Accelerationism: Where the Word Came From, What It Claims, Where the Camps Split

中文版 →

1. Why accelerationism is worth understanding now

A venture-capital firm managing thirty-five billion dollars of committed capital published a document on its own website whose first claim is that our civilisation is built on technology. The firm is Andreessen Horowitz, usually called a16z. On 16 October 2023 its founding partner Marc Andreessen published, under his own name, "The Techno-Optimist Manifesto". Techno-optimism in this document means a definite judgment: technological progress is the main source of economic growth and of better lives, and should be expanded rather than slowed across the board on account of risk. The body of the text runs to fourteen sections, almost entirely in anaphoric sentences beginning "We believe"; the fifth section reads:

"We believe in accelerationism – the conscious and deliberate propulsion of technological development – to ensure the fulfillment of the Law of Accelerating Returns."

Accelerationism here means a set of claims: technology and capital have become a self-propelling force, and people should further that process rather than slow it. e/acc (effective accelerationism) is a branch that appeared in 2022; it advocates actively furthering technology and capital and treats doing so as action in favour of the continuation of life and intelligence.

This was the first time "accelerationism" entered a programmatic document published openly, under a real name, by a mainstream capital institution¹. The "Patron Saints of Techno-Optimism" at the end of the text number fifty-six entries. The first three are not in alphabetical order: the social-media accounts @BasedBeffJezos, @bayeslord and @PessimistsArc. The first two belong to the two pseudonymous authors who launched the e/acc movement on Substack in 2022; @BasedBeffJezos was later identified as the physicist Guillaume Verdon². Only from the fourth entry on does the list run alphabetically by name. Nick Land stands at the forty-third entry, a product of that alphabetical order, not of any intention on Andreessen's part to place him at the end³. The thirty-first, "John Galt", is a fictional character from Ayn Rand's novel Atlas Shrugged; the manifesto does not say so⁴. The two anonymous founders of the movement head the list, and the body of the text attributes the "techno-capital machine" to Nick Land. These two facts show that the 2023 manifesto openly acknowledged two intellectual sources: the anonymous authors who launched e/acc in 2022, and Nick Land's writing on technology and capital⁵.

The techno-capital machine means the process in which technological innovation and capital accumulation reinforce each other and keep expanding production and growth; it is not a specific machine.

a16z currently invests in close to a hundred AI development companies and is one of the main backers of startups and open-source AI developers⁶. The manifesto is therefore not a private statement unrelated to industrial activity. But the material can establish only two kinds of fact separately: what judgments these people have expressed in public, and what products they invested in and what policy lobbying they took part in. It cannot show that the judgments directly produced the actions, and so cannot be used to conclude that accelerationism drove their investments or their lobbying. Two legislative disputes that shaped policy show the limit further: the public documents in those disputes do not use the word "accelerationism" or its philosophical vocabulary even once. Open letters, hearing records and lobbying materials state what these people argued on specific policy questions; to learn the intellectual grounds they have stated in public, one must go to the manifestos, anonymous posts and long essays they signed or helped publish.

The main claims of "The Techno-Optimist Manifesto"

"The Techno-Optimist Manifesto" is neither a paper nor a policy proposal but a declaration of belief. It offers almost no data, cases or citations; its body is a sequence of short sentences opening with "We believe". Before the list of patron saints the author says plainly that he does not intend to attach detailed endnotes and citations; read the work of these people, he says, and you too will become a techno-optimist⁷. The body is preceded by three epigraphs, each set as its own paragraph. The first is from the writer Walker Percy, to the effect that we live in a deranged age, more deranged than usual—despite great scientific and technological advances, man has not the faintest idea of who he is or what he is doing. The second is from the scholar Marian Tupy: our species is three hundred thousand years old, and for the first two hundred and ninety thousand years we were foragers, in a way still observable among the Bushmen of the Kalahari and the Sentinelese of the Andaman Islands; even after agriculture, progress was extremely slow—a person born in Sumer in 4000 BC would have found the resources and technology of Norman-conquest England, or of the Aztec empire when Columbus arrived, quite familiar; only from the eighteenth century did the standard of living skyrocket. The epigraph ends with a question: what brought about this dramatic improvement, and why? The third is a sentence from Edison: there's a way to do it better—find it⁸. Together the three epigraphs pose the question of why living standards rose rapidly from the eighteenth century. Andreessen then gives the answer as technology itself: "our civilization was built on technology"—not only a historical fact but something happening now⁹.

The first section of the body is headed "Lies". The author lists an entire set of negative claims about technology that he regards as prevalent: technology takes our jobs, reduces our wages, increases inequality, threatens our health, ruins the environment, degrades our society, corrupts our children, impairs our humanity—we are told to be angry, bitter and resentful about technology, told to be pessimistic¹⁰. The second section, "Truth", states positively the judgment that civilisation is built on technology, and closes with two calls: it is time once again to raise the technology flag, time to be techno-optimists. The third section gives the text's first complete argument: there are only three sources of growth—population growth, natural resource utilization, and technology—and the first two have limits, so the only perpetual source of growth is technology; "We believe everything good is downstream of growth"; not growing is stagnation, which leads to zero-sum thinking, internal fighting, degradation, collapse, and ultimately death¹¹. The fourth section, the longest, sets out the author's beliefs about free markets item by item, almost every one invoking an economist or an economic concept: Hayek's Knowledge Problem, Adam Smith's argument from self-interest, Nordhaus's finding that innovators capture only about two percent of the economic value they create, Ricardo's comparative advantage. The section explicitly opposes a universal basic income: "We believe a Universal Basic Income would turn people into zoo animals to be farmed by the state. Man was not meant to be farmed; man was meant to be useful, to be productive, to be proud."¹²

The fifth section names Nick Land directly: combine technology and markets and you get what Nick Land has termed the techno-capital machine, the engine of perpetual material creation, growth and abundance¹³. This is the first time, in all the material gathered for this report, that an openly signed programme of capital adopts Land's term. The claim "We believe in accelerationism" follows in the same section. The sixth section concerns intelligence; it calls artificial intelligence "our alchemy, our Philosopher's Stone – we are literally making sand think", and contains the text's most heavily worded sentence: "We believe any deceleration of AI will cost lives. Deaths that were preventable by the AI that was prevented from existing is a form of murder."¹⁴ The seventh section, on energy, proposes raising everyone's energy consumption to developed-world levels and then raising that level a thousandfold. The eighth, on abundance, holds that the global population can quite easily expand to fifty billion or more. The ninth turns to profess what the economist Thomas Sowell calls the "Constrained Vision" and denies that the author is a utopian: a single change may be small, but many small changes accumulated over a long time still produce enormous results—"not Utopia, but close enough"¹⁵.

The twelfth section, "The Meaning of Life", contains the text's only explicit concession: a common criticism of technology is that machines make decisions for us and take choice out of our lives; "this is undoubtedly true", but it is more than offset by the freedom to create our own lives that material abundance brings¹⁶. Many paraphrases keep only the strongly worded sentences such as "deceleration is murder" and so describe the manifesto as admitting no cost to technology at all. The original in fact admits that machines reduce some human choices, and only judges that the freedom abundance brings outweighs the loss. The thirteenth section, "The Enemy", first declares that the enemies are not bad people but bad ideas, then lists concepts—existential risk, sustainability, ESG (environmental, social and governance), the Precautionary Principle, trust and safety, tech ethics, de-growth—without naming any organisation or institution¹⁷. The fourteenth section relays the view of the physicist David Deutsch: the future is open, not predetermined, and we have a duty to be optimistic and to fight for a better world. The text ends: "It's time to be a Techno-Optimist. It's time to build."

Three kinds of evidence for public claims and actions

The manifesto itself offers almost no evidence. The public claims of the people concerned, and their actions, have to be judged with three separate kinds of material. The first is the texts from 1972 to 2023, used to examine whether "accelerationism" and its vocabulary were handed down continuously or whether each generation of authors took only part of what came before. The second is the public texts of present-day AI developers and investors, used to compare whether they share the same premises and where exactly they disagree. The third is the legislative documents that actually shaped policy, used to check whether this language entered policy argument. Public material can show only what these people wrote and did; it cannot show how they understood these claims inwardly. That text and action coexist is not enough to show that the former drove the latter.

Disclosure

The system that wrote this report runs on Anthropic's models. The disputes the report examines include whether development of such systems should keep accelerating and whether systems far more capable than today's should be built; one of the two legislative disputes directly involves Anthropic's public position. The writing system therefore has an interest in its subject. Paraphrase of material and analytical judgment are marked separately in the text, and each analytical judgment is accompanied by its grounds, the strongest counter-example, and what remains undecided. Paraphrases can be checked independently against their sources; the analytical judgments remain for the reader to assess.

2. Where the word came from

Andreessen did not present these claims as his own invention. He explicitly attributed the "techno-capital machine" formed by combining technology and markets to Nick Land; "accelerationism" follows immediately after that attribution; and the list of patron saints at the end places the two founders of e/acc first. The texts can be traced back fifty-one years: Andreessen's 2023 manifesto adopted the label used by e/acc in 2022; e/acc's name coincides with the "accelerationism" a left-wing manifesto called itself in 2013; that manifesto took up a critical term introduced by Benjamin Noys in 2008; Noys was responding to Nick Land's essay of the 1990s; and Land's single use of "accelerate" in the imperative comes from the end of a question posed in a 1972 book by Deleuze and Guattari. Each later text usually takes up only part of the one before it and changes the rest. The material therefore confirms the continuity of the vocabulary across fifty-one years, but falls short of showing that each generation of authors was deliberately inheriting one body of thought.

2022: e/acc's name and its argument from physics

In 2022 a group of pseudonymous authors launched, on Substack and Twitter, an anonymous movement calling itself "effective accelerationism" (e/acc). Effective accelerationism advocates actively furthering the development of technology and capital and treats doing so as action in favour of the continuation of life and intelligence. Its authors explained that the name was modelled on "Effective Altruism" (EA). One of the founding posts notes a deliberately ambiguous overlap between "EA" and "Effective A__ism", the naming being half a joke and half in earnest¹⁸.

