Dangerous Knowledge
Infohazards, containment, and the contrarian ecology
Consider a claim that may well be true and is certainly dangerous: that some human populations, under the selective pressure of their ancestral environments, drifted toward different distributions of cognitive and behavioral traits. Framed as biology, it is unremarkable. Nobody riots over lactose tolerance, sickle-cell resistance, or the map of skin pigmentation across latitudes; these are textbook cases of populations adapting to environments, and no one reads a moral hierarchy into the ability to digest milk. Extend the same reasoning one step — from digestion to disposition, from enzymes to aggression or verbal reasoning — and the room goes silent, then loud. The facts have not changed their character. The framing has.
This is the structure of an infohazard: knowledge that is technically true, or at least plausible, but does damage merely by being known. Nick Bostrom named the category in 2011 to capture the fact that information can be dangerous without being false. Some infohazards are adversarial — they hand a capability to a bad actor, like the synthesis route for a pathogen. Others harm unintentionally, corroding the mind that holds them or the society that circulates them. Human Biodiversity, or HBD, belongs to the second kind, and it is among the most radioactive concepts in contemporary discourse. A gene variant associated with aggression in one ecological niche becomes, three retweets later, a stereotype about an entire people. A hypothesis about selection for verbal reasoning in one historical community becomes fodder for a ranking of the races. What began as inquiry devolves into ideology, and the devolution is so reliable that most people conclude the only safe move is to refuse the knowledge altogether.
I want to resist that conclusion, because it is a bad general policy and this is a bad specific case. The refusal buys safety at a price it never names, and — more to the point — it misdiagnoses where the danger lives. The hazard here is not in the genetics. It is in the misframing. Change the frame and the poison drains out of the same facts.
The Illusion of the Single Ladder
The frame that makes HBD toxic is the one almost everyone imports without noticing: intelligence as a universal scalar, a single quantity on which every mind can be ranked from more to less. Grant that picture and any population difference in the trait becomes a difference in rank — a place higher or lower on one ladder that everyone is climbing. Superior and inferior follow immediately, because that is what a ladder is for. The essentialist hierarchy is not an abuse of the data; it is the honest output of the scalar model applied to the data. If you accept the ladder, the toxic reading is not a misreading. It is the reading.
So the repair has to happen at the level of the model, not the level of etiquette. And the model this book needs is already built. In the volume on minds I argue at length that intelligence is not a substance a system contains but a performance measure, and a performance measure is meaningless without a specified context — a game. Intelligence is effectiveness at achieving goals within the constraints of a given game; ask “how intelligent?” without naming the game and you have asked “how fast?” without naming the course. I will not re-derive that here — Intelligence Is a Game We Play does the work — but its consequence for HBD is exact and, I think, decisive.
Once intelligence is relational, the universal ladder simply has no referent. There is no single quantity across which populations could be ranked, because there is no context-free intelligence to rank. The question “which group is smarter?” becomes as malformed as “which is the better tool, a hammer or a telescope?” — a question that only sounds answerable because it smuggles in an unstated purpose. What remains, once the ladder is gone, is a different and far less explosive question: what game was being played, in a given environment, and which strategies won it?
Three Cases, Reframed
Watch what happens to the most-cited HBD claims when you run them through that question instead of through the ladder.
The Ashkenazi intelligence hypothesis, on the scalar model, reads: this group is smarter. Reframed, it reads: this population was, for centuries, funneled by law and persecution into a narrow band of occupations — moneylending, trade, administration — that rewarded a specific cluster of cognitive strategies, and selection in that constrained game shifted the distribution accordingly. The claim, if true, is now a claim about which game a people were forced to play, not a claim about their standing on a cosmic scoreboard.
The MAOA “warrior gene”, on the scalar model, reads: this group is violent. Reframed: in resource-scarce, clan-based environments, a fast and forceful aggression response was an adaptive strategy — the right move in that game — and the allele frequencies track the history of the environment, not a defect in the people carrying them. Move the same individuals into a large-scale market state and the strategy that once paid now misfires, which tells you the trait was always a strategy-context match and never a rank.
Clannishness versus individualism, on the scalar model, reads: some peoples are inherently tribal. Reframed: intense kinship loyalty was the winning strategy in small-scale societies where trust could not be outsourced to institutions, and low kinship intensity became the winning strategy in large-scale market states where it could. Neither is higher. They are adaptations to different games, and each looks pathological only when transplanted into the other’s arena.
In every case the essentialist hierarchy dissolves into a context-bound adaptation, and it dissolves not by denying the underlying fact but by describing it correctly. This gives three guardrails that fall straight out of the games model and hold whenever the model is applied honestly:
Context-dependence. Winning one game implies nothing about standing in all games. A strategy tuned to one environment carries no title that transfers.
The agency principle. Even where averages shift under selection, an average is a fact about a distribution, not a sentence passed on a person. Individuals play games far outside their ancestral niches all the time; the population statistic never licenses a prediction about the human in front of you.
No universal scalar. There is no one number that sums intelligence across all contexts. IQ is not that number — it is a score in one standardized game, useful within its game and mute outside it.
