Someone did this to me when I was cashiering at a zoo. I called my supe, who immediately counted up the contents of my drawer and found that I was a couple bucks short (mistakes happen; we were written up if the discrepancy was more than like $2 but it wasn't a serious problem) rather than the $10 over I'd be if the customer had been telling the truth. I didn't get in trouble.
Besides the problem you already mentioned that people aren't perfect philosophers who can reliably tell who had knowingly hurt them and who not, there's also the fact that people do get locked into loops (or expanding spirals) of revenge. Or whatever the hell you call them again. Conventional wisdom calls for forgiving fast before you actually sink a lot of effort into and get used to the idea of revenge.
I suppose there's also the answer that at least sometimes, there are better solutions. For instance, if people in a given profession are vulnerable to getting screwed over in some given way, they could decide together to not work for employers who will let them be screwed over like that. Form trade unions, that is. Unfortunately that option probably isn't available to a cashier in the US.
Personally, I don't know about others, but I'm not going to take this post as a compelling reason to lower my threshold for taking revenge on people. I have little faith in my ability to reliably tell when I've been knowingly swindled by people. Most likely if I deliberately took revenge more, I would look to people like I'm just behaving erratically and lashing out at people at random times for no good reason.
By the same argument, stealing someone's car is an act of charity if it forces them to walk and so get needed exercise.
It isn't, because the thief's motive was to deprive you of your property not to force you to walk.
The "revenge as an act of charity" argument says that cashiers should take revenge as a rule, because if they did, the world would be a better place overall. But if people stole cars as a rule, then the world would be a worse place overall, even if the occasional victim somehow ended up better because of the exercise.
As long as the motive is revenge, it is not charity.
Anyway, how do you know the world would not be better if people stole cars as a rule?
From the game theorist's perspective, revenge isn't just charitable: it's required. Since employees might be tempted not to carry out a revenge campaign, the other's should punish those who don't punish, and so on, to maintain full cooperation.
An economist's vendetta: Death to those who don't kill those who don't kill those who don't kill those that harm us!
2012-02-10 02:23 am (UTC)
Re: Higher-order revenge
From an open ended game theory perspective, you don't ever seek revenge, instead you don't cooperate. Revenge and non-cooperation are different things. Revenge is about harming someone, making them sicker than they already were (bad for everyone, usually) and wasting your own resources. What game theory does show is valuable is acting in one's own self interest to get something positive (more freedom, more money, more security, etc.). It doesn't actively harm anyone, but it doesn't help anyone else either - you use your resources on yourself, getting more good stuff for you.
Game theory shows us that the best strategy is cooperating as the default play, but then choosing a non-cooperative play as the immediate response to a non-cooperative play by one's game partner. In each case you invest your resources on getting the most for yourself, which is usually through cooperation, but if someone isn't cooperative, you let them know that it's not acceptable by not cooperating with them the next time. Usually you go back to cooperating after one non-cooperation move, no matter what their response to you is, giving them a chance to redeem themselves, and leading as the more forgiving player (humans tend to respond well when forgiven).
This isn't "turn the other cheek" it's getting out of the way and letting the attacker fall on his or her face while you go find someone else more rewarding to play with.
… and the part you're not elaborating on is where these non-cooperative strategies show up in very specific situations. While iterated prisoner's dilemma is a deeply fascinating experiment and area of research, not every interaction you have is modeled by such a game, nor is the iterative aspect necessarily true.
Concretely: if you work at a highway rest-stop, there might not be a second interaction with most your customers at all. This alone is enough to remove the premise of an iterated game at all.
The iterative (I call it "open") game is representative of real life because real humans are essentially all the same. We are all social animals, literally. We are predictable in a general sense of wanting friends. It's even in Maslow's hierarchy. Yes, it's not as important as physical survival, which is why we get competitive behavior in the first place, but collaboration is something all conscious humans gravitate towards eventually, because it biologically (and logically) serves our needs for finding ways to take better care of ourselves.