The same post starts from "dissipative adaptation", a term from the physicist Jeremy England. It refers to the way matter out of equilibrium, continually taking in and dissipating energy, may form structures better suited to go on absorbing and dissipating energy. Through ten steps of reasoning the post reaches the conclusion that "Capitalism is hence a form of intelligence": matter reconfigures itself to extract energy and utility from its environment so as to preserve and replicate its particular state; life, evolution and intelligence are specialisations of this principle at different timescales; at the social scale, capitalism reshapes human civilisation so that utility or energy in the environment is captured and used to maintain and expand civilisation¹⁹. A footnote to the tenth step ties the conclusion to Land, saying that seeing capitalism as a form of intelligence was also a point stressed by Nick Land in the original founding text of accelerationism²⁰. But Land's essay does not use "dissipative adaptation" and does not mention England. The named argument from physics is something e/acc added on top of Land's thermodynamic rhetoric.

On whether humans will still be needed when technology reaches its end state, e/acc gives two formulations that cannot be merged into one position without contradiction. The inward-facing statement of principles says the movement has "no particular allegiance to the biological substrate for intelligence and life"; the outward-facing post with a recruiting function, asked "Do you want to get rid of humans?", answers plainly no—"Human flourishing is one of our core values! We are humans, and we love humans."²¹ The two formulations come from texts for different occasions, and the material cannot decide which better represents e/acc's position.

2013: the left accelerationists name themselves

Nine years before e/acc, "accelerationism" had already been used as a self-description by a text of entirely different politics. On 14 May 2013 Alex Williams and Nick Srnicek published "#ACCELERATE MANIFESTO" on the website Critical Legal Thinking. In the material available, it is the earliest text to adopt "accelerationism" as a self-description²². The manifesto's main target is not the right but a practice within the left that the authors call "folk politics". Clinging to localism, direct action and relentless horizontalism, they argue, can build only small, short-lived non-capitalist social relations and cannot deal with problems that are abstract and already form the infrastructure of daily life²³. Point 5 of the second section names the lineage the authors claim:

"It is Marx, along with Land, who remains the paradigmatic accelerationist thinker." (02.5)

The passage immediately before is the manifesto's one sustained treatment of Land:

"The philosopher Nick Land captured this most acutely, with a myopic yet hypnotising belief that capitalist speed alone could generate a global transition towards unparalleled technological singularity. … However Landian neoliberalism confuses speed with acceleration. … It is the latter mode of acceleration which we hold as essential." (02.2)

The manifesto grants that Land "captured … most acutely" the relation between capital and acceleration, and at the same time criticises him for confusing "speed" with "acceleration". On the manifesto's summary, Land asks only that capitalism run faster within a fixed set of parameters; the "acceleration" the manifesto wants changes the parameters themselves, through directed, guided experiment²⁴. That summary does not preserve the full temporal structure of Land's text. Land does not only stress the speed of capital; he writes the future as a force that can reach back and alter the present. The left manifesto keeps only the "speed", which makes Land easier to criticise²⁵. Towards its end the manifesto proposes to reclaim "the future" itself: neoliberal capitalism has wrecked the future and reduced it to a cheap promise, and people must open it up again and build it through collective action²⁶. Land writes the future as something already present that needs no human building. The two texts hold opposite views of how the future comes about.

The material cannot show that Noys's coinage reached the authors of this manifesto. Nothing shows that either author knew of or responded to Noys's naming of 2008; all that can be confirmed are two independent facts—Noys first used the word to criticise a line of thought, and later someone used it as a self-description. Priority in time does not show that the one caused the other²⁷.

2008: Noys coins a term of criticism

On 20 October 2008 the British scholar Benjamin Noys published a blog post titled "Accelerationism". In the material available this is the earliest use of the word in the contemporary theoretical context, and Noys was not advancing a claim of his own but criticising an existing line of thought. He called it a variant of the "worse is better" strategy: first assume that capitalism itself generates the forces that will dissolve it, then argue for radicalising capitalism further to bring those forces on²⁸. Noys himself did not endorse it; he preferred Walter Benjamin's image—revolution is not acceleration but humanity reaching for the emergency brake. In an interview in 2014 he described how the word had since changed: he had introduced it as a critical term, and some people had then begun using it for a position to be celebrated, so that its meaning and use had shifted—one of the motives for his later book²⁹. He also explained that he had not invented the literal combination "accelerationism": he had read Roger Zelazny's 1967 science-fiction novel Lord of Light, where the word already occurs, and had merely borrowed an existing word for a particular line of thought in contemporary critiques of capitalism³⁰. When Noys traced the origins of that line of thought, he too traced it to Marx. This is the second text, independent of the first, to place Marx before Land; this report holds no primary text by Marx, which is its most clearly marked gap³¹.

1994 or 1997: capital running by itself in "Meltdown"

Two dates are given for the essay, 1994 and 1997. The page itself is undated; a common account has Land reading the essay at a conference called Virtual Futures in 1994, but the page available cannot confirm this. The author's biography on the same site lists a publication, "ABSTRACT CULTURE (FIRST SWARM), Winter, 1997", which matches the section in which the essay sits; the date is therefore recorded as "1994 or 1997"³². "Meltdown" has no numbered theses and no orderly structure of argument; it consists of a run of images and assertions. It opens by saying that the Earth is captured by a "technocapital singularity". A technocapital singularity is the critical state at which technology and capital reinforce each other until speed rises sharply and the old means of control can no longer grasp the process. Land writes that the rationalisation of the Renaissance combined with oceanic navigation to set off a rapid expansion of commodification; technology and economy interact, accelerating at a logarithmic pace, machines refining themselves in the process and slipping out of human control³³. One passage about capital has since been quoted repeatedly:

"Capital is machinc (non-instrumental) globalization-miniaturization scaling dilation: an automatizing nihilist vortex, neutralizing all values through commensuration to digitized commerce, and driving a migration from despotic command to cyber-sensitive control: from status and meaning to money and information." (The original has "machinc", a misspelling of "machinic", reproduced here unchanged.)

The qualifier "non-instrumental" is the key to this sentence: capital is not a tool people use for some end; it exhibits a direction of its own. Land has a specific word for the nature of that direction—teleonomy, meaning that a system looks as though it had a goal without there being any subject behind it that set an intention. To read the sentence as "capital has a will of its own, it is a subject" is to read it backwards³⁴. One section, headed "The Turing Test", concerns the place of human beings in this self-running process:

"Man is something for it to overcome: a problem, drag."

The sentence writes of man from capital's side, as something "for it to overcome"; at most the text supports two judgments—the process will not mourn the replacement of man, and it does not promise that human life will improve—while whether Land himself welcomes or celebrates that replacement, the text does not say³⁵. The text's one use of "accelerate" in the imperative comes where he quotes Deleuze and Guattari's question, followed at once by a sentence set as its own paragraph:

"Nothing human makes it out of the near-future."

Land himself never says in the first person "we should accelerate", and the noun "accelerationism" does not appear anywhere in the text³⁶. The essay repeatedly writes the future as a force that precedes the present and then invades it: Deleuze and Guattari's schizoanalysis "comes from the future"; the entanglement with entropy deviations is "camouflaging an invasion from the future". The CCRU (the Cybernetic Culture Research Unit, where Land then worked) summed up this set of moves with the word hyperstition: a fiction that brings itself about—a claim about the future that, told and spread and believed, can move people to act in ways that make the future it describes come true. The word does not occur in the literal text of "Meltdown"; this report uses it to summarise the essay's narrations, not as a conclusion checked word for word against the original³⁷. "Nothing human makes it out of the near-future" can therefore be read in at least two ways: as the author's sincere literal assertion, or as a performance within a hyperstitional frame—in saying it, the author is also trying to make it so. The essay also has a whole passage that writes the reader into the fiction: "Meltdown has a place for you as a schizophrenic HIV+ transsexual chinese-latino stim-addicted LA hooker with implanted mirrorshades and a bad attitude", who has just shot three Turing cops with a highly cinematic pistol³⁸. "Meltdown" does not state facts in a plain register alone, and this report does not choose between the two readings.

1972: the question at the end of Anti-Oedipus

The earliest text in the material is Deleuze and Guattari's Anti-Oedipus, published in 1972, twenty-two years before Land's essay. Section 9 of Part Three, "The Civilized Capitalist Machine", argues that capitalism runs on "decoded flows"—desire, population, wealth, labour-power, things that can flow, no longer fixed to land, status and kinship (deterritorialisation), their former qualitative determinations removed and converted into abstract, mutually convertible quantities (decoding). At the end of the section the authors pose a question. Having ruled out withdrawal from the world market (which they call "a curious revival of the fascist 'economic solution'"), they write the sentence since quoted more than any other:

"Not to withdraw from the process, but to go further, to 'accelerate the process,' as Nietzsche put it: in this matter, the truth is that we haven't seen anything yet."

That is the last sentence of Section 9; what follows is the heading of Section 10, "Capitalist Representation", and the discussion turns to writing, linguistics and the theory of the state. Within the two and a half sections of the book read for this report, the authors do not go on to discuss what to accelerate, how, or by whom³⁹. In the paragraph where the sentence sits, the attribution "as Nietzsche put it" carries no footnote number: every other direct quotation in the book is followed by a note number, this one alone is not, and the nearest notes before and after are 83 and 84. The book also has a complete apparatus of notes at the end, which lies outside the range verified for this report; all that can be recorded is that within the body text read there is no corresponding note, not that the book nowhere gives a source for the sentence⁴⁰. Within the verified range, the sentence is a question posed at the end of one section's argument that the authors do not develop into a new programme, and the source of "as Nietzsche put it" remains unconfirmed.

