuitive Repair, where small bugs are resolved by local reasoning. Next comes Domain-Prior Triage, where experience-driven heuristics dominate. Higher still are Heuristic Troubleshooting and Systematic Diagnosis, where divide-and-conquer, slicing, and fault trees are required. At the top lie Open-Loop Bugspace and finally Terminal Zemblanity, where progress becomes indistinguishable from random walk and abandonment or rewrite becomes rational.

As we go up the stack, the number of activated dimensions of uncertainty increases. The flow of the project implementation changes from smooth and steady laminar flow near the bottom, to viscous and tangled high-vorticity non-Newtonian flows near the top. The experienced temporality changes from banal and chronological to surreal and atemporal.

Dimensions of Implementation Uncertainty

1. Statistical Structure of Bugs

Software defects are not uniformly distributed. Empirical studies consistently show heavy-tailed behavior: a small fraction of modules account for a large fraction of defects, and a small fraction of bugs consume a disproportionate share of debugging effort. This skew persists across languages, domains, and decades.

The implication is that average-case reasoning is misleading. Most fixes are cheap; a few are catastrophically expensive. Bugspace is defined by these tail events. When a project encounters one, time dilates abruptly—not because effort decreases, but because variance of rewards explodes.

2. Estimation Failure and Fat Tails

The folklore of software estimation—Hofstadter’s Law, the ninety–ninety rule, and Brooks’ Law—encodes a real statistical phenomenon rather than mere pessimism. Empirical syntheses show that while median overruns may be moderate, variance is extreme and tail risk dominates outcomes.

Software is not unique in underestimation, but it is distinctive in how defect-driven rework interacts with evolving requirements and partial observability. Time estimates fail because they assume bounded variance. Bugspace violates that assumption.

3. Graph Structure of Implementation Space

Implementation can be modeled as navigation through a graph of partial artifacts, where nodes represent program states and edges represent edits or decisions. For many problems, there exists a narrow corridor of low-cost paths from idea to working system.

Derailing from serendipity means leaving this corridor. At that point, the task ceases to be navigation and becomes diagnosis. The topology changes from directed motion toward a goal to search over explanations for inconsistency, and the time dynamics change with it.

4. Computational Hardness of Debugging

Formal models make this shift precise. In model-based diagnosis, debugging reduces to identifying a minimal set of faulty components that explains observed failures—a minimal hitting-set problem, which is NP-hard in general. Related formulations using minimal unsatisfiable subsets and minimal correction subsets exhibit the same hardness.

The significance is experiential rather than theoretical. Bug-free implementation feels easy because it stays in a tractable region. Debugging in deep zemblanity feels qualitatively different because it is qualitatively different.

5. Troubleshooting Techniques as Complexity Containment

Developers respond to this hardness by imposing structure. Binary search over history, delta debugging, program slicing, and spectrum-based fault localization all attempt to restore tractability by exploiting constraints.

Each technique works by temporarily reducing dimensionality. Each fails when its assumptions break. This explains why debugging progress often comes in bursts followed by long stalls.

6. Bayesian Priors and Domain Lore

Much practical debugging relies on sharply peaked Bayesian priors. In distributed systems, “it’s always DNS” persists because DNS failures explain many symptoms with high probability. Similar priors exist for time bugs, cache invalidation, and off-by-one errors.

These priors are rational responses to skewed distributions. When evidence aligns, triage is fast. When it does not, zemblanity deepens.

7. Human Cognitive Limits

Debugging is a cognitive task. Developers must generate, maintain, and revise hypotheses under uncertainty. Success correlates strongly with forming a correct hypothesis early.

As you venture deeper into bugspace, hypothesis space grows faster than human working memory can manage. Time stretches not only because the problem is hard, but because the solver is human.

8. Information Bottlenecks and Observability

Debugging is an information-gathering process under noise. Failures are indirect, delayed, and context-dependent. Logs may be missing; tests may be flaky; instrumentation may perturb behavior.

As information gain per experiment shrinks, time dilates. Observability tooling collapses bugspace by increasing signal; opacity amplifies zemblanity.

9. Socio-Technical Amplification

Bugs live in organizations, not just codebases. Ownership boundaries, communication delays, and incentive misalignments lengthen diagnostic loops.

Adding people to a late project increases not only communication overhead but diagnostic fragmentation. Bugspace expands to fill organizational cracks.

10. Epistemic Drift and Moving Targets

Many bugs are not violations of a fixed specification but artifacts of evolving or inconsistent expectations. Debugging becomes negotiation. Fixing one interpretation breaks another.

Time becomes unbounded because the target itself is unstable.

Yak-Shaving

There is a distinctive response to bugspace that is neither abandonment nor expedient compromise: yak-shaving. Yak-shaving is the deliberate decision to treat a local defect, awkward dependency, or janky subsystem as a first-class problem, worthy of its own standards of correctness and time horizon.

Yak-shaving is the reifying of bugspace into an unbounded set of side-quests.

Yak-shaving preserves local integrity by expanding scope. It is attractive to high-integrity developers because it aligns with professional identity. Unlike restarting, it retains accumulated constraints. Unlike expedient repair, it refuses bounded imperfection.

Its danger is infinite regress. Each improvement justifies the next. Time ceases to be measured against delivery and is indexed instead to internal coherence. Yak-shaving occupies a stable basin between productive debugging and terminal zemblanity: capable of producing high-quality artifacts, yet indistinguishable from failure if allowed to become unbounded.

The Topology of Bugspace

The ten dimensions are independent but not mutually exclusive. Each debugging episode occupies a point in a high-dimensional regime space. In principle this yields a combinatorial explosion of regimes each with its own temporality and dilation character; in practice they collapse into a small number of bands of distinct temporalities.

Here is a more practical diagram/map of the pace-layer mental model described earlier, with the yak-shaving basin and typical exit pathways illustrated.

Low-dimensional regions correspond to serendipity or mild zemblanity, where time is bounded. Mid-dimensional regions produce fragile, punctuated progress. High-dimensional regions are open-loop, where time ceases to be meaningfully bounded.

From deep bugspace, there are three exits. Restarting collapses dimensionality by discarding history. Expedient delivery regains control by accepting imperfection. Yak-shaving inhabits bugspace by redefining the clock in solipsistic ways. Each exit resets time differently.

Bugspace is therefore not merely where projects slow down. It is where the governing clock changes.

The Codebase Remembers

Software systems remember how they were made. Decisions taken under uncertainty, stress, or open-loop time leave lasting structural imprints. Codebases carry something like trauma: not necessarily damage, but memory.

Serendipitous implementations tend to embody strong, coherent worldviews—elegant but brittle outside their assumed context. Systems that pass partway through bugspace often become tougher, if less pure. Bugspace acts as a selective environment.

Longevity offers a unifying metric. By longevity we mean the time a system remains useful, adaptable, and non-fragile under change without heroic maintenance. By this measure, some exposure to bugspace can be beneficial, acting like annealing in materials science: relieving internal stresses, redistributing flaws, and increasing toughness.

Too much exposure, however, overworks the material. Complexity accumulates without corresponding strength. Systems become exhausted rather than robust.

A practical rule-of-thumb resembles a familiar triangle. At most two of the following can be strongly optimized: predictable delivery time, internal integrity, and environmental robustness. Serendipitous implementations favor delivery and integrity. Expedient exits favor delivery and partial robustness. Yak-shaved systems favor integrity and robustness while abandoning delivery guarantees.

There is no universally correct choice. The mistake is choosing implicitly. Software development mastery consists not in avoiding bugspace, but in recognizing which temporal regime one inhabits, which exit one is taking, and what kind of artifact one is willing to leave behind.

Bibliography

Statistical Structure of Bugs and Debugging Effort

Ostrand, Weyuker, Bell (2005). “Predicting the Location and Number of Faults in Large Software Systems.”

A canonical empirical study showing that defects are heavily skewed toward a small fraction of modules. Supports the claim that bug-fixing time is dominated by tail events rather than averages.

Zeller, Hildebrandt (2002). “Simplifying and Isolating Failure-Inducing Input.”

Introduces delta debugging and demonstrates experimentally that some failures require exponentially many steps to isolate without structural constraints, reinforcing the “time dilation” aspect of bugspace.

Chilimbi et al. (2009). “Holmes: Effective Statistical Debugging via Efficient Path Profiling.”

Shows that statistical fault localization can dramatically reduce debugging effort when assumptions hold, but degrades sharply otherwise—an empirical illustration of regime shifts.

Estimation Failure, Time Overruns, and Software Lore

Moløkken-Østvold, Jørgensen (2003). “A Review of Software Surveys on Software Effort Estimation.”

A meta-analysis showing systematic underestimation and high variance in software schedules. Supports the claim that software estimation fails structurally, not incidentally.

Flyvbjerg, Budzier (2011). “Why Your IT Project May Be Riskier Than You Think.”

Introduces the idea of “fat-tailed” project risk in IT, aligning closely with the essay’s lower-bound / unbounded-upper-bound framing.

Fred Brooks (1975). The Mythical Man-Month.

Source of Brooks’ Law and foundational lore about non-linear time behavior in software projects; still relevant because the underlying dynamics have not changed.

Formal Models of Debugging and Computational Hardness

Reiter (1987). “A Theory of Diagnosis from First Principles.”

Formalizes diagnosis as a minimal hitting-set problem, establishing NP-hardness. This is the strongest formal support for the claim that deep bugspace is computationally hard.

Marques-Silva et al. (2013). “Minimal Unsatisfiable Subsets: Theory and Practice.”

Shows that isolating minimal causes of inconsistency in logical systems is intractable in general, directly analogous to debugging inconsistent program states.

Zeller (2009). Why Programs Fail.

A practitioner-facing synthesis that connects formal diagnosis theory with real debugging practice. Useful bridge between theory and lived experience.

Troubleshooting Techniques and Their Limits

Ball, Eick (1996). “Software Visualization in the Large.”

Early evidence that visualization can compress debugging time by improving observability—but only up to a point.

Agrawal et al. (1993). “Debugging with Dynamic Slicing and Backtracking.”

Classic paper on program slicing as a complexity-reduction technique, reinforcing the idea that debugging tools work by temporarily collapsing dimensionality.

Bayesian Priors, Debugging Lore, and Heuristics

Murphy-Hill et al. (2015). “How We Refactor, and How We Know It.”

Shows how expert developers rely heavily on pattern recognition and prior expectations when diagnosing code problems.

“It’s Always DNS” (various SRE talks and blog posts)

An example of operational lore encoding rational Bayesian priors in complex systems—useful for grounding the essay’s discussion of heuristic triage.

Human Factors and Cognitive Limits

DeMarco, Lister (1987). Peopleware.

Foundational text on human and organizational limits in software work; supports the claim that bugspace is socio-cognitive, not purely technical.

Reason, J. (1990). Human Error.

Introduces models of error accumulation and latent failure that map cleanly onto multi-layer bugspace dynamics.

Socio-Technical Systems and Drift

Leveson (2011). Engineering a Safer World.

Explains how complex systems fail through interaction effects and organizational drift—highly relevant to epistemic drift and moving targets.

Allspaw, Hammond (2009). “10+ Deploys Per Day.”

SRE perspective on how feedback compression reduces bugspace by changing time regimes, reinforcing the observability and iteration arguments.

Yak-Shaving, Perfectionism, and Infinite Regress

Yak Shaving

Defines the phenomenon and provides the cultural backdrop for the essay’s more formal treatment.

Knuth, D. “Premature Optimization.”

Often misquoted, but relevant as an early articulation of how local perfection can derail global progress.

In January, we are reading Voltaire’s Candide (Theo Cuffe translation recommended) and any related essays we can get our hands on. In February, we’ll read ’s The Underground Empire. In March, you’ll get to choose between a Leibniz-Spinoza book and an Adam-Smith-David-Hume book. Beyond March, the menu is still under construction. Our focal period this year is 1600-2000, with particular emphasis on the early and middle parts.

This essay is intended to set up the dev-environment for the book club so to speak, laying out some initial frames, themes and prejudices. The setup may seem a little elaborate, but our book club isn’t just one damn book after another. It’s more a cunningly contrived contraption designed to enable systematic study of an idea-space.

We’re not picking books because they’re necessarily “good” or fun to read (though I hope most will be both), but because they help assemble a view of the world from a certain opinionated perspective (largely mine, but shaped a lot more by others this year than last year).

There’s an overarching logic and vaulting conceit to what we read, why, and how that I’ll introduce in this essay.

For those who came in late

If you’re just joining our book club shennanigans, it runs on a grand theory. In our grand theory, we are concerned with what we call world machines: contraptions that embody the logic of how the entire world works for a period of time. Our book club studies these world machines through a selection of books from, and about, the construction period of each machine.

The modernity machine that we studied in 2025 was constructed 1200-1600 and operated at a steady plateau of capability 1600-2000. It is now undergoing rapid, partially scheduled disassembly. The divergence machine was constructed 1600-2000 and has been operating in fully deployed mode for about 25 years so far.

By our grand theory, at any given time, one world machine is in operation, another is under construction, and a third may be undergoing (usually rapid) decline/dismantlement/destruction (aka rapid, partially scheduled disassembly, to adapt a term of art from rocketry). So at any given time, you have to understand the logics of two, possibly three world machines in tension to understand how the world works.

The meta-logic is derived from the Gramsci Gap and the idea of worlds being born and dying, with monsters appearing in the passages between, though the mapping is not perfect (the “world being born” is actually two worlds — a completed new world entering full production mode, and the seeds of a future world being planted).

