A modified version of the Prisoner's Dilemma shows up in more than a few game shows like Trust Me, The Bank Job, Take It All, and Friend or Foe? with a twist. If both players cooperate, they split the pool of money. If both players betray each other, they get nothing. But if one betrays the other, the traitor gets all of the funds and the trusting player gets nothing. Does this alteration change what the optimal move is?
You absolutely must look at the free web game/simulation "The Evolution of Trust". It takes the Prisoner's Dilemma and explores what happens if you run a multiple games of it. What are the best strategies and how morale is born out of repeated interactions. It also lets you find the conditions when trust falls apart. It's based on the books "The Evolution of Cooperation!" and " The Complexity of Cooperation".
In that case, your choice is only about how you want the other to benefit. The true game is in convincing the other to give you at least half of the pot.
@@anarchisttechsupport6644 Trust is a weapon that can be used against you thats why we bugged all of our allies governments a friend is just someone whos not yet an enemy
@@jerrydickerson1111 yeah. . . Nash was a raving paranoid schitzophrenic. If he had been a hippy? Im pretty sure we'd find a way to have a better world.
@@jerrydickerson1111 Its not even about trust, its about business and making sure everyone is in the same boat should things go south. No matter what people do to each other, if anyone crosses a certain line, they all fall together. They will push that line to the limit though
The other cool thing about is this is when instead of having one game, you do multiple games back to back where your actions in one influence how the other person will act in the next. In that scenario, which is much closer to real life, the best strategy is to copy what the other person did on their last turn. It's the game theory equivalent of the Golden Rule, do unto others as you would have them do unto you.
If they're only going to play once, a player will always get more by defecting. But if they are going to interact with somebody multiple times; if they play once, then they play again and again, and we add up their scores, the strategy changes. In 1980 Robert Axelrod held a tournament where anyone could submit a strategy. Each strategy did 200 rounds against each other strategy. There were 14 strategy submitted. Plus the strategy 50/50 RANDOM. The winner was a strategy called tit for tat. TIT FOR TAT cooperates on the first round and from then on it just copies what the other person did last round. Why did it win? It's a simple strategy, but one thing is does is, it reciprocates quickly against defectors. Any strategy against tit for that that tries to take advantage of it, gets instantly punished and put into a bad situation. So if the other strategy keeps defecting, or even if it tries to go back to cooperating, it would have gained less than it would have if it had just kept cooperating with TIT FOR TAT. And if TIT FOR TAT didn't punish, TIT FOR TAT would have been worse off. So we might say TIT FOR TAT is "retaliating", it punished defection. A strategy that never defects first kind of looks like it's being nice, so we could say TIT FOR TAT is a nice strategy. And it seemed to be a good trait to have in this tournament, the top 8 strategies were nice, and the bottom 7 were not.
Another bright side is what happens when the prisoner's dilemma is iterated. When the prisoner's dilemma is repeated, the winning strategy is to cooperate and then mirror the opponent's behavior. The Tit-for-tat and similar strategies dominate iterated prisoners dilemma. In your gaming and teamwork analogy, consider the differences between a pick up team and a team you play with often. Each game has the same opportunities to benefit the individual at the expense of the team. But the regular team doesn't do this often because they want their teammates to play well with them in the future. This makes a huge difference. Global warming is actually the tragedy of the commons game. It's similar to the prisoner's dilemma, but for multiple players.
This is why faster multiple rounds with the same group is more "fun" than a single long round. Take Team Fortress 2, Heroes of the Storm 3-round Brawls, and DotA2. Yes a mixed fruit cart, but we don't really have many MOBAs (height of player toxicity) that are multi-round. In TF2 rounds can be very short, and the servers tend to roll over the groups into the next map/match without tossing everyone out. You get to know (by behaviors and actions) who are the fun people, and who are the jerks who keep taking the Engie's scrap/ammo. HotS is one of the shortest MOBA rounds currently out there. In the Brawls, these can be even shorter. With players doing a Best of 3. It also has Team EXP, and other kinds of "share kills" with team mates. DotA2 and LOL have extremely long rounds, with a great deal of selfish actions possible.
The issue with the iterated prisoner's dilemma is if you know when it ends - as after the last turn, there are no further consequences (in the game). As such, your best bet is to run the above strategy, and then betray on the last turn. However, If everyone knows this, then everyone will do it, and thus the game functionally ends at (end-1). Which means that the best strategy is to cooperate/tit-for-tat, until the (end-1) turn. et cetera, et cetera, until you get to betraying on all turns. The solution to THIS is to not have the game end at a pre-determined point; ie, an average length of a game will be 20 turns, but (say) it's actually a standard deviation based AROUND 20.
I think this shows internationally as well; because you do have some nations trying decently enough, not as much as they COULD be, but *some* to do something about climate change.
i'd argue that vlr explains the prisoner's dilemma better than extra credits even did. because, y'know, the actual choices tend to depend on what type of person you are and the history you have with the other participants.
I quit Overwatch over this exact kind of behaviour. After a while, it just destroys the game experience for everybody. It's a team game - and yet people only experience it on an individual level, so they hog resources that other players need, and fail to understand why their success in getting one or two kills just lost us the match - because they obtained those kills by overextending and leaving their backline undefended. Essentially, they just spent resources belonging to the entire team, in order to pursue their own personal glory. The whole concept of the game doesn't work when people are thinking selfishly like that - and the stats-driven progression and "glory" systems built into Overwatch only serve to *reinforce* the selfish behaviour.
when i used to play overwatch, i hated playing support. I was good at it, but in overwatch the support role is the babysitter role. If the team wins it because the dps killed everyone, but if the team loses its because the support didnt do good enough.
See, the secret of the Dilemma is the "no communication" part. Without that, there'd be no dilemma and the best choice could be made. We need to communicate with each other more. Pay attention and listen to people around you.
This episode. Coming out the same time that the rain forests in Brazil are on fire, so that cattle ranchers can get away with illegally clearing the land for business purposes.
ldsk Yeah, infinite or unknown iterations completely change the Nash Equilibrium, and are more in line with real world problems. Calling global warming a limited PD undermines the whole argument, as it implies that the only correct option is to do nothing.
@@onenof10 now to be fair, the explicitly say that the PD does not reflect real life scenarios like that, but that people involved in those real lode scenarios can often still think in the terms of the normal PD.
If they're only going to play once, a player will always get more by defecting. But if they are going to interact with somebody multiple times; if they play once, then they play again and again, and we add up their scores, the strategy changes. In 1980 Robert Axelrod held a tournament where anyone could submit a strategy. Each strategy did 200 rounds against each other strategy. There were 14 strategy submitted. Plus the strategy 50/50 RANDOM. The winner was a strategy called tit for tat. TIT FOR TAT cooperates on the first round and from then on it just copies what the other person did last round. Why did it win? It's a simple strategy, but one thing is does is, it reciprocates quickly against defectors. Any strategy against tit for that that tries to take advantage of it, gets instantly punished and put into a bad situation. So if the other strategy keeps defecting, or even if it tries to go back to cooperating, it would have gained less than it would have if it had just kept cooperating with TIT FOR TAT. And if TIT FOR TAT didn't punish, TIT FOR TAT would have been worse off. So we might say TIT FOR TAT is "retaliating", it punished defection. A strategy that never defects first kind of looks like it's being nice, so we could say TIT FOR TAT is a nice strategy. And it seemed to be a good trait to have in this tournament, the top 8 strategies were nice, and the bottom 7 were not.
Ya, the iterated PD shows a better reflection of reality, let's face it, we are more likely to behave like assholes to complete strangers, but with people that we are likely to have future interactions with, we are more likely to play the "tit-for-tat" strategy rather than the "always betray" strategy.
I love how people keep saying online "It's a forest, it's gonna burn at some point", when they're missing the key details of how the Amazon rainforest gets its name: RAIN! The humidity is high enough that no fire is needed to clear it, nor fire being able to, as the ground is constantly decomposing faster than the rate at which a fire could even start in California.
@@exurbian2420 Yeah, I know. This whole situation seems completely lost, and I just want some shred of hope. I just want some hope that maybe we can learn from our mistakes in time.
@@connorwalters3240 I don't know about you, but I'm hopeful to see people all over the place talking about this, even when news outlets won't. It shows that people everywhere really do have eyes on it, and maybe that's the pressure Brazil needs right now. Let's keep up the pressure.
