Sunday, 13 January 2019

Plan S implementation: in danger of mission creep?

The principles behind plan S have already sparked lots of debate, including an open letter denouncing the plan, based on objections that I found not very convincing. Now that the plan’s promoters have published their draft implementation guidance (and are inviting comments on it), the discussion can become more specific. Given the boldness of the principles, their implementation cannot be painless, and is bound to raise criticisms if not resistance. It is therefore both crucial and difficult to get the implementation right, and to inflict the minimum amount of pain that is necessary for achieving the goals.

This means that the implementation should be tightly focussed on open access and cost reduction, to the exclusion of other possible goals that a reform of the publishing system might have. I will discuss whether the draft implementation guidance is indeed focussed enough. But first, let me be more specific about what a minimal implementation might look like.

Sketching a minimal implementation


As I understand it, Plan S is about eliminating artificial restrictions to the distribution and reuse of scientific works, and the artificially high costs that come with such restrictions. To do this, it is necessary and sufficient that
  • articles are distributed under an appropriately permissive license,
  • paywalls are eliminated.
Both items are necessary: eliminating paywalls is not enough for allowing text and data mining, and mandating a permissive license for a number of articles is not enough for reducing costs, so long some indispensible articles remain accessible through subscriptions only. Together, these two items are sufficient for forcing journals to earn their money by doing actual work on articles, rather than just acting as gatekeepers.

Therefore, basically, the task is to eliminate perverse practices, not to mandate virtuous ones. Eliminating paywalls is the hard part: funders can mandate researchers to use permissive licenses, but they cannot directly mandate journals to tear down their paywalls. Moreover, there is the difficulty of formulating rules that cannot be evaded by clever publishers. For instance, the implementation guidance includes a clause forbidding mirror journals (paragraph 9.1), closing a potential loophole in the ban on hybrid journals. Still, it is surprising that the implementation guidance mandates so many detailed requirements. Let me examine whether these requirements are really justified, or whether their authors ceded to the temptation of aiming for more than their original goals.

Temptation 1: Focussing on quality


The quality of articles and of research itself has little to do with open access, and yet when a reform like Plan S is proposed, there is the temptation to discuss its possible effects on quality. The basic objection that reducing publishing costs would sacrifice quality is of course nonsense: the large costs of journals are ultimately due to the dysfunctional economics of a system where authors choose journals without worrying at all about costs.

Nevertheless, the implementation guidance includes a strong focus on quality, starting with its Point 4: “Supporting Quality Open Access Journals and Platforms”. And in Point 9.1, among the basic mandatory criteria for journals and platforms, we have a “solid system for review according to the standards within the relevant discipline”.

There is no reason for bundling open access with quality control: a quality-neutral flip to open access would already be great progress. Moreover, translating the focus on quality into concrete guidelines inevitably leads to mandating a specific quality-control mechanism, namely peer review. But who knows which other forms of quality control could emerge? For example, in an open system, text and data mining could let algorithms play an important role in quality control. And a crowdsourced quality control mechanism (cf StackExchange or Wikipedia) could conceivably be applied to scientific articles.

Temptation 2: Achieving funders’ dreams


Since Plan S emanates from a coalition of funders, there is the natural temptation to include requirements that seem convenient or desirable to funders, without having anything to do with open access. The obvious example is (Point 9.2) that “Metadata must include complete and reliable information on funding provided by cOAlition S funders”. It would be far-fetched to justify this clause from the need to track compliance with the open access mandate.

Moreover, Point 9.2 also mandates that publishers give details on their costs. While we certainly need transparent prices, transparent costs sound like a bad idea. First, costs are hard to define and to compute with any precision. Second, in an open publishing system, publishers compete on price (among other factors), and cost control should come from that competition. Finally, if Elsevier was charging only 500 euros per article rather than 5000, why should we care if they made 40% profit?

It is understandable that funders want to promote their good works, to streamline their operations, and to see what they get for their money. However, Plan S should not be used to these ends.

Temptation 3: Codifying best practices


The mandatory quality criteria (Point 9.2) read more like an attempt to codify and standardize today’s best practices, than like a minimal implementation of principles. Almost none of the criteria seems indispensible for achieving open access and eliminating paywalls. Actually, assuming that articles are published under a CC-BY license, requirements like having texts in machine-readable formats or having them archived for long-term preservation seem moot, as the license allows third parties to convert, archive and exploit the texts.

Moreover, an underlying assumption of some of these requirements is the existence of a reference version of the texts, typically the published version. But other publishing models are emerging: from living reviews that can always be modified and may not have final versions, to arXiv preprints that are never sent to any journal, to Wikipedia articles that may not have clearly defined authors. Rigid guidelines risk thwarting the evolution of the scientific article, and today’s best practices can become tomorrow’s bad habits.

Mission creep and its consequences


Going beyond what is strictly necessary to achieve its goals, Plan S opens itself to legitimate criticism that it asks too much from journals and platforms, which inevitably disfavours small or emergent players. The Coalition of Open Access Repositories has formulated criticism of this kind, and Angela Cochran too has raised many valid points.

Overreaching at the implementation stage could jeopardize the very success of Plan S. And even if the plan succeeded, it would replace the current system with a better but imperfect system that would be too rigid for having good prospects of further improvement. Fortunately, the implementation guidance is open to feedback, and there is some hope that Plan S can still be refocussed on its core mission.

No comments:

Post a Comment