The advantage of "curved" probability over linear probability is not the base chance. As hirch points out, both can easily be translated into each other, and it is true the curve can obsure the percentage.

The advantage is with the *modifiers*. With linear probabilities, a "+1" is absolute. Be it 1%, 5%, 10%, or 12.5% it is always the same regardless of the base chance. With "curved" probabilities, the "+1" has different percentage values depending where on the curve the base chance resides. This added capability allows games to institute "diminishing returns" on modifiers.