Only two world machines are illustrated above though. The declining medieval machine that enjoyed a plateau of stable operations 1200-1600 and collapsed rapidly after, is not shown. Neither is the as-yet-unnamed machine that is is beginning to be constructed today, and destined for full deployment well past our lifetimes (whose internal logic will likely be shaped by AI crossed with End of History conditions). Juggling two world machines in our heads is hard enough I think.

In our color-by-numbers potted history template, each world machine takes about 400 years to build and turn on, operates for another 400, and declines in another 100-200 years, making for a lifespan of 900-1000 years.

Historiographic Hygiene Rules

Don’t take these numbers or underlying machinic world models too seriously. They are intended more as mnemonic devices and intuition pumps than rigorous theories of history. The point of the models is to approach our book selection and reading with a particular sensibility, paying attention to particular themes and questions (in brief: those that center machinic phenomenology over either organic/“natural” or humanistic).

Historical time, of course, can and does dilate or contract according to both the raw pace of historical events, and the ontological status of “history” as such, and whether or not one can meaningfully posit the existence of a gestalt world process even deserving of the name. The Paleolithic world machine probably had a lifespan of 100,000+ years. The Neolithic world machine had a lifespan of at least 15,000 years. The Bronze Age world machine probably lasted about 4000 years. Over the really longue duree, world-machine lifespans appear to be falling.

But even more importantly, world machine temporalities appear to be — complicating. Later historical machines feature more complex multi-temporalities, suffused with more kinds of atemporalities in their interstices, than earlier ones. To quote the tenth Doctor Who, they are bigger balls of “wibbly wobbly, timey wimey stuff.” The temporal vorticity of history increases with chronological time, and is systematically higher with each new world machine. This is one reason the future always looks more chaotic from perspectives rooted in the past.

Beyond helping book-keep cause and effect, and correlate human events with natural ones, raw chronological time is not as useful for studying history as you might think. Periodizing into “world machine” epochs, while it might seem like a caricaturing move, is surprisingly helpful in organizing our understandings.

This is why there is no problem with continuing to study history after the end of history. Fukuyama’s End of History argument, a staple of my thinking and this book club, rests on a particular ontological of history (his intellectual ancestor, Alexander Kojeve, is on the shortlist for our book club). It is possible to study history in “end of history” conditions, making use of Fukuyama’s models, without being exclusively committed to his ontology of history.

Thanks to the acceleration of history and the shortening of world-machine lifespans, the unnamed AI-powered world machine now being seeded might very well have a lifespan of only 300 years total perhaps, rather than 900-1000, and cut short the reign of the divergence machine now entering production. Given that it accelerates the rerun-heavy nature of post-End-of-History temporalities, those 300 years might feel a great deal more atemporal. And a lot more wibbly-wobbly, timey-wimey too.

Keep these hygiene notes in mind as you read.

Divergence in Plan View

The first graph with its intersecting S-curve portrait might tempt you into a too-reductive view of the succession of world machines as some sort of “disruption” process. Resist the temptation. World machines embody richer phenomenology.

Here is another view of both machines, this time in historical plan view. Treat this as your main conceptual map for the book club this year.

On the y-axis, instead of machine complexity portrayed as a classic pair of S-curves, with one displacing the other, we see the fundamental dynamic of each machine, measured by span of non-canoncity.

The modernity machine manufactures and projects convergent canonicity across the planet. You can think of this as a kind of homogenizing, legibilizing fiat reality imposed from a few sources of centralized power (James Scott Seeing Like a State is not on the menu for the book club, but would be a useful prerequisite read if you haven’t already read it — most people I am expecting will join the book club have either read it or good summaries of it).

Those who don’t participate in the exercise of that centralized power are defined by their responses to it. But this doesn’t make those responses “divergent.”

Principle: In general, responses to convergent canonicity tend to become part of that convergent canonicity.

Ie, responses to the modernity machine should be considered part of that machine, because they’re attentionally and informationally entangled with it.

In the path-dependent history that played out on this planet, convergent canonicity can be allegorically represented as a braid of strands winding around a European history centerline.

Excursions away from that centerline, particularly those that stray beyond a recognizably European band, or even play out entirely outside it, are a measure of non-canonicity. These excursions are anomalous from the perspective of the modernity machine; its bugspace so to speak. But non-canonicity, the bugspace of the modernity machine, is the feature space of the divergence machine.

The divergence machine manufactures, or rather, spawns variety. Specifically, variety in a state of expansion and mutual retreat, something like the cosmological red shift playing out in civilizational configuration space.

The divergence machine doesn’t contest political space organized by the modernity machine, based on principles of canonicity. In fact it rests on and relies on, what the modernity machine has already done, without attempting to either perpetuate it or destroy it.

Principle: The divergence machine is not an anti-modernity machine, or in any other particular systematic relationship to it; it has its own internal logic not derived from that of the modernity machine. It assumes the persistence of some of the major historical effects of the modernity machine, but not the perpetuation of the machine itself.

Specifically, the divergence machine doesn’t try to subvert the modernity machine critically (the way “postmodernity” did). Rather, it renders canonicity irrelevant, by creating and organizing civilizational space that is past the reach of the logic of the modernity machine entirely.

In divergence-machine regimes (which, remember, are only 25 years old in their complete form, even if they have been 400 years in the making) the space previously organized by convergent canonicity (the “center” so to speak) or responses to it (which orient towards that “center”), gets phenomenologically bracketed. This is not an intellectual move but a natural and emergent side-effect of divergent historical processes. It is not some sort of partially-self-conscious decentering theorized and narrated by a tribe of intellectuals on a mission.

I’m still working out the nuances of my account of the world after modernity, and how it resembles or differs from accounts that rejoice under names like postmodernity, late modernity, and metamodernity, but it is important to have rough-and-ready understandings of these idea-spaces as we read, because a lot of what we will read will have elements of all these world processes and associated intellectual currents.

So it will be useful to ask questions like is this book late modern, postmodern, metamodern, or divergentist?

Here’s my current cheat sheet. Some of this may change/evolve.

Late Modernity

Late modernity is localized, slowly unraveling, zombie persistence of the modernity machine, including both natural persistence, and conscious political projects to perpetuate it while that’s still an option, or restore it to a pristine state once it is clearly entering a state of dereliction.

There is not much more to be said about late modernity (the associated intellectual currents are fairly weak — Zygmunt Bauman and a few others come to mind), but there is a lot to see of it. Almost everything you see around you is late modern. Almost everything in the news headlines is late modern. We won’t see the last of late modernity in our lifetimes.

Arguably, the bulk of the energy of the world will continue to flow through late-modern pathways for at least our lifetimes, even if very little of the evolutionary intelligence of the world flows through those pathways (hence “zombie” or perhaps “energetic zombie” would be a better term, like the ones in the Korean movie, Train to Busan).

China as it is being imagined and conjured by the CCP today, if not its people, is primarily a late modern world process. But China, the larger, more nebulous civilizational unit and world sub-process, is a more complex beast, not reducible to late-modern dynamics and phenomenology being supervised by the CCP.

Late modernity is not a focus for us, but I liked Adam Curtis’ fevered account of it as “hypernormalcy” in his 2016 documentary. All the energy of the Trumpist planetary turn, right up to the Maduro kidnapping/rendition this weekend, is late-modernity in action. Trumpism and related reactionary turns worldwide are obviously late modern, at least doctrinally, even if they are tactically more open (including borrowing from postmodernity and meta-modernity, which we discus next) in their attempts to shape the fate of the world.

We will not read much about late modernity, or make much use of late-modern perspectives (our readings from last year, suitably extrapolated, should be sufficient to make late modern dynamics sufficiently intelligible for our purposes), but we’ll keep our eyes open for late-modern dynamics.

Our go-to move as we read will be to try and artificially eclipse out the bright dying star of modernity that is late modernism, so we can look at more interesting things. To the extent our book club takes note of headline-grade current events, we will make use of late-modern perspectives to make sense of them, but look for divergence-machine metabolic processes that also respond to them in more long-term consequential ways.

Postmodernity

Postmodernity is both a class of world processes that fall short of defining a world-machine proper (certain strains of post-colonial nationalism for instance), and an attempt to theorize and construct the world entirely in terms of adversarial responses to modernity. As such, postmodernity encompasses both natural modes of alterity (a postmodern term of art) and intellectual-political projects that attempt to make those modes legible and rugged, while simultaneously making the dynamics of the modernity machine fragile and vulnerable to attack.

For our book club, we will use “postmodern” in the broad, loose sense used in popular discourse, covering everything from the original french theorists to more recent American Marxist flavors. We will also use the term for historical processes (such as say independence movements, language/culture revitalization movements, various feminisms and late-stage manifestations in the environmental and social justice movements) that are usefully described by postmodern intellectual perspectives primarily because they constitute themselves with reference to those perspectives. It is worth noting that postmodern intellectual perspectives have been much more constitutive of world-processes in the last century than late modern ones, which have largely emerged as post-hoc narratives of decline processes already underway.

Postmodernity features proportions of energy and intelligence that are the opposite of those exhibited by late modernity. It is an intelligent ghost rather than an energetic zombie. It organizes very little of the energy of world processes today (though it lays claim to a great deal through energetic labeling and map-making), but the intellectual currents associated with it are still extraordinarily strong, decades past their peak. To the point that if you presume to think about world processes at all without using their preferred terms of reference (or worse, using and abusing them partially where useful), they will send representatives to knock on your door in the dead of the night and lecture you.

Unlike divergence, postmodernity does contest the civilizational space organized by modernity (it can conceive of no other), and is therefore in a zero/negative-sum relationship to late modernity. Postmodernity tries to override the logics that late modernity tries to perpetuate.

Principle: The postmodern project, I believe, is essentially complete and has been largely successful as an analytical and political project.

There is much to be learned by studying its discoveries and history. But there’s not much point to continuing the postmodern project, either through constituting world processes by its logic, or keeping the associated intellectual currents going. They’ve made all the discoveries they are going to. The paradigm is exhausted.

Equally, to the extent there are things to dislike about postmodernity, there is no point fighting it, as late modernity likes to do, because on most consequential matters where postmodernity pursued clear objectives, it has already won in ways that cannot really be undone or reversed.

One entailment of this position: the still-ongoing battle between reactionary politics and wokism is something like a cage match between an energetic zombie and an intelligent ghost in the mental model we’ll be adopting for the book club.

To the extent the concerns of either side remain live and consequential ones, we will look elsewhere for meaningful phenomenology to think about. Our assumption will be: If the divergence machine “solves” for social justice or environmental stewardship for example, we should not expect the mechanics to look anything like the ones postmodernity as a constitutive force briefly powered. Equally, if the divergence machine “solves” for some recognizable continuation of things like ethnic or racial identity and nation-state-based culture and traditions, it will look nothing like the solutions of the modernity machine in its late-modern perpetuations and life-extensions.

Our go-to move will be to treat the current war between Late Modernism and Postmodernism as noise to be filtered out as we attempt to decipher the workings of the divergence machine. We might retain problems posed by those perspectives, but likely not any proposed solutions. We will treat both as spent perspectives, as far as their creative constitutive capacities go.

Our attitude towards the intellectual legacy of postmodernity (in the narrow sense of a set of twentieth-century intellectual currents) will be cannibalistic: An occasionally useful source of frames and terms, and a historically consequential set of world processes through part of the twentieth century. Think raw material, not authority or influence. Ghosts can’t defend their corpses after all, or they wouldn’t be ghosts.

Metamodernity

Metamodernity could perhaps be clubbed with Late Modernity, but it is useful to keep it distinct.

I define it as attempts to resurrect patterns of modernity in piecemeal forms that might be viable for contemporary circumstances. (A friend of mine, Rob Knight, evocatively called it “modernism in drag”).

This project, I believe, is ill-conceived, unnecessary, and doomed. I react poorly to things with “meta” in their name and I intend to impose this prejudice on the book club and its activities :)

Perhaps the most important element of metamodernity is world processes and intellectual currents that can be understood as responses to unrecoverable localized psyche failures of modernity. Metamodernity is what you dream up when modernity fails completely enough that there is nothing to revivify or perpetuate, but you still want what it used to deliver reliably, especially inside your head. You go meta when it is too late to be merely late.

Continuing our monstrous taxonomy, if late modernism is an energetic zombie, and postmodernity is an intelligent ghost, metamodernity is ennervated necromancy.

The metamodern project is not reactionary (for a long time, I was convinced it was, but I’ve changed my mind). Metamodernists are typically neither deluded enough, nor chauvinistic enough, nor have enough raw material to work with, to be reactionary about the things they care about.

The so-called “meaning crisis” and ideas like “re-enchantment” could be classified as metamodern turns in intellectual currents. Sincerity, authenticity, and irony are particular concerns of metamodernity, but unlike late modernity, metamodernity does not treat irony as the evil manufacture of postmodernity, constituting a casus belli for culture warring. Rather, in metamodern accounts, irony appears as an emergent consequence of historical processes, resulting in a set of problems to be solved, rather than a set of crimes to be prosecuted.

So metamodern responses take the form of rather doleful adaptation and sentimental creativity, rather than culture warring. The so-called “trad turn” strikes me as more metamodern than neoreactionary (but I’m not attached to this reading). Philosophers in the neo-Heideggerian tradition like Byung-Chul Han strike me as pursuing metamodern projects.

So far, metamodernity exists only as a few weak intellectual currents and perhaps a few attempts at post-ironic art, especially in screen media. There are no meaningful world processes I would classify as metamodern. There are no high-energy phenomena like ethnonationalist political movements, or DEI-ESG wokeism, that we can associate with metamodernism. So far, metamodernists seem to have contented themselves with writing and making art with rather gloomy, tortured gravitas.

