superrific
Legend of ZZL
- Messages
- 9,732
Very clear about what needs to be done. This is going to be another public service brought to you by superrific productions. It's going to be long and I will split it over multiple posts, because there is some background needed to grasp all the material. Let's start with democracy and game theory.
1. Game theory is what John Nash invented, as dramatized in A Beautiful Mind. It's not a special theory. It's simply a mathematical formulation of multi-actor competition. A "game" is simply a series of events in which the participants make repeated choices with knowledge of their payoffs and the payoffs of other actors. For instance, a simple game is husband-wife recreation. Husband likes football; wife likes opera. For them to stay happily married, husband has to do opera sometimes and wife has to do football (simple game so we're not thinking about external options). You can model the game mathematically and discover that it's impossible to predict what they will do on any weekend. They might be attending an opera or a football game. Both are solutions to the game.
But if the wife likes football somewhat, and the husband hates opera, the payoffs change. Now game theory tells you that they will end up focusing on football almost to the exclusion of opera. Wife's preferences get lost in the shuffle. Sound familiar?
2. A very important category of games are referred to under the umbrella term "Prisoner's Dilemma." Let me explain what that means, with two examples. The first and most famous example posits two prisoners being interrogated for a confession. The prosecution can nail them both for a lesser offense -- let's say something that gets them 2 years in jail. But he can't convict them of the serious crime (20 years, let's say) without a confession from one. So he offers each prisoner a deal: if you confess to the big crime, I'll drop that charge against you and let you plead to a lesser offense so you'll get only one year total, and the other guy gets 20. If you both confess, you'll each get 10 years. Usually the phrases are cooperate and defect. Cooperate in this case = not confess. Defect means take the deal.
Now, it's clearly best for the prisoners to not confess. But look at the payoffs. Let's call them Bob and Alice. If Alice confesses, she will either get 1 year or 10 years. Bob has the same payoff. But notice how the payoffs are distributed. If Alice confesses and Bob doesn't, she gets 1 year instead of 2. If she confesses and so does Bob, she gets 10 years instead of 20. So no matter what Bob does, it's in Alice's interest to confess. Well, Bob knows this. So even if Bob doesn't want to confess, he thinks Alice probably will because it's in her interest to do so. Note that the formalism usually posits no communication between Bob and Alice but that isn't a strict condition.
So, even though it is best for Alice and Bob to keep their mouths shut, they won't. They will end up getting 10 years each. That's the suboptimal equilibrium. Experiments have shown that people obey this logic in practice.
3. Democracy is a prisoners' dilemma game. It's better for us if we all cooperate: honest elections, respecting the outcome of elections, no propaganda or dirty tricks or lies. But the payoffs are such that every actor has an incentive to rig the game. If the other side is rigging, you have to rig. If the other side isn't, then you can gain by rigging. No matter what, rigging is the equilibrium.
This plays out in the gerrymandering fiascos. Why does California have to gerrymander? Because Texas will. No matter what California does, the people who rule Texas benefit from gerrymandering. And so too California. Not gerrymandering just allows Texas to gain influence. This is also why every state has winner-take-all assignment of electoral votes (well, almost every state and obviously the logic is pushing NE and ME to abandon their one at large vote). It makes no sense for a state like Wisconsin or North Carolina to assign EVs proportionately. The states are close enough that the EV will never differ by more than 1. NC has 16 EVs. If we did it proportionally, one party might get 9 EVs to the others' 7. Maybe 10-6. But the point is that NC will have gone from 16 net EVs to 4.
This is why the Supreme Court was so damn foolish to say gerrymandering was beyond the capacity of the federal courts, that the states could take care of it. The states cannot take care of it. Literally, the prisoners' dilemma structure (abbreviate to PD) makes it all but impossible. California and New York tried to lead the way on redistricting, and now they are backpedaling because the cooperate strategy is strictly dominated by defect. No matter what other states do, gerrymandering is always better for the party in charge.
