You do know about arxiv.org, right? This is the de facto place for preprints in all of the fields that I know of, and is pretty much removing the need for publications altogether. But it is definitely limited from a technology perspective. Search is poor, subscription types are limited, no comments or rankings (though many would argue that these last two are undesirable, at least on the original site), etc. Would be nice if you could provide a front end to the arXiv, but the amount of data on there is insane and you will need some hardcore computing power to keep up. You could start with the cs papers first, though.