/sci/ - Science & Math

[Anonymous No.14579250]

Is it ever rational to cooperate in the Prisoner's Dilemma?

[Anonymous No.14579257]

>>14579250
May i ask what the prisoners dilemma is?

[Anonymous No.14579260]

>>14579257
You may.

[Anonymous No.14579270]

>>14579257
It's a situation that comes up in everyday life. It's about whether you take advantage of someone, or try to work together with them.
1. if you both work together, you get a pretty good outcome
2. if one of you helps the other person, but the person doesn't help in return, then one person gets a great outcome and the other gets a very bad outcome
3. if neither of you help each other, you both get mediocre outcomes
Which is the rational thing to do, if you only care about your own interests? Cooperate, and hope that the other person doesn't cheat on you? Or do you defect?

[Anonymous No.14579291]

>>14579250
In the very contrived game known as the Prisoner's Dilemma, not really, especially if you don't know the other agent and can't communicate with them. In iterative games then it can be rational to cooperate because you can establish a reputation over multiple rounds. Furthermore, in the real world there are second and third order effects, like other people seeing that you cooperate and are willing to trust you, resulting in more opportunities. As well as emergent macro level phenomena such as widespread cooperation leading to higher order societal functioning, as seen in homogeneous nations like Japan.

[Anonymous No.14579306]

>>14579250
Every decision is perfectly rational in the moment that it is made.

[Anonymous No.14579313]

>>14579306
ummm what about when your parents conceived you?

[Anonymous No.14579322]

>>14579313
It was not a choice, hence it was not rational.

[Anonymous No.14579382]

>>14579250
No.
The only way to win those games is to act irrationally and trust the other player.

[Anonymous No.14579384]

>>14579382
If a choice gives you a better outcome that choice is by definition "rational"

[Anonymous No.14579397]

I've been thinking about this all day.
How serendipitous. Ty OP.

What happens if you have more than 2 people in the prisoners dilemma? There's 2 "teams" and each team has many people on it. Nobody can switch teams and the game never ends. Each day random amount of randomly selected people from each team are selected to play, and they can chose to either cooperate or betray. Every day there's new round. No team ever truly wins or loses, so the game never ends, but individuals will come up ahead or behind others to a greater degree, and to a lesser degree their team will come up ahead or behind the other team.

I think I got those rules right. This is based off a real world example, but I'd prefer to not talk about that just yet.
Ask OP asks, is there every a rational reason to cooperate?

[Anonymous No.14579398]

>>14579250
If you repeat the game forever and both of you sufficiently value the future then it's rational to cooperate. If you don't cooperate once, you're guaranteeing the other person will never again cooperate. If you cooperate once, you'll want to keep cooperating to get the benefit of future cooperation.

[Anonymous No.14579410]

>>14579384
I take it you one-box in newcomb's problem.

[Anonymous No.14579416]

>>14579250

the prisoners dilemma doesn't take into account future variables.
The logically sound mindset of always take makes sense on a single case basis.
however in the real world interactions don't stop there - especially in business and social situations.
over a long enough period the ability to leverage trust becomes more valuable than leveraging security.
i.e. people who have something and are willing to cut you in on it for mutual benefit will expand your opportunity to enter new deals

[Anonymous No.14579429]

>>14579410
Every rational person should.

[Anonymous No.14579452]

>>14579270
What if you sabotage the other person

[Anonymous No.14579461]

>>14579452
If you sabotage them, they can choose to sabotage you or help you (both of you make your decision without knowing what the other one chose). If they help you and you sabotage them, that's the best outcome possible for you and the worst outcome possible for them. But if they sabotage you back, that leads to a mediocre outcome, an outcome worse than if you had both helped each other.

[Anonymous No.14579484]

>>14579461
You present this as a binary for each player of cooperate/don't cooperate, but this doesn't account for active sabotage as opposed to passive non-cooperation. The dilemma seems basically non-applicable to real life.

[Anonymous No.14579499]

>>14579250
Have you ever considered following God's Law and just telling the truth? Also, not /sci/

[Anonymous No.14579500]

>>14579484
>but this doesn't account for active sabotage as opposed to passive non-cooperation
If you want, you can add a third option to the prisoner's dilemma. That would lead to an intermediary outcome. But passive non-cooperation still has the effect of sabotage, even if it isn't as aggressive or powerful.

