Aumann agreement theorem

Fragment of a discussion from User talk:Armondikov
Jump to navigation Jump to search

It's basically the same way as saying that if you take a calculator and type 2 + 2 you'll get 4, and if you take a different calculator that works the same way you'll also get 4. It's a methodology thing because in principle, if you get the right evidence communicated, Bayes' Theorem should converge no matter what the priors (i.e., previous beliefs) actually are - providing they're not 100.00%.

Scarlet A.pngnarchist18:24, 27 June 2012

Except it's nonsensical. As rational as we might be, humans are not mathmatical robots, and "rationality" is not math. Nor is it formal logic. It's a weighing of ideas, sure, but it can never be something every person shares. In fact, the way our brains work, you could yourself face the exact same puzzle 4 different times, and get different answers. Simply because that is how brains work.

Green mowse.pngGodotFire! Fire! Fire! (please send spare firefighters)04:09, 28 June 2012

Only if there are at least four different ways to put the puzzle together.

Mjollnir.svgListenerXTalkerX04:13, 28 June 2012

Generally, there are far more than 4 ways to put a puzzle together. The fact that the act of putting it together informs the final understanding or result, you will always be drawn into a world where nothing is "predictable", no matter who shares what information.

Rationality is not math. there is no one answer to a problem even if people had the exact same set of facts. You are still different human beings, with different brains and different synapses.

Green mowse.pngGodotFire! Fire! Fire! (please send spare firefighters)05:18, 28 June 2012

True, outside of mathematics most problems have more than one solution. One exception is jigsaw puzzles, which have only one solution, although there are many different ways to reach it.

Mjollnir.svgListenerXTalkerX05:33, 28 June 2012

It also depends on what you mean by "solution". the solution is also how it's solved, for many of us. the process. If you do the sides first, your experience is different and therefore your solution is in a tangable way, different. If you are going to be literal about a "puzzle" and a solution.

Math problems (usually) have only one solution, but math is an artificial construction we invented to define and describe our world. teh solution is actually contained within the definitions.

I know next to nothing about chaos theory, other than the new agey take, but it strikes me that theory would prove me wrong, if it were proven right, cause then there really is only one way everything can be. I think....

Green mowse.pngGodotFire! Fire! Fire! (please send spare firefighters)05:36, 28 June 2012

Oh. I thought we were talking about conclusions, since that is what the theorem covers. Obviously processes go all over the place, even in basic arithmetic.

Chaos theory is more about predictions — that you cannot predict the behavior of certain systems even if there is no randomness involved.

Mjollnir.svgListenerXTalkerX05:43, 28 June 2012
 
 

Still, I think the general concept that "agreeing to disagree" is a pitiful cop-out. You wouldn't have the conversation "I think gay people should be allowed to marry" "Well I think that's against God's law so they shouldn't" "Okay, that's your opinion, I respect that, let's agree to disagree". No. One of those is just wrong.

This is why I dislike debate. Two sides shouting at each other. How productive. Contrast with science where you have one side: science. And you have one solution: reality. Can you apply that method to "non science" areas (What's the best tax level? Should we have a death penalty?)? It's not easy, but neither should it be considered impossible and dismissed as even worthwhile trying. Sometimes it produces non-intuitive answers, such as when Yudkowsky said that from a utility perspective it would be preferable to torture one person for 50 years over 3^^^3 people getting dust in their eye - because it causes less net suffering overall. It's counter-intuitive compared to how we've evolved to think, but it's the result of a method.

Now, there are sometimes more than one solution. I know in modelling kinetic data with more than two processes this happens a lot. But you can compare the solutions in other ways, focusing on the "right" one. You don't just "agree to disagree" because two people used the same modelling method from different starting guesses and got two different answers with similar R2 - you bang on it again and again until you're right.

Scarlet A.pngmoral08:03, 28 June 2012

When it comes to disagreements over factual questions, one side is always "just wrong," and this can, in theory, be demonstrated; but the only way to demonstrate that an ethical proposition is "just wrong" is to prove that it is logically inconsistent.

Mjollnir.svgListenerXTalkerX16:12, 28 June 2012

If it was just a case of demonstrating it, that would be trivial. Even for an ethical position. The problem is convincing the other side, because that's where prior biases (fucking shoot me I am turning into EY) come in and make it more work than it should be. The flip side to that, of course, is that if your prior biases are in favour, it'll take far less work than it should to convince someone of it.

Scarlet A.pngpostate17:46, 28 June 2012

So I'm curious, why are you all so convinced that a "real" or "true" or "factual" point can be made in something like ethics or philosophy. (or am i misreading you). Just because you think something, and you are "rational" doesn't mean your something is logically correct, especially at the cost of other somethings.

