The US Office of Science and Technology Policy (OSTP), part of the Executive Office of the President, has issued a sweeping policy memo entitled, “Increasing Access to the Results of Federally Funded Scientific Research.” It directs all federal research agencies to develop and implement open access (OA) plans over the next 2-3 years.
While the basic concepts presented in the memo are relatively simple, there are significant administrative complexities. Having worked on federal administrative procedure systems for many years, I will look at the OSTP OA mandate from that perspective, especially how it might affect publishers and authors. There are several big challenges which I call three monsters and a gorilla.
The three monsters all have to do with flexibility, and there is a lot of flexibility in the OSTP memo, possibly too much. What one might call the default administrative structure of the mandated OA system is pretty simple, but it is just an optional starting point for system design. (An OA system is an ongoing system of behavior including the required actions by publishers and authors, not a computer system — although it certainly includes computer systems.)
In fact, the default OA system is simply the government-wide application of two existing agency programs. On the document side, it looks like the present OA system of the National Institutes of Health (NIH) with journal articles being made public after a 12-month embargo period. Research reports are not mentioned. On the data side, it looks like the present OA system of the National Science Foundation (NSF), which requires that every grant proposal include a data management plan. When the flexibility kicks in, however, the picture gets much more complicated, especially on the journal side because the agencies are free to choose other approaches — and there are many.
The first monster is the multiplicity of federal research organizations. Research is not a standalone federal activity. Rather, it is scattered among and within a host of mission-oriented agencies. Each of the primary agencies is specifically directed to develop an OA system (although cooperation is encouraged). This is already a relatively large number of separate OA systems that publishers might have to deal with. But there are potentially many more systems because there are numerous independent internal organizations that might develop their own systems.
For example, consider the press release endorsing the OSTP memo put out by the NSF. It includes endorsements from other research organization heads including the Office of Science of the Department of Energy. The Office of Science is about as big as NSF, but it is only one of a number of billion-dollar DOE research groups, each of which might develop its own OA system. So the question arises, How many OA systems will DOE have? The same is true for other primary agencies, as most have several big independent research organizations, each of which could develop its own OA system.
The monster question then is how many OA systems will the government have – one, 10, 20, 30, or more? Many publishers are likely to publish articles which in the aggregate flow from many different federal research organizations. For example, NSF and DOE jointly sponsor a lot of projects. Even single articles may have multiple agency sponsors, especially if they have multiple authors. The potential for complexity is large as publishers and authors may have to deal with many different federal OA systems.
The second monster is the multiplicity of possible OA system designs. The existing NIH PubMed Central (PMC) system is a complicated piece of administrative machinery with many features. Each of these features is in principle a variable such that other research agencies might choose to do things differently in ways that affect what publishers and authors have to do. This could lead to a complex array of requirements and conditions that varies from system to system, and we do not know how many OA systems there are going to be.
For example, PMC does not accept all federally funded articles; rather, it uses a complex screening process to certify journals based on certain submissions by publishers. Other agency systems might use different certification standards, such that a given journal might be included in some systems but not in others. Or they may use different application processes which would be burdensome for publishers. How authors interact with all these systems could be even more complex.
On the flexibility side, one of the most exciting prospects is OA systems that work as portals, not as repositories like PMC. Such systems would merely catalog agency-funded articles and direct users to the participating publisher’s website for access. There has been a lot of exploratory development of such systems, and the OSTP memorandum has language that speaks to this possibility. But such systems will likely not have the same publisher and author procedures as PMC; in fact, they may be quite different. Some research organizations may develop or adopt this portal model while others go the repository route — plus, there are various hybrid options.
The point is that not only might there be many different federal OA systems that publishers and authors (and users) have to deal with, but they may work in very different ways. The number of OA system design variables is very large, so there is a lot of flexibility.
The third monster is the multiplicity of disciplines. The OSTP memorandum explicitly provides for having different embargo periods for different disciplines. This is because there is evidence that the journals in these disciplines are financially sensitive to the embargo period in different ways.
Having discipline-specific embargo periods is an intriguing prospect, but it will not be easy to develop. To begin with, there is again the challenge of multiple agencies. Because the government is organized by mission, not by scientific discipline, most disciplines are funded by a number of different agencies. Almost everyone does some economics, computer science, biology, chemistry, and physics, for example. This raises the prospect that different agencies might have different embargo periods for the same discipline.
Then there is the challenge of formally identifying the discipline of every journal article. The British do it based on the sponsor because their sponsoring organizations are discipline-specific. US federal sponsors are not discipline-specific except perhaps at the program level. Going to that level would be a very complex process indeed. Then, too, there are various discipline categorization schemes available in the bibliographic and science studies world that might be useful, but one would have to be selected, made official, and properly applied. And, of course, different agencies might choose different schemes.
But the real challenge will be deciding on the proper embargo periods for each discipline once these are defined. There are presently no analytical or administrative procedures for doing this that I know of. There is a joke that every government official has three boxes on their desk — In, Out, and Too Hard. Deciding on discipline-specific embargo periods may be too hard. If not, then the agencies and the discipline-specific societies need to really jump on this challenge because there is a lot of work to do.
The gorilla is the budget crunch. The OSTP memorandum is emphatic that no new money should be requested to develop, field, and operate these new OA systems. The reality is even worse — namely, that there may be significant cuts to existing funding levels. This suggests that agencies may be looking for least-cost approaches. These constraints play strongly into the OA system design process; for example, they work against the NIH PMC model which is expensive and labor intensive, and perhaps work in favor of the portal model.
In my taxonomy of confusions — which was designed for cases like this — the OSTP OA policy memorandum creates a “matrix problem” of procedures. Multiple agencies considering many design features for multiple disciplines create a multi-dimensional matrix of possibilities with a huge number of cells. It is clearly a prescription for confusion. Hopefully, once the various agency draft OA system plans have been submitted, OSTP will take steps to consolidate and streamline them, or at least make them more uniform. It would also help if an inter-agency coordinating group were set up to try to do this from the beginning. Otherwise, Balkanization is a potentially serious problem for US federal OA.