Janice Gross Stein is the Belzberg Professor of Conflict Management and Founding Director of the Munk School of Global Affairs and Public Policy at the University of Toronto.
In this week’s conversation, Yascha Mounk and Janice Stein discuss whether rational choice theory has led us astray in understanding political behavior, why voters have lost interest in nuclear deterrence, and why cooperation, not rationality, is important in global politics.
This transcript has been condensed and slightly edited for clarity.
Yascha Mounk: I’ve been thinking about a strange way in which my perception of the world has shifted over the last 10 or 15 years. I don’t think my values have changed much, but my perception of how rational the world is has shifted. Ten or fifteen years ago, I had some sense that the people in charge were adults who made rational decisions and were responsive to incentives, and there was comfort in that. One reason to doubt that now is that increasingly erratic and extreme political leaders govern some of the most important nations in the world.
There’s also a subplot about how political science thinks about these things, which I know has grounded much of your work. The assumption is that we can identify what is in the rational self-interest of actors, whether politicians within a state or states within the international system, and that this allows us to predict what they will do. Have we been wrong to assume a baseline of rationality all along?
Janice Stein: You’ve raised one of the most fundamental questions in post-Enlightenment philosophy, economics, psychology, and political science. In the 20th century, for a variety of complex intellectual and sociological reasons, we came to believe that we were rational. That belief served many purposes. It was comforting, and it also served institutional and individual purposes.
Foundational work began in economics and psychology, most notably by Daniel Kahneman and Amos Tversky. Kahneman won the Nobel Prize, and Tversky would have if he had not died. Their message was: Wait a moment. In lab experiments, we are finding systematic departures from rationality. They are not random; they are systematic.
One concrete example is that when people are deeply in the domain of loss, they take big risks. You would not expect that. They are least conservative when their back is against the wall. A second foundational point is that we are not net asset calculators. If I ask you what your house is worth today, you probably will not compare it to what you paid for it. You will compare it to the peak of the market, which you have normalized for. You move yourself from a world of gains into a world of losses. That shift matters.
Microeconomics did not account for that. Political philosophy did not account for that. Much of international relations, focused on strategy and decision-making, did not account for that either. When this stream of thinking entered the field, and Bob Jervis was instrumental in bringing it into international relations, it was explosive.
It was explosive because it attacked the foundations of rationality. Are we never rational? Are we rational under some conditions and not others, which I think is closer to the truth? What do we give up when we abandon the built-in assumption that human beings are rational? From there, we can extend the debate to every area of politics and even to AI. In many ways, it all flows from this debate about rationality.
Mounk: That’s super interesting. There is an injury to human amour propre in saying that we are not super-rational beings. In international relations, the stakes were particularly high in a way that even readers of Kahneman and Tversky may not have had immediately in mind. This was the era of the Cold War. The logic of rational choice was reassuring in the nuclear standoff between the United States and the Soviet Union.
The reason why the proliferation of nuclear weapons on both sides was not supposed to lead to nuclear war and the annihilation of much of the world was the presence of mutually assured destruction. If I bomb you, you bomb me. I might be able to wipe out your cities, but you have nukes hiding in submarines around the world. You have a second-strike capability. So I have nothing to gain from a first strike.
The idea that every American president and every Soviet leader was a rational agent capable of understanding those incentives was, at the time, the assurance that the world was not going to end in a nuclear blast.
Stein: Let me take that example because it is so interesting and that will let us talk a little bit about different understandings of rationality. In the field of nuclear strategy, most people working in the field do not accept that argument, which is the nuclear revolution, MAD, mutual assured destruction, because they were looking for marginal advantage all the time.
What happens if the Soviet Union put multiple missiles, multiple warheads on one missile? Well, then there might be a calculation that they might have enough to overrun the survivable second strike of the United States. You have gone into these finely grained calculations, which we are still involved in to this day. This is what is shaping the debate as all the arms agreements have fallen apart and we have three big nuclear powers. It is called the Three Body Problem. It is all these fine-grained calculations.
But what do we know from looking at the documents in the archives and from the way leaders actually thought? Khrushchev and Kennedy never thought about who had the marginal advantage at that time. They were both terrified of what using even one nuclear weapon would mean. Probably the phrase that has stayed in my mind the longest is by Nikita Khrushchev who said, don’t pull on the thread of war too tightly because it will break.