Green mowse.pngGodotFire! Fire! Fire! (please send spare firefighters)17:59, 28 June 2012

You are definitely misreading me; I think that for any questions outside of the factual/naturalistic sphere, you just have to pick something to believe (preferably something that is logically consistent) and roll with it. These sorts of beliefs are categorically asserted without evidence and are dismissable without evidence.

Mjollnir.svgListenerXTalkerX18:08, 28 June 2012
 

You're getting into apples and oranges here. Unless you're going to pull a Sam Harris and argue that moral propositions are "facts" like physical law.

Nebuchadnezzar (talk)18:06, 28 June 2012

No, I didn't suggest you can treat moral judgements as "facts" - although you can demonstrate their consequences in a factual manner ("Gay marriage will lead to bestiality!!" - "Well, just watch it not do that."). But demonstrating something to be factually inaccurate, or demonstrating an ethical position to be badly thought out is fairly trivial next to convincing the believer that it's so.

Scarlet A.pngnarchist18:10, 28 June 2012
 
 
 

Why do you refer to science? Didn't you read Yudkowsky's anti-science rants? You are supposed to refer to Yudkowsky's interpretation of Bayes' theorem.

Also, you can't assume that utilitarianism is the "true" moral system, or the only possible framework to look at moral issues through. Hume's law can't be surpassed. No, not even by Yudkowsky. There is no correct moral theory. Certain utilitarian applications come across as rather horrible to the majority of us, so we choose not to apply them.

Baloney Detection (talk)23:26, 14 July 2012
 
 
 

It works for simple problems, but once you get into really complex questions, it starts to break down.

Nebuchadnezzar (talk)04:30, 28 June 2012
 

Don't expect too much from Armondikov on this. Clever as he may be in other circumstances, he has been drinking too much Yudkowskian kool-aid to get that the evidence can sometimes be tricky and not easily interpreted in a simple formula. It would be so much better if everyone (especially those stupid AI researchers and physicists who don't see what a genius Yudkowsky is) just agreed with whatever Yudkowsky says. Oh, and if you don't donate to the Singularity Institute even though you have heard of it, the singularity might torture you forever!

Baloney Detection (talk)23:01, 14 July 2012

You're a fucking idiot.

Scarlet A.pngpostate14:40, 15 July 2012

Tsk, tsk. Roko's basilisk gave you nightmares again?

Baloney Detection (talk)18:52, 15 July 2012

Yes, I can't sleep because I think a computer will eat me. You have it spot on exactly.

Scarlet A.pngnarchist19:00, 15 July 2012

I knew it!

Btw, your guru is still wrong about morality. There is no objective morality. It is fallacious to assume that utilitarianism is somehow the correct moral system.

Baloney Detection (talk)19:06, 15 July 2012

If you do insist on saying stupid things, I will treat you with sarcasm and contempt, you know this, right?

Scarlet A.pngtheist19:08, 15 July 2012

I will keep calling you out for being an apologist for a pseudoscience-promoting crank. I thought referring to Roko's basilisk would get you thinking about just where the stupidity can be found, but apparently not.

Baloney Detection (talk)19:12, 15 July 2012

You and the Cult of Yudkowsky forming on the LW talk page are setting up a very difficult false dichotomy to work with. To you, the world is split in two, and only two, starkly contrasted categories: you are either with us, or you are with the terrorists. You're labelling those who don't seem to fall in to the category of uncritically and unquestioningly despising Yudkowsky, anyone who signs up to LW - which is, after all, a website composed of individuals - and everything they say without regard to what it is that they say, as "apologists".

Please, grow up.

Scarlet A.pngpathetic19:31, 15 July 2012

Hahaha ok then, what is the accepted amount of criticism of LW? You show me. Heck, my latest thread on the talk page was a possible defense of LW, or at least an opportunity to clarify it. Face it, LW pushes plenty of fringe and crank ideas, and these are absolutely fair game, whatever you might think. Their belief that LW/SIAI is saving the world is just silly, as even someone there recognized.

Your reference to "individuals" ignores that the LW community has certain memes shared by most of the active LWers. It's akin to Thatcher's claim that society doesn't exist, only individuals and families do. Libertarians have a well-known tendency to ignore that individuals live in an environment called society.

And no, I don't despise "anyone who signs up to LW". I don't despise David Gerard for instance. Heck, I don't even despise Luke Muehlhauser, he can't help to have been raised a fundamentalist Christian and being unable to reject the basic way of thinking that comes with it.

Baloney Detection (talk)21:39, 15 July 2012

You mistake me for someone who gives more than a picofuck.

