For Full Details of the Model See Elsewhere

The purpose of this post is to consider how researchers using process-based models can best describe their methods, assuming that a given base model may be used in a series of papers.

A common situation in the current literature is that a full description of any particular model is given in the first paper where it is used, and then in subsequent papers we say something like “For full details of the model see Falster et al (2011).” The benefit of this approach is that it reduces repetition, and clearly identifies when an idea or set of ideas was first introduced. But there are at least two substantial downsides:

  1. Subsequent papers are not self-sufficient. By referring readers off to another paper, we leave them unable to properly evaluate the current work on its own.
  2. The details of the model may have changed slightly since the original publication, so it is often incorrect to refer readers to the original publication. And yet, spelling out all the subtle changes may be quite cumbersome.

What are the alternatives?

Option 1

Personally, I would like to see any paper using a process-based model to include a full description of that model and the code implementing it. I consider that the minimum amount of detail needed to understand and potentially reproduce the results in the current work. Most likely both would appear be included as supplementary materials. I should not have to go off and read another paper to understand the current one.

Following this approach, authors would maintain a living document of current code and model description and include a snapshot of that document with every paper they publish (presumably as supplementary material), as it was at the time of publication.

The main hurdle here is that there will be a lot of duplication in content among publications, most likely in the supplementary material but possibly even in the main text. And thus we meet the thorny monsters of academic publishing – copyright and licenses. This is not my area of expertise, so I’ve been reading up a little. And it seems my desire to include the same or slightly modified descriptions of a model across multiple publications might even be possible, but not under standard publishing agreements.

When we publish, it seems that most often we do not sign away copyright to the material per se, that stays us – the author. Rather we grant the journal (or publisher) an “Exclusive License” to publish the material. For the few journals where I checked (Journal of Ecology, Methods in Ecology and Evolution, and Science), the granting of an “Exclusive License” was both standard and explicitly prevents the type of re-use I am seeking. Also, the supplementary materials were covered under the same license agreement. (Update 2016.08.09 - no longer the case for Supp. Mat. in Methods in Ecology and Evolution, as described below).

If you can publish under a CC-BY license, however, the type of reuse I am seeking seems possible. According to one publisher Wiley, who publishes on behalf the British Ecological Society, the CC-BY license allows readers to “copy, distribute and transmit an article, adapt the article and make commercial and non-commercial use of the article, as long as the author is properly attributed”. Moreover, the license is offered under their OnlineOpen offer, where authors pay a fee to make their work open. Other open access journals like Plos and Nature communications also publish by default under CC-BY, while others like Science Advances offer CC-BY-NC as default and CC-BY as an optional extra.

So overall this seems very promising. Of course there are some restrictions when you publish under CC-BY, such as:

  • Any reuse or redistribution of a work must cite the original, while also making clear the license terms under which the work was published and indicating if changes were made.
  • You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.
  • I assume you cannot publish a re-used full model description in a journal requiring an standard “Exlusive license”, because you cannot offer an exclusive license.

Option 2

The second option would be to run a parallel publishing system, where model descriptions are kept archived away from journals and linked or referenced. The downsides of this approach are that

  1. the links are more like to rot over time
  2. the analysis paper is no longer stand alone
  3. Publishing a model description elsewhere in CC-BY format would not be possible if the same material was already published in a journal under an “Exclusive License”.

Feedback

I’m interested in your thoughts:

  • as a reader or reviewer, would you prefer to find a full description of the model in supplementary materials or do you prefer to go of and read the other paper?
  • as a journal editor or publisher, do you care if the supplementary material of our article exactly or closely resembles material that appears elsewhere?
  • does anyone know more about the legal status of SuppMat in different journals?
  • would publishing under CC-BY enable reuse of model descriptions across multiple publications?

Updates

  • Had some good discussion on twitter and email about this post.
  • In Feb 2016 we published an “Application Note” about the software package plant in the journal Methods in Ecology and Evolution. I was very pleased to discover that, ina dditon to offering an open source option, the latest version of the “Exclusive License” for publication in the journal explicitly excludes the Supplementary Materials. Instead, they ask you the author to grant the journal a “a non­exclusive license to host and disseminate this”. So we were able to publish a full account of our model as Supplementary Material, without removing our ability to reuse this material into the future.