'Various ideas about scientific publishing', written by Etienne Joly and published in BioMed Central's 'Open Access Now' forum, elaborates some possible solutions to the current dead-end faced by readers, researchers and publishers of scientific papers. The paper attempts to answer two main problems - the low scientific and literary quality of many published research articles, and the prohibitive cost of access.
The suggestions appear over-elaborate - it takes 3300 words to describe an alternative to what is currently 'write a paper, send it to a journal, get it refereed, repeat as necessary, then charge the reader for access' - but contains some good supporting arguments for a number of recurring themes.
The initial suggestion is that referees could lose their anonymity and get paid according to how well they've done. It's obviously a good way to weed out the poor reviewers, but increases the cost of publication.
The second proposal is making a paper open for online criticism in a post-publication but pre-finalisation stage. The problem I see here is that the author assumes that comments can be anonymous. Anyone with a weblog knows that anonymous comments are useless, except for rare exceptions. However, it's incredibly difficult to have the confidence in your own knowledge to be able to comment on someone else's scientific paper online (I've thought about it on a few occasions, and always chickened out), and even harder to make a criticism in full public view. If you're asking people to just pick out papers that they think are the best, though, it works nicely (see Faculty of 1000 and the homegrown biologging for two examples - or DayPop, Blogdex, Technorati for some others in a different field).
The third is that the readability of most papers sucks, basically. One reason is that scientists aren't supposed to make their writing sound interesting, just to get their points across (this piece is probably proof of that). Another is that a large proportion of the papers are written by researchers for whom English, the lingua franca of the scientific world, is a second language. Another is that the journals don't proofread their publications very well, and probably couldn't without an in-depth understanding of each subject area and a very low threshold for boredom. This theme is basically about making publishers work for their money (when delivery costs are minimal and referees don't get paid, as at present), or paying specialists to do it for them.
The final theme is the impact factor - an artificial way of judging how good scientists are, by looking at the names of the journals in which they have published. There's currently no way for a scientist to build up a reputation by publishing on their own website, and there's a lack of incentive to publication in open access journals - at least until the effects of open access on citation rates are fully comprehended. I'd like to think that the changes in attitude necessary for funding bodies to pay upfront for publication in open access journals will be accompanied by increased respect for these publications, and that the adoption of weblogs by professionals (and their probable focus on freely-available publications) will obviate the need for such an engineered solution to the problem of identifying high- and low-quality papers. Somehow, following this line of thought, the reputation of the papers has to become independent of the journals in which they are published. We seem to agree on the need for journals to become magazine-style, readable filters (even in paper form), while the manuscripts themselves (available online) retain the high value and scientific integrity that makes them so important to us all.