[ by Charles Cameron — popularizing game theory as a means of understanding significant currents in world affairs ]
You may pick up a few details about the origins of game theory and Prisoners Dilemma, but apart from that, the basic outlines offered by two Bew Yorker articles won’t contain too many surprises. What’s interesting is the role the New Yorker plays as a disseminator of knowledge: game theory, if I may put it this way, is joining the wider conversation.
In May this year, the New Yorker carried John Cassidy‘s piece, How Game Theory Explains the Leaks in the Trump White House.
Here’s the game theoretical background:
In 1950, Albert Tucker, a mathematician at Princeton, gave a talk to a group of Stanford psychologists about the rapidly developing science known as game theory. To illustrate one of his arguments, he invented a story about two criminals who had been arrested for a crime they had committed jointly.
In the story, the police interrogate the two prisoners separately. The prisoners have no means of communicating with each other, but they both understand that, if they each deny the crime, they will be charged with a much less serious offense, which carries a short prison sentence (one year, say). If they both confess, they will get a heavier punishment (five years). If one confesses to the crime and the other insists that he is innocent, the one who confesses will be let off, and his accomplice will get an even heavier punishment (ten years). Tucker posed the question: Should the men confess or deny?
When first confronted with this story, many people think that both criminals should insist on their innocence and escape with a minor conviction. The problem is that mutual denial isn’t consistent with individual self-interest. Take the first prisoner. If he believes that his accomplice is going to deny the crime, he can confess and get off scot-free. If he believes that his accomplice is going to confess, he should certainly confess, too, or he will end up receiving the heaviest punishment of all. In the language of game theory, confessing is a “dominant strategy.” Regardless of which strategy the other players adopt, it is the most rational option to choose. But it ends up producing a bad outcome for both players: five years in prison. If they had both stuck to mutual denial, they would have got just one year.
Then, the political application:
What does all this have to do with the Trump White House? Quite a lot, it turns out.
The issue is White House leaks, and game theory can explain the why of them:
Ever since Trump became President, the White House has leaked like a sieve. “The leaks come in all shapes and sizes: small leaks, real-time leaks, weaponized leaks, historical leaks,” Jonathan Swan, Axios’s White House correspondent, wrote this week. “Sensitive Oval Office conversations have leaked, and so have talks in cabinet meetings and the Situation Room. You name it, they leak it.” Mike Allen, Swan’s colleague at Axios, says, “we learn more about what’s going on inside the Trump White House in a week than we did in a year of the George W. Bush presidency.”
That may well be true, and game theory provides one explanation. By deliberately creating a factionalized, dog-eat-dog culture inside the White House, one that mimics how he ran his business and the premise of his reality-television show, Trump has turned the people who work for him into White House versions of the prisoners in Tucker’s story. With this in mind, it is to be expected that so many White House staffers would take actions that are damaging to the Administration, such as leaking explosive information.
One description of the internal conflicts in the WHite House:
“You have to realize that working here is kind of like being in a never-ending ‘Mexican Standoff,’ ” a White House official explained to Swan. “Everyone has guns (leaks) pointed at each other and it’s only a matter of time before someone shoots. There’s rarely a peaceful conclusion so you might as well shoot first.”
The questionably named Mexican standoff, in which several bandits with guns confront each other at close quarters, is just another version of the prisoners’ dilemma. Imagine yourself in the unfortunate position of being a White House official. If you believe your rivals are about to leak some damaging information, getting it out first is a rational form of self-defense. But, even if you don’t think a rival leak is coming, there is an incentive to spread damaging information about your opponents. Just like confessing, leaking is a dominant strategy.
So, Mexican Standoff, Prisoners Dilemma. The impossible solution?
In terms of game theory, you need to alter the rewards and punishments that individual staffers perceive to be attached to their actions, so that coöperation, rather than backstabbing, emerges as an equilibrium strategy.
And Cassidy’s conclusiom:
The prisoners’ dilemma illustrates how the process works. But, in this case, it could be renamed the Trump dilemma. He created it.
Okay, the second New Yorker piece, by Adam Davidson, Is Michael Cohen Turning on Donald Trump? — dated July 2nd, which triggered this post:
Trump’s former attorney Michael Cohen appears to be playing out the Prisoner’s Dilemma with the President in the most public and consequential way possible.
The most famous game-theory formula was developed in 1950, by two mathematicians, Melvin Dresher and Merrill Flood. But it was only later that another mathematician gave it the catchy name that made it famous: the Prisoner’s Dilemma. The idea is simple: two accused criminals have been arrested and are being interrogated separately. If they both stay silent, they’ll both get a year in jail. But, if one rats out the other, he could get away scot-free while his accomplice would spend three years in jail. The optimal outcome, in terms of total time served, is for both to remain silent. But, as Drescher and Flood posited, there is enormous likelihood that each will rat out the other. There are endless variations of the formula, tweaking the costs and benefits of silence and confession, but the core insight remains: if two people whose interests are mutually dependent on the actions of the other don’t fully trust each other, and don’t have the opportunity to secretly coördinate, they will end up behaving in ways that hurt both of them.
President Donald Trump and his former attorney Michael Cohen are currently playing out the Prisoner’s Dilemma in the most public and consequential way possible
That’s enough to get you started.
And my motive for writing this post? As I said in On two, one, seven plus or minus, and ten – towards infinity:
When I worked as senior analyst in a tiny think-shop, my boss would often ask me for an early indicator of some trend. My brain couldn’t handle that — I always needed two data points to see a pattern, and so I coined the mantra for myself, two is the first number.
These twin New Yorker articles mark a tidal level in the dissemination of knowledge: political scientists andd strategists already know this stuff, but the New Yorker now feels that bright orchestral musicians, humanities teachers, and media mavens, charity workers and foreign affairs correspondents — a few quick guesses at their readership, which must be large and various — are ready and thirsty to add it to their cconceptual vocabulary. That’s a quietly interesting marker in itself.
Oh ah, the New Yorker on July 2 also had a piece titled Will North Korea Play Nuclear Hide-and-Seek with Trump?. I suppose I’d best be on the loookout for other hide-and-seek references. Irony, n’dst ce pas?.