Can game theory tell us why nice guys finish last in politics?

I started writing this post over a week ago and a lot has happened since then. Although I do not address these recent events directly, hopefully it is still pertinent.

It has been tough to write lately. As I have stated previously, I am very concerned about what is happening in the country. Between the ongoing pandemic, continued social unrest, and the partisanship and the animosity that seems to pervade so much of life right now, it is hard to be optimistic about the road we are on. Reading the news and checking social media usually leads to discouragement and I constantly have to fight the urge to just disconnect from it all and pretend these things are not happening.

It is hard to know how to channel this concern into something positive. This blog has always been an attempt at that. However, when the topics turn political, I try to resist the urge to add to the divisiveness. This has been particularly difficult over the last couple of months. Between the conventions, the numerous “bombshell” reports, the fights over the Supreme Court vacancy, and the recent contentious debate, it seems that political controversy is unavoidable. Watching these battles have made a couple of things clear:

• Both parties view the stakes as particularly high for this election. They don’t see it as a choice between divergent governing philosophies, but rather as a choice between the preservation of our way of life and its destruction.

• Both parties see the other as corrupt and willing to stop at nothing to win. As a result, there are many on each side who are eager to fight fire with fire.

• Both parties seem happy to appeal to principle when it is politically expedient and willing to abandon those principles when they become inconvenient.

The idealist in me doesn’t like to view the political process as a game, but if we have no guiding principles and making sure our team wins is all that matters, then perhaps that is what it has become. This has been on my mind quite a bit lately, and perhaps that is why I have been so been drawn to the field of game theory.

The iterated prisoners' dilemma

In a recent post, I discussed the prisoners’ dilemma, a simple strategic game that can illustrates how self-interested decision making can produce undesirable outcomes. In this post I would like to look at an extension of this game called, the iterated prisoners’ dilemma, that can provide some insight into the mess that we are in as a country.

As a reminder, the prisoners’ dilemma is a game in which two players each decide between two actions, which I will refer to as “play nice” (formally known as cooperate) and “play dirty” (formally known as defect), and then they receive a payoff that depends on their choice as well as that of the other player. If both players play nice, then both receive a payout of 2. If both play dirty, they are worse off and receive a payout of 1. However, if one plays nice and the other plays dirty, then the one who plays nice gets nothing and the one who plays dirty gets a payout of 3. The choices and payoffs can be summarized using the matrix below.

Regardless of what the other player does, each player is better off playing dirty. So, counter-intuitively, rational players will choose to play dirty and everyone ends up worse off.

The iterated prisoner’s dilemma asks the question: What happens if you play the game over and over again? Will cooperation (i.e. playing nice) emerge as a rational outcome? In this repeated version, you can punish an opponent who plays dirty by playing dirty in response, this makes the incentives more complex than the one-off version and as a result, there is no optimal strategy. Instead, the game becomes a bit more like rock-paper-scissors in that a strategy that produces large payouts against one opposing strategy might perform very poorly against another.

Games like this can give us insight into competition between people, companies, parties and even countries. They illustrate how competitive advantages (as measured by the payouts) affect and are affected by the makeup of a population. Strategies that produce large payouts will proliferate and strategies that produce low payouts will either be abandoned or their adopters will be driven out.

In the iterated prisoner's dilemma, the average payout for the population is relatively easy to describe. If everyone one defects, the average payout is 1. If everyone cooperates, the average payout is 2. If one player cooperates and the other defects, the average payout is 1.5. In other words, the more cooperation that a population exhibits, the better off the population as a whole will be.

Determining which strategies will win out is a bit harder. It depends on the other strategies and their prevalence. Although there is no best strategy, there are certainly some strategies that are less successful than others. One way to figure out which strategies will fail is to create a pool of candidate strategies and then let them compete in a round-robin tournament for supremacy. Let’s look at a few common strategies and see how they fare against each other in these head to head matchups. Below I have produced a table showing the average payoff for eight different strategies (row) against each of those same strategies (column) after 100 matchups each.

Here are descriptions of each of the strategies:

Random p – This type of player chooses its action randomly by cooperating with probability p and defecting with probability (1-p). There are three variants above: p=0.1 (mostly defects), p=0.5 (half and half) and p=0.9 (mostly cooperates).

