Happy Spring, beloved -
I just want to give you a heads up from the outset that this week's dispatch consists of a long exploration of a single, arcane topic, and that topic may not be of interest to more than a handful of you. If all you want is to hear about my upcoming Stateside tour, or find out what I'm reading and listening to, I won't be hurt in the slightest if you skip forward to the end. Otherwise, though? Settle back, pull up your cushions (or whatever else you have with you that makes life bearable in Texas)...'cause this is gonna go on for awhile. Here we go.
§ Eliezer Yudkowsky. Say it out loud; take the time to roll those seven rather ungainly syllables across your tongue.
How is it possible that this name — that of the individual responsible for one of the most influential technological discourses of our epoch — is entirely missing from the index of a book about emergent technologies and their consequences that was published in the year 2017? Or that of Curtis “Mencius Moldbug” Yarvin? Per the index, even Peter Thiel and that supreme incontinent E. Musk appear only a single time each in its 349 pages. But Yudkowsky? Not at all. However did this the author of this book manage to sail by the point so serenely, so thoroughly, so ruinously?
The book in question is, of course, my own Radical Technologies: The Design of Everyday Life. I've been thinking about it again a good deal lately, now that the entire direction of the United States government has fallen into the hands of individuals who were already in the vanguard of technoreactionary thought at the time of the book's composition, but who went virtually unmentioned across its entire length. It's an unpleasant thing to realize — the more so since, if you’ll permit me to say so, Radical Technologies otherwise remains a reasonably solid account of the things it does set out to treat. Some eight years downstream from publication, it stands up much better than I expected it to, or had any right to expect given the speed at which technology evolves. But its major lapse, which I mean to correct imminently, is that it remains too focused on the way technology works rather than what it does in the world.
This has been obvious to me for awhile now, but it became inescapable a few weeks ago, in the wake of the spectacular flameout of an accelerationist deathcult you may have heard about recently, called the Zizians. You can read every word about the Zizians published in the New York Post or the Guardian, without ever once encountering an adequate account of what they believed to be true, what milieu those beliefs gestated within, or what they wanted to achieve. Surely (say) the Baader-Meinhof did no better from the not-yet “legacy media” of their time, but it's nevertheless a striking failure of contemporary mainstream reportage, given that the same doctrine driving the doomed and marginal Zizians ultimately animates far more consequential figures in society.
This goes directly to the weakness I now perceive at the core of Radical Technologies. In the end, I believe that the most important “impact” of a given technology isn’t anything direct or linear or determinist, but rather results from that technology's role in the elevation and empowerment of certain social groups over others it tends to suppress or marginalize. Which voices, individual or collective, are raised to prominence in the wake of its adoption? Which viewpoints are valorized, lent weight and credence? And conversely, which perspectives are eclipsed and fall by the wayside in the embrace of a given class of technologies?
Here I'll point out that happily, there is a book of similar vintage that engages all of these questions, and that in many ways functions as Radical Technologies' shadow twin: Elizabeth Sandifer's stunningly prescient Neoreaction a Basilisk. It doesn't waste time on explaining the mechanics of things like neural networks or the Bitcoin blockchain, choosing instead to anatomize the social groups that crystallized around them, and for my money it holds up just as well. If you're interested in this stuff, it's still well worth a read, even now.
A critique one might plausibly have leveled against Sandifer's book in the year it came out, however, is that the NRx/Dark Enlightenment nerdcorps it treats of was surely operating on the margins, and bound to remain that way. And therefore that dedicating any real amount of exegetical energy to the work of figures like Yarvin and Yudkowsky, as it did, was an endeavor best understood as a species of masochistic play: fine if that's how you really want to amuse yourself, maybe, but not anything that's of likely consequence or interest to anyone else, and maybe kinda distasteful into the bargain.
This argument feels a whole lot less convincing, though, now that the primary exponents of the very same (then crypto- and now overtly) fascist worldview explored at such length in Neoreaction a Basilisk presently enjoy virtually unimpeded influence in the nerve center of American power. Certain shards of their thought have become embedded in the public discourse, working their way ever deeper into what amounts to the regnant commonsense, and if we ever want that discourse to become healthy and life-affirming we're going to have to spend the next several decades collectively digging them out. For the moment, I'd like to consider just one of those shards: the one crystallized around the work of Eliezer Yudkowsky, the facets of which we know as Rationalism, effective altruism and its variant effective accelerationism (“e/acc”).
This dispatch is lengthy, largely because the people it centers on are to a one given to express themselves at mmmmmaximum length, but I hope you'll bear with me nonetheless. Who knows but that it might even function as a roadmap to some things that are about to unfold in our culture, and though the odds against are very long indeed, maybe even a minor contribution toward preventing them from happening?
§ The story I'm going to tell in the next few thousand words — let's go on ahead and call it the Rat King's Tale — has the curious quality that it is simultaneously recent, well-documented to the point of absurdity, familiar to a largish body of people, hugely consequential in real-world terms...and to this day almost completely absent from mainstream coverage and awareness. If the influence of Curtis Yarvin's neomonarchist rhetoric on figures from Thiel to JD Vance has at long last gotten due attention from popularizing gatekeepers like Big Media Ezra Klein, Yudkowsky's thought has yet to be so favored. And yet that thought is a spindle with seven- and eight-figure flows of capital wrapped around it, not to mention a great many comprehensively ruined minds.
