Game Theory and COVIDiocy

27 March 2020

Regular listener Susan L. from WA wrote to us with a very interesting question about game theory and COVID-19. She says:

I have recently been thinking about how to solve this problem (how to stop the President from interfering with the pandemic resolution, which could kill more people). It occurred to me that if one could discover a pattern in his behavior, and then create a disruption to the pattern, his actions might be stopped. That made me think of Game Theory, but I don't really know that much about it. I'm writing to you as experts...

Well I'm no expert on game theory, but it strikes me that a game theory analysis of collective behavior assumes a certain level of rationality amongst the agents making decisions in the game, and I just don't see how we can assume that when it comes to the president. First, we would need to identify what he wants, what is motivating him in the game (i.e. what would count as a 'win' for him). I don't think it's controversial to say that he is motivated by something like looking good, both as an end in itself and as a means to getting reelected.

If he were a rational agent, he would understand that looking good is achieved by doing things like tackling the crisis, getting necessary medical equipment to those who desperately need it, showing empathy for all those economically devastated, and generally acting like a leader. But instead, when he first was told about the impending threat of the virus back in early January, his first move was to ignore it. Then he called it a "hoax" and claimed it would just disappear one day "like a miracle." Now that he's gotten bored with a week or so of social distancing, he says he wants to "reopen" the economy and get people packing churches all over the country for Easter celebrations, even though our best infectious diseases experts say this would lead to the healthcare system being completely overwhelmed and many thousands more dying. That would not look good.

The impression I get is that he thinks the virus is something he can make a deal with (he does, after all, make the best deals!), as though he could get it to agree to a favorable timeline, or bully it into submission, just like his Republican colleagues. This is obviously so irrational and completely out of touch with reality that I just don't know how a game theory analysis would work with a player so fact-resistant. 

I reached out to one of our hosts and resident epistemologist, Ray Briggs, for more insight on the question. Here's what they had to say:

What an interesting question! I'm not quite sure how to solve this puzzle with game theory, but I have a few thoughts on where to start.

Game theory is the study of how individuals will behave in collective decision-making situations, where everybody's behavior plays a part in determining what happens. For a private individual who's young and healthy, the decision to isolate during a pandemic looks a lot like a move in a multi-player version of a game called "the Prisoner's Dilemma". You can either cooperate, and do something that costs you a little while helping those around you a lot, or defect, and bring yourself a small benefit at a greater cost to those around you. In this case, social isolation is a kind of cooperation: it can have real costs for your happiness, but it protects those around you from a devastating, out-of-control pandemic. Defecting is tempting; there's an immediate payoff to the individual. But if everyone cooperates, we're all better off than if everyone defects.

Leaders can encourage ordinary people to cooperate by choosing a coordinated action and encouraging (or ordering) everyone to take part. A political official who issues a stay-at-home shelter-in-place order (as many state governors and city mayors have done throughout the US) is doing something that makes sense as an attempt to achieve a good society-wide outcome. But what happens when you have a leader who isn't motivated by the common good?

To get a prediction from game theory, we'd have to answer several questions. First, what incentives does Trump respond to? Second, who are the other relevant players in this game (the CDC? FEMA? every member of Congress? all of us?), what do they want, and how do their actions influence the outcome? Third, how much can we trust the players to do a cost-benefit analysis of their options?

The Prisoner's Dilemma is useful because it helps us pull a simple and illuminating pattern out of a complex situation. I'm not sure how to give a similarly simple and illuminating account of the federal response to the COVID-19 pandemic, but I hope some game theorists are able to weigh in.

Thanks Ray! That's a much better answer than mine.
 
Finally, I reached out to an expert in game theory and economics. He had doubts about whether game theory would have a lot to say about this question. Psychology, he thought, might give us more insight:

It seems a situation where one just has to distract him or change his perceptions. He seems very driven by the latest things he sees on a certain channel—and I am not sure how one can change that from the outside. 

So there you have it, Susan! Thank you for your great question. It certainly gave us all something to think about. 

If you have a question for us, send it to comments@philosophytalk.org and we might just feature it on the blog!

Comments (2)


johnqeniac's picture

johnqeniac

Sunday, March 29, 2020 -- 12:48 PM

This actually raises many

This actually raises many interesting issues which would merit a full program on PhilTalk. I don't think the single sentence by the (anonymous) 'game theorist' (who is this guy?) does the topic justice. Trump's behavior and policies, aside from being almost universally regarded as morally bankrupt and repellant outside of 'crackerdom', seem quite childish, impulsive, and selfish generally, but his decision-making processes may be more complex than they are given credit for. About game theory, although lying is apparently outside the scope of classical mathematical game theory, nevertheless lying, deception, self-delusion, manipulation, cognitive biases, etc., have been formally included in game theory using formal logic in the related discipline of 'behavioral game theory, and thank god for that because all of human behavior crucially involves those components (almost certainly more than 'rationality').
But, without attempting to do a full set theoretic behavioral analysis to determine how one might manipulate trump's decisions most effectively, in answer to your un-named game theorist's one sentence analysis, how about this - kidnap Sean Hannity and replace him with an operative of Anthony Fauci?

Tim Smith's picture

Tim Smith

Wednesday, April 8, 2020 -- 5:58 AM

Still thinking about this...

Still thinking about this... very interesting ideas. I'm going to follow on this later this week. Really interesting question and good leads here to follow.