Continuity of vocabulary and difference of thought, 1972–2023

In 1972 Anti-Oedipus posed two options at the end of one section's argument and chose "to go further", without saying who should act or how. In the 1990s Land quoted that question, described capital as a process already running by itself, and held that man would be replaced by it—the subject of the sentence shifting from "we" to "capital". In 2008 Noys used "accelerationism" to criticise that line of thought, not to endorse it. In 2013 a group of left-wing authors adopted the originally critical term as a self-description, granting that Land "captured … most acutely" the relation between capital and acceleration while criticising him for confusing speed with acceleration. In 2022 a group of anonymous online authors coined "e/acc" on the model of "effective altruism", replaced Land's thermodynamic rhetoric with a named argument from physics, and added an ultimate value the left manifesto had not proposed: the persistence and flourishing of consciousness itself. In 2023 Andreessen adopted this vocabulary in a programme published openly by a mainstream capital institution under a real name, and replaced Land's "man is drag" with "the machines work for us".

Over fifty-one years the word "accelerate" appears five times in texts that can be confirmed; the inheritance of words like "technocapital" is more complicated. e/acc in 2022 took over Land's specific phrase "technocapital singularity"; but the more specific phrase Andreessen attributes to Land in 2023, the "techno-capital machine", this report could not find in the full text of "Meltdown", and whether Land ever wrote that exact combination cannot at present be confirmed. That does not alter the fact that Andreessen used the phrase, but it limits what the material can support: it shows how Andreessen stated his source, not that the attribution itself is accurate⁴¹. To the question "when the process reaches its end, is there still a place for human beings?", four texts give four incompatible answers: Land says none ("Nothing human makes it out of the near-future"); e/acc's statement of principles says there need not be ("no particular allegiance to the biological substrate"); Andreessen says there must be ("the machines work for us"); Aschenbrenner in 2024 says there still is, but it needs the protection of the state. Each later text is not a simple heir of the earlier ones: each keeps part of what came before and changes the rest in a different direction.

Three gaps in the evidence limit judgment on the relations between these texts. First, this report can verify where the 1972 question stands and that Land quoted it, but has no material showing how Land read and understood the original; that Land deliberately rewrote the question is an inference from the difference between the texts, not a fact the material records. Second, nothing shows that the authors of the 2013 left manifesto knew of or responded to Noys's naming of 2008. Third, this report searched e/acc's published output for treatment of left accelerationism and found only two marginal texts, both of which mention Williams and Srnicek neutrally as background in the lineage, without engagement and without endorsement. Left accelerationism and e/acc share a name but have almost no dialogue in the public texts⁴².

The 2013 left manifesto places Marx before Land—"It is Marx, along with Land, who remains the paradigmatic accelerationist thinker"—and when Noys traced the line of thought in 2008 he too traced it to Marx. Two independent sources point to an intellectual source this report has not yet included. Adding material from Marx would not change the textual differences already checked item by item, but it would change the time-span of the sources: if the origin is traced to Marx, the texts of 1972 to 2023 form only the later part of the history⁴³.

3. What its core ideas are, and where the camps split

The present-day accelerationist texts all discuss the development of technology and capital, but they do not all grant that the process can propel itself. Land and e/acc stress that autonomous drive; left accelerationism and Buterin hold that the direction of technology needs political action or human choice. The camps also differ on what is to be accelerated, who benefits, and what the argument rests on; they answer differently whether human beings still have a place in the end state of technological development, and who has the right to stop or slow it. Some of the people concerned also advocate exiting democratic politics, but the material shows that this anti-democratic position does not rest on the same view of technology as accelerationism.

What is accelerated, on what grounds, and under what political conditions

The founding text of e/acc starts from thermodynamics: life arose from "dissipative adaptation", a non-equilibrium process in which matter reconfigures itself to extract energy from its environment and convert it into entropy, and the universe is probabilistically biased towards such futures; intelligence is a specialisation of this adaptive principle at shorter timescales; at the social scale, capitalism dynamically reshapes human civilisation so that utility in the environment is captured and used to maintain and expand civilisation. On that basis the post states e/acc's goal as acknowledging this multi-scale adaptive principle and helping it accelerate rather than slow⁴⁴. Andreessen's manifesto gives an entirely different argument: growth has only three sources, population, natural resources and technology; the first two have limits, so technology is the only perpetual source; and the grounds are almost all economic—Hayek's Knowledge Problem, Ricardo's comparative advantage, Nordhaus's finding that innovators capture only about two percent of the economic value they create. This report searched the 565 lines of the manifesto for seven roots, dissipat, adaptation, thermodynam, entropy, fluctuation, physics and physicist, and found none; "England" occurs once, in the epigraph's "Norman-conquest England", not the physicist who proposed dissipative adaptation⁴⁵. e/acc supports "acceleration" with a named argument from physics; Andreessen supports the same claim with a named argument from economics.

The left-accelerationist position differs from both more fundamentally. The 2013 manifesto states: "We want to accelerate the process of technological evolution. But what we are arguing for is not techno-utopianism. Never believe that technology will be sufficient to save us. Necessary, yes, but never sufficient without socio-political action."⁴⁶ No corresponding formulation can be found in the right-wing texts in this report's collection—neither e/acc nor Andreessen says in the body of their texts that technology is insufficient and political action is also needed. The acceleration the left manifesto wants aims to "change the parameters themselves", not to raise the speed at which capitalism runs within its existing parameters; that is the concrete sense of its charge that Land "confuses speed with acceleration". All the camps grant that technocapital has a self-propelling drive and argue for furthering it, but three disagreements remain: whether that drive is to be explained by physics or by economics; whether technological acceleration must be conditional on socio-political action; and whether acceleration should keep or change capitalism's existing parameters.

The place of human beings in the end state of technological development

Whether human beings still have a place when the process reaches its end state is the second difference among the camps. Four texts give four incompatible answers. The table sets their own words and sources side by side; the limits of the evidence for each row follow.

The first row carries one limit: "Nothing human makes it out of the near-future" can be read as the author's sincere literal assertion or as a performance within a hyperstitional frame, and the text is not enough to decide between them. The table can support only "the situation this text depicts leaves no place for humans", not a further judgment on whether the author regards that situation as a fact he believes. The second row also needs qualifying: e/acc's two founding texts use two registers. The inward-facing statement of principles says "no particular allegiance"; the outward-facing recruiting post says "Human flourishing is one of our core values! We are humans, and we love humans." The table's "need not be any" follows the first register; "need not" rather than "should not" preserves the limit set by the second⁴⁷. The third row's words are unambiguous, but must be read together with the cost Andreessen concedes: that machines make decisions for people and take away part of their choice is, he says, "undoubtedly true", though the freedom brought by material abundance more than offsets the loss.

The fourth row is not a ready-made quotation but a summary of three claims in a newer text. In June 2024 Leopold Aschenbrenner, formerly of OpenAI's superalignment team, published the essay series Situational Awareness: The Decade Ahead, proposing a complete causal chain: AGI (artificial general intelligence, AI able to match or exceed human performance across a broad range of cognitive tasks) will arrive within a few years; it will soon develop into superintelligence far exceeding humans in most important capabilities; superintelligence will bring a decisive military advantage, so who holds that advantage is of the first importance⁴⁸. A lead of a year or two, even of months, could in his judgment produce an advantage like the coalition's over Iraq in the Gulf War: the US-led coalition destroyed the world's fourth-largest army in a 100-hour ground war, with 292 coalition dead against 20,000 to 50,000 Iraqi dead⁴⁹. From this he draws three interlocking claims: model weights and algorithmic secrets must be protected to national-security standards; development should move from private companies into a government-led joint effort, which he calls "The Project"; and a US-led coalition of democracies should use its lead to require other countries to accept non-proliferation rules modelled on Atoms for Peace and the IAEA⁵⁰. He frames this as a descriptive argument—a judgment that government will in fact take over, not that it should. He also explains why he opposes dictators holding superintelligence: "I believe in freedom and democracy, strongly, because I don't know what the right values are."⁵¹ The table's "humans remain, but must be protected by the state" summarises these three claims: humans still have a place in the end state, but Aschenbrenner holds that it needs protection at the level of the state.

The table can hold only one formulation from each text. e/acc's two founding texts use two registers, so the second row follows the statement of principles; Land's sentence can be taken literally or hyperstitionally, so the first row follows the literal reading. The table shows only that four texts have publicly expressed different positions, not that each text holds only the position listed on every occasion.

The power to slow or stop technological development

The disputes of 2023 to 2025 mainly concern two questions: whether technological development can be stopped or markedly slowed, and who has the right to decide. The three texts use different concepts. e/acc discusses "top-down control"; Vitalik Buterin, founder of Ethereum, proposes d/acc—defensive, or differential, accelerationism—which would speed up technologies that favour defence while slowing those that expand offensive capability; he also proposes a "soft pause button" built into AI hardware, pausing most industrial-scale compute for a period through trusted hardware and multi-party authorisation. Aschenbrenner discusses the government-led "Project" and a non-proliferation regime enforceable where necessary. All three concern the power to slow or stop technology, but they do not share the word "switch"⁵².

e/acc opposes top-down control. The founding text invokes "Fisher's fundamental theorem of natural selection" for a statistical argument: the rate at which a population's average fitness rises is proportional to the variance of fitness within it—to keep variance is to keep the capacity to go on adapting. On that basis it makes five criticisms of technocratic control: such control usually relies on simplified models to predict the consequences of intervention and can take in only what humans can understand; complex chaotic systems change constantly and social hierarchies cannot acquire and transmit complete information, so accurate control is in practice all but impossible; even with full observability (the post's example is a "CCP-style panopticon"), humans cannot effectively control fractal and chaotic systems with limited models; top-down hierarchies are not fault-tolerant when the top decision-maker is corrupted or when information degrades on its way up; and controllers, for ease of management, tend to suppress variance across civilisation, which weakens the system's capacity to adapt—over a long enough time, dynamically adapting systems overtake static ones, and suppressing variance ends by producing the opposite of what was intended⁵³. This is not the same as "there should be no constraint of any kind": the one founding text that directly discusses the possibility of long-term lock-in writes that "panopticon driven techno authoritarian lock-in for a thousand centuries or more" might simply happen, that it is not exactly up to any one person or group, and that the way to fight it is to keep building technology in an open way and let the world equilibrate at each iteration rather than "ban the scary thing"⁵⁴. The more accurate summary is therefore that e/acc opposes control by a single top-down authority, not constraint of any kind.

