This failed utterly.

Why? Once a company is beyond a certain size, employees maximize their expected value by spending time seeking internal resources. Budget, headcount, etc. If the lab is still part of the parent company, these games are still played.

You also have companies like Hulu, that exist at the pleasure of their "investors." This might work a little better, but Hulu can't act fully in self-interest.

Perhaps the best thing I can think of is a spinoff with initial funding, control of its own board, and an irrevocable IP grant. This would potentially be a huge head start. But then what happens if a competitor acquires your spinoff (for the IP grant) right away?

I guess all this seems more fraught with risk than the simple buy a random startup model.