Hopefully everyone is with me so far. The next post is where the rubber really hits the road.
1. Game theory is what John Nash invented, as dramatized in A Beautiful Mind. It's not a special theory. It's simply a mathematical formulation of multi-actor competition. A "game" is simply a series of events in which the participants make repeated choices with knowledge of their payoffs and the payoffs of other actors. For instance, a simple game is husband-wife recreation. Husband likes football; wife likes opera. For them to stay happily married, husband has to do opera sometimes and wife has to do football (simple game so we're not thinking about external options). You can model the game mathematically and discover that it's impossible to predict what they will do on any weekend. They might be attending an opera or a football game. Both are solutions to the game.
But if the wife likes football somewhat, and the husband hates opera, the payoffs change. Now game theory tells you that they will end up focusing on football almost to the exclusion of opera. Wife's preferences get lost in the shuffle. Sound familiar?
2. A very important category of games are referred to under the umbrella term "Prisoner's Dilemma." Let me explain what that means, with two examples. The first and most famous example posits two prisoners being interrogated for a confession. The prosecution can nail them both for a lesser offense -- let's say something that gets them 2 years in jail. But he can't convict them of the serious crime (20 years, let's say) without a confession from one. So he offers each prisoner a deal: if you confess to the big crime, I'll drop that charge against you and let you plead to a lesser offense so you'll get only one year total, and the other guy gets 20. If you both confess, you'll each get 10 years. Usually the phrases are cooperate and defect. Cooperate in this case = not confess. Defect means take the deal.
Now, it's clearly best for the prisoners to not confess. But look at the payoffs. Let's call them Bob and Alice. If Alice confesses, she will either get 1 year or 10 years. Bob has the same payoff. But notice how the payoffs are distributed. If Alice confesses and Bob doesn't, she gets 1 year instead of 2. If she confesses and so does Bob, she gets 10 years instead of 20. So no matter what Bob does, it's in Alice's interest to confess. Well, Bob knows this. So even if Bob doesn't want to confess, he thinks Alice probably will because it's in her interest to do so. Note that the formalism usually posits no communication between Bob and Alice but that isn't a strict condition.
So, even though it is best for Alice and Bob to keep their mouths shut, they won't. They will end up getting 10 years each. That's the suboptimal equilibrium. Experiments have shown that people obey this logic in practice.
3. Democracy is a prisoners' dilemma game. It's better for us if we all cooperate: honest elections, respecting the outcome of elections, no propaganda or dirty tricks or lies. But the payoffs are such that every actor has an incentive to rig the game. If the other side is rigging, you have to rig. If the other side isn't, then you can gain by rigging. No matter what, rigging is the equilibrium.
This plays out in the gerrymandering fiascos. Why does California have to gerrymander? Because Texas will. No matter what California does, the people who rule Texas benefit from gerrymandering. And so too California. Not gerrymandering just allows Texas to gain influence. This is also why every state has winner-take-all assignment of electoral votes (well, almost every state and obviously the logic is pushing NE and ME to abandon their one at large vote). It makes no sense for a state like Wisconsin or North Carolina to assign EVs proportionately. The states are close enough that the EV will never differ by more than 1. NC has 16 EVs. If we did it proportionally, one party might get 9 EVs to the others' 7. Maybe 10-6. But the point is that NC will have gone from 16 net EVs to 4.
This is why the Supreme Court was so damn foolish to say gerrymandering was beyond the capacity of the federal courts, that the states could take care of it. The states cannot take care of it. Literally, the prisoners' dilemma structure (abbreviate to PD) makes it all but impossible. California and New York tried to lead the way on redistricting, and now they are backpedaling because the cooperate strategy is strictly dominated by defect. No matter what other states do, gerrymandering is always better for the party in charge.
Hopefully everyone is with me so far. The next post is where the rubber really hits the road.