Comment by heckless | Hacker News Reader

Comment by heckless | Hacker News Reader

Parent Full thread

> And if we go by the assumptions that every possible state in this set of possible states is equally likely (a common assumption in classical probability theory)

I know I am being nitpicky, but despite the fact that yes, it's common to assume equal priors, this often has no basis---especially in regards to this question. With only one observation we know virtually nothing about the distribution of possible observations other than that the state that we observed is in a region with nonzero probability. We can't conclude anything else with any certainty at all.

I was happy to see that in the article the author discusses this point in particular.

I don't know if "tongue in cheek" is the right word, but I'm not exactly being philosophically earnest. There are a lot of assumptions involved in any discussion of these kinds of things, and I've come to see a kind of inevitably foolishness in pondering this kind of stuff. I don't mean that in a derogatory way.

That said, I definitely can relate to the gravity of existential yearnings.

I agree about the futility of trying to answer this question---that is what the entire article is about after all. But I am a practicing statistician so I feel the need to "well actually" sometimes when I see stuff like this. I can't help it.

I get that. I'm not the most fluent in probability theory, but I'm vaguely aware that there are all kinds of ways of going about it. Even just in setting up a probabilistic model, there are assumptions to be made. Then you have to make certain assumptions about what kind of probability theory you are using. Then there is the linking of probability theory to logic... Logics themselves have their own sets of assumptions and rules.

I don't think trying to answer the question is futile, but expecting to find an answer probably is. I think there's a good chance that we are in agreement here.

Reply on news.ycombinator.com