Oh, definitely. This is part of yet another GDon-Giuseppe dispute.With respect, I don't think that is the core of the Godfrey-Carrier dispute.At the Battle of Pharsalus, Caesar's army was way outnumbered by Pompey's. Caesar won unexpectedly, against the odds. What is the probability that Caesar actually won the battle? Isn't it 100%?
I agree, and that's what I think Godfrey is getting wrong. If I understand what is being argued on Vridar correctly: the Pearl Harbor attack occurred, no doubt about it. That's 100%. But if we use Bayes analysis (noting that there is no reason why we would for this particular case), then we aren't going to get a result of 100%. It's very very close, but as Bayes always allows for the addition of further information, it will never be 100%. Godfrey seems to see this as a failure of Bayes for doing history, whereas his interlocutors are simply pointing out that it is just how Bayes works.In a nutshell: I am willing to assume that Pearl Harbor happened. Is that unsurpassed confidence because of some way of knowing that is different in kind from other occasions of inference about the real world, or rather is it the same way of knowing informed by an extraordinary quantity of high quality and nearly unanimous evidence that Pearl Harbor happened?
And if the latter (as Carrier would answer, and so would I), is it useful for me to bear in mind that despite the height of my confidence, its foundation can be distinguished in principle from what underlies the appreciation of a tautology? Sometimes yes, sometimes not worth the bother in my opinion.
When yes, then Bayes will typically allow me to represent some reservation, something finitely less than unity (some uncertain reasoning does address minuscule probabilities - nuclear power plant failure modes for example). But Bayes also allows me simply to project the problem down to the possible worlds in which Pearl Harbor happened, and move on to more interesting uncertainties about WW II in the Pacific.
Statistics: Posted by GakuseiDon — Thu Dec 19, 2024 10:48 am