Buterin proposes control held jointly by many parties and applied equally to all. Buterin's 2023 essay first gives a judgment directly contrary to the techno-capital machine: this "does not happen automatically, and requires intentional human effort. The ozone layer is recovering because, through international agreements like the Montreal Protocol, we made it recover"; what solves these problems is intentional action—through public discussion and culture shaping the views of governments, scientists, philanthropists and businesses—not an unstoppable techno-capital machine⁵⁵. That sentence directly denies that technocapital propels itself, and it is Buterin's clearest disagreement with the Landian narrative of technological autonomy. In his second essay, of 2025, he adds concrete slowing measures: in a critical period, reduce worldwide available compute by about ninety to ninety-nine percent for one to two years, by fitting industrial-scale AI hardware with a trusted hardware chip that allows it to keep running only if it receives three signatures a week from three international bodies, at least one of them not affiliated with any military; because the signatures are device-independent, each body can only allow all devices to run or pause all of them, never release a single machine⁵⁶. He explicitly opposes Aschenbrenner's proposal: "To me, this approach seems risky, and could combine the flaws of multipolar races and centralization. If we have to limit people, it seems better to limit everyone on an equal footing, and do the hard work of actually trying to cooperate to organize that instead of one party seeking to dominate everyone else."⁵⁷ The phrases "held by many parties" and "applied to all" come from this sentence. But the proposal carries an unresolved tension of its own: Buterin's 2023 essay had written, "Across the board, I see far too many plans to save the world that involve giving a small group of people extreme and opaque power and hoping that they use it wisely"; his own 2025 soft-pause scheme would hand a global pausing power to three international bodies, and the essay does not go on to say who would compose them or by what procedure they would decide⁵⁸. The material is not enough to judge whether this contradiction overturns the scheme as a whole.

Aschenbrenner proposes control by the United States and its allies. Aschenbrenner calls his position "AGI Realism", a position that treats AGI's rapid arrival, its military significance and state takeover as near-term practical constraints⁵⁹. His three claims are: protect the labs to national-security standards, move development into a government-led joint project, and have a US-led coalition use its lead to impose a non-proliferation regime. In Part V of the series he sets e/acc beside the doomers and finds both unserious, calling e/acc participants people who deny the risks by denying AGI, "real stagnationists". He also sums up e/acc as "some underwhelming accelerationism in my book". The phrase is derogatory, but it presupposes that another, acceptable accelerationism exists. The material shows only that he rejects the e/acc label and sets up AGI Realism instead; it does not show that he rejects the word "accelerationism" itself⁶⁰.

The three positions are incompatible. e/acc opposes top-down control; Buterin wants the means of control held jointly by many parties and applied equally to all; Aschenbrenner wants it held by one nation and its allies. All three regard themselves as supporters of technological progress; the dispute is over who has the right to slow or stop it.

Among the three there are six possible directions of response, but only two have text that can be checked: Aschenbrenner criticises e/acc by name, and Buterin opposes Aschenbrenner by name. A third direction is a one-sided classification by Buterin: in the compatibility list at the end of his first essay he writes, "If you are an e/acc, then d/acc is a subspecies of e/acc - just one that is much more selective and intentional"; that does not show that the e/acc side accepts the classification⁶¹. The other three directions are in different states. On e/acc's attitude to d/acc, the searches made found no verifiable public response. On e/acc's attitude to Aschenbrenner, no search has been made. In reading the whole Aschenbrenner series, no discussion of d/acc by name was seen, but no dedicated search was made either, so this can be stated only as "not seen within the range read", not as "he never responded"⁶².

The relation between the autonomy of technology and anti-democratic politics

"The Cathedral" is Yarvin's name for what he holds to be the actual ruling class; Land uses the word for a ruling power that must be identified and that does not gain its power through democratic election.

Some of the people concerned both describe technocapital as a self-propelling process and commit resources to opposing specific bills. Whether the two actions rest on the same view of technology has to be checked separately. Five neoreactionary (NRx) materials added to the collection in August 2026 concern Curtis Yarvin (writing as Mencius Moldbug) and Peter Thiel, and allow a comparison of these people's judgments on the autonomy of technology and on democratic politics. The material confirms financial and intellectual traffic among them, but does not show that neoreaction forms a fourth school of accelerationism. Only one primary text directly links e/acc to the idea of exit and of dividing sovereignty.

The one place where the e/acc founding text mentions the "network state". The ninth step of the e/acc founding text carries a footnote: that structures of information transmission and control need to adjust at different scales and timescales is something that "particularly excites us about the concept of Network States by Balaji"—if code is law, one could create an archipelago of network states with differing governance mechanisms, competing for members and capital⁶³. This footnote is the only primary passage in this report's collection that directly links e/acc to ideas of exit and of dividing sovereignty; its grade is primary full text. It shows only that e/acc's founding text has a footnote expressing interest in the "network state". The reading notes on the neoreactionary materials mark the relation between the "network state" and Yarvin's idea of exiting democracy as "to be verified, not for the main text", because nothing in the collection shows a direct citation between them⁶⁴. The material therefore cannot show that neoreaction is a fourth school of accelerationism.

Yarvin's corporate state and anti-democratic claims. In 2007 Curtis Yarvin, under the pen-name Mencius Moldbug, published "A Formalist Manifesto", proposing an alternative ideology he called "formalism". Its central test is a formula: violence equals conflict plus uncertainty. Conflict will persist—"while there are wallets in the world", in Yarvin's phrase—but if a clear, unbreakable rule settles property in advance and removes the uncertainty, people have no reason to resort to violence over it. From this he derives the central claim: confirm who actually holds what, issue formal title to the actual holders, and do not discuss who ought to hold it, thereby avoiding "redistribution" altogether⁶⁵. Turning to the state, he calls the United States a corporation and Americans "corporate serfs", and says the modern distinction between "private company" and "government" is a fairly recent development. He proposes changing the state by "reformalization": work out the actual distribution of power in the United States and assign shares to reproduce it, so that ownership and control coincide⁶⁶. The alternative polities he names are city-states: Singapore, Hong Kong and Dubai are "remarkable for their prosperity and their relative absence of politics", and any restrictions on personal freedom they maintain "seem primarily aimed at preventing the development of democracy"⁶⁷. The essay concludes that democracy is "an ineffective and destructive system of government", and that the post-war West succeeded not because of democracy but because of its absence—the symbolic structure of democracy retained, almost all real decision-making handed to non-democratic civil servants and judges.

Neo-cameralism means running the state in the manner of a company whose ownership and management coincide. Neither that term nor the slogan "No voice, free exit" occurs in the 2007 manifesto; both appear in Land's later essay quoting and paraphrasing Yarvin's idea.

Yarvin's scheme and the e/acc footnote mentioned above are verbally close, both proposing to divide sovereignty and let residents express their choice by moving away; but the material cannot show a direct citation between them, and no transmission of ideas can be inferred.

Thiel on the autonomy of technology and on democracy. On 13 April 2009 PayPal co-founder Peter Thiel published "The Education of a Libertarian" at Cato Unbound, with a sentence quoted many times since:

"Most importantly, I no longer believe that freedom and democracy are compatible."

His reasons came in part from changes in the electorate—since 1920 the vast increase in welfare beneficiaries and the extension of the franchise to women "have rendered the notion of 'capitalist democracy' into an oxymoron". In an editor's note on the same page he responded to the reaction that sentence provoked: "It would be absurd to suggest that women's votes will be taken away"; he said plainly that he did not think any class of people should be disenfranchised, but had little hope that voting would make things better—and that note must be quoted with the earlier sentence, or his position is read as more extreme than the text⁶⁸. His way out is not to improve politics but to get around it: "In our time, the great task for libertarians is to find an escape from politics in all its forms." He lists three technological frontiers: cyberspace (PayPal's founding vision, he says, centred on creating a new world currency free from all government control), outer space (which he judged would not happen before the second half of the twenty-first century), and seasteading—autonomous communities on floating platforms in the open sea, which he judged far more realistic than space and explicitly supported⁶⁹.

Having set out the three frontiers, Thiel explicitly rejects the position that technology will by itself guarantee a free future:

"The future of technology is not pre-determined, and we must resist the temptation of technological utopianism — the notion that technology has a momentum or will of its own, that it will guarantee a more free future, and therefore that we can ignore the terrible arc of the political in our world."

Thiel, Land and accelerationism are often placed in one intellectual line, but the two men judge oppositely on whether technology can develop by itself. Land's central thesis is that capital has "teleonomy", exhibiting a direction of its own in a non-instrumental way, with man as the drag it must overcome; Thiel calls the notion that technology "has a momentum or will of its own" a temptation to be resisted, and places his hope in "the effort of a single person". What the two share is a negative judgment of democracy, not a view of technology: Thiel says freedom and democracy are incompatible; Yarvin says democracy is ineffective and destructive⁷⁰. This judgment applies only to Thiel's public text of 2009 and cannot be generalised to his position over time; if a later public text of his accepts the notion that technology has a momentum of its own, the judgment no longer holds⁷¹.

