It’s difficult to assess whether your objections are to what you call vague Bayesianism or to Bayesianism proper. For example, your point about the sky being blue, is an objection to the former but not the latter.
If I were to push back I would deny that Bayesianism is useless in debates for the existence of God. One can argue for plausible priors and likelihoods for competing hypotheses and evidence. It depends on your theory of probability, but it’s the dominant mode of discussion in the literature because it’s fruitful.
I would also challenge your claim that Bayesianism is just a useful tool that isn’t logically required. Maybe it is. It is a theorem of probability theory itself. So, it is a theorem of mathematics. Plus it’s pretty hard to deny it’s a norm of rationality as well given the Dutch book argument. There are moves you could make to deny those implications, but they seem pretty costly.
But like I said, I’m not sure whether these remarks are relevant because I’m not sure what the real target was. I do think your piece is right to point out the sloppiness of vague Bayesianism.
I'm about to clean the chinchilla cage, so I'll be brief:
1) I'm taking more aim at vague Bayesianism, yes.
2) I think (real) Bayesianism is useful for these discussions, but their use would be more limited. I think people don't fully understand Bayesianism and so they overuse it or use it poorly or use it where it's not fruitful
3) for your second paragraph, I agree. I will also say that it's probably normal in the field for reasons to do with the philosophers doing the research (their interests, etc) than it is because skeptical methods are bad. Skeptical methods lead to a dead end in terms of discourse, so in a weird darwinian way I'd expect this norm to take over. That doesn't mean it's wrong or good or bad though
4) for your third paragraph, I would have to get more specific (and so I'm liable to make mistakes!) and I may be conflating Bayesian modeling with modeling altogether. I.e when I say it's a posteriori, what I mean to say if your model gives 60% credence probability of something happening, and we repeat the process, it should happen 60% of the time. So, like, the validation of the model is a matter of fact, it's either useful/accurate (as much as it says it is) or it's not. But I'm also not fully confident than I'm using proper terminology and/or doing other conflation
5) the target, admittedly, is sloppy, I just want to reiterate the importance of filters like causality and relevance when we trot out Bayesian reasoning, which I don't think people do much. My next post on testimony should sort of bridge the theme here
It’s difficult to assess whether your objections are to what you call vague Bayesianism or to Bayesianism proper. For example, your point about the sky being blue, is an objection to the former but not the latter.
If I were to push back I would deny that Bayesianism is useless in debates for the existence of God. One can argue for plausible priors and likelihoods for competing hypotheses and evidence. It depends on your theory of probability, but it’s the dominant mode of discussion in the literature because it’s fruitful.
I would also challenge your claim that Bayesianism is just a useful tool that isn’t logically required. Maybe it is. It is a theorem of probability theory itself. So, it is a theorem of mathematics. Plus it’s pretty hard to deny it’s a norm of rationality as well given the Dutch book argument. There are moves you could make to deny those implications, but they seem pretty costly.
But like I said, I’m not sure whether these remarks are relevant because I’m not sure what the real target was. I do think your piece is right to point out the sloppiness of vague Bayesianism.
I'm about to clean the chinchilla cage, so I'll be brief:
1) I'm taking more aim at vague Bayesianism, yes.
2) I think (real) Bayesianism is useful for these discussions, but their use would be more limited. I think people don't fully understand Bayesianism and so they overuse it or use it poorly or use it where it's not fruitful
3) for your second paragraph, I agree. I will also say that it's probably normal in the field for reasons to do with the philosophers doing the research (their interests, etc) than it is because skeptical methods are bad. Skeptical methods lead to a dead end in terms of discourse, so in a weird darwinian way I'd expect this norm to take over. That doesn't mean it's wrong or good or bad though
4) for your third paragraph, I would have to get more specific (and so I'm liable to make mistakes!) and I may be conflating Bayesian modeling with modeling altogether. I.e when I say it's a posteriori, what I mean to say if your model gives 60% credence probability of something happening, and we repeat the process, it should happen 60% of the time. So, like, the validation of the model is a matter of fact, it's either useful/accurate (as much as it says it is) or it's not. But I'm also not fully confident than I'm using proper terminology and/or doing other conflation
5) the target, admittedly, is sloppy, I just want to reiterate the importance of filters like causality and relevance when we trot out Bayesian reasoning, which I don't think people do much. My next post on testimony should sort of bridge the theme here
I honestly agree, I just think that’s a tangled philosophical discussion and in the spirit of charity, conceding it!
I agree!
Great post! Been looking forward to reading it and it didn’t disappoint!
Thank you!