Stas has just managed to scare me.
I don't like the thought of millions of people dying and the world rendered permanently uninahbitable, disallowing for the survival of anyone but the folks in the super-long term shelters or "doomsday vaults".
I just took the idea of nuclear annihilation and brought it to an extreme. We should also enhance warheads with gold, zync, etc. to provide long-term contamination; in that case, the economic potential of the enemy would be crippled in entirety even if some of his people manage to survive.
But I'm not too thrilled about putting it on a dead man's switch. Dead man's switches can fail. It should be a command-activated system, rather than one which is periodically de-activated. Controlled from the ground, the space stations and in the future - from a permanent moon-base.
No, the deadman switch is not periodically de-activated.
It's permanently inactive, which makes it a lot safer.
Basically, the system registeres possible nuclear detonations on Earth through observation (I'm sure you could use GEOsats for that), and in case a multitude of detonations have occured on FUN territory, the computer then assumes that a nuclear assault has taken place.
What are the failure potentials?
1) mis-identification - pretty hard to have a multitude of megaton flares on a designated territory happen any other way than a nucelar war, so it's pretty safe to say it won't fail.
2) general computer failure or error that triggers death - the potential is not much greater than human anxiety failure in case of a misidentified threat amidst tensions.
3) attempt to disable the system by attacking GEOstationary satellites with KSATs - that is a clear sign of war and the system would act accordingly in case such an attack takes place
A special code could be devised that would immediately disable the system upon transmission, for destruction, deorbit or other necessary actions.