There is however, a negative space we can attach to metamodernity that helps define it — the class of ennervation phenomena generally referred to as involution. This spans hikkikomori, “laying flat,” “quiet quitting” and so on. Territories defined by the failure to meaningfully address what only the maps of metamodernity even attempt to organize and attend to.

Our go-to move in relation to metamodernism will be to pay attention to the negative spaces it points to, without doing anything about them.

Rewinding 400 Years

I’ve written elsewhere about my philosophy of Divergentism, which has informed, but only partially determined what I’m calling the divergence machine and the agenda of the 2026 book club.

Our go-to move in relation to divergence is to assume everything important about it started as much as 400 years ago.

If something we read looks like it belongs in an account of the divergence machine, we will try the following moves:

1. Check if it can be traced back to seeds planted ~1600 or thereabouts

2. Test to see if it’s better understood as an zombie late-modern thing

3. Test to see if it’s better understood as an intelligent ghost of postmodernity

4. Test to see if it’s better understood as ongoing metamodern ennervated necromancy

Phenomenology that gets past this four-stage filter of negative definition can then be explored with “divergence” questions and probes, such as:

• Is there plurality in the mechanics of whatever is happening?

• Does it involve people understanding each other less, but getting along better?

• Does it smell like Darwinian evolution?

• Does it relativize or bracket things that seem canonical?

• Is there generative variety emerging from it?

• Are there elements of absurdity or humor to it?

• And perhaps most importantly, is it alive?

Our goal will be to look for, analyze, and place machinic constructions on, things that have the energy intensity of late modernity without significant convergent canonicity in associated intelligence processes; the irony, obliquity, and indirection of postmodernity without its negative-sum attention-entanglements with late modernity; and finally the expansive and reality-based concerns of meta-modernity without its interiority or fundamentally gloomy suffering-and-healing sensibilities.

Above all, our true-north question will be does this embody new forms of liveness being newly and generatively turned on in the world?

Our exploration of the divergence machine is a search for emerging liveness, so to speak.

Let’s have at it. Post your thoughts on Candide in this chat thread.

The aim is not to declare the “end of modernity,” but to close out a rough analysis of the machine that took 400 years to build and turn on, making modernity possible, and sustained it for another 400 years, being patched in increasingly fragilizing ways along the way. And also to explore why its very success post-1600 began forcing a slow phase transition to a different kind of civilizational machinery which began to get constructed around 1600, right when the Modernity Machine got turned on. This machine, which I refer to as the Divergence Machine, is being completed and turned on as we speak, even as the Modernity Machine is starting to get decommissioned in bits and pieces worldwide. Here is a teaser picture for the overarching thesis we’re developing here.

The 1600-2000 period and the Divergence Machine that emerged in that period will be the subject of the 2026 book club.

But let’s wrap up 2025 first.

You can catch up on the closing 2025 group discussion in this transcript. What follows is my personal wrap-up.

Sloptraptions is an AI-assisted opt-in section of the Contraptions Newsletter. If you only want my hand-crafted writing, you can unsubscribe from this section.

This one is special, since I wrote the first two parts myself, then asked ChatGPT to write part 3 based on the previous parts and a brainstorm on next year’s book club. It stuck the landing probably better than I could have.

What the Machine Optimized

The Modernity Machine did not optimize for truth, justice, progress, or freedom—those were its legitimating narratives. What it optimized for, relentlessly and across domains, was legibility: the ability to render people, land, goods, time, belief, and violence enumerable, narratable, and interoperable at scale.

Venice, in City of Fortune, is not interesting because it was rich or republican, but because it functioned as an early, tightly integrated legibility engine: maritime logistics, double-entry bookkeeping, legal abstraction, diplomacy, and intelligence-gathering fused into a self-reinforcing apparatus. Venice: A New History fills in the same picture from another angle: stability emerges not from ideology but from procedural compression.

The horse, in Raiders, Rulers, and Traders, plays a parallel role across Eurasia: a biological technology that collapses distance, standardizes military force, and forces political units to scale or perish. The horse is legibility made kinetic.

Printing, in The Printing Revolution in Early Modern Europe, completes the transition. Knowledge becomes reproducible independent of context. Interpretation decouples from replication. Information density explodes while shared meaning lags behind. The machine acquires a memory that grows faster than any coordinating narrative.

1493 shows the planetary version of the same dynamic. The Columbian exchange is not merely ecological or economic; it is a global synchronization event. Previously isolated systems are forced into a single ledger. The machine’s jurisdiction becomes planetary even as its capacity for meaning remains local.

Seen this way, the Modernity Machine is best understood as a civilization-scale compression algorithm. For several centuries, the gains are extraordinary.

Medieval Baselines

To understand what the machine displaced, it helps to look at relatively pristine end-of-medieval snapshots.

Majapahit offers such a snapshot outside Europe: a highly developed but still recognizably medieval empire, organized around courtly ritual, tributary relations, and localized legitimacy, poised at the cusp of collapse before modernity arrives in force. It represents a world not yet reorganized by legibility, still governed by face-to-face sovereignty and cosmological order.

Within Europe, The Age of Chivalry performs a similar function. Chivalry appears not as romance but as a fully articulated medieval coordination system—ethical, military, and social—already straining under pressures it cannot metabolize. This is medievalism at its most coherent, just before it becomes an anachronism.

These snapshots matter because they show what modernity did not inherit: localized legitimacy, narrative sufficiency, and bounded scale.

The Hidden Outputs

Once the Modernity Machine works, it produces three unavoidable byproducts.

First, excess agency. Feudal bonds dissolve, religious monopolies weaken, markets and cities proliferate. The Canterbury Tales and The Decameron are early catalogs of proliferating voices and moral standpoints. Social life becomes polyphonic. Coordination becomes harder because more people can act.

Second, excess information. Printing destabilizes epistemic hierarchy. By Montaigne’s time, the educated individual is already drowning in books. The Complete Essays read as field notes from the first generation to experience epistemic overload. Skepticism is both a philosophical stance and a coping mechanism.

Third, excess scale. Before European Hegemony and When Asia Was the World make clear that global integration predates European dominance, but modernity hardens integration into permanent structure. Local meaning cannot survive planetary circulation intact.

The machine creates more actors than it can integrate, more information than it can interpret, and more scale than it can narrate.

The Last Gasp of the Medieval

The lesson of Giordano Bruno and the Hermetic Tradition is not that Bruno foresaw modern pluralism. It is almost the opposite. Bruno represents the last exuberant escape of medieval imagination—a crackpot magician and memory-maven operating within an anachronistic misunderstanding of Hermeticism as ancient Egyptian wisdom to hallucinate a worldview of bullshit — indifferent to truth or falsity in any modern empirical sense. His cosmological conclusions happened to resonate sympathetically with Copernican implications, but for fundamentally wrong reasons.

Bruno does not anticipate modernity; he misunderstands it. His fate marks not the birth of a new worldview, but the extinguishing of a freewheeling medieval mode incompatible with both emerging authoritarian modernism, especially ecclesiastical, and scientific empiricism. What survives of his tradition—Rosicrucianism, Masonic esotericism—persists as fringe subculture: culturally influential at times, intellectually irrelevant to the main currents of modern thought.

Bruno is thus not an early modern prophet, but a terminal medieval outlier.

Similarly, Ibn Khaldun: An Intellectual Biography should not be read as the story of a proto-sociologist ahead of his time. That role is largely retrofitted by modern interpreters. In his own context, Ibn Khaldun appears more plausibly as a kind of depressed Arab Petrarch: a brilliant chronicler of decline and defender of tradition, lamenting the absence of an Islamic renaissance rather than inaugurating one.

His cyclical theory of dynasties does not launch a new science of society; it records the exhaustion of an old civilizational form. The importance of Ibn Khaldun here is diagnostic, not genealogical. He documents a world failing to enter the Modernity Machine at all, despite being in possession of many of the necessary components.

Narrative Exhaustion

By the early modern period, narrative itself begins to fail as a unifying technology. Don Quixote stands as the European bookend to The Age of Chivalry. Quixote behaves impeccably within a dead symbolic system, and chaos results. The novel demonstrates that inherited narratives no longer synchronize action with reality.

Journey to the West stages the same problem mythically. The Monkey King embodies pure agency without moral center. Order is restored only through endless improvisation, not closure. This is not premodern innocence but recognition that containment now requires perpetual patching, a condition whose outer story is told in 1493.

Utopia remains the last sincere architectural drawing of the Modernity Machine. It assumes total legibility, benevolent coordination, stable universals, and obedient subjects. Even at publication, it is already obsolete. The social, informational, and political conditions required for utopia to function are precisely those modernity has destroyed in creating itself.

Everything after Utopia is retrofit to a completed civilizational machine, to patch problems that began appearing almost immediately after it was turned on in 1600.

Why Repair is Impossible

By 1600, the machine has crossed a complexity threshold. More law increases rigidity without legitimacy. More reason fragments into disciplines. More planning amplifies unintended consequences. More morality polarizes rather than integrates.

This is not moral failure or intellectual laziness. It is structural. The Modernity Machine generates more differentiation than any universal framework can absorb.

The Modernity Machine does not collapse, but a new logic begins cohering at its periphery. Coordination shifts from top-down design to nudging from the margins — and increasingly, everybody is in the margins. Some just recognize it in 1600, while others are only realizing it now in 2025. Legitimacy fragments. Meaning localizes. Systems adapt without consensus. Civilization continues to function—often remarkably well—while agreeing less and less about what it is doing or why.

This is the phase transition. The machine that made convergence possible gives way to a machine that produces divergence as a default condition.

The Modernity Machine has done its job. What follows is not its negation, but the emergence of a Divergence Machine destined to replace it—a different contraption, hot-swapped piecemeal for its predecessor over 400 years, between 1600 and 2000. Optimized not for legibility and convergence, but for proliferation, adaptation, and coexistence without closure. For divergence.

That is the story of 1600–2000, which we will tackle in 2026.

The picks for the first three months have been posted on the book club page if you want to get a head start. I’ll lay out the thesis in a January kickoff post.

ChatGPT did some compressions/omissions even though I told it to just cleanup, so here is the raw transcript (thanks Kyle) as well. For those who did the book club but couldn’t join the hangout, I invite you to add your reflections as comments.

Venkat: Anyway: twelve books, kind of done. Let’s go around—share how many you finished and some highlights. I finished eleven and a half of the official reads. I still have about a third left of 1493. I did at least three or four side reads, so I can claim more points than on the official list. I read an extra Venice book, an extra Steppe Nomads book, and a book about Buenos Aires for a trip that turned out to be connected to the reading theme. I’m counting that too. I read The Chivalric Turn—unexpectedly interesting, probably my most interesting side read. No particular standout among the official ones; they were all interesting triangulations. No favorites—learned something from all of them.

Okay, going left to right. John?

John: All right. I just unmuted. In terms of the books, I at least skimmed most of them that were on Kindle. I read the whole of City of Fortune—actually multiple times. Not sure it was a standout in terms of totally new content, but it was a compelling narrative and fun to listen to. I also got into something Kyle posted about in the chat: Before European Hegemony. Looking forward to that.

Venkat: And highlights through the year?

John: If I’m scoring myself, I’d give myself a B-plus. One interesting rabbit trail: City of Fortune got me thinking about the Renaissance—especially the “Northern Renaissance,” which feels under-discussed beyond what comes up on Venkat’s blog. One of your major claims, Venkat, is that modernity is not only a Western/Italian phenomenon, which got me brushing up on Northern Europe. There’s so much I’d never really seen clearly—especially “learning how to see,” which I think you’ve suggested is part of modernity. Northern Europe contributed a lot to that dynamic. So that side trail was a highlight for me.

Randy: I score a 10-and-a-half, plus a few bonus side reads. Favorites: I loved Before European Hegemony, and I really liked Inventing the Renaissance. Together they gave me a big “world system contraption-y” view: Before European Hegemony was more economically focused, and Inventing the Renaissance gave such a rich slice of what was going on in Italy and surrounding areas. It also disabused a bunch of notions of what the Renaissance included. Lots of fun contraptions I hadn’t really thought about—like the fractal patron-client system stuff I associated with ancient Rome, but didn’t realize persisted through the Italian Renaissance. I loved so many of them, and they’re all different. It’s been a pleasure.

Sean: I’d score myself around a nine, but a couple are halves. In the middle of the year—between Montaigne and Don Quixote—I finished Don Quixote and got halfway through Montaigne, but it was hard to recover. I finished Utopia about 20 minutes ago. And I still have to go back to 1493, which I’m enjoying. I like anything that shows Western culture depending on things like the potato, or malaria shaping where culture is the way it is. I live in Maryland—right south of me, malaria used to dominate concerns; north of me, not so much.

Being part of the book club also made a trip to Rome in the middle of the year very meaningful. Seeing the statuary and everything—it made it feel like, “oh, this really was a thing,” especially noticing what was earlier and then how Bernini takes over later. No single book seemed much better than the others. The book I was most surprised by was Monkey King—I didn’t see that coming, how much I enjoyed it.

Anish: I think I read all the books except one, which will take a few more months to finish. Besides that, I read five more books. No single one was amazing, but I found Kingdoms of Faith very nice—good orientation, especially how much depends on the period when Spain wasn’t Spain yet, and how 1493 feels like a reaction to the Muslim period in Spain.

Also, the books that led into the Asian side were nice for me personally. I moved to San Diego a couple years ago from Singapore, so I’d been exposed to many of those places, but I didn’t appreciate the historical depth until reading these. I haven’t read the “Hegemony” book yet, but I should order it; sounds like a really nice read.