If you have an evocative, powerful narrative, which is that nuclear weapons are unpredictable, despite all the climate scientists and everybody else, we cannot actually predict the damage with any precision because we have only used them twice. So do not go near them. If you believe that story, then it is irrational to use them. As soon as you move away from the story and you start to calculate marginal advantages, which is what economists do and what political scientists who were rational choice theorists did, we started to see that huge buildup of nuclear weapons to retain that marginal advantage. We are going back to that world again.
Mounk: The concern is not that world leaders might not act in a rational way. It is that this methodological approach to rational choice instructs people how to act. It might instruct people to act in ways that actually lead to much worse outcomes. There is a small story of this, which I think is slightly oversimplified, but it is directionally correct. The basic idea of Prisoner’s Dilemma is that two people have been arrested. If both of them confess, they are going to go to jail for a couple of years. If neither of them confesses, they go free because there is no proof that they did the crime. But the worst world is if you do not rat your friend out, but your friend rats you out. Then you go to jail for a long time and they won’t go at all; they had some kind of deal.
Stein: Let’s stop over that story for one second because we need to think about the implications of it. They are huge. The implication is that individual rationality will defeat collective rationality. You are still within the family of understandings of rationality, but they operate very differently. The incentive structure that operates for individuals will trump and will overcome what is a collective good and the collective benefit. That is a powerful insight for the way we structure a lot of institutions. We have to design protections for the collectively rational outcome that can defeat what is in the individual rational self-interest of leaders.
Mounk: That is a really helpful way of framing what the normative upshot of the Prisoner’s Dilemma is. The basic point is that there are many situations in which some action is individually rational. It is, under any circumstance, better for you to rat out your friend than not to rat out your friend. That is why that is always a Pareto optimal course of action. Collectively, it is a suboptimal outcome for both. It would be better for both of you not to rat each other out; institutions and norms can get us out of this world where we are always doing what is individually rational but collectively irrational.
I was getting at a different point, which is that in studies around the world, what is interesting is that there are actually background social norms which allow people to reach the collectively rational decision more often than you might expect. Most people do not rat out their friend in Prisoner’s Dilemma games, except one group that always rats them out, or that is among the most likely to rat them out, and that is economics grad students. If you ask economics grad students in a lab, perhaps not in real life, it shows an indication of a thesis you were putting earlier at a much more important scale of actual behavior of world leaders. The practice of rational choice itself can teach people to act in ways that are individually rational, but then perhaps collectively irrational.
Stein: Let us go back for a second and link that to a lot of evolutionary theory. It has always struck me that evolutionary explanations—and psychology is deeply into evolutionary psychology and that moves into politics—have always focused on survival. That has been translated into acting in your own self-interest to survive, which has emphasized being the first or moving first to keep the advantage.
What we have ignored in that whole story of evolutionary psychology is the huge advantage of cooperation that societies have when clans, for example, work together. There was another clan. There is a mixed story in the incentives to collaborate that you call those background social norms. If we want the economists in the room for this discussion, they will say incentives, but it does not really matter. There are huge advantages, evolutionary advantages to collaboration, which got buried in the literature.
Mounk: There is a great chapter on this among many other places in The Righteous Mind by Jonathan Haidt—I believe chapter nine—in which he defends the idea that group selection actually was a real part of evolution, which is something that evolutionary biologists for a number of decades really doubted, and connects it to a number of important political phenomena. Let me try and make the devil’s advocate argument for the ways in which rational choice is useful and helpful.
We hope you’re enjoying the podcast! If you’re a paying subscriber, you can set up the premium feed on your favorite podcast app at writing.yaschamounk.com/listen. This will give you ad-free access to the full conversation, plus all full episodes and bonus episodes we have in the works! If you aren’t, you can set up the free, limited version of the feed—or, better still, support the podcast by becoming a subscriber today!
