I would love to see experiments similar to this salary experiment in the vein of 4day/week Nhrs/day. I'd guess that schedules may be shifted so that some people cover different four day periods, and one interesting side effect might be fewer interruptions in the office on Mondays and Fridays. As an early-in-the office kind of person, I find that the first slice of uninterrupted hours in the morning are often the very most productive.
In general, I tend to be most productive doing actual programming when I'm in the office and it's fairly empty. That is, whether I'm in before most others or in after most others.
However, I have multiple roles and tasks, only one of which is development.
I definitely get a lot of insight during the regular day from fellow coworkers, and tend to do more "operational" (not developing) work better when I'm alongside peers than when I'm there alone late. It's also the kind of work where multiple people looking at one thing really helps, since one person might miss something important and another might notice something important.
So I think it depends on the kind of work you're doing and how you like to work. And that is partially why letting employees work when they want, and even choose where they work, can lead to them being more productive in general, at least as long as you can trust them to choose a schedule that makes them productive instead of one that lets them be lazy.
