When it comes to picking 1, 2 or 3 of the traits, it does seem that having all three is possible. In reading "Churchill and Orwell" by Ricks, it does seem that Churchill was willing and able to escalate indefinitely in the face of a mafioso type personality without himself being a mafioso, wanting good things, and not making dumb decisions (mostly; some of his tactical ideas in the war were questionable.) It seems that the trick is to have a long view that understands letting lower scale evils go to avoid escalation simply leads to future escalation, and a set of principles and preferences that place long run ethical goals (anti-mafioso type) above personal safety. In a sense the hero personality who stands up for what is right because it is right and views costs accrued in the pursuit virtuous, and possibly desirable. (If it gets too desirable you drift towards martyr complexes, but many people take pride in e.g. scars they got fighting for a good cause.)
Anyway, that might be a bit nit-picky, but I do think most game theoretic work misses out on why people might be willing to take one for the team, as it were, breaking people into predators and prey, Hawks vs Doves, instead of adequately examining the different types of people. Too much focus on the more concrete costs and benefits, to my mind.
I don't want to edit it, but perhaps "personal safety" above should be termed "cost avoidance". Personal safety is probably not the key concern of a national leader, although I am not sure the personality trait that drives you to refuse any peace treaty with Hitler is different from one that drives you to stand up to thugs who might kill you personally.
The impossible trifecta you describe is a case of something that can be served well by Hanson's governance-by-prediction-markets. What to do next when dealing with the mafioso, A or B? Easy, just look at which decision is supported by your pet conditional prediction market forecasting on A and B.
It's an interesting attempt, because in some sense a prediction market here is not CDT - the fact that the prediction market chose the outcome in question is evidence that it will choose other similar outcomes in the future, and thus perhaps it's capable of taking the locally wrong decision because of this? But I think it ultimately fails, because it's incapable of last-step execution - the whole idea is that you'd be willing in the end to do something purely destructive/vindictive/etc even if it brings you down, and a prediction market simply won't do it.
If you had a prediction market between choosing different decision algorithms, that might do better, but (1) we're fully in the realm of impossible solutions, and (2) even if we weren't, it still doesn't consider correlations with the past, so it's an incomplete system - you do perhaps consider what today's action says about tomorrow's, but tomorrow's will again 'start fresh' and today's doesn't consider yesterday. So it's at least imperfect. But it would then (if at all possible) be a start.
The optimal way to game Hanson's futarchy from the irrational dictator side is actually to always engineer situations where the market would offer probabilities around 50% between A and B - which is when it can't help decision makers with making a decision.
I doubt autocrats are that smart in practice, though.
I'm a strategy player. I miscalculate and 'declare wars' that I lose and that lead to a ragequit surprisingly often (Putin does not have the option to just ragequit and restart another game)
I'd be interested in a prediction market right now that asks "how many violent deaths in 2022? A - if we start prosecuting Putin in The Hague. B if we give him full immunity and Napoleon's house on St. Helena."
I was even assuming that the futarchy would not be directly gamed, but if it can be, things get really easy - it's not hard to credibly threaten to punish decisions you don't like in ways that will lower their expected payoffs (and the Scott Alexander play of "promise to pay off the vote-for-me contracts at 100' is always available, of course). Any serious version needs a ton of red teaming.
> 3. Benefits from the belief you have the mafioso nature, or otherwise might limitlessly escalate or do something disproportional if provoked.
I think this needs a lot of refinement. We don't actually need "limitless" or "disproportionate" response. We just need a response that makes our opponent's expected costs outweigh their benefits. You don't need to be a mafioso, you just need to not be a simpering, weakling pacifist.
To analogize who's primary target of a mafioso? Sure, they sometimes fight other mafiosos, but they prefer to target weak communities with no real means of fighting back. This is why they prey on marginalized people and communities. What they really don't want to fight is someone who's strong. Predators attack the weak. A well policed, economically connected neighborhood, however, is about the last place the mob wants to attack.
Democracies sometimes appear weak because of the nature of internal dissent and discourse, but I'd argue that democracies actually do deterrence fairly well and without even thinking about it. First, we're just that much stronger. Second, we can pretty easily make credible signals like deploying "tripwire" levels of forces to places we wouldn't really care about fighting over in other circumstances.
When a predatory country attacks another, it's usually based on this calculation. When it goes bad (Ukraine, Iraq->Kuwait, Germany -> USSR) it seems to come from dramatic misjudgment about the will and ability of the opponent to fight back.
I think the magic word in the Biden question is not "decisions", it's "approve". If I hate Biden then I might be forced to grudgingly agree that he was right on some specific issue, but I'll be damned if I'm going to contribute to something that sounds like an approval rating.
On the "pick two out of three" characteristics, I think you're counting Trump out too quickly from having all three. At the very least he's aware of the strategy and consciously attempting to have all three -- he's admitted to it quite a few times. You say he doesn't count because he's made _some_ poor decisions, which is true, but infallibility is a tough standard; I think he makes good decisions more often than bad ones (and no less often than most people who simply attempt 1 and 2), which is why he's a billionaire President and the rest of us aren't.
The biggest obstacle is democracy. It's easy to play the madman when you're a businessman staking his own money, it's much harder when you need the support of others. You need to signal "Hey look, I'm crazy" to your enemies while signalling "Wink wink, it's okay, I know what I'm doing" to the people on whom you rely. Putin can do this because he has private channels of communication to the only people who can actually depose him, but Trump can't, because he's dependant on actual voters and can only communicate through public channels.
"Note that he didn’t say that Biden should say that. He said he would have said it." --> Huh? Yes he did, three times, at 0:18, 0:30 and 0:41 in that video, "he should say".