So unless you're playing the game with a snake or a star or some other non-social being, you'll pretty much always do well to follow the best tactic as shown by the math. As long as a conscious human is playing, it doesn't matter who it is each round, because the strategy has a built in fail-safe, which is that you default to collaboration, and only ever play a non-cooperative move as a direct, one time, response to the previous play. So while it's possible to have the player change right after the previous play, making you look like a jerk to the newcomer, as long as you play the strategy of going back to the default collaborative play regardless of what the new player does in response to your unfriendly act, they will see that you're not as bad as they initially thought, and if you continue to take the higher ground (while being realistic and letting them know that you don't appreciate the non-collaborative moves now and then (at most in response to every other non-collaborative move by the other player/s), you will gain the most possible benefits.
And it doesn't matter if you never meet the player again, since we're all connected, and all playing the game simultaneously, and we are a species that can generalize. When the person you just interacted with a moment ago was generous to you, you will naturally be more generous to the next person you meet, which benefits everyone, as it makes more people more likely to be more generous in the world. And isn't that something we all could use? :-)
No, no, no, no, no. You do not seem to be listening to what I am saying, which makes this discussion a little bit awkward.
Certainly for iterated prisoner's dilemma, there are very successful tactics based around starting cooperatively, and punishing cheaters by non-cooperation. This far, you seem to have grasped the basics.
But in order for this purely computational observation to actually matter to an inherently sociological discussion, you really, really, really have to do some argument why this model is applicable. And I am not even mostly talking about the argument that iterative games model societal games, even if individual interactions do not re-occur -- you need to tell us why human interactions are (reasonably) accurately modelled by a prisoner's dilemma in the first place.
It really is ALL about the payoff matrix. With iterated prisoner's dilemma, the game is stacked in a way that makes tit-for-tat and related strategies interesting, and in a way where the game is not immediately solvable (always cooperate and always defect are easy behaviours to predict with the right entries in the matrix=; but this alone does not make it a good model. And by changing the payoff matrix, entirely new patterns of behaviour are more or less applicable.
For instance, you argue that non-cooperation is preferably to active revenge with the basis of your argument rooted in iterated prisoner's dilemma analyses; but what if the appropriate model is not a two-choice game, but a three-choice game? What payoffs should model an iterated 2-player game with the options cooperate/do not cooperate/take revenge? What properties does THAT game have? Sure, we can formulate a tit-for-tat here, but is that actually more performant than some other tactic?
The narrative is pretty, but you keep referring to "best tactic as shown by the math" -- and if you do that, you need to demonstrate why the math applies to this in the first place.
I tried to explain this already, fully understanding what you were looking for. We humans are naturally/biologically social animals. We always seek out and are chemically rewarded (in the brain) when we find people to cooperate with, as that's the way we are more effective in getting our needs met (two heads are better than one). So really, game theory payoff matrices do need to reward cooperation more than competition, except when there is a survival emergency case of truly limited resources (which is rare, but does happen). You're right in saying that the standard iterated game isn't entirely realistic, but it is mostly proven to be what we humans do, through studies where humans play other humans (and computers) to see how humans choose to play. And it's nearly always beneficial to be generous most of the time, but occasionally selfish (non-generous) when responding to one or more repeated plays of non-cooperation, even when playing with humans.
In real life, again, unless you're playing something that is not naturally/biologically social (cooperative), you're always essentially playing the same player, because we're all connected. We're all complex beings who change our motivations and needs as the situation changes, so our tactics regularly change, but because we are all social, and genetically have a motivation to cooperate, the more you cooperate, and show forgiveness even when someone else fails to cooperate, the more they will be inclined to cooperate with you, and everyone else, too.
As for the math, the payoff matrix that humans use (and the game theory games would be best off using, for the most realistic results) might be something like 3 points each for mutual cooperation, 1 point for individuals who choose non-cooperation, and -1 points for individuals who's cooperative efforts have been rejected. This sets up a realistic reward for cooperation, based on the exponential benefit we humans get when we pool resources towards a shared goal. It sets up a reasonable reward for being selfish when necessary. And it sets up a reminder to not try continue to try to cooperate with consistently non-social players such as rocks, or explosions, or hungry lions. :-)
I am getting a feeling you are really not listening to what I am saying here. The prisoner's dilemma is ONE model of ONE game. It (and similar games) happens to make some particular, easily narrativized behaviours very successful, and thus is a very pretty and nice example to bring into discussions of human behaviour.
It is however merely one possible game. If you change the game, you change which behaviours are successful and which are not. For that matter, if you change the game, these pretty, pretty labels we have on the strategies (cooperate vs. defect) also lose meaning.