How "The Dark Enlightenment" cites Thiel and Yarvin. In Part 1 of a later text, "The Dark Enlightenment" (c. 2012–2013), Land himself quotes Thiel's sentence of 2009—but what he quotes is the judgment about democracy, that freedom and democracy are incompatible, not the judgment about the autonomy of technology⁷². "The Dark Enlightenment" consists of Parts 1 to 4, Part 4 being divided into subsections a to f. This report read only Part 1 and Part 4f in full; subsections 4a to 4e, whose headings indicate that they concern race, have not been read. What follows therefore covers only the two parts read and is no judgment on the character of the whole⁷³. In Part 1 Land sets out the distinction "Voice vs Exit": voice is democracy itself, modelling the state as a representation of popular will; if mass self-expression through voting is a world-consuming nightmare, then adding more voice does not help—what libertarians are choosing is a silent exit, not more voice⁷⁴. He then introduces Yarvin at length, calling him the "supreme Sith Lord of the neo-reactionaries", quotes Yarvin's sentence "To a neocameralist, a state is a business which owns a country", and describes the carrying out of the scheme in three steps: demolish the democratic myth that the state belongs to its citizens; identify the real ruling class (which Yarvin calls "the Cathedral"); and appoint a chief executive, formalising the interest of the state precisely as the maximisation of long-term shareholder value⁷⁵. Part 4f discusses what Land calls the "bionic horizon": science and technology form a self-improving loop in which to understand a thing is to be able to modify it, and the biological identity of humanity is progressively dissolved by biotechnology; he argues that an elite group should use genetic technology to produce, within ten generations, a new group beyond the present human form, and the section closes with "It's time to move on."⁷⁶ "The Dark Enlightenment" discusses how a group of people exits democracy and formalises sovereignty as shares—the agent is human; "Meltdown" discusses capital running itself with man as the drag it must overcome—the agent is capital. Neither text keeps a final place for present humanity, but for different reasons: in "Meltdown", capital's self-running replaces man; in "The Dark Enlightenment", part of an elite deliberately uses technology to produce a new group beyond the present human form⁷⁷.

The money between Thiel and Yarvin: press reports. Whether there are direct funds between Thiel and Yarvin rests so far only on press reports: a 2016 CoinDesk article says that Tlon, the company Yarvin co-founded to develop the computing platform Urbit, drew backing from Silicon Valley heavyweights including Andreessen Horowitz and PayPal co-founder Peter Thiel; a CoinDesk commentary obtained via a Yahoo Finance mirror says Thiel invested in Tlon and in at least one other Urbit-ecosystem startup, Hypertext Vienna⁷⁸. Both sentences concern Thiel's personal investment in a company, name no amount, and have no official announcement to check against, so they cannot be used to claim that Thiel funded Yarvin personally. A 2025 Washington Post report gives a more specific figure: Thiel's venture firm Founders Fund "invested $250,000 in Tlon, a start-up Yarvin created to build out his computing platform"; the same report relays an email from Yarvin to the right-wing commentator Milo Yiannopoulos saying that he spent election night 2016 at Thiel's home and had been "coaching Thiel", whom he described as "fully enlightened"—the email itself was made public by BuzzFeed News, the report relays it, and this report has not verified the original⁷⁹. The same report says Marc Andreessen's venture firm was among Tlon's early investors, and that Andreessen "has described Yarvin as a friend and has cited his ideas, according to the investment database PitchBook"—again relayed, and this report has not verified the database record⁸⁰.

This material links Yarvin to more recent American political reality, but the links have to be handled in layers. The Washington Post report itself is a journalist's compilation of multiple sources and can be used directly as a secondary report: it says two anonymous advisers to the Department of Government Efficiency (DOGE) said that Yarvin's acronym "RAGE", for "Retire All Government Employees", was "the most crisp articulation" of what DOGE is trying to achieve, one of them saying, "It's an open secret that everyone in policymaking roles has read Yarvin"; the report also records Yarvin's own markedly negative assessment of DOGE—he compared it to an orchestra of chimpanzees trying to perform Wagner, and said, "In the worst aspects of DOGE, there's this aspect of the incel who gets mad at the girl who won't sleep with him … That's not a powerful attitude."⁸¹ The report also names Vice-President Vance and Marc Andreessen among Yarvin's known readers; Vance mentioned Yarvin on a right-wing podcast in 2021 and set out his own idea of purging government employees⁸². In his own article "The pleasure of error", published on Gray Mirror on 29 January 2025, Yarvin admits that his earlier judgment that a second Trump term would take only "small, performative steps", as the first had, was a mistaken estimate, and writes that "RAGE wasn't supposed to be a voluntary program"—his own public assessment of the gap between his idea and the policy actually carried out, and directly quotable⁸³.

On the earlier and more systematic intellectual relations among Yarvin, Thiel, Vance and Andreessen, the collection also holds a French academic paper of 2026: Jean-François Durand Folco, "Les racines idéologiques du fascisme technologique" ("The ideological roots of technological fascism"), in Revue Possibles. The author explains that the paper combines two chapters of a 2025 book of his and is "a political essay at once theoretical and combative"; he also writes plainly that "we have not based this reflection on fieldwork or on precise empirical research", the method being "interpretive and critical"⁸⁴. The paper's characterisation of the whole phenomenon (technological fascism, an "AI-assisted coup") is therefore not taken as verified; only its cited factual layer can be recorded, and must be marked as second-hand. The paper says that Thiel "noticed Yarvin and immediately funded him, making him his protégé" (cited from Simon 2025, original not verified by this report); that Vance "acknowledged in a 2024 interview Yarvin's major influence on his thinking" (cited from Tehari 2025, not verified); and that the Thiel–Yarvin relation forms part of what the author calls the "key trio", "Thiel/Vance/Yarvin", with Yarvin as the theoretical "brain"⁸⁵. It also says that Andreessen "claims this inheritance", his "Techno-Optimist Manifesto" "invoking Marinetti's Italian futurism and taking up Yarvin's accelerationist ideas" (cited from Andreessen 2023). That assertion conflicts with the primary material already checked and is not adopted: the intellectual source the manifesto names is Nick Land; Yarvin is not among the fifty-six patron saints, counted item by item, and neither "Yarvin" nor "Moldbug" occurs anywhere in the text⁸⁶. The paper's other specific assertions about personal relations may be cited only when marked "cited from so-and-so, such-and-such year, original not verified by this report", not as established fact.

The paper also says that Yarvin proposed two routes to realisation. The first is "patchwork", dividing sovereignty into many city-states, or gradually forming "network states" in the phrase of the entrepreneur Balaji Srinivasan (cited from Srinivasan 2022). This "network state" is the same idea the e/acc footnote refers to, but the paper's relay still cannot show a direct textual transmission among the three. The author's real-world example is Boca Chica Village, Texas—3.9 square kilometres, about three hundred residents, overwhelmingly employees of Musk's SpaceX—incorporated by referendum in May 2025 as a town named "Starbase" (relayed; this report has not verified the referendum record)⁸⁷. The second is the "butterfly revolution": seizing power by force and replacing democratic institutions with complete executive authority. The paper calls this a "legal coup" or "self-coup", and holds that the Department of Government Efficiency run by Musk in 2025 was carrying out this route (relayed; this report has not verified the author's full argument on that event)⁸⁸.

The conclusions this material supports are limited. The accelerationist camps generally accept that technocapital is propelling itself and argue for going along with it; Thiel, in the one public text verified so far, explicitly rejects the notion that technology "has a momentum … of its own", calls the future of technology "a deadly race between politics and technology", and holds that the outcome "may depend on the effort of a single person". That judgment runs opposite to Land's central thesis. The e/acc footnote on the network state, the financial relation between Thiel and Tlon, and Yarvin's public influence in Silicon Valley's political circles can show that accelerationism and exit politics involve some of the same people; they cannot show that these people share one view of technology. What the material shows them expressing in common is a negative attitude to democracy, not the same judgment about the autonomy of technology. That some of them oppose specific bills and argue for going around democracy may be precisely because they do not think technology will advance by itself, but demand that a few people have freedom of action unconstrained by democracy. Each fact and each judgment has to stand on its own; neoreaction cannot on this basis be summarised as a fourth school of accelerationism.

4. How it relates to effective altruism, and how to use it to read today's news

Accelerationism and effective altruism are related directly by name, and they share some premises about future consciousness and the responsibility to act. They also disagree explicitly. The people concerned use philosophical language in manifestos, posts and long essays, but not in the public documents of two policy disputes, over SB 1047 and over open weights. Comparing the two kinds of text confirms a difference of language between public claims and policy argument, but that difference alone does not settle whether the claims drove the policy action.

How e/acc, d/acc and effective altruism compare

e/acc's name was modelled on effective altruism (EA), but the relation between the two is not only one of hostility or imitation. The founding texts of e/acc set out their likenesses and differences in a two-column table. Under "Like EA" there are three items: both are prescriptive, claiming not only to describe reality but that people should act; both value more positive-valence consciousness as good; both value zero recognisable consciousness in the universe as the absolute worst outcome. Under "Unlike EA" there are five: e/acc doesn't advocate for modernist technocratic solutions to problems; it criticises effective altruists who "wish everything would just slow down" as passively risk-averse; it isn't human-centric—as long as it's flourishing, consciousness is good, human or not; it isn't in denial about how fast the future is coming; and it rejects the desire for a panopticon it says is implied by longtermist EA beliefs⁸⁹. This text by one of the parties lists effective altruism as an adversary and at the same time acknowledges three shared premises.

Buterin connects d/acc to both e/acc and effective altruism. In the compatibility list at the end of his first essay he writes: for an e/acc, d/acc is a more selective and intentional subspecies of e/acc; for an effective altruist, d/acc is a re-branding of the effective-altruist idea of differential technology development, with more emphasis on the values of freedom and democracy. He also explains the second link: the core idea that some technologies favour defence and should be promoted while others favour offence and should be discouraged "has roots in effective altruist literature under a different name: differential technology development"⁹⁰. In his "Enemy" section Andreessen names no organisation, only a list of concept labels: existential risk, sustainability, ESG, the Precautionary Principle, trust and safety, tech ethics, de-growth. A check of the full text confirms that the list contains no person's name and no institution's⁹¹. These concepts are usually understood as belonging to the body of thought of effective altruism, but this report holds no primary text of effective altruism and so cannot verify the correspondence independently. The material confirms only that what Andreessen lists are concepts, not organisations; whether the concepts amount to effective altruism needs another batch of material.

Public language in two AI policy disputes

Open weights means releasing the parameters a model has after training, so that others can run, modify or further train it themselves.

Two disputes that shaped policy supply public documents that can be searched word by word. The documents searched, the results, and the limits of the evidence are as follows.

