I wonder if we have dead man switches.. In a MAD world you would want to launch without human intervention from isolated systems.... I wonder how that would work... maybe it's not possible.

At least two, one for each Cold War power:

- ERCS: https://en.wikipedia.org/wiki/AN/DRC-8_Emergency_Rocket_Comm...

- Perimeter: https://en.wikipedia.org/wiki/Dead_Hand

I have always wondered, if I were at the helm of a government asserting MAD capabilities, I would probably just not hold up my end of the bargain. Threatening MAD makes total sense, but once your enemy has assured your own destruction, is there much sense in actually following through? All you are doing is assuring millions of innocent people die for the actions of some generals.

All this to say, that it is unthinkably inhuman to have MAD be an automated response.

Assuming your counter-attack doesn't make the difference between complex civilisation surviving and being wiped out on Earth, it has the potential benefit of preventing the initiating party from repeating their attack against other nations.

Even if the initial launch was a genuine mistake, there is some value in launching a retaliation because it prevents such mistakes (or "deliberate mistakes") from happening in the future. In fact, seeing how much the mistake costs the initiating party may be sufficient motivation for all the remaining countries to properly commit towards disarmament.

Your enemy might have intelligence that will allow them to learn that your MAD is a bluff, in which case they have all incentive to attack immediately.

Even modern conventional warfare utterly destroys countries, like Syria. So nothing to gain compared to science, technology and trade.

Though it's not what they've done so far, it's what they learned, and what they may have left behind that is more problematic.