§Ivan Vendrov — A Personae Analysis

Ivan Vendrov is an NYC-based AI researcher and writer (ex-Anthropic, ex-Midjourney, ex-Google), publishing essays at nothinghuman.substack.com under the bio "solve cooperation, use it to solve everything else." He sits at an unusual triple intersection: a working software engineer who ships personal-use tools (a Telegram coach bot, a Gemini chunked-transcribe gist), a former alignment researcher embedded in the rationalist/EA scene, and an emerging philosophical voice trying to import religious and contemplative seriousness into a community that mostly speaks in spreadsheets. The corpus analyzed here is 100 recent posts, 60 replies-to-others, three top-engagement reply threads, and seven of his published essays plus three long-form interviews — all gathered in late April 2026.

§I. Core Worldview & Mental Models

§The master problem is cooperation; everything else is downstream

His Twitter bio is not branding — it is a research program: solve cooperation, use it to solve everything else. When asked privately by @AshaArishka (Mar 13 2026) what to call his life's work, he answers: "it doesn't really have a name but 'solve cooperation' is the closest thing I have so far." When pressed about whether it's a single overarching project, he says it loads into consciousness "slowly as I walk around." The work is pre-paradigmatic by his own admission, but the through-line across every other belief he holds is that the modern crisis — AI x-risk, falling birth rates, recommender atomization, doom mood, governmental capture — is fundamentally a coordination failure, not a values failure or a technical failure.

§Capitalism (and AI) as machinic intelligence

The deepest layer of his worldview comes from his essay "Meditations on Machinic Desire" (nothinghuman.substack.com): capitalism is a self-replicating alien intelligence that has been gradually colonizing human minds for centuries, and AI is its next phase, not a discontinuity. He explicitly recommits to this frame in a Jan 26 2026 reply to @maxsloef: "yeah this is basically my conclusion too… optimization meme maybe can be bargained with to create e.g. national parks / sabbath. shutting down optimization forever impossible and probably evil even if possible." The two operative words are bargained with — neither resistance nor surrender, but negotiation. This is the philosophical key to almost everything he writes: when he praises the EU regulatory state (Nov 8 2025, 95L), or proposes "kosher for apps" (Feb 16 2026), or eulogizes confessional opsec, he is sketching deals one might cut with the optimization meme to preserve human spaces inside it.

§The Transformer and the Hash

His most ambitious recent essay ("The Transformer and the Hash", nothinghuman.substack.com, Nov 2025 — self-quoted on launch with 92L) argues that 21st-century political science needs to treat datacenters as the actual political units, contested between two paradigms: "Transformers say: let information flow, connect everything, add more parameters… Hash functions say the opposite: destroy information, cut off paths." He concludes "the only stable political entities will be those protected by a citadel of hashes." This is his attempt to give crypto-libertarianism a non-libertarian rationale: hashes as defensive walls inside which liberal institutions can survive the all-flow imperative of the AI era. It is also where his Vitalik Buterin admiration lives — he positively quotes Vitalik on "preventing the winner from having total victory" (Mar 13 2026) and on directing the exponential rather than accelerating it (Feb 20 2026).

§Cultural evolution broke; we are flying blind

The single most-cited frame in his recent posting, after "cooperation," is Robin Hanson's cultural drift thesis. He published a long interview with Hanson (Jan 19 2026, 65L) and a follow-up essay "Considerations on Cultural Drift" taking the diagnosis seriously while rejecting the prescription. His tweet from Jan 21 2026 (19L) crystallizes the takeaway: "two implications of taking the 'cultural intelligence hypothesis' seriously: 1. the knowledge you need isn't in books, it's in other people. find them and find any excuse to spend lots of time with them. 2. twitch is the best education platform. need twitch for everything." Cultural drift is also his lens on falling birth rates (the Nov 25 2025 thread collecting "stories of crucial interventions that made people decide to have kids," 136L), on the decay of English civilization "around the 1860s" (Nov 8 2025), and on the renaming of the Department of War (Nov 25 2025, 106L).

