wedrifid comments on Harry Potter and the Methods of Rationality discussion thread - Less Wrong

34 Post author: Unnamed 27 May 2010 12:10AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (866)

You are viewing a single comment's thread. Show more comments above.

Comment author: red75 30 June 2010 03:28:04AM 0 points [-]

Timeturner awesomeness for jail escape is pretty much discounted, when jailers know that one has it. And Harry risks being put under constant supervision because of his apparent disability of infancy.

Comment author: wedrifid 30 June 2010 09:21:28AM *  1 point [-]

Timeturner awesomeness for jail escape is pretty much discounted, when jailers know that one has it.

It never occurred to me to consider it a literal way to escape from jails. That's nearly useless. Instead, I consider it, among other things, as a nearly universally more effective med. kit. If someone just fell off the roof would you rather be able to bandage them up a bit or go back and tell them to watch their step?

And Harry risks being put under constant supervision because of his apparent disability of infancy.

... a good reason to not do things that are infantile... and when you do slip up you go back and give yourself a scolding so that you never do the infantile thing in the first place (but still give your past self the scolding note).

Comment author: Eliezer_Yudkowsky 30 June 2010 09:48:24AM 7 points [-]

Good heavens, Mr. Wedrifid, you can't change time! Do you think students would be allowed Time-Turners if that was possible? What if someone tried to change their test scores?

Comment author: wedrifid 30 June 2010 12:12:07PM 5 points [-]

Good heavens Mr. Yudkowsky, I thought the inventor of Timeless Decision Theory would have a better grasp on how being the kind of person who would make a certain decision can determine what happens, even when that decision never needs to get made, whether that be with Omega and his boxes or in the stable resolution of time loops.

In all the previous time related events, things worked out how they did because the situation in which Harry did not use the time turner was not stable. If Harry was different (for example, by being Hermione or by not having a Time Turner) then the stable, 'final version' given by the universe-time-loop-processor would be the simple one where he doesn't go mess with stuff. But it wasn't.

But lets say that for some bizarre reason Harry never found a note warning him about a stumbling risk and he didn't think to send one back later. At the very least we should find out a few seconds later that the friend fell off the building and landed on a great big padded mat.

Comment author: Eliezer_Yudkowsky 30 June 2010 07:38:09PM *  9 points [-]

Good heavens Mr. Yudkowsky, I thought the inventor of Timeless Decision Theory would have a better grasp on how being the kind of person who would make a certain decision can determine what happens

I do indeed.

I've written unpublished fiction about it.

From before TDT was invented, actually.

Harry has not worked all that stuff out yet.

He did work out one important principle so far.

It is called DO NOT MESS WITH TIME.

And considering that he got that result, you seem to have missed some of the implications for how time travel works in that universe which would make it potentially dangerous to try and blackmail reality.

Time travel was the first optimization process I considered which was truly alien enough to deanthropomorphize my thinking; evolutionary biology didn't do the job, but the unpublished story I was writing about time travel did.

What you're suggesting is a bit more potentially incredibly dangerous than you seem to think.

Comment author: wedrifid 01 July 2010 01:10:42AM 1 point [-]

What you're suggesting is a bit more potentially incredibly dangerous than you seem to think.

I think you are mistaking me for straw-wedrifid here. I saw the problem with trying to blackmail reality before he went ahead and actually tried it. But then I'm not eleven and while I am arrogant I am not nearly as arrogant as Harry seems to be.

Comment author: Vladimir_Nesov 30 June 2010 02:14:30PM *  3 points [-]

Exactly. To make certain situations impossible, you have to be the sort of person that makes the correct actions in the impossible situations, the actions making those situations impossible. (This is also at the core of bargaining.) You are not to take the money from two boxes in the Open Box Newcomb's problem, even if you clearly see that money is there in both of them (and if you have that property, then the situation will never arise).

