Geoffrey Bilder of CrossRef has a classic illustration showing the outline of a piece of paper with black lines cutting horizontally across it in a recognizable fashion. You know immediately — from the thick lines centered nicely toward the top to the two columns of thinner lines with a section in the northwest corner that has a clearer structure — that you’re looking at a scholarly article, with title, abstract, and a classic two-column layout.
Scholarly articles are further structured, with methods, results, and discussion sections providing a template for presenting scientific and scholarly findings.
Printed scholarly materials evolved these structures as a way to make reported scientific information more and more comparable as time passed. Yet, in the more engineered world of online journals, finding common interfaces for researchers remains a challenge, which in some ways may explain the constant craving for aggregated and common interfaces, whether it’s Google or Google Scholar for search, Ovid or PubMed Central or Sci-Hub for papers, or PDFs and their inherently standardized layouts for reading.
Editors and publishers often want to design pretty sites, but not especially usable sites. Some aesthetic choices can end up cast under the rubric of “usability” while diverging from what users expect. Many journals are design and technology islands, quirky and idiosyncratic, sometimes beloved for it. But dealing with divergent layouts, systems, and presentations creates what experts call “cognitive load,” a burden on thought that only impedes information flow and retention. Users dealing with highly complex information are likely less tolerant of cognitive burden around the information they want, which may partially explain the popularity of the PDF and the persistent appeal of simpler sites and interfaces.
Recently, during the course of business, I asked DeltaThink to review top journals (by impact factor, for convenience and a modicum of significance — take with an appropriate grain of salt) in 18 different scientific fields (Aerospace Engineering, Applied Mathematics, Astrophysics and Astronomy, Biochemistry, Biotechnology, Cell Biology, Chemistry, Ecology, Economics and Econometrics, Electrical and Electric Engineering, Endocrinology, Genetics, Geology, Immunology, Internal Medicine, Microbiology, Oncology, and Physics), and identify which commercial publishing platforms they’re on (e.g., HighWire, Atypon, ScienceDirect, self-hosted, etc.). Here are the high-level results for these 90 journals:
- In two fields, each of the top five journals is on a different platform.
- In seven fields, the top five journals are spread across four different platforms.
- In seven fields, the top five journals are hosted three different platforms.
- In two fields, the top five journals are on two different platforms.
- In no field are all top five journals on the same platform.
Most of the top journals in one field (electrical engineering) come from the same organization (IEEE), making this one of the fields with only two platforms supporting the top five journals. The other field with two platforms supporting the top five journals is cell biology, which is dominated by Cell Press and Nature.
There are many consequences for researchers navigating these divergent systems, including the chores of having to parse different designs, different logins, different technical demands (e.g., this site’s passwords require swear characters [#$!*&@] while this one requires 12 characters), and different features and feature deployments.
There are many consequences for researchers navigating these divergent systems.
New third-party systems like Altmetric, eLife Lens, ReadCube, Remarq, and others can provide a layer of common functionality or a standard interface across a set of journals. But even these are deployed unpredictably and without uniformity from the user perspective — I may see the Altmetric flower small, large, in a tab, or elsewhere, and so forth. I may use Lens on one journal, then a Lens-like design on another, and something completely divergent on the third.
All in all, it’s a confusing landscape for users.
Researchers sometimes try to eliminate these divergent interfaces themselves, with private PDF archives, reference manager software, and the like. Services like Mendeley once provided a virtual PDF locker for users, and now ResearchGate and Academia.edu are striving to provide a similar solution. The lure of UX consolidation remains strong enough to be interesting.
All this diversity in approach may be driving users toward consolidated experiences. While Sci-Hub is piracy on steroids, a single search and retrieval engine meets some core user needs. However, users are strange creatures, and aggregators of all sorts usually come up short of market domination, even if their services are perceived as free. At many institutions, librarians find themselves purchasing EBSCO and ProQuest and Ovid and the source journal because users develop preferences they won’t surrender, and the traffic still justifies the expense.
Even consolidation is fragmented.
Platforms are only the start of design, feature sets, system integrations, and various decisions publishers and technologists make. Even on a platform, journal instances can diverge. There are many different publishers involved in each field, as well (chart), each with their own strategic imperatives, design assertions, and system footprints. Perhaps we are still in the nascent age of digital publishing, and designs have not yet settled around a common set of interface standards and user expectations. Perhaps online is simply more dynamic and hard to tame, as the hardware, interfaces, and broader digital ecosystem all keep pushing publishers and designers to redefine themselves.
A higher degree of uniformity could be achieved in a number of ways. We have standardization all around us, from advertising standards from the IAB to technology standards. Visual standards are slowly evolving, but we could accelerate this by harmonizing the efforts. Interfaces matter, and a trade group devoted to defining the optimal layout principles and even a set of templates could help our users.
Perhaps we can forge a better path following the spirit of Geoff Bilder’s illustration, mentioned at the outset. Perhaps our sites could be as uniform and predictable as the papers we publish, with similar layouts, functionality, and standards — similar enough that scholarly sites become distinctive in and of themselves as scholarly in design and feature sets. We have long competed not on fancy layouts but on quality, relevance, and authority. Given where the real competitive advantages may exist, and the potential to improve usability while also reducing the risk of alienating users, more predictable sites with harmonizing features may only help.
Or, put another way, what would be the downside?