War Handwavium

This is a rather long-winded post that explores the Game Theory notion called the Prisoner’s Dilemma and several of its variations. As the variations get more complicated, they start to map to real-world topics, such as Laissez-Faire capitalism, the Invisible Hand of the Free Market, regulated economies, and Health Care Reform issues.

The article is somewhat sequential in nature. If you start reading the section on Health Care, you probably won’t understand the references made earlier, so I’d suggest reading it all in order if possible.

The Dilemma:

Long ago, far away, on the planet “Mervin”, the people were terrorized by a group of rogue cops. Every once in a while the cops would secretly arrest two people, keep them separated from each other, not allow them to communicate with each other, not allow them to communicate with anyone on the outside, not inform anyone on the outside that they were arrested, charge them with a phoney crime, and gives them a choice:

(1) Rat on your fellow prisoner or (2) clam up.

If one prisoner rats on the other and testifies in court that the other committed a crime, and if the other prisoner does nothing but clam up and maintain his innocence, the rat will go free, and the other will be sentenced to 20 years in prison.

If both prisoners rat on each other, then each gets a 5 year sentence.

If both prisoners remain silent, both will serve 6 months.

These are the circumstances of a problem in Game Theory called “The Prisoner’s Dilemma”.

Prisoner Point of View:

With our two prisoners named Alice and Bob, we have four possibilities.


Alice clams up. Bob clams up. => Alice 6 months. Bob 6 months.
Alice clams up. Bob rats. => Alice 20 years. Bob goes free.
Alice rats. Bob clams up. => Alice goes free. Bob 20 years.
Alice rats. Bob rats. => Alice 5 years. Bob 5 years.


If you’re Bob, what do you do? The first question is “What is your motivation?” Most likely you want to minimize your time in prison. But do you care about Alice? Is Alice your wife and you want to minimize her time. Or maybe Alice is your ex-wife and you want to maximize her time? Those are alternative scenarios, really. But they might help you consider that you as a prisoner might have different motivations depending on the circumstances, as might the other prisoner.

In the pure Prisoner’s Dilemma, you don’t know who the other prisoner is. And you’ll never see them again after you’re released. We will assume that since you don’t know who the prisoner is and probably don’t know them, that you will try to minimize your sentence and not care about their sentence.

This also means that the other prisoner probably doesn’t know you and doesn’t care about you. They’re only interested in minimizing THEIR sentence and don’t care what happens to you. Keep in mind that the other prisoner isn’t going to choose what is best for you, but what is best for them.

This can be hard for some people to grasp. Some will tend towards solutions where they assume the players would cooperate when no such incentive for them exists. A person might treat the prisoners as slaves to the outcome the person wants, rather than allowing the players to act in their own best selfish interest. If you’re having difficulty, then perhaps you need to imagine two players with sufficiently strong and obvious motivation. If it helps, pretend that the prisoners are two convicted criminals who both have long track records, and who don’t know each other, and don’t care about each other. Both will do whatever it takes to get out of prison as quickly as possible.

Actually, people have troubles with that too, because they can’t imagine themselves as a hardened criminal. So, how about this, imagine you are you, and you are a prisoner. And imagine that the other prisoner is someone who has a long list of violent crime convictions under his belt. If he can get out of jail as soon as possible, and has to throw you under the bus to do it, he will.

You’re Bob, the prisoner. What do you do?

First of all, just to clear this up right off the bat: these rogue cops are like a force of nature. They can’t be killed. They can’t be fought. They can’t be avoided. They’re like agents in the Matrix, and none of us are Neo. Obviously, if Barbary pirates are harassing your ships, you could send in the marines to have a polite conversation and straighten things out. But in the case of the prisoner’s dilemma, the rogue cops cannot be overpowered. So, you’ve got to choose to rat out or clam up.

You’ve got four possible prison sentences: get out immediately, get out in six months, get out in 5 years, get out in 20 years.

You’re interested in getting out immediately. To do that, you have to rat on the other prisoner, and they have to clam up. So you rat. But the other prisoner knows that if you rat on them and they clam up, they get 20 years, something they really don’t want. But if you rat on them and they rat on you too, they only get 5 years. 5 years is better than 20 years, so it is in their best interest to rat on you if you rat on them. In trying to get out immediately, you end up with a 5 year sentence.

The next best alternative to getting out immediately is getting out in 6 months. 6 months is better than 5 years, so you consider trying to get out in 6 months. To get out in six months, you have to clam up and Alice has to clam up. In that case, both of you get six months. But if you, Bob, clam up, Alice knows that if she rats on you, she’ll get out immediately. Getting out immediately is better for Alice than Alice getting out in six months. So Alice does what’s in her best interest and rats on you. She gets out immediately. And you get 20 years. In trying to get out in 6 months, you end up with a 20 year sentence.

If you try to get out immediately, Alice acting in her best interest will cause you to stay in prison for 5 years.

If you try to get out in 6 months, Alice acting in her best interest will cause you to stay in prison for 20 years.

Those are your two choices. And if you assume Alice is operating in her best interest, you have to settle for getting out in 5 years.

If you accept 5 years, you’ll rat on Alice. If, for whatever reason she is motivated to clam up, you’ll get out immediately. But if Alice is acting in her best interest, the best you can hope for is 5 years. That’s your best option.

Two prisoners, both operating in their best interest, and with no concern for the other prisoner, will rat each other out in this situation.

What makes it a bit of a “Dilemma” is partly the fact that with Alice and Bob both acting in their own best interests, neither can get the two best possible outcomes, getting out immediately, or getting out in six months. If both act in their own best interest and if there is no relationship between the two prisoners before, during, or after, no repurcussions for their decisions, then the best they can hope for is 5 years by betraying each other.

