Gary Brown posted an interesting experience on the XP mailing list yesterday. He described a team that set aside sensible hours in order to try to get more done. What happened? What we always predict will happen in that situation. The neat thing about Gary's experience is that they had metrics to back it up. Here's Gary:
Not only was PDT [personal development time, or research time] taken away, but people were asked to work nights and weekends as well as extra effort to work off a backlog of data transformation projects. This extra effort work was in addition to the normal project work. It was "voluntary", but the director made it clear that bonus money was at risk for those who didn't participate. I objected to this as long and as loudly as I dared. It happened anyway.
The developers soon figured out what was important, and started working on the backlog projects during the day, instead of their regular projects. They were tired, morale was low, and things were getting promoted to production without going through the normal QA process.
We measure velocity and defects reported each iteration. The charts show a dip in productivity, and a spike in defects. The other statistical anomaly of note was that unplanned time off also spiked (people calling in sick, basically).
For Gary, lots of overtime meant decreased productivity, more defects, and more people calling in sick. It's the textbook reason for energized work: sensible hours and work habits allow us to get more done.