One thing to keep in mind in discussions of this sort is that MAD was an outcome, not a policy. When you look at nuclear war from a game-theory perspective, there are two ways to deter a first strike. One is to ensure that any conceivable first strike would not eliminate all of your nuclear force, and that sufficient forces would remain to destroy your opponent. The other is to ensure that the first strike never hits–i.e. active defenses (interceptors and the like.)

The assumption during the Cold War was that there would be no way to completely stop a first strike, due to the massive saturation attack involved. With the modern “rogue state”, that’s not necessarily the case; an attack involving one warhead is much easier to handle than an attack involving one thousand warheads.

Beyond that, the Cold War thinking started out with the assumption that civilian casualties were inevitable; the emphasis was on minimization, not elimination. And again, in a modern context that’s not necessarily true. Let’s say that the North Korean government says “pay us one hundred billion dollars or Tokyo gets it.” Do you really think that we wouldn’t just pay them? Even if they’d be blown up afterwards, Tokyo would be gone–and it’s cold comfort to the survivors that the Norks followed their victims into hell.

(It’s kind of ironic. If we’re attacked by one hundred warheads and ten get through, we’d consider it a success. If we’re attacked by one warhead and it gets through, we’d consider it a failure…)