Again, some people will look at this and assume that the prisoner’s would automatically cooperate. That isn’t the case. If you’re having trouble, imagine Alice is a harderned criminal, and doesn’t give a hoot about you. Alice is solely interested in her own self interest. Hopefully it is obvious tht Alice will do whatever she can do to get out of prison as quickly as possible.

In that case, if you’re Bob, then you have to assume that Alice will squeal like a rat if it will let her get out of prison tomorrow rather than 5 years from now. And if you don’t know who Alice is, you have to assume she will act selfishly. Which means you have to act selfishly, or, if you don’t like that word, then “defensively”. Which means in a single, stand-alone game of the Prisoner’s Dilemma, two actors inten on their own best self interest will get the third worst result possible. The options are getting out immediately, in 6 months, in 5 years, and 20 years. Two selfish people will choose to rat each other out and both will get 5 years.

Because the alternative is for you to cooperate, and then Alice could still rat you out and she would go free immediately. So, she’s motivated to rat you out in case it lets her out immediately.

Iterated Prisoner’s Dilemma:

A variation exists called the “Iterated Prisoner’s Dilemma”. In this variation, you are paired with the same prisoner over and over (iterated). What happens in this variation can be interesting.

Say Alice and Bob are put through the same Prisoner’s Dilemma over and over. In a single iteration, as explained above, your best choice is to rat out the other prisoner and get 5 years. But say at some point, for whatever motivation there might be, Alice clams up. What does that tell you, Bob? If Alice were operating solely out of her own self interest, she would always rat on Bob. If Alice clams up, she’s taking a chance that she’ll get 20 years in prison. And the first time she does clam up, odds are that Bob will rat on her, and she’ll get 20 years.

Alice/Bob

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

(at some point Alice or Bob wonders if they’re in an iterated prisoner’s dilemma and one of them will test to see if they are)

clam/rat

20y/0y

At this point, Bob saw that Alice just clammed up. If Bob assumes that the other prisoner is always the same person, then the question to Bob is this: Why would Alice clam up when she knows she’ll get 20 years?

It could be that Alice is trying to send a message to Bob.

What is the message from Alice to Bob?

“I’m here and willing to cooperate with you if you cooperate with me”

So they go through another round. Bob’s decision making process is now different than a single iteration. He now knows that Alice may have risked 20 years to send a message offering cooperation to him. She wasn’t acting solely in her best interest. She was looking to improve the situation for both of them.

If Alice clams up again, Bob realizes that if he also clams up, they go free in 6 months.

And 6 months is better than 5 years.

But Bob also realizes that every time Alice clams up, it’s a chance for him to get out immediately by ratting on Alice.

So the next round comes up.

clam/rat

20y/0y

Alice tried to work with Bob, but Bob instead operated purely on his own self interest and without regard to what it did to Alice. Bob got out immediately. Alice got another 20 years.

For the next few rounds, Alice will likely not try to cooperate with Bob.

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

Bob may suddenly realize that he is getting 5 years in jail now that Alice isn’t helping him, and he could be getting 6 months if he worked with Alice. He might signal his willingness to cooperate with Alice by clamming up.

rat/clam

0y/20y

At this point Alice might consider working with Bob, or might wonder if it’s a trick of some kind. At some point they may both cooperate with one another.

clam/clam

6m/6m

If they continue this cooperation, they will now get 6 months sentences instead of 5 years.

clam/clam

6m/6m

clam/clam

6m/6m

clam/clam

6m/6m

If at any point one of the players decides to try to rat on the other, cooperation will likely end for several turns.

clam/rat

20y/0

Bob just ratted on Alice. He got out immediately. She got 20 years. Alice realizes that Bob isn’t cooperating and is instead operating in his own best interest, so Alice operates in her best interest. They both rat on each other and get 5 years each.

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

rat/rat

5y/5y

At some point one of the players might attempt to reestablish communications by clamming up, taking a 20 year sentence in order to signal to the other player their willingness to cooperate again.

clam/rat

20y/0

clam/clam

6m/6m

clam/clam

6m/6m

clam/clam

6m/6m

And on and on it goes, iterated.

The Iterations Add Up to Form a New Game

The way to look at this is not as separate iterations of the same game, but rather one long game that takes place over time. The one-time Prisoner’s Dilemma is a single shot game. It’s like playing Rock-Paper-Sciccors. You decide which shape you’ll take, and then on the count of three, you reveal it.

Iterated games are like one big game that takes place over several moves. Instead of Rock-Paper-Sciccors, think something like tic-tac-toe or checkers or chess. What really matters is the final totalled score, not just your most recent iteration.

Say you’ve got several players who each has a different strategy.

Robert has a simple strategy. He always rats. No matter what.

Charlie has a strategy of offering to cooperate once in a while, and if cooperation isn’t reciprocated, he stops cooperating for a while and rats.

When the two play against each other, they don’t know who the other player is. They don’t know what strategy they’ll use. No one ever knows who they’re playing against. You just see how they choose.

So, the game starts out with both players ratting each other out. At some point, Charlie tries to cooperate by clamming up. Robert doesn’t care. He always rats. Which means that most of the time, Charlie and Robert get the same five-year sentence, but every once in a while, Robert gets out immediately and Charlie gets 20 years.

Charlie/Robert
rat/rat
5/5
rat/rat
10/10
rat/rat
15/15
clam/rat
35/15
rat/rat
40/20
rat/rat
45/25
rat/rat
50/30
clam/rat
70/30

We can see that over time, Robert does a lot better than Charlie, because he always rats out Charlie. This is why it’s best to rat out your other player in single-play. In multi-iteration, using the same rat-stategy will allow you to do at least as good as, if not better than, your opponent.

