Newcomblike problem: it feels like there are two decisions to be made, but there is only one, in advance. If the Enemy knows you will never fire your missiles, They can pick off your cities one-by-one with impunity. But if you precommit to Mutual Assured Destruction if They ever cross the line, then it would be suicidal for Them to do so, so They won’t (if They are rational). If losing one city to a nuclear attack is unacceptable, then that is your line in the sand. You must precommit to full retaliation in that event, and make it known to the Enemy.
In a game of Chicken with your loved ones’ lives on the line, your best move is to defiantly rip off the steering wheel before the Enemy can do the same.
I’m oversimplifying a little, but nuclear deterrence is all about game theory like this.
Newcomblike problem: it feels like there are two decisions to be made, but there is only one, in advance. If the Enemy knows you will never fire your missiles, They can pick off your cities one-by-one with impunity. But if you precommit to Mutal Assured Destruction if They ever cross the line, then it would be suicidal for Them to do so, so They won’t (if They are rational). If losing one city to a nuclear attack is unacceptable, then that is your line in the sand. You must precommit to full retaliation in that event, and make it known to the Enemy.
Why is “I commit to a proportional response and I will sometimes choose to escalate slightly so the trade is unfavorable to you” not a viable in advance policy. As in, “you blow a city that has 1 million population, I will select a slightly better city of yours to blow up”.
Sure. I was simplifying, and said as much. Tit-for-tat is an excellent strategy in Prisoner’s Dilemma (which is not the same game as nuclear war). But when both players use it, it leads to death spirals as soon as one side defects, even on accident, even due to faulty intelligence. Both players continue to defect thereafter.
But this is not the same game. We know that starting down path leads to Our destruction anyway, so why not gamble on a massive strike to disarm the Enemy so there’s a chance Our destruction isn’t total? Maybe We can get them all, or at least survive what is left.
If the Enemy knows that a tit-for-tat trade is on the table, They might take the deal if They feel it’s worth it, declining to retaliate further. (The kind of autocrat willing to pay the cost in their own army’s blood to invade a democratic neighbor might just take that deal from time to time.) If that’s unacceptable to Us, then We must not offer that deal in the first place.
The problem with that is that it risks slow escalation. As gilch says, tit for tat has no stability margin for an accidental defect, which isn’t so great when rights aren’t perfectly defined. 1.1*tit for tat is even worse, since now the problem actively escalates.
The problem with nuclear war isn’t that someone loses, it’s that everyone loses. Getting a city of 900 thousand blown up after you blow up after you nuke a million of theirs isn’t a good trade and does not make you think “That went well, let’s do it again!”. Even if you see your initial nuke as justified and their retaliation as not justified, you can at least tell that they’re putting effort into de-escalating into escalating. If the next nuke that flies only targets 810k, then 729k, at least the total damage will be finite.
My thought with the “1.1*tit for tat” is simply it makes the aggressor always come out in a losing position for initiating. My thought is they can see the diverging series and know the eventual outcome and either choose to stop or not. But you are possibly right, maybe 0.9 * (tit for tat) is a better policy to pre-commit to.
And yeah, reading the rest of your argument: you’re right, no winners. The aggressor isn’t really going to feel they came out ‘ahead’ dealing with the disaster from losing a slightly worse city than the one they killed from the enemy. For that matter, recent events show that popular opinion of other nations is relevant. The country with slightly more damage who was not the aggressor may enjoy far more foreign aid, which it’s going to need for it’s citizens to survive.
Sure, assuming rational players with known capabilities. If it turns out one side has far fewer functioning warheads or delivery systems than expected, it’s possible that humans override the trigger conditions for a smaller response, or that even with a full response, the initial attack isn’t as complete as projected.
Newcomblike problem: it feels like there are two decisions to be made, but there is only one, in advance. If the Enemy knows you will never fire your missiles, They can pick off your cities one-by-one with impunity. But if you precommit to Mutual Assured Destruction if They ever cross the line, then it would be suicidal for Them to do so, so They won’t (if They are rational). If losing one city to a nuclear attack is unacceptable, then that is your line in the sand. You must precommit to full retaliation in that event, and make it known to the Enemy.
In a game of Chicken with your loved ones’ lives on the line, your best move is to defiantly rip off the steering wheel before the Enemy can do the same.
I’m oversimplifying a little, but nuclear deterrence is all about game theory like this.
Newcomblike problem: it feels like there are two decisions to be made, but there is only one, in advance. If the Enemy knows you will never fire your missiles, They can pick off your cities one-by-one with impunity. But if you precommit to Mutal Assured Destruction if They ever cross the line, then it would be suicidal for Them to do so, so They won’t (if They are rational). If losing one city to a nuclear attack is unacceptable, then that is your line in the sand. You must precommit to full retaliation in that event, and make it known to the Enemy.
Why is “I commit to a proportional response and I will sometimes choose to escalate slightly so the trade is unfavorable to you” not a viable in advance policy. As in, “you blow a city that has 1 million population, I will select a slightly better city of yours to blow up”.
Sure. I was simplifying, and said as much. Tit-for-tat is an excellent strategy in Prisoner’s Dilemma (which is not the same game as nuclear war). But when both players use it, it leads to death spirals as soon as one side defects, even on accident, even due to faulty intelligence. Both players continue to defect thereafter.
But this is not the same game. We know that starting down path leads to Our destruction anyway, so why not gamble on a massive strike to disarm the Enemy so there’s a chance Our destruction isn’t total? Maybe We can get them all, or at least survive what is left.
If the Enemy knows that a tit-for-tat trade is on the table, They might take the deal if They feel it’s worth it, declining to retaliate further. (The kind of autocrat willing to pay the cost in their own army’s blood to invade a democratic neighbor might just take that deal from time to time.) If that’s unacceptable to Us, then We must not offer that deal in the first place.
The problem with that is that it risks slow escalation. As gilch says, tit for tat has no stability margin for an accidental defect, which isn’t so great when rights aren’t perfectly defined. 1.1*tit for tat is even worse, since now the problem actively escalates.
The problem with nuclear war isn’t that someone loses, it’s that everyone loses. Getting a city of 900 thousand blown up after you blow up after you nuke a million of theirs isn’t a good trade and does not make you think “That went well, let’s do it again!”. Even if you see your initial nuke as justified and their retaliation as not justified, you can at least tell that they’re putting effort into de-escalating into escalating. If the next nuke that flies only targets 810k, then 729k, at least the total damage will be finite.
My thought with the “1.1*tit for tat” is simply it makes the aggressor always come out in a losing position for initiating. My thought is they can see the diverging series and know the eventual outcome and either choose to stop or not. But you are possibly right, maybe 0.9 * (tit for tat) is a better policy to pre-commit to.
And yeah, reading the rest of your argument: you’re right, no winners. The aggressor isn’t really going to feel they came out ‘ahead’ dealing with the disaster from losing a slightly worse city than the one they killed from the enemy. For that matter, recent events show that popular opinion of other nations is relevant. The country with slightly more damage who was not the aggressor may enjoy far more foreign aid, which it’s going to need for it’s citizens to survive.
Sure, assuming rational players with known capabilities. If it turns out one side has far fewer functioning warheads or delivery systems than expected, it’s possible that humans override the trigger conditions for a smaller response, or that even with a full response, the initial attack isn’t as complete as projected.