On the other hand, precise statements that are somewhere in the vicinity of the truth can be dangerous, because people tend to mistake precision for accuracy, and because modes of reasoning (e.g., formal logic) adapted to precise statements tend to be brittle -- one can deduce very wrong conclusions from slightly wrong premises.

A charitable reading of Smullyan would be that when a precise statement is made, he likes to examine it as closely as its precision allows, to avoid such dangers; and that a vague statement, so far as it's vague, is not worth the trouble of criticizing.

(For the avoidance of doubt: I think such a reading would probably be too charitable, and I upvoted Eliezer's comment.)