5:50 "A large number of psychological studies have shown, that at least when dealing with each other as active individuals, as specific human beings and we know other human being are at stake we tend to choose cooperation" I would love to read more about this. @Extra Credits It would be fantastic if you could put the sources for this to the description
This is why we have morals, /and/ why we have laws (well, some of them). To change the payoff structure, and break us out of prisoner's dilemmas, to the good of all.
Zero Escape: Virtue's Last Reward was definitely what came to mind. It's what introduced me to that concept. Also for those who haven't yet, definitely play 999 on the DS. That franchise involves a lot of other philosophical aspects.
when i was in highschool, we have to explain a math problem to our class in english (I'm french). I chosed the prisonners dilema... it brings so good memories
I first heard about the Prisoner's Dilemma in the game Zero Escape: Virtue's Last Reward, however, your video gave me a better understanding as to what it was trying to say. Thanks for making this video!
Here's the whole issue with game theory though; it relies on the premise that there is no communication, and that everyone must necessarily act in so-called """""""""""""rational""""""""""""" self-interest. It actually originated as an idea by a schizophrenic mathematician during the Cold War when it came to the threat of nuclear attack. He postulated that all people, no matter who they were, were always trying to deceive each other and trying to do the most to improve their own well-being, even at the expense of others, and the reason why Game Theory became so widely accepted was due to the rise of Neoliberalism and Laissez-Faire pseudo-economics from the likes of people like Milton Friedman, Murray Rothbard, and others following the long-discredited ideas of Hayek and Mises. A lot of this information is conveyed through the documentary "The Trap" by Adam Curtis. This is all also backed up by Ostrom's work "Collective Action and the Evolution of Social Norms"
I just scrolled through the comments to see if someone wrote something similar to me, which you mostly did. I think it is important to not interpret this 'optimal solution' as a Swiss army knife of decision making. In Reality some people feel bad about someone else having to spend 10 years in prison for them selves, while 7 or 2 years together both seem acceptable. Doing something bad to another person can be a cost, and here it is just assumed it is not. There is no real reason to call self focused behavior rational. Altruism is also is not always rational. It depends on context.
Thank you for covering this so well! You know, I really like the frame of both criminals bumping their fists while serving only a year each. It's the most beautiful thing when people who have every opportunity to betray each other, decide to work together instead. (Not that I condone the murder, I just appreciate trustworthiness and integrity a lot) (and honestly, the real logical fallacy here is thinking only in terms of material, countable things, instead of realising the value of other people in our life and our relationships to them. If we realised how beautiful is when we cooperate and people come together to create something amazing they couldn't have individually, we wouldn't just trade if away for meager material possessions.)
My father in law (a retired NYPD detective) told me how many of there interview techniques were really just convincing one prisoner that the other had talked and if they didn't talk the other would walk free and they would get the max sentence!
Great as always, Extra Credits! My favorites of your videos are the ones where you explain psychological concepts. I hope y'all have a good Halloween episode in mind. It's not close, but it'll come sooner than you think.
Yep. That's a neat little maxim I've heard a lot when it comes to being a mediator. You don't want either party to be fully happy but still agree to it (because the idea is that both had to give up something therefore it's most fair to both)
Relevant external source: look up "The Evolution of Trust" It's a nice little interactive web "game" which illustrates how this principle works in more depth.
The whole thing with "people actually tend to work together" was demonstrated in a variant, the Iterated Prisoner's Dilemma, which plays MULTIPLE games (like 100+) of Prisoner's Dilemma using various strategies
Cooperative board games can sometimes be a great remedy for the prisoner dillemma. They show that sometimes you doing that boring action, or not being able to be as awesome as you'd like a given turn, can hugely benefit the turns of other players and hence your chances of winning. On the flipside, a few board games that implement more social mechanics can show the Prisoner's Dilemma in action beautifully and make it a key component of the experience.
Hey james love your video on the prisoner's dilemma. I really want see you guys analysis the "all pay auction" i dont really know why it popped up in my head while watching this video but i really think its going to be a very thoughtful video. Love your guys and keep on doing the good work!
This reminds me of The Evolution of Trust (which was apparently also based on the Prisoner's Dilemma, but puts more of a focus on exactly how trust works). If you haven't played that already, I'd highly recommend it. I'm not sure whether to call it a game or not (the creator refers to it as "an interactive guide to the game theory of why & how we trust each other"), but it's web-based and it's free. Except instead of prisoners potentially facing jail time, it's a magical machine that multiplies the money you put into it and gives it to someone on the other side. If you both put money in, you both get more money back. If you put money in but the other person doesn't, you lose the money you put in, and the other person gets the magic multiplied money without having to put anything in themselves. It's probably right up your alley if you watched this until the end.
Soorian And the worst part is that I knew it yesterday through a friend from another country instead of knowing it through the news. It seems like they hide these type of stuff from us for some reason. It has been 17 days! And I knew it through Instagram smh
I saw this concept brought up in a evolution video! It states that being aggressive is sometimes bad. This is due to the fact that if an aggressive player (one who will attack to gain the food), meets another, they will both attack each other and result in not gaining any food and have a chance of dying. Then when introduced with passive players, they tend to level out at around the exact same population due to the same reasons in this video, specifically the outcomes coming from the punnet square playing out in a simulation. It’s a fun concept, and applicable to many, many situations!
I teach this subject in Microeconomics as the "Sounds good! You first!" problem. Nuclear disarmament? Sounds good! You first, because I'm not going to be the one without retaliatory power. Negative campaign ads? Sounds good, you first because they work and I'm not going to risk being the only one not running them. In a western movie... we agree to meet unarmed to discuss our differences. Sounds good! You first because I'm not going to be the only one unarmed! But, it also shows where coordination has occurred if we start talking anti-trust. Actions of self interested institutions that go against their self interest are a great place to look. I love this channel. Soooooo many topics. Plus... now I have another way to describe dominant strategies, "the choice that won't result in regret."
I got my degree in finance but have a special interest in game theory. I actually live here in New Jersey and my proudest moment was getting to meet the man himself, John Nash thanks to family friends. And to think it all started with my love of horror movies and seeing the parallels between the prisoner's dilemma and the plot of Saw V!
The basic concept of the Prisoner's Dilemma is the same as that of the Tragedy of the Commons, plus some other thought experiments. Fascinating stuff. It comes into play in evolutionary biology as well, though not as a matter of conscious decisions but naturally produced strategies.
One game I highly recommend is Virtue's Last Reward. It's the 2nd game of a trilogy originally released on DS for game 1 and 3DS/Vita for games 2 and 3, but eventually games the were brought to PS4, Vita, and Steam with game 1 getting HD, voice acting, and overall QoL improvements. But I digress. Virtue's Last Reward has the Prisoner's Dilemma play a huge part of its gameplay. Seriously, check out the Zero Escape trilogy if you haven't already.
I should also add that the entire trilogy is on sale until September 3rd for $18 total on PS4. $10 for the bundle for parts 1 and 2, and $8 for part 3.
There was 'Split or Steal' in Golden Balls, if both players split then they both share the money, if one splits and the other steals, the latter will get all the money, if they both steal then neither of them get the money. This was something that would happen at the end of the show where the prize was accumulating over the course of the show, there was once a prize of £100,000 ish. The problem is that the show is encouraging greed, if you split you either get half the total or nothing but if you steal you will either get all the money or nothing. However, the show made it even more complex by allowing the two people to talk to each other beforehand, this allows the contestants to try and convince the other to split by implying they will do the same before splitting.
Will you do a follow up video about the repeated prisoners dilemma? It's a great model that gives a very clean Game Theoretical explanation for the desire for cooperation you hinted at by the end. Though perhaps a little to mathematical.
Back when i was at school, some people came in to play some puzzle and board games with us. And the board game was a lot like this, if everyone chose 1, everyone went forward 1 spaces, if every one chose 2, no one went forward, and if both were chose, those who chose 1 stayed and those who chose 2 went forward twice.
