Wednesday, September 20, 2006

Chandler on the God probability calculator

My dissertation advisor, in response to the Unwin probability calculator:
Ah, just as I suspected:

Someone reviewing the book writes:

Furthermore, the most difficult issue of all problems with the bayesian approach to probability, the initial a priori probability, he skips over facilely by declaring it to be 1/2. This may perhaps be better defended than any other number, but the explanation here is lacking.

Presumably that is his subjective prior probability for there being a God (of some particular sort? Or any sort that might be given that name?)

And what about there being 2 beings who,together, create and govern things? Does he count that as 'God' too? Or does that get assigned a different and distinct prior probability? How about 12 such beings?

For myself, the prior probability of there being a single 'personal' [Plantinga type] God is quite low, and, consequently, the bayesian reasoning doesn't get me anywhere near 50/50 for the existence of a personal God.

(You can post this on your discussion page if you want to.)


VR: Of course, Chandler is ruining all the fun. To make matters more complicated, my guess is that Chandler's assessment of the sorts of evidence that is presented in Unwin's book has already helped to form his current antecedent probability for theism.

Another problem has to do with how arguments get weighted. In my book I turned Lewis's one argument into six. Did I thereby make his argument six times more powerful? Is inability to explain evil a more serious fault than inability to explain miracle claims (assuming that these inabilities do obtain, as the author suggests that it does)? I'll bet there aren't going to be a lot of problem of evil atheists who would agree to that. And then what miracle evidence are we counting. Good evidence that God delivered the Qu'ran in Arabic to Muhammad, if we had it, would I think diminish my probability for the claim that an omniscient, omnipotent, all-powerful God exists.

Still, I wonder what adjustments could be made to the calculator so that someone could do a Bayesian calculation on their own assessment of how likely it is that God exists. I'll stick with my initial claim that this is a pretty cool concept, however much it may need to be debugged.


Jason Pratt said...

It serves as a (very limited) illustration to what I've elsewhere called one of the few proper uses of BT on this subject: it can be helpful in giving a person a clearer idea of how different people (including himself) reach their composite intuitive expectations about what is and isn't true.

But that is _NOT_ the same as using it as a tool for apologetical dueling.

VR: {{Good evidence that God delivered the Qu'ran in Arabic to Muhammad, if we had it, would I think diminish my probability for the claim that an omniscient, omnipotent, all-powerful God exists.}}

um.... I think I may be missing something here. The Qu'ran _doesn't_ testify to the existence of an omniscient, omnipotent, all-powerful God??

(I can see how good evidence, if we had it, that God delivered the Qu'ran to Mohammed, would diminish a likelihood estimate for the claim that some specifically _Christian_ and maybe _Jewish_ claims were true. But I didn't recall the Koran claiming God to be having something other than one or all of those characteristics. Guess I need to actually read it someday... {g?})

Victor Reppert said...

I suppose God's giving good evidence for the Qu'ran would be a problem for theism only if God also allowed there to be good historical evidence for the Judeo-Christian revelation. If this were true, then I would want to say that "God" is confusing the human race, since these revelations are contradictory. At the risk of making Muslims mad (maybe I should think twice, just in case there are terrorists reading my blog), I don't think there is good historical evidence for the Qu'ran, or the book of Mormon either.