Compromise and Game Theory

I’m struggling to formulate this question, so I welcome constructive help on phrasing it well.

“Bipartisanship” is a buzzword. Nearly everyone in politics claims to want it, almost everyone claims to be trying for it. It happens far less often than people claim to want it.

I’m thinking of a couple of examples where compromise seems clearly possible but didn’t happen.

Support for “Dreamers,” or DACA, was overwhelming. Most Democrats and IIRC a majority of Republicans claimed to want to find a path to citizenship for Dreamers. But because folks couldn’t agree on other aspects of immigration, the only thing keeping DACA from being canceled is court decisions.

In North Carolina, nearly everyone agrees that teachers deserve raises. The governor proposed a ~9% average raise over two years, with all teachers receiving a minimum of 4%. Republicans countered with an average ~4% raise, with some teachers receiving no raise. Because the two sides couldn’t agree, teachers received a 0% raise.

My question–and this is the question I’m having trouble formulating–is this the result of rational action?

On one hand, it seems foolishness. Let’s say you have two sides:

  1. Side 1 wants pizza for everyone, cake for everyone, and no salad for anyone.
  2. Side 2 wants pizza for everyone, salad for everyone, and no cake for anyone.

The rational approach, I’d think, would be for the sides to get pizza for everyone first, and then settle the cake and salad questions separately. Or to settle the DACA question first, and the 4% raise question first, and then settle the other questions separately.

However, that’s not what happens. And maybe that’s rational? If side 1 doesn’t care as much about pizza as they do about cake, maybe it’s rational for them to hold the pizza hostage until they get their way on the cake issue.

I don’t know if this is making sense. I’d love to read some intelligent though on the game theory/economics/psychology of compromise, especially in the political arena.

bruce bueno de mesquita has written some interesting books about how game theory is involved in politics. His main message is that politicians have to keep the people who can keep them in power happy. In a dictatorship that is a small circle of military, business and political leaders. In a democracy that is a coalition of voters (more or less).

I think the republican party understands that their base voters feel they work for god, and view the democratic party as an abomination. Any attempt to cooperate with the democrats is seen as an allegiance with satan to roughly half of the GOP (and I’d assume a slight majority of republicans who vote in primaries since the party base make up roughly 50% of voters in a general election, but I’d assume over 51% in a primary due to higher turnout).

So the GOP politicians are not incentivized to cooperate with democrats on anything. If they do, they will be seen as working with devils. Richard Luger lost his primary in the Indiana senate because he worked with Barack Obama to secure nuclear weapons overseas. Chris Christie got a lot of pushback because he shook hands with Obama.

Democratic politicians deliver to their voters by passing legislation that expands civil/human/political rights as well as expand the welfare state and create a more sustainable society.

Republican politicians deliver to their voters by taking a scorched earth policy and triggering the libs, and being regressive on civil/human/political rights (curtailing black/gay/womens/voter/immigrant rights). At root the GOP has no incentive to pass legislation the way the democrats do. Democrats deliver by expanding things (human rights, the welfare state, etc). Republicans do not. Republicans benefit from obstructionism.

Fundamentally there is no incentive for GOP politicians to work with democrats and accomplish anything. Their voters will punish this since their voters view the GOP as their tribe and the democrats as inauthentic interlopers. A sizable % of republican voters really don’t even consider government run by democrats to be legitimate, as can be seen by the threat of militia action when a democrat wins a state or federal election (Virginia for example). At root, there is an undercurrent of a belief that if you aren’t a white conservative christian you have no right to power or citizenship. Cooperating with people who fall outside that box just means you lose an election.

Yesterday’s post on the “Conversible Economist” blog may be relevant: The Prisoner’s Dilemma: Celebrating its 70th Anniversary.

I disagree. Take the pay raise. The Dems want 9%. The Republicans want 4%, but importantly, they want ONLY 4% and nothing more. It really does nothing for them to agree to 4% only to have the Dems introduce another 5% pay raise bill tomorrow and the fight continues. Likewise if the Dems agree to the 4% bill, they likely know that they will get no Republican support for the rest of the raise that they want because they have just handed the Republicans everything that they wanted.

The fact that the parties reach a stalemate in these instances or any other means that each party would rather put things off for another day than give up what they really want and they believe that they cannot get what they want without their bargaining chip.

IOW, if the Republicans would rather have no DACA than DACA with no border wall funding, they are rational to refuse DACA. If the Dems would rather have no DACA if it means a border wall, then they are rational as well.

CGP Grey has a youtube video on this, explaining the necessary wrangling of a leader’s “keys to power” (i.e. the necessary ministers and executives and voting blocs he needs to maintain his rule, adapted largely from the very readable and interesting The Dictator’s Handbook: Why Bad Behavior is Almost Always Good Politics by de Mesquita. I recommend the video as a clear 18-minute summary of de Mesquita’s thesis.

