TylerJay

• TylerJayMar 17, 2015, 7:47 PM

  5 points

  in reply to: buybuydandavis’s comment on: [FINAL CHAPTER] Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 122

  And just a few lines before your last quote:

  Quirrell:

  “I did not have any friends like that when I was young.” Still the same emotionless voice. “What would have become of you, I wonder, if you had been alone?”

• TylerJayMar 17, 2015, 7:43 PM

  7 points

  in reply to: JenniferRM’s comment on: [FINAL CHAPTER] Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 122

  What I love about this twist is how it changes the interpretation of so many other things that were said throughout the story. For example:

  “Purposeless?” said Professor Quirrell. “Oh, but the madness of Dumbledore is not that he is purposeless, but that he has too many purposes.

  It turns out PQ was right in that the madness of Dumbledore was not purposeless, however much his going around and “snipping all the threads of destiny” to constrain future events would, to anyone without all the knowledge of prophecy, look like many divergent purposes. Even Dumbledore himself didn’t know how or why some of them fit into the whole picture. But it was all done in service of his one true goal. And if the service of that goal had involved killing Harry or framing Hermione? Well,

  “Who knows what the Headmaster thinks he has reason to do, when he has found reason to do so many strange things already.”

• TylerJayMar 15, 2015, 9:09 AM

  2 points

  on: [FINAL CHAPTER] Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 122

  I couldn’t be happier with the ending. So perfect.

  “I think that you always were, from the day I met you, my mysterious old wizard.”

  Thank you so much Eliezer. It’s been an amazing journey.

• TylerJayMar 15, 2015, 8:48 AM

  5 points

  on: Brag­ging Thread March 2015

  I derived Bayes’ Theorem and the basic rule for conditional probabilities today while trying to find the answer to a question.

  I had seen Bayes’ Theorem before, but never really understood it and certainly had never memorized it. The only relevant knowledge I had was:

  • That the syntax for “probability of A given B” is p(A|B)

  • That you can multiply independent probabilities to do AND, but aren’t allowed to do that if they are dependent

  I was surprised at how it followed directly from intuition and the last bullet point above. I put together a toy problem involving the probability that I go outside depending on if it’s raining or not and was able to derive Bayes’ Theorem essentially from first principles. Not a groundbreaking accomplishment or anything, but I know I’ll never forget any of it now!

  The line of reasoning was really simple once I was able to distill it:

  1. I can simulate the probability that I’m outside given that it’s raining by first rolling a die to see if it’s raining, and if it is, then rolling another die to see if I’m outside, since I already know it’s raining.

  2. These are clearly independent events, so I should be able to multiply their probabilities to get the probability of their conjunction.

  3. What does the conjunction mean? Intuitively, I’m outside and it’s raining exactly when each of those die rolls turned up True, so this multiplication must give the probability that I’m outside and it’s raining: p(outside|raining) * p(raining) = p(outside AND raining)

  4. And if that works, then symmetrically, this same probability p(outside AND raining) should also be equal to the probability that it’s raining given that I’m outside times the probability that I’m outside.

  5. Isolate p(outside|raining) on one side of the equation and Boom: Bayes’ Theorem.

  p(A|B) p(B) = p(A & B) = p(B|A) p(A)

  p(A|B) p(B) = p(B|A) p(A)

  p(A|B) = [p(B|A) * p(A)] /​ p(B)

• TylerJayMar 15, 2015, 7:52 AM

  0 points

  in reply to: Viliam_Bur’s comment on: Brag­ging Thread March 2015

  Congratulations!

  Out of curiosity: What do you think of Czech healthcare? I got appendicitis while visiting the Czech Republic and had to have my appendix out while there in a hospital that was built in the 1300s.

• TylerJayMar 9, 2015, 7:22 AM

  0 points

  in reply to: TobyBartels’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  Haha, yup, I gotcha. Thanks for the info.

• TylerJayMar 8, 2015, 1:47 AM

  3 points

  in reply to: johnswentworth’s comment on: False ther­mo­dy­namic miracles

  Just remember that this isn’t a boxing setup. This is just a way of seeing what an AI will do under a false belief. From what I can tell, the concerns you brought up about it trying to get out isn’t any different between the scenario when we simulate C3PO* and when we simulate C3PO. The problem of making a simulation indistinguishable from reality is a separate issue.

