[Goose]

I do think you may be underestimating the deterrent effect of even a small nuclear program. If North Korea could credibly nuke LA, SF, or Seattle, it almost doesn't matter that 98% of the country would be fine and NK would be nuked into oblivion. That being a real threat would mean we would have to treat them very differently and never back them into a corner where they would consider that trade off. 

Likewise if Ukraine had even a few nukes and a reliable way to deliver them to Moscow or St Petersburg, that drastically changes the calculus of the war. Yes Ukraine would then cease to exist. But It would mean that Russia would have to be careful not to back Ukraine into a position where that seems like a viable option. 

Remember the whole principle of MAD requires irrational actors. Say Russia decides to nuke the US to oblivion. The nukes are in the air and on their way. Should we fire back? Logical answer is no. We're already dead. Killing hundreds of millions more and rendering the planet more uninhabitable out of spite is not a rational move. But deterrence requires us to project and say we will retaliate whether we actually would or not. The true Game Theory Optimal outcome if I recall is to actually take the decision out of human hands. Broadcast to your enemy that if condition X is met then your Nukes have a Y% chance of being launched by a randomized program. Ukraine could say that if Russian forces advance within 10 miles of Kyiv the nuke may launch automatically at Moscow. How lucky does Putin feel?

IMO you have it 180 degrees wrong. The success of MAD requires rational actors. It requires that both "sides" decide that the benefit of a first-strike attempt on their opponent is outweighed by the probability of a retaliatory strike that would obliterate them. That is the "rational" decision. MAD fails when one of the actors irrationally decides the potential benefit is worth the risk. Your notional scenario that a "rational" actor would NOT retaliate and therefore a first strike is "risk free" would first require an irrational gamble by the opponent. First of all, both sides have done all sorts of things to ensure the response is broad-based enough that even if some individuals chose not to react, enough will that it won't matter. Even if the NCA does not command a launch it is probable somebody who has the "facts" in the chain will. No sane, rational actor would ever take that risk.

Having some kind of "doomsday" machine makes no sense. This has been debunked many times. Anybody who is willing to assume the NCA of the opponent will chose not to launch a retaliatory strike can also assume the doomsday machine is a bluff for the same reasons. If the NCA won't launch a strike themselves, why would the build a machine that would? Same result as launching a retaliatory strike personally, same moral onus. There is also the additional possibility that the machine may not work. How does it decide to launch without relying on sensors that can't be "fudged" by the owners? The less such a machine is able to be bypassed the more its response to false alarms becomes unacceptably high. Blowing up the world by accident has to be traded off against the ability to disable the machine. Nobody would risk it, the benefits aren't high enough.

I do agree that the possession of nuclear weapons by a state like North Korea or Iran is a "problem" that has to be taken into account. It makes them even more dangerous. If a direct conflict between the US and North Korea happens, it could become an issue. If North Korea is losing a conventional war with South Korea and the US, it could launch a nuclear strike on the US. It would know that 1) that strike wouldn't be sufficient to prevent an overwhelming nuclear response and 2) North Korea and its leadership would cease to exist.  There would be no reason for the US to do anything else because we would have already suffered the worst-case strike North Korea can deliver. If North Korea is at all rational, they won't do that. The problem is they may not be "rational enough" to realize it won't work, or they don't care.

The idea that "we can't afford to put them in a situation where they might use nuclear weapons" is unfortunately unrealistic. Since it is patently obvious that a rational North Korean regime would realize it wouldn't survive such a choice, only a "crazy" regime would do so. It is impossible to know "how crazy" a regime actually is. They can make all kinds of threats. Which ones are "serious"? How do you know? It isn't rational that we should let North Korea do anything it wants because they might resort to nuclear weapons if we don't. It also isn't rational that we should gratuitously get into an armed conflict with them. It is very rational to ensure they understand that if they stay in their box we won't bother them. If they do not, we will protect our allies. If we won't North Korea has a blank check.

IMO the much more likely scenario if we get into a conventional war with North Korea protecting say Japan or South Korea is that North Korea realizes nuking us is pointless. Nuking Japan or South Korea is a lot "safer". The US now has to make a choice about how we respond. So does the target of that nuclear strike. It would make no sense to continue the conventional war if North Korea is just continuing to nuke our allies (while we refrain from a nuclear response). How would we respond? It is the old "Would the US risk New York for Bonn?".  What this scenario requires is that the US must have the ability to quickly neutralize the nuclear delivery capability of North Korea at the start of the war using only conventional weapons. Of course, the danger there is that North Korea decides its nuclear weapons are "use it or lose it" and therefore uses them on day one.

Substitute Iran in the above scenarios with the constraint that Iran can't actually hit the continental US with a missile fired from Iran, at least right now. They could strike CONUS by using a commercial ship to get close enough and the launch a missile, or with an "old style SSG".