Nice – This type of player always plays nice (cooperates). This works well when facing another player that plays nice, but if the other player defects, then this player will tend to get a payoff of zero. As a result, this strategy does well with other cooperative strategies and very poorly otherwise.

Nasty – This player always plays dirty (defects). When the other player cooperates often, this results in big payoffs, but when the other player defects, this player does poorly. However, by always defecting, its payoff can never drop below 1.

Tit4Tat – This player plays nice during the first round and in every subsequent round chooses whatever the opponent chose in the previous round. So, if the opponent continues to play nice, this player will play nice. However, if the opponent defects, the Tit4Tat player punishes the opponent the next round by defecting. This prevents the Tit4Tat from being exploited as easily by an opponent that frequently defects. Notice that this strategy does just as well as the nice strategy against cooperative opponents and much better against nasty and random opponents.

Grim – The grim trigger strategy involves playing nice until the opponent defects and then punishing the opponent by defecting indefinitely from then on. This is similar to Tit4Tat, but much more punitive. As a result, it tends to do even better against strategies that defect often.

WSLS – The Win Stay – Lose Switch strategy does exactly what its name says. If the player got a payoff of 2 or 3 in the previous round, it sticks with that strategy. However, if it gets a payoff of 0 or 1, it switches its strategy.

If you look at the average payoffs (against all strategies) in the last column of the table above, you can get a sense of which strategies are likely to be successful. It seems that with a uniform mix of these strategies, the pure Nice and mostly nice (Random 0.5 and Random 0.9) strategies finish at the bottom of the heap. Meanwhile the Nasty and Grim Trigger strategies do much better. This means that if you create a population with all of these strategies and let them compete for survival (with losers being replaced by more successful strategies), the nice strategies will tend to lose out.

Simulation of a battle for survival

I decided to simulate this by creating a group of 100 different players evenly split between the eight strategies. I then let them compete in a round robin tournament (where everyone plays everyone). At the end of each round, I replaced the three lowest scoring strategies with random strategies and then repeated the process until the system reached equilibrium. This is similar to a famous experiment by political scientist Robert Axelrod in the 1980s.

In this simulated tournament, the random 0.9 strategy is the first to go (by the 5th generation). It plays nice most of the time, so it is easily exploited by the nasty strategies, and it also sets off the grim trigger leading to poor performance in their matchups. As this strategy becomes rarer, the scores for the other methods change. As a result, the random 0.5 strategy becomes the new worst performer. By the 10th generation, it vanishes as well. At that point, two of the more cooperative strategies are gone and the nice strategy is increasingly being exploited. Soon, it dies out as well.

The last random strategy is the next to go and by generation 25 there is an approximately even split between Grim, Nasty, Tit4Tat and WSLS. All of those strategies tend to defect quite often against the nasty strategy, but they play nice with each other. As a result, the Nasty strategy gets low payoffs and is forced out of the population. At that point, the three remaining strategies will never be the first to defect, so they cooperate with each other indefinitely and the balance stays fixed.

This is just one highly simplified experiment and with different initial strategies or slightly different random choices, the outcome might look a little different. However, it can still give us insight into how these different strategies interact and this general trend is quite consistent. With a richer set of initial strategies, this cooperative final outcome does not always occur, but the eradication of the random and pure “nice” strategy seems to be the norm. This suggests that in a competitive setting, nice guys do indeed finish last and players who play nice on principle can actually be driven from the population by continuous exploitation by dirty players. Ultimately, dirty players are usually driven out as well only to be replaced by pragmatic/punitive strategies. This all means that although cooperation is the best outcome for a population, it is fragile. It can easily collapse when defection is rewarded as it is in the prisoners' dilemma.

Implications for politics

What does this mean for politics? Well, suppose that playing nice represents honesty and playing dirty represents dishonesty. They say honesty is the best policy, so it is reasonable to expect that two honest players would produce the maximum total payoff. However, dishonest people can exploit the honest achieving great benefit for themselves while harming society as a whole. This is consistent with the payoffs of 3 (for the defector) and 0 (for the cooperator) in the game. Furthermore, when everyone is dishonest it leads to corruption and chaos, and everyone suffers just as defectors do in hte game. In other words, this game is a sensible description for the choice between telling the truth and lying.

