Having spent a lot of time both as a manager and as an individual contributor, here's the rub (at least for programmers): software developers are a classic market for lemons. The bigger the org and the more operational complexity the more space there is for poor developers to hide out and draw a pay check. But even at a small scale it's remarkably difficult to measure productivity actually.
Therefore, what you need is intrinsic motivation. You need engineers who really care about solving the right problem at the right time in the right way. 35 hours or 70 doesn't matter if they can't get that right, and it's very hard to have management that is qualified to make that judgement on people (especially as an org grows and management becomes a full time job). So the problem is that given this opacity, finding someone who works 70 hours a week is a better proxy than someone who seems excited about a 35-hour work week—the former are people who are clearly driven whereas the latter are literally everyone. It's far from a good metric, but perhaps the best one available to the pointy hairs of the world.
Speaking from personal experience, it wasn't until I had a kid that I found out what I could do in a 40-hour week. Once I had that time constraint, it forced me to be more efficient in a very deep and fundamental way from the core of my being. When I was 25 I would hit 5pm and think to myself: I still have another 8 hours to solve this problem before bed. Perhaps this is post-hoc justification, but I think that makes sense when you are just starting out and not really competent yet. Once you've passed the magic 10k hours or whatever, I think turning over problems in ones subconscious can provide a lot more of the value. So sleep / exercise / meditation can all help elevate your performance far more than extra hours—but only if your brain knows what it's doing.
In my experience the lemons have no problem working 70 hour weeks; they weren't working that hard to begin with and know that they need to appear productive since they can't be productive. Whereas employees who are productive are tired after 35 hours or want to get home and work on side projects.
Well, but here's the thing: you've figured out that you're more productive on shorter hours. So really you want to prevent those people who work long hours from doing so, because they'll actually do better.
Not much research in sofware AFAIK, but there's a bunch of research in other fields showing maximal output is at 40 hours a week, with rapid fall-off at longer working hours. See http://www.igda.org/?page=crunchsixlessons
So it's your job as a manager to prevent burn out and maximize output by encouraging people not to work too long.
Therefore, what you need is intrinsic motivation. You need engineers who really care about solving the right problem at the right time in the right way. 35 hours or 70 doesn't matter if they can't get that right, and it's very hard to have management that is qualified to make that judgement on people (especially as an org grows and management becomes a full time job). So the problem is that given this opacity, finding someone who works 70 hours a week is a better proxy than someone who seems excited about a 35-hour work week—the former are people who are clearly driven whereas the latter are literally everyone. It's far from a good metric, but perhaps the best one available to the pointy hairs of the world.
Speaking from personal experience, it wasn't until I had a kid that I found out what I could do in a 40-hour week. Once I had that time constraint, it forced me to be more efficient in a very deep and fundamental way from the core of my being. When I was 25 I would hit 5pm and think to myself: I still have another 8 hours to solve this problem before bed. Perhaps this is post-hoc justification, but I think that makes sense when you are just starting out and not really competent yet. Once you've passed the magic 10k hours or whatever, I think turning over problems in ones subconscious can provide a lot more of the value. So sleep / exercise / meditation can all help elevate your performance far more than extra hours—but only if your brain knows what it's doing.