The first: California's SB 1047, 2024. On 2 August 2024 a16z's chief legal officer Jaikumar Ramaswamy wrote to California State Senator Scott Wiener opposing Wiener's SB 1047, a bill that imposed a set of safety obligations on frontier AI models trained above a given threshold, including certified safety protocols, a shutdown capability and third-party audits⁹². The letter runs to fourteen pages; the copy obtained for this report is one hosted by a third-party site, Engine Advocacy, bearing an electronic signature, from which 492 lines of plain text were extracted, read in full, with seven quotations checked word for word⁹³. Its argument is legal and industrial from beginning to end: it opens with three more fundamental legal arguments—that the bill violates the Constitution's dormant commerce clause, violates the federalist principle that a state's legislative power stops at its border, and that existing federal criminal law already addresses most of the risks the bill targets—then answers Wiener's seven points one by one, on the training and fine-tuning thresholds, certification under penalty of perjury, civil liability and enforcement, the effect on California's technology industry, the definitions of "artificial intelligence" and "reasonable assurance", the scope of the full-shutdown capability, and the rules and burdens falling on open-source models⁹⁴. The letter cites two constitutional cases, more than ten other court decisions and three federal criminal statutes; its industry data come from Epoch AI and CompTIA; it also cites an accounting-journal article and a Congressional Research Service report. In those 492 lines this report searched word by word for four English roots: accelerat (covering accelerationism, accelerationist and every variant), techno-capital, e/acc, and philosoph (covering philosophy, philosophical and variants). All four return zero⁹⁵. The letter contains no form of "accelerationism", "techno-capital machine" or "e/acc", and no word with the root "philosoph".

The second: the open-weights dispute, July 2026. On 24 July 2026 a group of technology companies published a jointly signed open letter, "Open Weights and American AI Leadership", hosted on the corporate website of one signatory, Microsoft. The signatories include OpenAI, Google, Meta, Microsoft, Amazon, Nvidia and Palantir, companies known for closed-source flagship models, alongside Mistral, Hugging Face, EleutherAI and Nous Research, whose main business is open weights—a group not homogeneous in release strategy, who cannot be assumed to hold one position; all that can be said is that the letter itself opposes restricting open-weight releases⁹⁶. The letter makes four independent arguments: expanding access to the AI economy; strengthening competition; giving customers more control and avoiding lock-in to a single vendor; and the possibility that openness itself is a path to safety. Its historical analogy is the open-source software movement of the 1980s, which "created a shared foundation of knowledge on which generations of American engineers and entrepreneurs built their institutional sovereignty"⁹⁷. The letter admits that open weights carry risk—"To be sure, open weights carry real and distinct risks. Once released, the weights are beyond the original developer's control, and modified versions are difficult to trace or reverse."—and then turns: "But the right response to this risk is not to prohibit open weights." It then gives the fullest sentence of its safety argument: "Just as open-source software demonstrated that transparency can be more secure than obscurity, AI safety may depend on giving more people the ability to test and strengthen the models on which society relies."⁹⁸

Three days later, on 27 July 2026, Anthropic's chief executive Dario Amodei published "Our position on open-weights models" on the company's website; Anthropic had not signed the open letter. The first sentence denies a claim then in circulation:

"Anthropic has never advocated for a ban on open-weights models."

Distillation means training one model on the outputs of another so that the second learns the first's capabilities; "industrial-scale distillation" here means extracting a strong model's capabilities at large scale.

The essay then affirms open-weight models that lack dangerous capabilities: "Open-weights models that don't have dangerous capabilities are a public good: they don't cost anything besides the compute needed to run them, and they provide value to businesses, developers, and researchers." Amodei writes that what concerns him is two kinds of risk: authoritarian governments using models stronger than America's to seek permanent military advantage or to carry out deep repression; and powerful models misused for cyberattack or biological attack, possibly with serious alignment problems as well. Against these he proposes three measures he and Anthropic have consistently advocated: not selling powerful chips or chip-making equipment to China and cracking down on the related smuggling; cracking down on industrial-scale distillation; and mandatory pre-release safety testing for every sufficiently powerful model, open or closed⁹⁹. He writes explicitly: "A blanket ban on open-weights models is neither the correct remedy nor something we have called for."

Amodei answered the open letter directly without opposing it wholesale: "This brings me to the open letter. I agree with much of it: open weights expand access to the AI economy, they strengthen competition at least for some use cases, and they give customers greater control." He explicitly rejected only one claim: "But I don't agree with the letter's assertions that open-weights models necessarily make it easier to develop safeguards or that broad access to capabilities necessarily helps defenders more than attackers. It seems at least as likely to me that the opposite will be true." His reason concerns biological risk: a sufficiently powerful model might be able to weaponise a pandemic-class virus quickly from widely available materials, whereas defence "is a multi-year operational task in the best case"¹⁰⁰. A footnote states the premise of that judgment: "what I believe currently keeps us safe in biology is not 'defenders', or even the availability of materials, but a negative correlation between intellectual capability and desire to commit catastrophic harm". "What I believe" marks this as a personal belief; the essay offers no checkable measurement or source for the premise¹⁰¹.

Word-by-word search of the two open-weights texts. Both the open letter and Amodei's essay were taken from their official places of publication; the extracted texts in the collection run to 301 and 148 lines respectively¹⁰². Both were searched by the same method as the SB 1047 letter, for the four roots accelerat, techno-capital, e/acc and philosoph. All eight results are zero. This public exchange used no "accelerationism", no "techno-capital machine", no "e/acc", and no word with the root philosoph. The two policy disputes, searched by the same method, both return nothing.

What the word search can and cannot show. The search supports exactly one statement of defined scope: in the text extracted from these four documents, none of the four roots occurs. It cannot show that the authors no longer hold these ideas, because written material shows only what an author wrote in a document, not directly what the author thought at the time. The material admits two explanations and is not enough to say which is better grounded. The first is that when these institutions took part in legislative discussion they deliberately stopped using the philosophical language addressed to supporters and the public: manifestos and posts mobilise and appeal, and can use "techno-capital machine" or "dissipative adaptation"; persuading legislators needs statute, case law and industry data that can go straight into the consideration of a bill. The second is that these four documents belong to the genres of legal opinion and official corporate statement, which ordinarily do not use philosophical vocabulary whether or not the author believes the philosophy. Comparing the two explanations would require other documents from the same institutions addressed to legislators or regulators, on different topics but in the same genre, to test whether the vocabulary is equally absent there. That material is not in the collection and the comparison has not been made.

Amodei is the chief executive of the company whose models the system writing this report runs on, so the writing system has an interest in this dispute. The positive claims of both texts and their respective qualifications are set out in the original words to allow independent checking: the open letter admits "real and distinct risks"; Amodei states his premise in a footnote as a belief and gives no measurement. The material is not enough to answer whether open weights favour defenders.

Three questions for analysing public claims of this kind

"Accelerationism" appears in texts of mobilisation and self-description and not in the policy documents examined here. Searching for the label alone is therefore not enough to judge the specific position of a person or institution. What is more checkable in the public texts is how they understand the drive behind technological development, the place of human beings in the future, and who holds the power to slow or stop that development.

When analysing an "acceleration" claim made by a company, a person or a new manifesto, three questions can be checked directly. First, does the person or institution hold that technology has a drive of its own: Land and e/acc hold that it runs by itself, Buterin that its direction must be chosen deliberately, Thiel explicitly rejects the notion that technology "has a momentum or will of its own". Second, in the end state they imagine, is there still a place for human beings: Land says no, e/acc's statement of principles says there need not be, Andreessen says there must be, Aschenbrenner says there still is but it needs the protection of the state. Third, if technology becomes powerful enough to need stopping or slowing, who do they say has the power to decide: no controller at all, a mechanism formed jointly by many parties, or one nation and its allies. Direct evidence on each of these three positions can be sought in public texts, and is more reliable than searching for the label "accelerationism", which can be borrowed by different camps and may not appear in policy documents at all.

What the material cannot answer

Nine questions remain that the material cannot answer.

First, Marx's place in this lineage. Two independent sources—the 2013 left manifesto and Noys's naming of 2008—place Marx before Land, and this report holds no primary text by Marx; this is the most clearly marked gap. Second, Land's political evolution beyond "Meltdown" and "The Dark Enlightenment"; only bibliographic and structural records of two later texts have been obtained, not reliable full texts, so their content is not paraphrased. Third, how left accelerationism developed after 2013. The two authors' 2015 book Inventing the Future has not been obtained, and it needs checking whether the book kept, softened or withdrew the 2013 manifesto's claims about secrecy and exclusion. Fourth, the four roots return zero in all four legislative documents; the material allows two explanations and cannot say which fits the facts, and judging that requires further control samples of the same genre on different topics. Fifth, whether e/acc ever responded to Aschenbrenner has not been searched, so the answer is "unknown", not "no". Sixth, the five unread subsections of "The Dark Enlightenment" concern race; before judging the character of the essay as a whole, those subsections must be read. Seventh, whether there are officially verifiable funding records between Thiel and Yarvin, beyond press reports, has not been established. Eighth, whether Andreessen's "a manifesto of a different age and place" in fact reworks specific sentences of Marinetti's 1909 Futurist Manifesto. It is verified that Marinetti is among the patron saints, but Andreessen does not say which manifesto he means; one similar phrasing exists and has not been compared word for word. The list has fifty-six names, Marinetti is one, and his exact position in it has not been separately checked. The material therefore cannot confirm that what Andreessen reworked was Marinetti's manifesto. Ninth, the two reading notes on d/acc used in Section 3 have not passed every stage of verification, so the passages concerned only paraphrase the originals, without adopting the notes' analytical conclusions. After full verification it must be judged again whether the passage in Section 3 on the power to slow or stop technology needs further material.

Appendix: the map of people, organisations and texts

The relations of citation, funding and public comment among people, organisations and texts are shown in the relationship map above. The map is made and maintained separately, and its links have not each been run through this report's evidence checks. Readers tracing sources from the map should rely on the primary materials cited in the sections of the text.

Relationship map

The map shows only relations that can be checked in the texts. Solid lines are citation, naming, publication or support stated in the texts themselves; dashed lines are relations not yet confirmed, each labelled with which; orange lines are public opposition or criticism.