Game theory attempts to optimize outcomes.

An optimal outcome, however, depends on:

  1. How you measure success.
  2. The rules of the game.

But let me digress and then get back to that.

A nice thing with the prisoner’s dilemma is that even though the rules of the game are completely silly, and would seem to have very little applicability to the real world, it just so happens to have a sufficient parallel to the real world when you iterate the game out over thousands of games.

Specifically, what it points to is evolution. Cooperation is a naturally arising feature of a large number of strategies where the outcome can be gamed by forming alliances and sticking to them. For criminals, if you think you’ll break the law together again in the future then it’s the best outcome for both of you if you can get back to working together again, even if there’s a minor cost to that cooperation. In the real world, when you’re part of a tribe or family unit, we have a similar real-world pressure. And, one notes, if you look at your money you’ll see a bundle of arrows clutched together in the claws of an eagle, symbolizing the strategic value of alliances and how a clutch of allies is strong together while each individually weak, apart. The prisoner’s dilemma maps to real world pressures and realities.

But, note, the rules of the prisoner’s dilemma explicitly hardcode an advantage to cooperation. So one might say that it’s not meaningful for alliances to pop out of it. You could invent a different game and a different strategy would come out on top.

An example that I have given before is to compare Chess to Monopoly.

To win at these two games requires two very different techniques. In Chess, you attempt to optimally maneuver pieces to optimize offense and defensive capability, while disrupting your opponent’s ability to do so. If you can do that, you will win. Whereas, in Monopoly, you win largely by having the ability to accurately weigh the value of a particular piece of property, given the probabilities that a player will land on that property, and your odds of being able to create a set.

But “winning” is just as arbitrary as it is in the prisoner’s dilemma. There’s no objective reason, in the prisoner’s dilemma, for you to reward the prisoners for cooperating (minus the story element of the criminals expecting to come back to the same place, together again, in the future). Likewise, there’s no objective reason for the person with the last remaining King to win Chess, nor for the person with the most money to win Monopoly. Particularly if we look at Chess, we might note that both teams start with 16 pieces. By the end of the game, they might have 4 and 2, respectively. If you were to murder off ~13 of your friends and associates, would you view yourself as a winner in life? In Monopoly, as the loser, you will have turned $1500 into a variety of profitable ventures of even greater value than you started with. Oh darn!

One might also note, with Chess, that if you apply it to the real world and you’re the king - so it is in fact sort of important that you stay alive - then if you decide to fight it out with the king next door, on the battlefield, even if you don’t care about your troops and simply enjoy the glory of war, then after you win you now have fewer troops and the kings in the other two or three countries which surround you can happily step in and take over some of your land, knowing that you are short on warriors at the moment.

To be sure, a few guys in history got lucky with the “expanding empire” strategy. But, it is safe to say, the grand majority simply proved that it’s a stupid methodology for achieving that goal versus, say, marrying your daughter out and forming an alliance. Or, one might also consider, all the founders of the USA had to do to turn 13 independent states into a mega-country was to meet in a room and talk about how they’d all gain from it, then sign a document to do so. No deaths. No losing your daughter to some drunk asshole.

But, if you have locked in on playing Chess and you are being measured on your ability to win at Chess, on the basis of what we all popularly consider to be a “win”, then you would be stupid to not play in accordance with the rules. And if we killed you, if you lost, and we bred you out with other winners should you win, and maintained this over a few generations then we would develop a people with a nature inclined towards destroying everything that others have. It’s just the evolutionary effect of the rules, applied over an iterative cycle.

Politics is an iterative cycle. It breeds for better players, given the rules of the game, and the win condition. If those players, as they become better at it over their lives and over the generations, become more cooperative, more willing to compromise, etc. then it would be fair to say that that’s the winning strategy for that game. I mean, plausibly there is some other one that might also work, that simply hasn’t been trialed yet. But the safest conclusion would be that this particular rules and this particular measure of “winning” will produce this strategy as the optimum. And if they become less willing to compromise, less honest about their intentions, less practical in their aims, then it would be most reasonable to conclude that that strategy is the optimum, for this game.

Should those rules and that win condition be changed, though, you will get a different outcome.

In the ideal world, the rules and win conditions for the players of the game of politics would produce optimal legislation for the general public. And if we accept that that is the aim of our government, then even if you were able to vote out all of the people currently playing and replace them with people who are better, it would be best to predict that we will end up back where we are. If you filled a room of chess boards with the worlds kindest most friendly folks but only gave them food if they won a game, then you would get back to everyone trying to destroy the other player’s troops just as expediently as they could. It’s just the manifest outcome of that game.

Our system lasted for a good while but, I think, a few significant things changed the rules of the game and broke it. It was never, to begin with, meant to have parties and so it is reasonable to say that it started mildly broken and in need of repair. By the time we started to publicize each congressman’s discrete vote, the game was completely shot.