Scarlet A.pngpostate22:07, 15 July 2012

I can't resist commenting on this anymore. This whole week-long discussion has been a deranged laugh riot. It is utterly hilarious that you and David Gerard are being tarred as LW apologists. You two have incredibly well-thought-out and nuanced opinions on the LW source material, but it's all for naught, because apparently any defense of LW means that you're the William Lane Craig of Singularitarianism. I applaud your patience - I would have ragequit days ago. Makes me glad I'm (mostly) in lurk mode nowadays.

Also, I'm totally stealing "picofuck."

Tetronian you're clueless22:32, 15 July 2012

I'm also apparently a part of the Yudkowsky cult, despite the fact that I probably disagree with him on more things than ADK and David Gerard, mostly on esoteric cognitive science stuff. I think he's a computational fundamentalist, something which I see as a vestige of the days of early cog sci and GOFAI.

Nebuchadnezzar (talk)00:33, 16 July 2012

Oh, and he also buys into Santa Barbara brand evo psych, which means he endorses the hardest form of massive modularity. IMO, that's a bunch of hokum that flies in the face of the research in neuroscience, but then again, that's true of a lot of things that the hardcore AI-types think about the brain. I don't think Yudkowsky even realizes this, which is probably the worst part.

Nebuchadnezzar (talk)00:39, 16 July 2012

And what the fuck is this shit? Apparently PZ telling some computer programmers that they need to learn biology before commenting on how to simulate a brain can only be responded to with "PZ needs to learn more about computers".

Scarlet A.pngmoral15:34, 16 July 2012
 

Thanks for the links! Looks like I have some interesting reading ahead of me. At the moment, I don't know anything about either of these issues, so I don't really have anything to add to the discussion just yet. As a CS type myself, I'm pretty ignorant of cognitive science and philosophy, and I have more than a bit of catching up to do. (I've always been ignorant of these topics, but until recently I had a vastly overinflated sense of how much I knew.) I'll have to read at least those two articles and get back to you.

Tetronian you're clueless20:13, 16 July 2012
 

"I'm also apparently a part of the Yudkowsky cult..." Where have I said that? You are just making shit up.

Baloney Detection (talk)16:30, 16 July 2012
 
 

Can you be bothered to actually live by the graph on your profile page? Whenever I respond to you, and you (presumably) don't know how to reply in a reasonable way, you reply that you don't care or with an ad hominem.

You don't care? Fine. But then why did you even bother to post in the first place? Moron.

Baloney Detection (talk)16:27, 16 July 2012

Yes, I do live by the flow chart. But since you basically opened up with "don't expect much of this person, he's sucked into a cult" (which, by the way, is and ad hominem attack, whereas calling someone an idiot is just a crass insult - they're not the same thing) you left the chart by one of the exits pretty early and I had no particularly compelling desire to return to it, leaving the game open for excesses of contempt and sarcasm, which is far more enjoyable because frankly there's nothing better in the whole world than being snide to someone who thinks they're intelligent. I did give you a reasonable reply, one stating that you've set up a false dichotomy (and continue to do so), yet you did not really take this on board and instead tried to de-rail it onto something else.

Now, as I said before, I give about half a femto-fuck about Yudkowsky, yet you insisted on saying I practically worship him as a guru - without any evidence apart from the fact I have a fairly middle-of-the-road opinion that goes along the lines of "sometimes right, sometimes wrong, meh". Now, why did I post? Well, thanks to the magic of how forum threads work we can track up above, and you'll see two things, these are, in no particular order:

a) that this is my user page, so it's only polite to respond in some way. b) you made the first post digging something up that is a few weeks old (and also, started by someone else).

Evidence, which is there for people to see, doesn't quite fit your notions with respect to that. That and the fact you seem to be constantly attacking me, without much consideration for making a constructive point - "drinking too much Yudkowskian kool-aid" as you put it - then I'll recalibrate my fuck-o-meter to read between 200 and 800 milli-fucks.

So, in conclusions, you still have this "you are either with us or you are with the terrorists" mentality with respect to Less Wrong, and until you drop that I really have no inclination to give your personal obsession over everything Eliezer Yudkowsky says any more thought. Indeed, the only reason this post is as long as it is is because, as I said above, I like being snide to people who think they're intelligent.

Scarlet A.pngpathetic17:01, 16 July 2012

Well you are the one who complained that it's not fair to say that LW is a "claptrap of pseudoscience", despite the fact that LW promotes more than its fair share of pseudoscience, fringe ideas and crankery. Yudkowsky is not getting off the hook, get it? If he promotes pseudscience (he does), then he is called out on it.

And I clearly pointed out to you that I don't hate all LWers, yet you repeat that claim. Are you unable to comprehend or are you lying?

Baloney Detection (talk)17:12, 16 July 2012