I was just was reading the CC-BY-NC licenced textbook “Learning from Arguments” by Daniel Korman and remembered an old episode of the 80,000 Hours podcast (yes, the show that infamously gave the softball interview to SBF) discussing the problems with allowing infinite utility and figured it would be useful to spread this idea since not all refutations of Pascal’s Wagerare as definitive. The argument defeats itself because even if the probability of an anti-god reversing utilities that god assigns is infimitessimal, Pascal’s Wager shows that it too must be taken seriously. You can only believe in god if you somehow assign a 0 probability to anti-god but not to god or reject Pascal’s argument.
It’s complete bullshit. But the way it is bullshit is interesting. I had a response to it initially that was along the lines of well, there are lots of different gods, so why should you trust any of them. But if there is a very small chance of an infinite reward, that is still an infinite expected value. So shouldn’t you just flip a coin and choose one? A more sophisticated response is to say well, how do you know there isn’t a god of athiests that will reward athiests infinitely. If you accept Pascal’s Wager, then even if I grant you that the “god of athiests” a billion times less likely, you still can’t choose between it and the other gods because the expected value of any choice is infinite. So I can believe whatever I want to do and have the same expected reward. And if you don’t accept Pascal’s wager, then don’t talk to me until you have another reason to believe in your story. So you win either way by logic. And to paraphrase Lewis Carrol’s Achilles, “Then Logic would take them by the throat, and force them to concede the point! …Logic would tell them, “You can’t help yourself.”” 😁