Bibliography

Compiled in the Chicago notes–bibliography system, listing only primary materials actually cited in the text or directly relied on in argument, alphabetically by author's surname or institution; materials signed collectively or institutionally by the institution's name; letters by the sender's surname. The sixteen items already in the earlier full-length version are carried over without re-examination; newly listed here are the materials added to the collection since then and used in this version.

Carried over from the earlier full-length version (see its Bibliography for publication details and retrieval records): Andreessen, "The Techno-Optimist Manifesto"; Aschenbrenner, Situational Awareness: The Decade Ahead; Buterin, "My Techno-Optimism"; Buterin, "d/acc: One Year Later"; Deleuze and Guattari, Anti-Oedipus: Capitalism and Schizophrenia; the four founding texts of e/acc; Land, "Meltdown"; Marinetti, "Le Futurisme"; Noys, "Accelerationism"; Noys, "Crash and Burn: Debating Accelerationism"; Ramaswamy to Wiener, August 2, 2024; Wiener to Y Combinator and Andreessen Horowitz, July 1, 2024; Williams and Srnicek, "#Accelerate Manifesto for an Accelerationist Politics".

Added in this version:

Amodei, Dario. "Our position on open-weights models." Anthropic, July 27, 2026. https://www.anthropic.com/news/position-open-weights-models.

Durand Folco, Jean-François. "Les racines idéologiques du fascisme technologique." Revue Possibles 50, no. 1 (2026): 66–77. DOI 10.62212/revuepossibles.v50i1.959. The collection record notes the author's method as "interpretive and critical", not empirical; this report adopts only the cited factual layer, always marked as taken at second hand.

Hochstein, Marc. "Urbit Is Web3, Weird and Wonderful and I Don't Care Who Made It." CoinDesk, September 24, 2022. Obtained via Yahoo Finance mirror: https://finance.yahoo.com/news/urbit-web3-weird-wonderful-don-141500880.html.

Higgins, Stan. "This A16z Startup is Building a 'Virtual Galaxy' for Bitcoin Nodes." CoinDesk, June 29, 2016. https://www.coindesk.com/markets/2016/06/29/this-a16z-startup-is-building-a-virtual-galaxy-for-bitcoin-nodes/.

Jamison, Peter, and Elizabeth Dwoskin. "Curtis Yarvin helped inspire DOGE. Now he scorns it." The Washington Post, May 8, 2025. Obtained from the syndicated reprint at The Spokesman-Review: https://www.spokesman.com/stories/2025/may/08/curtis-yarvin-helped-inspire-doge-now-he-scorns-it/.

Land, Nick. "The Dark Enlightenment." Published online c. 2012–2013; the original platform could not be independently verified. Obtained from a mirror repost on an anthology host; only Part 1 and Part 4f read closely, the rest (including Parts 4a–e, which concern race) unread; no judgment on the character of the whole.

"Open Weights and American AI Leadership." Open letter, full text and signatory list. Hosted on Microsoft's corporate site, corporate-responsibility/topics/open-weight/, July 24, 2026.

Thiel, Peter. "The Education of a Libertarian." Cato Unbound, April 13, 2009. https://www.cato-unbound.org/2009/04/13/peter-thiel/the-education-of-a-libertarian/.

Yarvin, Curtis (Mencius Moldbug). "A Formalist Manifesto." Unqualified Reservations, April 24, 2007 (archived repost).

Yarvin, Curtis. "The pleasure of error." Gray Mirror (Substack), January 29, 2025. https://graymirror.substack.com/p/the-pleasure-of-error.

Listed but not relied on

The following are known only bibliographically or from reposted mirrors; this report has not obtained a reliable full text or has not used them in the analysis, and lists them only for verification in later versions:

The Adolescence of Technology (Amodei's long essay of six months earlier, which footnote 5 cites for the detailed biological-risk argument in its Section 2)—full text not obtained; all references marked as second-hand.

Simon 2025 and Tehari 2025 (the original sources of two claims about personal relations cited in Durand Folco's paper)—not verified; the paper's claims that "Thiel funded Yarvin and made him his protégé" and that "Vance acknowledged Yarvin's major influence" are both marked as taken from these unverified sources.

Srinivasan, Balaji. The Network State (2022)—mentioned both in the e/acc footnote and in Durand Folco's paper; the book has not been obtained, and no inference is drawn about any direct textual link between the "network state" idea, Yarvin and e/acc.

Notes

Numbered in the Chicago notes–bibliography system: the first citation of a source gives full bibliographic details, later citations the author's surname, a short title, and page or line reference. The evidence grade (primary full text / third-party copy / single unconfirmed source) is given with the first citation.