The poison drains away once you abandon the illusion of a single ladder of intelligence. The right question was never who is more intelligent? It is what game is being played, and which strategies win it? Populations differ because environments differ, games differ, and strategies differ. Say that, and HBD becomes a descriptive account of how humans adapt — not a prescriptive ranking of human worth.
Containment, Not Suppression
Notice what this maneuver did and did not do. It did not delete a single fact. The allele frequencies are whatever they are; the selective histories happened or they didn’t; the reframing takes no position on any empirical question and forbids no line of inquiry. What it neutralized was the interpretive hazard — the scalar frame that turned every finding into a verdict. This is the distinction I want to make the method claim of the chapter: containment versus suppression.
Suppression attacks the facts. It declares the questions off-limits, the data unpublishable, the researchers unemployable, and it treats the knowledge itself as the contaminant to be sealed away. Containment attacks the framing. It leaves the facts in open circulation and disarms the frame that made them explosive — the way you make a claim safe not by burning the document but by teaching people to read it. Bostrom’s own vocabulary points this way: the danger of an infohazard is a property of how information meets a mind, and there is more than one place to intervene. Suppression intervenes on the information. Containment intervenes on the meeting.
The two are not equally cheap, and the difference is the whole argument. Suppression looks costless because its price is paid elsewhere and later — in the questions that never get asked, the corrections that never arrive, the epistemic blind spots that institutions grow around a forbidden zone and then forget they are blind. That bill comes due in full in The Forbidden Pattern, where I trace what an entire culture pays for the topics it has agreed not to see. Here the point is narrower: when containment is available — when a reframing genuinely defuses the hazard while preserving the truth — it is strictly better than suppression, because it keeps the facts and pays none of suppression’s hidden costs. HBD is a case where containment is available. Not every case is. But the burden of proof runs the other way from where the culture places it: suppression should be the last resort, not the first reflex.
The Carriers
Containment is a property of frames, but frames do not float free. Radioactive content moves through the world on carriers — the particular people who take it upon themselves to say the unsayable — and the carrier shapes the payload. A true fact delivered by one kind of speaker informs; the same fact delivered by another detonates. So a full account of dangerous knowledge has to include an ecology of the people who traffic in it, because contrarianism is not one thing. It decomposes into strategies with sharply different risk profiles, and the differences are instructive.
Taboo-flipping takes a socially uncomfortable subject — sex work, kink, polyamory — and dissects it with clinical curiosity, aiming to normalize open discussion and expose hypocrisy. Aella is the type case. The risk is survivable: these taboos are provocative enough to draw attention but not so radioactive that debate becomes impossible. This is contrarianism operating near the containment ideal — the frame (“let us just look at the thing plainly”) does most of the work, and the hazard stays low.
Provocation-laundering starts from the most explosive taboo available, then reframes it through abstract analogies — diamonds, lab-grown meat — before running the same logical move in safer domains. The aim is to be known as the one who will say what others won’t. Decker is the exemplar, and the strategy is misnamed by its own practitioners, because it does not launder anything. Once the association is made, the safe analogies do not wash the original provocation out; they keep it in circulation under a thin coat of respectability. The risk is radioactive, and unlike taboo-flipping the danger is not contained by the frame — the frame is a delivery mechanism for it.
Cynical nihilism does not flip taboos so much as tear down whole fields — IQ, academia, formal risk models. The payoff is authority as a professional skeptic, the clearinghouse for iconoclasm; Taleb is the standing example. The risk is volatile: the same performance reads as truth-telling to one audience and crankery to another, and which reading dominates can swing with the topic and the decade.
The ironist, finally, packages fringe or taboo ideas in layers of satire, science-fictional thought experiment, and studied deniability — just exploring ideas — while cultivating an audience that takes the exploration in deadly earnest. Curtis Yarvin writing as Moldbug is the pure case; the early Yudkowsky had a strain of it too. The risk is ambiguous, and ambiguity is the strategy’s entire engine. The mask can work for years. But it is a mask, and the deniability is load-bearing, which means it can fail: eventually, for most who wear it, it slips.
These grade cleanly from survivable to fatal, and the grading is the lesson. Contrarianism is a tool. Used well, it breaks idols — it forces a culture to confront the assumptions it mistook for facts, and it is very nearly the only thing that reliably does. Used badly, it brands the user permanently with one grotesque provocation and discredits the truth he was carrying along with him. Some contrarians break idols. Others just set themselves on fire and call it illumination — and the tragedy is that the crowd, watching the second kind burn, learns to fear the first kind too.
The Hinge
So the same body of facts can be handled or weaponized, and the difference is not in the facts. It is in the frame that meets them and the carrier that moves them. Containment is what handling looks like: keep the truth, disarm the interpretation, and refuse the false comfort of suppression whose price is only deferred. It is a discipline — and like any discipline it can be practiced or abandoned.
What abandonment looks like is the subject of everything that follows. The reframing move I have used here to defuse a hazard is, mechanically, the same move that can be run in reverse to manufacture one: choose the frame that inflames rather than the frame that informs, and you have converted a fact into a weapon. The next part of this volume is a catalog of that reverse operation — the tactics by which discourse is turned against truth on purpose, beginning with The Grammar of Bad Faith. Everything up to now has been the ecology of how patterns live and spread. What comes next is how they are wielded.