In my country at least, but I'm sure it is the same for most, researchers survive by putting their quota (fixed by the ministry of research) of publications in well renowned journals. The risk of having someone "steal" your idea (in the very loose definition) and beat you to the publication, for instance by having a bigger research team, thus putting you out of a job, is a real one.

In an ideal world your system would be perfect, putting the advancement of knowledge before everything else (as open source does with software), but researchers have to look out for themselves before all, considering how their performance is evaluated and how that affects their livelihood. Unlike open source software, there aren't a lot of ways to monetize scientific publications. I would enjoy seeing things change, but change is slow to come. Especially in an area as pachydermic and anchored in traditions as academia.

I really like Authorea and I will use it for non-published work, but I feel that limiting private publications so much is going to limit adoption. But then again what do I know, I guess we will see how it goes.