Even observing “This is only believed by a fringe group of of low-status people not taken seriously by society at large” is sufficient to assign it an extremely low prior probability(single-digit percentages?) without even attempting to consider direct evidence for or against the proposition.
You may be right. But is there no better methology for estimating priors than to say “surely this prior must be in the single digit percentages”. It seems very method-less.
The method is “set your prior probability equal to society’s judgement of its probability”. For anything where society as a whole has an opinion at all, it’s a better approximation than any other one-piece-of-evidence method around, and is thus a great way to set your priors.
Even observing “This is only believed by a fringe group of of low-status people not taken seriously by society at large” is sufficient to assign it an extremely low prior probability(single-digit percentages?) without even attempting to consider direct evidence for or against the proposition.
You may be right. But is there no better methology for estimating priors than to say “surely this prior must be in the single digit percentages”. It seems very method-less.
The method is “set your prior probability equal to society’s judgement of its probability”. For anything where society as a whole has an opinion at all, it’s a better approximation than any other one-piece-of-evidence method around, and is thus a great way to set your priors.
As a general statement, this doesn’t seem to be true. I am also not quite sure what—in real life—is a “one-piece-of-evidence method”.