Scott Alexander, (Alexander 2020)


Socratic grilling is “like Socratic questioning, but a little harsher and more confrontational in order to get to the point more quickly” (Alexander 2020). Scott Alexander encourages leaning in to such conversations without shutting them down with an “I know more than you do”- or “Just because”-type responses.


The learning done through Socratic questioning is worth the potential friction of such a conversation. Be careful that the person is asking questions in good faith rather than attempting to Control the conversation.


Imagine an kid in school first hearing about germ theory. The conversation might go something like this:

Teacher: Many diseases like the common cold are spread by germs, when one infected person contacts another.

Student: But I got a cold a few weeks ago, and I never touch anyone except my family members. And none of them were sick.

Teacher: You don’t need to actually touch someone. Sometimes it can spread through mucus droplets in the air.

Student: And one time I was camping in the woods for a month, and then I got a cold, even though I hadn’t been around anybody.

Teacher: If it was spring, you might have gotten allergies. Allergies can feel a lot like a cold, but they aren’t spread by germs.

Student: It was fall.

Teacher: Then maybe it was an unusual allergy, or some other condition.

Student: Hey, wait. If germs are spread from person to person on touch, why doesn’t the government just mandate one week when nobody is allowed to touch anyone else? Then all the germs will die and we’ll never have to worry about germs again.

Teacher: That’s a good question. A lot of germs have what’s called reservoirs in the environment, where they live when they’re not infecting humans. Even if the government tried your plan, probably most sicknesses would come back from their reservoirs.

Student: I’ve never seen a germ reservoir. Where are they?

Teacher: They’re not literal reservoirs like a water reservoir, that’s just what we call it when germs live in bats or raccoons or something.

Notice a few things about this conversation.

First, it’s really good that it happened. The student was clearly confused at many points. First, he had direct evidence that seemed to contradict the teacher’s claims that germs only spread by touch. Second, he had a sort of efficient-market-style confusion: germ theory seems to imply an easy way to eliminate all sicknesses forever, so why hasn’t someone picked this low-hanging fruit? Third, he was confused by an awkward term – he thought germ reservoirs were lakes full of germs in the hills somewhere.

All of these confusions are totally understandable. In fact, they’re a really good sign he’s paying attention – that he’s trying to figure out what germ theory really means and how it interacts with the rest of his worldview. They’re the direct opposite of guessing the teacher’s password. One day this kid is going to be an amazing scientist.

Second, to a hostile observer, it would sound like the student was challenging the teacher. Every time the teacher tried to explain germ theory, the student “pounced” on a supposed inconsistency. When the teacher tried to explain the inconsistency, the student challenged her explanations. At times he almost seems to be mocking the teacher. Without contextual clues – and without an appreciation for how confused young kids can be sometimes – it could sound like this kid is an arrogant know-it-all who thinks he’s checkmated biologists and proven that germ theory can’t possibly be true. Or that he thinks that he, a mere schoolchild, can come up with a novel way to end all sickness forever that nobody else ever thought of.

And the thesis of this post is that you must never, ever say that. Saying that is so bad. Smack down that student once, say “I think I know more about germ theory than you do”, make him feel like he challenged your authority and that’s bad – and the best case scenario is he will never ask questions to resolve his confusion again. The worst case scenario is that he stops feeling the confusion entirely, or stops thinking of forcing things to fit together and make sense as a desirable goal to have.

One of the most important rationalist skills is “noticing your confusion”. But that depends on an even more important proto-skill of wanting things to make sense. If you lose that skill – if it stops bothering you and seeming like a problem when things don’t make sense to you – you will never notice your confusion and you will never become a good scientist or a good anything-else-that-requires-independent-thought. And interpreting an attempt to explore dissonance as a status grab that needs to be knocked down is absolutely fatal for that skill. Instead, you need to think of it as Socratic grilling – like Socratic questioning, but a little harsher and more confrontational in order to get to the point more quickly.

Tolerating this is harder than it sounds. Most people can stay helpful for one or two iterations. But most people are bad at explaining things, so one or two iterations isn’t always enough I’ve had times when I need five or ten question-answer rounds with a teacher in order to understand what they’re telling me. The process sounds a lot like “The thing you just said is obviously wrong”…“no, that explanation you gave doesn’t make sense, you’re still obviously wrong”…“you keep saying the same thing over and over again, and it keeps being obviously wrong”…“no, that’s irrelevant to the point that’s bothering me”…“no, that’s also irrelevant, you keep saying an obviously wrong thing”…“Oh! That word means something totally different from what I thought it meant, now your statement makes total sense.”

But it’s harder even than that. Sometimes there is a vast inferential distance (a) between you and the place where your teacher’s model makes sense, and you need to go through a process as laborious as converting a religious person to a materialist worldview (or vice versa) before the gap gets closed. The process of learning to really appreciate communism, or libertarianism, or whatever, coming from a diametrically opposed philosophy, looks a lot like dozens of questions about “but isn’t that an atrocity?” “wouldn’t this inevitably lead to dystopia?” and hearing what your interlocutor has to answer. It’s so, so tempting to round this off to them trying to gotcha you (as indeed sometimes it will be) and assume they’re not really committed to trying to understand.


Alexander, Scott. 2020. “Socratic Grilling.” Slate Star Codex.