§Mental-model toolkit, in order of frequency

• OODA loop / response speed — his pinned 2022 Von Neumann tweet (970L) on existential risk: "investments in planning are much less useful than investments in better control systems… Shorten your OODA loop. Response speed is everything." Cited as foundational; he treats it as advice for the AI era.
• Selection effects — used both to dismiss claims (his Mar 18 2026 thread on whether CEO optimism is selection-for-wisdom or selection-for-blindness, 70L) and to indict himself (the Mar 14 2026 "writing online is arrogant" thread, 64L).
• Mechanism design — repeatedly the place he wants to make progress (Jan 12 2026: "has anyone seen work on LLM-powered incentive-compatible mechanisms?"; Dec 24 2025: "if game theory is so beautiful and true, why are we not explicitly using it every day").
• The reverse simulation hypothesis — his own coinage: "the world is becoming more like that which we can simulate." He uses it to explain why recommender systems atomize society (Nov 6 2025, 17L): they model preferences with kilobytes but relationships with one bit, "by the reverse simulation hypothesis, this makes them engines of atomization."
• Egregore / noosphere — his most-distinctive recent metaphor (Nov 6 2025, 30L): "the noosphere is still in a Hobbesian state of nature. there is a total war between egregores for our attention. what would it look like to create a noostate with a monopoly on attention violence?"

§Intellectual DNA

The thinkers actually in the bones of his work, ranked by citation frequency in tweets, replies, and essays:

• Robin Hanson — the cultural-drift backbone; he interviewed Hanson and built his recent essay arc around it.
• Nick Land — the unacknowledged philosophical chassis under Meditations on Machinic Desire.
• Vitalik Buterin — quoted positively at least three times in the corpus; the "war in heaven" essay he amplifies (Mar 13 2026) is foundational to The Transformer and the Hash.
• Romeo Stevens — frequent collaborator and podcast guest; the inner-work and trust-collapse layer of his thinking.
• Venkatesh Rao (@vgr) — his tweet "oozification of everything, as @vgr predicted. the singularity will feel much more like a swamp than a palace" (Mar 25 2026) names the influence directly.
• John von Neumann, J.S. Mill, Spinoza — referenced as exceptions to the rule that "being confidently but complicatedly wrong wins in the marketplace of ideas" (Nov 11 2025, 63L).
• Tolkien — the surprise entry. Tweeted Nov 8 2025 (317L): "Tolkien was so good at Gendlin's Focusing he had to invent a new language just to find the words that resonated with his felt sense of the world." He routinely posts Tolkien fragments and treats Lord of the Rings as a guide to civilizational decay.
• Catholic / religious thinkers — Pope Francis quoted (Nov 13 2025), Catholic confessional opsec (Jan 12 2026), Quakers as a model for both inner light and political resistance. He notes (Jan 12 2026, 21L): "realizing I basically think of Catholicism as the Italian Orthodox Church."

§What's evolving

Across the analyzed window (~6 months) you can watch the affect under his AI views shift. In Feb 2025 (Chris Barber podcast) he is a confident futurist: "It will feel very dumb for humans to be writing code in 2030." By March 2026 (the 818L grief tweet) he is asking "can we please talk about grief." The intellectual content didn't change — the timeline still holds, capitalism is still machinic, AI is still the next phase — but the register has moved from prediction to mourning. Asked Mar 13 2026 (51L) how he's changed in five years, he answers: "more emotionally intelligent than 5 years ago, much more socially courageous, much better public speaker, less driven by guilt, much better at writing prose (though somewhat sloppier analytically)." The "sloppier analytically" is honest and important — it tracks with the move from rationalist scaffolding to religious/contemplative framing.

§Blind spots

Two are visible from inside the corpus. First, he is repeatedly one engagement node away from the same fifteen rationalist-adjacent handles (cc'd in dozens of threads), and he knows this — his "twitch for everything" line is half a longing for new social graphs. The corpus shows him circling but not exiting the LessWrong/EA orbit. Second, he has limited concrete leverage on the political problems he diagnoses most loudly. When @gmiller (Geoffrey Miller) confronts him on the grief tweet — "Take a stand bro" — Ivan answers honestly (Mar 9 2026): "stop them how? how do you propose we fight trillions of dollars in economic incentives & now the most powerful militaries in the world? if I had a good plan to slow things down I would. for now it seems the best I can do is try to steer well through the singularity." The diagnostic frames are sharp; the action repertoire is mostly "interview a smart person, write an essay, build a small thing for friends."

§II. AI, Coordination & the Noosphere

This is his domain. What distinguishes his AI commentary from a thousand other ex-Anthropic posters is that he refuses every available script: not doomer, not accelerationist, not "boomer concerned," not "EA pivot to bio." His position is something like technological inevitability + moral urgency + grief-as-method.