[Anonymous No.14579501]

>>14579499
>Also, not /sci/
Thanks for labeling your post.

[Anonymous No.14579509]

>>14579500
I feel like this lack of understanding of real world dynamics is why game theory nerds get stuck in academia until they get tenure or have to leave in shame and eke out life on adjunct gigs

[Anonymous No.14579519]

>>14579509
Let me guess, if you put the effort into writing your real counterargument instead of shitposting (defecting), you fear that I would defect by shitposting against you instead?

[Anonymous No.14579543]

>>14579250
>Is it ever rational to cooperate in the Prisoner's Dilemma?
prisoners dilemma is well understood and completely solved in game theory
your question is redundant, and so is the thread

apply yourself

[Anonymous No.14579666]

>>14579543
Is it rational to cooperate in the iterated prisoner's dilemma when the number of remaining iterations is unknown?

[Anonymous No.14579699]

>>14579250
Depends on how many iterations.

[Anonymous No.14579712]

>>14579499
>not /sci/
Game theory is maths, even though anons aren't actually using any.

[Anonymous No.14579823]

>>14579699
Can you elaborate? You mean like >>14579666?

[Anonymous No.14580223]

Here is a formal definition of the repeated prisoners dilemma for rational actors.
Fiona
Carmichael(2004) A Guide to Game Theory, Prentice Hall, (203-216).


My areas of interest concern the game theory, and behavioural economics when the payoffs are asymmetric

https://www.researchgate.net/publication/265796653\_Alibi\\\_games\_the\_Asymmetric\_Prisoner's\_Dilemmas

and more generally the topology of the 2x2 games that surround the prisoners dilemma

https://www.researchgate.net/publication/266995029\_The\\\_Topology\_of\_the\_2x2\_games\_A\_New\_Periodic\_Table

[Anonymous No.14580235]

>>14580223
that's really interesting. in the PDF, what does r mean in the second to last equation?

[Anonymous No.14580241]

>>14580235
whoops forgot to cut the last line out, should end with

where F is denoted the discount rate and r the annualized rate of return of the asset

[Anonymous No.14580267]

>>14579250
Philosophy belongs on /his/

[Anonymous No.14580271]

>>14580267
Thanks for mentioning it.
This is a thread about math (game theory).

[Anonymous No.14580274]

>>14579410
Brainlet here. I just read about this. I don’t get it. Why wouldn’t you pick both boxes? It always pays out better than one box regardless of the prediction.

[Anonymous No.14580285]

>>14580274
It's just an ambiguous problem, and it depends on what "godlike" powers the predictor has. If he's literally God, he can time travel and change the past, so you should one-box and play along with the almighty predictor. If he's just been really lucky with his guesses, you should two-box because there's a thousand bucks more in it for you.

[Anonymous No.14580327]

>>14580285
Gotcha. Thanks.

[Anonymous No.14580520]

>>14580327
>>14580285
>It's just an ambiguous problem
No, it isn't ambiguous. Time travel or perfect prediction accuracy aren't required to make it work either. The problem is a result of expected utility not matching up with the dominance principle. It's a genuine paradox in that sense.

[Anonymous No.14580742]

>>14580274
The real solution is to precommit your behavior to a conditional that will win you money either way... e.g. if I buy a lotto ticket and win the lottery, I'll two box, but if I buy a lotto ticket and don't win, I'll one box.

[Anonymous No.14581855]

Okay, Newcombs problem, but the boxes are transparent and you can see if the money is there or not. Should this change your strategy at all? Why or why not?

[Anonymous No.14582286]

>>14580274
No it doesn't. When you pick one, you get a million dollars. When you pick both you only get a thousand.

[Anonymous No.14582297]

>>14580285
>>14580520
>It's a genuine paradox in that sense.
The apparent paradox only comes from some people's inability to understand that the behaviour of others depends on their own behaviour. The fact that such people decide what is right and what is wrong is the reason why the world is such an unjust mess today.

[Anonymous No.14582305]

>>14579416
There is a version of this called the iterated prisoner's dilemma which does this, the single prisoner's dilemma does not do so on purpose.

[Anonymous No.14582306]

>>14579250
better question: what are some real life examples of true prisoner dilemmas that show that Game Theory actually does have real life applications?

