From a game-theoretic perspective, if the other side knew you thought that way then they should launch on your watch.
MAD only works if both sides believe the other is willing to retaliate. If one side is willing to push the button and the other is not willing to retaliate, then the side willing to push the button nukes the other and takes over the world.
If you can be absolutely certain the other side never finds out you aren’t willing to retaliate, then yours is the optimal policy.
MAD only works if both sides believe the other is willing to retaliate.
“Willing” can be unpacked.
Having the other party believe you are operating under a mixed strategy would be optimal, so long as: a) each side values the other side winning more than mutual destruction, which as humans they probably do, and b) accidental/irrational launches are possible but not significantly higher when facing a perceived mixed strategy.
If, say, the USSR and the USA were willing to strike first to win, but not willing to incur a 95% risk of mutual destruction for a 5% chance of total victory, the optimal retaliatory strategy is to (have the other believe you will) retaliate based on a roll of 1d20 - a roll of a natural one has one refrain from retaliating. That way, an accidental launch has a 5% chance of not destroying the world.
In practice, declaring a mixed strategy will probably be seen as setting up an excuse to update one’s actions based on the expected payoff considering the circumstances that have happened—i.e. to use CDT rather than TDT. Declaring an updateless strategy is a good way to convey one is operating under a mixed one.
From a game-theoretic perspective, if the other side knew you thought that way then they should launch on your watch.
MAD only works if both sides believe the other is willing to retaliate. If one side is willing to push the button and the other is not willing to retaliate, then the side willing to push the button nukes the other and takes over the world.
If you can be absolutely certain the other side never finds out you aren’t willing to retaliate, then yours is the optimal policy.
“Willing” can be unpacked.
Having the other party believe you are operating under a mixed strategy would be optimal, so long as: a) each side values the other side winning more than mutual destruction, which as humans they probably do, and b) accidental/irrational launches are possible but not significantly higher when facing a perceived mixed strategy.
If, say, the USSR and the USA were willing to strike first to win, but not willing to incur a 95% risk of mutual destruction for a 5% chance of total victory, the optimal retaliatory strategy is to (have the other believe you will) retaliate based on a roll of 1d20 - a roll of a natural one has one refrain from retaliating. That way, an accidental launch has a 5% chance of not destroying the world.
In practice, declaring a mixed strategy will probably be seen as setting up an excuse to update one’s actions based on the expected payoff considering the circumstances that have happened—i.e. to use CDT rather than TDT. Declaring an updateless strategy is a good way to convey one is operating under a mixed one.