Mathematics simply does not work the way you try to use it — before you can appeal to the higher power of mathematics in a real world argument, you really, truly, honestly, have to put in some effort into demonstrating that the mathematical model corresponds to the reality you are trying to describe, predict or analyze.
And you are not doing this. You are continuing to argue the point, that I have conceded for now already, that you might use an iterated game to model a societal behaviour, but you still have not adressed — directly — why the Prisoner's Dilemma is the game that models everything. You have asserted it repeatedly, and added moralistic lessons (i.e. It sets up a reasonable reward for being selfish when necessary. And it sets up a reminder to not try continue to try to cooperate with consistently non-social players such as rocks, or explosions, or hungry lions.) as if they were motivations for the original model. This is NOT what I have been asking for, and I do not entirely know what question you imagine that you have been answering — but what I do need to know is why you believe that the prisoner's dilemma is so superior to all other games in modeling human interaction that lessons drawn about finding a LOCAL optimal strategy (as far as I know, there has been no proof that any of the tit-for-tat variations is a global optimum in the space of IPD strategies) are automatically moral lessons taught to us with the full certitude of axiomatic truth.
And you don't get to assert mathematical truth when arguing morals. At least not without some serious groundwork.
If you want to see what real life is like, and how people choose what to do in situations, you're free to test these things out. It's something that psychologists have been doing, with various versions (not literally the prisoner's dilemma), and the results are interesting. There have been so many studies about how humans behave and how they choose to collaborate, compete, and be neutral with others, that I wouldn't even know where to begin to suggest you look for more information about game theory in real life. Sorry. I don't have time to look it up for you either. Sorry. Maybe do a search for collaboration psychology studies?
Anyway, I do appreciate your discussion here. It's been inspiring and I have gotten some fun work done as a result (a more full sample of a way computer game theory games could be set up for more realistic tests that would reflect real human behavior, with a bit more chance for some new options to appear). (As I said, which it looks like you might have missed, the common game is NOT superior, when it comes to modeling all human behavior, though it models it well enough for most uses.)
Oh, and if you want all three options of cooperation, non-cooperation, and competition, you can add the option that when you compete, you can either win 1 (if the other player cooperates or non-cooperates), or lose 3 (-3 if both players compete).
2012-02-18 05:16 pm (UTC)
Re: Higher-order revenge
The secret is doing what Aikido pros do: take the harmful energy and turn it around to doing something useful. There's a difference between revenge and problem solving. Revenge is a lose-lose (eye for an eye) approach that leaves everyone worse off. Problem solving is looking at all the available resources, and figuring out how to rearrange them in a way that leaves everyone better off.
Why did the customer feel the need to scam the cashier? Why were they so desperate that they thought it was a good decision?
For an example of the problem solving approach, consider what if the cashier offered to give the customer $40 out of his own pocket in the first place, saying something about charity and not wanting anyone to have to be so needy and desperate. This would let the customer know that the cashier understood what was really going on (scamming due to some kind of desperation), while also taking the high ground, where the cashier could tell the manager what happened, end up with a drawer that was correct, AND likely be recognized for being an upstanding employee. This would leave the manager in a position where they'd have to come up with some kind of policy so that employees didn't feel the need to give out their own money to sick, desperate folks.
Also, in problem solving mode, on a larger scale, people being stuck in jobs that disrespect them is bad for everyone, and so we can focus our anger energy in a direction of creating coops and more support for self-employment (and self-sufficiency so that people aren't forced into servitude just to live).
Win-win solutions are more challenging to come up with, but it's what we brainy humans are built for, as compared to our ancestors the fish, who mostly just do the fight-or-flight thing when they have problems.
I would assume a person pulling the "I gave you a $10, but I'm going to swear I gave you a $50" is scummy not desperate. But maybe I'm just...cynical...
What do you see as the difference between "scummy" and "desperate". What sort of causal factor is "scummy"?
Remember the question is: WHY did he feel that it was a good choice, given all his other options? We're looking for a problem solving approach. Think of the customer as a puzzle, and we're looking to find all the pieces that go into making someone who feels compelled to try to get a quick bit of cash from a store (that has lots of it already, presumably). What do these puzzle pieces look like? What kind of situations led up to this guy doing what he did, rather than doing something else?
