Another Problem with the NAP
Deterrence policies tell some person/group that actions in a certain category will be met with a certain response. The point is to prevent the deterred party from performing some evil action that would otherwise be performed, without the consequences. The consequences chosen will specifically be things that would be of questionable morality without the announced declaration, because otherwise they'd just be expected. (No need to tell thieves that if we catch them in the act, we won't allow them to continue.)
Two examples would be to tell Saddam that if he nukes Israel, we will kill every last member of his extended family, and a policy of sending every nuke we have at the USSR should it send a single nuke at the USA.
Are these specific policies justified? That's debatable. They have to be evaluated by how effective they will be, what they will prevent, and what we will have to do should they fail (to not follow up would make all future deterrence policies ineffective and is generally not an option).
But by the NAP, they involve initiating force against people who did not initiate force against us. Killing Saddam's family if Saddam attacks us, or blowing up Russian cities should the Russian military fire a nuke, respectively. The NAP cannot accommodate deterrence policies. Whether these specific ones are right or not, the NAP fails to include a general case argument why all deterrence policies that initiate force must be wrong, and is thus an unreasonable way to approach the issue.
If a deterrence policy has an action that wouldn't actually be moral to do, what happens when it comes time to do the thing threatened?
If we don't do it, it has no power because the bad guys will realise it's just talk. If we do do it, it's not moral?