If you have any questions or issues setting up the full podcast feed on a third-party app, please email leonora.barclay@persuasion.community
I am not sure that you disagree with them, but it will be helpful for you to tell us when to call on this tool and when not to call on this tool. The way that I try to explain to students what that method is—and it is not a method I use in my own work, but there are some partitions of it that I find to be insightful about the world—is to ask: what should a map do? I can say that I look at Google Maps and this is a terrible map. There are lots of things about the world that it does not show. It really simplifies the world. There are ways in which it distorts what is actually in the world. It hides some of the most interesting things about the world. I can stand in the Campo in Siena and it will give me the shape of the Campo, but it will completely miss how beautiful the square is.
Maps are terrible, but the answer to that is no. Maps have a particular kind of purpose. If a map always had to be one-to-one, if the map was exactly at scale, it would be completely useless for the kind of purpose that it is actually serving. It would not help to orient you about the broad outline of a terrain that makes it easier to say, should I take this highway or that highway in order to get from Ottawa to Toronto?
The loss of detail is in fact part of the point. A good rational choice model is one that takes very simple assumptions and reveals a lot of how people act. It is never going to reveal 100 percent, just as a map does not reveal 100 percent of the reality. But it is directionally correct. It says that you make these few little assumptions and something really interesting follows from that. It is not 100 percent correct; it does not give you everything. But it gives you something about the logic of things that you might otherwise miss. That is my attempt to represent what the tradition is up to. To what extent do you agree with that? To what extent do you disagree with that? When should we call upon this?
Stein: Your argument is in fact an argument for the value of abstraction and models because the world is so complex. If we just mirrored the complexity of the world, we would be overwhelmed. We do not have the capacity even to process that kind of information. But when you were describing it, you used a set of phrases, Yascha.
You may want to take this back because you made it easy for me. You said it is directionally correct because it tells us more or less most of the time the way people will behave. But what if the models are not directionally correct? How do we know? One of the ways to know is to use history as a living laboratory and to go look at the way people actually behaved under a set of conditions. Now, that is not scientifically satisfying. So how do you get out of that trap? Well, you do a lot of cases from history and work. Again and again, if the assumptions violate the empirical patterns, you should, in my view, either limit the assumptions to a set of conditions or revise the assumptions.
But that is not what the rational choice theorists in politics did. The argument came back, well, they are just “as if” assumptions. So it does not matter if they are empirically correct or not; it is an “as if” exercise.
Why do I find that a very, very difficult argument to accept? For two important reasons. Models are simplified assumptions. They are also stories, and they have narrative power. If we tell stories over and over where we know the assumptions are not sustained by empirical evidence, then I think as academics we have a responsibility to go back and look at the assumptions and say, where do we think these assumptions are most likely to hold and where are they not? That is much, much better.
The world helped because we have evidence in front of us now that we have leaders who barely meet the minimum of rationality, and so it becomes a more plausible argument to make. But until that happened, there was a resistance. There was the Milton Friedman “as if” argument: I do not really care how people behave. Whereas for Keynes, that was not his argument. Keynes talked about “animal spirits” that operated; he could not quantify them, he could not measure them. Often in his argument, it was hard to know exactly when the animal spirits would be unleashed, but he knew they were there. We had to tell a story about the animal spirits just as much as we had to tell a story about disciplined, rational calculators who were always able, through some method which they never specified, to make the most optimizing decision.
I have spent my life asking people, how do you make the decisions in your life? Who do you want to partner with? Did you calculate relative cost and benefit?
No, not really.
Which graduate school did you want to go to?
Well, I had a kind of a list.
I said, well, how did you measure across the dimensions? How do you compare?
Well, I don’t know, but I got to an answer.
When you think about what the risks are here, rational choice models are stories where the assumptions are sometimes upheld but often not, and they rely on a method that almost nobody uses other than in formal studies. That is a very dangerous model to take into the world of practical politics. The one that is most striking is nuclear politics, but you and I can think of many others.
Mounk: I would love to hear more examples where you think a rational choice approach goes wrong. I am sure there are many, and I am sure I will argue on many of them. Let me share one where I was dismissive of a kind of rational choice approach, and I think that it was somewhat vindicated.