• TylerJayMar 7, 2015, 4:16 AM

  2 points

  in reply to: johnswentworth’s comment on: False ther­mo­dy­namic miracles

  Here’s my explanation of it. Let me know if this helps with your concerns at all:

  Imagine we have an AI design we want to test. Call this AI C3PO, and let its utility function be U(A) where A is a world-state from the set of all possible world-states. And let the super-unlikely-event-happening-at-the-specified-time described in the post be w such that w = true if it happens and w = false if it doesn’t happen. Then let A be a world state in the subset of all world-states A in which w = true. Basically, A is A given that w happened (this is how we simulate a “false belief” by only allowing the AI to consider worlds in which w = true). Finally, let C be a constant.

  The proposal is that we create a variant of C3PO, C3PO* that has the utility function:

  U*(A) = P(!w) * C + P(w) * (U(A*))
  

  If the AI is boxed such that it cannot affect the probability of w occurring and it cannot modify its own utility function, then maximizing U is exactly the same as maximizing U once event w has occurred (ie. with false belief w). In this way, we are able to perfectly simulate C3P0 to find out what it would do if w were true, but we don’t actually have to convince it that w* is true.

• TylerJayMar 7, 2015, 3:38 AM

  4 points

  in reply to: Stuart_Armstrong’s comment on: False ther­mo­dy­namic miracles

  Thanks. I understand now. Just needed to sleep on it, and today, your explanation makes sense.

  Basically, the AI’s actions don’t matter if the unlikely event doesn’t happen, so it will take whatever actions would maximize its utility if the event did happen. This maximizes expected utility

  Maximizing [P(no TM) C + P(TM) u(TM, A))] is the same as maximizing u(A) under assumption TM.

  What links here?

  • False ther­mo­dy­namic miracles by Stuart_Armstrong (Mar 5, 2015, 5:04 PM; 19 points)

• TylerJayMar 7, 2015, 3:21 AM

  0 points

  in reply to: TobyBartels’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  Nope, you’re right. It’s not definitive. In my original comment, I just said I thought I remembered reading somewhere that you couldn’t fit >30 hrs into a day, and the passage I quoted is where I got that impression. If /​r/​hpmor thinks it’s possible that TTs let you fit up to 48 hrs in a day, then I have high confidence there wasn’t anything explicitly forbidding it in the story.

• TylerJayMar 6, 2015, 12:01 AM

  1 point

  in reply to: TobyBartels’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  Ch 17:

  Harry:

  ″...Sorry to ask but I was wondering, is it possible to get more than six hours if you use more than one Time-Turner? Because it’s pretty impressive if you’re doing all that on just thirty hours a day.”

  Dumbledore:

  “I’m afraid Time doesn’t like being stretched out too much,” said Dumbledore after the slight pause, “and yet we ourselves seem to be a little too large for it, and so it’s a constant struggle to fit our lives into Time.”

• TylerJayMar 5, 2015, 9:20 PM

  1 point

  on: HPMOR Wrap Par­ties: Re­sources, In­for­ma­tion and Discussion

  Anyone interested in a Central/​Southern California wrap-party? Even if it’s pretty informal?

HPMOR Wrap Par­ties: Re­sources, In­for­ma­tion and Dis­cus­sion [link to Main post]

• TylerJayMar 5, 2015, 9:12 PM

  1 point

  on: HPMOR Wrap Par­ties: Re­sources, In­for­ma­tion and Discussion

  Just wanted to say that I almost missed this because I only check Main every other month or so. There are a lot of people who only really browse Discussion, and that’s where pretty much all of the HPMOR discussion goes on anyway on LW. Can you x-post? I’m going to make a discussion post linking this Main post just for visibility. If you x-post it, I can delete it, just let me know.

  Thanks for organizing this!

• TylerJayMar 5, 2015, 9:03 PM

  7 points

  on: False ther­mo­dy­namic miracles

  This is really interesting. I thought I understood it and I wanted to verify that by trying to summarize it (and maybe help others too) but now I’m not so sure…

  Edit: Just to save anybody the reading time, my reasoning is false below. After sleeping on it, I see my mistake. Nothing below the “False Thermodynamic Miracles” subheading made sense to me yesterday because I thought the purpose of the setup was to have an “off switch” on the simulated AI under the false belief (letting it see the result of the signal after some time period). I get it now though. Maximizing “[P(no miracle) C] + [P(miracle) u(A) given miracle]” is the same as maximizing “u(A) given miracle”. So the AI will act as if the miracle happened, because there’s no cost associated with those actions if it didn’t happen—only benefits if it did.