[Anonymous No.14582427]

>>14582306
I've been hesitant to mention this because it may derail the thread, but I think intentionally using a logical fallacy to win an argument is a perfect example of prisoner's dilemma. And just like the real prisoner's dilemma, if both sides decide to be illogical in an argument, nobody wins, because no rational arguments can be made. (like congress) The part I'm not sure I should mention is I think racism itself is an example of prisoner's dilemma too, with some modified rules. see >>14579397 for pathetic attempt. I know it's not right tho.
I'm not saying people are consciously choosing to be racist because it helps them. I think the choice happens on a subconscious level and most aren't even aware of it, but there are exceptions. see >>14579543 for that point being made for me.
Now the problem with this when both sides see the clear benefits to racism, then you get racism on both sides. This can and DOES happen, tho black people are surrounded by more anti-racism propaganda so that may have an attenuating effect. When both sides chose to act irrationally due to the rational benefit of doing so, then discourse becomes impossible, and the same is true with both sides choose to act/think with racism and prejudice.

[Anonymous No.14582636]

>>14581855
Bump

[Anonymous No.14582642]

>>14582427
Define "racism". It's not a well defined term at all, so you should just say what you mean instead of using a that word.

[Anonymous No.14582713]

>>14579250
yes if you're playing it with your clone
if you choose to cooperate, your clone will probably too

[Anonymous No.14582719]

https://youtu.be/S0qjK3TWZE8

Solved game

[Anonymous No.14583440]

>>14579291
>homogeneous nations
xD

[solarfair00;;;;5006 No.14583823]

>>14579509
>I feel like this lack of understanding of real world dynamics
https://www.youtube.com/watch?v=LJS7Igvk6ZM

>>14579519
>Let me guess, if you put the effort into writing your real counterargument instead of shitposting (defecting), you fear that I would defect by shitposting against you instead?
He doesn't have a counterargument so he just pretends he lost because he "wasn't trying", not because he was bad.

>>14579306
>Every decision is perfectly rational in the moment that it is made.
Sunk Cost Fallacy

>>14579397
>Nobody can switch teams and the game never ends.
https://www.youtube.com/watch?v=AcBTOU7RvbU&t=105

>Ask OP asks, is there every a rational reason to cooperate?

Yes, because of:
>>14579291
>Furthermore, in the real world there are second and third order effects

You can communicate with the teams to affect them. If, alternatively, the teams are so large that you can't communicate with them as wholes, then you can make special signals for "enlightened" members of each team (shibboleths, ichthys, etc.) to subtly display and distinguish themselves from the rabble.

>>14582427
>intentionally using a logical fallacy to win an argument
Yes, this is an example, but you missed the concept of approximation and fat tails. Not every pit bull may maul you, but assuming they will and avoiding them is still rational.

Conversely, if there ever is a pit bull that might not maul you, it can voluntarily choose to countersignal to escape anti-pit-bull prejudice, for example Larry Elder. (Ironically in turn receiving prejudice FROM OTHER PIT BULLS).

>>14581855
>you can see Should this change your strategy at all?
You think you can see it but how do you know it's not a trick by the chooser?

>>14582306
>what are some real life examples of true prisoner dilemmas that show that Game Theory actually does have real life applications?
They are all secret because no one wants to admit to using them. This is, ironically, itself an example of the prisoner's dilemma.

[Anonymous No.14583863]

>>14579250
Not science

[Anonymous No.14584312]

>>14581855
I would still one box.

[Anonymous No.14584316]

>>14582297
>it's not a paradox because the people who have a different theory of rationality than me are wrong
So what is your answer? 1 box or 2 box?

[Anonymous No.14584325]

>>14582306
Pretty much any interaction with other people where you could try to take advantage of each other involves a form of the PD. One example would be not littering in a public park. Another is not cheating on your taxes.

[Anonymous No.14584331]

>>14583863
it's math you dimwit

[Anonymous No.14584343]

>>14584325
>One example would be not littering in a public park.
How about cutting carbon emissions to curb global warming?

[Anonymous No.14584349]

>>14584343
oh yeah. that's a much better one.
another one is virtue signaling stupid shit. If nobody virtue signalled, the whole world would be better off, but if you do it, it benefits you personally.