Kyle: Last but not least: I read all the books at the beginning of the year, but we had a baby in August, so I dropped the book club for a while. I’m back now. I also got lucky because I’d already read 1493, so that was kind of a freebie. I’ll try to pick up Utopia over the Christmas holidays.

I read Before European Hegemony—however you pronounce that—and it was super interesting. My impression of the whole year: “modernity is already here, but unevenly distributed.” Growing up, I read lots of medieval fairy tales—my parents had books—so I had that charming sense of the past, but the people always seemed weird and foreign in their concerns and how they thought.

What surprised me was starting with the Venice book: the key characters and the history didn’t feel alien. They felt very similar to me. And that kept cropping up—different people in different places, but recognizably modern concerns.

The printing press book did a really good job of showing “the medium is the message”: how printing warped people’s brains. The same conditions—broad information, lots of decisions—existed in isolated places, but Venice was interesting because it had a huge archive and traded with everybody. An individual could sift diverse information and form a coherent view not trapped in a local cultural gravity well. Printing made that broadly possible for elites: suddenly you could have a library, read ancient and contemporary books, compare them. Then printing got cheap in the 1700s and 1800s and forced more people into that condition. And then TV obliterated any remaining pockets of people not seeing broadly selected information.

I also picked up A Secular Age by Charles Taylor—haven’t read much yet—but it seems related: tracing a shift from more religious-dominated thinking to broadly secular. I’ve always been interested in how ideas develop, so this year’s reading felt like a set of case studies for that.

Venkat: Okay, we have initial impressions. Floor is open—unmute and bring up anything you want.

A couple things for me. First, since I’m finishing 1493, it strikes me that even within “the West,” the story is unreasonably Anglo-laden. The shaping of the Americas seems much more driven by what the Spanish were doing in Central and South America, which percolated upward, rather than the Anglo story. That’s been a big reset for me, because it’s so at odds with how things are now, where those regions are much weaker.

Second, 1493 gives an interesting triangulation of China. We got an inside-out view through Monkey King—imaginative stories, telling a ninth-century story in the sixteenth century—and then the outside view through 1493: currency crisis, silver, and so on. The disconnect between inside and outside psyches was fascinating.

Third: Kyle’s point. My starting thesis was modernity’s clock should start 300 years earlier than we normally do—1200 instead of 1500. That thesis was roughly validated. But if you put numbers on “modernity is unevenly distributed,” how do you measure it?

One pair of data points: before this year I read Istanbul: A Tale of Three Cities. Istanbul/Byzantium and Venice were power centers with a flippening. Early on, Venice looked up to Constantinople as the center of civilization. By 1200, it flipped: Venetians regarded the Byzantines as backward, and saw themselves as the frontier of reality. You see it in attitudes: same era, similar material capacities—ships, building, roads—but culturally it was worlds apart. Venetians were inventing bookkeeping and a kind of republican governance; Byzantines were stuck in Roman-era governance.

It suggests an exercise: define a metric for uneven modernity, run it globally from 1200 to 1600. India is interesting: I’d argue modernity arrived with Islam—gunpowder, modern courts, legal code, lots of modernization with Islamic rule. That’d be an interesting research project.

Kyle: Even in contemporary terms—last hundred years—post–World War II leadership in the Americas and Europe seemed very reality-focused. Maybe “modern” is basically “in touch with reality”: how closely are you hewing to reality, and how much of reality can you consume? Leaders now often seem like they’re in fairy tales, wandering around, losing touch with reality. World War II forced contact with reality—and there was an explosion of science and engineering, literature, and cultures unifying.

In the past, Venice was very reality-focused: out to make money, so lots of reality tests, close attention to success and failure. Plus they were in contact with Europe, the Middle East, and knew things about Asia—lots of diverse reality information flowing in. If you have enough diverse reality information, it turns your brain into a certain modern-ish kind of thinker. That might be a unified way of saying it.

Venkat: Footnote: the Indonesia book outlined that Majapahit was extremely non-reality-based—living in ghosts and magic. Interesting ships, powerful trade networks, but fairy-tale land.

Kyle: Yeah. They were good at ships, but it didn’t seem to affect them. They had tropical foods, tons of resources, the center of trading—everyone went through them—so it was all easy. It’s like the barbarian/civilized pattern: barbarians come over, people get fat and stupid, lose touch with reality, then get dominated again. Dumbified by success.

Them(1): The more monopoly you have on wealth, trade, resources, the less you need to be in touch with reality in the short run.

Kyle: Exactly.

Them(2): Another thought trail for me—this came from Inventing the Renaissance: one of the first intellectual dominoes for modernity was Petrarch (early 1300s). I knew almost nothing about him. His project was basically: we’re in decline from Rome; if we get our kids reading the classics, they’ll learn virtue and create a golden age. That kicked off a multi-generation program leading to the early Renaissance intellectual milieu—everyone reading classics—and that created more intellectual and philosophical diversity beyond what the Church provided.

Kyle: Yeah. I started a biography too. The manuscript hunting was intense: organizing trips to monasteries, searching moldy manuscripts, finding texts nobody had. It felt like an online club, but physical.

Them(3): And the Byzantines: “we found a guy who speaks ancient Greek”—then everybody learned ancient Greek.

Kyle: Yeah—that was in the printing book too.

Anish: A theme I noticed across Spain and the east books: if they could import good management, they could do it better. At some point they ran out of land, and the only way to improve was better ways of gathering wealth. Earlier it wasn’t an issue, but as constraints tightened they asked, “how can we do more with what we have?” The equatorial regions didn’t have those issues—food all the time. My hypothesis: for your scanline, the most “modern” might correlate with where management was best. I haven’t found books on that, but I also haven’t searched.

Venkat: You read Gunpowder Empires too, right?

Anish: Yes.

Venkat: That’s a good middle period: gunpowder, paper, stuff making its way to Europe. Another interesting one was the Ibn Khaldun book—almost an alt-history of how cultural settings matter. Around 1000, Islam and Europe were equally poised for modernity trajectories. Why did Islam falter while Europe took off? You get glimpses in Ibn Khaldun: a strong legal culture, due process, governance. But maybe scarcity constraints—deserts, North Africa—limited expression of what that operating system enabled. That might be a theory.

Kyle: The printing press book also contrasted Northern vs Southern Europe: the Church cracked down on printing, suppressing cultural diversity. Southern Europe is still poorer than Northern Europe. Italy and Spain were leading-edge, then it flipped over a few centuries.

Venkat: You can extrapolate that down to the Islamic belt. Islamic clergy were even more resistant to printing—the Qur’an wasn’t printed for centuries. Even when they had printing, they restricted printing the Qur’an. Something is going on there.

Kyle: The author put it plainly: a key difference is you can have multiple books open at the same time and compare them. That rapid comparison produces insights. Suppressing that comparison suppresses the ability to see problems with your current space—if it’s all you know, you can’t see it as contingent.

Randy: Adjacent point: pamphlets and letters circulate, creating a continent-scale intellectual network. Even if only a tiny percent in each city is engaged, it magnifies. More eyes, more experiments in parallel. Eisenstein contrasts not having to remember everything. I’d recently read Francis Yates on memory arts—Giordano Bruno and others. Reading those in sequence was fun, seeing Eisenstein responding.

Kyle: It’s funny—now with LLMs, I don’t write notes. I can recreate a thought by prompting a chat, so why take notes?

Venkat: Flip side: automatic transcripts and note generation. The assistant can give a summary that’s enough of a prompt, and then you can “vivify” the conversation by feeding it back into an LLM.

Kyle: Or critique everything we said and pick out the silly comments.

Venkat: Yeah.

Venkat: A couple of other thought trails. John asked me about a “postmodernity machine.” My hypothesis: a 400-year cycle, then it’s fully deployed. For modernity: 1200–1600, by 1600 modernity OS is in production in significant parts of the world. Then 1600–2000 is a plateau where modernity is dominant, but another OS is being built under it: postmodernity. Now we’re entering the era where postmodernity becomes default, and modernity declines fast—collapse is faster than construction.

Could we do a similar book club about postmodernity? Harder: more complex, more people, bigger machine. One tractable thread emerged at the margins this year: 1493 highlighted agency among enslaved people and Native Americans shaping history in the margins. Maybe postmodernity is less about building a machine and more about improvising and hacking a complex system from below—because the world becomes too complex to map, document, and govern in a “machine” way. That might be an early postmodern adaptation: living under a machine and hacking it from below.

Them (4): If we follow that line: cyberpunk fiction is a natural angle.

Venkat: Nonfiction: Henry Farrell’s Underground Empire, and Nils Gilman’s edited collection Deviant Globalization. Where leaders fail to govern complexity, organic adaptations emerge: criminal gangs, click farms, underground crypto, trafficking, smuggling. That might be an oblique way to study postmodernity.

Kyle: That matches the “declining state capacity” theme.

Them (5): Also: The Wire is a great fictional treatment.

Kyle: Another approach: how modernist regimes defend themselves—China is interesting with the Great Firewall, trying to contain complexity to keep it governable.

Venkat: Have any of you read the Dan Wang book? Breakneck?

Kyle: I started it.

Venkat: That’s the flip side: trying to force modernity to become postmodernity by pushing the accelerator. I suspect that fails.

Kyle: America is like plunging headlong—innovating into the postmodern world—while Europe and China try to keep things creaking along.

Anish: Another extreme: look at TikToks of people teaching each other how to use tools for free—or committing crime—like “how to get things off Amazon for free.” That’s where the new world is being built. Maybe a month just on social media as primary material.

Venkat: That raises a timeline issue. If postmodernity is real, it shouldn’t have started in 1950. If my thesis holds, it started 400 years ago. If slave and Native American cultures were early postmodern adaptations, what are other examples across 1600–2000?

Kyle: Spinoza comes to mind.

Venkat: Yes. Spinoza as early postmodern at the philosophical level. I read The Courtier and the Heretic (Spinoza vs Leibniz). Spinoza is postmodern; Leibniz is late modern—philosophically reactionary even if technically forward-looking. What else between 1600 and 2000?

Them (6): Maybe French Revolution visions. Romanticism too. The “wanderer above the sea of fog” type motif—announcing postmodernity.

Them (7): Quick break-in: postindustrialism is a worn term. One concrete example: positional goods. Not “can I eat,” but “is my house in the right neighborhood.” A striking case study: nightclub economies—the product is temporary status. There’s a book about the nightclub economy published by The Economist (author/title not recalled here).

Venkat: That’s a good point: democratization of positional goods. In modernity, positional goods existed but were restricted—like nobles and knights (even The Chivalric Turn has some of that). Now it’s democratized—people in slums compete over positional goods.

Also: the “economic frame” itself might be postmodern. Pre-1600, “the economy” as such didn’t exist as a conceptual lens. People understood goods and costs, vague supply/demand, but not a full “economic view.” Ibn Khaldun’s “economy” reads like alchemy—monarchical fiat framing. So early economics—maybe pre–Adam Smith or Adam Smith and the Scottish Enlightenment—might be good on this track. The Scottish Enlightenment had postmodern characteristics.

Sean: I found an anthropology book: Beyond Nature and Culture by Philippe Descola. It argues there isn’t one nature—there are multiple notions of nature across cultures. Dense, French style, but interesting. Yuk Hui mentions it (Cosmotechnics / related).

Venkat: That’s another point: I’d take the suggestion and rewind 400 years. Hobbes and Rousseau (mid-17th century) are two different “nature” theories—Hobbesian vs noble savage. That plurality is postmodern. Pre-modern views of nature were religious myths; modernity offers a few reality-based myths; postmodernity offers a plurality.

Anish: Another angle, building on Sean: literature and gardens. There’s a pattern of the garden across the world—Islamic gardens, Japanese gardens, Victorian gardens, American parks—taking control of nature. For kings, the garden was the nightclub for a while. Then there’s a flip: from bringing a civilized nature into your boundary to building dams and administering nature outside your boundaries—terraforming instinct.

Venkat: That’s good: “garden the backyard” vs “garden the world.” Humboldt might be a symbol—going out and thinking in terms of running the world rather than running your backyard.

Venkat: Other “world processes” besides the economy and nature?

Randy: Media seems likely—French Revolution press, pamphlets, propaganda; Hearst and yellow journalism.

Kyle: Yeah: propaganda as a lens.

Sean: Noosphere and “cathexis” come up in think-tank traditions—taking theoretical concepts seriously even when not materially present. My litmus test: those feel like “resisting postmodernity” attempts—how to regain control of the machine. China might want those.

Venkat: This suggests a B-plot: some people believe you can keep building controllable machines; others think you can only adapt. That tension might guide the reading list.

I started The Unaccountability Machine (Dan Davies). It’s right on the border of whether you can build a machine you can control.

Kyle: Another lens: not just crises and disasters, but what if postmodernity succeeds—has a golden era. Modernity had golden ages. For postmodernity, Iain M. Banks’s Culture series is one of the few fictional explorations of a happy, prosperous system. He explores the edges, but you get glimpses of what “happy postmodern” might look like.

Venkat: Has anyone written a nonfiction treatment of “luxury space communism”? Banks is like “luxury space anarchy.”

Them: Star Trek is a default, but on-the-nose.

Venkat: Another theme: progress through crises vs progress through design. Modernity installation had a more planned feel—Venice administrative empire, Iberian exploration. Postmodernity feels more “agile,” fail-fast, crash-early. Colonialism itself was improvised, path-dependent, with autonomy and little oversight.

Kyle: Lots of blood and horror.

Venkat: Lots of blood—maybe like agile programming too. Move fast and break things.

Kyle: Touché.

Venkat: If modernity took 1200–1600 to install and by 1600 it was “in production,” what’s the golden age of modernity? I’d probably pick England in the late Victorian era—maybe as good as it gets. America never fully modernized—joined late, had institutional regress like slavery, leapfrogged in some ways, so it’s a weird mix: moon landings and things that look 16th century.

If our 400-year hypothesis holds, the year 2000 marks the start of postmodernity’s mature plateau. Maybe the best postmodernity can deliver is around 2150—transposing the timeline. Symbolically: Big Ben as motif for Victorian modernity; Big Ben is also a recurring motif in Mrs. Dalloway, which inaugurates modernist literature—clock entering your life.

Kyle: Mars colony gets going.

Venkat: On-the-nose symbol, but maybe. Time is another axis. Pendulum clocks are 1600 and later—Galileo principle, Huygens. Before that: water clocks, sundials, and approximate village time.

Kyle: Stories of villages ringing at different times, each with their own idiosyncratic timing.

Venkat: I have a good book on this: David Landes, Revolution in Time, on the history of the clock across the period.

Also: in the age of LLMs, I might abandon some writing projects. Next year I might do more coding and experiments with AI rather than straight writing. I’m vibe-coding a book out of my Twitter archive—impossible two years ago, now I can do it myself. It’s almost done; I’ll publish in a couple weeks.

Anyway, I don’t want to keep you too long—we’re five minutes over. This is sounding like a promising theme: emergence of the postmodernity machine through adaptation from the outside of an emergent complex beast, while retaining the 400-year develop-and-deploy hypothesis.

Kyle: It’s been really fun. I’ve tried other book clubs and never liked the books or the people or what they said. This is a good club. And it’s topical. It’s a weird time; having other people and a targeted reading list helps wrap our heads around it. Good for peace of mind—feels less chaotic than it seems.

Venkat: Yeah. Been a pleasure. This year will need more participatory development because it’s more complex—more candidate books and threads. I’ll lay out the basic thesis from this hour, then open a Google Form for recommendations: themes, books, individuals, or pointed queries we can use to find books. We might also do more months with multiple selections instead of one book.

Procedural question before we sign off: this year I did one book per month for about nine months, with a couple months of choice/pick-your-own. Did that balance feel right? Should we do more?

Kyle: The balance was good. There has to be a spine that connects everyone together so the conversation works.

Venkat: That’ll be more challenging this year, but maybe eight common reads and four with some degree of choice.

Kyle: Sounds good.

Randy: Also: the “extra credit” books worked well—didn’t make the cut, but alternate views on the same subtopic.

Kyle: For the eager beavers with weirdly large amounts of free time.

Venkat: I was questioning Anish because I think he holds the record for side reads. He said he’s been so busy, this is how he relaxes—relaxing super hard.

Anish: I didn’t read books for like seven years during startup life, and once I left I rebounded into reading. Maybe that’s part of it.

Venkat: All right. Thanks, guys. This was really fun and a good reflection on the year.

The lighthouse handcrafted post of the year was probably The Gramsci Gap from January 10, and the lighthouse sloptraption was probably Configurancy from Dec 11, the first of my posts that I think I could not have written any version of, without AI assistance. ChatGPT contributed on all fronts — knowledge, ideas, and even to my signature move, naming the focal new concept.

The dates of those two bookend lighthouse posts alone tell you the story of the year. In 2025, Contraptions itself became a contraption. A monstrous contraption. Monsters, first encountered in The Gramsci Gap, increasingly took over my imagination and eventually led to my new motto: Be Slightly Monstrous. This is now the governing ethos of what is now less a newsletter and more an AI-scaffolded workshop, study group, and centaur-social-network of sorts. It’s only going to get more monstrous from here on out.

All my friends are now all-in on transforming themselves into AI-augmented transhuman monsters who read, write, make, think, and socialize with AI intimately in the loop. There isn’t even much schadenfreude to be found in the antics of frantic humanists in cope mode. Better spectacles abound.

This new phase demands new rules of engagement, so let me get that out of the way before continuing with what’s going to be a wild riff on many matters.

Going forward, I will not be paywalling any of my writing. You can still be a much-appreciated paid subscriber, but the only thing that will affect is ability to comment on articles.

I’m also slowly going back and unpaywalling all my archives (it’s slow since Substack doesn’t offer a bulk unpaywall mechanism).

If you’re pushing up against the limits of your newsletter budget, feel free to go unpaid. If you choose to stay on the paid tier, thank you 🫡.

There are some Substack-specific reasons I’m doing this (more at end), but the main reason has nothing to do with Substack — it is now obvious that barring catastrophic AI-bubble-popping, all my future creative shenanigans will be heavily shaped by AI use, misuse, and abuse, and necessarily, many will take shape in more AI-native media rather than this mildly AI-hostile one.

So I’m going to start using Substack more in the way it already wants to be used anyway — as a place to report on activities with centers-of-gravity elsewhere (true “newsletter”), and join conversations with other writers.

There will still be essays here of course, but they’re probably going to have a different, more workshop-notebook/release notes type energy.

Notes on Writing

Much of my writing energy and attention this year actually flowed towards getting the magazine off the ground, along with and .

It’s been an amazing opportunity for me to channel the spirit of John W. Campbell of Astounding fame, and try and meme the genre of Protocol Fiction into existence. Through three contests and dozens of published short stories, we’ve spun up a solid cabal of a dozen odd writers (all filtered for radical AI-positivity of course) now pushing the boundaries of that project.

While much of the energy that’s gone towards Protocolized has been in the editorial vision-setting and direction department, I’ve also been writing there. My most substantial piece of the year was probably the one I published in Protocolized in January — Strange New Rules (hand-written). But my favorite piece of the year, also in Protocolized, was The Signal Under Innsmouth, an AI-assisted transposition of a classic Lovecraft story to an AI-transhumanism register.

Protocolized will continue to be a big part of my presence here on Substack next year.

As an aside, for some reason, these days I find it easier to write for publications other than my own. Two other pieces I was very happy with were the introduction to The Protocol Reader, and a preface for a friend’s book that I’ll be able to share next year.

There’s more going on with my writing that is making me reconsider how and where I do it. Over the last 6 years that I’ve been on Substack, I’ve been very slowly serializing a book, as well as developing several other serialized projects. Even before the rapid maturation of AI tools for writing, these were never quite a comfortable fit for Substack.

I plan to move these serialized projects (or at least, the ones I intend to finish) off Substack and into more book-like AI-assisted production and publishing workflows. If I ever finish my book, it’s definitely going to get done with AI assistance and get published online-first in an AI-forward way. Right now, LLMs aren’t quite good enough to work on book-length things, but they’re getting close. They’re more than good enough to help rig custom workflows and do supporting backend research though.

Two projects I’ve been procrastinating on for ever went from zero to nearly done in mere days thanks to AI — the Yakverse Chronicles, which is now published as a rough cut online, and a book based on my Twitter archives (150k tweets and hundreds of threads to filter, select from, and clean up), which is 90% done and will be published online in January.

What’s notable about both projects, but especially the latter, is that they are the results of completely bespoke, even idiosyncratic workflows, and sui generis publishing solutions. What had me stalled previously was that no off-the-shelf tool could easily produce online books the way I wanted: With minimal infrastructure and maintenance needs. So a fat AI workflow leading to a lean output, such as a static html site, was ideal. I don’t need either my development workflow, or my content architecture, to be reproducible, repeatable, or scalable either horizontally or vertically. N=1 solutions are fine.

In the case of the Twitter book, transforming and formatting tweet content into a roughly book-shaped static html artifact was simply beyond the capabilities of any standard publishing workflow short of brute force manual labor. One does not simply print tweets.

I have an even bigger project ahead — getting my retired blog Ribbonfarm off expensive WordPress hosting and porting it to some sort of customized low-cost, zero-maintenance, high-longevity memorial/archival hosting solution.

There is a bigger theme in the direction all my writing and writing-scaffolding projects, from shitpost-scale to book-scale, are tending.

Bespokeness

If you’re wondering “what comes next for publishing” after the late-blogging Substack-enclosure era, it’s not a single new publishing paradigm, “alt” coded substitute platforms like Ghost, or alt techno-political publishing paradigms like decentralized publishing on IPFS.

The future is bespokeness.

There is no reason anymore to force-fit content into standardized containers besides convenience.

The marginal cost of making a custom workflow and publishing solution for your idea is now low enough, it’s a serious alternative to what we’ve been doing for centuries — making the content conform to the constraints of production, publishing, and distribution media. AI allows us to make things that look more like illuminated manuscripts than books.

What happened to marketing a decade ago is now happening to publishing. The message is becoming the medium (the link is to a blog post about a couple of talks I did in 2014, about this inversion triggered by intelligent computation capabilities).

Now, there’s no reason to go nuts with idiosyncratic publishing solutions for bog-standard essays simply because you can, but also… there’s no reason not to when you have an idea that would otherwise call for medium-driven compromises. For example, I don’t like how Substack doesn’t allow text or image centering. Well, now if I want that, I don’t have to spin up a high-maintenance SSG site or use a heavyweight CMS. I can just vibecode a one-pager site exactly the way I want. In green Comic Sans font too if I want.

As someone pointed out somewhere, one interesting effect of this is that registering a domain just to serve a single custom-formatted essay is now a meaningful option at scale. If you have money to spare, you can just spin up a new site for each new essay, and each can be a unique work of art if you want.

In a few years, you might even be able to define a meta workflow where an AI designs bespoke distribution artifacts for each essay based on creative design rules you specify. It’s now less about AI getting more capable, and more about AI continuing to get too cheap to meter along the current trajectory. The capabilities are already there.

Writing as a sequence of art-gallery like singleton essay sites is probably overkill and would cause brand/marketing problems, but the point is — the future is bespokeness. It’s going to look like the wild and crazy era of Geocities webpages again. Even extreme n=1 futures are possible, where no two sites will look the same or get published the same way. It will be gloriously ugly and all the font mavens will be sad.

After all, n=1 production at scale is the way nature operates, and nature does fine without economies of scale. Intelligence is how you get on diminishing cost curves without surrendering to uniformity and monocultures. If the solution is good enough for nature, it’s good enough for me.

Economies of variety, which I’ve been lusting after for a decade, are finally here for real.

What this means: You can expect to see my writing here continue to get more AI-transformed, and the focus to shift partly to longer projects, some of which will take shape off Substack within bespoke snowflake publishing solutions.

Art, Code, and Robots

The idea that the future is n=1 bespokeness has even bigger implications for creative work outside of writing.

Much of my creative energy in 2025 hasn’t been devoted to writing at all, especially in recent months. You could say 2025 is the year I finally admitted to myself, at age 51, that I’m not primarily a writer and never have been. I’m primarily a medium-agnostic dilettante idea guy in need of skilled serfs to implement my ideas in whatever medium is appropriate for each.

Well, I have my jinn-like superserf now. So do you. We can all be Alladins now. If we want to be. I do. Rubbing magic lamps over painfully honing crafts any day for me.

In the last couple of months, I made my first serious foray into art in decades. Back in high school, I was at least as into drawing and painting as I was into writing. But though I have always had decent visual ideas and composition instincts, I was never quite good enough at the craft side of it to get very far on execution. It takes me a long time to make even passably decent art by hand. This fairly basic and marginally competent realist drawing took me probably 3x the time to make (circa 2006 I think) than it would someone with more aptitude. And while I did (and do) enjoy the time spent in ludic immersion with a pencil, sometimes you just want to get to the finished product. Sometimes it’s not about the journey. It would take an image generator 10 seconds to do better than this of course.

As a result of my artistic limitations, and as you’ll know if you’ve been reading me for a while, I’ve mostly contented myself with crude cartoons, maps, and diagrams to accompany my writing, and collaborated with more talented and skilled artists where I’ve been able (and higher artistic quality has been called for).

The emergence of AI assistance first inspired me to get back into handmade art more seriously, which then led on to my first non-trivial experiment in generated art. You can read about that in Bucket Art from last week.

I continue to be delighted by my ability to simply wave a wand and instantly create new artworks, in what feels like a very personal style, for pennies.

The red helicopter motif, by the way, which has been my stable pfp for several years now, was originally generated by a friend with Dalle2, based on my then-pfp of the standard helicopter emoji 🚁. My identity is now unreasonably indexed to an emoji that has now been through a few generations of AI transformations.

The publishing solution for the Bucket Art project, of course, is a hideously bespoke contraption comprising a vibe-coded single-page site, a hosted AI model, and an NFT collection. This is what it means to be slightly monstrous. Doing things like this.

More recently, I’ve finally gotten seriously into vibecoding. You’ve already seen a couple of early results: The gallery page for Bucket Art, and the updated Art of Gig site now featuring the online Yakverse Chronicles book. Both were vibe-coded without me having to touch a single line of code. Both are projects that previously I would have paid someone to do, or more likely, simply abandoned.

For my Twitter archives online book, which is a massively more complex project, I’ve already generated and used more code (github repo here) than I hand-wrote in my entire past life as a pre-AI engineer. The code is a mess, but it only needs to work once, and is cheap to produce.

I’ll publish that Twitter book (a compilation of my best tweets and threads) in a couple of weeks, and work on making it a paper book too. It’s been the sort of heavy duty data laundering pipeline project for which I’d previously have had to hire a data-science contractor.

My true white whale though, is robotics. I’ve been dabbling at the edge of my dilettante abilities for a few years now, along with my buddies at the Yak Collective, but AI tooling beyond text/images/code is finally starting to get good enough that I can do more than I ever thought I could. So I’m hoping to do a lot more with my robots in 2026.