This experiment suggests that in a highly competitive setting like the political arena, people who are always honest on principle will have a hard time competing. They can be exploited by the dishonest and as a result they are likely to be driven out. Pathological liars can cause mayhem for a time, but they too will ultimately struggle to compete and ultimately the arena will be filled by the pragmatists who lie when it suits their needs, to punish others or when they think it will give them a leg up.

You could make a similar argument for “bipartisanship” (cooperating) and “hyper-partisanship” (defecting). Again, the iterated prisoner’s dilemma can serve as a reasonable, albeit simplistic, model for competitive partisan interactions. If the politicians who are willing to pursue bipartisan solutions can be exploited by the hyper-partisans who prefer to score points with their base over working toward real solutions, then the result will be increasingly partisan legislative bodies and ultimately political gridlock.

This raises a couple of questions in my mind:

Is this outcome acceptable? To me, the idea that in the long run people of integrity have no chance in succeeding in politics even if they are the ones who would most effectively govern is extremely troublesome and yet it is easy to imagine the experiment playing out in the political realm. Even if most of the political population is inclined to cooperate, an insurgent outsider could come in and play by a different set of rules. They could exploit the “weakness” of the cooperators and ride it to a series of short term victories. If that insurgent happens to be on your side, it would feel like success, but it cannot last. Others would be tempted to abandon their principles and play dirty as well. The pendulum would swing back and forth and the dirty tactics that seemed justified when used by your side, would be much less palatable when they are inevitably used against you. Those early victories would turn into a war of attrition in which both sides lose. In order to survive, it would be necessary to adopt a punitive strategy in which opponents are punished for their dirty tactics with dirty tactics of your own. This would lead to a cycle of escalating retribution that would destroy any vestiges of the cooperative order. A willingness to resort to dishonesty and corruption would become the norm. And even if a cooperative truce emerges, it would always be fragile, able to collapse at any moment into the chaos of tit for tat.

If this seems farfetched, it shouldn’t. A 2018 study by the Pew Research Center found that only 4% of the population trusted that members of congress “provide fair and accurate information to the public” all or most of the time and only 46 % believed that they do that “some of the time”. Meanwhile 81% of Americans believe that members of congress act unethically “some of the time” with an additional 16% believing that they are unethical “only a little”. This suggests that we already have a serious problem on our hands.

What allows these undesirable outcomes to occur? There are a couple of key aspects of this game theoretic experiment that make desirable outcome of principled cooperation so vulnerable. The first is the payoff structure. When the payoffs for defection are greater than the payoffs for cooperation regardless of the opponent’s strategy, it creates an incentive for people to embrace playing dirty. The second is the fact that some members of the population are permitted to change their strategy. This flexibility allows problematic strategies to spread when they are profitable.

Is there anything we can do about it? Call me an idealist, but I don’t think that things have to be this way. There are things we can do to address both of those vulnerabilities in our political system. For the latter concern, this flexibility is the result of adopting strategies based on expediency rather than principle. In a political or social context, this feature will be present as long as behaviors like honesty and cooperation are pursued not for their own sake or because of some ethical code, but because sometimes they are beneficial. If people stop standing up for what is right because it is right, and instead only do it when it helps their cause, then we leave our entire social structure vulnerable to invasion and transformation by strategies that are more morally flexible. Perhaps seeing this play out is what motivated Solomon to write the words from Ecclesiastes 7:15: “In my vain life I have seen everything. There is a righteous man who perishes in his righteousness, and there is a wicked man who prolongs his life in his evildoing.” In contrast, if people were willing to do what is right even when it costs them and to support politicians who do the same, it would prevent this sort of take over from occurring.

For the former concern, the payoffs don’t have to be structured this way and if they shift, decreasing the benefits of defection and increasing the benefits of cooperation, then the outcome changes and principled strategies can persist. In a democratic system the payoffs are not fixed. We, the voters, can influence the payoffs by deciding who gets elected. If we expect and demand that our leaders be men and women of character, who abide by a moral code, who hold fast to their principles, who are honest, and who are willing to stand up for what is right even when it is inconvenient or unpopular to do so, then we can change the incentives that drive us toward this undesirable outcome. We can increase the payoff for cooperation and decrease the payoff for defection. On the other hand, if we don’t demand integrity from our leaders, and we just support “winners”, we will end up with leaders who lie to our faces when it is convenient to do so.

