The canonical example of an incentive gone wrong is of the manager offering to pay $10 for every bug found by QA or fixed by the developers. The results were exactly what you'd expect. To quote from Dilbert: "I'm gonna code me a minivan this afternoon!". But you can fix the problem:

• QA gets paid for every confirmed bug found. They have to show that the test case conforms to requirements and the code's behavior doesn't.
• Developers get paid for clean test runs. They don't get to claim more than one clean run per issue reported by QA.

Current Mood: bored