When Donald Trump won the first presidential election in 2016, the Republican Party was still full of people who really did not share his ideology and who I think were quite depressed. The easiest task for any decent political journalist in Washington, D.C., was to corner some senator or some representatives in one of the bars around Capitol Hill and get them to tell you anonymously, of course, how much they hated Donald Trump and how miserable their life was. It seemed to me at the time that if they were being honorable, they would quit or criticize the president. They probably would even have gotten a lot of good things out of it. They could have gone and been fellows at the Munk School at the University of Toronto, which you lead, or perhaps at SAIS in Washington, D.C., where I teach. They might have gotten a nice gig as an on-air commentator at CNN or MSNBC.
But they did not. The very simple model that students of Congress had applied for a long time best predicted their behavior, which is that they did whatever they thought was needed in order to get reelected. They prioritized the reelection over everything else. That is a specific conception of rationality. One of the things that puzzles me is why it is that those members of Congress so preferred this particular set of incentives—what it took to get reelected—over these other incentives, like perhaps cashing out in some smart way and having social prestige and having a lot of followers. They clearly care about power as a particular kind of goal, rather than a different basket of goals they might have, and that is interesting in itself.
This simple model that I would have laughed at in 2014 or 2015—that a legislator is just going to do whatever it takes to get them reelected—explains, to a first approximation, 99.5 percent of behavior. I feel like that particular model in that particular moment held up better than my prediction, which was to go criticize him and take the job at CNN and be a fellow at a fancy place and join some corporate board.
Stein: I agree with you. I think the model is vindicated. There is still a very large and interesting literature on whistleblowers and the huge costs that whistleblowers pay, even when they are exposing behavior that is morally reprehensible. People know that they pay these costs. There is going to be isolation; they are going to lose their job; they are going to lose their friends. There are powerful inhibitors against actually blowing the whistle even when they see things.
Part of this speaks to the power of rationality. You calculate the costs, and sometimes they are emotional. It does not matter; they are costs, and you do exactly what rational choice theorists will tell you they are going to do. Another area where I see this is the unwillingness of students to come forward sometimes when there is egregious behavior, either by their peers or by a faculty member. They will not come forward because they are well aware that there are going to be social costs afterwards for doing it.
When we started our conversation, I said to you that the recognition for me—which was like a kind of blinding light—is that for some of the biggest questions of international security, rational choice was not going to get us there. It raised more questions than it answered because it raises this question that we are talking about right now: when are people rational? There is a fertile ground for a conversation that crosses almost all our fields, including political philosophy and theory.
Is it when the costs are so blindingly obvious to them that they are rational? Or is it when the costs for engaging in it are so painful as they understand it that they therefore do not take that action? There is a whole set of categories where people are rational.
Where I think the really critical work has to come is in distinguishing those and separating out the conditions that are likely to promote optimizing behavior where the measurement is easy. It is pretty easy to figure out the cost. That is what Khrushchev was telling us: Boy, do not pull this war. It is very easy for me. I do not care how many more nuclear weapons you have than I do, JFK; it is going to be really bad for you and it is going to be even worse for me, so let us not do this. I come back to narrative and story; certain kinds of stories will enable rationality, and certain kinds of stories will actually move us away from it.
Mounk: One of the problems with a rational choice approach is just about how broadly you define the incentives and the rational choice that comes in. There is a very straightforward, famous puzzle in political science, which is: why does anybody go to vote? The chance of being run over by a car on the way to the polling station is higher than the chance that your vote is going to swing the election, at least in large-scale elections at the national scale. So, why would you do it?
One way to say it is that clearly voters are not making a rational choice; they are doing it for other kinds of reasons. Rational choice scholars come back and say that it is a rational choice; it is just that one of the things they consider is the kind of nice civic glow they get from voting. But then you might say we are smuggling in this much broader conception of rationality. It is one way of thinking about this simply: human beings are deeply responsive to incentives, but the nature of incentives is much more complicated than a rational choice approach usually explains.
To go back to the puzzle about why it is that rank-and-file members of the Republican Party acted the way they did after 2016, the key thing to understand might be what Arthur Brooks argued. He argued that each human being is driven by one of four different fundamental drivers: money, power, pleasure, and status. Different people are going to maximize on different dimensions.
That is a soft rational choice approach in a way, but the key thing you have to understand is: is this person driven by power or by something else? The reason why these representatives were not willing to give up being reelected is that they like money and they like status, but what they really like is power, and so they are going to maximize political power over those other things. Is that a more helpful way of thinking about it, or do you worry that that is going to have some of the same flaws?
