= 27d567bc2d48d9f40e9ccc20ea370b15)
ciphergoth comments on Eutopia is Scary - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (121)
Ugh, agreed.
I think P(newage|poly) - P(newage) > P(rationalist|poly) - P(rationalist) > 0.
I also think P(poly|rationalist) - P(poly) >> P(rationalist|poly) - P(rationalist), which is why we see it as a Common Interest.
As an aside, I've been reading your blog since (I think) before you joined LessWrong; like Wei Dai, you're one of the connections I've made to a different community that has appeared here. I usually read it through RSS, which I think broke. You also appear to have abandoned your earlier blog posts?
I think P(X|E) - P(X) is the wrong measure - should be the log likelihood ratio log(P(E|X)) - log(P(E|NOT X))
I was feeling uncomfortable about that myself.
In all likelihood, I shouldn't be using probability at all, because probability theory doesn't capture cause and effect well. Thinking back, what I should have said is just that rationalists are more likely to adopt polyamory than polyamorists are likely to adopt rationalism. The actual ratios of each are less relevant.
To be clear, this is almost the same as the formula you gave; I'm just using the log odds ratios formulation of Bayes theorem
LOR(X|E) = LOR(X) + log(P(E|X)) - log(P(E|NOT X))
where LOR(X) = log(P(X)/P(¬X))
in other words LOR(X|E) - LOR(X) = log(P(E|X)) - log(P(E|NOT X)) the log-likelihood ratio, the weight of evidence you need to update from one to the other.