An opinion piece in the ACM Communications by two CS professors describing the current situation in CS publishing: unlike other academic disciplines that emphasize publishing in peer-reviewed journals, CS as a discipline emphasizes publication at conferences.
They theorize this is due to a number of factors:
* Conferences give faster review and publication turnarounds than journals (implied but not stated: CS is a rapidly moving field where this is particularly vital).
* Publicity. "The best way to get your sub-discipline to know about your results is to publish them in the leading conference for that subdiscipline."
And it is problematic for a few reasons:
* Conference papers are usually limited to be shorter than journal ones, meaning that it's harder to explain results in reproducible detail.
* Conference papers are often not reviewed as thoroughly as journal ones.
* CS as a discipline has splintered into so many subfields and their corresponding conferences that presenting at a conference doesn't actually disseminate work to everyone who should see it.
The authors go on to suggest that the CS community shift their focus to journal publication for more thoughtful certification of quality work, and give a number of things that could support such a shift.
* Use centralized web archives to store papers publicly online
* Speed up journal review cycles
* Make everyone who submits a paper "pay" to have that paper reviewed by reviewing papers themselves
* Allow multiple certifications per paper -- that is, make it ok for a paper to get reviewed and approved by two or more publications.