Having just finished a semester as a student for the first time since 2002, I can say that advancements in web-based research tools have made research more efficient and enjoyable. I use Google Scholar all the time and, being at a research university, I can find most articles I want online. In the past semester I never once had to go to the library to hunt through a dead-tree journal. Integraing Zotero (for bibliography management) and Diigo (for web highlighting) into Firefox have also made my life easier. (I wish there was a Diigo extension for my brain so I could highlight any information anywhere, regardless of the medium. I think that goes well beyond Web 2.0.)
Research tools have come a long way, but I have some ideas about how much farther they need to go. I feel like most web research tools are still stuck in "Web 1.0," and not taking advantage of current technology. I've been envisioning a system for research publishing, indexing, and connecting that I think is possible to build. If you happen to read this and can do something to make it a reality, by all means steal every idea you can. Academia will thank you.
Let's start by considering the publishers. Journals give value to academia by ensuring quality for published research and delivering research to an audience. In exchange, journals take copyright of articles and restrict access to ensure they keep a sustainable business model. As an open-source guy, I don't particularly like the system but I can understand it. A "research 2.0" model can't simply do away with publishers; journals must retain their identity and they must still coordinate peer review of articles.
So imagine a wiki-like site where each article is a page. Unlike a wiki, the responsibility of posting each article would lie with a partner journal and the content would be static. The formatting of each article would be 100% consistent. Every citation in the article would link to the corresponding entry in the article's references, and each reference entry is a link to that article.
To protect the journals, no full articles could be posted for a period of 3-5 years after they're published. The journals would make an entry for the article, including an abstract and the article's reference list (so the links to other articles would be established), but the content of the article would be protected by the publisher, much as it is now. Researchers can't afford to be 3-5 years behind the latest literature, so they would still have to pay for content and keep the journals in business. Journals can't afford to lose readers, so they'd want to be part of this "research 2.0" system.
I like how Google Scholar can figure out how often articles are cited by other articles, although I'm not sure how well it works. In "research 2.0," the consistency of the links would allow for careful and accurate analysis of the relationship between references. Now here's where the 2.0 part really comes in: users could create an account on the site that allows them to mark articles as read, and articles would be "rated" based on how often they've been cited by other highly-rated articles. The site could analyze the articles a person has read, what has been cited in those readings, and recommend other articles to read. Just like a recommendation system at Amazon or Netflix, the system would use the knowledge of what I and others have already read to recommend what I should read next. Heck, I'd even pay to subscribe to such a system, and the journals could all take a cut in exchange for their cooperation.
I can't think of a single technological barrier to making this happen, and wouldn't be surprised if engineers on the Google Scholar team have already thought of it. I'm thinking about building a personal wiki to keep track of what I've read and what those readings have linked to, but I'm not sure if entering all that information is worth all the time and trouble it's sure to take.