The best way to fight against it is to have a "safety card" - something that would prevent or minimize the high loss in case if you cooperate and the other side does not
The thing people don't seem to understand about Genji, is that a half-way decent Genji operates like a simple mechanism. Healing goes in, killing the entire enemy team goes out. It's one of Overwatch's few skill checks for healers that haven't been patched out when ranking up, since a worse healer won't understand this, and will force the Genji to either switch to a character they are worse at, or lose. Even when the healer knows what to do conceptually, Genji is always flying all over the map, and is pretty hard to keep consistent heals on. Unfortunately, Genji always being low on health became the meme, not the solution to the problem.
Before the responses come in, I actually play Genji pretty rarely, before I quit Overwatch. I mained Reaper and Lucio, as well as Bastion, when I could get away with it.
A "simplified" situation that I have thought is the free market. Companies have two choices: high prices and low prices, where low prices trumps high prices, but high prices for both is better than low prices. In this case, the consumer wants the companies to not cooperate (form trusts, manipulate the market, etc.) in order to benefit from low prices. Another example is warfare. Take, as a particular example, the trench. Trenches were meant to help soldiers survive as guns became accurate enough for musket-era warfare to be too deadly. However, since both sides used trenches, WW1 bogged down into a war of attrition, arguably worse for both sides than if a swift victory for one was made possible by the absence of trenches.
I teach game theory, including of course the prisoner's dilemma. This video is pretty accurate (although, of course, simplified, especially in regards to WHY cooperation occurs in the real world - plenty of comments have mentioned iteration, but there are also evolutionary versions of the game with cooperation, or prior-commitment devices, among other things). One correction I'd make is that the way you've described climate change is more of a public-goods game, not the prisoner's dilemma. It's similar in many ways but not the same. Anyone interested in game theory should check out Dixit and Nalebuff's Art of Strategy for a not-too-mathematical introduction.
This is completely different from the Prisoner Dilemma (singular/non-possessive) that was described by the Spoony One in his RPG Vlog series Counter Monkey. This dilemma is about imprisoning enemies who have surrendered and have asked to be spared. The question comes with whether the players should actually spare them or kill them as they stand. This often leads to conflicting perspectives between players of different allignments, e.g. lawful-good vs. chaotic-neutral.
This dilemma (or one similar to it) comes up often in the anime Kakegurui, where players involved in a gambling game have to decide whether to focus on themselves, or to achieve some sort of group/team victory.
I was sure the episode would include at least a passing reference of Virtue's Last Reward, seeing as how 90% of the game is built on the back of the Prisoner's Dilemna. You pull off a better explanation than they do.
Did some comment reading, which eventually made me think about children (with siblings) and how they're exposed to the prisoner dilemma at an early age. I'd argue this is due to less than stellar parenting, or guardians just blaming the easiest target. Imagine two young kids, and one of them did something they're not allowed to. Maybe they took a cookie, or didn't clean up after themselves. Doesn't matter so long as they know they did it. Mom finds out, doesn't know who it was and goes to ask. If the culprit blames their sibling and the sibling doesn't blame back, they might dodge a scolding by putting it on their sibling. If the sibling blames back, mom might decide to just scold them both, neither, or just tell them not to do it again. But if the sibling blames the culprit and the culprit stays quiet, they get the full scolding by themselves. Later when they're older, they'll learn of option 1, which is to both stay quiet. An unwritten rule among siblings which is only broken when one sibling has an axe to grind with the other. The result would still be the same as option 4, but at least there's no blame game. It may even be prudent to be quickest on the draw of blaming the other. I've experienced this myself in elementary school. On occasion where I would be fed up with getting bullied during the break, I'd go to the teacher on duty to complain about it and tell who is bullying me. Thing is, the bully would run past me to the teacher and then blame me for bullying them, or something else I didn't do. The teacher was terrible at her job, and whenever people came to complain she'd just scold the first person getting blamed, even if the culprit did the blaming, and everyone knew it. Which is why I stopped reporting bullying in elementary, as the bully would simply beat me to the teacher and blame me for something and then I'd get scolded and they'd get away with it. So not only might children learn to blame the other for a chance of getting away with it, they might also learn to be the first to point the finger as the guardian might not investigate further and just go straight to the scolding of whoever got blamed.
This represents a defining factor of good leadership: When one is able to predict the impacts on various participants without consulting them, and come up with the best option for the best outcome, even if it compromises their position.
After watching this Video, I would like to suggest a video on the methodology and psychology of making comebacks in PvP team games. I'm a League of Legends player (and personal experience has admittedly generated the desire for this video) obviously this topic applies to MOBAs but I think it is also applicable to games such as Destiny's Crucible, CSGO, and others . I know it's a little more on the gameplay side of things than most of your videos, which are more game creation/theory focused, but I figured I'd put it out there because hey, why not?
Does anybody else remember the TV game show Golden Balls? It was this exact same premise. The who final contestant get 2 balls each, one says "Split" the other "Steal". If they both Split, they split the pot between them, if they both Steal, the leave with nothing, if one Splits and the other Steals then the one who Stole gets all the money.
Mirroring what some people have said it would be great if you could talk about the iterated prisoner's dilemma. For those who want to know what that is and how it changes things here is the short and sweet of it. Rather than looking at one huge life changing decision you look at a series of decisions involving the same actors. So let's say rather than Alice and Bob talking about confessing/ratting to one murder they instead repeatedly get brought in for petty theft (that word on the street is they both did but again no proof.) So the equation is: A and B silent : 1 week community service A rats B silent: A free B 4 months in jail A silent B rats : B free A 4 months in jail A and B rats : Both spend 1 month in jail And once both of them are out of jail in 2 months they will be brought back in for questioning every time. (For the sake of argument we won't say they suffer repeat offender laws increasing the severity of the punishment.) So now knowing that no matter what you pick you will have to deal with this person again in the future what becomes the optimal move? If you look at an individual case it is still to rat out your partner, but if you know that you have to come back and do this again AND that they will remember you ratting them out it changes a bit. An extremely common and successful strategy in this case becomes Tit-for-tat. You remain silent unless your partner rats you out and then you rat him out until he stays silent while you rat him out, and then you go back to being silent until he rats you out again. If you both start silent you continue staying silent and both get the minimum punishment. If your partner tries to make a cheap buck at your expense then you return the favor and both suffer till s/he fixes it. Ok most people can stop reading here but for those who want to go even deeper there is one more addition to this problem. Other pairs of thieves, or competition from people who are not Alice or Bob who also steal and who also have to make this choice. So in this addition we will note that time spent in jail is not time spent robbing and making money. If Alice and Bob keep spending 1 month in jail while other people keep doing community service all the good targets will get taken and Alice and Bob won't get to take any of that phat loot. They might even starve because they can't steal enough to keep a food on the table because again all the prime targets keep getting taken. So now you have to look at the problem not only knowing that you have to make this decision multiple times, but also that the person will remember previous decisions, and even further that other people will have to make this choice too. All that put together builds up into a long term evolutionary scenario where you get things like a thieves code to not rat out other thieves. Because when all thieves know this rule they all prosper over thieves who don't. I know I simplified and cut a lot but it is an interesting way to help bring this problem closer to how decisions in real life work and how we end up where we are. Please EC talk more about this! And if you read this far down into my post... well thank you. I hope you found it worth your time.
Ah yes, Prisoner Dilemma. My class in uni failed that test miserably when asked to choose car or bus commutes in a thought experiment where more cars lead to larger traffic jams but choosing the bus gives a longer individual commute.
This is why the jack box trivia murder party mini game with the taking money bags thing is always doomed to end with everyone dead, it follows this exact principle
I love love love that you incorporated "guy who while being questioned by homicide detectives will not stop unloading crates" Doesn't matter to this guy!
One detail: in the PD, the reason what you should do for yourself is different from what you should do overall is because switching from the former (cooperate) to the latter (betray) gives you a small benefit (lower jail sentence) while causing a larger penalty to the others, making the switch good for you but bad for everyone in total.
A lot of people have brought up how the iterated prisoner’s dilemma can produce the both cooperate outcome in the long run. However, it’s actually possible for two perfectly selfish and perfectly logical actors to cooperate even in the one-shot case if they understand each other well enough. Consider playing the dilemma against a clone of yourself. Even if you don’t care about what happens to the clone, you know that whatever decision you make, your clone will make too (because you both have the same brain and the same information). Knowing this, you can play the prisoner’s dilemma as if you were choosing for both players and so your choice is between both cooperate and both defect. Extending this, you can reason that if you are playing a prisoner’s dilemma with someone you think is likely to reason in this way, then you should cooperate because the decision theory that produces cooperate/cooperate gets better outcomes than defect/defect (this is called ‘functional decision theory’ as opposed to ‘causal decision theory’, which is what the video presumes the players are using).