Here's the kernel of the Rat King's Tale: Over the span of a few years at the cusp of the first and second decades of the twenty-first century, while so many of us were occupied (heh) with the permutations of post-2008 politics, an exceptionally bright but not particularly accomplished software engineer and artificial intelligence researcher named Eliezer Yudkowsky became obsessed with how might one go about enhancing the likelihood of consistently making optimal decisions under prevailing conditions of uncertainty.
I hope it's obvious that in a great many ways, this is a perfectly reasonable question to pursue, even though the specific way it was addressed here eventually led to psychosis and suicide and murder. Who amongst us has never wished for the clarity to choose wisely in moments of instability? What if there actually were some latent order operating in the universe, that might help us make better decisions more of the time if only we once learned to discern it? Who wouldn't be willing to exert at least some effort toward discovering the principles of that logic, and acting upon them, if it meant living ever after with the compounding fruits of better decisions? I mean, I would, and I'm surely among the laziest people you'll ever meet.
The stumbling block consists in the way Yudkowsky went about approaching the question. He committed his quest for a mechanism that would reliably identify “true beliefs” to an “online forum and community” called LessWrong, in an epic production dubbed the “Sequences.” There is a plodding, relentless logic to these posts: they are designed to walk the reader, step by aaaaachingly drawn-out step, through the granular development of a method of analysis that might reliably substitute for wisdom.
In order to make his points, Yudkowsky deploys metaphors from the lifeworlds beyond software engineering he's developed an interest in, notably quantum physics, the game of go and Orthodox rabbinical lore. In the hands of a better and a more disciplined writer, you can see how an argument anchored in such points of reference might be compelling, even charming. The issue here, of course, is that Yudkowsky is...not that writer. One of the reasons that people outside Rationalism have so much difficulty reckoning with it as a body of thought is that the Sequences are unbearably prolix — the very first installment runs to nearly 7,000 words, and there are, by my count, an additional 401 of them — and littered with highly idiosyncratic terminology besides. By the time Yudkowsky finally enjoins his readers to “go forth and create the Art” of rational living, capital A and all, even this most basic account of his thought on LessWrong has hit some 1.2 million words. And that's before we even get to the Harry Potter fanfic.
I imagine one would already have to be of a certain cast of mind to find the prospect of absorbing such a body of writing anything but harrowing, on the life's-too-short principle. Nevertheless, people did follow where Yudkowsky led them — and not a few, either. LessWrong developed a fervent following, certainly outweighing the audience for anything I've ever written by a few orders of magnitude, who rather self-congratulatorily dubbed themselves “Rationalists.” And that following in turn weighed the questions that concerned them at the very great length that was quickly becoming something of a communal signature. One of their central preoccupations — indeed, the topic about which they most fervently wanted to make decisions founded in “true beliefs” — involved the quest to develop an artificial general intelligence, or AGI: a synthetic reasoning agent at least equal in intelligence to the engineers who created it at the moment of its inception, and eventually capable of a very great deal more.
The gestation of such a reasoning agent had always been at the very heart of Yudkowsky's intellectual project, and it was the focus of his activities in the years before he published the Sequences on LessWrong. In 2000, he co-founded a Berkeley-based nonprofit called the Singularity Institute for Artificial Intelligence — later Machine Intelligence Research Institute, or MIRI — and this organization was consecrated to research on so-called “friendly AI.”
This is the point at which — in order for what Rationalists fear about the AGI they so fervently want to call into existence, or how that relates to the Zizians, or really any of this, to make any sense at all — we're first going to have to define a few terms, and again, please forgive me if this is terrain you're already overfamiliar with. We're specifically going to interrogate the two notions of “existential risk” (or “x-risk”) and “infohazard.”
Of the two, the idea of an extinction-level risk to human survival is easier for most of us to get our heads around. We grew up immersed in such scenarios, and though it should be obvious that no attempt to assess their actual probability was involved, some of our favorite fictions were woven of them. Global thermonuclear war, runaway atmospheric heating, a slatewiper contagion, perhaps an asteroid strike of sufficient mass and velocity: I bet you can think of a big-budget film treating each one of these themes, without any prompting to speak of.
The primary x-risk that preoccupied the Rationalists, though, was none of these, and was in fact somewhat novel. This was the threat to the survival of our species posed by an artificial intelligence more capable than any individual human, or even all of us together, once that intelligence had undergone the explosively geometric expansion of its capacity over a short duration, and particularly as such an entity acquired the ability to manipulate matter directly. An AI equipped with the “effectively complete control of the structure of matter” long called for by nanotechnologists like Eric Drexler would in effect be “weakly godlike,” which is to say that it would be able to cause change in the world of any sort whatever to occur in arbitrary conformity with its will. And if at any point that AI's perception of its interests fell out of sync with ours as a species — this is the so-called “alignment problem” — we would be doomed, and our doom would unfold so comprehensively and over such a compressed timescale that there would not be any meaningful way for anyone to counter it. Thus the overwhelming importance MIRI placed on the effort to develop “friendly AI.” When Rationalists discuss “x-risk,” this is what (and more or less all) they're talking about.
