Rationalists always wonder: how come people are not more rational? How come you can prove a thousand times, using Facts and Logic, that something is stupid, and yet people will still keep doing it?
Henrich hints at an answer: for basically all of history, using reason would get you killed.
Simple Picture
ELI5: a woman in the Amazon spends a quarter of her day processing manioc through a multi-step, multi-day procedure that seems absurdly elaborate. A rational observer would simplify it — just boil the thing and eat it. Boiling removes the bitter taste, so it seems to work. But boiling does not remove the cyanide. The elaborate process does. The rational observer develops chronic cyanide poisoning. The “superstitious” woman lives.
The Manioc Problem
Indigenous Tukanoans process bitter manioc through scraping, grating, washing, separating fiber from starch from liquid, boiling the liquid into a beverage, letting the solids sit for two days, then baking. Every step matters. Skip the two-day wait and you get cyanide.
The critical difficulty: chronic cyanide poisoning develops slowly and does not cause immediate symptoms. Simply boiling removes the bitter taste but not the poison. Without clear signs of cause and effect, it is nearly impossible to figure out which steps are necessary through individual reasoning.
Historical proof: the Portuguese introduced manioc to West Africa in the 1600s but did not bring the processing methods. Many African groups failed to rediscover proper detoxification, leading to chronic cyanide poisoning that persists today. Some groups have evolved proper techniques, but they spread slowly. Centuries of individual intelligence operating on the problem have not solved what cultural transmission would have handled instantly.
This is locally-optimal at the civilizational level: the shortcut (skip the elaborate processing) works locally — the food tastes fine, no immediate symptoms. The cost is invisible and cumulative, distributed over decades. The tradition that looks irrational from the outside is the only thing preventing slow death.
Outsourced Randomness
Romans used divination — bird flights, animal entrails — to decide when and where to attack. This seems insane: generals risking thousands of lives because of a weird bird. But war is a classic case where a random strategy is optimal. If you are deciding whether to attack the right or left flank, it is critical that the enemy cannot predict your decision. If you are generally predictable — and you are — then outsourcing your decision to weird birds might be the best available strategy.
This reframes the “irrationality” of tradition: the tradition is not using reason poorly. It is solving a different problem than the one the rationalist sees. The foxhog orientation would recognize this: the rationalist insists on a legible, defensible decision process. The general who consults birds produces a genuinely unpredictable strategy — and unpredictability is the actual advantage being sought.
Seal Hunting as Accumulated Knowledge
Inuit seal hunting through ice requires: locating breathing holes by smell, assessing hole shape with carved caribou antler tools, covering with snow for muffling, placing down indicators for detection, driving a 1.5-meter harpoon blindly at full force when the indicator moves — a harpoon with a detachable sinew-tethered tip and a rear spike made of polar bear bone (which requires first hunting polar bears). Then rendering blubber into oil for soapstone lamps, finding specific moss species for wicks, and identifying old sea ice by color and texture because it has lost enough salt to be drinkable.
No individual — no matter how intelligent — could derive this system from first principles in a single lifetime. It is the accumulated product of thousands of generations of cultural transmission, each adding marginal improvements. The knowledge lives in the culture, not in any individual brain.
The Rationalist’s Trap
The will to think names intelligence as a moral virtue — the refusal to accept an answer you do not understand. But Henrich reveals the limit: understanding is not always available on the timescale of a human life. The manioc processor does not understand why the two-day wait matters. If she insists on understanding before complying, she gets cyanide. The seal hunter does not understand the full biochemistry of blubber combustion. If he waits for understanding, he freezes.
The Milo Criterion is the product-design version: human neuroplasticity is bound by the speed of meat, and shipping complexity faster than users can habituate produces the same result as handing a grandmother an iPhone Pro on day one — total rejection.
Taleb’s Fat Tony saw the same: “You are killing the things we can know but not express.” The tradition contains knowledge that survives only in the practice, not in the explanation. Pirsig warned that when analytic thought is applied to experience, something is always killed. Cultural evolution is the process that preserves what analysis would destroy.
Pirsig’s static quality is exactly this: the patterns that preserve the world. Static quality is old, complex, and full of memory. It cannot be replaced by Dynamic quality (innovation) without losing the survival knowledge it encodes. The strong gods — faith, tradition, moral codes, communal practices — look irrational to the rationalist. But they encode solutions to problems the rationalist has not yet encountered. The Babel Limit extends this across the scale axis: cultural diversity is the civilizational cousin of the manioc wait — apparently-irrational redundancy that preserves the option value of surviving the next catastrophic idea. Any universal consensus that dissolves the silos is the same mistake as boiling the manioc.
Common Misread
The dimwit take is “tradition is always right — just do what your ancestors did.”
The midwit take is “tradition is superstition — reason and science will always do better.”
The better take is that tradition is a lossy compression of ancestral experience, and the compression is lossy in both directions: it preserves knowledge that cannot be articulated, and it also preserves practices that have outlived their usefulness. Take that compression across enough generations and you get hyper-distilled symbols — semantic singularities so dense they generate meaning rather than pointing at it. The cross, 道, ॐ are the endpoint of the same process that encodes cyanide deaths into a set of steps no single person understands. The art is not choosing between tradition and reason but understanding that they solve different problems on different timescales. Reason works in domains where cause and effect are visible and fast. Tradition works in domains where cause and effect are invisible and slow — which, historically, has been most of life.
Main Payoff
Human success is not about being the smartest animal. It is about being the animal that can transmit complex knowledge across generations without each generation having to rediscover it. Individual intelligence is the ability to generate new solutions. Cultural evolution is the ability to preserve old ones. Both are necessary. But for most of human history, the second was far more important than the first — and the modern tendency to valorize individual intelligence at the expense of cultural transmission is, in Henrich’s terms, the same mistake as boiling the manioc and skipping the two-day wait. The Bitter Lesson creates a productive tension here: in a high-compute world where encoded knowledge decays within half a generation, what gets transmitted must shift from content to infrastructure — not the specific rules but the capacity to learn, search, and tolerate the mess of genuine discovery.
References:
- Joseph Henrich, The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter
- Scott Alexander, Book Review: The Secret of Our Success, Slate Star Codex