Your strategic calculations are defective. You are balancing 3 versus 30 as if they are equally likely. You are prepared to assign a probability of 1 to 3 cities (i.e. go ahead with a strike), but you cannot assign anything like 1 to the 30 cities (and even 30 is a guesstimate). Given the 77 year record on non-use, it is probably more like .001 (still too high IMHO) but definitely tipping the balance against preemptive strike.

You're fired!