Thoughts on publishing

From Murray Wiki
Revision as of 14:49, 7 July 2017 by Murray (talk | contribs)
Jump to navigationJump to search

This page collects some thoughts on publishing based on my experience in going from publishing in "engineering" disciplines (specifically electrical engineering and mechanical engineering, with an emphasis on control systems and robotics) and "scientific" disciplines (more specifically the biological sciences).

Publishing in Controls and Robotics

Most of my career has been doing work in the area of control systems and robotics. Here is the way that publishing tends to work:

  • Graduate students and postdocs write short papers on preliminary results as conference papers. Conference papers are submitted as 6-8 page papers and peer-reviewed to determine what is accepted for presentation. Most talks at conferences are given by graduate students and postdocs and all but a very small number of talks are based on review of submitted papers. At a big conference, there might be 2000 people with 800 papers presented (in 15 parallel sessions, 20 min/talk). Every paper will appear in the proceedings. Papers are submitted about 9 months before the conference, reviews come back after 3-4 months and final versions (addressing the reviewer comments, but not re-reviewed) are due a few months before the conference. Acceptance rates at big conferences are 50-60%, at small (single track) conferences about 20-30%.
  • Collections of conference papers are pulled together into journal papers. The results in journal papers have typically appeared in preliminary form in conference papers, but the journal paper has a more complete set of results. Journals are relatively slow to publish: from first submission to final publication as a paper (typically after 1 major review and a second minor revision) might take 1-2 years (but remember that the results are probably available in conference paper form well before they are submitted to the journal => people in the field know what everyone is working on years before the final results come out in a journal).
  • Most of the journals that we publish in are handled by technical societies (eg, the IEEE Control Systems Society) and the editors, associate editors and reviewers are all volunteers (active researchers in academia, industry or government). It is rare that a paper would not be sent out to review (unless it were clearly wrong, duplicative or trivial). A typical cover letter for a journal paper just says that we are submitting this paper to this journal and lists the authors.
  • Preprints of results are used at every stage. In my group, we typically post submissions to conference and journals on our web page at the time of submission. For conference papers these results are usual work that has occurred in the last 3-6 months => we let people know what we are doing almost at the same time as we are doing it. We typically don't use preprint servers; we just post on our web page. The results will eventually come out in the conference and then the journal, so we don't worry about DOIs for our preprints.

The IEEE has written up some nice author guidelines that summarize their approach. Here's an excerpt:

It is common in technical publishing for material to be presented at various stages of evolution. For example, early ideas may be published in a workshop; more developed work in conference proceedings; and the fully developed study may be published in a journal. However, IEEE guidelines require that authors fully cite their prior work. authors must be able to demonstrate signicant advances from prior publications.

Publishing in Synthetic Biology

When I first started publishing in journals that are more science oriented, I was surprised to find that papers were handled very differently. This is particularly true in the biological sciences, where conferences consist mainly of invited talks by PIs, there are no conference papers or preprints, and some groups are very secretive about their results until they are published in journal form (so often presentations at open conferences are results that have already appeared in a journal). The notion of being "scooped" is quite prevalent and there is a lot of emphasis about publishing in "glamour" journals (high impact factor). Since this doesn't fit the style of the work that we do in my group, I have adopted the following approach to publishing in this environment:

  • When we submit an abstract to a conference, we write a technical report with the results that we are going to present. The title, authors and abstract of the conference submission match the technical report, and the report contains all of the data you would find in a paper. The results are still preliminary and we may not have all of the explanation and controls. We post the report on a web page or (more commonly) on the bioRxiv preprint server at the time of submission (usually).
  • When we submit a paper to a journal, we acknowledge that the results have appeared in preliminary form in one or more preprints. All journals we publish in allow preprints and so this is not a problem. We do have to be careful about publishing results that have appeared in a peer-reviewed form (such as a traditional controls conference). Typically we would include those results only in the supplementary materials and that seems to be fine with everyone (and we are upfront about the fact that the results have appeared elsewhere).
  • The first author on a paper is typically the corresponding author on the paper. This is usually a graduate student or postdoc, and we did get some questions from journals about this in the beginning. It seems OK and generally hasn't been a problem. Senior authors still appear at the end, just without the dagger or whatever annotation is used for the corresponding author.
  • I still haven't figured out the "game" of talking to editors and writing cover letters. To me, our results should stand on their own and all journal submissions should be sent out to review. In many cases I have been fortunate to have co-authors who understand the system better than I do and they are able to provide advice to students and postdocs on how to "pitch" their results.
  • I try to tell students joining my group that I am not particularly interested in (nor good at) publishing in high impact factor journals so that they don't get surprised by this when they start to publish their work. I also tell them that if they want to attend a conference, they have to write up a technical report before we submit. I'll occasionally relent on this when the student commits to writing the paper before the conference actually occurs, but sometimes that doesn't happen and so I have been getting increasingly strict about requiring the full technical report before submission. (If you aren't able/willing to write down what you have done at an appropriate level of detail, then you don't get to submit for possible presentation -:).

