So if we’re not ideal bayesian reasoners, how do we make progress? It seems like things need to come from outside bayesianism somehow. Or is it just a lottery that we can’t ever do better at except by having more people actively do bayesian reasoning about stuff? If there is something outside bayesian reasoning/epistemology and new ideas come from there, then it seems like it does things bayesianism does not, or is better at it. And such a thing would either contradict bayesianism at worst or make it (partially) obsolete at best.
Sorry for the delayed replay. My comments are being held for review.
There aren’t just two possibilities “ideal bayesian reasoning” and “useless rubbish”. There is a huge range of heuristics, ad-hoc models, evolved instincts, and everything else in the mix. These are all ‘outside bayesianism’ and while the collection is almost certainly worse than ideal bayesian reasoning, they are not useless.
That also doesn’t mean that we can only improve by having more people actively do bayesian reasoning about stuff, though there are certainly many cases where people would be better off actively doing bayesian reasoning.
There are many ways to improve incredibly complex systems such as human minds and their interactions. It’s far from certain that applying more bayesian reasoning is the best way. We are definitely not capable of reaching the ideal, and will have to settle for something imperfect. Maybe there is a better approximation than “try bayesian reasoning as far as our limited human brains can handle”, maybe there is not.
The main point is that we don’t know anything better, and pretty much everything else that we do know looks worse. However, there is a lot that we don’t know, and far more that we don’t even know that we don’t know.
One thing that is pretty clear is that an ideal bayesian reasoner would distribute some probability across all non-self-contradictory hypotheses that you can express in text of bounded length. There are only finitely many of them, so a failure to include some of them would be a pretty major departure from idealness.
I agree with lots of what you said, but, to focus:
The main point is that we don’t know anything better, and pretty much everything else that we do know looks worse.
This, I think, is wrong. Not a little bit wrong, but outright wrong, and has been wrong for decades. If you want to believe and repeat this, I can provide some standards I think you should meet before being so confident. I think you should stop claiming it (unless you can answer about why other contenders are wrong, which would be big news philosophically if you could).
That’s a big claim on my part, so I will make a point and then give you some alternatives:
Claim: There is no one who can put forward a specific, integrated version of Bayesianism that will: take responsibility for defending it, put the best and strongest arguments forward, and debate with any philosopher who disagrees.
(Integrated here means it comes with all necessary parts for it to work; e.g., if one needs a model for hypothesis generation, then at the very least there are known fleshed out ways of doing that. Also, debates don’t need to happen twice, we can simply point to an argument made in the ongoing (academic and non-academic) literature and check whether the argument is resolved or not.)
Alternatives:
Critical Rationalism (Karl Popper). CR gave us the basis of modern science (last 100yrs or so) so should be taken as a serious option. Popper tried to engage with bayesians and inductivists but some of his criticisms never received answers.
Critical Fallibilism (Elliot Temple). CF is descended from CR and some other philosophies. It has original and important contributions from Elliot which solve some problems with CR, like eliminating credences completely (Popper tried and failed to get rid of ‘strong’ and ‘weak’ arguments fully, which is one of CF’s criticisms of CR).
Does the fact that CF eliminates credences completely mean that whoever is applying CF in practice never considers credence for anything, never reports credence for anything, never asks others for credence for anything? Does CF recommend that nobody ever considers credence?
It can be useful to communicate about credences with other people who believe in them. But ideally everyone should move on to better approaches and choose between competing ideas for qualitative reasons not quantitative credence differences.
So if we’re not ideal bayesian reasoners, how do we make progress? It seems like things need to come from outside bayesianism somehow. Or is it just a lottery that we can’t ever do better at except by having more people actively do bayesian reasoning about stuff? If there is something outside bayesian reasoning/epistemology and new ideas come from there, then it seems like it does things bayesianism does not, or is better at it. And such a thing would either contradict bayesianism at worst or make it (partially) obsolete at best.
Sorry for the delayed replay. My comments are being held for review.
There aren’t just two possibilities “ideal bayesian reasoning” and “useless rubbish”. There is a huge range of heuristics, ad-hoc models, evolved instincts, and everything else in the mix. These are all ‘outside bayesianism’ and while the collection is almost certainly worse than ideal bayesian reasoning, they are not useless.
That also doesn’t mean that we can only improve by having more people actively do bayesian reasoning about stuff, though there are certainly many cases where people would be better off actively doing bayesian reasoning.
There are many ways to improve incredibly complex systems such as human minds and their interactions. It’s far from certain that applying more bayesian reasoning is the best way. We are definitely not capable of reaching the ideal, and will have to settle for something imperfect. Maybe there is a better approximation than “try bayesian reasoning as far as our limited human brains can handle”, maybe there is not.
The main point is that we don’t know anything better, and pretty much everything else that we do know looks worse. However, there is a lot that we don’t know, and far more that we don’t even know that we don’t know.
One thing that is pretty clear is that an ideal bayesian reasoner would distribute some probability across all non-self-contradictory hypotheses that you can express in text of bounded length. There are only finitely many of them, so a failure to include some of them would be a pretty major departure from idealness.
I agree with lots of what you said, but, to focus:
This, I think, is wrong. Not a little bit wrong, but outright wrong, and has been wrong for decades. If you want to believe and repeat this, I can provide some standards I think you should meet before being so confident. I think you should stop claiming it (unless you can answer about why other contenders are wrong, which would be big news philosophically if you could).
That’s a big claim on my part, so I will make a point and then give you some alternatives:
Claim: There is no one who can put forward a specific, integrated version of Bayesianism that will: take responsibility for defending it, put the best and strongest arguments forward, and debate with any philosopher who disagrees.
(Integrated here means it comes with all necessary parts for it to work; e.g., if one needs a model for hypothesis generation, then at the very least there are known fleshed out ways of doing that. Also, debates don’t need to happen twice, we can simply point to an argument made in the ongoing (academic and non-academic) literature and check whether the argument is resolved or not.)
Alternatives:
Critical Rationalism (Karl Popper). CR gave us the basis of modern science (last 100yrs or so) so should be taken as a serious option. Popper tried to engage with bayesians and inductivists but some of his criticisms never received answers.
Critical Fallibilism (Elliot Temple). CF is descended from CR and some other philosophies. It has original and important contributions from Elliot which solve some problems with CR, like eliminating credences completely (Popper tried and failed to get rid of ‘strong’ and ‘weak’ arguments fully, which is one of CF’s criticisms of CR).
Elliot has been critical of LessWrong for a decade plus FWIW and even though points like the one I made above (not original to me) have been brought up to LW for years, there is little-to-no engagement.
Does the fact that CF eliminates credences completely mean that whoever is applying CF in practice never considers credence for anything, never reports credence for anything, never asks others for credence for anything? Does CF recommend that nobody ever considers credence?
It can be useful to communicate about credences with other people who believe in them. But ideally everyone should move on to better approaches and choose between competing ideas for qualitative reasons not quantitative credence differences.