Say you're a mayor, and you're trying to decide whether to get smart guns for your city's police force. (Assume again that such guns do become at least as reliable as purely mechanical guns.)
There's an upside to such a decision: About 10 percent of all police officer shootings happen with the officer's own weapon (according to 1990s data). Sometimes the shooter might have his own weapon and might use the officer's weapon just to make tracing harder; but sometimes the shooter starts out unarmed and seizes the gun from the officer in a struggle. If the officer has a personalized gun, the officer's life could be saved.
But there's also a downside, and part of the downside is EMP risk: The last thing you want is for your police force to be entirely unarmed in case of an EMP attack and its aftermath. Do you take steps to try to protect against such risk, for instance by insisting that you only get smart guns if they operate in "work mechanically if the electronics fails" mode, or if they can be "dumbed down" using some simple mechanical toolkit that all officers in the field will have on them? Do you make sure that each police car has a mechanically locked purely mechanical gun just in case? (You can't just store such guns at the station, I think, given that officers may be stranded far from the station.) Do you do this even if the steps involve some modest expense, or some modest delay in getting the personalized gun?
Or do you not focus on the risk, because it's too speculative? If the answer is that such speculative risks should be ignored, would you likewise ignore the risk if you're deciding whether to spend city funds to harden some local infrastructure against EMP?