As part of the Miller Institute here at Berkeley, I've had the wonderful opportunity to meet weekly with scientists from across disciplines (including chemistry, physics, astronomy, biology, and mathematics). One topic that comes up a lot at our discussions is how to address the challenges, while maintaining the benefits, of peer-review.
In science we design experiments, test hypotheses, draw conclusions, write up manuscripts interpreting the results, and then submit our research findings for peer-review. This process takes different forms in different disciplines. In Biology, generally, peer-review takes the form of submitting a manuscript to a journal, the editors take a cursory read through the research and decide if it fits the scope of the journal. If it does, the editors seek out experts in the field of study and request that these researchers read the article and write a summary of their professional opinions of the strengths and weaknesses of the article, specifically assessing whether the methods and analyses are sound, and should be published, or whether more/different work should be done. These reviewers are, typically, anonymous, meaning that the authors will not know who has submitted suggestions regarding the submitted research. Often it takes a couple rounds of comments from the reviewers/editors, revisions and resubmissions, and then - yay! - the paper is published. The time from submission to publication can take as little as a few weeks, or upwards of 12 months.
In my opinion, the purpose of peer review is to strengthen Scientific research, specifically identifying and addressing concerns about methodology and interpretations that might lead to faulty conclusions. Peer-review can also catch cases of academic dishonesty (including plagiarism, and falsifying data). It isn't always perfect (see examples of falsifying caught after publication, here, here, here, and here), but generally peer-review seems to be a good tool for facilitating the self-correcting nature of Science.
A benefit often touted of anonymous peer-review is that reviewers (especially new scientists) need not worry about being retaliated against if they bring up concerns regarding a manuscript. A major concern about the traditional peer-review, is that the time between submission and publication can drag out the sharing of research-findings, and hinder scientific advances.
It wasn't until I spoke with scientists from other fields, specifically astrophysics and astronomy, that I learned about a completely different way to achieve peer-review, using a pre-print server called the arXiv (arxiv.org).
I haven't used the arXiv, but my understanding of it, as explained by my colleagues, is that researchers upload their draft manuscripts to the arXiv as soon as the paper is in a complete or near-complete form (for example, at the same time as I would submit a manuscript for peer-review to a journal). Others might even submit abstracts or preliminary data, to stake a claim on a particular research topic. The arXiv is then published daily, completely open-access, so anyone can immediately review the most-recent research, add their own comments to the site, and the authors can upload revised versions, as they choose to address comments from the community. While some authors choose to also submit to peer-reviewed journals, it is not required of authors in all fields; submitting to the arXiv is sufficient (for academic success) in some disciplines.
In addition to getting scientific research out immediately, another benefit of the arXiv are that the reviewers are not anonymous, so discussions can ensue about particular concerns. However, because there is no filter as to what can be uploaded to the arXiv, it is entirely possible for manuscripts with no scientific merit to get wide-distribution and, perhaps to those not familiar with the format of the arXiv, the facade of scientific approval.
The debate in Biology, and especially Computational/Theoretical/Quantitative Biology going on now is whether the benefits of the arXiv outweigh any drawbacks. Most importantly, how beneficial is it to get research to the community as quickly as possible?
There is one other drawback to the arXiv that makes me, as a potential submitter, very nervous: being scooped.
A paper is "scooped" if someone else publishes the same (or very similar) concept before you get a chance to publish yours. But, wait, if it is on the arXiv, isn't that documentation that I had the idea first? Well, yes, but... the arXiv isn't commonly used in Biology yet, so it isn't clear how important or how much priority will be given to authors who publish there before "traditional" peer review. This is especially concerning if the novelty of the paper is the idea (which is easy to reproduce with the same or different data) versus a method (which is more difficult to replicate). Maybe this isn't a valid concern, because anonymous reviewers could, one might argue, just as easily "scoop" ideas from a manuscript they have reviewed. Furthermore, perhaps posting ideas/research early might facilitate more collaborations instead of competitions between research groups.
All said, I think that submitting to pre-print servers can be a very valuable tool for facilitating scientific discourse and advances. Will I start submitting there? We will have to wait and see.
See also discussions here: http://gcbias.org/2012/07/22/arxiving-our-papers/.
And, from the above post, a link to finding which journals allow what kind of archiving: http://www.sherpa.ac.uk/romeo/