If Robert plays against another Robert (always rats out), they will both always rat, but neither will ever get ahead of the other.

rat/rat
5/5
rat/rat
10/10
rat/rat
15/15
rat/rat
20/20
rat/rat
25/25

If Charlie plays against another Charlie (cooperator) , then when they cooperate, they’ll both accumlate a much lower amount of time in prison per turn.

rat/rat
5/5/
rat/rat
10/10
rat/rat
15/15
clam/rat
35/15
at this point, Charlie clammed to offer to cooperate with his other player.
The second Charlie sees it at a sign to cooperate, and clams up.
clam/clam
35.5/15.5
At this point, if they cooperate, they’ll only get half a year per turn
clam/clam
36/16
clam/clam
36.5/16.5
clam/clam
37/17

At this point, the strategy you use is in part determined by what sort of strategy the other prisoner uses. If they decide to always rat, and you try to cooperate but fail, then at some point, you should switch to always ratting. Otherwise, you’ll keep getting 20 year sentences and the other will get out immediately.

If you’re lucky enough to get paired up with someone who cooperates, then you can cooperate and minimize how much prison time you accumulte per turn.

This is an odd twist to the game. The restrictions of Prisoner’s Dilemma prevent communication beforehand between the players. There is no way to know who the other play is or what general strategy they will use. Therefore your best bet is to assume they’re the worst case, most selfish player and choose to rat him out. Maybe he won’t rat you, but if not, good for you, you get out immediately.

In an iterated game, how you play can depend on the strategy exhibited by the other prisoner. You can’t communicate before hand to find out what sort of player they are. So you have to watch how they play and then go from there. If they always reject your attempts at cooperation, then you should stop cooperating yourself. If they cooperate from time to time, but occaissionally rat you out, then you have to watch out that they don’t get too far ahead of you by betrayign your trust.

All of which boils down to your strategy in an iterated Prisoner’s Dilemma depends on the other prisoner. Once you’ve established how the other prisoner plays, you can try to cooperate or not.

Iterated Prisoner’s Dilemma in a Population

So the Iterated Prisoner’s Dilemma is like a single game of chess. It’s a bunch of moves over time that accumulate into some final score. How you play will depend on how the other person plays.

So, now, lets expand it one more level. Imagine instead of playing multiple iterations of a single game with one person, you instead are playing many games, each with multiple iterations and each game is with a different person. The prisoners are selected from an entire alphabet of people: Alice, Bob, Charlie, Dave, Eve, Fred, George, Harry, Ida, Juliet, all the way down to Zed and beyond.

Different cops pick you up and drop you into different prisons. Sometimes the prisons start looking familiar, so you start wondering if they’re picking up the same people each time. Maybe the other prisoner in that particular prison is always the same prisoner. Maybe it’s always a different prisoner. You can’t find out until after you try several iterations. At first, you’ll rat out your fellow prisoner. But after a few iterations in the same prison, you might try to cooperate and see if they follow along. It may be that it’s always the same person, and they get your message for cooperation. Or it may be that it’s always different prisoners that are paired up, so you can never predict what their strategy is.

If you use the chess metaphor, it’s like you’re in the park, with several different chess boards in front of you, but you can never see who is on the other side of the board playing against you. Each board might have an assigned player, and you can try cooperating. Or each board be played by random people walking through the park, and they only play one move, and then leave.

Now, at this point, you might be asking yourself, what the heck is the point? You don’t live in a world where rogue police officers arrest you and another prisoner over and over again and keep giving you this same offer. You may play chess, but you don’t play with a bunch of different boards in the park and not know who the other player is. So, what does the iterated prisoner dilemma in a population do?

It’s getting close to a model for laissez-faire capitalism.

Laissez-Faire Capitalism

Laissez Faire capitalists don’t want any sort of government regulation on commerce. Instead, they want consumers to work with retailers through a form of the iterated prisoner dilemma within a population.

For example, Alice is our customer. She goes into a Laissez-Faire grocery shop, which takes the place of Bob. Alice buys some apples from the shop. They’re full of worms. She buys apples somewhere else. They’re full of worms. She buys apples somewhere else. They’re pretty good. Alice continues to buy at that shop every week, iterating the cooperation between them. Alice gives the shopkeeper money, the shopkeeper keeps giving Alice good apples. They keep cooperating.

As soon as the shopkeeper stops selling good apples, Alice stops cooperating with the shop. Alice goes and finds a different store.

This iterated prisoner’s dilemma in a population fairly closely models Laissez Faire capitalism.

Alice is playing several different iterated games with many different shops. She’s got the grocery store, the gas station, the hardware store, the car dealer, the auto mechanic, the clothing store, a bunch of restaurants, all the stores in the mall, and so on.

Alice is playing a different iterated prisoner’s dilemma game with each store. She doesn’t know how the other stores will behave. She doesn’t know if they will cooperate. She doesn’t know if they will try to cheat her. And her only way to find out is to play several turns with the store and see how they act. Alice has to keep all these different games in mind as she shops.

Caveat Emptor.

Laissez Faire capitalists want the world of economics to operate this way. They want the only communication between Alice and all those stores to be her money and their goods. They do not want Alice to be able to force the store to do anything. They do not want Alice to be able to force the store to play a certain strategy. They do not want anyone to be able to force the stores to cooperate. The only recourse Alice has to spending money at a store and getting shoddy goods is to stop shopping there and go somewhere else.

And the Laissez-Faire capitalists will tell you that shops will then be motivated out of their own self-interest to cooperate with their customers. They call this the “Invisible Hand” of the Free Market.

Think about that for a moment. When you’re playing iterated Prisoner’s Dilemma with a single person, your strategy depends on their strategy. You will have to risk cooperation, and risk 20 years in prison, to find out if they are willing to cooperate. If they are, then you can cooperate, but at any moment, they may be motivated by short term greed to rat you out. At which point you have to rat them out for a while. If they never cooperate, you would be a fool to extend an attempt at cooperation to them, because they’ll just rat you out and go free while you stay in prison for 20 years.