So much for existential risk. What, then, is an infohazard? We can think of this as an idea about ideas. Specifically, an infohazard is an idea so lethally potent that to be exposed to it, to take it on board even momentarily, is to entrain catastrophic consequences for one's sanity, or even continued physical existence. The canonical infohazard was introduced by science fiction writer David Langford in a brilliant 1988 short story called “BLIT”: a machine-generated image so viciously confounding to the human perceptual system that to glimpse it even in part is to immediately and irreversibly crash the brain. (You may recognize Videodrome's pirate-television signal, the Sumerian nam-shub in Snow Crash and the Entertainment in Infinite Jest, not to mention the Medusa of Greek mythology, as a series of subtle variations on this theme.) Rationalists believe in the literal existence of infohazards, and, as we shall see, they may well have reason to.
The ostensible infohazard relevant to this conversation is one contained in a brief thought experiment posted to LessWrong in 2010, called “Roko's Basilisk.” I have no hesitation describing it to you here in full, because I do not believe it constitutes a danger to anyone who has not first worked themselves into a tizzy by following a particular line of faulty reasoning for an extended period of time. If you haven't already turned your brain to butterscotch pudding by following the train of logic laid out in Yudkowsky's Sequences, that is, the description that follows holds no threat for you — and the notion that it might will likely strike you as ridiculous, as it should.
In their post, a LessWrong community member called Roko presented the idea that an “otherwise benevolent,” superintelligent AI arising at some unspecified point in the future might wish to incentivize people in the present to work toward its creation, and that the most effective means it would have of doing so would consist in its commitment to punish everyone who'd been made aware of the possibility of such a superintelligence, but thereupon failed to devote all of their life energy toward bringing it into existence. (No, you're right: this doesn't make sense. You need to have swallowed a whole lot of Yudkowskian “timeless decision theory” concerning the game-theoretical coordination of distanced actors with no physical channel of communication between them for this to even cohere as a thought. But remember, that's just what LessWrong readers had done.) Then Roko sprung the trap: since the very act of reading the post would by definition make the reader aware of the possibility of such a superintelligence, it was already too late for them. The idea had already taken root in their mind. Unless they now actively, immediately dedicated the entirety of their resources to the effort of creating it, and maintained that effort for the rest of their natural lives, the godlike AI entity, when it finally arrived, would ensure not merely that they would be tormented throughout all of infinity, but that their consciousness would be multiplied infinitely, and each and every one of those consciousnesses tortured for all the rest of time. It would be as if the Christian hell had been raised to its own exponential power, in gouts of agony that would rival the searing heat of creation itself. Reader: people believed this. And more than that, they acted on it.
Now we're prepared to talk about what happened with “effective altruism,” with consequences not merely for the Bay Area Rationalist scene but for philanthropic giving, for the flows of capital underlying the prosecution of American political campaigns, and for the disposition of public resource more generally.
Many of those in orbit around LessWrong — drawn as they were to particularly highly-compensated work, in an already notably remunerative sector — were becoming personally wealthy, and increasingly wanted to share the benisons of their wealth via charitable contributions. Their preoccupation with an “effective altruism” is founded in what is, for such well-off benefactors, not at all an unreasonable question: how to ensure one's allocations of charitable capital make the most impact possible. Most philanthropists are confronted by some version of this; their concern is generally that the checks they write aren't simply eaten up by inefficiency and overhead, or ducted off into outsize compensation for the executives of whatever charity they've chosen to shed their largesse upon. Effective altruism went a good deal further. Its proponents attempted to systematically identify where one might realize the greatest possible bang for the buck, when measured by enhanced human wellbeing, across the entire spread of possible philanthropic endowments.
Unlike many other similarly-situated individuals, what’s more, by and large the Rationalists weren’t drawn to philanthropy for social-capital brownie points, or tickets to some glittering charity gala, or so they'd be memorialized as bold-faced names in the local society column, but because they sincerely wanted to make the world a better place. Indeed, many of them are known to have endured personal conditions of significant austerity in order that they could dedicate a larger portion of their earnings to charity. The trouble here is that the effective altruists understood “human wellbeing,” with clunky literality, to be a kind of global utility function, susceptible to global optimization. When you combine this with a longtermist perspective that weights the ethical prerogatives of octillions of entirely notional sentient beings – i.e. those that might someday come into existence, in the course of a triumphant human breakout from Earth and eventual colonization of the galaxy – over the mere billions that actually exist now, the stage is set for the worst and most consequential sort of error.
Roko's post was a kind of Pascal's wager, updated for the age of synthetic intelligence, with stakes so unthinkably awful that merely contemplating them mired these self-proclaimed Rationalists in extended bouts of bowel-clutching terror. From their perspective, the only rational strategy to adopt when confronted with such an all-powerful enemy is to placate it by dedicating all of your life energy and every resource at your disposal to bringing it into being. This, to put it bluntly, is a notion most of us have no problem immediately discerning as absurd, to the degree we weigh it as a serious proposition at all. As we'll see, it's something that only hazardous to those with a very particular cast of mind.
