Physical Cryptographic Warhead Verification
lnsp.mit.edu
lnsp.mit.edu
The thing about a one-time-pad is that it's only secure if you use it "one time". A malicious weapon inspector can make multiple measurements of known objects using the foil to deduce the foil's structure and, if he knows enough about how your system works, use that to deduce specifics of the warheads it's meant to verify.
I suspect this is just a bad analogy and this system would be fairly hard to crack, but their explanation should probably make a point of not comparing the foil to a one-time-pad, or explain why knowing the key doesn't give away protected knowledge.
Zero knowledge means your peer participates in the authentication without needing you to inform them of your nuclear secrets, and them reading positive communication without needing to tell you a secret.
So the inspector should provide the foil sample, and the foil should be comprised of a publicly known composition, meanwhile they should still control a secret of some kind, and use a combination of the additional retained secret, and the exposed material as their own authoritative means of secretly confirming whether they can prove that know the difference between an authentic positive or negative result.
This way, the weapons producer cannot forge an arbitrarily desired result, by providing both the means of measurement, and the measured object.
The inspector must decide the degree of proof. The manufacturer must submit to a reasonable revelation of partial details.
This concept merely has the manufacturer stating that the foil signature is valid, because the foil signature is valid.
And if the opponent removes the Nurf Hat--claiming they decommissioned the nuke--how would you know it's true?
EDIT: For an explanation of how this might be done, see:
http://crypto.stackexchange.com/questions/2249/how-does-one-...
Also tackles this problem, from what I understand.