How you play iterated Prisoner’s Dilemma with one person depends on how that person plays and depends on whether or not they are willing to cooperate. Earlier you saw Robert playing against Charlie. Robert always ratted. Charlie would try to cooperate. At the end of the game, Robert was way better off than Charlie.

How you play depends on the other player’s strategy. That doesn’t change just because you start to play multiple games of iterated Prisoner’s Dilemma with different people. It actually becomes more difficult because you have to keep track of every individual store’s strategy. And at any point, they would get a short term benefit for ratting you out.

So, how you play iterated Prisoner’s Dilemma in a Population still depends on the strategy of each person you’re playing against. There is no singular “best” solution that says people must cooperate.

And even if it were true, some stores only have one-time buyers. People buying a house usually intend to buy it and stay there for thirty years. If they sell it and buy a different house, odds are they’re not buying it from the same person who sold them the first house. People on the road might stop at a gas station and buy gas at a place they’ll never come back to. Other people are allergic to peanuts and will die if they eat them, so can’t take a risk at a store that maybe they didn’t label their food products properly.

For all your purchases, customers would have to always be distrustful of the shop failing to cooperate and taking advantage of them. They’d have to check everything they buy for hidden flaws. They’d have to check all the hamburger meat to make sure it doesn’t have sawdust or kangaroo meat in it. They’d have to check their gasoline to make sure it doesn’t have any water or other thinners in it. When they buy a case of beer, they’d have to check each bottle as they drank it over time to make sure one of the bottles didn’t have something nasty in it. And they’d have to track this for each store, since each store is a different player, and a different potential strategy.

And the only way to communicate in Laissez-Faire capitalism is with your money. You as a customer must risk purchasing something and can only tell if it’s a quality product after you purchase it. No refunds. No returns. No nothing. If you don’t like it, your only allowed recourse is to go shop somewhere else.

So, the constant mental state of a laissez-faire capitalist would have to be a constant distrust that his fellow man is about to screw him over, a constant skepticism of humanity, a constant low level paranoia.

This reflects the personality of most of the laissez-faire capitalists that I’ve run into. The more distrustful and paranoid they are, the more they insist on absolutely no government oversight on commerce.

The Population Responds Proactively to the Iterated Prisoner’s Dilemma

Back in the land of Mervin, the people start to get the notion that the entire population is being raided by rogue cops at random intervals. Random people are being picked up, put into holding cells and told to either rat on their fellow citizen or clam up, and then are given a sentence.

But the people start to figure out what is happening to them and try to minimize the damage. One option is to create their own prison system and then institute a law that says anyone picked up by these rogue cops must cooperate with their fellow prisoner, and then they’ll both only stay in for 6 months. It isn’t a perfect solution. But it’s certainly better than 5 years. But how do they enforce it? If you rat on your fellow prisoner, you go into the local Mervin prison for 30 years.

Suddenly, the game theory table changes. (mp) represents a sentence in Mervin Prison.


Alice clams up. Bob clams up. => Alice 6 months. Bob 6 months.
Alice clams up. Bob rats. => Alice 20 years. Bob 30 years (mp)
Alice rats. Bob clams up. => Alice 30 years(mp) Bob 20 years.
Alice rats. Bob rats. => Alice 5 years. Bob 5 years.


If you’re Bob and want the best outcome for yourself, the best outcome is 6 months, which is to clam up. If Alice thinks you’re clamming up, her best response is to clam up too, and you both get 6 months. If you clam up and Alice rats on you, you get 20 years, which sucks, but Alice gets 30 years in Mervin Prison as soon as she gets out of Agent Smith’s prison. So, if Alice thinks you’re clamming up, she would want to clam up too.

If you’re Bob and you consider ratting on Alice, you know there’s 30 years in Mervin prison waiting for you whenever the Agents let you out. And 30 years is much worse than 6 months, so you don’t want to even think about ratting on your fellow prisoner.

This makes the solution stable. No matter how greedy you or the other prisoner are, the best solution for the both of you is to clam up.

This law created by the people of Mervin is called a “Strategic Move”. It alters the cost of a choice by the players so that it changes the game’s outcome. In the case of the Prisoner’s Dilemma, it makes the game have a stable solution, and more importantly, that solution is the best one could hope for if everyone were cooperating in the unregulated version of the game.

i.e. if everyone were cooperating in an unregulated game, everyone would get 6 month sentences. By regulating the game, everyone still gets 6 month sentences. But by regulating the game, everyone cooperates even if its a one-time game, a short iteration game, an iterated game with the same people, or a randomly iterated game with people randomly selected from a population.

Roberto the rat would want to change his strategy. He always ratted out his fellow prisoner because he would in the worst case tie with the prisoner and best case come out a lot better than the other prisoner. With this regulation added, Roberto no longer wants to rat because he will come out far worse than his fellow prisoner. So, Roberto changes his ways and adopts a strategy of cooperating.

Charlie the cooperator can now cooperate on the first round, knowing that any prisoner on teh other side will want to cooperate with him and get six months rather than rat him out and face 30 years back in Mervin’s prison.

Regulating the Laissez-Faire Prisoner’s Dilemma

What happens when you apply regulation to the Iterated Prisoner’s Dilemma in the population of Mervin that is the economy?

Alice no longer has to be paranoid about whether or not the apples in the store are safe. The Food and Drug Administration has minimum standards for cleanliness and so on for food. And Peter doesn’t have to worry whether a candy bar has peanuts in it or not, because the FDA has strict labeling requirements for peanuts and other ingredients that cause allergic reactions. And so on. THe people of Mervin implement regulations that remove any short term benefit the store might gain for betraying it’s customer, and then a minimum level of trust is implied in transactions.

This is especially useful for people traveling across the country and who might by gas once in some station in Ohio and who can’t really use their purchasing dollars as a communications channel in an iterated transaction. There’s only one transaction, but people can still trust a minimum level of quality or assume that the regulation agencies in Mervin will impose a fine that would cost the company more than whatever short term benefit they gained from cheating their customer.