Stein: Let me tell you why I started with that, because I am playing a sophisticated trick on you in this answer: you can rationalize any outcome by reasoning backwards and constructing a set of incentives that will explain the outcome. That, to me, is a trap. It is an intellectual trap, and it actually becomes, at some level, a normative trap too.
That is why I pushed back. You just said they are really motivated by power. But are you going to tell me that being one of 435 members of the House gives you a lot of power? That is a really difficult argument to make. Then, well, but they did not break ranks. You end up with a kind of regressive reasoning back, and you can explain almost any outcome. If you cannot tell me in advance what incentives people are responding to, then what is the value of the whole thing? Let us just go find better, more complex models that are not so satisfying but build in at least an additional level of assumptions so that we get a more accurate understanding.
I will tell you how I explain voters. There are powerful norms that motivate people to vote because they value being a citizen in a democracy. It is paradoxical if you are using microeconomic reasoning, but it is not paradoxical if you have a larger conception of society that is, in part, norm-driven on certain key issues that matter to people. That is why I come back and say: the story you tell has long-term consequences for our self-understanding in our society.
Mounk: I was playing devil’s advocate, but as you know, I agree with 90 percent of what you have said. How do we apply this to this moment? We are in a situation in which we have the political leaders we do: Donald Trump in the United States, Vladimir Putin in Russia, and Xi Jinping in China. We have a lot of people losing trust in the idea that their leaders and the institutions are acting in very rational ways. We are at a moment in which many of the norms that have actually governed and constrained the behavior of political leaders seem to be on the way out as well. What kind of intellectual toolkit should we grasp for understanding and analyzing this political moment?
Stein: I think that is the question of our times, and that should be our vocation. For us in the social sciences and in political philosophy, it is the most fundamental question because getting the answers more or less right will explain where we are a decade from now. Again, I think a story about norms that supersede incentives is a fundamental part of what we have to reckon with. That is an empirical question; it is not a philosophical question only. It is an important empirical one. When do people respond to norms and give priority to norms rather than self-interest? If that is not part of the research agenda, people are not going to hear that discussion.
I have had conversations with many of my liberal friends—let me put the hot-button issue on the table. I have had many conversations with my liberal friends who have nothing but scorn for people who vote for people who are norm-breakers and for the exclusionary language that they use about these people. I cannot use some of the words because I would offend the sensibilities of some of our listeners, but there is no recognition, Yascha, that part of the way the liberals behaved beforehand is in part a contributor to the distrust that people have of institutions. There was a self-righteous superiority where liberals condescended to those they considered the “great unwashed,” who were ignorant and uneducated and drove trucks and only had high school degrees and worked in the industrial economy and really did not understand these niceties of politics. The attitude was: just leave those to us because we know better.
Now, if you engage in those kinds of conversations—and that is not to claim in any way that that is the sole experience, but it is one of the streams that feeds into the river—that is not an incentive question. That kind of behavior is a norm-driven question. For any candidate to say that all the voters for my opponent, or 50 percent of the voters for my opponent, are “deplorables,” that is norm-breaking behavior for a leader of a democratic society. People who are labeled “deplorables” hear it and are unforgiving and are angry and resentful, and then they recalibrate what democracy means to them. It is not about who defends their interests, because I think we would have a very difficult case arguing that Trump in any way has advanced the interests of the voters who are his base. That is not the right question. It is a different question, and that is why I say we need to move beyond rationality. It is not that it is not important, but it is never the whole story. The question becomes: who speaks for me? Who represents me? Who hears me?
Mounk: The way I think about this is that in American politics, there is sometimes the idea of a “beer test,” which goes back to 2004 when there was a poll which showed that more people would like to have a beer with George W. Bush than with John Kerry. I always found that to be a kind of cop-out. I do not think people care about whether they would enjoy having a beer with the president.
I think there is a kind of inverse beer test that makes a lot more sense to me, which is: if a president showed up in my home today, and I did not have time to change the paintings on my wall, and I talked to the president the way that I talk to my neighbors, would they like me? Would they think this is a decent person whose interests I share, whose concerns I take seriously? Or would they say—as Gordon Brown said in a hot mic moment during his reelection campaign—“What a bigoted woman?”