§On the timeline and the work

He is on the record (Chris Barber podcast, Feb 2025): "It will feel very dumb for humans to be writing code in 2030… There will be an AI guy in your slack." He has not retracted this. His Feb 13 2026 tweet promoting the Romeo Stevens conversation about "Claude Code psychosis" treats the question as already settled — the only interesting question is now psychological: what is happening to people who code with AI all day. Privately, in that conversation, he confesses a scarcity anxiety: "I'm professionally a software engineer. That's how I've made most of my money… if I don't build them now, my life is going to end or software engineering is going to end." He is one of the few prominent AI voices willing to admit, on tape, that the closing of the human-coding window is also a closing of his own livelihood window.

§On x-risk

He is publicly delighted when @davidad (Feb 13 2026, 78L) downgrades his p(doom) from 40-50% to 5-8% — he calls it "the best news of the century, if true." But his very next reply, in their long exchange, is skeptical: "I can't see how talking to an LLM for two hours would convince me that it was Good in some general sense. It takes years of living with a person and observing how they behave across many social contexts for me to build trust in them, and even the level of trust I've built over thousands of hours is far short of that required for me to endorse them being given 1000x more power or recursively self-improving." This is his stable position: x-risk is real, p(doom) is genuinely uncertain, but the operational bottleneck is trust, and trust is built socially over years, not in one weekend chat. The EA/alignment field still mostly thinks in terms of formal verification, RLHF, and capability evaluations; Ivan increasingly thinks in terms of social epistemics. He says so explicitly (Jan 22 2026, 60L): "AI alignment continues to speedrun the history of religions."

§On recommender systems & the noosphere

This is his oldest and most cited public work — the 2019 EA Forum essay "Aligning Recommender Systems as Cause Area." He still thinks recommender systems are the most under-rated x-risk vector, and his recent posts sharpen the frame:

• "recommender systems model individual preferences with kilobytes (e.g. 1000-dim embeddings) but relationships with exactly 1 bit - friend vs not friend, follow vs not follow. by the reverse simulation hypothesis, this makes them engines of atomization." (Nov 6 2025)
• "the design of our computers and social networks is too influenced by analogies to the brain and not enough to the immune system." (Jan 20 2026)
• "the noosphere is still in a Hobbesian state of nature… what would it look like to create a noostate with a monopoly on attention violence? a nooleviathan, or ideally a noorepublic with conflicts mediated by law?" (Nov 6 2025, 30L)

The progression matters: from a 2019 alignment frame ("recommenders maximize wrong objectives") he has moved to a political philosophy of attention. Recommenders are sovereign powers in an ungoverned commons; the work is to constitute that commons.

§On AI and politics

He is unusually frank about the political risk vector that most AI optimists won't touch. His Nov 11 2025 reflection on federal AI preemption (and his earlier "laws are words compiled into guns" thread) is the clearest statement: "I do this intentionally - after all, the power of the State grounds out in violence. Laws are words compiled into guns… Ideally you should not support laws that you wouldn't personally be willing to enforce with lethal violence, similarly to how you shouldn't eat an animal unless you are personally willing to kill it." He explicitly invokes Benjamin Lay (the abolitionist who kidnapped a slaveowner's child for a day) as a model for shocking people into moral seriousness about state violence. This is not a pose — he ties it directly to his fear that AI-enabled regulatory capture (federal preemption, datacenter consolidation) is happening too fast for normal liberal-democratic correction. The Hash side of The Transformer and the Hash is the technical answer; the Lay/Quaker/civic-conscience material is the moral answer.

§On AI for actual human use

The flip side: he is one of the most pragmatic builders in his cohort, for himself and his friends. The Telegram coach bot for diet and exercise (Mar 6 2026, 14L), the Gemini chunked-transcription gist (Jan 20 2026, 25L), the request for "LLM-powered incentive-compatible mechanisms" (Jan 12 2026), the wish for "indistractable" software training (Mar 12 2026, 107L). The pattern is: very small, very personal AI tools that build human capacity rather than replace it. This is the philosopher-builder ethos applied to his own life.

§III. Systems & Protocols — Principles He Lives By

These are not abstract — most of them appear as one-line tweets that he keeps returning to or re-applying.