Quick note: While no amount of direct knowledge about your opponents decision in a prisoner's dilemma (from nothing to perfect information) will ever make it a good idea to stay silent, mutual information *will* do the trick. If you don't directly have knowledge of what your opponent will do,but you have some degree of confidence that they will think as you do and decide as you do, cooperation can become Nash optimal. The obvious example is if you know your opponent is a perfect clone of you and will mirror your every choice, then then the right decision is clearly to cooperate. But if you're only partially sure your opponent will mirror you, well then you have to work out the math, but depending on the degree of shared principles you have with your opponent and the specific costs of your dillemna it can be possible that cooperation is optimal even with imperfect information. This more closely represents our real world interactions with strangers, where our lack of direct knowledge of other's decisions can still be colored by our mutual humanity, shared cultural experiences, common frames of reference, etc. It's a 'selfish' reason to cooperate even without taking empathy or care for one's fellow humans into account.
unfortunately I am afraid big governments are much more disaffected by the world around them than most of history, and capitalism (society of mutual self-interest in theory) is more widespread than ever, and we also have big companies which are largely unregulated due to being able to evade the jurisdiction of any particular governments' laws. all of these are not likely to change without appealing to self-interest and greed, which means I don't see a way we can escape the nash equilibrium, I am an aspiring game-theorist and philosopher, but not a professional, what do you think?
Another good example that comes up in the real world is things like fishing stocks. Everyone benefits if you don't over-fish, but if you have some people who overfish and some who do not, the ones who do make out better and are in a better financial position to switch industries if stocks collapse. Unfortunately the way around this is regulation with greater penalties than the rewards for overfishing, which is often really hard to pass because of the metagame.
“Betrayal” rather than “Cooperation” is strictly the rational choice and dominant strategy in a single-turn Prisoner’s Dilemma, and “Betray-Betray” is the Nash Equilibrium. But the rational decision changes when you’re trying to maximize payout over multiple continuous turns, and you’re able to respond to the other player’s previous moves. It’s under those circumstances that cooperation becomes a viable, often preferable play. In an iterative Prisoner’s Dilemma, the increased payout of a single-turn betrayal is short-term and usually not worth losing the chance to reap the long-term rewards of mutual cooperation over many turns. Hence, it’s a very bad strategy to always betray, and compel the other player to respond with betrayal. Instead, when statisticians run simulations on iterative Prisoner’s Dilemma games, the strategies that work best for good and bad players aren’t absolute strategies like “Always Cooperate” or “Always Betray”, but rather adaptive strategies that respond to the actions of other players. More specifically, the best performing strategies when played against multiple player behavior types are reciprocal strategies, sometimes called “copycat” or “tit-for-tat” - in short, you copy the other players player’s previous move. The strategy rewards players for cooperation without getting played for a chump - because the moment the other player betrays, the player simply adjusts to playing betrayal until the other player goes back to cooperation. What’s really interesting is that this iterative version of the Prisoner’s Dilemma is hypothesized to contribute to the development of morality and pro-social behavior, and to demonstrate why reciprocity is at the heart of so much human interaction. As the theory goes, early humans encountered many circumstances which operated like an iterative Prisoner’s Dilemma, and Evolutionary forces pushed them to develop behavior which would desire mutual cooperation but demand trust. If that trust is betrayed, our grievance makes us stop cooperating and want recompense, to disincentivize betrayal and bad faith. In this way, complex cooperative human relationships were enabled by the principle of reciprocity, and we didn’t have to succumb to the negative consequences of a standard, single-turn Prisoner’s Dilemma.
I fear, that these badly researched videos will be the new standard for this channel. They totally missed the fact, that something like team games are in their nature cooperative, not antagonistic. In case of prisoner's diliema it is never assumed, that the robbers could make plan for this situation beforehand and that would make it cooperative. Also, when you mentioned evolution of behavior, memes (not the internet version) also comes from this philosophy of cooperation. Actions or behavior that shows lack of malicious intentions are good shortcuts for establishing trust.
It boils down to trust really. Prisoner's Dilemma shows why trust is a valuable commodity, because you get players more willing to choose a way that gives you points (if the choice is reworded so that we're talking gains instead of losses). If you can establish trust, then you'll win more, and part of that establishment is by not taking advantage of it for short-term gains.
Cooperation until betrayal is the best community strategy mathematically (start cooperating until you get betrayed). Cooperation till betrayal will always win in the long run on nash equilibrium cases. The prisoners dilema would work with cooperative strategies if the worst case scenario didn't had a life sentence. In other words, if the punishment is less intense than an infinite cost punishment, statistically is very easy to defend this cooperative strategy.
The prisoner's dilemma is one of the inspirations for a subfield of Behavioural Economics called public choice theory. This explains why the government is so often corrupt, ineffective, and just downright bad at their job
A modified version of the Prisoner's Dilemma shows up in more than a few game shows like Trust Me, The Bank Job, Take It All, and Friend or Foe? with a twist. If both players cooperate, they split the pool of money. If both players betray each other, they get nothing. But if one betrays the other, the traitor gets all of the funds and the trusting player gets nothing. Does this alteration change what the optimal move is?
I'd say it doesn't. Either way the reasoning behind any given decision is the same.
You absolutely must look at the free web game/simulation "The Evolution of Trust".
It takes the Prisoner's Dilemma and explores what happens if you run a multiple games of it. What are the best strategies and how morale is born out of repeated interactions. It also lets you find the conditions when trust falls apart. It's based on the books "The Evolution of Cooperation!" and " The Complexity of Cooperation".
Hey extra credit ever play virtues last reward it uses the prisoners dilemma its a really interesting game you should check it out
In that case, your choice is only about how you want the other to benefit. The true game is in convincing the other to give you at least half of the pot.
hi extra credits could you do me a favour and do a shoutoout on the next video
Ha, my professor just covered this in International Relations. Basically, treaties between countries are essentially giant prisoners' dilemmas.
The only way we build anything is with trust. Remember that.
@@anarchisttechsupport6644 preaching to the choir, dude.
@@anarchisttechsupport6644 Trust is a weapon that can be used against you thats why we bugged all of our allies governments a friend is just someone whos not yet an enemy
@@jerrydickerson1111 yeah. . . Nash was a raving paranoid schitzophrenic. If he had been a hippy? Im pretty sure we'd find a way to have a better world.
@@jerrydickerson1111 Its not even about trust, its about business and making sure everyone is in the same boat should things go south. No matter what people do to each other, if anyone crosses a certain line, they all fall together. They will push that line to the limit though
The other cool thing about is this is when instead of having one game, you do multiple games back to back where your actions in one influence how the other person will act in the next. In that scenario, which is much closer to real life, the best strategy is to copy what the other person did on their last turn. It's the game theory equivalent of the Golden Rule, do unto others as you would have them do unto you.
You absolutely must look the free web game/simulation "The Evolution of Trust" (you need around 30 minutes).
It get things to the next level.
Except you need to be able to forgive mistakes, otherwise there is a downward cycle the moment someone screws up.
Another game to illustrate this is Zero Escape: Virtues Last Reward.
The number of players in a game also effects it too.
If they're only going to play once, a player will always get more by defecting. But if they are going to interact with somebody multiple times; if they play once, then they play again and again, and we add up their scores, the strategy changes.
In 1980 Robert Axelrod held a tournament where anyone could submit a strategy. Each strategy did 200 rounds against each other strategy. There were 14 strategy submitted. Plus the strategy 50/50 RANDOM.
The winner was a strategy called tit for tat. TIT FOR TAT cooperates on the first round and from then on it just copies what the other person did last round. Why did it win? It's a simple strategy, but one thing is does is, it reciprocates quickly against defectors. Any strategy against tit for that that tries to take advantage of it, gets instantly punished and put into a bad situation. So if the other strategy keeps defecting, or even if it tries to go back to cooperating, it would have gained less than it would have if it had just kept cooperating with TIT FOR TAT. And if TIT FOR TAT didn't punish, TIT FOR TAT would have been worse off. So we might say TIT FOR TAT is "retaliating", it punished defection.
