The internet allows for a much more powerful system than the current journal system, much more powerful than even an open journal system.
Some things I'd like to see in a unified online open system:
* Hyperlinking between papers
* Discussion threads for papers
* Collaborative mark ups of papers, so that difficult papers can be communally dissected and fleshed out, or so that students can work through a paper and provide a mark up to ease the reading for other students
* An ongoing wiki for every subfield, detailing current outstanding problems, papers to read to get up to speed, most recent progress, etc, as well as curating accepted knowledge. Wikis should also be able to be marked up by students, so that difficult material can be broken down and fleshed out for the sake of other students.
You've got some great ideas, but some of them are stymied by academic ego. For example, in peer review, reviews are generally unnamed. A discussion thread based system can suffer from what peer review did - people refuse to comment negatively with an attached name.
Many of these suggestions are being attempted in journals like PLoS One (plosone.org) or through groups like Faculty of 1000 (http://f1000.com/) for commenting/discussion. Many journals now have comment fields on their articles, and you'll notice that most of them go empty. All the discussion happens behind closed doors before publication in anonymous peer review. Nobody wants to stake their reputation on something not polished.
edit: I partially take-back my suggestion of Faculty of 1000. Apparently, they are not open to unaffiliated and require a subscription. I like what they're doing, but it shouldn't be so ivory-toweresque.
In the ideal world we would have hypertext as Ted Nelson intended it, but in this world contributing to such a thread does not pay the bills for a scientist. He is financially better of by writing a three page paper that is mostly filler around a valid argument. So, that is what he will do.
Some things I'd like to see in a unified online open system:
* Hyperlinking between papers
* Discussion threads for papers
* Collaborative mark ups of papers, so that difficult papers can be communally dissected and fleshed out, or so that students can work through a paper and provide a mark up to ease the reading for other students
* An ongoing wiki for every subfield, detailing current outstanding problems, papers to read to get up to speed, most recent progress, etc, as well as curating accepted knowledge. Wikis should also be able to be marked up by students, so that difficult material can be broken down and fleshed out for the sake of other students.