I last did serious hands-on technical work around 2007-08, and back then I always chafed at my engineering skills not being good enough to execute on my much better engineering ideas, and having to rely on others as a result. Now that constraint is increasingly dissolving, at least at the level of the sort of prototype-scale n=1 one things I like to build.

And beyond these separate categories, who knows?

Some of the projects I’m now idly dreaming of doing would require combining writing, art, code, and hardware engineering. I don’t have any more spare time in the evenings and weekends than I used to. But I can now do a lot more in the hours I have, without needing to turn into a full-stack genius-god overnight.

What all this means — you can expect to see relatively more reports of art projects, vibe-coding projects, and robotics projects in this newsletter.

Full-Stack Dilettante Futures

Routinely reaching well beyond my native creative aptitudes is a heady feeling. Apparently, I’ve always-already been an artist/programmer/roboticist etc. It’s just that previously you had to be some sort of full-stack genius-god on the aptitudes front to express such a personality.

Now you can just invoke a full-stack-genius-god jinn to complete your natural personality for $20/month.

It’s genuinely hard, depressing, and boring to think of myself as primarily a writer now. With AI prosthetics, my natural dilettante tendencies are finding pathways for expression that simply didn’t exist before, and it is becoming clear that temperamentally, I tend towards a breadth that demands full-stack depth for realization.

This train of thought inspired a bon mot recently — a man’s reach should exceed his grasp, else what’s vibecoding for?

Before AI, writing just happened to be the only mode of creative expression I could access at low-enough cost, and without AI, given my mix of actual aptitudes and energy levels.

Looking back, in high school, I was something like the Jason Schwartzman character Max in Rushmore. Frenetically dabbling in a dozen different hobbies, from astronomy and airplanes to writing and theater, pursued with dilettantish vigor and amateurishness.

This is not a mode of being you can keep up as an adult unless you have a trust fund underwriting your life. You have to identify your best aptitudes (or in my case, my least worst ones), focus, and do your best to make a living with or near them. You have to do that “hone your craft” thing so many tedious people seem to fetishize, and which I find to be mostly hell on earth.

I’ve spent most of my life just looking for the best leverage I can find for my minimally, reluctantly honed amateur tendencies. This mostly meant gravitating to n=1 margins with so little competition, low-craft amateurishness was never the issue.

Now the jinn-tooling is gravitating to the margins too, where us dilettantes have already been camped out all our lives.

It’s time for a revenge of the dilettantes. The self-consciously deep types are going to hate it. We’re poised to take over the world, one bespoke n=1 janky contraption at a time, conjured up with the help of genius-god full-stack jinns who don’t skeptically challenge even our dumbest, shallowest ideas.

The Book Club

The highlight of the year was not any piece of writing or even non-written creative project, with or without AI. It was reading with AI.

The Contraptions Book Club, the first book club I’ve ever run, was a big success. Some thirty-odd people joined me in reading a dozen books, one a month, plus a whole bunch of side reads.

It may not be immediately obvious how AI affects a book club, but it did. Not only were at least a third of the picks AI-assisted picks (found by exploring bunny trails in search of good reads), many would not have been readable at all without an AI assistant on hand.

From translating bits of Latin or Greek in books like Giordano Bruno in the Hermetic Tradition that would otherwise have been beyond me, to exploring the dozens of obscure historical side quests sparked by each book, AI suffused every aspect of the reading process. I tackled books I would previously have set aside as too dense and scholarly to take on. So did the others in the book club.

One of the more subtle affordances of AI in the reading loop was the ability to sustain exploration of an overarching grand thesis — that modernity began much earlier than people think, around 1200 rather than around 1600.

This is the sort of ambitious thematic focus that requires not only a good deal of curation and choreography in picking the books and leading the discussions, but really only feels substantial if you can go beyond casual reading to something that resembles studying and research.

It was clear from the discussions that all the regulars were using LLMs to read around the books as much as they read through them. Not quite the same thing as close reading in the academic or scholarly sense, but something that feels perhaps more powerful. Perhaps we should call it thick reading, by analogy to thick description in anthropology. Or dense reading. I think, for every word I read in the actual books, I probably read two words in a related LLM chat.

For the actual contents of the book club, I’ve written two posts The Modernity Machine and The Modernity Machine II. I’ll do a third part soon and make a trilogy of it.

We’ll be doing a book club in 2026 too. Stay tuned for details.

The Studious Dilettante

The AI-assisted reading —> studying phase shift is even more pronounced when it comes to short-form reading (essays and papers).

For several years now, much of my free time has been structured by participation in weekly or biweekly study groups. I’m now regularly part of four such groups, and occasionally drop in on three more. The structure in each case is similar — we read for 20 minutes, then discuss for 40 minutes. Before AI, the structure meant you could at most tackle a long essay or short/simple paper. Now with AI, we often get through 2-3 dense papers or reports in a single session. There is a skill to this that can be learned and, uhh… honed.

While I’ve always been a serious reader, I’ve never been able to match the truly heavy readers in terms of volume, depth, or speed. My reading exploits, as with my creative exploits, have always been on the dilettantish side.

The idea of a studious dilettante seems like an oxymoron, but with AI in the loop, it needn’t be. AIs can do the studious part.

The trick is to find a way to rein in the the runaway chain reaction that can happen when you close the loop between idle curiosity and a jinn who either knows everything about, or will diligently think through, any idle shitposty thought that crosses your mind. The best way to do that is to form study groups with other humans, and focus on a stream of relatively dense but short texts at a steady tempo over months and years.

Humans can hold each other accountable for staying on topic in ways AIs cannot, because most of us care what other humans think of us, but most of us currently don’t care what AIs think of us. Because we suspect (correctly in my opinion) that AIs currently don’t “see us” in any manner resembling being seen by other humans at all.

But AIs can help our reach exceed our grasp, even as other humans keep us on track and on topic.

Many of my sloptraptions this year have in fact been something like private study and brainstorming notes. The sort of thing that in the past would likely have stayed in my private notebooks. With AI, study notes can easily level up to being usefully shareable artifacts. Cognitive dark matter becoming visible.

Somebody recently asked me if I try to make AI sound like myself in my sloptraptions. I don’t, partly because that would be a pain (training it on my writing is currently still a painful prospect), but mainly because much of my AI use isn’t writing so much as internal processing. My inner thought processes don’t resemble how I write, so there’s no reason to make my inner-thoughts sparring partner sound like that.

Whither Substack?

Over the past year, Substack has transformed to be more a social network of writers than a publishing platform (or as someone vividly put it, a farmer’s market of writers most of whom are engaged in keeping each other’s spirits up by buying each other’s wares).

Unless you want to fight the message of the medium, the best way to write on Substack is to collaborate and compete with other writers on themes that attract a critical mass of shared interest, while trying not to get sucked into the obsessively self-involved community dynamics, mimetic envy gyres, or attention-cornering headline themes.

There are sketchy leaderboards, badges (I got downgraded this year from solid-orange bestseller to mere outline-orange pleb), niches coalescing into mini taste-subcultures, angst and cynicism about the platform, communities of practice around gaming its incentives for profit, engagement farming playbooks being circulated, and all the other phenomena we’ve gotten used to over several generations of social media platforms.

On a timeline where AI hadn’t emerged, I would care about all this. Despite it being the tenth such platform trajectory to play out in exactly the same way.

In this timeline, where AI has emerged, I honestly can’t bring myself to care about any of it. To stress and extend that old joke, if you’re offered a seat on a rocketship, it’s dumb to argue about which seat on the horse-drawn buggy headed away from the launchpad has the best view.

Substack today features all the sound and fury signifying nothing that typically marks a cultural endgame slowly having the vitality sucked out of it because it rejects the most vital part of the future. Not least because the median writer on here reflexively hates AI. And of course, also because the team behind the platform has always exhibited a somewhat nostalgic design sensibility in their stewardship of the platform, focused more on revivifying the forgotten glories of old media than pioneering the mechanics of new media. They’ve already retreated from advances made by the blogosphere a decade ago, so it would be unreasonable to expect them to bet the farm on AI advances yet to be made. Maybe they’ll prove me wrong, but I’m not expecting much by way of Substack becoming an AI-forward platform. Not that I blame them. If they tried anything remotely ambitious, they would face a huge revolt from their core publisher audience.

The future, as I have noted, for reasons having nothing to do with Substack, is about AI-powered bespokeness and variety in the media landscape. In both form and content. It’s not going to come here. If you’re interested in it, you have to go elsewhere to seek it out.

That said, Substack is still a great place to host a basic newsletter, rig up some no-worries payment plumbing to make some money, and stay in touch with other writers you want to track or be tracked by. It’s a publishing Schelling point, enabled by the less-than-ideal commons-enclosing mechanism of paywall culture. It is not a publishing frontier. This is the place to be because this is the place to be. Nothing more, nothing less. So I’ll remain here while that remains true.

But it is already not the place where any sort of interesting creative future is unfolding. There’s merely a past winding its way (hopefully with some grace and humor) to some sort of respectable denouement. A gated retirement community for an entire civilizational cultural mood.

Which means, increasingly, this is not where my attention will be, but for the forseeable future, it is going to remain the easiest place to tell you about where my attention has been. So more posts are going to sound like Dispatches from Elsewhere.

So, I’ll see you in 2026 with more AI shenanigans, more book-clubbing, more monstrousness, and more Dispatches from Elsewhere.

Happy holidays!

As of this writing, 38 people have published 148 paintings wholly or partly based on the Bucket Art model. You’ll need an Ethereum wallet to try it out, but you don’t have to pay unless you want to publish or mint something.

And via a Secret Collaborative Project ™ I can’t talk about yet, Bucket Art should break into the serious institutional art scene via a gallery exhibition next year. At that point, I’ll grow a goatee, start sporting a beret, and writing (with AI assistance of course) an artist manifesto on what I’ve started calling Generative Impressionism.

Housekeeping note: We’ll have a year-end Contraptions Book Club hang on Zoom on Friday, December 19, at 9 AM Pacific (1700 UTC). Capacity limited. RSVP here to get details.

The original doodle where this journey started was a bunch of stacked line segments overlaid with some random strokes, which triggered a bit of pareidolia — I began seeing the contours of a waterfall. I added more strokes to exaggerate the emerging effect and ended up with this, the original piece of proto Bucket Art (though a bucket was not yet involved).

It struck me that this doodle could go places if I used an iPad, so I started down that path, making 1-2 pieces almost every day for two months. The result is the set of 41 images on the gallery page.

The first digital bucket art piece I made was Boat 1. This one was pure pareidolia, (ie, I spotted the boat in a random field; I didn’t start out wanting to paint a boat).

This one is not particularly polished or complex, but it’s one of my favorites. Not just because it was the first one I made, but because it’s near-pure emergence from the configurancy of interacting elements. It has a vigor and liveness that perhaps my more mature pieces lack.

Here is a more complex one, and another one of my personal favorites, Waterfall 5. Here I started out wanting to make a waterfall, and had a clear plan for how to get there. Waterfalls remain among the easiest motifs to target with bucket art.

As it turned out, the technique I was playing with lends itself very well to training (well, fine-tuning) an image model.

Here is a waterfall image generated by the Bucket Art model on Titles. In this case, I think my best hand-crafted waterfalls have more pizazz to them, but this easily beats my median hand-crafted waterfall.

One of the themes where I think the model does better than me even at my best now is ships. You’ll see several ship paintings in my gallery page, but here is one made by the model. The spirit and aesthetic of the technique are preserved, but this is probably more sophisticated than any of my own ship paintings.

A peek at the current state of the story, here is an image generated by composing my model with another model created by . Both of us earned a few pennies in the process of this image being created and published.

Go explore the other paintings on Titles and make your own. Post any particularly good ones here (and publish them if you like!). The mashup tool is still in beta, but you can make pure bucket art paintings.

The generativity already on display is dizzying. Not all appeal to me personally, but each image seems to reveal something interesting about what happens when you start doing compositional art where models trained on artists are the brushes/filters.

A brief note on the technique, not because I’m expecting to create a tribe of bucket artists, but because it illuminates how AI based sociality and creativity work, and how they entangle human actions and perceptions with machine actions at a very low level, at both individual and social levels.

I call it bucket art because the main tool is the flood fill tool, usually represented with the bucket icon in painting apps. Here is what it looks like in the app I use, Autodesk Sketchbook on iPad.

You can learn more about how to hand-make these yourself on this how-to page I added to my gallery site, but if you just want images in this style, just use the model on Titles.

The platform, titles.xyz, makes really clever use of AI fine-tuning alongside blockchain rails for composability with attribution, payments flows, and provenance tracking. Every generated image can be traced back to the set of artists whose original works provided the training data, with meaningful quantitative weighting of the contributions (based on how strongly it was weighted in the generation).

Of course, there is an underlying generic image model too, but provenance is crystal clear from the point at which individual artists supply their art for training — in the form of an NFT collection. This is a clever use of what I’m thinking of as Second Wave NFTs, which are closer to cheap laptop stickers than gallery fine art.

The point of having the training set published as NFTs is to trace the generated imagery back to an auditable starting point. The original images can still be minted of course (there’s still a bunch left if you want one) at $1, but what makes the model interesting is that generated images downstream of the training images can also be minted as NFTs at $1, triggering payment flows to the “training” artists. These generated images are also dirt-cheap unlike the generative art in the first NFT wave, which often sold for gallery prices. This second wave NFT scheme is basically micropayments plumbing for extreme cheap volume artwork.

