The website would get far too many terrible papers.
Many papers would get no reviews.
Many reviews would be terrible, and done by non-trustworthy people.
The problem with the system you are suggesting is that it feels a lot like wikipedia, and who "wins" on wikipedia has little to do quality, and a lot to do with who is willing to spend the most time editing / wiki-lawyering.
Sadly, the evidence so far strongly suggests that you're right, and this idealistic approach won't work. That evidence is from PLoS ONE (which remember is the main example of how many innovations do work). They have had the ability for people to comment on and evaluate their papers for a couple of years now, and it's hardly ever used.
The website would get far too many terrible papers.
Many papers would get no reviews.
Many reviews would be terrible, and done by non-trustworthy people.
The problem with the system you are suggesting is that it feels a lot like wikipedia, and who "wins" on wikipedia has little to do quality, and a lot to do with who is willing to spend the most time editing / wiki-lawyering.