Comment author: CronoDAS 30 June 2010 03:02:54PM *  2 points [-]

Harry tried "being the kind of person who would make a certain decision" when using the Time Turner. The result was DO NOT MESS WITH TIME.

Comment author: thomblake 30 June 2010 03:24:11PM 2 points [-]

He was trying to create a stable time loop, which had consequences along the same lines as the Outcome Pump - there's no way to know which stable time loop you'll get.

However, if he was using a "being the kind of person" strategy, we might expect he'd avoid being the sort of person who would pass along "DO NOT MESS WITH TIME".

Comment author: Sniffnoy 30 June 2010 03:38:24PM 2 points [-]

Yes, to repeat what I said earlier, this seems easy to avoid by replacing his "blank paper" condition with a more general "anything other than a pair of numbers in the given range" condition. I have to suppose Eliezer had him get that specific message because it wouldn't be good for the story if Harry noticed this fact. Though even if he does take that approach, as with the outcome pump, there's still other possibilities, because they can screw with Harry's ability to execute his intended algorithm.

Comment author: Vladimir_Nesov 30 June 2010 10:41:10PM 2 points [-]

Though even if he does take that approach, as with the outcome pump, there's still other possibilities, because they can screw with Harry's ability to execute his intended algorithm.

Yes, like it turning out that he was predetermined to die at the time of the experiment, and never complete it.

Comment author: wedrifid 30 June 2010 03:38:58PM *  0 points [-]

He did the same thing every time he used the time turner. I mean every time, even the times when he was being a good boy and using it to manage sleep. the universe doesn't care whether it is a conversation with McGonnagal, juggling bullies and pies, someone falling off the roof or just bed time.

What does matter to the universe is whether the agent in the time loop is interacting with the time loop in a way that is complex and improbable. That is, factoring large primes should give unpredictable outcomes, long detailed tricks like throwing pies and playing with bullies should be slightly safer, giving yourself a time out simpler again and pre-sheduled study and sleep breaks right down at the bottom of the scale.

I argue that all the instance of time turner differ use differ only in degree. There are many things to do with a time turner that are far, far less disruptive, complex or unstable than what Hermione did when attending multiple classes. Given her interaction with other people who would be encountering her other self there are butterfly effects that would need to be resolved by the system. If Harry set up a smart system to communicate with himself unobtrusively things may be simpler to predict. He could send himself SMS messages (when outside Hogwarts) or use one of those coins to send messages.

Comment author: Vladimir_Nesov 30 June 2010 09:58:18PM *  3 points [-]

What does matter to the universe is whether the agent in the time loop is interacting with the time loop in a way that is complex and improbable. That is, factoring large primes should give unpredictable outcomes, long detailed tricks like throwing pies and playing with bullies should be slightly safer, giving yourself a time out simpler again and pre-scheduled study and sleep breaks right down at the bottom of the scale.

This "scale" sounds extremely anthropomorphic.

Comment author: red75 30 June 2010 01:09:01PM 0 points [-]

As Eliezer said already, timeturner can't change the past. One generally can't even calculate probability of desired outcome of timeturning... Ouch. This is discussed already.

Comment author: NancyLebovitz 30 June 2010 08:10:34PM *  1 point [-]

If anyone wants some sf on the subject, I recommend Leiber's The Big Time and "Try and Change the Past".

They're both based on the same premise. The timeline is changeable but highly resistant. Humans can't change it (that's the short story) but there are two superhuman sides (called Snakes and Spiders, but never seen onstage) which recruit humans who are willing to be cut out of their timelines just before they die.

Comment author: Vladimir_Nesov 30 June 2010 02:39:28PM 1 point [-]

One generally can't even calculate probability of desired outcome of timeturning...

Probability is in the mind. It is the thing that is being calculated.

Comment author: red75 30 June 2010 04:17:16PM 0 points [-]

And the thing is "DO NOT MESS WITH TIME".