I think it is when they feel like this person would be nice to my face, but then go back to the people of their ideological persuasion or their social class and say, what a terrible person, that they say, why should I trust this politician? Why should I trust the institutions run by these politicians?
Stein: That is not a calculation of interest. You just told a profoundly important social and normative story about trust. Trust, as we all know, is the glue of all social interaction. Without trust, there is no society, and no institution survives when members lose trust in it and how it works.
That is a normative story. It is much more, for me, to come back to what we are talking about: the limits of rationality. It is not that we are not at times all of us very busy calculating our own self-interest, but it is not what we do all the time. It is not what we do on some of the most important issues in society.
Mounk: We share a fundamental intellectual sensibility and diagnosis of this moment, which can feel lonely at this time. It is both to recognize that there are, in fact, many institutions which are important that we should defend, and that universities, especially in the United States, are deeply flawed places at the moment. Some of the most important institutions in our society—to recognize that the CDC has acted in some really quite disastrous ways during the pandemic—really failed to live up to their mission. For example, the catastrophic failure to roll out accurate and widely available tests for COVID early on because of bureaucratic obstacles.
Stein: When COVID started, the Chief Medical Officer of Health came out to do a press conference, as every government does, and she said, do not wear masks. They asked why, and she said, well, there is no evidence that masks will help against this sort of virus. Except, of course, if you went to the ERs where COVID patients were swamping the wards, there was not a single health professional who did not have a mask on. Afterwards, there were two hypotheses about why she said that, and both of them raise profound difficulties.
The less benign hypothesis is that we did not have enough masks in the country and the priority was to make sure that the frontline workers had them, and so you do not tell the truth to the public. Boy, if that is accurate, we are in deep, deep trouble for the future. I do not think that was her motive because she could have said to everybody, we do not have enough masks. Go home, cut up a pillowcase. Make masks for yourselves and your kids until we can get enough masks into the country.
The second argument is even more troubling to me: there was no evidence that masks worked. Why was there no evidence that masks worked? Because nobody had ever done an RCT on masks. But there is common sense—for years in operating rooms and infectious environments, healthcare professionals wore masks.
That piece of advice did more to undermine confidence and public trust. There is another way that could have been handled, and it comes again to how we think about public discourse and public language. She could have said, we do not know much about this virus. It is not for no reason it is called a novel virus. I am telling you, I do not know. I am going to give you advice today, but I will be back in two weeks because we have an army of people studying this, and I hope my advice changes because that means we will be learning something. I want you to stay in step with me as we learn more over time, and I will share whatever we learn with you. To me, that is a deep flaw: officials who are fearful that they will be pilloried by the public for making a mistake and fail to talk to citizens.
Mounk: Elites wonder why people do not trust them, but actually we should also wonder why elites do not trust the people. Of course, there is a very similar debate about Anthony Fauci and his advice that masks were not effective at the same stage of the pandemic early on in the United States. I think this discussion will be familiar to listeners here. The dilemma that you pointed out is true on either assumption. If the point was that they thought, we do not have the best, most rigorous scientific study to prove that masks are effective, and therefore we are going to tell them that they are not effective, there is a mistake of scientific reasoning going on there.
More importantly, it is this inability to communicate in the right way—to trust people and say, look, this is a fast-moving situation. We are going to try and give you the most accurate information we can at every point, but really, we do not yet fully know. We have the best scientists working on this, and we will update you as soon as we can, but we trust you with uncertainty. It is not having the trust that the public can understand that complexity. It is saying, no, we have to have a simple message; we have a media trainer over here and a PR consultant. It is not that the motivations are nefarious, but it is the lack of trust in people’s ability to understand.
The same thing is true in the other assumption, if you think that what was going on was a “noble lie.” Again, I think I understand the underlying motivation. If you feel that the doctors and nurses in the ICU desperately need this personal protective equipment and they do not have access to it, and if we tell people masks are really effective against the virus, everybody is going to try and go out and get masks—so perhaps we should fudge the issue a little bit because I am speaking to the head of this hospital and they are saying we do not have masks. I get it; I do not think the public health officials are corrupt in the sense of wanting to make money or that they do not care about people dying. I get it is a hard position to be in.
