The art of being less wrong.
The LessWrong community founded by Eliezer Yudkowsky to teach people to think clearly — and, above all, to solve the problem of aligning superhuman AI before it kills everyone. Critics call it a movement that mistakes reason for a personality and apocalypse for an argument.
Rationalism is the internet subculture that grew out of Eliezer Yudkowsky's million-word blog series, “The Sequences,” and the forum LessWrong. It promises a toolkit — Bayesian probability, the study of cognitive bias, ruthless utilitarian calculation — for overcoming the flaws in human reasoning. In practice it became the staging ground for the AI-safety movement, for effective altruism, and for some of Silicon Valley's stranger offshoots: Roko's Basilisk, the neo-reactionary right, and a Bay Area murder case. Critics including Émile P. Torres and Timnit Gebru argue that beneath the language of logic sits a familiar shape — an apocalyptic sect with a prophet, a scripture, and a salvation narrative.
Rationalism begins with a flattering premise: human reasoning is a buggy inheritance from evolution, and it can be debugged. Our brains run on heuristics that misfire — we anchor, we rationalize, we confuse what is comforting with what is true. The rationalist project is to study these failure modes systematically and train them away, using probability theory as the gold standard of correct thought.
The movement crystallized around Eliezer Yudkowsky, a self-taught AI researcher who, beginning in 2006, wrote a sprawling sequence of blog posts on cognition, science, and ethics. Collected as The Sequences, they became the founding scripture of a community that gathered first on the blog Overcoming Bias, then on its own forum, LessWrong. The name is the whole ethos: you can never be right, only less wrong.
What makes rationalism more than a self-improvement hobby is where the reasoning is meant to lead. Yudkowsky's deeper conviction is that superintelligent AI is coming, that most ways of building it end in human extinction, and that the people who grasp this have a moral duty to work on “alignment” above all else. Clear thinking, in this frame, is not an end in itself — it is the only thing standing between humanity and a machine god that does not share our values.
This gave the community a center of gravity and a sense of mission. It also gave it institutions: the Machine Intelligence Research Institute (MIRI), the AI-safety wing of Silicon Valley, and — directly downstream — the effective altruism movement, with which rationalism remains deeply intertwined. Sam Altman has credited Yudkowsky with getting many people interested in AGI in the first place.
Critics argue that a community which names itself after rationality has built in the perfect defense: to disagree is, by definition, to be irrational. Émile P. Torres describes the movement not as a school of reason but as “pseudointellectual reductionism, utilitarianism, and worship of advanced technologies.” Scott Alexander, one of its most prominent writers, half-joked that the community's defining belief is that “Eliezer Yudkowsky is the rightful caliph.”
The cultural critique sharpened as the offshoots accumulated: Roko's Basilisk, a thought experiment so disturbing it was banned; the neo-reactionary politics that flourished in adjacent comment sections; and, in 2025, the Zizians — a splinter group linked by reporting to a string of killings. For its detractors, this is what happens when a movement teaches that any conclusion is acceptable if the math works out, and that the fate of the cosmos hangs on getting the math right.
The doctrines that recur across the movement's founders, theorists, and builders.
Beliefs are not true or false but more or less probable. The ideal rationalist updates those probabilities by Bayes' theorem as evidence arrives — reasoning is bookkeeping done correctly.
Human cognition is riddled with systematic biases. Knowing the catalog of these failure modes — and catching yourself committing them — is the core discipline of the movement.
Building superhuman AI that shares human values is treated as the single most important task in history. Get it wrong and everyone dies; get it right and the future is unimaginably good.
When intuition and arithmetic conflict, trust the arithmetic. Moral feelings are unreliable; expected-value calculation, however cold the result, is the higher authority.
Follow an argument to its logical end, no matter how strange or uncomfortable the destination. Refusing to do so is dismissed as cowardice or “normalcy bias.”
Our models are always approximations of reality. Confusing the model for the world is the root error rationalists train themselves to notice and correct.
