I should add to that - not just results but quality of results. Software development is a tricky profession to measure for quality - outside of bug count (which may be heavily influenced by quality of testing) and code coverage numbers (those can miss some really nasty edge cases) the only thing you can measure is how long it takes to make change to existing code.

I have seen numerous cases where the person implementing the code is rewarded and next one comes along, and rewrites the code since first one is too hard to change. He gets rewarded too and cycle continues.