When it comes to picking 1, 2 or 3 of the traits, it does seem that having all three is possible. In reading "Churchill and Orwell" by Ricks, it does seem that Churchill was willing and able to escalate indefinitely in the face of a mafioso type personality without himself being a mafioso, wanting good things, and not making dumb decisions (mostly; some of his tactical ideas in the war were questionable.) It seems that the trick is to have a long view that understands letting lower scale evils go to avoid escalation simply leads to future escalation, and a set of principles and preferences that place long run ethical goals (anti-mafioso type) above personal safety. In a sense the hero personality who stands up for what is right because it is right and views costs accrued in the pursuit virtuous, and possibly desirable. (If it gets too desirable you drift towards martyr complexes, but many people take pride in e.g. scars they got fighting for a good cause.)
Anyway, that might be a bit nit-picky, but I do think most game theoretic work misses out on why people might be willing to take one for the team, as it were, breaking people into predators and prey, Hawks vs Doves, instead of adequately examining the different types of people. Too much focus on the more concrete costs and benefits, to my mind.
I don't want to edit it, but perhaps "personal safety" above should be termed "cost avoidance". Personal safety is probably not the key concern of a national leader, although I am not sure the personality trait that drives you to refuse any peace treaty with Hitler is different from one that drives you to stand up to thugs who might kill you personally.
The impossible trifecta you describe is a case of something that can be served well by Hanson's governance-by-prediction-markets. What to do next when dealing with the mafioso, A or B? Easy, just look at which decision is supported by your pet conditional prediction market forecasting on A and B.
It's an interesting attempt, because in some sense a prediction market here is not CDT - the fact that the prediction market chose the outcome in question is evidence that it will choose other similar outcomes in the future, and thus perhaps it's capable of taking the locally wrong decision because of this? But I think it ultimately fails, because it's incapable of last-step execution - the whole idea is that you'd be willing in the end to do something purely destructive/vindictive/etc even if it brings you down, and a prediction market simply won't do it.
If you had a prediction market between choosing different decision algorithms, that might do better, but (1) we're fully in the realm of impossible solutions, and (2) even if we weren't, it still doesn't consider correlations with the past, so it's an incomplete system - you do perhaps consider what today's action says about tomorrow's, but tomorrow's will again 'start fresh' and today's doesn't consider yesterday. So it's at least imperfect. But it would then (if at all possible) be a start.
The optimal way to game Hanson's futarchy from the irrational dictator side is actually to always engineer situations where the market would offer probabilities around 50% between A and B - which is when it can't help decision makers with making a decision.
I doubt autocrats are that smart in practice, though.
I'm a strategy player. I miscalculate and 'declare wars' that I lose and that lead to a ragequit surprisingly often (Putin does not have the option to just ragequit and restart another game)
I'd be interested in a prediction market right now that asks "how many violent deaths in 2022? A - if we start prosecuting Putin in The Hague. B if we give him full immunity and Napoleon's house on St. Helena."
I was even assuming that the futarchy would not be directly gamed, but if it can be, things get really easy - it's not hard to credibly threaten to punish decisions you don't like in ways that will lower their expected payoffs (and the Scott Alexander play of "promise to pay off the vote-for-me contracts at 100' is always available, of course). Any serious version needs a ton of red teaming.
> 3. Benefits from the belief you have the mafioso nature, or otherwise might limitlessly escalate or do something disproportional if provoked.
I think this needs a lot of refinement. We don't actually need "limitless" or "disproportionate" response. We just need a response that makes our opponent's expected costs outweigh their benefits. You don't need to be a mafioso, you just need to not be a simpering, weakling pacifist.
To analogize who's primary target of a mafioso? Sure, they sometimes fight other mafiosos, but they prefer to target weak communities with no real means of fighting back. This is why they prey on marginalized people and communities. What they really don't want to fight is someone who's strong. Predators attack the weak. A well policed, economically connected neighborhood, however, is about the last place the mob wants to attack.
Democracies sometimes appear weak because of the nature of internal dissent and discourse, but I'd argue that democracies actually do deterrence fairly well and without even thinking about it. First, we're just that much stronger. Second, we can pretty easily make credible signals like deploying "tripwire" levels of forces to places we wouldn't really care about fighting over in other circumstances.
When a predatory country attacks another, it's usually based on this calculation. When it goes bad (Ukraine, Iraq->Kuwait, Germany -> USSR) it seems to come from dramatic misjudgment about the will and ability of the opponent to fight back.
I think the magic word in the Biden question is not "decisions", it's "approve". If I hate Biden then I might be forced to grudgingly agree that he was right on some specific issue, but I'll be damned if I'm going to contribute to something that sounds like an approval rating.
On the "pick two out of three" characteristics, I think you're counting Trump out too quickly from having all three. At the very least he's aware of the strategy and consciously attempting to have all three -- he's admitted to it quite a few times. You say he doesn't count because he's made _some_ poor decisions, which is true, but infallibility is a tough standard; I think he makes good decisions more often than bad ones (and no less often than most people who simply attempt 1 and 2), which is why he's a billionaire President and the rest of us aren't.
The biggest obstacle is democracy. It's easy to play the madman when you're a businessman staking his own money, it's much harder when you need the support of others. You need to signal "Hey look, I'm crazy" to your enemies while signalling "Wink wink, it's okay, I know what I'm doing" to the people on whom you rely. Putin can do this because he has private channels of communication to the only people who can actually depose him, but Trump can't, because he's dependant on actual voters and can only communicate through public channels.
"Note that he didn’t say that Biden should say that. He said he would have said it." --> Huh? Yes he did, three times, at 0:18, 0:30 and 0:41 in that video, "he should say".