Jason_t's flight home...

[PartyGirlUK]

[ QUOTE ]
i still dont understand people who are actually arguing that (2,2) is correct long term. i am perfectly happing to pick $100 every day and gaurantee myself the $97. its like people have a tough time grasping the fact that $97 > $2.

[/ QUOTE ]

You will not guarantee yourself $97 if you choose $100 every time.

[xorbie]

[ QUOTE ]
[ QUOTE ]
i still dont understand people who are actually arguing that (2,2) is correct long term. i am perfectly happing to pick $100 every day and gaurantee myself the $97. its like people have a tough time grasping the fact that $97 > $2.

[/ QUOTE ]

You will not guarantee yourself $97 if you choose $100 every time.

[/ QUOTE ]

orly? i choose $100. you choose $99. next day i choose $100 again. you choose $99 again. i choose $100 again. you break this cycle why?

[Phaedrus]

[ QUOTE ]
its like people have a tough time grasping the fact that $97 > $2.

[/ QUOTE ]

Some people (mostly economics undergrads) would rather both get $2 than risk the other guy getting $101 when they only get $97. No way you're gonna exploit me, dude!

[JaredL]

[ QUOTE ]
[ QUOTE ]
its like people have a tough time grasping the fact that $97 > $2.

[/ QUOTE ]

Some people (mostly economics undergrads) would rather both get $2 than risk the other guy getting $101 when they only get $97. No way you're gonna exploit me, dude!

[/ QUOTE ]

This reminds me of last year. Ariel Rubenstein came to give a talk. He was doing a bunch of experiments where people were playing standard games, PD, dictator, ultimatum etc. He was using response time to describe behavior.

In the ultimatum game (player 1 proposes a split of $2, player 2 accepts keeping his share, or refuses and both get nothing) some players offered 0 (oddly they weren't actually playing for anything, when something is at stake this pretty much never happens) he referred to these people as "casualties of game theory."

Jared

[CallMeIshmael]

[ QUOTE ]
[ QUOTE ]
If you think the answer is anything but 2, you are wrong.

[/ QUOTE ]

I guess you can be forgiven for thinking this. After all you read it in a book written by a PhD and it is standard fare in elementary game theory classes. It's even described in the Wikipedia (in slightly different form) with this answer.

However, the situation is not as straightforward as you may have been led to believe.

It only takes a little searching on the internet with terms like "backward induction, fallacy, paradox" to find that it has long been known that there is a problem with the classical game theory answer.

Quine apparently trashed this general argument as early as 1953, but it is difficult to understand and certainly beyond the scope of an introductory text.

Here's a more recent paper: link

Is there anything awry here? No, in the sense that the proof is correct that there is the unique subgame-perfect Nash equilibrium here of constant mutual defection. Yes, in that the explanatory force of the subgame-perfect Nash equilibrium has dissipated, which also explains its empirical failure in this case as well.


Translation: The proof is correct according to classical game theory but the answer it arrives at is [censored].

[/ QUOTE ]

You act like I dont agree with this post.

Is the answer to the problem 2? Yes.

Is it retarded? Yes

Does it have any real world application? Almost certainly not.

Do I think its a neat problem? Yes, thats why it was posted.

Would I ever actually write 2? No, of course not. 99 is almost certainly the correct response.

[CallMeIshmael]

[ QUOTE ]
Some people (mostly economics undergrads) would rather both get $2 than risk the other guy getting $101 when they only get $97. No way you're gonna exploit me, dude!

[/ QUOTE ]

No one cares about the other guy getting 101. We care about being able to get 100 instead of 97. If you're going to argue, at least understand the problem, please.

[El Diablo]

CMI,

Wait a second. You agree w/ the post Phaedrus made? If that is the case, I don't understand a number of the posts you have made in this thread.

For example, you have said that even in an infinite trial version of this game, Nash and Jason will not arrive at 100/100. I don't see how you can say that and agree w/ Phaedrus' post.

[CallMeIshmael]

[ QUOTE ]
CMI,

Wait a second. You agree w/ the post Phaedrus made? If that is the case, I don't understand a number of the posts you have made in this thread.

For example, you have said that even in an infinite trial version of this game, Nash and Jason will not arrive at 100/100. I don't see how you can say that and agree w/ Phaedrus' post.

[/ QUOTE ]

For some reason or another, people are ignoring the fact that this is a very specfiic situation.

His post argues that backward induction doesnt work in real life with a ton of trials. His post argues that the answer that BI gives with a large number of trials is [censored].

I dont disagree with that.

2 is clearly a 'bad' answer.

BUT... in the situation of 100+ levels of rationality (we dont actually need infinite levels), 2 is also what they would write, and the 'correct' response.

The reason this answer is [censored], is because 100+ levels of rationality does not occur in real life. But, it doesnt matter. It was stated in the problem, so, within the context of the problem, its an assumption.



You had posted some time ago (I never read the post at first, but madtown quoted it to me on aim), saying like "this problem is trivial if you make the assumptions given. I gave CMI more credit than that. And assumed the problem was about how GT solutions dont have practical value"

This is false.

I had shown this problem to some friends (mostly 2p2ers) over aim, and then explained to them the GT solution. All of them agreed that it was 'cool.'

That is why I posted it. Its interesting to see how the 'solution' seems so dumb.

I did NOT post it for people like you, who have GT knowledge. The problem is trivial and boring. But, I think its safe to assume that the majority of OOTers dont hae this knowledge.

It might have been a mistake to make up the story, since it certainly gave it a hint of 'realness.' I would say this was a mistake on my part, and now regret it.


(I need to go for an afternoon run, so I cant respond for like an hour)

[El Diablo]

CMI,

You did not answer my question.

Earlier in the thread (in response to me) you said that even given an infinite trial scenario, the answer would remain 2,2.

If you believe that, I don't see how you agree w/ what Phaedrus posted.

[suzzer99]

I just want to say again that that cartoon rocked. Best laugh of the day.