The recent kerfuffle caused by Butler's article in Nature seems to have gotten quite a number of stones rolling. What I thought would be a rather slow process seems to be speeding up considerably because of Nature's rather unclever attack on PLoS. Cameron Neylon's blog post "What I missed on my holiday or why I like refereeing for PLoS One" points out what the real danger of PLoS One is for basically all traditional journals: Today's system of scientific journals started as a way to effectively use a scarce resource, printed paper. Soon thereafter, the publishers realized there were big bucks to be made and increased the number of journals to today's approx. 24,000. Today, there is no technical reason any more why you couldn't have all the 2.5 million papers science puts out every year in a single database. It doesn't take an Einstein to realize that PLoS One is currently the only contender in the race for who will provide this database. For all the involved, it is equally clear what the many advantages of such a database would be. Consequently, traditional publishers are rightfully concerned that their customerbase is slowly dissappearing.
I'm now not alone anymore in believeing that we are seeing the beginning of the end of traditional journals. The acceptance of PLoS One is a quantitative marker of this development and the positive reactions I get from virtually everyone involved (even scientific editors at traditional journals) underscores the numbers.
Precurser to this publishing reform was access reform: scientific papers are the result of publicly funded research and should be publicly accessible. This reform appears now to be well underway and will probably conclude in 2-3 years. Both reform movements have their base in the more general open science movement. The goal of this reform movement is to have full public access not only to the published papers, but also to the raw data, ideas and reagents for sharing among scientists. There are still plenty of problems which have to be worked out before open science can become a reality, if it is even feasible. One of the more easy to solve problems (one that is shared with publishing reform) is that of how to attribute credit. If we all publish in the same database and share ideas online, how can two scientists competing for the same position or grant be assessed objectively? Cameron Neylon just returned from a conference on open science and writes in his summary: I would disagree here and argue that a multivariate portfolio is exactly what is required. Different universities/employers will focus on different aspects of a researcher and value some of his/her contributions more than others. I don't think there can be too many measures to capture the complexity of scientific output. I'd like to see an aggregating service, maybe based on services like  OpenID, where a flexible portfolio can be organized such that employers can easily search for the traits they are looking for and find or compare the people who maximize their efforts on these traits.
The analogous problem to comparing researchers is that of comparing papers. I have already written about this problem and I think it is easy to solve. I think most researchers would gladly pay for a service which has a track record of picking the most interesting, groundbreaking and well-done papers from the 2.5 million every year. Today's professional editors would be a great pool from which such services could recruit employees.
Like dino-oil, there's still some use in long-dead structures.