But two things: number one, it is really short-sighted about the effect of that. As you are saying, it delayed all of the entrepreneurial spirit that people then developed, of saying, I will just cut up my pillowcase and turn that into a mask, or factories that are producing other things saying, perhaps we can start producing masks instead. It is not thinking of all the broader effects on society.
Most importantly, it is not trusting people to be public-spirited. What a different appeal it would have been to say, hey, look, masks are helpful—or at least we think they are helpful. The evidence is not yet completely clear. If you have some way of getting a mask at home, produce your own cloth mask. We are going to try and do everything to produce more masks so we have the best masks to give out very soon. But in the meanwhile, please do not buy them up. The doctors and the nurses who are keeping your friends and relatives safe if they have a serious case of COVID—they are the ones that need them. That would be trusting the public. On either of the assumptions—either that they do not trust the public to understand the complexity of evidence available at the time, or they do not trust the public to act in a public-spirited way—it is actually a lack of trust in the public that led to that misstep.
Stein: That is right. We agree entirely. But that itself, I think, is a deep flaw. It is an elite problem: I know more, I am more sophisticated, I understand the complexity, so I am not going to trust you, the citizen. Once you, as leaders, lose trust in the community that you lead—it does not mean you share every complexity, but you are truthful. You have the capacity to acknowledge, I do not know.
I find with students that is the most trust-building statement I can make in the classroom. They ask me something and I say, you know what? I do not know. I have to go read, and I will be back to you. That creates a kind of shared trust. I believe it is the loss of trust in the public that has gotten us to where we are. If I had a voice in who we select as the next group of leaders in 2028, my highest criterion would be: who trusts the public? Who does not condescend to voters? Who does not believe voters are ignorant?
Let me switch gears and tell you why I think this matters so much—because you and I both agree that we are in the middle of a disintegrating international order. Whether you think the next one has not yet been born or not, clearly we are at a moment where linear thinking is not very helpful. If you love history, we can have a lot of fun identifying what these nonlinear breakpoints in history were in which things changed.
Just think about what this means: all our linear probabilistic models will not work. Let me take two sentences to tell you why I feel so strongly about this. It is an entirely different way of thinking about rationality, because rationality has probability built deeply into it. It is subjective, expected utility. “Expected” is just a nice word for “probability”: what is the likelihood? Linear means the future is going to look a lot like the past.
Well, you and I do not think that right now. You do not think that about AI. I do not think that about the way our international institutions and practices are going to represent themselves. So, linear thinking is not a good way to go forward.
There is a second thing: probability is going to be a very poor guide. What do we need to really be intellectually honest about probability? We need to have a large number of trials and understand the probability distribution. Otherwise, we really should not do it if we are being intellectually honest. We do not have that for AI; it is too new. We do not have it about this world that we are leaving behind, and we do not know yet what we are moving toward. So, what do you do? How do you use your analytic tools?
Mounk: That is not only a question about rational choice. Rational choice is one way of trying to predict the future. We know roughly what the incentives are, and so we can build the model and predict it forward. A very different way that political scientists try to understand the world, which in some ways is more dominant today than rational choice, is high-end statistical reasoning. You construct databases which show statistical regularities.
Then you use them to make predictions about the future. To go back to 2016, one of those was to show The Party Decides, a very well-regarded book in American politics, which said that activists seem to play a role and there is often a crowd favorite who is quite high in the polls early on. But in the end, the thing that really matters is the decisions of donors and party insiders and elected officials. They are the ones who actually decide; the party decides who is going to be the nominee.
For 2016, all these political scientists were saying, Donald Trump, it is not going to happen. In the end, he is not going to get the Republican nomination. This is one prominent example of a broader problem where if you are at a moment of paradigm shift, looking back at the statistics of the last 10 years or the last 50 years to predict what is going to happen in the next five or 10 years is not going to help.
Stein: I just asked somebody who is a very experienced political operative who is going to be the nominee in 2028. He said, whoever is most closely connected to the big donors, that is who it will be. I am not sure that is the right answer.