A failure of accountability

We saw a particularly egregious example of this over the past few weeks. In 2016, President Obama’s nomination of Merrick Garland to the Supreme Court was blocked by the Republican Senate majority. According to Senate rules, they had this authority, but rather than telling the truth, that they opposed the nominee because they had the power to do so and that they hoped that the next president’s nominee would be more appealing, they instead pointed to a so-called “principle” that judicial nominees should not be confirmed during an election year. At that point, “the American people should be allowed to decide,” they said. Fast forward to today, and these same senators are now rushing a new Supreme Court nominee through the approval process as quickly as possible mere weeks before an election. The “principle” was quickly abandoned as soon as it was beneficial to do so. The justification? That it was OK because of the Democrats behavior during the Brett Kavanaugh nomination. In other words: TIT FOR TAT; you play dirty, I play dirty. Here are some clips showing the change in rhetoric from 2016 to 2020: https://www.youtube.com/watch?v=OuOLandsO_0

I am less concerned with the election year appointment controversy than I am with the rampant dishonesty. Based on my political leanings, it is very likely that I will agree more with Gorsuch and ACB than I would have with Garland and a potential Democratic nominee. The supposed “don’t nominate justices during an election year” principle made little sense in 2016 and just as it makes little sense now, but that is beside the point. The fact that dozens of senators were ok with concocting a dubious precedent in 2016 and that those same senators seem perfectly comfortable abandoning that precedent 4 years later is deeply troubling. (Don’t believe me? Check out these stunning quotes from Lindsey Graham: https://thehill.com/homenews/media/517247-video-of-lindsey-graham-arguing-against-nominating-a-supreme-court-justice-in). Remarkably, rather than acknowledging the obvious contradictions, many of these senators have defended their actions, refused to admit their mistakes and many of them are now asking for our votes.

Do you know what would be even more troubling than being lied to? It would be more troubling if we were to just accept that all politicians are liars and support them anyway. Now, I know what you are thinking, “the other side does it too!” How can I support them? And you are absolutely right, dishonesty is not a uniquely Republican phenomenon. You can find plenty of egregious examples of dishonesty and hypocrisy from the Democrats during previous political battles. (Although you would be hard pressed to top the brazenness of that Lindsey Graham video.) I’m not arguing for or against either party. To be frank, I don’t see many appealing options in either party right now. We can debate the lesser of two evils, but that is not really my point.

The heart of the matter

I am arguing against the sort of blind loyalty to any party that leads to accepting any amount of dishonesty or corruption because the “other guys” are bad too and we can’t let their policies win. This sort of loyalty allows candidates to earn our support by checking a few ideological boxes without regard for their character or competence. It is willing to accept just about anything as long as it leads to a short-term policy win. It creates an environment that allows dishonesty and corruption to flourish and drives those who are principled out of the political arena. It leads to punitive responses from both sides that make everyone worse off. The iterated prisoners’ dilemma tells us exactly where that sort of moral flexibility leads, and it is not a good place.

If recent events are any indication, we may already be there. Lately I have seen quite a few influential people arguing that policy is all that matters. They trivialize concerns over character and integrity by suggesting that they are really about nothing more than personality and style. They dismiss concerns about dishonesty, incivility, and moral failings by saying that “no one is perfect”. “War is hell”, they say and “the only way we are going to win is by fighting fire with fire.” They may be right about the last part. The prisoner’s dilemma suggests that doing the right thing may not be a winning strategy. But is winning the only thing that matters? What if winning means sacrificing the very identity we set out to protect?

Continuing to downplay the importance of integrity and character will only promote the sort of dysfunctional environment in which problematic strategies (e.g. Nasty, Tit for Tat, Grim) strategies succeed while the desirable principled ones (e.g. Nice) are driven out. On the other hand, if we were to refuse to support politicians who lack integrity, to stop viewing politics through the lens of us vs. them and more through the lens of right vs. wrong, to quit defending misconduct from politicians just because we think they might help us achieve our political goals, then we might just see better politicians rise to the top.

Now I am not saying that policies don’t matter or that character alone is sufficient. Good policies are essential and some measure of policy agreement is necessary for a politician to be worth supporting in my eyes (although I think there is some room for debate about how much agreement and/or which policies are essential). Nor does it mean that a politician must be blameless. No one is perfect. But demanding perfection is not the same as demanding minimal level of integrity and a minimal standard of personal conduct. An effective leader needs character and competence and sound policies. If we only emphasize the latter at the expense of the former, this will lead to the extinction of the sort of principled politicians that we so desperately need.

