Research (as a whole) is a lot like mining. The human race can put in so many hours of work researching new areas, and the majority of it does not produce immediately fruitful work. Only a very small percentage of the work done results in "breakthrough" technologies that are the main reason money is invested into research anyway. But I don't understand why people think that if we put less time into it, we will cut out all of the garbage and only bring out the really good stuff. You have to wade through a lot of dirt to find a diamond, and cutting your crew down won't help that. It helps to look in the right places, but at the end of the day, you really don't know what will end up producing magnificent results.
In other words, not everybody is doing great research work, and they know it.
I say this as a Phd student who is "on leave" and instead doing a startup. I spent about a year as a Phd student and I thought the whole system is mind bogglingly inefficient, I felt like I'm wasting my time. This was in E.Europe, so not at a top US school --- I'm sure doing a Phd there is great.
Coupled with low pay and the lack of a payoff at the end just made it a very easy decision to do a startup instead: ultra efficient mode of operation, equally low pay, but at least there is the slim chance of a payoff.