I have been thinking hard about uncertainty for the last couple of years. First of all, people hate it. It is similar to rational choice in only one minor but important way. Rational choice was as powerful as it was because it gave you control; it made a world predictable. Large datasets are as attractive as they are because they give you control; they allow you to structure things and you can generate these predictions. That is psychologically very satisfying because you do not feel you are tumbling in a universe which you cannot structure in any way.
But those methods are not appropriate to the world that we are moving into because we just do not know. That is where I start from. We do not know. So what do we do? We tell stories. You could call them scenarios. They are structured, disciplined stories. You tell them and you ask yourself: what are plausible ways of thinking where we will be in the next 10 years? What makes sense to you, Yascha? What are three possible worlds?
The next question might be: tell me how you get from your first possible world back to today. If you cannot imagine a single possible path to get there, it is probably not very possible if you cannot reverse engineer your way back. That is also narrative power. As you reverse engineer your way back, there are key decisions along that path, or key landmarks. When you get there, it is like the Turing test in AI. Here is a test, and if we pass this test, we are going one way rather than another way. Identifying those markers will tell us: are we moving more directionally here rather than there? If so, then we adjust our strategies and the things we do to mitigate the risks. But that is fundamentally based on human capacity to imagine. Right now, we have a monopoly on a capacity for structured imagination.
Mounk: The question I was going to ask earlier, and it is a mark of a great conversation where there are always four branches and you have to prioritize between branches. I am sad to let that other branch go, but I do want to return to this branch: I think we share the sensibility that there are many things in the old order that we want to rescue in various ways.
We don’t think we shouldn’t have a CDC. We don’t think that we shouldn’t have NATO. At the same time, it is clear that the order is crumbling and that the people who are simply saying the order is fine and we need to make some cosmetic changes—that eventually Trump will leave office and leave the political scene, and then we can go back to normal—are deluded.
The fundamental challenge is: how do we balance those things? How do we recognize that there is fundamental change that is needed without throwing the baby out with the bathwater? Should we think through scenarios? Tell me how to think about this. What are some of the scenarios, good and bad, for how this is going to go?
Stein: Let’s say that if we were thinking about this and we had a blackboard and we could draw the stories that we were telling. We could think about a world with three or two and a half big powers that are more or less led by thugs. Let me put it bluntly. But what is the underneath layer? How does that underneath layer self-organize? Do we put all our eggs in strengthening international institutions that I think are less than effective? Long past their best-before dates.
Or do we think differently about standing up a set of institutions just underneath that are smaller, that are nimbler, that don’t build in vetoes? Mark Carney spoke about this at Davos to some degree. It is in many ways a very misunderstood speech because it has so many dimensions. But one of the things he talked about was a world just beneath that level that was what we would call variable geometry. You and I are both interested in narrative and stories. I am going to come and start a project with you to stand up a big research project on how disciplined narrative can help us think through our future. We could probably think about 10 other colleagues that we would invite from different parts of the world. We get that project going; we wouldn’t worry too much about the ones who tell someone, this is all going to blow over, don’t do anything.
That is one vision of the world: that we invest all our effort toward these next 10 years in propping up the UN, propping up NATO, and other institutions. Some of these are the holy grail. That is one. The second one is no, we don’t do that. There is another world out there. There are still thugs at the top, but right underneath them, there is a plethora of small or nimble international institutions. Climate gets thought about that way and nuclear weapons get thought about that way, and they are not the same members in every group. There is a kind of test of effectiveness that we use for each of those. That is number two. Number three is the thugs self-immolating, or two of the thugs self-immolating, or one of the thugs has self-immolated. What difference would that make to what we do in level two?
Then we start thinking about the paths backward. What would it take? What would we have to do today to get to stage number one, number two, or number three? That is hard work. That is where it is really disciplined thinking, and it requires a knowledge of political institutions: how they work, how you build coalitions, what is doable, how you keep a coalition together. That is the essence of democratic politics, isn’t it? And what norms really matter?
In the rest of this conversation, Yascha and Janice discuss the role of mid-sized powers in global politics, whether the European Union can be saved, and how to stay intellectually curious in cynical times. This part of the conversation is reserved for paying subscribers…
Listen to this episode with a 7-day free trial
Subscribe to Yascha Mounk to listen to this post and get 7 days of free access to the full post archives.
