Unlike most other academic fields, refereed conferences in computer science are generally the most prestigious publication venues. Some people have argued computer science should "grow up" and adopt journals as the main venue of publication, and that chairs and deans should base hiring and promotion decisions on candidate's journal publication record as opposed to conference publications.a,b
While I share a lot of the sentiments and goals of the people critical of our publication culture, I disagree with the conclusion that we should transition to a classical journal-based model similar to that of other fields. I believe conferences offer a number of unique advantages that have helped make computer science dynamic and successful, and can continue to do so in the future.
I wholeheartedly disagree. I also can see some good points you make (i.e. the rotating "gatekeepers"), but I don't see a clear reason why our model is really better. Quite to the contrary, I believe that the double role of our conferences (network/new results vs. publication) harms both.
1. We do not really include the whole community at conferences but reject a lot of papers and thereby people. I just cannot believe that for many conferences 80% of the submissions are not interesting enough to be discussed at the conference. These rejection rates also lead to papers being resubmitted, sometimes several times, which in turn leads to very old results being presented. I'd rather see the new stuff.
2. Conference reviews are a one-shot thing. There is no way to really get into a discussion with the reviewers as in a journal. Rebuttals don't really solve this. And I don't see the advantages of hybrid models.
So while journals are not perfect either, a stronger focus on them would be an improvement for the whole community.
I think this is a classic example of arguing about the solution without defining the problem. As I see it, the problem that we need to address has (at least) 4 dimensions:
1. We need a means for researchers to publish unfinished work, which may or may not be of 'publishable' quality so that they can get comments from and discuss this with their peers. In some disciplines, this is accomplished through conference publication.
2. We have an overload of published papers - far too many for active researchers in a field to read - so we need to have fewer, better quality, more definitive publications. For example, rather than a PhD student publishing 4 or 5 papers in the course of their work, they should publish a single paper at the end. Of course, reducing the number of publications has significant implications for hiring, tenure and promotion processes. This is a problem for all scientific disciplines, not just computer science.
3. We need mechanisms for social, face to face interactions within the research community to support community building. Conferences and workshops play this role and it may be difficult for people to get funding for more informal replacement activities.
4. We need to ensure that research without excessive costs. Sadly, many journals by commercial publishers are very expensive and have restricted open-access policies. Publishing more in such journals rather than in open-access conferences would be doing the community a disservice.
The current system is not, in my view, working in a number of important respects and this makes it harder to pursue research that is unfashionable, unusual and which does not conform with the expectations of the community. So, please let's have a discussion about the problems that we face rather than premature arguments about the most appropriate solutions.
Displaying all 2 comments