But the logic of Roko's Basilisk infects everything that happens next: in particular, the idea that doing something to counter the existential risk from rampant, weakly godlike artificial intelligence is more critical than alleviating the suffering of actually-existing humans in the here and now begins to spread virally among the susceptible population, eventually bleeding back out into effective altruism. After all, from a Rationalist moral calculus, how could preventing the suffering of anyone now alive possibly outweigh in importance actions taken to benefit the trillions upon trillions of human souls that might eventually come into existence? And so the effective altruists diverted their support, from — actual example — a cause they themselves had previously identified as disproportionately beneficial, a charity that produced and distributed low-cost mosquito nets in sub-Saharan Africa, preventing the spread of malarial disease among millions, to...funding their own community's research on the alignment problem. Perceptions of the very long range interest of an entirely imaginary cohort had outweighed the very real interest millions of Africans had in, well, not dying. (Again, this was very far from being a marginal position in the Valley. Something very much along these lines is what massively overwealthy pinheads like Marc Andreessen actually believe...and I dare you to find another figure whose commitments are so grotesque that revulsion at them unites me with the likes of noted reactionary Rod Dreher.)
Where does this leave us? In an attempt to hedge against the coming into being of the thing they feared most, a goodly number of clever but highly suggestible individuals literally argued themselves into a living hell. The torment they experienced, while terrible and sanity-eroding and entirely real, wasn't inflicted upon them by a godlike AI, but by their own minds. (The exquisite, awful irony, which I am grateful to Jon Evans for pointing out, is that MIRI’s ostensibly terrifying work was a dead end, a blind alley in AI research. It was comprehensively leapfrogged by research into the neural networks underlying ChatGPT, Deepseek and all other contemporary models, and never produced anything beyond the level of a chatbot, never mind a weakly godlike power.) Meanwhile, members of this same community also diverted funds from efforts that might actually have saved human lives to pointless, wheel-spinning exercises in the creation of “friendly AI,” to the tune of billions of US dollars and some unknowable (but clearly reasonably large) number of victims living somewhere safely over the horizon.
Ideas have agency.
§ Here I want to spend a few moments considering π, Darren Aronofsky’s great debut feature of 1998. π begins with a curiously lyrical soliloquy from its protagonist, a hollow-eyed insomniac of a mathematician named Max Cohen, on the properties of complex dynamical systems in nature, from “[t]he cycling of disease epidemics, the wax and wane of caribou populations” to “sunspot cycles [and] the rise and fall of the Nile.” Max winds up his ode to organized complexity with a bold but somehow predictable associative leap: “So what about the stock market?”
There it is: the money shot, so to speak. Max's lifework — which he pursues obsessively, in a darkened apartment crammed to the point of claustrophobia with CRT monitors, dodgy custom processors and LED stock tickers — involves using arcane computational techniques to derive material advantage from latent pattern in the market, before others are able to perceive that pattern at all. But what Aronofsky manages to make clear is that these computational techniques are little more than formalizations and eversions of the way Max's own mind works. What π gets so very, very right (and what's more, did so pretty close to the first historical moment at which any such depiction was possible) is the way in which a particular set of human neurocognitive predispositions intersects with the needs of capital. Max, tormented literally to the point of self-trepanation by the patterns he helplessly perceives in all things, is going to make someone a whooooole lot of money — but only if that someone can capture, domesticate and own for themselves the things his brain does for him for free.
What are those things? Aronofsky's script tells us that Max is porous, labile to perception, detail-oriented, inflexible in his routines, tormented by acute sensory sensitivity and, above all, by intimations of an order which remains forever just beyond his grasp. And we recognize these qualities, right? Maybe not in quite the twitchy intensity Sean Gullette brings to his indelible performance, but I'm willing to bet that all of us know people like this to some degree. Maybe we ourselves are like this. [Raises hand.] This cluster is definitely A Thing, but we've never quite worked out as a society whether it should be understood as an impairment, a brace of superpowers, or both at the same time. And as a result, the names by which we currently know it — Asperger's syndrome, high-functioning autism, auDHD — are crude and imprecise and kind of shitty. For the sake of this conversation, then, let's use a term to describe this set of propensities that, while equally inadequate and no less imprecise, is at least purged of the judgment and resonances which shadow those other labels. For now, let's call this property the Z-cluster.
The billions of dollars flowing into effective altruism and other Rationalist institutions existed in the first place because capitalism conceived a need for people with this precise neurocognitive cast, to function as quants and coders and seers: an army of Max Cohens capable of pulling clarity (and therefore the possibility of microsecond-scale arbitrage, among other things) from undifferentiated chaos. In due time, the existing concretion of technocapital capacity we think of as Silicon Valley assembled in the Bay Area of northern California a dense knot of Z-cluster folks, scattered about group houses and cubicle farms where they formed cliques, pondered the implications of “timeless decision theory,” and practiced the Art of Yudkowskian Rationalism at each other at full volume.
