What I had in mind was something like the following: X is separated/proper iff for all curves C and all maps f : C \ c -> X, f extends to C in at most/exactly one way. Is there a good reason why this cannot possibly be true?