Why Are All the Smart People So Bad at History?

You know the type. MIT-trained, Substack-fluent, AI-curious. They can quote Hume, dabble in Bayesianism, and confidently wield phrases like "regression discontinuity" and "effective optimization landscape" in casual conversation. They’re very concerned about the future. Superintelligence. Existential risk. Civilizational stability. The longtermist dream: guiding the ship of humanity through narrow straits toward the sunlit uplands of utopia.

They’re also - far too often - terrible at history.

That might sound glib, but it’s a problem. Not because history is sacred, or because the past is a perfect guide to the future. It’s a problem because these folks, who wield so much intellectual and - increasingly - institutional power, treat history as either a minor footnote or a convenient prop. And when you treat history like that, you don’t just misread it. You weaponize it. You flatten it into caricature, cherry-pick to justify your models, and bulldoze over inconvenient truths.

This is a subculture that praises nuance and complexity in physics and economics but laps up the most simplistic historical narratives imaginable. Roman Empire falls = decadence. French Revolution = chaos. World War II = moral clarity. Cold War = MAD worked. Colonization = unfortunate but efficient. Civil Rights Movement = MLK was nice. Every great disruption or transformation boiled down to a graph or a game-theoretic insight. History as analytics. History as abstract inputs. History, sans contradiction, contingency, and culture.

Systems Over Stories

The structure of rationalist and technocratic thinking incentivizes a flattened historical consciousness. They favor systems over stories. They trust models over memories. They crave optimization, not interpretation.

It’s why you’ll see someone argue that we should re-engineer democracy because "Athens failed" - without once reckoning with who got to vote in Athens, what slavery meant, what ostracism did, or how empire shaped its economy. It’s why someone else will spin a counterfactual where World War I never happens, and suddenly we get interplanetary civilization by 1950. It’s not just wrong. It’s alien. It’s the past as speculative fanfiction, stripped of causality and consequence.

The epistemic sin here is not ignorance. It’s certainty. There’s no humility in these histories. No pause. No sitting with the messiness. There’s a hunger to make history tractable, computable, gamified. The impulse is always to turn it into a set of levers you can pull, or knobs you can tweak. It’s not "what happened," it’s "how we could have made it better."

The Counterfactual Trap

This attitude infects Effective Altruism in particular. EA has many strengths - moral seriousness, data-driven thinking, global scope - but its engagement with history often feels like a means to an end. EA loves counterfactuals: What if we’d prevented malaria in 1850? What if we’d stopped Hitler in 1923? What if Bismarck had implemented universal basic income? These aren’t idle questions. They’re foundational to EA’s whole worldview: that we can model the outcomes, score the possibilities, and act accordingly.

But history isn’t a clean dataset. It doesn’t behave like a lab experiment. You can’t rerun the 20th century with a tweak and expect a stable result. Contingency is everything. People make choices. Systems react. Culture evolves. Timing matters. Change one node and the network shifts in ways you can't predict. There’s no control group for history. You don’t get to A/B test the Russian Revolution.

Technocrats and rationalists often talk like you can.

This isn't just annoying. It's dangerous. Because when you treat history as an optimization problem, you miss what it's actually good for: perspective. Understanding how systems fail. How ideologies collapse. How revolutions devour their children. How power accumulates and corrodes. History teaches scale. It teaches caution. And, yes, it teaches radical possibility—but not the frictionless kind.

Mess, Memory, and Models

The irony is that the very people who talk most about "existential risk" often have the thinnest understanding of the very events that came closest to ending civilization. Ask your average longtermist about the Cuban Missile Crisis and they might give you a Bayesian estimate of nuclear launch probability. But ask them about Khrushchev’s internal politics, or the role of Vasily Arkhipov, or the military chain of command on October 27th, 1962 - and you’ll get blank stares. That stuff doesn’t fit in the model. Too messy. Too human.

But history is the study of human mess.

And that mess matters. Because we’re not going to navigate the 21st century with clean models. The next pandemics, climate shocks, political collapses—they’re not going to play by the rules of probability theory. They’ll be shaped by culture, miscommunication, inertia, ego. Just like the last ones.

Toward Better Ancestors

So why does this subculture keep making the same mistake?

Part of it is structural. The communities that birthed these ideas - Silicon Valley, elite academia, the rationalist blogosphere - value abstraction over narration. They trust math. They don’t trust emotion. They see history departments as soft, ideological, unserious. But the answer to bad history isn’t no history. It’s better history.

Another part is psychological. A lot of rationalists come from outsider positions. Smart kids who didn’t fit in. People who needed the world to make more sense than it did. For them, historical narrative isn’t helpful - it’s painful. It’s a flood of irrationality. A constant reminder that things don’t work the way they should. Their coping mechanism is to reframe everything into a puzzle they can solve.

But the biggest reason is ideological. Whether they admit it or not, many of these thinkers have a Whig view of the world. A belief in progress as default. A sense that, barring catastrophe, things trend upward. This isn’t unique to rationalists - it’s embedded in liberal technocracy too. But it’s especially dangerous for people who want to think in Expected Value terms. Progress becomes quantifiable. The arc of history bends not toward justice, but toward utility.

This is how you end up with smart people arguing that colonialism was net good because it increased GDP. Or that Bostrom’s paperclip maximizer is a more important topic than anti-colonial resistance. Or that we should ignore current political crises in favor of "trajectory changes" in the deep future.

But history has a long memory. And a sharp edge.

You don’t get to ignore the past until you need it to justify your simulations. You don’t get to treat historical trauma as background noise in your utilitarian calculus. And you certainly don’t get to build utopias on the bodies of forgotten revolutions.

If you want to think clearly about the future, you have to think clearly about the past. Not in a romantic way. Not in a didactic way. But in a serious, detailed, critical way. You have to engage with historians. Economic historians. Cultural historians. Social historians. Postcolonial theorists. People who study the margins.

And yes, that means reading things that aren’t on LessWrong. It means reading history books. From Gibbons to Tuchman, from McPherson to Diamond. It means sitting with stories that don’t resolve neatly. It means learning how to think historically.

Because history, at its best, doesn’t tell you what to do. It tells you what’s been done. What’s been tried. What’s been broken. What’s been survived.

It teaches you that every utopia has a border. Every policy has a shadow. Every breakthrough has a cost.

The smartest people in the room should know that. They should hunger for that kind of knowledge. Not because it’s efficient. Because it’s real.

The future deserves better ancestors.