  As I understand it, the basic premise is that the AI has a utility function u(w,A) where “w” is whether or not the specific thermodynamic miracle (TM) occurred and A represents a regular/​normal input to a utility function (state of the universe, action taken, or whatever).

  u(w,A) = C (a low constant) if w is false (TM did not happen), and u(w,A) is essentially “u(A) given w” when w is true (ie. it’s a “normal” kind of utility function, dominated by things other than the TM, though it still has to incorporate the fact that the TM happened into its model of the world).

  So, the signal is sent, and it calculates its expected utility for either scenario. If w = false, u(w,A) = a low constant, and if w = true, it determines an action A that maximizes the utility function u(A)|w. As long as 2^-100 u( A ) > C, it starts taking action A.

  Is this right so far?

  Then, after “ON” has gone through, it must adapt itself and backtrack on its previous behaviour.

  This mechanism seems underspecified to me (though maybe it’s obvious to others). Why would it backtrack (or what do you mean by backtrack)? Eventually, it observes that w = false (that “ON” went through unchanged) and that its actions are no longer beneficial, so it just stops doing anything, right? The process terminates or it goes to standby?

  If we want the AI to actually backtrack, wouldn’t we need to modify the utility function so that it takes a configuration parameter as well? u(w,A,X) s.t. X = true iff the AI is in its initial configuration, and set it so that u(F,A,F) = 0, but u(F,A,T) = C ? (this probably counts as just an implementation detail, but I figured I’d mention it anyway)

  The part that makes me feel like I didn’t actually understand it at all is:

  However, the only possible world that it cares about, now, is one in which the “ON” signal was overwritten by a thermodynamic miracle… into exactly the same “ON” signal. As long as the possibility of this is non-zero, the AI will behave as if it’s certain.

  Can you explain this a bit more? Do you mean that even after it observes “ON”, it doesn’t stop acting as if w = true? That P(TM that overwrites ON with ON) * u(A)|w > C ? If that’s the case, then it would never backtrack, right? So it’s essentially a full simulation of an AI under the assumption w, but with the knowledge that w is incredibly unlikely, and no built-in halting condition?

  Thanks

• TylerJayMar 5, 2015, 7:33 PM

  0 points

  in reply to: Velorien’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  Thought so. Seems like a pretty close call to me. Thanks.

  Although, really, it seems a bit contrived. If QQ was identified to the wards as the defense professor, wouldn’t that be what the Hogwarts security system sees?

• TylerJayMar 5, 2015, 8:33 AM

  3 points

  in reply to: Rob Bensinger’s comment on: An­nounce­ment: The Se­quences eBook will be re­leased in mid-March

  I don’t know much about copyright law, but I still don’t understand whether it’s okay to use this in a classroom setting. If it’s a free course or nonprofit, it seems it’s allowed, but what about:

  • A paid course offered by a for-profit institution where the reading material is linked to or provided for no additional (explicit) fee?

  • A course where excerpts are part of a reader that costs money?

  I’ve been wondering this in general, not just related to this particular e-book, so if you know the answers, I’d appreciate it. Thanks!

• TylerJayMar 5, 2015, 8:26 AM

  7 points

  in reply to: Benquo’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  And Monroe was in Slytherin. That was a piece of intentionally leaked information so that the “smart” people could deduce that he was Monroe.

  What links here?

  • TobyBartels's comment on Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116 by Gondolinian (Mar 5, 2015, 11:45 PM; 0 points)

• TylerJayMar 5, 2015, 8:21 AM

  4 points

  in reply to: Benquo’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  I’ve been wondering about this for a while, but never had a chance to do the required detective work. Were Harry and the Defense Professor in the castle at the time? If so, when Dumbles said “Find Tom Riddle...” he most definitely learned that HP and QQ were both Tom Riddles.

  So… were they in the castle at the time? Or was QQ at St. Mungo’s or the DMLE?

• TylerJayMar 5, 2015, 8:17 AM

  2 points

  in reply to: William_Quixote’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, March 2015, chap­ter 116

  I see lots of people doing that on /​r/​hpmor and I’ve never been confused by it, strangely enough. It’s not just you. You probably absorbed it unconsciously.