This question showed up in the close review queue for C++:
I was leaving a comment as to the inappropriateness of the
c++ tag for this question, when I went to the question and found this:
The question does not have the
c++ tag, and has never been revised. Surely any ninja revision would be over a week old, far too old to influence the close queue filtering.
Diamond mod Oded was able to confirm that the honeypot messes with tags sometimes.
That's bad, because it messes with the quality of the question and you can no longer make a judgement about the user's reaction when you show them an alternate version of the question. In this case I was leaving a negative comment using the review queue, and it did show up on the real question! Of course I deleted it because I realized that it wasn't deserved, but if I was only paying enough attention to decide the question was ok (I do know enough Java for that), then I might not have visited the question.
If you're going to make a fake version of a question to show the reviewer, then review actions including commenting need to be disassociated from the real question. In order to not discard useful comments, you may need to drop the facade if the reviewer tries to leave a comment, and send them to the real question to decide whether it still applies.
Related but different: Close vote review queue tag filter broken? (different because there the system recognized that the tag wasn't present, and passed it through the filter anyway)
Edit: Neither of the linked questions, or answers, address comments left on faked answers via the review audit system. There's a solitary comment that mentions the issue, but nothing is resolved, since it isn't answered. This bug report is separate from those two.