I would guess that it is because it gives some freedom to the compiler. e.g. If you have two pointers 'foo' and 'bar' that point to two separate structures (e.g

I would guess that it is because it gives some freedom to the compiler. e.g. If you have two pointers 'foo' and 'bar' that point to two separate structures (e.g. two arrays of ints), the compiler can always assume that the pointers, even with some adds/subtracts, will never 'collide', i.e. foo will never == bar, regardless of their relative memory positions.