Seems pretty straightforward.

Then why are their so many Laissez Faire capitalists on the internet? Well, for one, the internet attracts all sorts of crazy kooks. For another, the regulatory agencies in Mervin have to somehow answer to the people of Mervin. Which means there really is no “best” set of regulations for many transactions. So not everyone is going to agree. This means folks have to learn how to negotiate with a group of people they don’t agree with, on the internet. And most people know that usually ends in kicking, screaming, a diaper change or two, and someone put in the corner for a timeout.

Laissez-Faire capitalists want to negotiate with an extremely limited channel, money. They tend to have two messages: I will buy from you. I won’t buy from you. That’s it. If someone with this extremely limited messaging system is thrust into a world of real people trying to agree on some minimum standards for a product, their brain implodes. They can handle “Yes” and “No”. When they’ve got millions of people all putting in different opinions of what the minimum standard should be for, say, mining operations, the possible options are limitless. How much arsenic do you want to allow into the water supply? How much mercury? Do you measure it by an absolute amount (tons per year) or something relative like pounds per ton of coal produced. It seems like the yes/no capacity of the laissez faire capitalist gets overwhelmed by an infinite number of options and a population where everyone has a different opinion.

This simplistic yes/no filter towards the world might actually explain in part why the person is drawn to laissez-faire capitalism in the first place. It’s a lot easier to model it in your mind. Everyone can do what they want or don’t want. And communication is limited to yes/no messages. There are no complexities of trying to create a system of government, figure out how to count votes, create political offices of representation, and deal with law passages.

If you cannot understand the complexity of a regulated Iterated Prisoner’s Dilemma in a Population, then maybe you will convince yourself that the Iterated Prisoner’s Dilemma forces people to cooperate anyway (via the Invisible Hand, usually) and pretend we don’t need any of the stinking regulation and gummint.

Laissez-faire capitalists often want a rudimentary government with an almost dogmatic approach to law. Physical violence is illegal. Force is illegal. Everything else is allowed. The end. Again, this may very well reflect the fact that laissez-faire capitalists can’t hold in their mind something much more complex, like government in any modern first world country today. These countries are far more sophisticated than the laissez-faire capitalist would ever want, but these governments produce far more benefits than the rudimentary, yes/no, no violence, no force, everything else is allowed, tree-house clubs that the laissez-faire folks use to model the world.

But even more fundamentally, as I mentioned before, it’s been my experience of the extreme libertarian individuals that I’ve run into that they exhibit a certain level of mistrust of their fellow man, mistrust of any transactions, and certainly mistrust of anyone else making a strategic move to alter their choice of actions. THey don’t trust the watchers, because they don’t trust people in general.

But interestingly enough, their arguments against regulation are arguments for the individual right to betray another.And this betrayal is usually in a short sighted choice of betraying trust in an iterated prisoner’s dilemma.

For example, a extreme libertarian, say Larry, might complain that government-regulated apples have to be a certain minimum quality now, and Larry the Libertarian complains that he could make more money if there weren’t all this apple-regulation. He could make more money if he could sell apples of lesser quality, rather than having to throw them out. But if Alice is buying apples, she wants a certain minimum quality, and if Larry sells her shoddy apples, Alice is betrayed in the transaction, and Larry is the Betrayer. Larry, like many extreme libertarians, will argue that it’ll all settle out all on its own if enough iterations of unregulated prisoner dilemmas are performed. i.e. if enough apple buyers and apple sellers exchange enough apples and money that the store with the best cheapest apples will make the most profit, and the others will go out of business or improve their apple quality.

But not everyone buys apples repeatedly from the same place to make a substantial effect on the seller. Many times the transaction acts as a single play Prisoner’s Dilemma and the two players will never see each other again.

The Laissez-Faire capitalist will tell you that a “bad reputation” will follow bad vendors and make their sales go down. But now they’re replacing government regulations that people have input on to some sort of vague “reputation” system that somehow allows everyone to know about not only every store, but every brand within that store. And does so in a fair and honest method, and does so for zero money. And do it perfectly.

How do I know Laissez Fair capitalists claim their reputation system works for free and operates perfectly? Because Laissez Faire capitalists demand that no one be taxed to pay for government regulations and enforcement agencies. They will tell you that such overhead isn’t needed. The market will magically (perhaps by their magical “Invisible Hand”) sort it all out without any additional costs. So this reputiation system will magically operate at zero cost. And why will it be perfect information? Because any Laissez Faire capitalist will tell you that any Government Agency will always make mistakes. And they can’t stand bureaucracy because they are always making the wrong decision. Therefore, on can assume that this reputation system will contain perfect information about every possible item in the world that is for sale. And distribution it perfectly, and without cost, to every person in the world.

Don’t Tread On Me!

Another Laissez Faire argument against any form of regulation is the “Don’t Tread on Me!” response. Regulation is a form of Tyranny, they say. Often what happens is that ultra libertarians like Larry will talk about their Freedom and Liberty being trampled by regulatory groups. But what it really means is Larry wants the right to screw someone over without the law telling him he can’t and fining him for trying it.

Imagine Larry in the land of Mervin after the local prison is created and the law that says cooperate or go to prison for 30 years. At some point, Larry gets picked up by the rogue cops, and knows that the laws of Mervin say that if he doesn’t cooperate with his fellow prisoner, he’ll get 30 years. So, he cooperates. He and the other prisoner get 6 months. When Larry gets out, he HOWLS with the injustice of the laws of Mervin forcing him to serve 6 months in the agents prison when he could have gotten out scot free in zero time. The laws of Mervin, Larry will proclaim, have stolen his liberty and freedom.

The problem is, Larry wants the freedom to betray his fellow prisoner, get out of jail immediately, let his fellow prisoner go to jail for 20 years, and then have the people of Mervin not hassle him about betraying of of their citizens. That’s what he’s calling his “liberty” and “freedom”. It’s his freedom to screw his fellow citizen.

