There is this basic idea (I think from an old blogpost that Eliezer wrote) that if someone says there are goblins in the closet, dismissing them outright is confusing rationality with trust in commonly held claims, whereas the truly rational thing is to just open the closet and look.
I think this is correct in principle but not applicable in many real-world cases. The real reason why even rational people routinely dismiss many weird explanations for things isn’t that they have sufficient evidence against them, it’s that the weird explanation is inconsistent with a large set of high confidence beliefs that they currently hold. If someone tells me that they can talk to their deceased parents, I’m probably not going to invest the time to test whether they can obtain novel information this way; I’m just going to assume they’re delusional because I’m confident spirits don’t exist.
That said, if that someone helped write the logical induction paper, I personally would probably hear them out regardless of how weird the thing sounds. Nonetheless, I think it remains true that dismissing beliefs without considering the evidence is often necessary in practice.
If someone tells me that they can talk to their deceased parents, I’m probably not going to invest the time to test whether they can obtain novel information this way; I’m just going to assume they’re delusional because I’m confident spirits don’t exist.
This is failing to track ambiguity in what’s being refered to. If there’s something confusing happening—something that seems important or interesting, but that you don’t yet have words to well-articulate it—then you try to say what you can (e.g. by talking about “demons”). In your scenario, you don’t know exactly what you’re dismissing. You can confidently dismiss, in the absence of extraordinary evidence, that (1) their parents’s brains have been rotting in the ground, and (2) they are talking with their parents, in the same way you talk to a present friend; you can’t confidently dismiss, for example, that they are, from their conscious perspective, gaining information by conversing with an entity that’s naturally thought of as their parents (which we might later describe as, they have separate structure in them, not integrated with their “self”, that encoded thought patterns from their parents, blah blah blah etc.). You can say “oh well yes of course if it’s *just a metaphor* maybe I don’t want to dismiss them”, but the point is that from a partially pre-theoretic confusion, it’s not clear what’s a metaphor and it requires further work to disambiguate what’s a metaphor.
There is this basic idea (I think from an old blogpost that Eliezer wrote) that if someone says there are goblins in the closet, dismissing them outright is confusing rationality with trust in commonly held claims, whereas the truly rational thing is to just open the closet and look.
I think this is correct in principle but not applicable in many real-world cases. The real reason why even rational people routinely dismiss many weird explanations for things isn’t that they have sufficient evidence against them, it’s that the weird explanation is inconsistent with a large set of high confidence beliefs that they currently hold. If someone tells me that they can talk to their deceased parents, I’m probably not going to invest the time to test whether they can obtain novel information this way; I’m just going to assume they’re delusional because I’m confident spirits don’t exist.
That said, if that someone helped write the logical induction paper, I personally would probably hear them out regardless of how weird the thing sounds. Nonetheless, I think it remains true that dismissing beliefs without considering the evidence is often necessary in practice.
This is failing to track ambiguity in what’s being refered to. If there’s something confusing happening—something that seems important or interesting, but that you don’t yet have words to well-articulate it—then you try to say what you can (e.g. by talking about “demons”). In your scenario, you don’t know exactly what you’re dismissing. You can confidently dismiss, in the absence of extraordinary evidence, that (1) their parents’s brains have been rotting in the ground, and (2) they are talking with their parents, in the same way you talk to a present friend; you can’t confidently dismiss, for example, that they are, from their conscious perspective, gaining information by conversing with an entity that’s naturally thought of as their parents (which we might later describe as, they have separate structure in them, not integrated with their “self”, that encoded thought patterns from their parents, blah blah blah etc.). You can say “oh well yes of course if it’s *just a metaphor* maybe I don’t want to dismiss them”, but the point is that from a partially pre-theoretic confusion, it’s not clear what’s a metaphor and it requires further work to disambiguate what’s a metaphor.
As the joke goes, there’s nothing crazy about talking to dead people. When dead people respond, then you start worrying.