With any project like 20% time, you should expect that almost everything produced will be worthless. So what? All you need are 1-2 big hits to justify the entire program.
Letting employees work on what they think will be valuable is a bottom up approach that can reveal knowledge that's inaccessible to managers.
Some comments on here say that the program 'only' created Gmail. That's been a massive success for Google. That + the smaller successes it has led to likely justify all the projects that didn't pan out.