It will not get better without better rules. And those rules are not populist.

What you’re doing wrong is trying to map the game theory behind passing legislation with the game theory behind choices that have direct consequences to everyone involved. I get you’re trying to simplify to discuss, but you break the question by simplifying this way.

I’ll also look at the pay raise question. Nobody voting will lose or gain money based on the vote outcome. They are voting based on whether it will cost them supporters. So, the public wants a pay rise for teachers. “How much?” is now the question. Republicans think their supporters want a more low ball offer. Only a tiny fraction of supporters will be pissed they didn’t get a bigger raise. This is nothing like “pizza for all, salad for none!”

The pay rise thing can’t simply be a game between a couple of political parties. Surely the teachers themselves have, at least ostensibly or in negotiations, a hard line below which they will go on strike, resulting in a massive cost to the politicians who let things get that far. “0% raise” doesn’t seem like a realistic outcome.

The pay rise thing can’t simply be a game between a couple of political parties. Surely the teachers themselves have, at least ostensibly or in negotiations, a hard line below which they will go on strike, resulting in a massive cost to the politicians who let things get that far. “0% raise” doesn’t seem like a realistic outcome.

That’s the direction things are moving.

One possible explanation for the NCGA’s behavior is that they want a crisis point. They may believe that teachers will overplay our hand by striking, and that they can turn public opinion against a voting bloc that’s been a thorn in their side since they gained power.

Edit: note that teachers don’t have a negotiating body. NC law provides that any contract signed between a labor organization and a government body is unenforceable in court.

There is a lot of interesting game theory in legislatures. The example about DACA is easy — the GOP, while claiming to support Dreamers, wants to hold out their support as a bargaining chip. They won’t vote to help Dreamers unless they get something in return.

One of the classic legislation paradoxes involves cyclical choice.. Suppose legislature wants to choose among A, B, C and will do so with two votes. ⅓ of voters prefer A to B and B to C, i.e. A>B>C. ⅓ prefer B>C>A; and ⅓ prefer C>A>B. If I’m an A supporter and control the docket I will arrange that “Is B better than C?” be voted on first. In answer to that, if I’m one of the B>C>A guys, I’ll encourage my team to vote against B in that first vote!That way we get our 2nd choice © instead of bottom choice (A).

A related tactic is called “loving a bill to death.” For example, a big controversy in health insurance is whether abortion coverage is mandatory. Some D’s might have been willing to sacrifice “free abortions” to get an otherwise good bill passed. But opponents of a health bill insisted on a vote on an abortion amendment, hoping that inclusion of the mandate would kill Catholic support.


A frequent situation is where the majority of the legislature supports a bill (e.g. the McCain-Feingold Bipartisan Campaign Reform of 2002) but a majority of the majority party does NOT support it.  Only rarely is the majority leadership overcome, but McCain-Feingold is such an example: 
[QUOTE=Wikipedia]
... In just the second successful use [in the House of Reps] of the discharge petition since the 1980s, a mixture of Democrats and Republicans defied Speaker Dennis Hastert and passed a campaign finance reform bill. The House approved the bill with a 240-189 vote, sending the bill to the Senate. The bill passed the Senate in a 60-40 vote, the bare minimum required to overcome the filibuster.
[/QUOTE]

Despite that the bill passed by 240-189, it was difficult to get 218 signatures for the discharge petition: the signatures are public and Congressmen are loath to go on record defying their leadership.

Finally, note that the Iterated Prisoner’s Dilemma was already understood by Mesolithic Man: “They probably have a lot more of that nice obsidian. Maybe if we don’t bash their brains out this time, they’ll keep coming back for more trading.”

The issue posed by the OP is one of goal setting not solution. Using the Prisoner’s Dilemma game model, if the goal is common benefit then cooperation is the best strategy. If the goal is individual opportunity then always defect until the resource is depleted then quit.

A better game model to describe the political community is Predator/Prey. Politicians are a small population of co-predatory omnivores who prey on a common voter population. The predator dilemma is that if it impoverishes the prey base it will starve. But, the political predators are omnivorous, so have sustenance other than the voter prey. In this case the voters form a commons. The game is best described by Garrett Hardin in The Tragedy of the Commons.

Since the problem posed by the OP is one of goal setting rather than solution. The choice of goal (OP side1/side2) should be made by voter referendum, not the politicians, because it effects only the voters. It is then the task of the politicians to achieve the chosen goal within the resources available. I believe game theory indicates a cooperative process is most likely to yield political success.

As I understand it, in 2018 Congress was ready to help the “Dreamers”. Then the President threatened to veto the bill, and the deal collapsed.

If you are wondering whether the bill’s eventual demise was the product of rational action, the question would be whether President Trump’s actions were rational.

~Max