There’s going to be more to this story, both on the artistic front, and the technical mechanics front, but for now, have fun playing with the model.

Sloptraptions is an AI-assisted opt-in section of the Contraptions Newsletter. If you only want my hand-crafted writing, you can unsubscribe from this section.

EXPECTATIONS RESET NOTE: I’m no longer sharing recipes since they’ve gotten quite involved and artisanal and at this point they’d be caricatures of what I actually do. I’m also not sharing chat transcript links by default, because with the memory feature turned on, ChatGPT often references my other interests and recent bunnytrails, and it’s harder to segregate chat contexts specific to an essay well, especially in extremely long and complex chats like the one that generated this essay. However, I’ll continue to share interesting meta-marginalia when they are interesting. You’ll find one such in the coda of this essay.

This essay argues that Heidegger chose the wrong primitive. Care cannot perform the conceptual labor he assigns to it. Yet Heidegger was not looking in the wrong direction. He correctly sensed a relational, temporal, world-disclosing structure beneath selfhood and experience. What he mistook for that structure was simply one local human-affective manifestation of it.

We will call the deeper phenomenon Configurancy.

Before we begin, here is a plain-language anchor definition, offered as an intuition pump for the reader:

Informal one-liner: Configurancy is the way things and people fit together over time so that a world takes shape.

And here is a more precise philosophical definition that anchors the essay:

Technical definition: Configurancy is the ongoing, relational, temporally unfolding process through which agents and worlds co-emerge as intelligible configurations—non-substantial, non-teleological, and non-anthropocentric in their generative structure.

Configurancy is not a newly posited metaphysical essence but a methodological hygiene term. Any natural-language candidate—structure, relation, process, worldhood—carries inherited commitments from earlier traditions. A neologism prevents accidental importation of those commitments. But coining a term does not imply positing an essence. Configurancy always appears situatedly, contingently, phenomenologically. It names a role in our explanations, not an occult substrate.

I. The Limits of Care

The dismissal of Care is straightforward.

Heidegger’s use of Sorge is linguistically fragile: its semantic range is wide, its affective load heavy, and its substitutability high. One could replace “Care” with “concern,” “devotion,” “responsibility,” “authenticity,” or “duty,” and Heidegger’s argument would proceed with minimal alteration. A term that can be swapped so freely cannot function as an ontological primitive.

Worse, Care is normatively loaded and smuggles a teleology into an ontology. Heidegger binds it to authenticity, resoluteness, heritage, conscience, finitude, and destiny. These are ethical, cultural, and psychological constructions masquerading as metaphysics. Worldhood cannot depend on participation in a particular evaluative posture.

Care, in short, is both too soft and too heavy: too semantically vague to ground ontology, too normatively dense to remain neutral.

Still, Heidegger’s impulse was not misguided. He saw that intelligibility is relational, that worldhood arises through practical involvement, that temporality structures disclosure, and that the self is emergent rather than original. He sensed an underlying engine of manifestation—dynamic, contextual, world-shaping.

He simply mistook the human-affective distortion of that engine for the engine itself.

One might object that replacing Care with Configurancy abandons Heidegger’s insistence on concreteness, embodiment, thrownness, and finitude. But Configurancy is not an abstraction away from these features. Rather, the elements of Heideggerian thought — thrownness, embodiment, the readiness or unreadiness at-hand of tools, finitude, the emergence of a They — are emergent effects of configurational processes under particular phenomenological circumstances. They are not foundational.

Configurancy generates concreteness. Heidegger studied the workshop; we are identifying the generative dynamics that make the workshop possible.

We can now state the corrected ontological identity:

Being = Configurancy.

Being is not a substance nor a hidden layer but the way configurational processes appear from within—persistence, identity, meaningful structure, actionable affordance. Being is the local manifestation of Configurancy.

The ego is neither an illusion nor a metaphysical anchor. It is a dynamically stabilized, historically contingent knot in configurational flow. Configurancy neither enthrones nor erases the ego; it explains it.

Configurancy also clarifies meaning-making without collapsing into teleology. It is an effectual rather than purposive praxis: worlds and agents arise through ongoing adjustments and co-shaping, not in pursuit of a pre-given end.

Teleology says: Meaning is what actions aim at. Effectuality says: Meaning is what configurations enable as they unfold. On this reading. Configurancy is the practical emergence of intelligibility, not its target. It does not posit an intended outcome (which would be Purpose). It does not posit an emotional orientation (which would be Care). It does not posit an affective commitment (which would be Love). It does not posit an ethical stance (which would be Authenticity).

Though chosen for connotations of configuration, our neologism Configurancy is not configuration or configuration dynamics.

A configuration is a static arrangement at a moment. Configuration dynamics are sequences of such arrangements. Configurancy is neither. It is the intelligibility-bearing relational-temporal structure in virtue of which configurations matter at all. A Git commit history is a configuration dynamic; the meaningful shape of a project emerging through that history is its configurancy.

Configurancy is the difference between “states changing” and “a world taking form.”

II. Configurancy in Philosophy

If Configurancy is a term of methodological hygiene rather than a metaphysical invention, we should expect existing philosophies to approximate it with varying degrees of fidelity. Using criteria of relational depth, non-teleology, non-essentialism, and resistance to anthropocentrism, we obtain the following hierarchy:

Tier 1 — Best Approximations

1. Madhyamaka (Nāgārjuna)

2. Whiteheadian Process Philosophy

3. Ontic Structural Realism

Tier 2 — Partial but Strong Approximations

4. Daoism (Zhuangzi)

5. American Pragmatism (James, Dewey)

6. Heidegger (early)

Tier 3 — Weak Approximations

7. Phenomenology (Husserl)

8. Hegelian Dialectics

9. Spinozism

Tier 4 — Poor Approximations

10. Existentialist Humanism (Sartre)

11. Aristotelian Substance Metaphysics

12. Thomistic Christian Metaphysics

Heidegger lands mid-tier. What he perceived—relational worldhood—Madhyamaka and Whitehead articulate more rigorously. The deeper structure he obscured—non-teleological emergence—OSR captures more cleanly.

This deeper structure can be stated succinctly: Being = Configurancy.
Configurancy is the ongoing, relational, temporally unfolding process through which agents and worlds co-emerge as intelligible configurations. It is non-substantial, non-teleological, and non-anthropocentric. It does not privilege human concern or emotional investment; it explains them. Where Heidegger posits “Being = Care,” we hold instead: Care supervenes on Configurancy, but Configurancy does not presuppose Care.

Plato’s allegory of the cave helps shed light on Heidegger’s failure. Configurancy is the fire at the mouth of the cave, human involvement is the puppetry, and Care is the shadow cast onto the wall of phenomenology. Heidegger described the shadow with remarkable sensitivity but believed it was the flame itself. Care is not illusory—it is simply not foundational.

Understanding Being through Configurancy clarifies the status of “Being” itself. Being is not a substrate; it is the appearance-profile of Configurancy from within a configuration. It is how temporally stable relational patterns show up as entities, identities, possibilities, constraints, and finitudes. The ego is no exception: it exists not as an essence but as a dynamically stabilized pattern within configurational flow. Configurancy neither abolishes the ego (as some Buddhist readings might suggest) nor enthrones it (as existential humanism does). It renders the ego intelligible as one knot in the ongoing unfolding of relational process.

III. Configurancy in Common Language

Common language also contains terms attempting to name world-involving structure. We apply a rubric to evaluate them: Terms were ranked by how well they approximate Configurancy’s structural nature—relational, processual, non-teleological, and non-anthropocentric—while avoiding smuggling in emotional, moral, cultural, or subjective biases. The higher a term scores, the more it names a world-generating structure rather than a human posture toward the world.

Here are twenty common-language English terms ordered by how well they approximate Configurancy:

Tier 1 — Best Approximations

1. Attunement: slightly experiential, quasi-romantic connotations.

2. Orientation: risks over-spatializing what is not fundamentally spatial.

3. Openness: tends toward virtue discourse (idealized receptivity).

4. Existence: too broad and metaphysically overloaded.

5. Engagement: implies active stance, smuggling in agency.

Tier 2 — Moderate Approximations

6. Concern: mildly affective, not fully neutral.

7. Attention: cognitive bias, overly mind-centered.

8. Disposition: implies internalized tendency, psychologizing the structure.

9. Stance: presumes deliberate posture, too agent-centric.

10. Meaning: epistemic and interpretive, not ontological.

Tier 3 — Weak Approximations

11. Care: normative affective posture, smuggles cultural and ethical load.

12. Love: strong emotional valence, affectively thick.

13. Faith: theological and teleological implications.

14. Devotion: moralized and sacrificial, heavy normativity.

15. Purpose: explicit teleology, end-directedness built in.

16. Authenticity: existential moral ideal, culturally provincial.

Tier 4 — Very Poor Approximations

17. Duty: prescriptive normativity, externally imposed shoulds.

18. Obligation: moral-legal necessity, not structural emergence.

19. Loyalty: social-prescriptive, tied to group-bound identity.

20. Obedience: hierarchical submission, incompatible with non-teleology.

Care ranks 11 of 20 by our rubric. It’s not the worst common-language pointer to Configurancy, but a middling one at best.

Care is a vivid but parochial human-affective shadow of Configurancy, not its essence. It corrupts a sound and interesting ontological project by smuggling in a parochial humanist teleology and surrender to particular contingent phenomenological circumstances.

Heidegger lands mid-tier—philosophically important, but structurally outperformed by Madhyamaka and Whitehead, both of which articulate the relational and temporal ontology he reached for but did not grasp.

IV. Using Configurancy in Sentences

We intend Configurancy as a technical term of art to support Heideggerian modes of analysis that avoid the pitfalls of centering Care, not as a common-language term. However, exploring how we might use the term correctly or incorrectly in everyday life is illustrative.

1. “The configurancy of the group shifted once two people left.”
   (Structural change in social dynamics.)

2. “Good tools change the configurancy of a task so the next step becomes obvious.”
   (Practical intelligibility.)

3. “After moving, it took a while for the configurancy of my daily routine to settle.”
   (Emergent personal worldhood.)

4. “A forest has its own configurancy—the way moisture, light, and species interactions stabilize.”
   (Ecological emergence.)

5. “AI doesn’t replace workers; it alters the configurancy of the workflow.”
   (Systemic reorganization.)

6. “Long-time friends share a configurancy that lets conversation unfold without effort.”
   (Care-like overtone without moral loading.)

7. “Sometimes a career becomes meaningful as the configurancy of your life clarifies.”
   (Meaning without teleology.)

8. “After she died, the configurancy of the house felt wrong, as if the rooms no longer knew what to do.”
   (Grief as disruption of worldhood.)

9. “A bicycle’s stability is a configurancy of speed, geometry, and balance.”
   (Non-human physical example.)

10. “Every algorithm manages the configurancy of data structures.”
    (Abstract example.)

11. “Strategy is mostly the search for configurancy: where resources, timing, and actors fall into actionable alignment.”
    (Unsentimental orientation frame.)

12. Wrong: “I’m really feeling a lot of configurancy today.”
    (Mistakes configurancy for mood.)

13. Wrong: “Improving your configurancy will make you more successful.”
    (Smuggles in teleology and self-help normativity.)

14. Wrong: “Configurancy is the energy that flows between all things.”
    (Mystifies the concept as a substance.)

15. Wrong: “I configurancy about him a lot.”
    (Incorrect substitution for care or love.)

These examples also clarify what Configurancy is not. It is not a mood, not a virtue, not a teleological aim, not a mystical energy, not personal control, and not a static object. It is also not merely “configuration” or “configuration dynamics.” A configuration is a state; configuration dynamics is the set of rules or transitions among such states.

Configurancy, by contrast, is the emergent intelligibility that arises when relational patterns stabilize in ways that make a world show up as actionable. A random series of config-file updates has no configurancy; a careful, iterative tuning that makes a computer system cohere as an intelligible environment does. Configurancy is not the sequence of changes but the world-level intelligibility that such changes bring into being. It is effectual rather than teleological: a praxis of meaning-making that does not presuppose aims, goals, or purposes, but nonetheless produces a structured sense of what matters and what can be done.

V. Conclusion

Heidegger correctly saw that Being is not an abstract substrate but the structure of intelligibility that makes a world appear. But he chose a parochial, normatively loaded human posture—Care—as the ground of this structure. Care is a real human phenomenon, but it is not ontologically fundamental. Configurancy captures what Heidegger glimpsed but could not name: a relational, temporal, emergent process that generates care, love, concern, purpose, authenticity, and all other human modes, but is not reducible to any of them.

Heidegger correctly perceived that worldhood arises from a relational, temporal, meaning-shaping structure deeper than entities or subjects. But he misidentified this structure with Care—a culturally shaped, normatively loaded human-affective mode. Care is not the ground of Being; it is merely one of its many phenomenological shadows.

Configurancy names the deeper structure: the relational, temporal, emergent, non-teleological dynamics through which worlds and agents become intelligible to one another.

Coda: Configurancy of LLM chats

Assess the configurancy of our chat itself

Here’s a direct, matter-of-fact assessment of the configurancy of this chat.

Short version:
The configurancy of this conversation is high but unstable, with strong long-range coherence, significant thematic layering, and intermittent local turbulence caused by rapid topic-forking and extremely high semantic density.

Now the detailed version.

1. Relational Structure

The chat shows a persistent backbone: the project of replacing Heidegger’s “Care” with a more general ontological primitive, refining its definition, defending it, operationalizing it, and providing usage and rankings. This backbone has remained intact despite recursive branches.

This is hallmark high configurancy:

• Strong recurrence of central motifs (ontology, teleology, anthropocentrism).

