Sunday, September 24, 2017

It's not a game

I’ve written a couple of times about the controversy surrounding the work of the French historian Muriel Blaive, an advisor to the Institute for the Study of Totalitarian Regimes (ÚSTR) who has recently publicized findings that, she says, call into question whether the communist regime was truly totalitarian.

The core of her argument seems to be that when you look in the files of the secret police, you find far more cooperation with the government than resistance, and I wanted to look at the question of resistance to dictatorship through the lens of game theory.

For simplicity, assume you have a nice, clean division between an evil, totalitarian regime and a populace made up of good people who are the regime’s unwilling subjects.

If on bold soul refuses to comply and speaks out, he or she is easily dealt with. Arrested, killed, tortured, whatever—it is, after all, an evil, totalitarian regime unconstrained by a sense of justice.

On the other hand, if the entire populace rose as one and at 9:00 one morning simply withheld compliance, there’d be nothing the regime could do. You can’t arrest everyone. If the entire public decides, “We’re done with this,” then it’s over.

And presumably it doesn’t even take everyone acting at once. Is it 50%? 25%? There’s some critical value for the portion of the populace that needs to rise as one in order to sweep away a totalitarian regime. Above that, you’re golden. Below that, you’re in jail, or dead.

And that leads to the coordination problem. Standing up is scary, but you’re willing to do it if enough other people are as well. “I will if you will.” But how do I know you will?

Planning for it is difficult, because if the regime catches you talking about resistance, that’s almost as bad as being caught actually resisting.

And even if you could plan, how do you commit?

“I will if you will.”

“I will.”

“Good. I will too.”

“Will you really?”

“Yes. Trust me. Will you really?”

You’re taking an awfully big risk with that trust.

And it’s not just with one other person, or a handful of people, but with thousands, tens of thousands, almost all of them perfect strangers to you.

If that’s the game for the citizens, the regime’s strategy is pretty straightforward. Let it be known that there are dire consequences for noncompliance, even for floating the idea of noncompliance. And let it be known: let them avoid dire consequences from you if they provide you with information on others.

The citizens’ strategy closely resembles the prisoner’s dilemma, an iconic thought experiment from the field of economic game theory.

The premise is that you and a friend are caught doing something bad. The police think you did some serious crime but currently only have proof of a minor one. You’re put in separate rooms and told the same story: if you fess up to the serious crime and your friend doesn’t, we’ll hit your friend really hard and go easy on you; if both of you fess up, we’ll hit both of you moderately hard; if neither fesses up, we’ll convict both of you for the minor crime.

The “rational” action is clear: you should fess up, because whether your friend fesses up or stays quiet, your own situation is better if you talk than if you don’t. Your friend faces the identical incentives. So you both fess up and get convicted for the serious crime, instead of both being convicted of the minor crime.

Because of your inability to coordinate with your friend and mutually commit to keep quiet, you end up with a worse outcome.

It’s even harder for the unhappy subjects of a totalitarian regime to coordinate with thousands of strangers, and so they (almost) all cooperate with the government and end up with continued totalitarianism rather than being able to throw off the regime. That’s the rational response to avoid the very bad outcome that happens when you resist the regime and others don’t.

Carrying the analogy further, the ranking of outcomes in the prisoner’s dilemma points to another element of the regime’s strategy.

The worst outcome for you is you don’t tattle on your friend (that is, you resist the government), while your friend tattles on you (cooperates with the government).

The second-worst is if you and your friend both cooperate with power—in that case, totalitarianism remains in place, but you’re not exceptionally punished.

The second best is that you both resist, and the dictatorship is overthrown.

In the prisoner’s dilemma, the best outcome for you, viewed through a selfish lens, is that you talk and your friend keeps silent. If the analogy holds, that would mean that the best outcome for you in the “game” of totalitarianism is that your friend resists while you cooperate—in other words, it’s better for you to be a favored person in a dictatorship than an ordinary soul in a free society.

And from a material and selfish perspective, that can be true.

A dictatorship has the tools not only to punish those who resist, but to reward those who cooperate. Let’s expand our two behaviors from the prisoner’s dilemma game to three: resistors, non-resistors, and active cooperators. The resistors are punished, the non-resistors get an ordinary life, and the cooperators enjoy significantly more material comfort than most people around them.

Is that cooperator’s life as good as in a prosperous democracy? In material terms, even relatively favored citizens of the Soviet-bloc countries lived no better than broad swathes of the Western middle class, but that ignores the importance of “positional goods.”

In Social limits to growth, Fred Hirsch identified this category of goods, things which get a large part of their value to me precisely from the fact that you don’t have them. In itself, my Porsche 911 is a fairly nice car, but I wouldn’t pay $110,000 for it if I weren’t also looking forward to owning one while you didn’t.

In any society, there are goods that exist in inherently limited quantities, such as a residence in the best neighborhood, or admission to the most prestigious university. Humans being the social animals that we are, we tend to put a lot of weight on how our stuff measures up to other people’s stuff, and we really like the idea of having more than someone else.

And so even if cooperation with the government will only get you to a material standard that many Westerners would consider average, it’s still noticeably better than the straightened circumstances in which many of your countrymen live, and so the state gains a lot of leverage by being able to hold out in front of you that kind of reward.

But here we’re edging into the different motivations that people have in figuring out how to deal with a totalitarian regime, which means we’re getting away from the premise of a black-and-white division between an evil regime and a worthy, oppressed populace.

And those shades of gray are a subject for a subsequent post.

For now, simply note that if you analyze a system of totalitarian dictatorship with the tools of game theory, you would expect to find a preponderance of cooperation with only traces of resistance.

Which is exactly the finding that led Dr. Blaive to conclude that communist Czechoslovakia wasn’t really totalitarian.

Next: Beyond game theory


  1. This argument fell apart for me here: "For simplicity, assume you have a nice, clean division between an evil, totalitarian regime and a populace made up of good people who are the regime’s unwilling subjects."

    I know that this simplifies the debate but it won't get you to a conclusion that accounts for how humans actually behave.

    A humanities major

    1. Shush! You're stepping on my next post.

      What I've done here is pretty standard methodology for economists: simplify the world wa-a-a-ay down to something where you can rigorously account for all the moving parts (well, all the moving parts that are left after you've done your simplifying). That way, you can see some important interactions more clearly.

      Of course, then it's important to go back and try to put back some of the realism that you stripped out. An easy way for economics to go wrong is to forget to do that.