Keeping up with the Literature

There's an interesting post over at Corante on trying to keep up with the scientific literature. The blunt fact of the matter is that there is so much literature being published that I suspect 90%+ is ignored by the vast majority of researchers. It's not malicious, it's just that if you're not careful, you'd spend your entire life reading other people's work and never doing your own. Once you make the decision to ignore 1 paper or 1 journal, you may as well go the whole hog and just skim read the contents on a couple of major journals whenever you wander past them. I'm not advocating for this method, it just seems the only way to survive trying to 'drink from a firehose'. May I also go out on a limb here and say that 'the least publishable unit' correlates pretty well with 'the least interesting paper' and hence volume is in no way indicative of quality.

One alternative that i've wondered about is a system of trusted authours in a communal space. Imagine an on-line publication repository where it's free to upload your work, and collate comments i.e. make peer-review happen in real time much like a blog entry - then apply a few criteria:

1. Make a personal list of all the authours you rate as primary
2. Make a similar list for those you consider secondary
3. Use an aggregator to keep track of your primary and secondary authors
4. Filter out comments on your papers to those you consider first/secondary (comments from people you don't respect are considered noise) - perhaps throw in an anonymyser here, and then publish v2.0 if and when you think it is warranted
   
5. Create key word searches that check citations and references by author
6. Throw in options such as "keep an eye on my primary author's primary authors"

Yes, i know it sounds cliquey but guess what? it mostly is by default already. If you could get a reasonable number of people doing it, it could work out to be an interesting way of keeping up with the key authors in your field as well as your discipline in general.
An interesting property would be that journals could become more prestigious i.e. they select papers based on a common theme where the papers are more valuable because of the juxtaposition of other papers. The editors create value by linking the pieces together to make a statement about new research directions and findings.
Now that's a journal i'd like to subscribe to!
Anyway, just $0.02 worth from me. I suspect that we'll see innovation in this sector come from China anyway, they don't need to buy into this whole system of ludicrously expensive journals and could start something really useful using the internet as the starting point not as an up-start competitor.