The people of Mervin created the law in order to get the best deal for everyone, 6 months for both prisoners. Larry is complaining because he sees he could have gotten out immediately if it hadn’t been for the regulations. What Larry fails to grasp (or just doesn’t care about), is that to get out immediately, the other prisoner, his fellow citizen of Mervin, would have to have stayed in prisone for 20 years.

In short, Larry is complaining that he couldn’t get what was best for HIM because the regulations tried to make sure that the result was what was best for EVERYONE. Poor Larry. The people of Mervin just keep trampling his rights, his liberties, his freedoms.

I’m not even sure exactly how Larry justifies the whole transaction in his mind. I think it’s because he is so shortsighed that he doesn’t see the betrayal he’s committing against his fellow prisoner. All he sees is what’s in it for him. He could get out immediately if he rats out his fellow prisoner, or he could sit in jail for 6 months. I don’t think Larry, or the extreme libertarians or extreme right wingers, really consider anyone’s welfare beyond their own.

It’s like their capacity for empathy, for seeing things from other people’s shoes, got stunted at age 4 when “Mine!” was the battle cry.

Larry can only see the world in terms of what’s best for him. So he thinks everyone else operates that way too. Larry cannot grasp that when the people of Mervin made that law requireing everyone to cooperate, they were all agreeing to do what was FAIR FOR EVERYONE, rather than to pick one person and say he gets to go immediately and the other has to wait 20 years.

The Invisible Hand

Another common Laissez Faire argument against government regulation is that the “invisible hand” will sort it all out without any regulation.

The Invisible Hand is a fantasy in laissez-faire capitalism that says in a completely unregulated market, buyers and sellers will cooperate in their own self interest and reach the most efficient outcome possible. Using Game Theory terms, they claim that the Invisible Hand causes unregulated iterated prisoner’s dillemma in a population to always cooperate out of their own selfish interests.

But we’ve already shown this isn’t true. In a non-iterated game, players will tend towards ratting each other out for fear that the other player will try to get out immediately and keep them in for 20 years. 5 years is far better than 20, so they rat.

And very few economic transactions are an iterated Prisoner’s Dilemma between the same two players, with an infinite supply of alternative stores to go to if they try to cheat you. You may pay your cable bill every month, which is an iterated game, but if there is no other choice for service, then the one thing a Laissez Faire person could do when the other player cheats them is removed. They’ve got no other vendor to go to for the same product or service.

If there are few alternatives for a vendor, that alters the game to the point that the vendor can cheat teh customer, but the customer cannot leave and go spend their money somewhere else. In an unregulated iterated prisoner’s dilemma in a population, your only means of communication is to spend money or not spend money. And if you can only get cable TV through one company, then your one means of communication has been severed.

The Unregulated Iterated Prisoner’s Dilemma only achieves cooperation over time. And it only works if both players are cooperative. If one of the players is Robert the Rat, and if he’s the only player in town for the service you want, well, again, the game theory shows the customer gets ratted out by the vendor and no “invisible hand” comes in to fix it.

The “Invisible Hand” of capitalism is a fairy tale that laissez-faire capitalists use to convince themselves that even one-time unregulated economic transactions will result in cooperation. And we’ve already shown that they won’t.

Unregulated Iterated Prisoner’s Dilemma in a Population yeilds non-ideal outcomes.

In the real-world of economics with many buyers and many sellers, most transactions are not part of an iterated game. In the real world of the free market, there is no invisible hand to encourage folks to cooperate and achieve the best outcome.

The invisible hand doesn’t stand up to the game theory of the Prisoner’s Dilemma. The invisible hadn doesn’t exist.

In their defense, the term “Invisible Hand” was invented by Adam Smith in his book “The Wealth of Nations” published in 1776. It wasn’t until Antoine Augustin Cournot published “Researches into the Mathematical Principles of the Theory of Wealth” in 1838 that the general concept of Game Theory came into being. And even then, game theory did not really exist as a unique field until John von Neumann published a series of papers in 1928. The thought of the Prisoner’s Dilemma didn’t exist until Merrill Flood and Melvin Dresher working at RAND in 1950 came up with the concept. (Merrill + Melvin = “Mervin”, in case you were wondering)

Adam Smith used the term “invisible hand” in a very limited sense. The term only occurs once in his book, and in that situation, it was used to describe how a businessman would be inclined to trade locally (rather than internationally) out of an affection for his fellow countrymen. It turns out that in the modern world where American jobs are getting exported to countries with cheaper foreign labor, that isn’t always the case.

Economists then took this “Invisible Hand” term and used it to describe something far greater than Smith ever intended. They used it to describe a fantasy world where unregulated, free markets would always tend towards the most efficient solution due to this “invisible hand”.

They called it that, because they didn’t really have a formal way to study it, such as an unregulated iterated prisoner’s dilemma in a population. What the formal approach shows us is that there is indeed a tendancy for cooperation when there are enough iterations going on in a small enough population that buyers and sellers know each other. But this isn’t an accurate model of the whole economy.

Not every transaction is part of a long chain of iterated transactions. Not every buyer or seller is going to automatically cooperate because some “invisible hand” made them do it. And the “invisible hand” doesn’t change the game theory table of a prisoner’s dilemma so that suddenly it’s in everyone’s best interest to cooperate.

As shown above, in one-time transactions and in short iterations or spotty iterations, prisoners revert to their own self interest, based on the assumption that the other prisoner will try to operate in his own self interest. If you try to cooperate, but he rats on you, you get 20 years and he goes free. The “invisible hand” doesn’t suddenly impose a punishmen on the rat.

That’s what regulation does. It imposes a punishment on the rat to the point that it is in the rat’s best interest to cooperate. It doesn’t happen by some magical invisible hand. It happens because the people of Mervin mandated it to be so.

