> Can you give a reference to back up your interpretation?
Alas, no. I do recall being astonished at my claim, and then being convinced by a colleague (which was backed up by plain language on the CMI page, in my memory...) but now that I'm re-reading (current and archive.org'd) I cannot find such a thing. Disturbing. Yet relieving. Fuck my memory.
Your interpretation is correct. The official problem statement [1] is "does P = NP?", so either "P = NP" or "P != NP" is a valid answer.
Can you give a reference to back up your interpretation?