The cost of production

Last week an interesting discussion on the cost of copy editing popped up over on Stephen Curry’s blog. In addition the comment thread at the recent Science editorial seems to make this post somewhat timely. I used to manage the copy editing of a good portion of physical science related content from Springer from 2002 – 2005. I’m also currently in the process of setting up a new online-only journal.

Back in the early 2000s we were getting pages copy edited for about $20 per page. Right now we are looking at costs of between $10 and $35 dollars per page depending on the level of copy editing required, so for a 10 page article the cost could come in at about $350, or about 10% of the cost of an OA publication in an outlet like PLoS.

High quality copy editing means significant work on grammar and clarity. The kind of functional checking for things like consistency in figure labeling, and so forth, is cheaper, and many typesetters have tools that can augment the checking for this kind of consistency: s1,$/Figure/Fig\./. If the burden of these costs were pushed back to authors, then authors who know that their English is sub-par, would have an incentive to get their language checked prior to submission. This certainly happens in some labs, and there are many services available for authors to pay to get this work done.

As we know most articles are not read much, and are cited less, even in journals such as Nature. If we could chose to on copy edit the items of the literature where it is going to provide significant value, I believe there would be an economy of effort. It is somewhat of a catch-22 in that one would want to apply the effort to articles that will attract attention, without knowing ahead of time which articles will attract attention. Wikipedia represents an interesting model for aligning effort and attention. There is a proposal to create a wikipedia journal, and in such a scenario one would be able to do sub editing post-publication.

I believe that the cost of copy editing should be placed on the authors, and that it does not represent an overly significant value add that publishers bring to the dissemination process. I feel that in terms of issues that we have with the scientific literature, lack of access to underlying data is a much bigger problem, than low-quality articles being difficult to read.

In terms of the cost of production of scientific content, I feel the discussion around copy editing is a red herring. There are a couple of new initiatives that are really interesting. PeerJ is trying a model of charging $100 per year per author, scholastica are charging $10 per submission. This question of where we will get to if charges normalise towards the $1500 dollar mark seems to me to be missing the point. Creating content on the web, especially where authors are not paid for their work, could be substantially reduced to the level of a few dollars per article. This would be disruptive for the existing publishing industry. As Jason Hoyt points out in a [recent presentation][ffd], this is also a model that the incumbents would be almost certainly unable to follow. [ffd]: http://www.slideshare.net/jasonhoyt/a-framework-for-disruption

What about maintaining high quality? Well, there are a number of confounding factors with the current system. One of these factors is that the cost of production is usually separated from profits for STM content. One of the reasons for this is that the market is non-transparent. Contracts with libraries usually include a non-disclosure clause, so no one really knows how much other customers are paying for the same product. Vendors providing typesetting or hosting services have the same non-disclosure clauses for publishers, so publishers don’t know how much their competitors are paying for production costs.

You can get rid of the first of these by having article processing fees, rather than a subscription model. You can get rid of the second of these if you invest in building your own publishing infrastructure, like scholastica have done, and PeerJ are doing.

There is still the issue of reputation, and what we call quality of production. To get well structured research content one currently ideally wants the output to be in NLM XML, and that does require some processing of the input from the authors. I believe that structured editing tools can be built that will provide a compelling experience for authors, and that will create content that can be ready to be online immediately. We have been waiting for such tools for a long time, and there is a history of attempts to create such tools, but I remain optimistic. One of course needs appropriate incentives to authors to take up such tools, faster publication times, and cheaper publication costs may provide sufficient incentives.

In this future, it may be possible to completely separate the cost of production from the activities that are associated with signification of high scientific quality, a grand un-bundling of the roles of the current STM publisher. This kind of scenario is already playing out in the print news media. I think it is an attractive future that offers benefits to research, and it is one that I look forward to.