A strategy that never defects first kind of looks like it's being nice, so we could say TIT FOR TAT is a nice strategy. And it seemed to be a good trait to have in this tournament, the top 8 strategies were nice, and the bottom 7 were not.
3:52 To be fair, the Mercy probably shouldn't be healing the enemy Roadhog
nobody would protect her in her team so she switched side.
If only we could get a legible health bar in-game...
Roadhog’s Gas is more a last ditch effort not a primary healing source.
Kaius Foxx it’s an enemy not her teammate
Lol
Another bright side is what happens when the prisoner's dilemma is iterated. When the prisoner's dilemma is repeated, the winning strategy is to cooperate and then mirror the opponent's behavior. The Tit-for-tat and similar strategies dominate iterated prisoners dilemma.
In your gaming and teamwork analogy, consider the differences between a pick up team and a team you play with often. Each game has the same opportunities to benefit the individual at the expense of the team. But the regular team doesn't do this often because they want their teammates to play well with them in the future. This makes a huge difference.
Global warming is actually the tragedy of the commons game. It's similar to the prisoner's dilemma, but for multiple players.
This is why faster multiple rounds with the same group is more "fun" than a single long round. Take Team Fortress 2, Heroes of the Storm 3-round Brawls, and DotA2. Yes a mixed fruit cart, but we don't really have many MOBAs (height of player toxicity) that are multi-round.
In TF2 rounds can be very short, and the servers tend to roll over the groups into the next map/match without tossing everyone out. You get to know (by behaviors and actions) who are the fun people, and who are the jerks who keep taking the Engie's scrap/ammo.
HotS is one of the shortest MOBA rounds currently out there. In the Brawls, these can be even shorter. With players doing a Best of 3. It also has Team EXP, and other kinds of "share kills" with team mates.
DotA2 and LOL have extremely long rounds, with a great deal of selfish actions possible.
The issue with the iterated prisoner's dilemma is if you know when it ends - as after the last turn, there are no further consequences (in the game). As such, your best bet is to run the above strategy, and then betray on the last turn. However, If everyone knows this, then everyone will do it, and thus the game functionally ends at (end-1). Which means that the best strategy is to cooperate/tit-for-tat, until the (end-1) turn. et cetera, et cetera, until you get to betraying on all turns.
The solution to THIS is to not have the game end at a pre-determined point; ie, an average length of a game will be 20 turns, but (say) it's actually a standard deviation based AROUND 20.
I think this shows internationally as well; because you do have some nations trying decently enough, not as much as they COULD be, but *some* to do something about climate change.
DID SOMEBODY SAY VIRTUES LAST REWARD?
I mean....nobody did but that is pretty much 10000% prisoner's diliemma and I just like talking about VLR
It was a good game!
It taught me quite well the whole prisioner's dilemma situation in a entertaining way by maling you choose
i'd argue that vlr explains the prisoner's dilemma better than extra credits even did. because, y'know, the actual choices tend to depend on what type of person you are and the history you have with the other participants.
@@sol_ARG yeah, but to be fair VLR had a 15hr campaign to help illustrate the concequences of each scenario in the chart
I literally finished that game 2 days ago and watched this expecting it to be mentioned, but it was not :( have any of you played ZTD?
i was hoping someone would mention this game in the comments!! zero escape is a fantastic series
I main support in Overwatch, and the Genji example is something I deal with on a daily basis. Or at least every time I play Overwatch.
Andrew Beehler Yeah that really struck a chord
I quit Overwatch over this exact kind of behaviour. After a while, it just destroys the game experience for everybody.
It's a team game - and yet people only experience it on an individual level, so they hog resources that other players need, and fail to understand why their success in getting one or two kills just lost us the match - because they obtained those kills by overextending and leaving their backline undefended. Essentially, they just spent resources belonging to the entire team, in order to pursue their own personal glory.
The whole concept of the game doesn't work when people are thinking selfishly like that - and the stats-driven progression and "glory" systems built into Overwatch only serve to *reinforce* the selfish behaviour.
when i used to play overwatch, i hated playing support. I was good at it, but in overwatch the support role is the babysitter role. If the team wins it because the dps killed everyone, but if the team loses its because the support didnt do good enough.
I need Healing XD
Dashes into a Fight, I NEED HEALING!
See, the secret of the Dilemma is the "no communication" part. Without that, there'd be no dilemma and the best choice could be made.
We need to communicate with each other more. Pay attention and listen to people around you.
0:33 ZOEY IS THE LAW
We need art of Zoey in a tiny Judge's Helmet
The thumbnail tho...
*I nEeD hEaLiNg*
i иееd неаliиg
TF2 Heavy: MEDIC!
TF2 Scout: I need a medic here!
TF2 Medic: WHO DO I HEAL?!
Ahem.
*MEEEEEEEEEEEEDIIIIIIIIIIIC*
@@501ststormtrooper9 and that is why resto druids exists
As a mercy main who's taken a break from overwatch, I would like to let you know, the "I Need Healing" made me jump out of my skin.
This episode. Coming out the same time that the rain forests in Brazil are on fire, so that cattle ranchers can get away with illegally clearing the land for business purposes.
Please, I hope people just upvote this all the way to the top!
to the top you go good sir.
Good luck getting any of that to change for the next three years, with Mr. "I would kill my son if he was gay" in charge.
@@Inogat Thanks =)
Hope we can fix this by spreading awareness of what's happening, at the very least
You should have talked about the iterated prisoner's dilemma.
ldsk Yeah, infinite or unknown iterations completely change the Nash Equilibrium, and are more in line with real world problems. Calling global warming a limited PD undermines the whole argument, as it implies that the only correct option is to do nothing.
@@onenof10 now to be fair, the explicitly say that the PD does not reflect real life scenarios like that, but that people involved in those real lode scenarios can often still think in the terms of the normal PD.
If they're only going to play once, a player will always get more by defecting. But if they are going to interact with somebody multiple times; if they play once, then they play again and again, and we add up their scores, the strategy changes.
In 1980 Robert Axelrod held a tournament where anyone could submit a strategy. Each strategy did 200 rounds against each other strategy. There were 14 strategy submitted. Plus the strategy 50/50 RANDOM.
The winner was a strategy called tit for tat. TIT FOR TAT cooperates on the first round and from then on it just copies what the other person did last round. Why did it win? It's a simple strategy, but one thing is does is, it reciprocates quickly against defectors. Any strategy against tit for that that tries to take advantage of it, gets instantly punished and put into a bad situation. So if the other strategy keeps defecting, or even if it tries to go back to cooperating, it would have gained less than it would have if it had just kept cooperating with TIT FOR TAT. And if TIT FOR TAT didn't punish, TIT FOR TAT would have been worse off. So we might say TIT FOR TAT is "retaliating", it punished defection.
A strategy that never defects first kind of looks like it's being nice, so we could say TIT FOR TAT is a nice strategy. And it seemed to be a good trait to have in this tournament, the top 8 strategies were nice, and the bottom 7 were not.
Ya, the iterated PD shows a better reflection of reality, let's face it, we are more likely to behave like assholes to complete strangers, but with people that we are likely to have future interactions with, we are more likely to play the "tit-for-tat" strategy rather than the "always betray" strategy.
@@ahmh8411 Oh wow. Thanks for this; now I know the optimum strategy to choose when playing Universal Paperclips.
The Amazon is on fire. We've already collectively sold the other out.
I love how people keep saying online "It's a forest, it's gonna burn at some point", when they're missing the key details of how the Amazon rainforest gets its name: RAIN! The humidity is high enough that no fire is needed to clear it, nor fire being able to, as the ground is constantly decomposing faster than the rate at which a fire could even start in California.
the current wildfire is nothing compared to the daily practices of logging and cattle industries who move in on the forest
@@exurbian2420 Yeah, I know. This whole situation seems completely lost, and I just want some shred of hope. I just want some hope that maybe we can learn from our mistakes in time.
@@connorwalters3240 I don't know about you, but I'm hopeful to see people all over the place talking about this, even when news outlets won't. It shows that people everywhere really do have eyes on it, and maybe that's the pressure Brazil needs right now.
Let's keep up the pressure.
kirknay
It burns annually.
