Have we reached peak publication?

May 28 2014 Published by under ideas, life in the lab, postdoc, science

Peak oil is when "the maximum extraction of oil is reached, after which the rate of production is expected to enter terminal decline". This, of course, won't happen to papers. There will probably always be papers to be written and things to be discovered.

But back in the days, you could get into Science with a smart experiment that you had already published and add a model and some equations*. Nowadays, getting into C/N/S requires an almost endless amount of experiments and an even more endless amount of control experiments. You can argue that this is because reviewers and editors require people to do more and more, but I think it is also because fields move forward and questions become more complicated and more narrow. In my field** it's clearly more complicated than just measuring glutamate levels in the accumbens and saying that that drives any type of motor output. It's necessary to figure out exactly which cells and which inputs onto these cells are involved in a behavioral output (or even better: which synapses from which cell!). And not just if they happen to be involved, but rather if these cells and the inputs they receive are necessary for this particular behavior. And then obviously a boatload of control experiments that don't even end up in the actual paper***.

This makes me wonder if someday soon, it will no longer be possible within the duration of one PhD project or one post-doc to gather enough data for one C/N/S paper? Is this a bad thing and do we in someway need to reset the criteria?


* not to say that this wasn't a very important Science paper, but just to illustrate the amount of work necessary to get into Science.

** I realize that illustrating which field I am in may make it even easier for readers and followers to deduce who I am. This is a choice that I've made because it just takes me too much time to come up with bunny hopping analogies to what I want to illustrate. And also because if you've followed me here and on twitter it's really probably not that hard to figure out who I am anyway. Which I've decided is okay. Hi mom!

*** Did you know that nowadays we call Supplementary Data "Extended Data"?


11 responses so far

  • Jim Woodgett says:

    Increases in efficiencies in data acquisition rates keep approximate pace with increased expectations (and perhaps partly drive these expectations) but data is not information and the rate-limiting step is usually the parsing of said data and it's wrangling into something that makes sense (often succumbing to cherry-picking). The number of papers has also increased over time as the options for publishing have increased and the barriers to publish have decreased. But the expansion has largely derived through "new" journals rather than fattening of those that existed in the 1980's. This broadening of the literature (along with a distinct lowering of overall quality per paper) was itself enabled through new search and discovery tools which are, in of themselves, good. However, there are clear imbalances between how science is disseminated and the actual publishing industry. There are clearly corrections and disruptions in play, but until we are able to uncouple our desire to measure scientific output in the form of manuscript quanta from what is actually of value, we will be chained to the traditional models and perpetuate the misleading connection between where something is published and its actual value.

    On the point of the output of a PhD, we've already got there in terms of (at least 1 paper not equalling 1 PhD). Perhaps this realization will actually help in moving us away from counting papers as a bizarre form of volumetric currency (salt) and instead look at what material the papers are composed of (precious metals or tin).

  • In the physiological fields of biomedical research, high-profile journal publications generally require a complex assembly of data supporting a broad hypothesis using a multitude of technical approaches: genetics, physiological measurements, manipulation of relevant physiological parameters in the intact organism, genomic or other high-throughput measurements, etc. It can be tough for any one scientist to master and perform all of these techniques, and thus collaborative efforts involving multiple trainees in the same (or in different) labs is frequently required.

    DoucheMonkey should be here any minute to explain how this is ruining science.

    • babyattachmode says:

      Yes, I'm well aware of the need to collaborate and that this can result in multiple first and/or last authors. But I keep wondering if the amount of work required to get that desired high-impact paper that seems necessary to continue in science is increasing to a point where it is nearly impossible to get that, even if it's a shared first author paper.

    • drugmonkey says:

      that is not the part that is ruining science, pp.

  • Dave says:

    While many CNS papers clearly require a lot of work/figures, very frequently they do not and, instead, are describing something extremely novel and/or something of extraordinarily broad importance. If you are in the right place at the right time, it can happen during a PhD or post-doc.

    And this 'problem' is not restricted to CNS. Far from it, in fact, and I think CNS gets dealt with unfairly on this topic. High..ish IF society journals are some of the worst offenders, in my experience. I just got reviews back in a very nice field-specific journal (IF 8.5) and we are going to have to spend 3 - 4 months doing the extra work needed to satisfy just one reviewer. This paper already has 9 multi-panel main figures and 10 supplementary figures + suppl methods etc, and although the requested experiments are worthwhile, with all the data combined, it's probably 'too much' for this journal.

  • Agree with post and comments... IMO re: "someday soon, it will no longer be possible within the duration of one PhD project or one post-doc .." that day has come and gone. The odds make it virtually impossible to publish in a high-profile journal without something approaching 10-person years of work. As far as the right place and right time goes, "try and be lucky" is good advice for any field.

    I joke with my lab folk that the first thing CNS journals do when evaluating a manuscript is to weigh it (you know what I mean digerati, print everything out including suppl). Not really joking though.

    • becca says:

      A couple of recent articles for our journal clubs used some from glam mags which had 60+ and 80+ panels (counting the supplemental stuff). In one case, it was showing via EVERY conceivable technique that two things interacted (co-IPs, colocalization, fluorescently tagged, in situ and in cells, ect.). In another case, the experiments spanned C. Elegans, mouse and human samples. In neither case was the science moving forward in a very innovative fashion, it was just the demonstration of the relationships were very robust (which, admittedly, is nice to see. Just... in a very practical sense, overkill).

      • drugmonkey says:

        Didn't you see becca? PP said it was *required*. So clearly it must be. Couldn't be anything else going on here.

  • drugmonkey says:

    Stamp collecting that brags on how many different countries one has visited is still mere stamp collecting. True innovation does not depend on technological masturbation.

  • Cynric says:

    Stepping aside from the glamour issue, there's no peak in sight for the number of new articles published annually. It's been increasing at a remarkable steady rate of 3% per year for the last two or three centuries:


Leave a Reply