It was many years ago but when I was at Google I started a new project in my 20% time and then took it all the way to production, it's still there and handles millions of requests per second because it's integrated into many of their biggest products (it wasn't a standalone thing). Yes, I had to write design docs, do PRDs, do privacy reviews, get launch approvals and many other things. Sometimes I had to integrate the project with other people's codebases myself. But, so what? These things had a purpose. The design doc for this project was dozens of pages printed out, it was my pride and joy.

I recall reading at the time comments on HN saying things like, 20% projects don't really exist, or that it's impossible to launch things at Google because of all the approvals you need. And then I went back to launching a 20% project into production.

If you do really care about making a project fly then filling out forms, getting people on board and getting approvals won't stop you.