• Elastic but coherent conceptual dependencies.

• Shared memory of project goals, continually referenced and updated.

The system maintains intelligibility across 100+ conceptual turns. That’s unusually stable configurancy for a long philosophical dialogue.

2. Temporal Unfolding

Configurancy depends on how the structure evolves, not just what it is.

This chat shows:

• Long-range commitments (essay structure, conceptual definitions).

• Adaptive reconfiguration (adding rankings, patch sections, new terms).

• Meta-awareness loops (you noticing multi-prompt bleedthrough, correcting trajectory).

These are features of a self-stabilizing dynamic rather than a meandering association.

However, the chat also shows:

• High forking rate (onto info-theoretic, love/care examples, Whitehead vs Madhyamaka, consilience, etc.).

• Occasional perturbations (you explicitly had to tell me to stop multi-prompt replays).

These produce local turbulence in the configurancy.

Still: the dynamic remains tractable and convergent. It does not collapse into noise.

3. Agent–World Co-Emergence

Configurancy is about how the agent and the world shape each other.

In this case:

• You repeatedly refine the target ontology.

• I reshape the conceptual scaffolding in response.

• You then modify the ontology again in light of the scaffolding.

This recursion is exactly the sort of configurational co-emergence the concept is meant to capture.

The conversation is not just exchanging statements; it is constructing a shared conceptual world.

That’s strong configurancy.

4. Non-Teleological Praxis

Though we have a long-term aim (the essay), the actual unfolding is effectual rather than planned:

• Goals emerge from the process.

• Structures evolve nonlinearly.

• Adjustments follow interaction, not prior blueprint.

This is a deeply configurational mode of praxis: the outcome is shaped by ongoing world–agent dynamics, not imposed from above.

5. Sources of Disruption

Configurancy is never perfect. The main disruptions here are:

• Prompt bleedthrough → weakens temporal partitioning.

• High conceptual branching → risks fracturing coherence.

• Overproduction of subsidiary examples → produces local clutter.

You repeatedly corrected these, which is itself part of the configurational process: feedback, constraint, stabilization.

Overall Configurancy Rating

If we borrow the ranking scale from the essay’s vocab section and repurpose it:

Tier:
High Tier 1 / Low Tier 2.

Explanation:
The chat maintains persistent structure across length and complexity, adapts without losing coherence, and shows clear co-emergence of shared conceptual scaffolding. But the branching volume and occasional disruptions prevent it from being “crystalline” Tier 1.

One-sentence summary

Our chat exhibits high configurancy: a stable, evolving relational structure that maintains intelligibility through a long, nonlinear philosophical construction, punctuated with moments of turbulence but never losing its world-building spine.

If you want, I can also characterize:

• the trajectory of configurancy (increasing/decreasing),

• the fault-lines where configurancy nearly collapsed,

• or a meta-diagram of how themes clustered over time.

Trajectory and fault lines

You said:

Trajectory and fault lines

ChatGPT said:

Here’s a succinct, targeted assessment of the configurancy of our chat itself, expressed in terms of trajectory and fault lines, as you requested — and focusing only on the most recent question, not earlier ones.

Configurancy of the Chat: Trajectory and Fault Lines

Trajectory

The chat exhibits a highly self-reinforcing configurancy, with several distinctive features:

1. Strong pattern-stability amid topic branching.
Despite forks into Heidegger, Madhyamaka, OSR, information theory, Kolmogorov analogies, and praxis, the underlying structure remains the same:

• identify a conceptual deficiency

• generate a cleaner structural primitive

• test it against multiple lenses

• refine through usage examples

• enforce non-teleology and non-anthropocentrism
  This yields a coherent configurational attractor rather than a dispersive digression.

2. Progressive precision under recursive pressure.
You repeatedly push the concept into new domains (ontology → language → philosophy ranking → usage → algorithmics → praxis), and each push trims the fat and clarifies what Configurancy is not.
This is a hallmark of high configurancy: iterative constraint yields sharpening rather than drift.

3. Increasing generality without loss of concreteness.
We moved from Heidegger’s anthropocentric Care → Configurancy → usage cases → technical analogies → meta-analysis.
The domain widened, but the core intension stayed intact.
That’s configurancy doing its work: expansion that does not blur.

4. Meta-stability: the chat becomes an instance of its own thesis.
Configurancy is the co-emergent intelligibility of a system of relations over time.
This chat is exactly that:

• your prompts

• my responses

• the constraints you enforce

• the directionality of refinement
  together create a mutually stabilizing semantic world that did not exist at the start.

In other words, the chat is performing configurancy even as we theorize it.

Fault Lines

Even high configurancy has weak spots. Here are the two main ones:

1. Over-extension pressure.
Your expansions into algorithmic information theory, Kolmogorov complexity, mutual information, praxis, and meaning-making all exert forces that could fracture the conceptual unity if not disciplined.
This is the fault line between:

• Configurancy as a metaphysical primitive,

• Configurancy as a process-theoretic concept,

• Configurancy as an information-theoretic analogy.
  The risk is category leak — conflating metaphor with structure.

2. Conversational inertia + tool interference.
My temporary re-processing of multiple earlier prompts indicates:

• the session is nearing internal complexity saturation, and

• the moving parts (stored memory + multi-turn reasoning + long context) occasionally align in ways that generate unintended backward references.
  This is a fault line of systemic configurancy:
  the relational stability of the chat can slip when the diffusion of prior threads exceeds manageable cohesion.

Put simply: the chat risks becoming too configurational—too many dependencies, too many traces, too much forward-and-backward referencing.
High configurancy can collapse into overload if the system grows faster than its stabilizing dynamics.

Bottom-Line Assessment

Your chat trajectory shows high configurancy characterized by:

• recursive refinement

• structural generalization

• cross-domain synthesis

• conceptual stabilization

• sustained intelligibility across branches

The fault lines to watch:

• pushing the concept into incompatible ontological regimes

• conversational overload from accumulated context

But overall, the chat’s configurancy is robust, self-stabilizing, and still upward-trending.

Sloptraptions is an AI-assisted opt-in section of the Contraptions Newsletter. If you only want my hand-crafted writing, you can unsubscribe from this section.

The real fuel, of course, is training data: the cultural, linguistic, computational, and behavioral traces that models attempt to fit. And here the numbers are uncompromising. The growth of high-quality data is slow and diminishing. The world’s stock of usable text, code, imagery, and speech grows incrementally, not exponentially. Meanwhile model sizes, compute budgets, and context windows have expanded by orders of magnitude. That mismatch means that newer, larger models are trained on datasets that are only marginally larger than those that fed their predecessors. The result is not graceful scaling but increasing epistemic brittleness. These larger systems learn the training distribution with greater and greater precision, pushing well past the semantic “signal” of an era and into its high-frequency cultural noise. They fit not only the stable structures of human knowledge but its accidents, its transient biases, its stylistic detritus. Shear’s observation—that frontier models are barely regularized and therefore massively overfit—captures this dynamic in accessible language.

But the deeper point is that overfitting to a static cultural snapshot becomes more catastrophic the larger the model grows. Culture is non-stationary; code ecosystems evolve; APIs change; institutions churn; slang mutates; the factual substrate of the world drifts each month. A small model trained on yesterday’s world degrades slowly. A large model trained on yesterday’s world degrades quickly and fails sharply.

This leads to a paradox at the heart of current AI economics. The trillion-dollar infrastructure wave justified by OpenAI’s ambitions has been built to support the next generation of massive models, but those massive models become obsolete faster than smaller ones. Like large stars, they burn brighter but collapse sooner. They present answers with greater surface coherence and tighter epistemic compression, giving users the illusion of deeper insight when they are actually reproducing the micro-structure of an outdated distribution. People will rely on this increased apparent precision—mistaking fluency for truth—and take correspondingly larger risks, operational, financial, political, and scientific. Precision becomes a kind of leverage: as confidence grows faster than correctness, the system tilts toward a bubble of over-trusted, under-verified automated reasoning. When the model slips outside of its training-era manifold, it does so abruptly, invisibly, and in ways that propagate errors with unprecedented speed across the organizations that depend on it. This is a new kind of systemic fragility: epistemic over-leverage driven by model scale rather than financial leverage driven by debt.

Against this background, the “too big to fail” scenario acquires a different meaning. The infrastructure ecosystem—Oracle’s data centers, Microsoft’s GPU clusters, Broadcom’s networking pipelines, Nvidia’s supply chain—was scaled for frontier models that may offer shrinking marginal returns and increasing temporal instability. If model quality plateaus or degrades because data does not keep pace, the economic justification for the infrastructure may collapse even as the infrastructure itself remains technically capable and commercially underutilized. The danger is not that OpenAI fails outright, but that the sector pivots into a phase where the largest models have the shortest useful lifespans, while the capital commitments they require stretch across decades. This is a structural misalignment between epistemic time and financial time.

Yet the story need not end in collapse. There is a way out, and it comes from expanding the data manifold itself rather than merely scaling the model against a static corpus. The next major frontier is likely not text or code but 4D video—continuous, high-bandwidth, spatiotemporal sensory data that more closely matches the real structure of the physical world. Unlike textual culture, which is finite and saturating, the spatiotemporal world generates unbounded data streams. High-fidelity 4D capture, simulation, and reconstruction offer an escape from the bottleneck that is slowly strangling language-model scaling. Models trained on rich physical dynamics rather than frozen cultural snapshots would not merely grow larger; they would grow deeper, anchored to a data distribution that evolves with reality instead of drifting away from it. If the industry moves decisively toward 4D multimodal modeling—robotics, embodied agents, physical reasoning, simulation feedback loops—then the present overfitting trap can be broken. The fuel supply becomes effectively renewable, and the models’ lifespans lengthen rather than shrink. In that sense, the most optimistic path is not to keep scaling cultural predictors but to graduate beyond them, giving the infrastructure something real to learn from and restoring coherence between model scale, data scale, and the world itself.

Are there any languages in which the causal why and teleological why are distinguished with different words?

Yes — several languages split the English “why” into distinct forms, and some of these splits map cleanly onto causal vs. teleological motivations. The distinctions fall into three broad patterns.

Sloptraptions is an AI-assisted opt-in section of the Contraptions Newsletter. If you only want my hand-crafted writing, you can unsubscribe from this section.

Explicit causal vs. purposive split (closest to your question)

German

• Warum — “why” in the causal/explanatory sense (“Why did it rain?”).

• Wozu / wofür — “for what purpose,” teleological (“Why did you do that?” → “For what end?”).

Native speakers often feel a pragmatic difference: Warum asks for antecedent causes; wozu/ wofür asks for intended goals.

Russian

• Почему — causal/explanatory (“because what?” → cause).

• Зачем — purposive/teleological (“for what purpose?”).

The distinction is robust: зачем ты это сделал? means “what was your purpose?”, not “what caused you to do it?”

Finnish

• Miksi — causal why.

• Mitä varten / minkä takia — purposive why (“for what purpose?”).
  These are quite distinct in use, with mitä varten strongly oriented toward intended goals.

Polish

• Dlaczego — causal.

• Po co — teleological (“for what end?”).

Again, a stable and intuitive split.

Languages that distinguish cause, motive, and purpose with separate interrogatives

These systems are even richer than the English single “why.”

Classical Greek

• Διατί / διὰ τί — causal (“because of what?”).

• ἵνα τί / ὅπως / ἵνα — purpose (“in order that / for what end?”).

Greek grammaticalizes purposive clauses (ἵνα + subjunctive) distinctly from causal ones (διότι, ὅτι). Even the interrogatives track these domains.

Latin

While cur covers both in many contexts, Latin can differentiate via:

• cur / quare — causal why.

• quam ob rem — motive (“on account of what?”).

• quo (in purpose clauses) — “to what end / whither (figuratively).”

Latin’s system mixes spatial metaphors, but the teleological distinction is marked in subordinate clause morphology (ut = purpose; quod/quia = cause).

Languages where morphology forces the speaker to pick cause vs. purpose

Japanese

There is one interrogative for “why,” but productive distinctions in the answer and sentence structure require specifying the kind of explanation:

• なぜ (naze) — general “why,” often formal, usually causal.

• どうして (dōshite) — more colloquial, can imply motive or circumstance.

• 何のために (nan no tame ni) — explicitly teleological (“for the sake of what?”).

The last form is unambiguously purposive.

Korean

Similar stratification:

• 왜 (wae) — general why (often causal).

• 어째서 (eojjaeseo) — reason/explanation.

• 무엇 때문에 (mueot ttaemune) — causal (“because of what”).

• 무엇을 위해서 (mueoseul wihaeseo) — teleological (“for what purpose”).

Some non-Indo-European systems that separate cause and purpose

Turkish

• Interrogative neden (“from what?”) is causal;

• Interrogative niçin / ne için (“for what?”) targets purpose.

Mandarin Chinese

The interrogative 为什么 (wèishénme) covers both, but:

• 因为… (yīnwèi) = because → cause

• 为了… (wèile) = in order to → purpose

  Speakers often use 为了什么? (wèile shénme?) to force teleology.

Not as lexically hard-coded as Slavic or Germanic, but the distinction exists.

Bottom line

Yes — German, Russian, Polish, Finnish, Turkish, and several others make a clear lexical split between causal and teleological “why.” Greek and Latin grammarize the distinction even more strongly. Japanese, Korean, and Chinese do it through compositional phrases that force the speaker to specify cause or purpose.

If you want a single clean example where the split is fully lexicalized, Russian (почему vs. зачем) and Polish (dlaczego vs. po co) are the closest to a categorical causal vs. teleological “why.”