Excerpts from

Prisoner’s Dilemma

William Poundstone


In June 1949, Flood wanted to buy a used Buick from a RAND employee who was moving back East. Buyer and seller were friends. They weren’t looking to cheat each other, just to agree on a fair price for the car. How should they set a price?

As it happened, Flood and the seller knew a used-car dealer. They took the car to the dealer and asked him his selling and buying price for the car in “as is” condition. The difference, the dealer’s profit, was a gain the buyer and seller could split between themselves.

Let’s say the dealer’s buying price was $500, just to have a concrete figure. The seller could, if he wanted, sell the car to the dealer for that price. Likewise, the buyer could buy a car just as good as the Buick from the dealer for the dealer’s selling price – say, $800. In a transaction handled by the dealer, the dealer’s cut would be $300. By not going through the dealer, the buyer and seller have an extra $300 to split between themselves.

How should the buyer and seller split the $300 profit? They could divide it right down the middle. The sale price would be the dealer’s buying price of $500 plus half of the $300 profit, or $650. The seller would get an extra $150 for his car, and the buyer would get an $800 car for only $650.

That sounds fair. In point of fact, it’s what the two RAND employees did. The only thing is, it’s not a unique position. Both buyer and seller are in a position to veto any price. Should either party want to make an issue of it, they could demand a different split.

The buyer could be obstinate and insist that he won’t pay any more than $600... or $550, or $525, or even $501. The owner could tell him to go take a hike, but still, if the owner goes to the dealer, he’ll only get $500. He’s hurting himself by not accepting the buyer’s offer, no matter how low (in excess of the dealer’s price) it is.

It works the other way, too. The seller can be just as obstinate and name a price near the dealer’s selling price. The peculiar thing is that the party who is more unreasonable is apt to get the better of the deal. This isn’t news to used-car dealers, but it’s a little disturbing.

The underlying thread of connecting many of Flood’s observations and experiments was “splitting the gain.” When people can cooperate to secure an extra gain, how do they split it among themselves? Flood cooked up what he thought was a pretty good experiment. He offered two RAND secretaries the following deal: he would either give the first secretary a cash prize (say, $100) or give both secretaries a larger prize (say, $150) provided they could agree how to split the larger amount between themselves and would tell Flood their reasoning.

This experiment differs from the Buick sale because the first secretary alone is privileged and can secure the $100 without any assistance from the other. The other secretary is guaranteed nothing unless the first secretary cooperates. The problem, Flood assumed, was divvying up the extra $50. He supposed that they would split the difference, as in the car sale. The privileged secretary would get $125 and the other, $25. The secretaries didn’t see it that way. They agreed to split the total $150 down the middle! Flood concluded that the social relationship of the parties made a big difference in how they acted.

Even kinship was no guarantee of cooperation, however. Flood needed one of his three teenaged children to baby-sit and held a “reverse auction” for the job. The child agreeing to baby-sit for the lowest pay would get the job. The opening bid was set at $4.00. Flood encouraged the children to come to an agreement among themselves to avoid a bidding war (this is what the von Neumann-Morgenstern theory of n-person games assumes). Despite the fact that the children were given several days to confer, they came to no agreement and bid against each other. The job went for a low bid of 90 cents.

Flood noted, “This is probably an extreme example, although not really so extreme when the magnitude of the children’s error is compared with that made by mature nations at war because of inability to split-the-difference. I have noticed very similar “irrational” behaviour in many other real life situations since August 1949 and find it to be commonplace rather than rare.”

William Poundstone, Prisoner’s Dilemma, Doubleday, NY 1992, pp. 101-103.

      Main Directory      

–– The Heretical Press ––