Publication, publication, publication

I thought in this post I thought I would write about academic publication, focussing on the journal article or "paper", I may try to introduce a tortured analogy at some point. This is all rather topical because some people in stem cell research have just complained loudly about the unfairness of it all. In fact there's a whole slew of comment on this around at the moment by, for example, Cameron Neylon, Russ Swan, Suzan Mazur, and Mark Henderson. My goal here is to explain to the lay reader scientific publishing, what on earth we're all so ventilated about and drop in a couple of comments for practioners.

As an university scientist I, my boss, my students, would do research. Every once in a while we would consider it appropriate to publish a paper on this work. This was important because through our careers those papers are a measure of our academic worth, when you apply for a job the appointment panel will go through the list of your papers to get an idea of how a good a scientist you are. As a personal rule of thumb I reckoned on an average one paper per person per year. This is a bit low (even for me, since I have my name on 29 papers over an 18 year active research career), it varies with academic discipline, and even within academic disciplines.

So what does it look like? Well, you can see one of mine here. There's a bunch of authors whose functions are opaque to the reader, a set of fairly standard sections which roughly comprise: Abstract, Introduction, Methods, Results, Discussion, Conclusions, References.

Your paper will be decorated with little markers pointing to papers in the reference section. The idea is that you indicate the support for a particular statement or idea via the references. The paper I linked to above is described like this:

Brujic, J., S. F. Edwards, D. V. Grinev, I. Hopkinson, D. Brujic, and H. A. Makse. “3D bulk measurements of the force distribution in a compressed emulsion system.” Faraday  Discussions 123, (2003), 207-220.  

That's to say: a list of authors, a title, the journal it appears in, the volume number, year of publication and the page range - it pins the paper down pretty thoroughly. I can go and find the paper (and make up my own mind as to whether the reference was appropriate). In a way this is what peer-review system is about, it isn't really about correctness in anything other than the broadest sense, it's about reputation and discoverability.

Just so you know, in the list of authors above: Jasna Brujic was the PhD student who did the experimental work, Sir Sam Edwards is a theoretician who works on granular materials, Dmitri Grinev worked with Sir Sam on the theory, I supervised Jasna, Djordje Brujic is Jasna's dad and wrote the image analysis code and Hernan Makse is a computer simulator of granular materials.

After you've written the paper, you send it off to an academic journal of your choice (there's a big field to choose from, and practioners know the relative prestige of each of these journals and there are published Impact Factors which attempt to quantify this). The journals send it off to roughly three other academics for "peer-review", on the basis of whose reports they will accept or reject the paper. If rejected you'll likely send it off to another, less prestigious, journal. At the same time you will curse the anonymous reviewer that lead to this ignominy, a bit like this, in fact.

I've peer-reviewed papers, my approach is as follows: read paper, check for obvious lunacy, check for obvious previous publication, check to see if you're referenced, write a few lines of recommendation to the editor, which in total  takes me a couple of hours or so. I make a more in-depth reading of a paper if I'm trying to replicate results or, as I have done once before, been writing a review in which case I repeat calculations and re-plot data. This level of effort doesn't seem worthwhile for an anonymous activity with no payment; reading referees reports on my papers then it would appear my approach is par for the course on peer-review.

In truth the real test of a scientific paper is what happens after it's published, there are three possibilities:

1. everybody ignores it, 
2. they refer to it in their papers to point out it was wrong, 
3. they refer to it in their papers to support their work.

Academic search engines will tell you the aggregate of possibilities 2 and 3, that's to say the citations: the number of times your paper occurs in the reference sections of other papers. This is the exciting bit, most of my papers have somewhere under 10 citations, there are a few with 30 or so citations and a couple with 60 or so. Actually this is a habit I carry over to blogging, since bit.ly links come with statistics on how many times they have been followed I can get exciting real-time feedback of how many people retweet a link to my blog posts on twitter, and how many people have at least looked at my post. As far as I can tell my blog posts are better read than my papers.

Academic publishing is pretty lucrative for commercial publishing organisations, this report cites profit margins of 30%, and there is a more general discussion of costs in the UK here. It's all a bit odd really academics, like me, write articles for free, we send them to journals (whose academic editorial boards are often unpaid) who then send them out to more academics to review (for free), we then buy back our material in the form of journal subscriptions, which can be very pricey (£1000 per annum per 12 issue journal is not uncommon). The latest wheeze is to replace journal subscriptions with an "open access" model, whereby the author pays the journal to publish a paper.

Really academic publishing is all about reputation, your reputation as a scientist depends on how many articles you get published in high reputation journals as a proxy for your own reputation and the absolute quality of the paper you have written. But do we really need specialist journals any more? You can see how easy it was for me to make the paper I referenced above visible. I could have made my paper visible on a blog, and interested people could post their comments (like peer-review), I could promote my paper through twitter. We could have a soup of articles re-sliced by keywords, spread across the web, or leave it to individuals to curate their own sets of papers. We could leave it to academic departments to host the papers of their staff, they're paying through the nose for library access, and these days as often as not they're hosting electronic reprints already on the personal web pages of their staff. The programming solution site, stackoverflow.com has an interesting reputation model, which seems to work well - couldn't this be adapted for academic use?

Are you missing a tortured analogy? How about this: the current publication model is like buying all the ingredients for a cake, making a cake, then taking the cake to a shop who then charge you to take the cake away again. We should break free of the hegemony of the cake shop!