The specialized terms you need to read the movement in its own words.
Founders, theorists, builders, and financiers of the project.
Founded LessWrong and MIRI; wrote The Sequences and HPMOR; co-authored “If Anyone Builds It, Everyone Dies” (2025). Critics describe the community around him as cult-like; Scott Alexander joked he is its “rightful caliph.”
Wikipedia →
Pen name of the psychiatrist behind Slate Star Codex and Astral Codex Ten — the movement's most-read writer and its bridge to the tech elite. Deleted his blog in 2020 over a New York Times piece preparing to publish his legal name.
Wikipedia →
Economist who co-founded Overcoming Bias with Yudkowsky. Critics point to his posts musing about “redistribution” of sex and a piece on “gentle silent rape” as examples of the movement's willingness to follow utilitarian logic to disturbing ends.
Wikipedia →
Writing as “Mencius Moldbug,” Yarvin founded the neo-reactionary (NRx) movement. Critics note his anti-democratic ideas circulated and found an audience on and around rationalist platforms.
Wikipedia →
Turing Award winner and Meta's chief AI scientist; a prominent skeptic of rationalist AI-doom claims. A frequent foil for Yudkowsky in public sparring over whether advanced AI poses an extinction risk.
Wikipedia →The essays, declarations, and books that built the doctrine.
The collected Sequences — the movement's founding canon on probability, bias, and clear thinking. Free to read in full.
Read ↗The community forum that is the movement itself — where The Sequences live and where the AI-safety canon was written.
Read ↗The most widely read rationalist blog and the movement's main on-ramp for Silicon Valley readers.
Read ↗A ~660,000-word fanfiction that turns Hogwarts into a rationalist parable — and the movement's most effective recruiting tool.
Read ↗The original LessWrong post posing the dust-specks dilemma — a canonical example of “shut up and multiply.”
Read ↗MIRI's flagship argument that building superhuman AI will, with current methods, kill every person on Earth.
Read ↗The whole rationalist community… believes that Eliezer Yudkowsky is the rightful caliph.— Scott Alexander (paraphrased)
Shut up and multiply.— Eliezer Yudkowsky, The Sequences
If anyone builds it, everyone dies.— Yudkowsky & Soares, 2025
[Rationalism is] pseudointellectual reductionism, utilitarianism, and worship of advanced technologies.— Émile P. Torres
Rationalism markets itself as the cure for human irrationality. Its critics — among them Émile P. Torres, Timnit Gebru, and a wave of journalists covering the movement's offshoots — argue that it is better understood as a belief system: one with a prophet, a scripture, a salvation narrative, and a striking record of producing exactly the irrational behavior it claims to inoculate against.
Reporting and former members describe communal “group houses,” intense internal dating networks, authority deference to Yudkowsky, and rituals like “doom circles.” Critics argue the apocalyptic stakes and in-group jargon produce textbook high-control dynamics.
Timnit Gebru and other AI-ethics researchers argue the movement's fixation on speculative extinction (“AI safety”) is used to wave away the documented harms of AI now — bias, surveillance, labor exploitation, environmental cost — as a lesser distraction.
Critics note that rationalist and adjacent platforms incubated Curtis Yarvin's neo-reactionary politics and recurring eugenics-flavored arguments — the “forbidden” conclusions that “taking ideas seriously” licensed people to entertain.
The movement spawned Roko's Basilisk, a thought experiment so distressing it was banned — and, per Rolling Stone and WIRED, the Zizians, a splinter group linked to multiple killings. Critics argue these are not random aberrations but the predictable output of a culture that prizes extreme logic over moral intuition.
Transhumanism does not stand alone. It feeds — and is fed by — the other letters of TESCREAL.
People, organizations, and the ideologies this one bleeds into. Drag nodes, hover to isolate, scroll to zoom — or pop it out fullscreen.
Primary documents and reporting used on this page. See the full References library → for the complete source list.