Help with a weird bayesian result regarding advisors

28 Views Asked by At

I'm thinking about cases where you receive advice from two people who are (i) independent, and (ii) supposed to be competent on getting questions within this domain correct to a probability of >.5

Suppose, for example, that they are each .8 competent at getting questions right.

Person 1 tells you: P!

Then, Person 2 tells you: P!

Intuitively, you should become .8 confident in P after hearing from person 1. (Suppose you were entirely agnostic beforehand, with no particular view).

And, intuitively, you should become MORE confident in P after hearing from person 2.

But if you try to incorporate these two pieces of evidence by multiplying them together -- as in (0.8)*(0.8) -- then you get a LOWER confidence in proposition P than when you only had evidence from one advisor.

What gives?