They now seem to think that their management has obtained good enough crystal balls to be able to determine beforehand which of the potential products will succeed and which will fail. Meaning that they can scrap the 20% time and put all their "wood behind the few arrows" they have determined will succeed (you know, like Google+).
You need your engineers to put up with mundane, boring work. Paying them to spend a minority of their time on an interesting project helps.
There was a good image I saw on reddit awhile back that I can't find off hand. On the left was titled what most people view success and failure as, and it had a single branch, you fail or you succeed. Then it had what successful people view it as, and it was a chain of failure after failure finally leading to a success.
If you eliminate failure, you eliminate the success you gain from learning from failures.
Isn't that begging the question in regards to whether or not you can know, ahead of time, which projects will or won't become viable and have a supporting business case?
See also the "people simply empty out" article posted on HN. Google seems to be falling for the PHB fallacy that people are just machines that crank out code. Perhaps that is what google is becoming, but I certainly wouldn't want to work there.