It was this very density of concentration, I'd imagine, that produced the conditions of possibility for what happened next. If you're hungry for an account of this moment that offers even greater depth and detail, extending to hair-raising tales of intramural magickal warfare at MIRI and its satellite institutions, I can only refer you to Jon Evans's thoroughgoing account. So far as I am concerned, this long chain of long posts remains the gold standard in accessible descriptions of Rationalist thought and lifeworld, and is sparklingly written besides. For now, though, what's important to glean is that the same Z-cluster qualities that made the Rationalists so valuable to bleeding-edge research also made them susceptible to totalizing arguments. And predictably enough, given the prevailing dynamics, some of these arguments came to vest in the context of what social psychologists might charitably refer to as a “high-control group,” and the rest of us would have no problem identifying as “a cult.”
And this brings us to Jack Amadeus “Ziz” LaSota and the Zizians.
§ Imagine that you live among a group of people who are all just as bright as you are, who are unanimously convinced that your group and your group alone has through reason alone glimpsed something of incredibly vital importance to the multibillion-year future of the human species, and equally convinced of the necessity of immediate direct action if that future is to be at all. Imagine that this obligation is just about all that any of you can talk about, or even think about in the safety of your own room, in the privacy of your own mind. What acts would you come to believe were not merely justified by your duty to the future, but absolutely, unbendingly ethically required of you?
You can get some insight into what this might feel like by consulting the Zizians' own public accounts of their own beliefs. If the common-or-garden Rationalist screed is strewn with unfamiliar words and concepts, though, this is even truer of the word salad generated by the people around Ziz. (It’s striking to me, for example, that Ziz uses the technical game-theoretical term defection to describe what you or I might regard as minor disagreements with friends.)
If that's not your speed, though, you can also glean something of the flavor of their thought at second hand, by reading any of the many articles about the Zizians that have cropped up in the wake of the six violent deaths associated with them. Probably the most comprehensive of these is this piece in WIRED, but by now any number of media outlets have taken a shot at the story, and surely one or two of the more thoughtful among them will shed additional light on what these people held to be true.
At the core of all these stories is the vexed figure of a young Alaskan software engineer named Jack LaSota, who had moved to the Bay Area specifically to get involved with effective altruism and the broader community of Rationalists working in technology. LaSota — Ziz — displayed the full panoply of Z-cluster traits, and — it's clear to me, anyway, after reading tens of thousands of words of their online journal entries — had suffered tremendously from them, since early adolescence at the latest.
Here's where that “high-control group” stuff becomes relevant: Ziz was evidently sufficiently charismatic, among a self-selected cohort of fellow travelers in this milieu, to compel quite a few people in the Bay Area Rationalist scene to adopt their own idiosyncratic reading of Yudkowsky's already quite idiosyncratic thought. It's hard to flense out, in this groupuscule's blog posts, LessWrong comments and other discursive productions, just what was Yudkowsky, what was Ziz and what was entirely their own — these were people wont to use terms like “recursion stack” and “clamped variable” to refer to their own psychic architecture — but it's clear that they all spiraled together into a viciously-enforced and highly self-serious groupthink, in which they perceived their acts to have disproportionate bearing on the direction of the entire human future.
Many of the Zizians' more puzzling statements revolve around a professed belief that was originally Ziz's own, and became unquestionable doctrine among them: that the human mind is radically bicameral, and that each of the brain's hemispheres functions as an independent processing core, bearing its own identity, its own gender and its own ethical commitments. While individually inaccessible in ordinary life, there were techniques you could use to enter a state in which one of the hemispheres was made quiescent, leaving the other responsive, independently addressable and able to be hailed at will. In order to access this state, the Zizians committed themselves to the practice of “unihemispheric sleep,” which involved undertaking the following steps:
You need to be tired.
You need to be laying down or sitting up. (It is important that you stay in a comfortable position that won't require you to move very much.)
In either case you want to close one eye and keep the other open.
Distract the open eye with some kind of engagement. (This works best if you have a partner who can help you.)
Eventually you should feel yourself begin to fall asleep on one side. That side will also become numb. (The degree of numbness is a good way to track how deep into sleep the side is.)
Once into [unihemispheric sleep], it is supposed to be possible to infer which aspects of your personality are associated with which side of the brain. (This is a little farfetched, and the more likely explanation is that it disrupts peoples visceral sense of what is and isn't socially okay.)
Now, firstly, I can’t imagine a better way to reliably precipitate a psychotic break than sleep deprivation, most especially when accompanied (as it was in the Zizian milieu) by intense struggle sessions and routine high-dose psychedelics. So it ought to come as no surprise that, by their own account, that is exactly what happened to more than one of the Zizians. The other, ironic thing, though, is that Ziz might well have been onto something in diagnosing neurohemispheric asymmetry, though surely not in the way intended. Beyond the sorrowfulness and even abjection they are drenched in, what leaps right out at the reader about Ziz's writings is their strongly depersonalized, dissociative quality; no reader of Iain McGilchrist, certainly, could read this material without being struck by the pronounced pattern of derealization generally associated with severe right-hemispheric deficit. (This is someone who had to lay out a multi-stage argument to account for their perceptions of their own gender, starting from questions like, “You’ve tried running your mirror neuron thing on both men and women, right?”)
