Fixed point representation is more useful than most programmer know. Many lossy compression algorithms (such as MP3) can be done in fixed point computations. However, there are a few problems associated with fixed point computation. First of all, the smallest and largest absolute values that can be represented by a fixed point representation are limited by the implicit fixed binary point. This makes fixed point representation not very useful for applications where the range of numbers can be huge.
This is why we have floating point numbers.