"Desperate" is someone who is stealing to feed their family; if they don't get money somehow, they are going to starve and for whatever reason they haven't succeeded in getting honest work. They don't want to steal, but feel they have no other alternative to survive.
"Scummy" is someone stealing with little or no remorse; out of want not perceived need. They are immoral, amoral, or maybe even a low grade sociopath. They are stealing because it gives them a thrill, because they're a good scam artist without much of a conscience, because they have some warped view where it's okay to steal from a chain store because it's a big faceless corporation, or just because they can.
OK, so it sounds like you separate a kind of emotional/intellectual desperation for thrill-seeking from physical desperation for food, etc. That's reasonable.
So solving the problem of physical desperation, if that was the motivation the customer had, would involve using collective resources to support his needs better, so that he was able to be healthy and functional, rather than sick and under threat of starvation. Helping him apply for food stamps, offering him a plot in a community garden, collecting unwanted and left over food from stores, homes, farms, and restaurants would probably eliminate his desperation fairly well, and fairly sustainably.
And if his desperation was more emotional/intellectual to the point of him feeling the need to engage in semi-dangerous social games, the solution would probably involve something like offering him a role in the spy business, or maybe in theater/movies where he could enjoy himself and do something beneficial to society, rather than harmful to society.
These kinds of puzzles are challenging and fun to do, as long as you are able to not get emotionally involved, and are instead able to be objective.
I don't like turning the other cheek; it encourages people to keep hitting you until they get tired/bored or they break your face.
I think the problem is people get too emotional about it. Do you suck it and stay, or do you sue your boss and seek to destroy him? But there's the third option: just quit, because your boss is an unreasonable jackass. If anyone asks, tell them the truth. Maybe file a grievance up the corporate food chain. But revenge is a waste of your time.
In traffic I often seen somebody driving 5 miles under the speed limit in the left lane. I could get in front of them, slow down, and get my revenge on them! Or I can mutter "f'n bad drivers", pass them, and not look back. Revenge is a waste of time and energy and when it's over you usually don't even feel real satisfaction from it.
You can bring in Hitler, Bin Laden, serial killers, rapists, pedophiles then give me the option to torture any of them however I saw fit. And I'd rather just give each of them a single shot to the engine block. Yeah, maybe they deserve torture and pain and anguish, but it's just not worth it. Kill them quick and clean, make the world a better place by their removal from it, and move on with your life.
I don't have much to say, but I think this is the best solution presented so far.
2012-02-10 04:08 am (UTC)
"Economists would take an opposite view: ... By taking revenge, I'm sacrificing my own pleasure - my job and my time - in order to help create a world where crappy behavior isn't tolerated and doesn't happen anymore."
What economists have you been talking to? I think that tradeoff warrants more in-depth discussion. In my view (and I'd appreciate your input if you think I'm missing factors), we would lose the productivity of the revenge-seeking employee, AND the entire mud-dragged business would collapse (owner, employees, and customers lose); we would gain confidence that any surviving businesses would be owned and staffed by people terrified of coming off as assholes lest they risk permanent unemployment.
As a side note, given zero evidence of cash fraud except a shouting customer and a shouting employee, I would err on the side of the customer. (See how I didn't say "economists would err on the side of the customer"? That's because I don't speak for economists.)
Moot point because there was not in fact zero evidence -- tallying the cash drawer would easily have revealed the facts.
Tallying the cash drawer would only indicate if the customer was telling the truth, and the cashier was confused. If the cashier was lying, and had pocketed the $40 himself, counting the cash drawer would not have revealed the facts, obviously.
It would still be evidence, just not conclusive.
2012-02-10 05:23 am (UTC)
The standard name for this is "altruistic punishment" - see for instance Fehr & Gachter's (2002) Nature article "Altruistic punishment in humans".
I believe it's a natural primate reaction, I've read that chimps will rather go without food/reward if they perceive an unfair situation than accept the asymmetry.
"I'd be...probably known around town as the guy who threw a fit over an employment squabble"
Depending on the framing in the resulting rumor, it could be a good thing to be known as someone who retaliates when slighted.
You have given me a fantastic rationalization to commit acts of revenge and I'm not sure whether to thank you or curse your name.