David Hobby wrote: > > Yes, that's the kind of thing I was thinking of. Alberto > was talking about probability. Since all probabilities > sum to one, that might well imply that each god got > probability zero. > No, there may be infinite a priori gods, but they can form a converging sequence, like 1/2, 1/4, 1/8, ...
BTW, in Bayesian analysis, you can even consider an "improper prior", and assign to an enumerable quantity of gods the _same_ probability, and end up, after observations, with a proper probability distribution. Like this: imagine a sequence of gods labeled 1,2,... and assign to each of them the same a priori probability [this is an improper prior - there is no such distribution]. Then, let's do an experiment that will succeed for the n-th god with probability 1/2^n. If this experiment succeeds, the "a posteriori" probability will be the "bona fide" p(n) = 1/2^n. Alberto Monteiro _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