Unfortunately, that is a tough pill to swallow. It may mean supporting candidates who have little chance of winning but who are the best person for the job. It may mean voting for someone you partially disagree with when the alternative is someone with no regard for right and wrong. It may even mean taking some losses for a time. I have to believe that is better than the alternative. Right now, it feels like the predominant philosophy in the political realm is “the end justifies the means.” If it helps my cause, then it is worth it, even if that means crossing a few boundaries and breaking a few rules. I saw this exemplified in an op-ed I read recently (it’s a year old but was new to me) where the author came to the following conclusion:

“Progressives understand that culture war means discrediting their opponents and weakening or destroying their institutions. Conservatives should approach the culture war with a similar realism. Civility and decency are secondary values. They regulate compliance with an established order and orthodoxy. We should seek to use these values to enforce our order and our orthodoxy, not pretend that they could ever be neutral. To recognize that enmity is real is its own kind of moral duty.”

In other words, the columnist argues that if your opponents are willing to play dirty, you should play dirty back. TIT FOR TAT. It is OK to sacrifice civility and decency if it means winning the culture war and achieving the type of social order you believe is best. The end justifies the means. People like the columnist believe that sort of approach is the only way to win the culture war, so they believe that playing dirty is justified. And these people hold a great deal of influence across the political spectrum. But, if compromising your integrity and endorsing or even participating in wrongdoing is the cost of victory, is it worth it? Could it be that this isn’t a battle we aren’t supposed to win?

I would like to think that we live in a world where enough people still care about things like honesty and integrity that there is room for principled people to flourish. But even if that is not the case, when faced with the choice between being the chump who always “plays nice” and is exploited by the morally flexible pragmatist who gets what he wants by ignoring his conscience, abandoning his principles and playing dirty, my conscience leaves me with only one option. In Matthew 5:38-45, Jesus taught his followers saying:

“You have heard that it was said, ‘An eye for an eye and a tooth for a tooth.’ But I say to you, Do not resist the one who is evil. But if anyone slaps you on the right cheek, turn to him the other also. And if anyone would sue you and take your tunic, let him have your cloak as well.  And if anyone forces you to go one mile, go with him two miles. Give to the one who begs from you, and do not refuse the one who would borrow from you. “You have heard that it was said, ‘You shall love your neighbor and hate your enemy.’ But I say to you, Love your enemies and pray for those who persecute you, so that you may be sons of your Father who is in heaven.

In other words, for Christians there is no room for tit for tat. It would be better to be the chump. Sure, you might be better off playing dirty in the short run, and you might even win. But, Matthew 16:24-26 reminds us that there is more to life than winning:

Then Jesus told his disciples, “If anyone would come after me, let him deny himself and take up his cross and follow me. For whoever would save his life will lose it, but whoever loses his life for my sake will find it. For what will it profit a man if he gains the whole world and forfeits his soul?

I, for one, would like to keep my soul. If that means losing a few battles in the process, then so be it.

PS I don’t think this passage is teaching that Christians should be pushovers. Nor do I want to be a chump. The Bible also says that Jesus’ followers are to be “be wise as serpents and innocent as doves”. This means knowing how to play the game, but doing so with integrity. I think the point is that we need to be willing to let go of our “right” to retaliate and instead pursue to a higher aim: living as children of God and representing our Father well even if that comes at a cost.

PPS In a time when "owning", "calling out" and "canceling" seem to be the widely accepted ways of treating your political opponents, calls to love our enemies feel so rare. Perhaps that is why I found this article by David French (the columnist derided in the oped above) on the president's recent COVID diagnosis to be a breath of fresh air. Here is an excerpt, but it is definitely worth reading the whole thing:

The quest for justice must be tempered by a fervent desire (a love, even) for mercy. At the heart of the Christian Gospel is the availability of grace so profound that not one of us receives the justice we so richly deserve. How many times have each of us done wrong and prayed fervently that we should not suffer the natural and logical consequences of our own foolishness? How many times have we felt immense relief when a gracious and loving God has spared us when our actions cried out for a punishment immensely greater than we received? And should we not seek that same blessing for others—including when those others may be political enemies who’ve behaved recklessly or maliciously?