Comparatives are in general unambiguous. For example, in interpreting the sentence "John is taller than every girl is”, we need to compare John’s height with the height of the tallest girl, while in interpreting the sentence “John is less tall than every girl is”, we need to compare John’s height with the height of the shortest girl. Intriguingly, in a context in which there is a highway with a speed limit of 50 mph and a minimum required speed of 35 mph, the sentence “Lucinda was driving less fast than allowed” is ambiguous between two readings: her speed was below the speed limit, or her speed was below the minimum required speed. In this talk, I show that neither scope ambiguity nor the assumption of different silent operators can account for this kind of ambiguity. Rather, I propose that (i) based on interval arithmetic, there is a generalized recipe to derive the truth conditions of all kinds of comparatives, and this recipe does not involve any scope interaction or choice of silent operators, and (ii) the ambiguity of Lucinda examples is based on the use of permission-related modals in their than-clause, which leads to a mention-some/mention-all ambiguity for the than-clause. (I'll discuss some additional data and provide an exploratory discussion on the conditions of licensing mention-some answerhood.)