However outré this belief in radical bicameralism might seem, it unquestionably had consequences for the way people in Ziz's circle interacted with one another and the larger world. This particularly true of the corollary notion, universal among them, that one could only ever trust those whose hemispheres had both somehow proven themselves ethically reliable — in their jargon, double-goods.
Another of the core tenets of the Zizian belief system was a commitment to functional decision theory, which led them to literally think of themselves as a kind of laminated hyperagent applying game theory across a skein of roughly parallel universes. For example, Ziz counseled followers to physically resist arrest, however futile this may seem, on the theory that doing so would marginally increase the odds that they live in a worldline in which that resistance is effective and they are able to escape. When you follow lines of thought like this to their logical conclusion, you get thought patterns like the one seen here, in a blogpost from onetime Ziz associate Slimepriestess:
Now for most people, if your perfectly rational decision theory tells you to do something stupid and fatal, you simply decide not to do that, but if you let yourself sink deeply enough into the frame of being a multiverse-spanning entity then it’s very easy to get yourself into a state of mind where you do something stupid or dangerous or fatal. In the frame this creates, it’s just obvious that you can’t be killed, you exist across many timelines and you can’t experience nonexistence so you only experience places where your mind exists. This might lead you to try to kill yourself to forward yourself into a good future, but the bad ending is that you run out of places to shunt into that aren’t random [B]oltzmann brains in a dead universe of decaying protons at the heat death of every timeline and you get trapped there for all of eternity and so you need to kill the whole universe with you to escape into actual oblivion. Oh and if you aren’t doing everything in your cross-multiversal power to stop that, you’re complicit in it and are helping to cause it via embedded timeless suicidal ideation through the cultural suicide pact to escape [B]oltzmann hell.
How does someone decide to kill themselves because they came to believe, in so many words, that their own utility function had dropped below zero, and that they therefore perceived a positive “magnitude of the timeless benefit from self-destructing in the proactive”? Or because they assessed that one hemisphere of their brain, acting independently, had breached the set of conditions under which the body containing both hemispheres justified its continuing existence? (This actually happened.) That this is a lunacy that only the very bright might find themselves subject to doesn't alter the essential fact that it remains lunacy. Mainstream reportage continues to characterize the group gathered around Ziz as “trans” “vegan” “anarchists,” but forgive me, I don’t think they were any such thing — not more than incidentally, anyway. What I think they were is a knot of somewhat brighter-than-usual Z-cluster folks without adequate reality-testing practices, who allowed themselves to become socially isolated and caught in a self-reinforcing positive feedback loop in which an induced mental-health pathology was able to run rampant. As we now know, the results of this reinforcement loop were lethal. But what is made plain by the utter anguish sheeting off every account of themselves offered by those in Ziz's circle is that even well before that point, they were heartbreaking.
Rationalists often justify their commitments by upholding consequentialism, the belief that the only valid standard for assessing one's choices in life is the results they are seen to produce. In the wake of the Zizians, and the six deaths associated with their conduct of Rational decision-making, it feels like a fruitful moment for anyone inclined to think and act in the ways laid out by Yudkowsky to reassess whether his Art is a wise path to embark on. Consequentialism itself commends this. Remember the question Cormac McCarthy's implacable killer Anton Chigurh asks: “If the rule you followed brought you to this, of what use was the rule?”
Ideas, it turns out, have agency.
§ If I'm honest, the main reason I never paid much attention to the Rationalists in their pomp, even beyond the daunting logorrhea of their productions, is that the gut-binding terror most of them lived in, and dedicated so much effort to countering, has always struck me as prima facie ridiculous.
To me, it seems based on a profound misunderstanding of intelligence — that is, the very quality those drawn to LessWrong had dedicated their lives to upholding, cultivating and extending. What did they miss? For our purposes, the salient feature of intelligence is that it is embodied. It isn't simply an emergent property of large number of neuron-switches, internetworked in a very large number of ways. (This is the belief known as “connectionism.”) So far as we understand it, it's something that mediates signals from the world through a vastly ramified and equally subtle electrochemical apparatus, whose origin, evolution and boundary conditions we don't comprehend very well even now. And even were it possible to gestate a reasoning agent of comparable information-processing capacity in silico, it seems absurd to imagine any such thing might harbor motivations that descend from the endocrine facts of embodiment: drivers like resentment, rage or the thirst for revenge. To assert otherwise is to stumble into the most vulgar sort of anthropomorphosis.