The “invisible hand” may very well be pointing to something economists had a gut feeling understanding about, the notion that iterated prisoner dilemmas tend towards cooperation. But the economy isn’t all that iterated. It’s a lot of one-time or few-time transactions with different people. And the game theory shows the “invisible hand” doesn’t encourage cooperation in those instances.

If I had to take a stab at what the “invisible hand” is in the minds of Laissez Faire capitalists, it’s some notion of “reputation” that occurs outside the yes/no comunications channel. And this “reputation” collects for each vendor’s behaviour, and then customers avoid the bad and buy from the good vendors. But it is some sort of “natural” proces. Any sort of formal, people organizing to make it happen, will usually anger the laissez fair capitalist. The government certainly can’t do it, because the Laissez Faire capitalist will tell you the government can’t do anything right.

So Laissez Faire capitalists take one time and few iterated prisoner Dilemma games within a population and with a wave of their magical “invisible hand”, they transform them all into Iterated Prisoner Dilemmas between two individuals.

Anyone who invokes the “Invisible Hand” doesn’t understand the formal study of economics using game theory and the Prisoner’s Dilemma. And has a rather slippery grasp of reality.

Government Overhead Versus Personal Sacrifice

The other argument of a laissez-faire capitalist against regulation is that government regulation creates overhead and therefore they claim it is impossible for governmetn regulation to be as cheap or good or whatever as a Laissez-Faire system. They will claim that an unregulated iterated prisoner’s dilemma will always be better than a regulated iterated prisoner’s dilemma.

This argument contains two implicit premises and both of them are false.

Premise number 1: an Unregulated Iterated Prisoner’s Dilemma in a Population will naturally enforce cooperation. i.e. the “Invisible Hand”.

Premise number 2: This “invisible hand” is 100% efficient, has no overhead, creates no wasted costs, etc.

i.e. They claim that the Invisible Hand is (1) perfect and (2) free.

Well, they’re wrong on both counts.

(1) We already saw that in an Unregulated Iterated Prisoner’s Dilemma in a Population, that ratting out your fellow prisoner is still an option. When Robert the Rat is paired up with Charlie the Cooperator, Charlie gets screwed. Cooperation is not required by the game. And most transactions are not iterated repeatedly with the same vendor anyway. Many transactions are one-time or once-in-a-blue-moon exchanges. At which point, the result is the third worst solution. 5 years for both.

(2) There is waste associated with the “Free Hand”. Every time Charlie goes into a new store, he’s playing a new Prisoner’s Dilema with a new player. He has to risk cooperating, risk buying some merchandise to find out if it’s really good or not. Some stores won’t be good quality. And Charlie will have to go somewhere else and try to find a good quality product. Every time Charlie buys something shoddy, that’s the OVERHEAD of using the Free Hand to get cooperation, rather than using government regulation.

The Laissez Faire types often talk about how a store’s “reputation” will force them to stay in line. Apparently this reputation system will be (1) perfect and (2) free as well. Perfect in that it will cover every brand of product in every store, and report accurately in every case, and it will be freely available to every customer. Except that’s impossible.

Health Care as an Iterated Prisoner’s Dilemma in a Population

This whole thing about the regulated iterated prisoner’s dilemma in a population came to me as the health care debate was raging. At which point I realized that health care is probably one of the best, real world examples of a prisoner’s dilemma.

Instead of police rounding people up randomly, it’s disease aflicting people randomly.

Instead of getting sentenced to differnt years of prison, people are forced to pay different amounts of money to pay for their, and possibly others, health care.

What happened is that the people of Mervin started getting sick. When someone got really sick it usually cost them lots of money to get better. Say $20k. (instead of 20 years). There are lots of people in Mervin, and only some of them would get sick every year. Eventually people started noticing an iterated pattern of getting sick. And they wondered if there was some way they could deal with it proactively, before more people got sick.

Turns out the solution is cooperation. Everyone pitches in a little bit of money into a pool. When someone gets sick, they use that pool of money to pay for treatment.

The problem is that some healthy people don’t want to pitch money into this common pool. They want to keep their money and take their chances on not getting sick. And they don’t want any meddling government to tell them they have to do that.

So, what happened is that some people tried to start their own private pools of money. Private insurance corporations. But the problem was all that did was move the cheating to the insurance corporations. People put money into the pool. Then when someone got sick, the corporation wouldn’t pay for their health care. Some of the people who stayed healthy didn’t mind the corporations doing this too much because these healthy individuals didn’t really want to pitch money into the pool as long as they were healthy. As long as it wasn’t them.

A number of health insurance companies in California have been found to turn down between 20% to 40% of all health insurance claims.

This is simply a reflection of the individual desire to cheat and betray their fellow prisoner in the scenario. They get away scot free, and the person they betrayed gets 20 years.

What this clearly indicates is that the solution to the health care problem must be a government solution. Regulations must change the game theory chart in a strategic move so that betraying the sick is not rewarded. If you are an insurance company and you fail to pay a valid claim, you get fined for more than the claim was. If an insurance company refuses to allow a person to join their ranks due to a preexisting condition, they pay a fine larger than they would have saved. And so on. The incentives to cheat your fellow human beings must be removed at a regulatory, universal, level.

And a government option to cover the 40 million or so uninsured should be made available so that people aren’t using the emergency room for treatment, raising costs for everyone else. Preventitive care is much cheaper than letting the problem get so bad that it needs treatment in an emergency room, and then teh hospital is stuck with an unpaid bill that gets passed on to its other patients. Preventive care, routine care, is far cheaper than filling the emergency room with uninsured people who can’t pay their bill but couldn’t afford to get preventative care.

This means that regulations will have to be handled politically, which means that no everyone will be happy with how they are handled. But that doesn’t mean the regulated solution will be worse than the unregulated solution. What it means is people will have to do some work to keep it on track, but it should be easily possible to make government mandated healthcare reform better for EVERYONE.

