So if you are trying to work out which hypothesis is simpler, how do you do that? You use your personal intuition.
I was using Solomonoff Induction as an example of a system that uses Occamian priors. My question was on those who assert that they don’t use Occamian priors at all, or for that matter any other type of objective prior. This usually seems to lead either to rejecting Bayesian epistemology in general or to asserting that any arbitrary prior works. I actually have no problem (in theory) rejecting Bayesian epistemology, as long as you still use some sort of probability-based reasoning.
When I referred to “personal intuitions” I meant controversial or arbitrary-sounding personal intuitions, such as “I feel there’s a god” or “I feel abortion is immoral” and then using those intuitions not as some sort of evidence but as priors. I get why someone would perhaps use universal intuitions as priors, along the lines of “there exists an external material world”, but why use an intuition where you know the next person over likely has a different intuition?
Your choice isn’t “Statistically correct prior” versus “Arbitrary prior”, your choice in the real world is between arbitrary priors and nothing at all.
When I referred to “personal intuitions” I meant controversial or arbitrary-sounding personal intuitions, such as “I feel there’s a god” or “I feel abortion is immoral” and then using those intuitions not as some sort of evidence but as priors.
I think most people just hold things like faith and emotions higher than logic and probability. Asking, say “how do you know that murder is wrong?” would, I imagine, freak out some people who aren’t philosophers. The whole idea that belief in god is a matter of probability is not held by many people, and moreso with moral questions. Most people, including intelleginet, educated people, do not seem to think that any justification for political opinions is needed except ‘anyone who disagrees with me is evil/stupid’.
Its actually worse than this—there are people who are deeply uncomfortable with having a notion of truth at all, because if there is a notion of truth, then some people are right and some people are wrong, and the idea that people might be wrong about something is offensive.
I was using Solomonoff Induction as an example of a system that uses Occamian priors. My question was on those who assert that they don’t use Occamian priors at all, or for that matter any other type of objective prior. This usually seems to lead either to rejecting Bayesian epistemology in general or to asserting that any arbitrary prior works. I actually have no problem (in theory) rejecting Bayesian epistemology, as long as you still use some sort of probability-based reasoning.
When I referred to “personal intuitions” I meant controversial or arbitrary-sounding personal intuitions, such as “I feel there’s a god” or “I feel abortion is immoral” and then using those intuitions not as some sort of evidence but as priors. I get why someone would perhaps use universal intuitions as priors, along the lines of “there exists an external material world”, but why use an intuition where you know the next person over likely has a different intuition?
Your choice isn’t “Statistically correct prior” versus “Arbitrary prior”, your choice in the real world is between arbitrary priors and nothing at all.
I think most people just hold things like faith and emotions higher than logic and probability. Asking, say “how do you know that murder is wrong?” would, I imagine, freak out some people who aren’t philosophers. The whole idea that belief in god is a matter of probability is not held by many people, and moreso with moral questions. Most people, including intelleginet, educated people, do not seem to think that any justification for political opinions is needed except ‘anyone who disagrees with me is evil/stupid’.
Its actually worse than this—there are people who are deeply uncomfortable with having a notion of truth at all, because if there is a notion of truth, then some people are right and some people are wrong, and the idea that people might be wrong about something is offensive.