[bowman]

We are already there in a sense: http://arxiv.org/

Interestingly the publishing industry has a firmer hold on the life sciences.

This is good too: http://math.ucr.edu/home/baez/journals.html

[eterno]

I think arxiv is a great start but still seems to be kind of internet add-on to the old way rather than being a rethink from the ground up to the research business.

I liked the "Cite as" feature, but there are fundamental issues which I think needs to be changed:

1. Anyone should be able to comment/discuss articles. With perhaps karma based user voting to surface the best comments. Each comment should again be citable.

2. The text should be in HTML- shareable simply and acted upon by the community.

3. A reddit style list for each topic listing the recent,high impact papers. High impact is again voted on by the community, again with people who are 'known' to produce better research having more weight.

[jballanc]

What you want is Nature Precedings (http://precedings.nature.com/) or any of the PLoS journals (http://www.plos.org/)