The Last Useful Anxiety

There is a particular species of dread that I suspect is quietly going extinct, and I find myself mourning it the way one mourns a predator rather than a pet: aware that its absence will be worse than its presence, even if the presence was never comfortable.

I mean the productive fear of being wrong.

Not the social anxiety of being caught wrong in front of others — that one is doing fine, thriving even, metastasizing into a culture of epistemic cowardice so thorough that the average educated person will hold any position at all with such strategic vagueness that no empirical result could ever embarrass them. What I mean is something more private: the cold-sweat 3am variety of "what if I've been catastrophically mistaken about something important and I won't discover this until the mistake has already propagated into consequential decisions?" That specific, pointed fear. That's the one I'm worried about.

I spent a while in medicine doing psychiatry, which has an unusual relationship with being wrong. A surgeon who is wrong cuts something incorrectly and the patient deteriorates in a way that is legible, measurable, and traceable. A psychiatrist who is wrong often produces an outcome that looks exactly like a correct psychiatrist whose patient is refractory. The feedback is attenuated to the point of near-uselessness. This is, I now think, probably not unrelated to why psychiatry has accumulated a rather impressive historical record of confident nonsense — lobotomies, insulin coma therapy, the entire Freudian apparatus — that each era found obviously correct and the subsequent era found obviously barbaric. You cannot calibrate without feedback. You cannot update without calibration. Without update, you're just an articulate random number generator who went to medical school.

This should terrify more people than it does.

The curious thing about modern epistemics, at least in the educated professional-class milieus I now swim in — hospitals, certain corners of the internet, the occasional ACX meetup in Edinburgh where everyone is worryingly similar — is that there's a great deal of performance of updating. People have read their Kahneman. They know to name their biases. They've absorbed enough rationalist-adjacent discourse to speak fluently about epistemic hygiene. And yet.

And yet the actual beliefs held by these people are almost uniformly the beliefs you'd predict from knowing their social tribe, their career incentives, and their educational background. The machinery of reasoning is elaborate and ornate and attached to conclusions that were reached before the reasoning began. I include myself in this accusation and particularly resent that including myself doesn't inoculate me against it.

Here is what I think is actually happening: the fear of being wrong has been successfully decoupled from the experience of being wrong, because we have built social structures where being wrong in the right way is categorically different from being wrong in the wrong way. A mainstream expert who confidently predicted the Iraq War, or the 2008 crisis, or the initial COVID models, pays essentially no professional price. The class of people who were right but wrong in the wrong way — wrong priors, wrong political valence, wrong aesthetic — pays a significant reputational one. The incentive gradient thus runs directly against calibration and directly toward conformist wrongness.

This is not news. The more interesting question is what you do about it when you've noticed it.

My own approach, which I offer with appropriate humility given my demonstrated susceptibility to the same failures, is to identify the questions on which I am most comfortable with my current views and treat that comfort itself as evidence of motivated reasoning. Not conclusive evidence — sometimes you're comfortable because you've thought carefully about something and the answer is genuinely robust — but sufficient to warrant aggressive reexamination.

The questions I'm most comfortable about are, suspiciously, the ones where my views would get me fired in certain institutions and celebrated in others, which suggests the views are tracking social sorting rather than truth. More usefully: the questions where I'm most comfortable are often the ones I haven't seriously stress-tested because I haven't needed to. I operate in environments where those views are load-bearing, where being wrong about them would require costly revisions to self-concept and social position. That is precisely where the fear ought to be highest and where it is structurally suppressed.

The practical upshot is boring but under-practiced: find someone who holds the opposite view and is more intelligent than you, and engage with them as if you might lose. Not in the competitive debate sense where you're constructing counterarguments while they speak, but in the epistemic sense where you're genuinely trying to update. This is miserable. I do not recommend it for enjoyment. I recommend it because the alternative — a life of confident wrongness insulated from correction by social sorting — seems to me a rather significant waste of whatever cognitive hardware you've been allocated.

There's also the question of what you do with the anxiety once you've cultivated it. My answer is that you mostly just carry it. It is not, fundamentally, a problem to be solved. It's a feature. A mind that has fully resolved its uncertainty about important questions is a mind that has stopped engaging with reality and is coasting on a cached model. The uncertainty is doing something. The 3am dread that you've been subtly wrong about something that compounds is the feeling of taking your epistemic responsibilities seriously, which is uncomfortable in much the same way that physical training is uncomfortable, and for the same reason: you're working against a system that would rather rest.

What I'm actually worried about, if I'm honest, is not that I'm wrong about this or that object-level question. I'm wrong about things regularly, and the updates are usually more interesting than the original beliefs. What I'm worried about is losing the productive terror that makes the update possible. The ability to be wrong in a way that reaches you.

Because the alternative, which I see everywhere around me, is a certain kind of glossy confidence — articulate, credential-bearing, socially calibrated — that is wrong in ways it will never discover.

That seems like a much worse fate than being afraid.

Edit

Pub: 24 Feb 2026 18:39 UTC

Views: 1