We love Stack Overflow. The first time you use Stack Overflow, you realize that this is how Q&A forums should work. No long trails of useless comments like "I have the same problem" or "Thanks", and no endless digging through hundreds of comments until one finally finds the ultimate answer to the original question (or not). Once you start using Stack Overflow a little seriously and start contributing, you realize why it works so well: it makes it rewarding for those who help. Analyzing the reasons why it is so rewarding, whether it is the fact that questions are reviews and rated, resulting in mostly high quality questions, or whether it is the mere collection of reputation, is beyond this blog post. But because it is so rewarding, there are sufficiently many people who are willing to spend time helping others, and hence anyone coming with a question can get it answered in often a very short amount of time -- quite often in significantly less than 10 minutes.
This positive community behavior that emerges from such a well-crafted, yet simple mechanism of reviews, votes, and reputation, has inspired us to ask the question: can the same, general idea be applied to scientific publishing and could this lead to a better way to further the scientific cause? In our experience, discussing publications is almost as important as reading them. Yet, today there are no obvious places to discuss papers with those who are most vitally needed for a fruitful discussion: the authors and their peers in the same field. Conference papers get a chance to be discussed with the authors at the conference itself, but even there there often simply isn't enough time for questions. Hallways conversations at conferences are extremely useful, and certainly solve this problem in part, but what about all those readers who couldn't attend the conference or read the paper much later.
This past Sunday we released a new version of BibBase. This version adds significant new capabilities to the site incl. several features inspired by Stack Overflow and which we hope will allow us to test the hypothesis that Stack Overflow-like discourse is useful for scientific publishing as well. The features include the ability to respond to publications on BibBase, vote publications up or down in terms of their scientific merit, and vote on responses based on their degree of being constructive and contributing to the understanding of the paper. Authors of papers or responses that get voted up receive and increase in their "expertise" in the topic of the paper, as defined by its keywords. There are currently no distinguished types of responses, but anything from a review, to a question, to a constructive remark seem appropriate.
Keywords have gotten a much more prominent role. To reflect that, we have also created separate pages for all papers on a specific keyword. These papers are sorted by their votes. This means that keyword pages indicate which papers on the topic are the most highly regarded -- at least as measured by BibBase's users. In several sub-fields of computer sciences and other disciplines have people created designated community pages, where they manually curate list of publications in the area and often also enable some degree of discussion. BibBase keywords automate some of this, and we hope this will make it easier for more such pages to be created. Consider a keyword like "neural nets". Anyone who uses BibBase and published in this area and uses that keyword in his or her publication list, whether it is in bibtex, on Zotero, or Mendeley, will have his or her papers included on the corresponding, automatically generated and always up-to-date keyword page on BibBase.
Ultimately, we believe that scientific publishing can benefit from an accelerated means of discussing new publications, long before the actual conference where they are published takes place, or the journal where they are bound to appear comes out. Furthermore, the creation of the Internet has raised the question whether journal and conference based publishing is even still required. Many reasons speak against this model that has been established during a time in which it was far from trivial to disseminate new ideas. This is no longer true today and we all ought to reconsider who scientific publishing should look like, provided all the new means we have with the Internet.
Eventually we hope that publishing papers on ones web site and participating in a fair and public review online, can lead to a new way of peer reviewed publishing that avoids unnecessary delays in publishing, low quality of reviews, and unnecessary uncertainty about the true value of specific publications.
If you believe in this being the right direction to go, then please give the new BibBase a try and let us know what you think. We won't get there without your feedback.