Though the ordinary “semi-formalization” of UDASSA seems to rely on Turing machines, which are already mathematical objects. I find the idea that there is some universal Turing machine endlessly running all programs in parallel (because to be clear, many never halt) that exists somewhere in the ether… a little unintuitive, personally, even as an AIT researcher.
Yeah, the observation that the universe seems maybe well-predicted by a program running on some UTM is a subset of the observation that the universe seems amendable to mathematical description and compression. So the former observation isn’t really an explanation for the latter, just a kind of restatement. We’d need an argument for why a prior over random programs running on an UTM should be preferred over a prior over random strings. Why does the universe have structure? The Universal Prior isn’t an answer to that question. It’s just an attempt to write down a sensible prior that takes the observation that the universe is structured and apparently predictable into account.
(I did discuss earlier versions of this argument at least a few years ago, so at least I’m confident I couldn’t have been subconsciously influenced by that thread specifically!)
But it has been discussed here:
https://www.lesswrong.com/posts/3LcyoqNTJuCZ65MbL/mo-putera-s-shortform?commentId=Lif85TC2zJgmQieZH
Yes—I would say this is one of the many paradoxes solved by UDASSA.
Hm, nice rabbit hole.
Though the ordinary “semi-formalization” of UDASSA seems to rely on Turing machines, which are already mathematical objects. I find the idea that there is some universal Turing machine endlessly running all programs in parallel (because to be clear, many never halt) that exists somewhere in the ether… a little unintuitive, personally, even as an AIT researcher.
Yeah, the observation that the universe seems maybe well-predicted by a program running on some UTM is a subset of the observation that the universe seems amendable to mathematical description and compression. So the former observation isn’t really an explanation for the latter, just a kind of restatement. We’d need an argument for why a prior over random programs running on an UTM should be preferred over a prior over random strings. Why does the universe have structure? The Universal Prior isn’t an answer to that question. It’s just an attempt to write down a sensible prior that takes the observation that the universe is structured and apparently predictable into account.
Noted! Will check out the comments later.
(I did discuss earlier versions of this argument at least a few years ago, so at least I’m confident I couldn’t have been subconsciously influenced by that thread specifically!)
(EDIT: Upon a skim they discuss a pretty different set of considerations here, closer to what Hamming calls “You see what you look for”)
It doesn’t seem to be the same as your argument.