Look at a world fire map
5:50 "A large number of psychological studies have shown, that at least when dealing with each other as active individuals, as specific human beings and we know other human being are at stake we tend to choose cooperation"
I would love to read more about this.
@Extra Credits It would be fantastic if you could put the sources for this to the description
This is why we have morals, /and/ why we have laws (well, some of them). To change the payoff structure, and break us out of prisoner's dilemmas, to the good of all.
The dilemma hinges on the fact that they cannot work together. This is why it's so important to communicate and work together.
Zero Escape: Virtue's Last Reward was definitely what came to mind. It's what introduced me to that concept.
Also for those who haven't yet, definitely play 999 on the DS. That franchise involves a lot of other philosophical aspects.
Lexyvil such a good series!! I love the use of psychology and mind fuckery in the series
when i was in highschool, we have to explain a math problem to our class in english (I'm french). I chosed the prisonners dilema... it brings so good memories
I first heard about the Prisoner's Dilemma in the game Zero Escape: Virtue's Last Reward, however, your video gave me a better understanding as to what it was trying to say. Thanks for making this video!
0:43 keep an eye on the peanut butter... 0:47 it's pronounced gif. XD
@DerpToni616372 I was quoting the video, I personally don't care how ppl pronounce it
No it's pronounced gif
Here's the whole issue with game theory though; it relies on the premise that there is no communication, and that everyone must necessarily act in so-called """""""""""""rational""""""""""""" self-interest. It actually originated as an idea by a schizophrenic mathematician during the Cold War when it came to the threat of nuclear attack. He postulated that all people, no matter who they were, were always trying to deceive each other and trying to do the most to improve their own well-being, even at the expense of others, and the reason why Game Theory became so widely accepted was due to the rise of Neoliberalism and Laissez-Faire pseudo-economics from the likes of people like Milton Friedman, Murray Rothbard, and others following the long-discredited ideas of Hayek and Mises.
A lot of this information is conveyed through the documentary "The Trap" by Adam Curtis. This is all also backed up by Ostrom's work "Collective Action and the Evolution of
Social Norms"
I just scrolled through the comments to see if someone wrote something similar to me, which you mostly did.
I think it is important to not interpret this 'optimal solution' as a Swiss army knife of decision making.
In Reality some people feel bad about someone else having to spend 10 years in prison for them selves, while 7 or 2 years together both seem acceptable. Doing something bad to another person can be a cost, and here it is just assumed it is not.
There is no real reason to call self focused behavior rational. Altruism is also is not always rational. It depends on context.
I love the excessive use of the law and order meme
* visible belt noises *
Thank you for covering this so well!
You know, I really like the frame of both criminals bumping their fists while serving only a year each. It's the most beautiful thing when people who have every opportunity to betray each other, decide to work together instead. (Not that I condone the murder, I just appreciate trustworthiness and integrity a lot)
(and honestly, the real logical fallacy here is thinking only in terms of material, countable things, instead of realising the value of other people in our life and our relationships to them. If we realised how beautiful is when we cooperate and people come together to create something amazing they couldn't have individually, we wouldn't just trade if away for meager material possessions.)
"I need healing"
-Every Genji
I REQUIRE HEALING
*NOW BITHC*
i require the ability to recover from my wounds
My father in law (a retired NYPD detective) told me how many of there interview techniques were really just convincing one prisoner that the other had talked and if they didn't talk the other would walk free and they would get the max sentence!
If the police are interviewing you it's because they can't prove anything.
Great as always, Extra Credits! My favorites of your videos are the ones where you explain psychological concepts. I hope y'all have a good Halloween episode in mind. It's not close, but it'll come sooner than you think.
i was honestly expecting them to discuss the prisoner dilemma in D&D
which is really an alignment dilemma
You guys should do an episode about the horrifying work environment of play testing
i'm honestly a little disappointed that there was no reference to virtue's last reward
I love you guys :) Please don't change and don't you DAAARE even consider stopping your channel!! I send my regards from Sweden
I remember something I once saw in A Dance with Drangons: "In a good agreement both parts aren't satisfied". Or something like that.
Yep. That's a neat little maxim I've heard a lot when it comes to being a mediator. You don't want either party to be fully happy but still agree to it (because the idea is that both had to give up something therefore it's most fair to both)
Relevant external source:
look up "The Evolution of Trust"
It's a nice little interactive web "game" which illustrates how this principle works in more depth.
I was thinking of exactly that. I just couldn't remember what it was called.
Loving the "It was Walpole" easter egg at 2:30
The whole thing with "people actually tend to work together" was demonstrated in a variant, the Iterated Prisoner's Dilemma, which plays MULTIPLE games (like 100+) of Prisoner's Dilemma using various strategies
Cooperative board games can sometimes be a great remedy for the prisoner dillemma. They show that sometimes you doing that boring action, or not being able to be as awesome as you'd like a given turn, can hugely benefit the turns of other players and hence your chances of winning.
On the flipside, a few board games that implement more social mechanics can show the Prisoner's Dilemma in action beautifully and make it a key component of the experience.
Hey james love your video on the prisoner's dilemma. I really want see you guys analysis the "all pay auction" i dont really know why it popped up in my head while watching this video but i really think its going to be a very thoughtful video. Love your guys and keep on doing the good work!
Best storytelling I've ever seen for the Nash equilibrium. And I've seen it a few times.
Seriously I highly recommend looking into this further, this rabbit hole goes deeper and it's quite interesting
This reminds me of The Evolution of Trust (which was apparently also based on the Prisoner's Dilemma, but puts more of a focus on exactly how trust works). If you haven't played that already, I'd highly recommend it. I'm not sure whether to call it a game or not (the creator refers to it as "an interactive guide to the game theory of why & how we trust each other"), but it's web-based and it's free. Except instead of prisoners potentially facing jail time, it's a magical machine that multiplies the money you put into it and gives it to someone on the other side. If you both put money in, you both get more money back. If you put money in but the other person doesn't, you lose the money you put in, and the other person gets the magic multiplied money without having to put anything in themselves. It's probably right up your alley if you watched this until the end.
And as this video was released, the Amazon is on fire.
Soorian And the worst part is that I knew it yesterday through a friend from another country instead of knowing it through the news. It seems like they hide these type of stuff from us for some reason. It has been 17 days! And I knew it through Instagram smh
I saw this concept brought up in a evolution video! It states that being aggressive is sometimes bad. This is due to the fact that if an aggressive player (one who will attack to gain the food), meets another, they will both attack each other and result in not gaining any food and have a chance of dying. Then when introduced with passive players, they tend to level out at around the exact same population due to the same reasons in this video, specifically the outcomes coming from the punnet square playing out in a simulation. It’s a fun concept, and applicable to many, many situations!
I teach this subject in Microeconomics as the "Sounds good! You first!" problem. Nuclear disarmament? Sounds good! You first, because I'm not going to be the one without retaliatory power. Negative campaign ads? Sounds good, you first because they work and I'm not going to risk being the only one not running them. In a western movie... we agree to meet unarmed to discuss our differences. Sounds good! You first because I'm not going to be the only one unarmed!
But, it also shows where coordination has occurred if we start talking anti-trust. Actions of self interested institutions that go against their self interest are a great place to look.
I love this channel. Soooooo many topics. Plus... now I have another way to describe dominant strategies, "the choice that won't result in regret."
I got my degree in finance but have a special interest in game theory. I actually live here in New Jersey and my proudest moment was getting to meet the man himself, John Nash thanks to family friends. And to think it all started with my love of horror movies and seeing the parallels between the prisoner's dilemma and the plot of Saw V!
The basic concept of the Prisoner's Dilemma is the same as that of the Tragedy of the Commons, plus some other thought experiments. Fascinating stuff. It comes into play in evolutionary biology as well, though not as a matter of conscious decisions but naturally produced strategies.
Excellent episode. Please more game-theory!
One game I highly recommend is Virtue's Last Reward. It's the 2nd game of a trilogy originally released on DS for game 1 and 3DS/Vita for games 2 and 3, but eventually games the were brought to PS4, Vita, and Steam with game 1 getting HD, voice acting, and overall QoL improvements. But I digress. Virtue's Last Reward has the Prisoner's Dilemma play a huge part of its gameplay. Seriously, check out the Zero Escape trilogy if you haven't already.
I should also add that the entire trilogy is on sale until September 3rd for $18 total on PS4. $10 for the bundle for parts 1 and 2, and $8 for part 3.