One of my pet peeves is when students mention the term "impact factor" when discussing where we should send our work. This typically results in a 10 min monologue from me about why this is the wrong approach. Here's a typical e-mail response to a student on that topic:

My two cents:

  • I don’t care about impact factor and would suggest that we not bother with that as a metric. I’ll spare you the long version of the rant, but the considerations should be (a) fit of the journal to the primary contributions of the paper, (b) quality of the review process, (c) speed of the review process (submission → acceptance). Also, we should not submit to any journal that doesn’t allow free distribution of preprints.
  • To me, this is a paper whose contributions are related to *engineering* (how to effectively engineer biological circuits) and not about *science* (understanding the natural world). Hence, I would suggest submitting to journals that are dedicated to advances in engineering (along with science is OK, but not dominated by science).
  • Finally, although this is a practice that I very much dislike (and personally refuse to participate in), if you want this to appear in a “flashy" journal, then we probably do need to call up someone on the editorial board of whatever journal we submit to and get them to agree to shepherd the paper through the process. <Redacted> thinks this is important and he seems to be good at publishing in these types of journals, so worth considering (but don’t ask me to have any part in it -:).

-richard

p.s. (because I can’t resist) If you haven’t read what Bruce Alberts has to say about impact factor and why it is the wrong measure (“Impact Factor Distortions”), you should: http://www.sciencemag.org/content/340/6134/787

Preprint servers

I'm a big fan of bioRxiv. I use its alerts systems to find out about papers in synthetic biology and this is the fastest way to let me know about results that I might care about. I recently got asked some question about preprint servers and I have sent these to a few people. Here are the questions and my responses:

Q: Do you think posting of papers on preprint servers (such as arXiv and bioRxiv) makes a valuable contribution to science and scientific publishing? Anything youąd like to share about the underlying basis of your views would be helpful.

  • A: Yes. Early dissemination of results helps the community know what others are doing and build on those results and/or avoid trying to solve the same problems that others are already working. In many engineering disciplines, it is common to post preprints, then conference papers, then journal papers in online forums so that people can see the work evolve over time. I support this approach.

Q: Have you ever posted a paper on a preprint server? Or are you thinking of doing so? In either case, what would be your main motivations and goals?

  • A: Yes. I post *all* of my submissions to conferences and journals online, either on my own web page or (for relevant articles) on bioRxiv.

Q: If you have posted a paper, did you find doing so useful or helpful overall? If so, in what way?

  • A: Others saw our work when we posted it online and we were able to refer to that work in presentations, posters, etc for people who wanted more information.

Q: Have you ever commented on a paper that another group posted? If so, was the discussion productive?

  • A: I generally don’t bother reading or writing comments on papers. To me, this is not the main purpose of a preprint server.

Q: Have you found preprints posted by others useful for informing your research?

  • A: Absolutely. I routinely forward articles posted in bioRxiv to my group and it is my primary way of keeping track of new results. If someone wants me to know about their work, their best bet is to post on bioRxiv/synbio.

Q: Do you have any concerns about potential widespread adoption of preprint servers in biomedicine that you would like us to take into account?

  • A: Nope. I’m a fan (and have been publishing this way for 30 years).

Some articles about impact factor (and why you shouldn't use it)

Other useful information

Retrieved from "https://murray.cds.caltech.edu/index.php?title=Thoughts_on_publishing&oldid=21452"