Paging Dr. Strangelove ... Over at Defense Tech, Jeffrey Lewis has a wonderfully digressive post that starts with Cheney's request that the Pentagon draw up plans to attack Iran and ends with a discussion of how "organizations abstract reality in order to manage it" -- in this case, the reality in question being the statistical failure rates of nuclear warheads.

The redundancy that those rates lead to is shocking. For instance, when General Lee Butler did a review of SIOP in the early 1990s, he found that "one [target] was slated to be hit by 69 consecutive nuclear weapons." Why 69? Because for an underground bunker, a warhead has a kill probability of only 4%. Consequently, for the kill probability on the target itself to be 94%, you had to nuke it 69 times.

Hopefully whatever plans the Pentagon has drawn up for Iran will not incorporate that level of redundancy. But I doubt it: we may have more efficient weapons now for targeting bunkers, but there are also many more bunkers in Iran, some of which lie near densely populated areas like Tehran.

Suffice it to say, then, that if Iran really does sponsor a terrorist attack against us, and if we really do strike its bunkers in response with nuclear weapons, we will have forfeited whatever moral authority we had to the dry and altogether base precision of statistical certainty.