The hardest of choices require the strongest of wills
A variant of this has been used as the endgame of a few game shows. Most noticeably the British show Shafted,from 2001.
Brandon Martin-Moore There was also the game show “Golden Balls”
There was 'Split or Steal' in Golden Balls, if both players split then they both share the money, if one splits and the other steals, the latter will get all the money, if they both steal then neither of them get the money. This was something that would happen at the end of the show where the prize was accumulating over the course of the show, there was once a prize of £100,000 ish. The problem is that the show is encouraging greed, if you split you either get half the total or nothing but if you steal you will either get all the money or nothing. However, the show made it even more complex by allowing the two people to talk to each other beforehand, this allows the contestants to try and convince the other to split by implying they will do the same before splitting.
When I Saw the 2x2 square, I was instantly reminded of The Evolution of Trust, which is pretty similar.
Will you do a follow up video about the repeated prisoners dilemma? It's a great model that gives a very clean Game Theoretical explanation for the desire for cooperation you hinted at by the end. Though perhaps a little to mathematical.
Back when i was at school, some people came in to play some puzzle and board games with us. And the board game was a lot like this, if everyone chose 1, everyone went forward 1 spaces, if every one chose 2, no one went forward, and if both were chose, those who chose 1 stayed and those who chose 2 went forward twice.
No game explains this better than Virtue's Last Reward in my opinion.
The best way to fight against it is to have a "safety card" - something that would prevent or minimize the high loss in case if you cooperate and the other side does not
The thing people don't seem to understand about Genji, is that a half-way decent Genji operates like a simple mechanism. Healing goes in, killing the entire enemy team goes out. It's one of Overwatch's few skill checks for healers that haven't been patched out when ranking up, since a worse healer won't understand this, and will force the Genji to either switch to a character they are worse at, or lose. Even when the healer knows what to do conceptually, Genji is always flying all over the map, and is pretty hard to keep consistent heals on. Unfortunately, Genji always being low on health became the meme, not the solution to the problem.
Before the responses come in, I actually play Genji pretty rarely, before I quit Overwatch. I mained Reaper and Lucio, as well as Bastion, when I could get away with it.
Also found a lot in Zero Escape games and is what Battle Royale is basically run on too :)
Prisoner’s Dilemma also applies to situations like businesses setting (cutting) prices, but in those situations society benefits.
Man, that climate change riff makes me sad considering the situation in the Amazon right now.
A "simplified" situation that I have thought is the free market. Companies have two choices: high prices and low prices, where low prices trumps high prices, but high prices for both is better than low prices. In this case, the consumer wants the companies to not cooperate (form trusts, manipulate the market, etc.) in order to benefit from low prices.
Another example is warfare. Take, as a particular example, the trench. Trenches were meant to help soldiers survive as guns became accurate enough for musket-era warfare to be too deadly. However, since both sides used trenches, WW1 bogged down into a war of attrition, arguably worse for both sides than if a swift victory for one was made possible by the absence of trenches.
I teach game theory, including of course the prisoner's dilemma. This video is pretty accurate (although, of course, simplified, especially in regards to WHY cooperation occurs in the real world - plenty of comments have mentioned iteration, but there are also evolutionary versions of the game with cooperation, or prior-commitment devices, among other things). One correction I'd make is that the way you've described climate change is more of a public-goods game, not the prisoner's dilemma. It's similar in many ways but not the same. Anyone interested in game theory should check out Dixit and Nalebuff's Art of Strategy for a not-too-mathematical introduction.
The correct option is they both invoke their right to remain silent and demand to see their attorneys.
This is completely different from the Prisoner Dilemma (singular/non-possessive) that was described by the Spoony One in his RPG Vlog series Counter Monkey. This dilemma is about imprisoning enemies who have surrendered and have asked to be spared. The question comes with whether the players should actually spare them or kill them as they stand. This often leads to conflicting perspectives between players of different allignments, e.g. lawful-good vs. chaotic-neutral.
This dilemma (or one similar to it) comes up often in the anime Kakegurui, where players involved in a gambling game have to decide whether to focus on themselves, or to achieve some sort of group/team victory.
I was sure the episode would include at least a passing reference of Virtue's Last Reward, seeing as how 90% of the game is built on the back of the Prisoner's Dilemna.
You pull off a better explanation than they do.
Great video alqays love them
Did some comment reading, which eventually made me think about children (with siblings) and how they're exposed to the prisoner dilemma at an early age. I'd argue this is due to less than stellar parenting, or guardians just blaming the easiest target.
Imagine two young kids, and one of them did something they're not allowed to. Maybe they took a cookie, or didn't clean up after themselves. Doesn't matter so long as they know they did it.
Mom finds out, doesn't know who it was and goes to ask. If the culprit blames their sibling and the sibling doesn't blame back, they might dodge a scolding by putting it on their sibling. If the sibling blames back, mom might decide to just scold them both, neither, or just tell them not to do it again. But if the sibling blames the culprit and the culprit stays quiet, they get the full scolding by themselves. Later when they're older, they'll learn of option 1, which is to both stay quiet. An unwritten rule among siblings which is only broken when one sibling has an axe to grind with the other. The result would still be the same as option 4, but at least there's no blame game.
It may even be prudent to be quickest on the draw of blaming the other. I've experienced this myself in elementary school. On occasion where I would be fed up with getting bullied during the break, I'd go to the teacher on duty to complain about it and tell who is bullying me. Thing is, the bully would run past me to the teacher and then blame me for bullying them, or something else I didn't do. The teacher was terrible at her job, and whenever people came to complain she'd just scold the first person getting blamed, even if the culprit did the blaming, and everyone knew it. Which is why I stopped reporting bullying in elementary, as the bully would simply beat me to the teacher and blame me for something and then I'd get scolded and they'd get away with it.
So not only might children learn to blame the other for a chance of getting away with it, they might also learn to be the first to point the finger as the guardian might not investigate further and just go straight to the scolding of whoever got blamed.
This represents a defining factor of good leadership: When one is able to predict the impacts on various participants without consulting them, and come up with the best option for the best outcome, even if it compromises their position.
Interesting this has been mentioned in media recently. Good timing
After watching this Video, I would like to suggest a video on the methodology and psychology of making comebacks in PvP team games. I'm a League of Legends player (and personal experience has admittedly generated the desire for this video) obviously this topic applies to MOBAs but I think it is also applicable to games such as Destiny's Crucible, CSGO, and others . I know it's a little more on the gameplay side of things than most of your videos, which are more game creation/theory focused, but I figured I'd put it out there because hey, why not?
Does anybody else remember the TV game show Golden Balls? It was this exact same premise. The who final contestant get 2 balls each, one says "Split" the other "Steal". If they both Split, they split the pot between them, if they both Steal, the leave with nothing, if one Splits and the other Steals then the one who Stole gets all the money.
....and this is going in one of my Climate Change lectures. Thanks team!
Came for an interesting take on Game Theory and Decision-Making, got a lazy mantra on climate change. I think this qualifies as a click-bait.
Mirroring what some people have said it would be great if you could talk about the iterated prisoner's dilemma.
For those who want to know what that is and how it changes things here is the short and sweet of it.
Rather than looking at one huge life changing decision you look at a series of decisions involving the same actors.
So let's say rather than Alice and Bob talking about confessing/ratting to one murder they instead repeatedly get brought in for petty theft (that word on the street is they both did but again no proof.)
So the equation is:
A and B silent : 1 week community service
A rats B silent: A free B 4 months in jail
A silent B rats : B free A 4 months in jail
A and B rats : Both spend 1 month in jail
And once both of them are out of jail in 2 months they will be brought back in for questioning every time. (For the sake of argument we won't say they suffer repeat offender laws increasing the severity of the punishment.)
So now knowing that no matter what you pick you will have to deal with this person again in the future what becomes the optimal move?
If you look at an individual case it is still to rat out your partner, but if you know that you have to come back and do this again AND that they will remember you ratting them out it changes a bit.
An extremely common and successful strategy in this case becomes Tit-for-tat.
You remain silent unless your partner rats you out and then you rat him out until he stays silent while you rat him out, and then you go back to being silent until he rats you out again. If you both start silent you continue staying silent and both get the minimum punishment. If your partner tries to make a cheap buck at your expense then you return the favor and both suffer till s/he fixes it.
