Recently, Todd Carpenter posted a penetrating Kitchen article discussing the fact that PLoS, SPARC, and the Open Access Scholarly Publishers Association (OASPA) have collaborated to develop a visual guide to the spectrum of OA options entitled “How Open Is It?” The main component of the document is a table showing various types of openness a publisher might provide. Each of these types is then described on a scale of more or less open versus closed access.
Todd’s specific focus was on reuse policies, but he notes that the overall goal of this document is to reduce confusion by clarifying the meaning of OA, because OA comes in many forms. Since confusion due to complexity is my field, I decided to make a few observations, which mostly involve quantifying the complexity, especially in the context of designing a journal publishing system. Understanding why an issue is complex is often helpful.
To begin with, the table includes six columns, each of which is supposed to be a type of potential openness. These range from the familiar use of embargoes and gold OA all the way to the rather esoteric issue of degrees of machine readability. For each column there is one case deemed closed access, plus either three or four options which are said to display progressive degrees of openness. I call this sort of thing an essential variable analysis (EVA) because every publisher has to choose a case from each of these variables when designing their system for a given journal, or a set of journals. In fact, EVA is typical for any design problem.
So we see that there are six variables (the columns) and either four or five cases for each variable. Choosing one case from each column defines a possible design for a journal or system of journals. Assuming the cases are all independent of one another, we find there are a whopping 8,000 different possible combinations, each combination a possible publishing system design. Mathematically, this is basically a six dimensional matrix. In reality, some choices constrain others, but if anything these constraints make the design problem more complex, not less.
Moreover, there are probably a number of significant dimensions that are not included. Some of the cases are probably dimensions in their own right, or at least under-represented. For example, only two embargo periods are included, namely six and 12 months. In contrast, existing journals and journal systems have embargo periods ranging from two months to over 30, all of which are viable degrees of openness. This complex flexibility is not shown.
Likewise, hybrid journals are listed as a single case, even though the degree of openness varies widely, depending on the percentage of open articles. Also, hybrids are ranked as less open than a 12-month embargo period, but it is arguable that having many articles available immediately is more open than having all available only after waiting a year. More deeply, combining gold and green OA in a single column dimension is probably wrong. Hybrids may be the road to gold OA, while embargoes are not.
Perhaps many of the columns and cases are like this — multi-dimensional or poorly represented. But I leave that judgment to the subject matter experts. The point is that OA design is far more complex than this table suggests.