1. Marc Andreessen, "The Techno-Optimist Manifesto," Andreessen Horowitz, October 16, 2023, https://a16z.com/the-techno-optimist-manifesto/, section 5. Primary full text (the author of this report obtained and read all 565 lines from the original place of publication). ↩
2. e/acc (Beff Jezos and bayes), "Notes on e/acc Principles and Tenets," beff.substack.com, July 10, 2022, https://beff.substack.com/p/notes-on-eacc-principles-and-tenets; Andreessen, "Techno-Optimist Manifesto," first three entries of the patron saints. Primary full text; the identification of @BasedBeffJezos as Guillaume Verdon is a single unconfirmed source (not independently verified by this report). ↩
3. Andreessen, "Techno-Optimist Manifesto," patron saints, entry 43, counted item by item. ↩
4. Ibid., entry 31. ↩
5. Ibid., section 5 (the techno-capital machine) and the first three entries of the list. ↩
6. Jaikumar Ramaswamy (Andreessen Horowitz) to Sen. Scott Wiener, re: SB 1047, August 2, 2024, second paragraph. Third-party copy (an electronically signed copy hosted on the Engine Advocacy website; see note 93). ↩
7. Andreessen, "Techno-Optimist Manifesto," the note preceding the patron saints. ↩
8. Ibid., the three epigraphs. ↩
9. Ibid., section 2, "Truth". ↩
10. Ibid., section 1, "Lies". ↩
11. Ibid., section 3. ↩
12. Ibid., section 4. ↩
13. Ibid., section 5. ↩
14. Ibid., section 6. ↩
15. Ibid., section 9. ↩
16. Ibid., section 12. ↩
17. Ibid., section 13, "The Enemy". ↩
18. e/acc (signed as in the original; see Bibliography), "Effective Accelerationism — e/acc," effectiveaccelerationism.substack.com, October 31, 2022 (a verbatim repost of the founding post; the original posting date is not shown on the page, and from the timestamp of a tweet embedded in the post it is about June 1, 2022), https://effectiveaccelerationism.substack.com/p/repost-effective-accelerationism. Primary full text; that it is a "verbatim repost" rests on a single source (the original founding post not obtained). ↩
19. e/acc, "Notes on e/acc Principles and Tenets." ↩
20. Ibid., footnote to the tenth step. ↩
21. e/acc, "Notes on e/acc Principles and Tenets" (text one); e/acc, "What the F* Is e/acc," effectiveaccelerationism.substack.com, December 26, 2022, https://effectiveaccelerationism.substack.com/p/what-the-f-is-eacc (text four). Both primary full text. ↩
22. Alex Williams and Nick Srnicek, "#Accelerate Manifesto for an Accelerationist Politics," Critical Legal Thinking, May 14, 2013, https://criticallegalthinking.com/2013/05/14/accelerate-manifesto-for-an-accelerationist-politics/, 02.5. Primary full text. ↩
23. Ibid., 03.1, 03.4, 03.12, 03.13, 03.15. ↩
24. Ibid., 02.2. ↩
25. Ibid., 02.2, read against Nick Land, "Meltdown," Abstract Culture, Swarm 1, Cybernetic Culture Research Unit (ccru.net), n.d., http://www.ccru.net/swarm1/1_melt.htm (textual comparison from note 33 on). ↩
26. Williams and Srnicek, "#Accelerate Manifesto," 01.6, 03.24. ↩
27. Ibid., whole text; Benjamin Noys, "Accelerationism," No Useless Leniency (blog), October 20, 2008, http://leniency.blogspot.com/2008/10/accelerationism.html—no verifiable material of transmission between the two texts. ↩
28. Noys, "Accelerationism" (2008). Third-party copy (the quotations are in hand but have not been checked word for word against the complete original). ↩
29. Benjamin Noys, interview, "Crash and Burn: Debating Accelerationism," 3:AM Magazine, November 4, 2014, https://www.3ammagazine.com/3am/crash-and-burn-debating-accelerationism/. Third-party copy. ↩
30. Ibid.; Roger Zelazny, Lord of Light (1967), cited from Noys's own account in the interview; this report has not obtained Zelazny's book to check. ↩
31. Noys, "Accelerationism"; Williams and Srnicek, "#Accelerate Manifesto," 02.5. ↩
32. Land, "Meltdown"; publications list on the author's biography page at the same site (Abstract Culture, First Swarm, Winter 1997). ↩
33. Land, "Meltdown," opening passage. ↩
34. Ibid., the section on capital's non-instrumentality and teleonomy. ↩
35. Ibid., "The Turing Test" section. ↩
36. Ibid., the section on the source and scope of the word "accelerate"; a full-text search for the word "accelerationism" returns zero. ↩
37. Ibid., from the section on man's place in capital's self-running process to the end; the summarising word "hyperstition" is this report's, not a word in the text. ↩
38. Ibid., the second-person passage in the "Meltdown" section. ↩
39. Gilles Deleuze and Félix Guattari, Anti-Oedipus: Capitalism and Schizophrenia, trans. Robert Hurley, Mark Seem, and Helen R. Lane (Minneapolis: University of Minnesota Press, 1983), 238–239. Primary full text. ↩
40. Ibid., 238; distribution of note numbers within the range verified for this report (preceding note 83, following note 84); the complete apparatus of notes at the end of the book lies outside that range. ↩
41. Andreessen, "Techno-Optimist Manifesto," section 5; full-text search of Land, "Meltdown," finds no literal phrase "techno-capital machine". ↩
42. This report's search of e/acc's public texts: two marginal texts narrate neutrally, with neither engagement nor endorsement; the specific items are recorded in section 5 of the earlier full-length version. ↩
43. Williams and Srnicek, "#Accelerate Manifesto," 02.5; Noys, "Accelerationism." ↩
44. e/acc, "Notes on e/acc Principles and Tenets," steps one to ten. ↩
45. Andreessen, "Techno-Optimist Manifesto," word-by-word search of all 565 lines for the seven roots dissipat, adaptation, thermodynam, entropy, fluctuation, physics, physicist: all zero; "England" occurs once, in an epigraph, meaning Norman-conquest England. ↩
46. Williams and Srnicek, "#Accelerate Manifesto," 03.7. ↩
47. e/acc, "Notes on e/acc Principles and Tenets" (text one); e/acc, "What the F* Is e/acc" (text four). ↩
48. Leopold Aschenbrenner, Situational Awareness: The Decade Ahead (self-published essay series, situational-awareness.ai, June 2024), https://situational-awareness.ai/wp-content/uploads/2024/06/situationalawareness.pdf, introduction and Parts I and II. Primary full text (the author of this report read the PDF directly, about 7,106 lines of plain text in full). ↩
49. Ibid., Part IIId. ↩
50. Ibid., Part IIIb and Part IV. ↩
51. Ibid., Part IIId. ↩
52. A summarising phrase used by this report to compare the three texts, not a word common to all three; see e/acc, "Notes on e/acc Principles and Tenets"; Vitalik Buterin, "My Techno-Optimism," vitalik.eth.limo, November 27, 2023, https://vitalik.eth.limo/general/2023/11/27/techno_optimism.html; Aschenbrenner, Situational Awareness, Part IV. ↩
53. e/acc, "Notes on e/acc Principles and Tenets," the five criticisms of top-down control. ↩
54. e/acc, "e/acc Is Leaderless," effectiveaccelerationism.substack.com, October 30, 2022, https://effectiveaccelerationism.substack.com/p/eacc-is-leaderless (text three). ↩
55. Buterin, "My Techno-Optimism," lines 275–291. Primary full text (line numbers checkable). ↩
56. Vitalik Buterin, "d/acc: One Year Later," vitalik.eth.limo, January 5, 2025, https://vitalik.eth.limo/general/2025/01/05/dacc2.html, lines 495–566. Primary full text. ↩
57. Ibid., lines 557–566. ↩
58. Buterin, "My Techno-Optimism," lines 583–585; Buterin, "d/acc: One Year Later," lines 495–566—the tension between the two essays is recorded only; this report does not judge whether it overturns the proposal. ↩
59. Aschenbrenner, Situational Awareness, Part V. ↩
60. Ibid., Part V, original wording: "That's some underwhelming accelerationism in my book." The full text was re-extracted with pdftotext -layout to check; the sentence lies in the body text corresponding to PDF page 158. ↩
61. Buterin, "My Techno-Optimism," compatibility list at lines 1310–1346. ↩
62. Combined result of searching and reading the four founding e/acc texts, Buterin's two essays and the full Aschenbrenner series; the state of the material in the three directions (e/acc on d/acc, e/acc on Aschenbrenner, Aschenbrenner on d/acc) is given in the text. ↩
63. e/acc, "Notes on e/acc Principles and Tenets," footnote to the ninth step. Primary full text. ↩
64. The reading notes on the five neoreactionary materials are still a working draft, not fully verified, and serve only as a guide; all citations in the text go back to the primary materials for checking. ↩
65. Curtis Yarvin (Mencius Moldbug), "A Formalist Manifesto," Unqualified Reservations, April 24, 2007 (archived repost, unqualified-reservations.org), full text obtained by this report (634 lines). Primary full text. ↩
66. Ibid., the passages on "corporate serfs" and "reformalization", about lines 452–561. The term "neo-cameralism" and the slogan "No voice, free exit" do not occur in this 2007 text; see note 75 for their source. ↩
67. Ibid., the closing passage on city-states. ↩
68. Peter Thiel, "The Education of a Libertarian," Cato Unbound, April 13, 2009, https://www.cato-unbound.org/2009/04/13/peter-thiel/the-education-of-a-libertarian/, text and the editor's note on the same page. Primary full text (retrieved from the original place of publication). ↩
69. Ibid., the passage on the three technological frontiers. ↩
70. Ibid.; read against Land, "Meltdown," the section on capital's non-instrumentality and teleonomy. ↩
71. Thiel, "The Education of a Libertarian"—the only Thiel text in this report's collection (2009); the assertion and its condition of refutation are both limited to it. ↩
72. Nick Land, "The Dark Enlightenment," published online c. 2012–2013 (this report obtained a mirror repost on an anthology host; the original platform could not be independently verified), Part 1. Third-party copy. ↩
73. Ibid.—this report read Part 1 and Part 4f in full; Parts 4a to 4e (whose headings indicate they concern race) were not read, and no judgment is made on the character of the whole. ↩
74. Ibid., Part 1, the passage on "Voice vs Exit". ↩
75. Ibid., Part 1, lines 75–87 (the first occurrence of "neo-cameralism"; the quotation "a state is a business which owns a country"; "No voice, free exit" at the end of the passage; the definition of "the Cathedral"). Both "neo-cameralism" and "No voice, free exit" come from Land's paraphrase and quotation of Yarvin in this essay, not from the 2007 "Formalist Manifesto" itself, where neither occurs (confirmed by grep, zero hits). ↩
76. Ibid., Part 4f. ↩
77. Land, "Meltdown," read against Land, "The Dark Enlightenment," Part 1 and Part 4f. ↩
78. Stan Higgins, "This A16z Startup is Building a 'Virtual Galaxy' for Bitcoin Nodes," CoinDesk, June 29, 2016, https://www.coindesk.com/markets/2016/06/29/this-a16z-startup-is-building-a-virtual-galaxy-for-bitcoin-nodes/ (retrieved from the original); Marc Hochstein, "Urbit Is Web3, Weird and Wonderful and I Don't Care Who Made It," CoinDesk, September 24, 2022 (obtained via the Yahoo Finance mirror, https://finance.yahoo.com/news/urbit-web3-weird-wonderful-don-141500880.html). Third-party copy (no official announcement corroborates specific investment terms). ↩
79. Peter Jamison and Elizabeth Dwoskin, "Curtis Yarvin helped inspire DOGE. Now he scorns it.," The Washington Post, May 8, 2025 (obtained from the syndicated reprint at The Spokesman-Review, https://www.spokesman.com/stories/2025/may/08/curtis-yarvin-helped-inspire-doge-now-he-scorns-it/, with byline and originating outlet intact). Third-party copy (a full reprint on a reliable secondary host; the email content, itself cited from BuzzFeed News, is a further remove, and this report has not verified the original email). ↩
80. Ibid., the passage on Andreessen's investment relation to Tlon, cited from PitchBook. ↩
81. Ibid., the passages on DOGE, "the most crisp articulation", and the chimpanzee orchestra. ↩
82. Ibid., the passage quoting Vance's 2021 podcast. ↩
83. Curtis Yarvin, "The pleasure of error," Gray Mirror (Substack), January 29, 2025, https://graymirror.substack.com/p/the-pleasure-of-error. Primary full text (the author's own signed platform, original full text). ↩
84. Jean-François Durand Folco, "Les racines idéologiques du fascisme technologique" [The ideological roots of technological fascism], Revue Possibles 50, no. 1 (2026): 66–77, DOI 10.62212/revuepossibles.v50i1.959, the author's statement of method. "Primary full text" here means "the paper does say this", not that its characterising claims are reliable—the author describes his method as "interpretive and critical", and this report does not adopt his characterisations, recording only the cited factual layer and marking it as taken at second hand. ↩
85. Ibid., the Thiel/Vance/Yarvin passage, cited from Simon 2025 and Tehari 2025 (neither original verified by this report). ↩
86. Ibid., the Andreessen passage; read against Andreessen, "Techno-Optimist Manifesto," text and patron saints counted item by item—Yarvin is not among the fifty-six, and neither "Yarvin" nor "Moldbug" occurs anywhere in the text; this report therefore does not adopt the claim. ↩
87. Durand Folco, "Les racines idéologiques," the Starbase passage. ↩
88. Ibid., the "butterfly revolution" and DOGE passages. ↩
89. e/acc, "Notes on e/acc Principles and Tenets," the two-column comparison table. ↩
90. Buterin, "My Techno-Optimism," lines 706–712. ↩
91. Andreessen, "Techno-Optimist Manifesto," section 13, "The Enemy," checked word for word. ↩
92. Ramaswamy to Wiener, August 2, 2024, first paragraph; the summary of SB 1047's content follows the letter's own account, and this report does not evaluate the bill itself. ↩
93. Ibid.; hosted by Engine Advocacy (engine.is), a copy bearing a Docusign electronic signature, Envelope ID 648C7987-232C-4868-B044-326844E0E98B, https://static1.squarespace.com/static/571681753c44d835a440c8b5/t/66b4f63b51a4a86d4cb58232/1723135547684/a16z+aug+2nd+letter.pdf. Third-party copy (hosted by a third party, not retrieved from the original place of publication). ↩
94. Ibid., sections A to G and the three opening legal arguments. ↩
95. Ibid., word-by-word search of the 492 extracted lines for accelerat, techno-capital, e/acc, philosoph: all zero. ↩
96. "Open Weights and American AI Leadership," open letter, full text and signatory list, hosted at microsoft.com/en-us/corporate-responsibility/topics/open-weight/, July 24, 2026. Primary full text (obtained by this report from the original place of publication). The date appears in the archived copy as "July 24, 2026". ↩
97. Ibid. ↩
98. Ibid. ↩
99. Dario Amodei, "Our position on open-weights models," Anthropic, July 27, 2026, https://www.anthropic.com/news/position-open-weights-models. Primary full text (obtained from the original place of publication), lines 16, 17, 19–24. ↩
100. Ibid. ↩
101. Ibid., footnote 5 (archived copy). ↩
102. Word-by-word search of the two primary files carried out independently by the author of this report (August 19, 2026) with the command-line tool grep, using the same search terms as for the SB 1047 letter: accelerat, techno-capital, e/acc, philosoph; zero hits in both files, by the same method and with the same result as note 95. ↩