[AISWorld] Practical suggestions for improving journal cycle times

[Paul Ralph]

Dear all,

Here are some ways to reduce review cycle times:

1) Give reviewers only two options: reject or accept with minor revisions.
2) Limit revisions to one cycle, i.e., manuscript, revision one, galle
proofs, published. No revisions two and three.
3) Direct reviewers specifically to evaluate methodology and rigour, rather
than respond to tone (see http://www.paulgraham.com/disagree.html).
4) For a 'minor revision' decision, insist on list of specific action items
rather than a vague discussion.
5) Limit review periods to one month.
6) Officially suspend dysfunctional reviewers from authorship, i.e., if
someone, fails to complete a review or does a terrible job, they lose the
right to submit papers to that outlet for one year. Of course, this has to
come with a limit, e.g., the right to refuse more than two reviews per
year. It also has to be transparent; silently blacklisting people
contributes to nepotism (see recommendation 12).
7) Flatten the editorial hierarchy - one paper doesn't need both an SE and
an AE. One editor per paper is enough.
8) Limit editors to one month of decision time. Dismiss editors who can't
make these deadlines and suspend their authorship privileges (see
recommendation 6).
9) Abandon blind review. Blind review is supposed to free junior reviewers
to reject the papers of their more powerful peers without repercussion.
This obviously isn't working. It's protecting bad reviewers from
well-deserved backlash. Knowing your name is on a review encourages you to
stick to actionable suggestions rather than name calling and quibbling
about tone.
10) Stop peer-reviewing position papers. Peer review is a system for
checking the methodological rigour of empirical research, not for analyzing
essays. Treating a position paper as a "peer reviewed contribution" is
absurd. Journals are for empirical science. If you want to share an
opinion, start a blog.
11) Develop a clear set of desk-reject rules that allows more desk rejects.
Publish them, let them be challenged and continually evolve them. If these
policies are regularly updated, they'll save everyone time and drive up
research standards. For example, we might reject any interview-only study
based on less than 10 hours of interviews.
12) Make no exceptions. Exceptions will inevitably apply more often to more
powerful academics, increasing nepotism.

None of these suggestions are particularly novel or inventive. Common-sense
improvements like these are only resisted because of the incorrect belief
that anything that simplifies review will reduce quality. A simpler, more
direct review process will encourage everyone to focus on key issues –
methodology and results rather than framing, positioning and tone –
increasing quality.

P.S. Long review cycles are not caused by poor reviewer incentives. This is
a red herring, designed to divert criticism of the extraordinarily
inefficient way we review papers, and the editors-in-chief who have the
authority to improve it but choose not to.

—
Dr. Paul Ralph
Lecturer in Computer Science, University of Auckland
http://paulralph.name