There is a very interesting discussion going on about open access publishing in a physics blog. The author also links to a recent editorial in Nature Physics.
A lot of the discussions are on how to certify content after submission to a pre-print server and how expensive should the whole process really be. Going through it the thing that most impressed me is that everyone seems to accept naturally the usefulness of a preprint server. Ken Muldrew in one of the comments says:
"something like the arXiv is sort of a bridge between conference talks and publications; a new phenomenon that doesn’t replace the old ways but rather adds to them, like email as it relates to phone conversations and page-written letters."
So why are there not more bioinformatics manuscripts in pre-print servers ? Most bioinformatic/computational biology journals accept submissions from papers already in pre-print servers. I subscribe to the arXiv feed on quantitative biology but most papers seem to a bit away from biology when compared to work published in say Bioinformatics, PLoS Comp Bio and BMC Bionformatics. This recent manuscript about horizontal gene transfer shows that arXiv does accept the type of work that I might participate in so I will try to in the future submit there first.