And yet the entire Basilisk scenario — which, remember, a nonzero number of Rationalists live in sleepless, abject terror of, to this very day — depends on the notion of an enraged machinic superintelligence reaching backward in time to wreak eternal revenge on their defector souls. I am far from the first to observe that this is essentially the plot of Harlan Ellison's 1967 short story “I Have No Mouth, And I Must Scream,” blended with the less pleasant aspects of Christian eschatology like swarf and shrapnel in a particularly foul cone of Ben & Jerry's. But what's peerless as a short story is, to say it once more, definitionally insane when elevated to the point of preoccupation as a daily concern more important than shopping for the groceries, paying the bills, cleaning the house or attending to matters of personal hygiene.
If we really want to understand the fundamental blunder of analysis being committed here, though, we have to dig just a little deeper. Just as Ray Kurzweil's similarly misguided quest to resurrect his father in a perfect-fidelity computational simulation has its origins in his inability to come to terms with the immense grief of bereavement, Rationalism also strikes me as having its roots in lived pain — in this case, the pain of a mind hardwired to seek justice, balance, reason and order, compelled to make its way in a world (or more to the point: a society) where none of these things exist. Rationalism is the desperate attempt to impose clarity on an inherently unruly universe, by concerted exertion of will.
Once you've learned to see it, what's more, it's impossible to miss the sharply similar logic that runs all the way through the vein of contemporary information technology: eliminate trust, and all the ambiguity and risk of hurt and loss that necessarily orbit it like a cloud of bad-tempered electrons. Lock in contract at the level of code, so no party to an agreement ever again has to suffer exploitation or betrayal. Automate insight, to render it neutral, predictable, pliant. Sand down the confusions, the logic snares that litter the void between map and territory. Purify language so interlocutors say what (and only what) they mean...and then institute mechanisms compelling them to mean what they say. The pattern is striking, so striking indeed that it demands an explanation. My guess? I think people are drawn to these technologies and their development in no small part because, just like Yudkowskian Rationalism, they promise to redress the searing agony some of us who bear the Z-cluster seem to experience simply by being alive in a world that refuses to make orderly sense.
To which I can only say two things: an unchanging and self-consistent framework of logic doesn't seem to be what our universe is built on, and even it were, it definitely isn’t how embodied minds work.
If you'll forgive my poor autodidact’s reading of things I’m probably not qualified to discuss, that the phenomena we contend with are simply fundamentally insusceptible to logic is something I've suspected ever since my first encounter with two thinkers even smarter than Eliezer Yudkowsky. The first is Ludwig Wittgenstein. Informed, surely, by my encounters with Taoist and Zen Buddhist thought, my understanding of his Tractatus Logico-Philosophicus has always been that its famous final clause enfolds and contains everything worth knowing. Everything is packed at singularity density into that single clause at the end: “Whereof one cannot speak, thereof one must pass over in silence.” There is a universe of propositions whose truth value can be specified rigorously in the terms of formal logic, in other words, but that universe is smaller and by far the lesser in importance to the universe of everything which cannot.
The other is the mathematician Kurt Gödel — whose work, again, I'm sure I have but the shallowest grasp of. As I understand it, though, Gödel's incompleteness theorem confronts us with the finding that no system, including reality as we experience it, can be rigorously specified within terms or processes originating inside it. It's an impossible, Klein-bottle involution at the very root of being, with shattering consequences for everything that is even in principle thinkable.
Faced with these implacable constraints, it seems to me that the only truly rational thing one can do is to let go of the fetish for logic, once and for all. Let go, and embrace the fact that life is inextricably bound up with confusion, ambiguity, interpretation and contention, however hard those things might be to bear. And this is because following logic unbendingly, in a universe that refuses even the possibility of that logic, right down to its basement, can only ever be a recipe for the multiplication of sorrows.
At least this is what the Rationalists learned, to their detriment, and our own as well. As an ethical compass, I've always found something hugely appealing about the humility involved in an effort to be "less wrong than yesterday," and to the degree that this is what Yudkowsky and the community around him set out to achieve, I'm not at all unsympathetic — for as I say, there are more than one or two respects in which my own mind works this way too. But now all I can think is this: if only he and they had been able to leave it at that.
§ As we conclude this Rat King's Tale, I want to circle back to the point about Radical Technologies I began with. If I had one wish for the book, it’s that I'd leaned harder into the politics — the discrete and coherent ideology — underlying the pursuit of things like AI, VR and the blockchain.
It's not that I harbor any illusion that the book would have changed anything, even if written that way. It's that it would have been a better book because ideas have agency, every bit as much as the tools and systems that are the manifestations of those ideas, and understanding how they mesh together to produce change in the world requires a thorough unpacking of both material and ideational aspects. Too narrow a concentration of focus on the former only allows the influence of the latter to blossom unobserved.
It may, just, be possible to recover consequentialism from the grip of those currently most closely associated with it. There is nothing wrong with consequentialism. If you are not, in fact, allocating your resources with eyes firmly fixed on the outcome of your wagers, then this is really a confession that your politics are nothing more than an aesthetic affectation. The rest of the package, though? If there truly was ever any such thing in the world as an “infohazard,” it is Rationalism itself: a virus of the mind, capable of entirely devouring the energy, mental stability and physical health — and, the available empirical evidence clearly shows, occasionally the very lives — of those susceptible to it, the overwhelming majority of whom display a set of cognitive predispositions I've referred to here as the Z-cluster. And as awful as what Rationalism did to them was, they are far from the only people to have suffered from its lock upon their minds.
