Philoctetes, we'll have to agree to differ on this issue. If a clue generates more than one letter for a message, I regard that as frustrating for the solver if he/she has the wrong one, not an extra challenge; especially if there are other ambiguities or difficulties in detecting the letters. If I get a wrong letter through careless parsing of the clue, that is my fault; if I get a wrong letter because the setter has overlooked or ignored other possibilities, that is the setter's fault. As a setter I strive to avoid such ambiguities, and if a test solver points any out I will change the clue. I won't claim to always succeed, but I'm happy to take on board criticisms of my clues where there has been ambiguity.
In this puzzle there is also an ambiguous entry (and therefore an ambiguity in the 'wrong' letter). At the time I thought it was there for thematic reasons, but now I've applied the instruction I don't see that it is, in which case it doesn't appear to have any point apart from making a central part of the instruction harder to see.
However, I'm not trying to make a big issue out of a few clues that I think could have been better. In other respects the puzzle is an excellent one. The clues are varied and have excellent surfaces, and a generous sprinkling of anagrams helps the solver to get off to a good start. The endgame is also very good, requiring of the solver some observation and perhaps a bit of lateral thinking. What I eventually found was not what I was initially looking for, and it was lovely PDM.