Sometimes I believe different things than other people that I respect. This is pretty concerning because wrong exists, and I don’t want to be wrong, so conditional on me being wrong I’d like to know that I am.
I sometimes enjoy arguments. Talking through an idea with someone, having them push back on places and refining your argument in response, can be helpful. Sometimes I’m also not in the mood to have an argument about something, and in this case having such an argument won’t generally result in a good conversation.
The preconditions for having a good argument are pretty narrow. I’m going to spell them out here in hopes of facilitating more good arguments. The basic recipe for a good argument is as follows:
- Establish crisply what the big disagreement is. Preferably the disagreement is actionable in some way.
- Note that it doesn’t actually have to be a disagreement to be a good argument.
- For instance, if Alek and JJ are both unsure of whether X or not X is true, then they can have a good conversation where they analyze arguments for and against X and try to figure out which of
is true. For simplicity I’ll assume that Alek and JJ believe different things for the rest of the discussion.
- Establish that both parties are interested in discussing their position, and open to having it be challenged, and are theoretically open to revising their position (even if they don’t expect to).
- If you think that you’re not currently in the right mental state for an argument, say so!
- Disclose “COIs”.
- What I really mean is cognitive biases that you should be suspected of.
- E.g., if the answer to a question would cause me to change my life plans in some drastic way, then maybe I’m prone to sunk costing / motivated reasoning.
- Both parties share their argument for their conclusion.
- Iteratively make implicit premises explicit until both parties agree that the conclusion follows from the premise.
- Identify cruxes — premises that if you believed, it’d make you change your mind about the whole issue.
- Unpack cruxes until someone changes their mind or you find a premise that you can’t decompose further, but that you could think of ways to test empirically.
- Think of ways to resolve the different in intution — what additional data could you collect, what do you predict, and if you saw something different would you change your mind?
- Collect that data.
- Update your beliefs as follows
, where you’ve precommited to the liklihood ratio, and can’t say post-hoc oh actually is different than I thought.
This is pretty involved, and probably not very many people are willing to follow this algorithm. In such cases, it’s best to politely yet firmly state that I have a policy against having arguments without these preconditions being met.
Also note that there’s a different between having a good argument and explaining your position to someone. Sometimes, person Y will be interested in hearing why I believe X, which is something that Y hasn’t thought about much. In such a case, it’s probably not appropriate to have an “argument” — it’s not a “fair match” because you’ve thought about X too much more than them. But anyways explaining X can still be valuable.