The short story is that scientist need "credit", as there are much more people that positions (not even to get a good position, just to get any). But the credit is given only for two things: publishing in already well-respected journals and getting citations.
In principle there is not problem with establishing an truly open journal. In practice - almost no one does (because doing anything but writing papers will very likely result in loosing academic job - as there is no credit for that, regardless how important).
However, there are people trying change the system; I wrote about it here: