will enable researchers to share, discuss and cite their early findings. It provides a lightly moderated and relatively informal channel for scientists to disseminate information, especially recent experimental results and emerging conclusions. In this sense, it is designed to complement traditional peer-reviewed journals, allowing researchers to make informal communications such as conference papers or presentations more widely available and enabling them to be formally cited. This, in turn, allows them to solicit community feedback and establish priority over their results or ideas.
A bioinformatics resesarcher, Pedro Beltrão, who has been participating in the beta testing, has a posting about it here. Extracts:
I have been participating in the beta for some months now and as it is mentioned in the editorial it will be openly available starting next week. All documents are citable (have DOIs), are not peer-reviewed (in the formal sense) and are archived under a creative commons license (derivatives allowed). The site has the community features (tagging/commenting/rating/RSS feeds) that you would expect and that will hopefully allow for requesting and providing comments on early findings. In summary an nicer version of ArXive for biomedical research
A framework for open science (in biology) can now go from blogs/wikis to pre-print server to peer-reviewed journals. Many ideas might die along the way and many collaborations might form by connecting early findings in an unexpected way.
Despite the long-time success of subject-based preprint servers in other disciplines (arXiv in physics, RePEc in economics, etc.), there has been very little interest to date, and indeed some active hostility, to the idea in biomedical research. I would guess that it is much more likely to attract interest from areas like bioinformatics than in medical research (where publishing un-refereed material is particularly problematic). If it does succeed (and presumably Nature has at least sufficient positive feedback from the beta trials to think it worth launching publicly), it would also be the first preprint server started and run by a commercial organisation, though Nature do say:
We anticipate that the content will be mirrored by academic partner organizations, several of whom have been involved with us in developing this service. As well as allowing it to become incorporated into the substantial information hubs already provided by these organizations, this federated approach will also help to ensure the long-term availability of the content — and act as a practical guarantee of the Nature Publishing Group’s pledge not to charge readers for access.
 e.g. see Rightscom’s 2005 report for JISC, Disciplinary Differences and Needs