There needs to be a canonical short clear explanation of "here's why Solomonoff induction is a mathematical curiosity that has no bearing on anything real" that could be pointed at whenever it comes up. I've been tempted to write one, but my queue is already infinite...
-
-
@ProfJayDaigle, might you be implored to write something like this?1 reply 0 retweets 2 likes -
Something like that is in my queue, actually--look at how I end my post on Bayesian inference. I'm not sure I'll land exactly where you would on the issue, and I certainly don't think it will be the canonical anything: I'm an interested amateur here, not an expert.
2 replies 0 retweets 1 like -
Replying to @ProfJayDaigle @Meaningness and
But I do think Solomonoff induction is both theoretically interesting and much less practically useful than some people give it credit for. (Of course, as a mathematician, "theoretically interesting" is enough to make it, well, interesting).
2 replies 0 retweets 1 like -
Yes, I found it interesting enough to take a graduate seminar in it, which was one of best mathematical experiences I’ve had. The math is gorgeous and I leveled up my proof ability during it. (That was in 1982 I think...)
1 reply 0 retweets 1 like -
Replying to @Meaningness @ProfJayDaigle and
For me, there would be two hard parts in writing this. One would be deciding how much detail to go into (= who the audience is, roughly). It’d be hard for me to keep it to a reasonable length and prevent it turning into a textbook.
1 reply 0 retweets 1 like -
Replying to @Meaningness @ProfJayDaigle and
The other is, it would take some research to address the “yeah ok it’s uncomputable but we can approximate it” line. My intuition is there should be a strong theorem saying “no you can’t” but I don’t think that existed in 1982.
1 reply 0 retweets 1 like -
Replying to @Meaningness @ProfJayDaigle and
Chaitin may have such a theorem but I’ve read only summaries of his work and don’t remember seeing exactly that. I’d also feel I had investigate Scmidthuber’s stuff, which I’m pretty sure is wishful thinking but I don’t actually know.
1 reply 0 retweets 1 like -
Replying to @Meaningness @ProfJayDaigle and
If there actually isn’t such a theorem I’d be tempted to try to prove one :(
1 reply 0 retweets 0 likes -
Coincidentally, Google Scholar decided to ping me with a link to this paper just this morning: https://www.gwern.net/docs/rl/2017-leike.pdf … . I only read the abstract but looks relevant.
2 replies 0 retweets 2 likes
Yes indeed, thank you very much!
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.