In academia, the big business is about publishing papers. There are different kinds of papers: conference papers, journal articles, technical reports, monographs.
Those kinds of papers are part of various academic social contracts:
- Depending on the research community, one kind is more valued than others. For instance, in computer science, conference papers are more valued than in some other disciplines [1].
- Papers are even linked according to community conventions, for instance, it is OK to have the same content in a technical report that is considered unpublished and in a peer-reviewed journal article.
In some disciplines, there is a practice called “journal extension”. Authors can re-publish a work that has already been published in a conference, later, in a journal. One practice is that the extension should contain some new material, e.g. one third of new content. These journal extensions exist for good, bad and ugly reasons. Let’s list the good reasons first:
- The authors discover important things to tell the world based on the feedback received at the conference
- The conference paper is only a short paper (2-4 pages), with only a preliminary presentation of the work.
Now the bad reasons:
- The conference paper has a length limit, preventing the appropriate presentation of the work. The sensible solution would be to allow for more pages in the conference papers. Note that we don’t live anymore in a world where proceedings are physical and of fixed size.
- The promotion committees only count journal papers. The sensible solution would be to have clear guidelines for committees, which would be specific per field.
And finally the hugly reasons:
- This makes more entries on the CV.
- This makes more money for predatory journals
Beyond that, the practice of journal extensions has many side-effects:
- It is confusing for readers, who don’t know what’s best to read for a given piece of work
- It decreases the value of journals, because one has no incentive to follow a journal if it contains too much refurbished stuff. This is essential: E. Berger’s popular csrankings.org does not contain journals because of the absence of systematic novelty.
- It is often a waste of time for everybody: for authors when they add uninteresting content in their journal extension, for reviewers when they review not-novel content, for editors, etc.
For all these reasons, I think we should avoid making journal extensions, in order to all focus on writing high-quality novel papers in the first place.
Naturally, the natural consequence is to setup journals that do not contain journal extension papers, but only, fresh, novel, papers (aka “journal-first” papers).
Finally, science is a social process [2] and I am aware that the journal extension practice is well spread in the some communities, including mine. As such, getting rid of journal extensions is disruptive and it will not be easy to change.
What do you think?
--Martin Monperrus
Stockholm, Jan 24 2019
[1] Conference Paper Selectivity and Impact (CACM)
[2] Social processes and proofs of theorems and programs
Credits: Thanks to Denys Poshyvanyk and Robert Feldt for insightful remarks