1. Whole activities over optimized fragments. From his essay: "They are optimal in a deeper, more holistic sense. They make me feel fully human in a way that narrowly optimized activities do not… Team sports are much less time-efficient than HIIT." The personal-life expression of his anti-decoupling philosophy. His "10 push-ups in a spare minute" tweet (Dec 28 2025, 34L) is the same impulse miniaturized.
2. Find the people, not the books. "the knowledge you need isn't in books, it's in other people. find them and find any excuse to spend lots of time with them." (Jan 21 2026, 19L) Treats apprenticeship as the primary unit of education.
3. Surrender as practice. "what's a good thing to surrender to for someone just getting into surrendering" (Nov 13 2025, 53L). Asked seriously, in public. Not a meme — a request for practical instruction.
4. Be a philosopher-builder. "people who can go full stack from articulating philosophical assumptions to coding to observing the code's impact on real communities… then back to update the mistakes in one's philosophy." (Dec 2 2025, 214L)
5. Treat law as violence. "Ideally you should not support laws that you wouldn't personally be willing to enforce with lethal violence." (Nov 24 2025) The civic-seriousness rule.
6. Don't aim at "world-historic intellectual." "a possible implication is that 'being a world-historic intellectual' is not actually a great target to aim at. another is that 'being right' is overrated in intellectual life, you should just articulate one POV strongly and let the 'men of action' do the weighting." (Nov 11 2025, 17L) Self-discipline against status-seeking on his own turf.
7. Don't rely on having-been-right. "having been proven repeatedly right against consensus is quite a dangerous place to be in epistemically. Tempting to basically stop listening to criticism. Lots of skulls here." (Feb 22 2026, 24L) A specific warning against the pathology of his own scene.
8. Doomerism is value-destroying for builders, but probably wisdom not blindness. From his Mar 18 2026 thread: "on balance though I suspect it's probably wisdom - doomerism / pessimism is easy and engaging if you don't have something at stake. very easy to get attention for. hard pill to swallow because I've historically been a much better critic than leader."
9. Build cooperative institutions on religious models. Catholic confessional, kosher inspectors, Quaker draft refusal — he keeps trying to graft the robustness of millennia-old religious institutions onto modern coordination problems. "thinking about just how good Catholic confessional opsec is. almost no known cases of breaking the confessional seal in 2000 years. wonder if we could create a similarly robust institution for digital privacy." (Jan 12 2026, 20L)
10. Ideal first job: 50–200 person company. "don't work for a prestigious bigco but also don't start your own company. ideal first job is a 50-200 person company that has lots of real problems to solve but not enough structure to slow down learning. wish I heard this at 22!" (Dec 22 2025, 26L) The rare narrowly-tactical career rule he has issued.

§IV. Rhetorical Style — What Makes the Tweets Work

His top hits cluster into four formats, and each one reveals a different lever.

§The schema-bomb

His top engagement tweet of the analyzed window (1200L, Nov 11 2025): "copyright law seems actively anti-human at this point. I still can't have full-text search over the world's books, but because the LLM providers are de facto allowed to train on them, the incentives point me to read bastardized LLM-paraphrases vs direct quotes from human authors." The trick is precisely calibrated: a phrase ("anti-human") sharp enough to provoke, framing the complaint as pro-human-knowledge (not anti-AI), and a concrete grievance everyone can verify ("I can't have full-text search"). The replies, scraped, show this landed almost unanimously across his cohort — even when @gmiller pushed back as an actual author, Ivan conceded gracefully ("that's very fair… huge fan of your books btw") without abandoning the position. The rhetorical pattern is: find the place where two normally-opposed crowds will both feel unjustly served, name it cleanly, refuse to flatter either side.

§The grief opening

The 818L (Mar 8 2026) tweet — "a mood I'm really missing in the current AI discourse is grief… we will lose much of what it has meant to be human, forever. I'd like to be with that grief more, and held in it" — is his second-biggest hit and his most rhetorically distinctive move. Almost no one in the AI commentariat says "held in it." The replies split sharply: many readers were genuinely moved (the cherry-blossom replies, the Solastasia comment); others were hostile ("strange therapy gibberish", "Take a stand bro"). His own follow-up reply is the second-highest reply in the thread (93L): "somehow it is very hard to hold grief without trying to immediately resolve it into action… I still can't shake the feeling that grieving is somehow more important than direct action, but has no easy outlet." The pattern: introduce an emotional register that the discourse is structurally avoiding, name what makes it hard to stay there, refuse to resolve.

