su3su2u1-deactivated20160226 asked: I wrote a post about computability/time travel/HPMOR (its titled something like chapter 14). If you get a chance, let me know if you spot anything obviously wrong there. I'm pretty sure I'm right, but I'd hate to mislead people.

su3su2u1:

hot-gay-rationalist:

nostalgebraist:

I didn’t see anything wrong in the post (which, for others’ reference, here).

I guess you could be wrong in your criticism of Harry’s statement about “Turing computability,” but only because I have no idea what I means by that statement.

Like, it seems obvious that physicists talk about GR possibly allowing closed timelike curves and this doesn’t make the equations of GR any less capable of being approximated on a Turing machine, so the idea “a computer couldn’t simulate this” seems clearly wrong?  (There might be more than one self-consistent possibility for what happens on the CTC, but at worst that would be indeterminacy that means you’d need more initial data to uniquely solve the equations, which isn’t really a computability issue?)

On the other hand, maybe he’s assuming that the simulation is running in polynomial time (in some reasonable sense) in the external-to-simulation universe, and that by doing too much time travel you could slow it down a lot by forcing it to do NP computations in order to find a self-consistent state (the famous “you can do NP-complete computations with a time machine” thing)?  But why would that be a problem?  Even if you did something that made it took a million times longer (in the external universe) to compute more of the internal universe, you wouldn’t be able to notice that from the internal universe.  (EY is a huge fan of Permutation City so this should be a familiar point to him)

I can’t think of any possibly meaning for Harry’s statement that makes it true, but I’m so unsure about what it’s supposed to mean that I don’t feel like i can confidently declare it false.

At any rate, CTCs are not straightforwardly turing-computable: you have to do the bruteforce thing of searching for all self-consistent universes. Maybe that’s what he meant.

snip 

But I think P = NP is pretty obvious? The example he tried with the prime factorisation thing for instance. Get any NP-complete problem, find a well-order of the possible solutions, then start bruteforcing each one. Do so for 5h50min. Then, if you still haven’t found a solution, send it back to past!you. Past!you starts from where you left off, bruteforcing for 5h50min, and so on. The only stable time loop is the one where you received the actual solution from the future, since we can verify that a given solution is the correct one for an NP-complete problem in polynomial time.

Or, you know, maybe you receive a note with “DO NOT MESS WITH TIME” written in your shaky handwriting from the future. That can work too.

To the first, if closed time like curves aren’t Turing computable than GR isn’t, the Godel metric is a solution of GR with CTCs in it, and its totally classical.  Even doing inserts of all possible Harry/AntiHarry pairs is something an non-deterministic Turing machine should be able to do, and if you can map it on to quantum mechanics (see my post), then it should probably be in something like class  BQP (bounded error quantum polynomial time).  Either way, anything in BQP or NP can be done on a deterministic turing machine, it just takes a lot of time.  

Regarding P = NP with the time-turner- he gives the rules of the time turner as only allowing you to go back 6 hours in a day and only 3 operations per-day max.  So for ever 24 hours, you get 30 hours of time total to work, which is approximately just mapping t in a scaling law to 24/30*t (approximately because the scaling is discontinuous).  

Its possible I’ve misunderstood the rule (I’m only up to this chapter), and you can just turn time back 6 hours every 6 hours, in which case, yes, you can effectively create an infinite amount of time for yourself, so all algorithms take 0 time. Well, any algorithm that can be done in a wizard’s lifetime.  After that, normal scaling will resume. 

Edit: In this case, then hypercomputation (non-computable stuff) IS possible if you have an infinite amount of memory.  Although in effect you are limited to a wizard’s lifetime I guess. 

I’m really tired and could be wrong, but I think you’re not getting the P = NP thing – the idea isn’t that you get extra time, the idea is that you set a rule for yourself about what you will do at the end of the loop, and the only self-consistent “story” involving following that rule is the one where you get the answer to the problem.

(If you hadn’t found the answer by the time limit, you’d send something different back than what you actually got, but that isn’t self-consistent.  So the only self-consistent possibility is one where you do find the answer by the time limit.)

What you end up with is a way to get the answer immediately to any computation you can check in 6 hours, which could be a problem in NP whose solution would take much more than 6 hours to find.

I think this is basically the same idea Scott Aaronson talks about in the time travel chapter of Quantum Computing Since Democritus (although the one he’s talking about is something from David Deutsch involving probabilistic decisions, but the idea is the same (the only self-consistent loop is the one where you get the answer out at the end).

Anonymous asked: I can't find the author notes anymore either... do you remember where they were originally stored?

All I know is that there’s an archive here, but it only goes back to Chapter 18.

su3su2u1-deactivated20160226 asked: I wrote a post about computability/time travel/HPMOR (its titled something like chapter 14). If you get a chance, let me know if you spot anything obviously wrong there. I'm pretty sure I'm right, but I'd hate to mislead people.

hot-gay-rationalist:

nostalgebraist:

I didn’t see anything wrong in the post (which, for others’ reference, here).

I guess you could be wrong in your criticism of Harry’s statement about “Turing computability,” but only because I have no idea what I means by that statement.

Like, it seems obvious that physicists talk about GR possibly allowing closed timelike curves and this doesn’t make the equations of GR any less capable of being approximated on a Turing machine, so the idea “a computer couldn’t simulate this” seems clearly wrong?  (There might be more than one self-consistent possibility for what happens on the CTC, but at worst that would be indeterminacy that means you’d need more initial data to uniquely solve the equations, which isn’t really a computability issue?)