[Goose]

I do think you may be underestimating the deterrent effect of even a small nuclear program. If North Korea could credibly nuke LA, SF, or Seattle, it almost doesn't matter that 98% of the country would be fine and NK would be nuked into oblivion. That being a real threat would mean we would have to treat them very differently and never back them into a corner where they would consider that trade off. 

Likewise if Ukraine had even a few nukes and a reliable way to deliver them to Moscow or St Petersburg, that drastically changes the calculus of the war. Yes Ukraine would then cease to exist. But It would mean that Russia would have to be careful not to back Ukraine into a position where that seems like a viable option. 

Remember the whole principle of MAD requires irrational actors. Say Russia decides to nuke the US to oblivion. The nukes are in the air and on their way. Should we fire back? Logical answer is no. We're already dead. Killing hundreds of millions more and rendering the planet more uninhabitable out of spite is not a rational move. But deterrence requires us to project and say we will retaliate whether we actually would or not. The true Game Theory Optimal outcome if I recall is to actually take the decision out of human hands. Broadcast to your enemy that if condition X is met then your Nukes have a Y% chance of being launched by a randomized program. Ukraine could say that if Russian forces advance within 10 miles of Kyiv the nuke may launch automatically at Moscow. How lucky does Putin feel?

IMO you have it 180 degrees wrong. The success of MAD requires rational actors. It requires that both "sides" decide that the benefit of a first-strike attempt on their opponent is outweighed by the probability of a retaliatory strike that would obliterate them. That is the "rational" decision. MAD fails when one of the actors irrationally decides the potential benefit is worth the risk. Your notional scenario that a "rational" actor would NOT retaliate and therefore a first strike is "risk free" would first require an irrational gamble by the opponent. First of all, both sides have done all sorts of things to ensure the response is broad-based enough that even if some individuals chose not to react, enough will that it won't matter. Even if the NCA does not command a launch it is probable somebody who has the "facts" in the chain will. No sane, rational actor would ever take that risk.

Having some kind of "doomsday" machine makes no sense. This has been debunked many times. Anybody who is willing to assume the NCA of the opponent will chose not to launch a retaliatory strike can also assume the doomsday machine is a bluff for the same reasons. If the NCA won't launch a strike themselves, why would the build a machine that would? Same result as launching a retaliatory strike personally, same moral onus. There is also the additional possibility that the machine may not work. How does it decide to launch without relying on sensors that can't be "fudged" by the owners? The less such a machine is able to be bypassed the more its response to false alarms becomes unacceptably high. Blowing up the world by accident has to be traded off against the ability to disable the machine. Nobody would risk it, the benefits aren't high enough.

I do agree that the possession of nuclear weapons by a state like North Korea or Iran is a "problem" that has to be taken into account. It makes them even more dangerous. If a direct conflict between the US and North Korea happens, it could become an issue. If North Korea is losing a conventional war with South Korea and the US, it could launch a nuclear strike on the US. It would know that 1) that strike wouldn't be sufficient to prevent an overwhelming nuclear response and 2) North Korea and its leadership would cease to exist.  There would be no reason for the US to do anything else because we would have already suffered the worst-case strike North Korea can deliver. If North Korea is at all rational, they won't do that. The problem is they may not be "rational enough" to realize it won't work, or they don't care.

The idea that "we can't afford to put them in a situation where they might use nuclear weapons" is unfortunately unrealistic. Since it is patently obvious that a rational North Korean regime would realize it wouldn't survive such a choice, only a "crazy" regime would do so. It is impossible to know "how crazy" a regime actually is. They can make all kinds of threats. Which ones are "serious"? How do you know? It isn't rational that we should let North Korea do anything it wants because they might resort to nuclear weapons if we don't. It also isn't rational that we should gratuitously get into an armed conflict with them. It is very rational to ensure they understand that if they stay in their box we won't bother them. If they do not, we will protect our allies. If we won't North Korea has a blank check.

IMO the much more likely scenario if we get into a conventional war with North Korea protecting say Japan or South Korea is that North Korea realizes nuking us is pointless. Nuking Japan or South Korea is a lot "safer". The US now has to make a choice about how we respond. So does the target of that nuclear strike. It would make no sense to continue the conventional war if North Korea is just continuing to nuke our allies (while we refrain from a nuclear response). How would we respond? It is the old "Would the US risk New York for Bonn?".  What this scenario requires is that the US must have the ability to quickly neutralize the nuclear delivery capability of North Korea at the start of the war using only conventional weapons. Of course, the danger there is that North Korea decides its nuclear weapons are "use it or lose it" and therefore uses them on day one.

Substitute Iran in the above scenarios with the constraint that Iran can't actually hit the continental US with a missile fired from Iran, at least right now. They could strike CONUS by using a commercial ship to get close enough and the launch a missile, or with an "old style SSG".