A commonly shared piece of wisdom in the LessWrong community is to say or do the obvious things. Normally, this is treated as an unambiguously good thing to do, for example, see Nate Soares’s “Obvious advice”.

But I think there’s another genre of “obvious things” that requires more nuance: namely, the background assumptions that are so obvious to each of us or terms we hear so often they feel mundane. I think it’s uncontroversial to say that (“obviously”) what is obvious to you is often not obvious to others. The problem is that, what is obvious or not to others is itself not always obvious.

I think these obvious things often serve as an important barrier to good-faith conservation. I also think they serve to make people feel excluded from the communities I’m involved in: when speaking to people (especially younger people) who feel excluded from AI Safety, EA, or even Constellation, probably the most cited reason is that they feel either stupid for not knowing about obvious things or deliberately unwelcome because they did not agree with said obvious things. And I empathize with this; when people assume that the things I believe are so “obviously” ridiculous that no sane person would have them, I feel a deep sense of indignation that sometimes even explodes into an angry LessWrong post.

There are topics that are known to be obvious to some but controversial around in many of the Bay Area circles I frequent. Generally, because people have met others who (loudly/publically) disagree, these facts are known to be controversial. Classics include race, gender, and electoral politics. Various topics on AI Safety also fall under this category: the various Pause AI groups, SB 1047, and the usefulness of current interpretability methods come to mind.

In a sense, these topics are “easy” to handle in conversation, because people know to check for known controversies. In contrast, I think there are topics that are obvious to many people around me, that are

But I think there are controversial topics that are not obvious to many people I’ve interacted with, but are generally not known as such. The one that comes up most often is a form of implicit America-first beliefs: many people assume that it is “obvious” that we should support US chip export controls (or even to encourage the US labs race ahead) in order to make sure the US “wins the AI race” against China, while in fact many people do not share this America-first belief (even people from other western countries, especially in the last 2-3 years). Another such topic is whether or not it’s good to work for AI labs under common-sense ethics or deontological views (though recently, this specific debate has become more public). And there’s also the perennial issue of different levels of knowledge of jargon or technical knowledge

Sometimes people state nonobvious things as obvious in order to manufacture consensus: almost everyone is susceptible to social pressure, and it is only polite to agree. Sometimes people state nonobvious things as obvious in order to skip to what they consider the “important parts”: there's a reason a classic joke in math-heavy academic fields is that if you ever get really stuck with a proof, you should say that the result is so obvious even a baby could see it. And sometimes it really is to exclude people: I’d go so far as to say that most conversations would be worse if you’d have to restate every single assumption from the start.

But most of the time, my guess is that people assume obvious things because they are obvious to them, and not for another reason. Most people I know want to have honest, good faith conversations with others, even if they disagree.

I don’t have a complete solution to this problem. If I had to come up with something, I’d say that both speakers of such obvious facts and listeners irritated by the speaker’s assumptions should try to meet each other half way: that is, there are cheap cultural norms that both parties could follow to improve communication in the presence of differing obvious assumptions.

I think the speaker in conversations has some degree of responsibility for noticing when listeners seem confused, and allowing space for questions. Another useful tool for speakers in this context is original seeing/rationalist taboo: describing the object of the conversation without the standard jargon or shorthand. I think the listener also has responsibility for noticing their own irritation or confusion, and bringing it up in a polite and curious way.

For both parties, it’s important to remember to treat conversation parties with charity. People come from vastly different cultures that make different background assumptions. Sometimes the speaker is correct about their assumptions and sometimes they are not; curiously engaging rather than assuming malintent is more likely to figure out what is true.