Not sure that user generated review comments are a good idea - it would be to hard to convince people (me included), that they'd be a good replacement for peer-review. You don't know the user's expertise, you have no guarantee that they have read the paper. Granted that last one is sometimes not true in real world publishing either - but at least the social pressure is there. It could also lead to gaming, asking your colleagues to give good reviews, or worse, companies that are paid to write good reviews, same as what has happened with Amazon etc.

Academia is largely a social construct, that's why it's hard to replace with 100% technology.