Worth noting that this was written before Eliezer's "Toolbox-thinking and Law-thinking"; several LW commenters seem to feel that that the T-T & L-T post addresses the criticisms in this post (based on the comments at https://www.lesswrong.com/posts/gsQjde3qeZw36arYE/nostalgebraist-bayes-a-kinda-sorta-masterpost … ).
“If your method requires infinite storage space and computation speed to use, the problem is not that finite beings aren’t sufficiently rational, it’s that your method doesn’t work.”
(@nostalgebraist re Bayes, but applies to many rationalist fantasies.)http://nostalgebraist.tumblr.com/post/161645122124/bayes-a-kinda-sorta-masterpost …
-
-
-
It doesn’t seem to me that EY’s post addresses the issues at all. But, after much effort, I don’t understand his point, so I might be missing something.
-
Afaict, EY is saying “PT is a set of axioms. If you accept the axioms, then you have to accept their mathematical consequences.” Which no one disputes. The question is, where do the axioms apply? Which he doesn’t seem to address at all.
End of conversation
New conversation -
-
-
It just occured to me that rationalisms are kind of deontological epistemologies. Not only the business of trying to reduce cognition to adherence to rules but also emotional dynamics seems similar. Does it make sense to you?
-
Hmm, I’m not sure… say more, maybe? Bayesian rationalism is similar to utilitarianism, formally. I can’t make any sense of critical rationalism, but maybe it’s similar to virtue ethics?
-
I meant that it reduces coginition to following certain rules, in the way deontology reduces morality to following rules. But maybe the analogies with utilitarianism are stronger. Anyway, self-flagelation for falling short of an impossible and impractical ideal ensues.
-
Yes, definitely that!
End of conversation
New conversation -
-
-
I’m probably missing the point here but there are plenty of puzzles with solutions that are unsolvable regardless of endless compiting power , see Yanofskyhttps://mitpress.mit.edu/books/outer-limits-reason …
- 3 more replies
New conversation -
-
-
I think I've put my finger on a definitive refutation of Bayesian philosophy!
@DavidDeutschOxf@seanmcarroll While *physical* reductionism might be true, *theory* reductionism (trying to reduce high-level explanations to lower-level ones) is false! See: https://www.scottaaronson.com/blog/?p=3943#comments … -
The ladder of abstraction for computer science seems to be ‘Probability&Stats’ (most concrete) -> ‘Computational Complexity’ (middle-level) and ‘Computational Logic’ (most abstract). But theory reductionism is false! So the higher-level domains aren't reducible to probabilities!
-
The art of reasoning is not arranged in a hierarchy like physics is! Rather, it's a network of coherent methods. No single foundation at the bottom! Set theory is *not* the foundation of pure math! And nor is probability theory the foundation of applied math!
-
Should it be used as the basis of the scientific principle? No, just ask Judea Pearl. Should it be used as the basis of logic? No, just ask David Chapman. Should it be used for the basis human cognition? No: https://medium.com/intuitionmachine/why-probability-theory-should-be-thrown-under-the-bus-36e5d69a34c9 … . These are all different aspects though.
-
Ha, well I wouldn't go so far as to say 'throw it under the bus'. Probability&Stats is a valid component of computer science. But it's just one component.
-
I'm referring to its use as being the underlying mechanism for explaining human cognition. It is of course a valid tool for analyzing data .
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.