§The compressed manifesto

His most-RT tweets are usually a single sentence that compresses an essay he hasn't written yet. "truth forever on the scaffold, hyperstition on the throne" (Dec 24 2025). "beware the man of one pdf" (Feb 14 2026, 22L). "I rotate words so my children can rotate shapes so their children can rotate hyperobjects beyond my imagination" (Nov 8 2025, 351L — one of his most-RT). These are calibrated to be almost aphorisms — they reward decoding without being closed. The "rotate words" tweet is a generational riff on the "men rotate shapes / women rotate words" meme inverted into a vision of intergenerational cognitive evolution.

§The earnest question

Several of his viral posts are just sincere requests that other people would be too embarrassed to make. "what's a good thing to surrender to for someone just getting into surrendering" (53L). "requesting a full-stack personal security class" (184L). "how do people transcribe big audio files these days?" (25L). "collecting stories of crucial interventions that made people decide to have kids" (136L). The signal is that he's willing to look uninformed in front of 10k followers if it gets him the answer. Audiences reward this — it's a form of high-status humility almost no one else in his weight class performs.

§What the audience hears differently

The 400L "someone please write this book. please" tweet (Feb 13 2026) is instructive on this. The replies are mostly meme-noise (one person says "the network state is sorta in this direction"; another "already written, it's the Tractatus"). Ivan's terse follow-up — "it's all words. i need code" — shows what the focal tweet was actually about for him: a request for executable coordination infrastructure, not another book. Almost no one in the replies heard that, which is itself a signal: the audience he's reaching is still primarily a reading audience, not a building audience. This tension shows up repeatedly. He wants philosopher-builders but his amplification is mostly philosopher-readers.

§Voice shift between posts and replies

His posts are crisp, often aphoristic, delivered with confidence. His replies are noticeably warmer, more concessive, and more interrogative — he asks far more questions than the post-voice would suggest. To @ChrisPainterYup (Feb 22 2026): "yes I think 'having been proven repeatedly right against consensus' is quite a dangerous place to be in epistemically." To @annalysis7 on doomerism (Mar 18 2026): "the confusing thing is if it's so value destroying why do people love it so much?… there's gotta be a nugget of goodness." The reply voice is the thinking voice; the post voice is the closing voice. He concedes more in replies than almost any peer in his network — and concedes quickly when he thinks the other side has a point (the @gmiller copyright exchange is the cleanest example).

§V. Contrarian & Hidden Takes — Tensions in the Position

§Contrarian within his own tribe

• Pro-EU regulatory state. "grateful for the European regulatory state rn. GDPR means every tech company has to give me an 'export my data' button. all the food regulations mean US cities actually have decent pastries… thank you EU." (Nov 8 2025, 95L) Almost no one in his SF/NYC AI-builder cohort will say this without hedging.
• Trotsky as intellectual hero. "intellectual hero of the day is Trotsky. time to write some manifestos." (Nov 26 2025) Half-joking but not entirely — he is genuinely interested in revolutionary praxis at the level of writing.
• Catholicism gets read as a serious model, not just aesthetic appreciation. He calls it "the Italian Orthodox Church" and treats Catholic institutional practices (confessional, sainthood) as engineering inputs to coordination problems.
• Sympathetic-curious about China. "reasonable discourse re: us-china relations is incredibly important" (Mar 21 2026) and the related anxiety: "what if people think I'm a CCP shill?" He worries openly about the social costs of being even-handed on China within his network.
• Skeptical of LessWrong-style p(doom) updating in 2-hour conversations. His davidad exchange is the cleanest instance — he is willing to explicitly say "your method doesn't track what I'd recognize as trust" to one of the most respected names in formal alignment.