Ok most people can stop reading here but for those who want to go even deeper there is one more addition to this problem.
Other pairs of thieves, or competition from people who are not Alice or Bob who also steal and who also have to make this choice.
So in this addition we will note that time spent in jail is not time spent robbing and making money. If Alice and Bob keep spending 1 month in jail while other people keep doing community service all the good targets will get taken and Alice and Bob won't get to take any of that phat loot. They might even starve because they can't steal enough to keep a food on the table because again all the prime targets keep getting taken.
So now you have to look at the problem not only knowing that you have to make this decision multiple times, but also that the person will remember previous decisions, and even further that other people will have to make this choice too. All that put together builds up into a long term evolutionary scenario where you get things like a thieves code to not rat out other thieves. Because when all thieves know this rule they all prosper over thieves who don't.
I know I simplified and cut a lot but it is an interesting way to help bring this problem closer to how decisions in real life work and how we end up where we are. Please EC talk more about this!
And if you read this far down into my post... well thank you. I hope you found it worth your time.
Alice,bob and I NEED HEALING
outstanding episode!
Ah yes, Prisoner Dilemma. My class in uni failed that test miserably when asked to choose car or bus commutes in a thought experiment where more cars lead to larger traffic jams but choosing the bus gives a longer individual commute.
This is why the jack box trivia murder party mini game with the taking money bags thing is always doomed to end with everyone dead, it follows this exact principle
I love love love that you incorporated "guy who while being questioned by homicide detectives will not stop unloading crates"
Doesn't matter to this guy!
Good presentation!
I love you guys keep up the good work.
One detail: in the PD, the reason what you should do for yourself is different from what you should do overall is because switching from the former (cooperate) to the latter (betray) gives you a small benefit (lower jail sentence) while causing a larger penalty to the others, making the switch good for you but bad for everyone in total.
A lot of people have brought up how the iterated prisoner’s dilemma can produce the both cooperate outcome in the long run. However, it’s actually possible for two perfectly selfish and perfectly logical actors to cooperate even in the one-shot case if they understand each other well enough.
Consider playing the dilemma against a clone of yourself. Even if you don’t care about what happens to the clone, you know that whatever decision you make, your clone will make too (because you both have the same brain and the same information). Knowing this, you can play the prisoner’s dilemma as if you were choosing for both players and so your choice is between both cooperate and both defect.
Extending this, you can reason that if you are playing a prisoner’s dilemma with someone you think is likely to reason in this way, then you should cooperate because the decision theory that produces cooperate/cooperate gets better outcomes than defect/defect (this is called ‘functional decision theory’ as opposed to ‘causal decision theory’, which is what the video presumes the players are using).
Quick note: While no amount of direct knowledge about your opponents decision in a prisoner's dilemma (from nothing to perfect information) will ever make it a good idea to stay silent, mutual information *will* do the trick. If you don't directly have knowledge of what your opponent will do,but you have some degree of confidence that they will think as you do and decide as you do, cooperation can become Nash optimal.
The obvious example is if you know your opponent is a perfect clone of you and will mirror your every choice, then then the right decision is clearly to cooperate. But if you're only partially sure your opponent will mirror you, well then you have to work out the math, but depending on the degree of shared principles you have with your opponent and the specific costs of your dillemna it can be possible that cooperation is optimal even with imperfect information.
This more closely represents our real world interactions with strangers, where our lack of direct knowledge of other's decisions can still be colored by our mutual humanity, shared cultural experiences, common frames of reference, etc. It's a 'selfish' reason to cooperate even without taking empathy or care for one's fellow humans into account.
unfortunately I am afraid big governments are much more disaffected by the world around them than most of history, and capitalism (society of mutual self-interest in theory) is more widespread than ever, and we also have big companies which are largely unregulated due to being able to evade the jurisdiction of any particular governments' laws. all of these are not likely to change without appealing to self-interest and greed, which means I don't see a way we can escape the nash equilibrium, I am an aspiring game-theorist and philosopher, but not a professional, what do you think?
Basically the dilemma is Split or Steal from the British TV show Golden Balls
1:41 You sons of guns. X )
Teamwork makes the dream work
Another good example that comes up in the real world is things like fishing stocks. Everyone benefits if you don't over-fish, but if you have some people who overfish and some who do not, the ones who do make out better and are in a better financial position to switch industries if stocks collapse. Unfortunately the way around this is regulation with greater penalties than the rewards for overfishing, which is often really hard to pass because of the metagame.
“Betrayal” rather than “Cooperation” is strictly the rational choice and dominant strategy in a single-turn Prisoner’s Dilemma, and “Betray-Betray” is the Nash Equilibrium.
But the rational decision changes when you’re trying to maximize payout over multiple continuous turns, and you’re able to respond to the other player’s previous moves. It’s under those circumstances that cooperation becomes a viable, often preferable play.
In an iterative Prisoner’s Dilemma, the increased payout of a single-turn betrayal is short-term and usually not worth losing the chance to reap the long-term rewards of mutual cooperation over many turns.
Hence, it’s a very bad strategy to always betray, and compel the other player to respond with betrayal.
Instead, when statisticians run simulations on iterative Prisoner’s Dilemma games, the strategies that work best for good and bad players aren’t absolute strategies like “Always Cooperate” or “Always Betray”, but rather adaptive strategies that respond to the actions of other players.
More specifically, the best performing strategies when played against multiple player behavior types are reciprocal strategies, sometimes called “copycat” or “tit-for-tat” - in short, you copy the other players player’s previous move. The strategy rewards players for cooperation without getting played for a chump - because the moment the other player betrays, the player simply adjusts to playing betrayal until the other player goes back to cooperation.
What’s really interesting is that this iterative version of the Prisoner’s Dilemma is hypothesized to contribute to the development of morality and pro-social behavior, and to demonstrate why reciprocity is at the heart of so much human interaction.
As the theory goes, early humans encountered many circumstances which operated like an iterative Prisoner’s Dilemma, and Evolutionary forces pushed them to develop behavior which would desire mutual cooperation but demand trust. If that trust is betrayed, our grievance makes us stop cooperating and want recompense, to disincentivize betrayal and bad faith. In this way, complex cooperative human relationships were enabled by the principle of reciprocity, and we didn’t have to succumb to the negative consequences of a standard, single-turn Prisoner’s Dilemma.
I fear, that these badly researched videos will be the new standard for this channel. They totally missed the fact, that something like team games are in their nature cooperative, not antagonistic. In case of prisoner's diliema it is never assumed, that the robbers could make plan for this situation beforehand and that would make it cooperative.
Also, when you mentioned evolution of behavior, memes (not the internet version) also comes from this philosophy of cooperation. Actions or behavior that shows lack of malicious intentions are good shortcuts for establishing trust.
Simply video thst blow my mind well done
It boils down to trust really. Prisoner's Dilemma shows why trust is a valuable commodity, because you get players more willing to choose a way that gives you points (if the choice is reworded so that we're talking gains instead of losses). If you can establish trust, then you'll win more, and part of that establishment is by not taking advantage of it for short-term gains.
Really good episode
Thank you for summarizing a week and a half of a college class I took.
Cooperation until betrayal is the best community strategy mathematically (start cooperating until you get betrayed). Cooperation till betrayal will always win in the long run on nash equilibrium cases. The prisoners dilema would work with cooperative strategies if the worst case scenario didn't had a life sentence. In other words, if the punishment is less intense than an infinite cost punishment, statistically is very easy to defend this cooperative strategy.
The prisoner's dilemma is one of the inspirations for a subfield of Behavioural Economics called public choice theory. This explains why the government is so often corrupt, ineffective, and just downright bad at their job
this is exactly the dilemma that "split or steal" is based on. look it up on youtube it´s so interesting to see how people act in these situations
I learned "prisoner" dilemma, during my second year of "International Relations" major. Pretty interesting.
Did they also cover the iterated version? Seems to me that it'd be a logical thing to follow up with.
@@Roxor128 sure, did.
I learned all about this from Virtue's Last Reward, and it's still amusing hearing about it here.
Alright. I’m about to make some captions for this... maybe?
Also great video!
i'd love to see more game theory stuff!
I thought I was done with the prisoner's dilemma when I finished the 999 trilogy, but here I am.