The interesting thing here is that the people most strongly objecting to Health Care Reform are objecting on the same grounds that people object to any regulated iterated prisoners dilemma. The complain the same way the laissez faire extremists complain. FREEDOM! LIBERTY! They shout. They shouldn’t have to have health care reform done to them against their will. But as the Prisoner’s Dilemma shows, what they’re really demanding is the right to betray their fellow citizens who are in hospital beds instead of prisons. They shout on about Death Panels! About Killing Grandma! When real people in town halls get up to report some tragedy that happened to them because private insurance cheated them, the laissez faire extremists heckle them, mock them, and try to shout them down. They don’t care about Freedom or Liberty for everyone, they just want to have the right to betray their fellow prisoner, and not have to pitch any money into a truly regulated insurance pool, take their chances, and pass the cost onto everyone else.0

In the prisoner’s dilemma, getting nabbed by the rogue cops and getting sick are like a sort of reverse lottery. If you’re number comes up, it’s bad news for you. If your number doesn’t come up, nothing bad happens to you at all. In the long term, you’re going to get sick. But if you’re thinking short term only, you’d rather not pay any money into the pool, cross your fingers, and hope you don’t get sick.

But if you do that, there is a probability that you DO get sick, and that means there is a probability that the bills will be too high for you to pay, you’ll end up not paying the hospital and declaring bankruptcy, and the hospital will have to pass along that expense to everyone else.

That is like ratting on your fellow prisoner, and they try to cooperate. You end up paying nothing, and they get left holding the bill of $20k

What seems to be behind the biggest screamers is the same distrust of their fellow humans. They’d rather take a chance at on getting sick on the Prisoner’s Dilemma than trust that some regulatory group will do what’s best for them. And they see a government mandate to cooperate from a short-sighted viewpoint. If everyone else is cooperting, then the libertarian sees the opportunity to betray his fellow prisoner. He’d rather no pay anything in and take a chance that he won’t get sick.

This isn’t a long term solution. This is a short term, I want to betray my fellow human beings, not pay anything extra into the pot, and then if I get sick and can’t pay my bill, I’ll just pass that along to my fellow prisoners, solution. If everyone cooperates because the government mandates it, then everyone comes out ahead.

Summary

In one-time and short-iterated Prisoner’s Dilemma, the best outcome to expect is 5 years, not zero years.

In an unregulated iterated prisoner’s dilemma, two players may start to cooperate with one another. Though the payoff for betrayal remains, and may be used.

In a unregulated iterated prisoner’s dilemma in a population, the players are picked randomly, and may not see each other after prison, so cooperation can be spotty.

In a regulated Iterated Prisoner’s Dilemma in a Population, the population uses government to create a strategic move so that cooperating is always the best payoff for prisoners.

The advantage of regulating the Prisoner’s Dilemma game is that it creates a stable outcome where everyone cooperates, whether it’s a single-play game, a short iteration game, or a multi-iterated game plaed against a random population.

A regulated game will have the overhead associate with creating and maintaining teh regulations. An unregulated game will have the overhead associate with the costs of establishing cooparation through sacrifices and leaps of faith. An unregulated game will also have the overhead that comes from players not wanting to cooperate in a one-shot or short-iteration game.

In a regulated game, even a one-shot game, players will cooperate and serve 6 months. In an unregulated game a one-shot game will likely have both players rat and serve 5 years. An unregulated one-shot game is 4.5 years less efficient than a regulated one-shot game.

Laissez-Faire capitalists will tend to model their notion of “Free Market” using an unregulated iterated prisoner’s dilemma between the same two prisoners. The reality is that the Free Market is more like an Unregulated Semi-Iterated Prisoner’s Dilemma within a Population. Many transactions are one-off transactions with no incentive for the players to cooperate. Some are short iterations. Others might be many-iterations games, but there might not be many alternative players to choose from should your current vendor betray you. Monopolies prevent customers from punishing betraying vendors, because teh customer can’t leave teh betraying vendor.

Laissez-Faire supporters tend to use the concept of the “invisible hand” to explain why players start to cooperate. But in reality, cooperation only works as part of a long chain of iterated transactions. In single-play games of the Prisoner’s Dilemma, there is no Invisible Hand forcing people to cooperate. In fact, selfish actors with no concern for the other prisoner’s welfare, should be counted on to betray you to get himself the best deal. The “invisible hand” is basically an attempt by laissez faire capitalists to say that even though game theory says that a one-time prisoner’s dilemma would result in non-cooperation, that the “invisible hand” would always cause cooperation. In short, the “invisible hand” is a lie.

Health Care is essentially an example of a Semi-Iterated Prisoner’s Dilemma within a Population. In an Unregulated version, private industries have shown their incentive to betray their own customers, essentially acting as the prisoner who rats out the other prisoner. 20% to 40% of private insurance claims in California were turned down by the private health insurance companies. Half of all bankruptcies in America are due to medical bills. 80% of those bankruptcies were people who had health insurance. Individuals tend to risk not getting insurance to save themselves money. When they get sick, they end up in the emergency room, get a huge bill, and many end up declaring bankruptcy. This leaves the hospital with huge unpaide bills that must be passed along to all the other patients. So betrayal can occur at the company level (insurance trying to drop customers who cost them more than they want to pay) and at the individual level (healthy people risking not getting any insurance, then when they get sick, they resort to emergency room care, and create a huge bill that is passed on to everyone)

The idea of regulation would mandate that individuals cooperate, removing the incentive to betray your fellow human beings. It would require that insurance pay all claims, not reject customers due to preexisting conditions, and not charge them more either. It would also mandate that everyone have some form of insurance, providing a public option for those who can’t get it any other way, which would then prevent small problems from being ignored until they become emergency room issues that the hospital has to pay for.