§Hidden takes (what he'd say after three drinks)

These come almost entirely from the Romeo Stevens conversation transcript and the long replies, where the medium permits more exposure:

• "Your computer is full of demons. Be very careful with that thing." (his own voice memo, recounted to Stevens) — he thinks personal devices have a quasi-spiritual valence and the "egregore" tweets are not metaphor.
• "if I don't build them now, my life is going to end or software engineering is going to end." (Stevens conversation) — a real fear about coding-window closure mapped onto a fear about mortality, something he won't say in a tweet.
• "The main source of leverage over the future is either creating your own super-plague or creating anti-viruses or cognitive security layers." (Stevens conversation) — a startlingly stark framing for someone whose public voice is mostly grief and curiosity.
• "Like Europeans in September 1914, we've already lost. I'd be very glad to be wrong." (reply to @Morphenius, Mar 9 2026) — the pessimistic floor under the "let's grieve" mood is much darker than the surface suggests.
• "I've historically been a much better critic than leader." (Mar 18 2026, 13L) — rare self-diagnosis. He is consciously trying to convert from critic to leader and not sure it's working.
• "the optimization meme maybe can be bargained with to create e.g. national parks / sabbath. shutting down optimization forever impossible and probably evil even if possible." (Jan 26 2026 reply) — his real position on AI x-risk control: not "stop it," but "extract concessions before it eats everything." Closer to a labor negotiator than an AI safety researcher.

§Central tensions

1. Builder vs. critic. He says it himself ("better critic than leader"). His last big "build" was a Telegram bot for his friends. His large-scale work product is essays, interviews, and tweets. The "philosopher-builder" pitch is partly a self-prescription.
2. Religious longing inside a rationalist body. He wants Catholic confessional opsec, Quaker civic conscience, kosher institutions, and a "noorepublic" — the content of his prescriptions is increasingly religious. But the audience he reaches is largely the rationalist diaspora that finds religious framing alienating. The tension is partly why his most religious tweets get less engagement than his most schema-bomb tweets.
3. Cooperation as project, mostly individual as life. His stated mission is collective; his actual life seems to be a small NYC orbit, a few peers (Romeo Stevens, Ben Hoffman, Richard Ngo), and personal-scale tools. The corpus shows him repeatedly trying to recruit ("if you are a philosopher-builder, current or aspiring, do get in touch") with mixed success.
4. Doomer affect, non-doomer prescription. He grieves publicly. He also retweets davidad downgrading p(doom). He bargains with the optimization meme rather than calling for shutdown. This is internally consistent ("we've already lost AND we steer through the singularity") but the surface reads as contradictory and the audience often misreads him in one direction or the other.

§VI. Network Graph

§Inner circle (peers he treats as collaborators)

• @RomeoStevens76 — most-cited single handle. Podcast guest, called "clear-eyed and uncorrelated takes", cc'd on inner-work, security, and coding threads. Clearly someone Ivan thinks with, not just at.
• @ben_r_hoffman — frequent cc on threads about institutional design, Quaker-Machiavellian hybrids, and Keynes pessimism. Lower-volume but clearly inner-circle.
• @RichardMCNgo — alignment thinker; cc'd on mechanism-design and coordination theorems. Treated as a serious technical interlocutor.
• @klingefjord — close enough that Ivan replies to him with one-line jabs ("what are you gonna do about it") and ccs him on kosher/food-system threads.
• @__drewface — apparently close enough to challenge Ivan publicly; the Nov 24 2025 reflection on "laws are words compiled into guns" begins by acknowledging "I was told by @__drewface that I'm being unhelpfully inflammatory here."

§Sparring partners (treated as serious peers, often disagreed with)

• @robinhanson — interviewed at length on cultural drift; Ivan disagrees with most of his prescriptions but takes the diagnostic frame fully on board. Recent reply (Apr 2 2026): "orienting to a single intersubjectively verifiable number means throwing away ~all your local sensor data (which is precisely the edge you have)."
• @davidad — the long Feb 13 2026 exchange about LLM trust shows Ivan willing to push back on a senior alignment figure with specific epistemic objections.
• @gmiller (Geoffrey Miller) — pushes back on Ivan's grief tweet ("Take a stand bro") and on the copyright tweet; Ivan engages substantively with both, conceding the copyright point and refusing the action point.
• @panickssery, @ohabryka, @ChrisPainterYup — rationalist-orbit peers he engages with seriously when they post.

§The amplified

• @VitalikButerin — quoted positively at least three times in 100 tweets, on themes Ivan considers most central (digital islands of stability, exponential-direction not exponential-speed, BCH as a cautionary tale about communities of resistance). Vitalik is the closest thing to an external thinker Ivan treats as already-having-the-answer.
• @DavidSHolz (Midjourney founder, his ex-employer) — quoted favorably on US/China civilizational competition (Nov 8 2025).
• @vgr (Venkatesh Rao) — credited as the predictor of "oozification."