Comment author: wedrifid 30 June 2010 04:44:23PM 0 points [-]

That is a significant quote from the FanFic but I am having difficulty seeing it as relevant to Vladmir's statement.

Comment author: red75 30 June 2010 05:50:09PM 0 points [-]

I found Vladimir's utterance (I'm not sure of word's connotations, I use it in pragmatics sense) incomprehensible on my current level of understanding his intentions and his ways of expressing toughts. So I've took literal meaning of his words into current context. However, beside joke part my message points on difficulties in dealing with stable states of closed time loops (the thing).

Comment author: wedrifid 30 June 2010 06:04:23PM 1 point [-]

I found Vladimir's utterance (I'm not sure of word's connotations, I use it in pragmatics sense) incomprehensible

This is not an unusual occurrence. V thinks clearly, and thoroughly but presents his conclusions in a way that assumes a similar thinking style and a lot of shared prior knowledge. Petty things like 'intermediate steps' are not necessarily included.

Comment author: wedrifid 30 June 2010 01:19:23PM 0 points [-]

One generally can't even calculate probability of desired outcome of timeturning...

I disagree with this statement. See earlier discussion of 'idiot god' by Richard.

Comment author: red75 30 June 2010 01:53:07PM 0 points [-]

Those closed time loops are weird. I considered timeturing before outcome, but, yes, even in that case one can be told by one's timeturned twin, that all is set up for good outcome. And timeturing after desired outcome... should be done unconditionally, as you can't know it is not you who caused this outcome. Weird.

Comment author: wedrifid 30 June 2010 03:22:11PM 1 point [-]

Definitely weird. A related consideration is that I would always give reasons for any advice I give my former self. That cuts off a large swath of potential stable loops that consist of me giving myself advice for absolutely no reason at all except that it happens to be stable. The better the reasons I have been given myself the less likely it is that the self perpetuating cycle is a completely arbitrary cycle.

For example, I wouldn't have sent back "Don't mess with time". I would have sent "the universe doesn't particularly care about your rules and plans you arrogant little git! What's more likely, guessing your way through 128 bit encryption or something seriously nasty that distracts you from your games, such as <what just happened>? That's right. Think." (Yes, I'd include the 'arrogant git' part. That is information I would clearly need to be reminded of!)

Now, not all scary situations give me the chance to write an explanation but a large swath of the probability mass does. While I would still follow the hastily written directive I would also know that to write that particularly message something really bad must be happening. Without having a predetermined policy for giving details I would have no idea whether the message meant something bad almost happened or not. (It also means that I am far less likely to get such a message - I'll probably get one of the many possible detailed messages.)

Comment author: red75 30 June 2010 05:26:07PM *  0 points [-]

The problem is that you aren't source of advice, you are one of constraints to be satisfied. Any message, that you will reproduce with picometer precision and that will create stable state, will do. Precision isn't a problem in deterministic world, and maybe in quantum one too (if our neurons are sufficiently classical), but I'm hesitant to estimate influence of one's preferences on stable state.

Comment author: wedrifid 30 June 2010 05:45:12PM 0 points [-]

The problem is that you aren't source of advice, you are one of constraints to be satisfied.

I am both. The advice that I will choose to give is determined by the same physics that allows me to breath.

Regarding quantum effects - the uncertainty effects can be amplified based on the elimination of unstable loops. Most obviously when my behavior is determined by a quantum coin. The way that plays out looks seriously when pictured in 4 dimensions.

Comment author: red75 30 June 2010 06:04:39PM 0 points [-]

I am both.

Not necessarily.

In casual loops, information simply exists as a consequence of the peculiar topology of the space-time manifold.

Self-existing objects and auto-generated information in chronology-violating space-times: A philosophical discussion

Comment author: wedrifid 30 June 2010 06:19:29PM 0 points [-]

I read the link and make the same claim I made previously: I am both. The advice that I will choose to give is determined by the same physics that allows me to breath.