Via an article from Nivi. A fast good read. - Sanjeev Singh

I'm skeptical about this research. - Igor Krivokon

(1) It might be easy to check the productivity on Ford conveyor. How could you possibly set an experiment in software engineering? 2 teams working on the same product, using different work weeks? I don't believe it was ever done. (2) They assume linear scale. Good = time x productivity. Again, it works for Ford, but not for real world software development. You need to deliver more at certain times. Hence variable work week should be more efficient than const 40-hour - Igor Krivokon

(3) I don't know of any worthy software project that was delivered by 9-to-5 workers - Igor Krivokon

Igor, I agree with (2). Sometimes a subteam is on the critical path and needs to go into crunch to deliver something that unblocks other teams. regarding (3), Ruby on Rails seems super popular, though I've never used it, and 37signals claim to abide by many of the ideas in this presentation. I also don't believe NASA's software team, widely regarded as pretty good, works at a 60-hour pace. - Sanjeev Singh

There is a lot of good software that is delivered at crunch mode rates, but that may have more to do with the quality of the people than anything else. A 10:1 productivity difference easily swamps a 30% hit, especially when you take into account the smaller teams better people allow you to have. - Sanjeev Singh