On the other hand, maybe he’s assuming that the simulation is running in polynomial time (in some reasonable sense) in the external-to-simulation universe, and that by doing too much time travel you could slow it down a lot by forcing it to do NP computations in order to find a self-consistent state (the famous “you can do NP-complete computations with a time machine” thing)?  But why would that be a problem?  Even if you did something that made it took a million times longer (in the external universe) to compute more of the internal universe, you wouldn’t be able to notice that from the internal universe.  (EY is a huge fan of Permutation City so this should be a familiar point to him)

I can’t think of any possibly meaning for Harry’s statement that makes it true, but I’m so unsure about what it’s supposed to mean that I don’t feel like i can confidently declare it false.

Iirc Eliezer said he didn’t really explicitly think of the NP/bruteforce thing wrt time-travel until after writing that chapter. At any rate, CTCs are not straightforwardly turing-computable: you have to do the bruteforce thing of searching for all self-consistent universes. Maybe that’s what he meant.

But I think P = NP is pretty obvious? The example he tried with the prime factorisation thing for instance. Get any NP-complete problem, find a well-order of the possible solutions, then start bruteforcing each one. Do so for 5h50min. Then, if you still haven’t found a solution, send it back to past!you. Past!you starts from where you left off, bruteforcing for 5h50min, and so on. The only stable time loop is the one where you received the actual solution from the future, since we can verify that a given solution is the correct one for an NP-complete problem in polynomial time.

Or, you know, maybe you receive a note with “DO NOT MESS WITH TIME” written in your shaky handwriting from the future. That can work too.

I’m confused by “CTCs are not straightforwardly turing-computable: you have to do the bruteforce thing of searching for all self-consistent universes.”  Words like “brute force” are computational complexity words to me, but “Turing computable” isn’t.  As long as a Turing machine can do something, that thing is Turing computable, even if it has to do it by brute force (and thus takes a lot of time or space or whatever).

Are you saying that Harry is talking about computational complexity (and isn’t using the term “Turing computable” in the usual sense), or that he is talking about Turing computability in the usual sense and there is some issue with that and CTCs?

Anonymous asked: Isn't this the chapter in which he explicitly says that Harry is probably wrong in the author notes?

I don’t know – the archives of old HPMOR Author’s Notes that I’ve found only go back to Chapter 18 or so.  Do you have a link to the place where he said this?

su3su2u1-deactivated20160226 asked: I wrote a post about computability/time travel/HPMOR (its titled something like chapter 14). If you get a chance, let me know if you spot anything obviously wrong there. I'm pretty sure I'm right, but I'd hate to mislead people.

I didn’t see anything wrong in the post (which, for others’ reference, here).

I guess you could be wrong in your criticism of Harry’s statement about “Turing computability,” but only because I have no idea what I means by that statement.

Like, it seems obvious that physicists talk about GR possibly allowing closed timelike curves and this doesn’t make the equations of GR any less capable of being approximated on a Turing machine, so the idea “a computer couldn’t simulate this” seems clearly wrong?  (There might be more than one self-consistent possibility for what happens on the CTC, but at worst that would be indeterminacy that means you’d need more initial data to uniquely solve the equations, which isn’t really a computability issue?)

On the other hand, maybe he’s assuming that the simulation is running in polynomial time (in some reasonable sense) in the external-to-simulation universe, and that by doing too much time travel you could slow it down a lot by forcing it to do NP computations in order to find a self-consistent state (the famous “you can do NP-complete computations with a time machine” thing)?  But why would that be a problem?  Even if you did something that made it took a million times longer (in the external universe) to compute more of the internal universe, you wouldn’t be able to notice that from the internal universe.  (EY is a huge fan of Permutation City so this should be a familiar point to him)

I can’t think of any possibly meaning for Harry’s statement that makes it true, but I’m so unsure about what it’s supposed to mean that I don’t feel like i can confidently declare it false.

Anonymous asked: Def. agree with your latest LessWrong post. One of the reasons I tend to rail so hard against Big Yud (and Mike Darwin/cryonics in other contexts) is that I feel like they are taking advantage of people I find to be largely decent, good natured people. But I think you are off when you say the rationalist community sends off the 'wrong' social signals- its about the people they want to bring in-group. Lots of people DO respond to those signals.

I agree with you about that.  I addressed that among other things in this long response to anononthewater (warning for everyone: please read the first paragraph of that post before deciding if you want to read the rest).

anononthewater-deactivated20140 asked: re: last post about lesswrong: I think part of her power came from being so hardcore, tho. Like, bland niceness gives you a handful of followers, getting your academic radical qrrr theory to the point where cis men should all stop being cis will alienate 90% and make 10% devoted? like, her signals were offputting to passers-by too, in a way you tend to speculate about the community-building powers of strangeness over at LW.

(Under the cut for more pontificating, this time involving many more specifics about myc.  I know there are several reasons some people might not want to read this; it’s under a cut for a reason.  TBH I feel kind of weird about talking stuff about this from the somewhat outside perspective I have on it, and also I have had a bit to drink and may not be as capable of leaving well enough alone as I should be.  But for better or for worse it’s on my mind, so here we go.)

Keep reading

blurds asked: Less Wrong: Not Socially Adept Enough To Manipulate You (except that Yud clearly knows all the right words to say to -his- chosen audience)

Pretty much, though my larger point was “they also seem to have nothing to hide by manipulating you (with, as always, the possible exception of Yud)”

beforeness asked: I don't know how to phrase it, but your post about LW, even though I broadly agree with it, kind of irked me, I'm not sure how the surface signal stuff really plays into all of it, when the reason the LW community is so good has nothing to do with that, and everything to do with the quality of the people there - surely if LW doesn't have these kinds of things happen that's all that matters, since it has its own language which hypothetically would could be used by manipulative people in same way?

(Cut for the same reason the last one was cut)

Keep reading