Thursday, December 12, 2013

Thoughts on scientific publishing, from someone without tenure, or a Nobel prize.

There has been a lot of heated discussion about recent Nobel prize winner, Randy Schekman's article, "How journals like Nature, Cell and Science are damaging science".  In it he argues that these types of journals preferentially publish science that is sensational over science that is important.

Of course they do.
Of course Nature, Cell, and Science choose to publish articles that they think are both scientifically accurate and going to be read by a lot of people. NCS have a brand to promote, and an audience to cater to that expects exciting articles of broad interest. I like to read or skim through articles that aren't in my own field in Science. They have no reason not to publish the articles that will bring in the most readership, and every incentive to seek out the "hottest" research topics.

I have no problem with a journal choosing to publish only a certain kind of research (chemistry, physics, biology, evolution, and so on), or choosing to publish articles that they think will bring in the most readership. It only becomes a problem when the quality of the science is overlooked in favor of the sensationalism. Publishing inaccurate science should not be tolerated.

The real problem.
Schekman argues that part of the problem with "high impact" journals that choose to prioritize flashy results over real scientific advances is that hiring, tenure, and promotion committees view publication in these journals as a measure of the scientific merit of a paper. I don't see this as a problem with the flashy journals, but in the people who assume publication in such a journal should be equated with scientific excellence.

As a postdoc on the job market, I am acutely aware of this, and worried about it. You'll note from my publications that I don't have a single NCS paper. I'm pretty sure that not having those NCS papers has/will hurt my prospects this year. I hope that some schools will still recognize my ability to do good science, to conceptualize projects and bring them to completion, and to successfully mentor students, and will offer me a position despite the lack of NCS-publications. Perhaps foolishly, I do believe there are departments that value science more than the journal where it was published. And, with the development of more detailed, and manuscript-specific, metrics, including downloads, citations, shares, page views, and saves, I think it will be easier in the future to dispense with using journal IF as a measure of scientific merit.

Don't get me wrong, I'm not naive. I know I'm not going to be considered at some institutions because I don't have publications in the NCS journals. While "luxury" journals are valued at "luxury" institutions, it is also important to remember that "luxury" institutions (as pointed out over at Telliamed Revisited) can also be used inappropriately to infer quality of a person and their research. Case and point: I was told on an interview, "It would be really great for our department to hire someone from Berkeley, like you." Ugh. Nothing to make you feel frustrated with the system like thinking you may have been recruited solely because you came from University A instead of University B.

So where does that leave us?
Tenured, well-known, professors have been chiming in on the issue, but will that affect how I proceed with my scientific trajectory. I am confident that any position I advocate cannot be attacked on the basis of, "I've already made it." I can assure you, in this tumultuous academic job environment, I have not "made it". At this stage in my career these are the four things regarding scientific publishing that are important to me:

1. Scientific papers, results, and datasets should be publicly accessible.
2. Classifying research by field is useful.
3. There is value in peer review (pre and/or post).

4. Reviews should be public.

1. Scientific papers, results and datasets should be publicly accessible
For me, the most important aspect of publishing my work is that it should be publicly accessible and reproducible. Nearly all journals now offer an option to pay (sometimes a prohibitive amount) to allow papers to be freely available, and many are requiring data to be submitted with publication. Assuming I have all the funding in the world (ha!), where will I submit?

2. Classifying research by field is useful.

Well, I'll submit wherever I think the audience for my paper will be the best. I tend to find new publications through google scholar and pubMed searches, but on occasion, I'll scan through the recently published articles in journals with a focus that is near my discipline. So, as an early-career scientist, I will publish in journals that are in my field, or through venues where my work can be classified (tagged, labeled) in broad categories, so people who weren't already looking for my specific paper might find it.

3. There is value in peer review (pre and/or post)
I have found peer-reviewed comments to be immensely valuable for my accepted manuscripts. Sometimes this delayed publication for months, and I didn't always agree with the reviewer comments, but that doesn't mean the process itself was worthless. I don't think that the current system is the best, but I also don't think it is terrible. One of the challenges with the peer review process is the huge lag time between submitting a manuscript and it finally being published. Most journals don't have a good way around this. One option would be to go with a system like F1000 or PeerJ, where manuscripts can be archived and immediately available upon submission.

4. Reviews should be public.
For journals with pre-publication peer review, I wish that all papers were published with the pre-publication reviewer comments. I think it would be immensely valuable to see what other scientists thought of the work before it was published. For all work, I think it will be beneficial to have a single location where post-publication comments can be collected, reviewed, and published. This would likely need to be monitored in some way to prevent it from turning into a free-for-all by trolls. 

So, will I publish in Nature, Cell or Science? 
I don't know.

But, I can tell you that I know better than to judge the value of a paper, or a scientist, by the journals they're published in, and that is where we have to start. 


Adam Retchless said...

" I'll submit wherever I think the audience for my paper will be the best."

I think that Science and Nature are useful journals since they reach a different audience than any other journal. The sad thing is that they're publishing "exciting" research regardless of whether it is appropriate for their audience.

Most people who read Science or Nature cannot understand most of the original research reports published there. The audience is essentially intelligent laymen -- whether they are journalists, educators, policymakers, or scientists from other disciplines...but for some reason these magazines publish work that is targeted at experts.

Both the authors and the editors need to think more carefully about what is appropriate content for Science and Nature. This is particularly bad when they publish what is essentially speculation (such as the arsenic paper in Science, or the "non-random mutation" paper in Nature that I've talked with you about). Another common faux pas of these journals is that they publish research that relies on extensive data collection and analysis -- and then severely constrain the space available to describe the research.

I think that the publishers are starting to address these problems -- for instance I like how Science is now publishing one-page summaries for research that they are publishing online. But ultimately, I think that these journals should be used specifically for short papers with a broad (non-expert) audience.

mathbionerd said...

"I think that these journals should be used specifically for short papers with a broad (non-expert) audience."

Agreed. Should scientists be writing these? Or science writers? I do like having articles with a focus on science, with primary results, but as it is now, it is often difficult to judge the methods, but adding more would make them not accessible to a broader audience.

I really don't know.