For a very simple illustration, imagine you are in a room with 6 others. 6 of you have dollars, and one has a gun with one bullet. There is a Nash equilibrium where each of you give up your dollar and a NEwhere you say “shoot one of us am we’ll kill you”.
Yes. And because coordination is difficult, the natural equilibrium is “what happened the last time in a similar situation”. If you know that, historically, when a similar situation happened, in 80% cases the people paid; in 5% cases one person resisted and got shot, then the shooter got killed; in 15% cases one person resisted and got shot, then the shooter said “actually, your information is wrong, and I have more bullets (and in 5% the remaining people paid; in 5% they called his bluff and killed him; and in 5% the shooter actually had more bullets, killed one more person, then everyone paid)… I guess you would pay.
On the other hand, if the historic experience is that in 95% cases the shooter gave up, and in 5% cases shot one person and then got killed… then you would probably resist.
And this may be the key difference between traditional democracies that, dunno, perhaps simply got lucky the first few times, and then the Shelling point became “most people will defend the democratic institutions, the traitors will lose and end up in prison”. And traditionally non-democratic societies, where everyone expects that “the rules will only be followed as long as it benefits the most powerful person… otherwise that person will say ‘fuck the rules’, most people will join their side, and the rest will be killed”. In other words, the popular expectation about who would join the rebelion is a self-fulfilling prophecy (if you expect that no one would join, you don’t want to be the only one; if you expect that most people would, you don’t want to stay on the losing side).
Establishing the institutions may be the easy part, changing the expectations may be the difficult one.
As an example, most people in USA expected that life would mostly go on as usual both when Trump won in 2017 and when he lost in 2021. A few people were hysterical, there were even protesters marching on the Capitol, but it was always obvious that the army will support the democratic transition. (And when half of population and the entire army are against you, you have already lost.)
I totally agree that “autocracy is always and everywhere an expectation phenomenon”. My favorite piece of evidence is how quickly regimes collapse when the leader is terminally ill. Nothing has changed but you found out the Shah has cancer so you immediately throw down your arms. Because “Hello prince, i killed people for your dad now rob the people to pay me” doesn’t work. Clearly, repression is motivated by the expectation the incumbent will win and pay you back in the future.
Yes, if people expect democracy to fail it probably will. But the inverse is not true. People expecting democracy to succeed is not nearly a sufficient condition for its success, and such expectations are more common than successful democratizations. The Russians really expected to democratize in 1992, and their experiment failed. The French really expected to democratize in 1789 and didn’t. The Ethiopians I talk to today really expect Ethiopia to stay democratic and it obviously won’t.
The US didn’t just believe in themselves and win the gun game. They denied coercive capacity to the president and distributed it among state governors. They then constrained the governors with the threat of tariffs to prevent secession or shirking. The governors are specialized leader-restraining elites with coordination capacity, the ability to punish each other for shirking, and they are competitively selected.
The Chilean regime had strong expectations of democratic continuity but a terrible constitution that gave Allende the presidency with 35% of the popular vote, leading to collapse into autocracy.
On the other hand, society wide expectation flips are surprsingly common. I agree that it’s weird, but it’s true. Autocratic regimes (not leaders) are very shot lived. The oldest autocratic regime today is Saudi Arabia, which became a state around 1920. Even Saudi is currently in a massive consolidation crisis. The CCP is ancient at ~80 years old, and also in a consolidation crisis. Consolidation means the leader is systematically removing competent elites to cement control. When a consolidated leader dies regimes often collapse. Most autocratic states have not had a regime last 30 years. So it seems like the autocats expectation equilibrium would be very stable, but empirically it is quite unstable.
One could argue that the expectation that any regime will keep power is weaker than the expectation that “democracy will backslide into autocracy”. I think that’s a stretch, but this post is already too long.
Yes. And because coordination is difficult, the natural equilibrium is “what happened the last time in a similar situation”. If you know that, historically, when a similar situation happened, in 80% cases the people paid; in 5% cases one person resisted and got shot, then the shooter got killed; in 15% cases one person resisted and got shot, then the shooter said “actually, your information is wrong, and I have more bullets (and in 5% the remaining people paid; in 5% they called his bluff and killed him; and in 5% the shooter actually had more bullets, killed one more person, then everyone paid)… I guess you would pay.
On the other hand, if the historic experience is that in 95% cases the shooter gave up, and in 5% cases shot one person and then got killed… then you would probably resist.
And this may be the key difference between traditional democracies that, dunno, perhaps simply got lucky the first few times, and then the Shelling point became “most people will defend the democratic institutions, the traitors will lose and end up in prison”. And traditionally non-democratic societies, where everyone expects that “the rules will only be followed as long as it benefits the most powerful person… otherwise that person will say ‘fuck the rules’, most people will join their side, and the rest will be killed”. In other words, the popular expectation about who would join the rebelion is a self-fulfilling prophecy (if you expect that no one would join, you don’t want to be the only one; if you expect that most people would, you don’t want to stay on the losing side).
Establishing the institutions may be the easy part, changing the expectations may be the difficult one.
As an example, most people in USA expected that life would mostly go on as usual both when Trump won in 2017 and when he lost in 2021. A few people were hysterical, there were even protesters marching on the Capitol, but it was always obvious that the army will support the democratic transition. (And when half of population and the entire army are against you, you have already lost.)
I totally agree that “autocracy is always and everywhere an expectation phenomenon”. My favorite piece of evidence is how quickly regimes collapse when the leader is terminally ill. Nothing has changed but you found out the Shah has cancer so you immediately throw down your arms. Because “Hello prince, i killed people for your dad now rob the people to pay me” doesn’t work. Clearly, repression is motivated by the expectation the incumbent will win and pay you back in the future.
Yes, if people expect democracy to fail it probably will. But the inverse is not true. People expecting democracy to succeed is not nearly a sufficient condition for its success, and such expectations are more common than successful democratizations. The Russians really expected to democratize in 1992, and their experiment failed. The French really expected to democratize in 1789 and didn’t. The Ethiopians I talk to today really expect Ethiopia to stay democratic and it obviously won’t.
The US didn’t just believe in themselves and win the gun game. They denied coercive capacity to the president and distributed it among state governors. They then constrained the governors with the threat of tariffs to prevent secession or shirking. The governors are specialized leader-restraining elites with coordination capacity, the ability to punish each other for shirking, and they are competitively selected.
The Chilean regime had strong expectations of democratic continuity but a terrible constitution that gave Allende the presidency with 35% of the popular vote, leading to collapse into autocracy.
On the other hand, society wide expectation flips are surprsingly common. I agree that it’s weird, but it’s true. Autocratic regimes (not leaders) are very shot lived. The oldest autocratic regime today is Saudi Arabia, which became a state around 1920. Even Saudi is currently in a massive consolidation crisis. The CCP is ancient at ~80 years old, and also in a consolidation crisis. Consolidation means the leader is systematically removing competent elites to cement control. When a consolidated leader dies regimes often collapse. Most autocratic states have not had a regime last 30 years. So it seems like the autocats expectation equilibrium would be very stable, but empirically it is quite unstable.
One could argue that the expectation that any regime will keep power is weaker than the expectation that “democracy will backslide into autocracy”. I think that’s a stretch, but this post is already too long.