In the case of Google's 20%. The numbers are easy to quantify, how much does 20% of the engineering staff's time cost? How much money has the output of 20% time earned? It might be that Google thought the work was too unfocused, and outside of some pretty obvious early big hits, fairly little else of value was coming out of the work year after year.
The best way to deal with it, I think, is to have people submit proposals, pick the best, then take those people out of regular work for a few months and put them into a "lab" where they come up with an MVP. If it looks good at that stage, invest more in the idea.