My point is that the design is wrong. Nobody expects [] as lvalue to update a value. Your argument is descriptive, mine is prespcriptive. I am saying that the C++ committee is wrong on this one (or whoever designed it this way)