Quite. To believe that the inscriptions in any way dictate where the treasure is, or could be, by mathematical or lexical analysis, is the fallacy. The inscriptions are paradoxical, and so are the human reactions to them. If the inscriptions were in a foreign language the reader did not understand, they would no longer result in a paradox.
Self referential statements are excluded from logic.
I think you mean that your system of logic is incapable of handling self-referential sentences.
I had this kind of argument with a Sudoku setter, who claimed any Sudoku puzzle that could not be solved by pure logic was somehow "improper". What he meant was the set of solving rules he knew (or possibly anybody knows) can't solve all solvable grids, necessitating trial-and-error (which he regarded as improper) - but if a grid of this type still has one unique solution, it is still valid and there may be some undiscovered steps of logic that could solve it.
I regard trial-and-error as valid - it's the equivalent of
reductio ad absurdum. The trials are not random: one starts with a cell that only has two possibilities, choose one possibility and follow its logical consequences until it reaches a contradiction (hopefully before the logical chain comes to the impasse of another bifurcation), or the solution.