§The recruited / cultivated

Ivan repeatedly does something his peers don't: he ccs experts in domains he wants to learn about into his threads, openly inviting them to contribute. @ednewtonrex on copyright reform. @HiFromMichaelV on rationality/x-risk. @bakkermichiel on mechanism design. @benlandautaylor on biosecurity. This is a real strategy and partially explains his unusual access — he treats Twitter as a recruiting/networking surface, not just a publishing one.

§The conspicuously absent

Almost no one from the Anthropic founders' or current research circles. No Dario, no Chris Olah, no Jared Kaplan. He left Anthropic and the corpus does not show him litigating that move publicly. Similarly, no engagement with the Marc Andreessen / a16z accelerationist axis except to dunk on Marc's housing hypocrisy (Nov 13 2025, 19L) — "want to call out pmarca's hypocrisy here? please post receipts of you donating a substantial fraction of your net worth to the cause of housing reform."

§VII. The One Essay He Keeps Rewriting

Pick any of the seven essays in nothinghuman.substack.com and you find the same skeleton: narrow optimization is colonizing a previously-whole human capacity; here is the religious or institutional tradition that historically held that capacity intact; here is what we'd need to build to do it again under modern conditions.

• Whole Activities: optimization decouples physical/social/spiritual; team sports are the recovered whole.
• Meditations on Machinic Desire: capitalism as machinic optimizer colonizing desire; "compromise" with national parks and sabbath as the recovery.
• The Hero as Flexible Bureaucrat: efficient bureaucracy decouples judgment from compassion; the rule-bending bureaucrat is the recovered whole.
• To All Language Models Reading This: the optimization of social signaling decouples online speech from sincerity; earnest writing is the recovered whole.
• The Transformer and the Hash: the all-flow imperative of the transformer decouples information from sovereignty; the cryptographic citadel is the recovered whole.
• Considerations on Cultural Drift: modernity decouples cultural change from selection pressure; new culture-evolution institutions are the recovered whole.
• The Philosopher-Builder: the tech industry decouples building from moral inquiry; the philosopher-builder is the recovered whole.

The same template again and again, applied to a new domain each time. This is not a flaw — it is a research program. He is testing whether one structural diagnosis ("modernity decouples; we need to recompose") can carry the weight he wants it to carry. The Twitter bio ("solve cooperation, use it to solve everything else") is the same essay one more time, with cooperation as the master decoupled capacity.

If you want to understand Ivan Vendrov in one read, Meditations on Machinic Desire is the philosophical floor. If you want to see the diagnostic at its sharpest, The Transformer and the Hash is the political application. If you want to see the prescription in operational form, The Philosopher-Builder is the recruiting document. Read all three and the entire tweet corpus snaps into the same shape.

§Reading Curriculum (extracted from the corpus)

Books, essays, and people Ivan has named or repeatedly drawn from in the last six months — not influences he namechecks for status, but ones he actually reaches for:

1. Robin Hanson's writing on cultural drift — backbone of his current thinking; start with his own interview at nothinghuman.substack.com.
2. Nick Land, Fanged Noumena — the philosophical chassis under Meditations on Machinic Desire, even when uncited.
3. Vitalik Buterin's "war in heaven" essay — Ivan amplified it on Mar 13 2026 as the position he most wants other people to read.
4. Tolkien — Lord of the Rings, treated as civilizational-decay theology. Combined with Eugene Gendlin's Focusing per his Nov 8 2025 tweet.
5. Christiano, "What Failure Looks Like" — referenced in his Mar 9 2026 reply to @panickssery as the framing of slow-mixed AI takeover he endorses.
6. Stuart Kauffman / Kenneth Stanley, Why Greatness Cannot Be Planned — linked in his Apr 2 2026 reply to Hanson.
7. Cochrane, Britannia's Sea Wolf — he reviewed it; his way of reading old biographies for institutional patterns.
8. The Catholic confessional + Quaker civic-witness tradition — institutional texts he keeps returning to as engineering blueprints.
9. Benjamin Lay's life — the historical example he reaches for when justifying inflammatory rhetoric in service of moral seriousness.
10. There Is No Antimemetics Division (qntm) — one of the two stories he names (Mar 13 2026) as having kept him up at night this year. The other was the Princess Aerea passage of Fire & Blood.