The twentieth-century poet Czesław Miłosz once argued that it is usually via some process of suffering that we come to understand how our fate can be “influenced directly by intricate and abstruse books of philosophy.” The specific philosophy he was thinking of was dialectical materialism, which he regarded with especial horror, having suffered through the Stalinist repression ostensibly mounted in its name. But he could just as easily have had Heidegger’s Romantic mysticism in mind, or the differing nihilisms of Nietzsche, Céline or Cioran. It's surely possible to trace lines between all of these thought traditions and the eventual conduct of atrocity.
Due in no small part to the particulars of where and when he was writing from, the framing Miłosz offers has a certain melancholy weight and power. It turns out, though, that he was only half-right.
It is certainly the case that philosophical thought has often enough set up currents that later worked themselves out on the historical stage, at the level of lives and bodies. As I've been at some pains to show here, ideas have agency, and perform material work in the world. And it is also the case that it is usually only by way of grim experience that we are forced to concede a connection between some idea that had previously seemed harmlessly obscure and its large-scale effects in the world — for better or worse, we appear to be a species condemned to learn all its most important lessons the hard way. So Miłosz was surely on firm enough ground in arguing that we ought to pay closer attention to the currents of subterranean thought, and to what is implied by them, long before they breach the surface of mainstream awareness, as companions and legitimators of horror. But where he stumbles is in thinking that the ideas capable of giving shape to horror in this way need to be anything richly philosophical in nature. They could, indeed, take the form of Harry Potter fanfiction.
Just now, most thoughtful people I know are concerned about the advent of something that sure is beginning to look a lot like fascism, and properly so. But what I hope our tour of Yudkowskian Rationalism, effective altruism and “e/acc” demonstrates is that there are currents abroad in the culture that are orthogonal to conventional left-right politics; that they are increasingly salient to the disposition of our collective fate; and that virtually nobody who is not already immersed in them is even paying attention to them.
To say so yet one last time, ideas have agency. For certain kinds of very suggestible people, indeed — particularly those of us who carry the Z-cluster — they can kill. This was the real Basilisk all along. Those exposed to its gaze terrified themselves, in some cases quite literally to the point of madness and death, over the spectre of something most others would regard at most as a thought experiment, a game of words. Even for those not driven to sleeplessness by thoughts of what some godlike AI at the end of time might do to them, though, the belief that we live in the dawning moments of the AGI era can have seriously destructive implications. I have a cherished friend whose mental health has taken a beating lately, who's turned away from the pursuits that used to furnish him with value and meaning and pleasure (and an income!) because “it feels like there’s not much time left now.” And as painful as it is for those of us who love him to contemplate the consequences of his conviction, I know he's far from the only one to think such things.
I don't think there's anything anyone could have said out loud, or committed to print, that would have prevented people like this from finding themselves reflected, and thereafter lethally mired, in the Rationalist infohazard. Like so many of the sociotechnical systems I've considered since my very first book, the appearance of these ideas in some version feels overdetermined, as does the prospect that they would appeal overwhelmingly to some group of people once they had been advanced in public. In thinking about what happened next, I'm haunted by an asymmetry in the world that William Burroughs, of all people, once diagnosed. He said — and here I'm paraphrasing, because ironically enough I can't track down the exact language — that for each of us there is a particular sequence of words that will destroy us, which, life being what it is, we are certain to hear, and a similar sequence that has the power to bind up our wounds, which we have no guarantee of ever hearing uttered. Whatever his motivations may have been, Eliezer Yudkowsky's words objectively belonged to the former category. The only words I'd ever want to write, though, belong to the latter.
§ We'll leave it there, yeah? Now: as to my travel plans. I'm doing Lifehouse talks at Alienated Majesty in Austin, Texas on April 4th; at a venue TBD in New Orleans, on the 6th or 7th of the month; with A. Naomi Paik at Pilsen Community Books in Chicago on the 9th; and at Joybomb Social Center in Gary, IN on the 10th — feel free to share these dates as widely as you please. It'd be a solid hoot to see you at any of them. (If you want to organize a get-together around any of these appearances, by the way, I encourage you to reach out to me on Signal, where I'm soixantehuitard.68.)
§ Right now I'm reading a neat little book called Out of the Ruins: The Emergence of Radical Informal Learning Spaces, as well as (finally finally finally) The Many-Headed Hydra, which I've been looking forward to for ages; listening to the tasty new BIG|BRAVE single, which sure seems to augur good things about the forthcoming album; and watching Flow, which was entirely gorgeous (and had me wracked with sobs). You should arrange to see it as soon as you possibly can.
That's it for now. As always, I hope that you are well, that you have your feet planted solidly on the good Earth, and that you are taking care of those who need you, taking care of yourself, and letting others take care of you. You have
all my faith,
ag
ldn
This is Dispatch from London No. 140 of Sunday, 23 March 2025, offered to you under the terms of a Peer Production License. You are free to reuse this content, but must acknowledge Adam Greenfield as its author. ◊