Is the journal article fit for purpose, or stuck in the past?

Journal articles remain, in many disciplines (though not all), the most important way in which new scholarly information is acquired.1–4 What is surprising is that they have changed so little over the years. The way that people communicate has been transformed by a whole succession of technological developments enabled by the Web, the mobile phone, and the interface between them. Newspaper and magazine articles have got shorter as readers become busier (or have a shorter attention span). Encyclopedias have morphed into Wikipedia, with its very different notion of ‘authority’.5 Textbooks have evolved to serve a much more modular style of teaching and learning – perhaps the extreme example is a customizable textbook such as McGraw-Hill’s Primis.6 In Japan, we even see highly condensed ‘text novels’ which are read on people’s mobile phones.7 But journal articles remain, in their underlying form, very much the same as they have been for decades; if anything, they are getting longer even as readers’ time is getting shorter.2 This is not to say that articles in e-journals are not enhanced, relative to their print predecessors. As long ago as the late 1990s, the SuperJournal project correctly identified that one of the features that would be of most value was linking8 – and many publishers now provide DOIs for their articles, aiding long-term retrieval; some (though not so many) also include DOIs in their reference lists. Some even provide links to supplementary data or other material, or to external databases or other information sources. E-journal articles are also, by their nature, easier to find than the print equivalents – particularly if the publisher takes the trouble to ensure they are fully indexed by all the major search engines,9 that they have all the right metadata,10 and that the content is clearly signalled (rather than cryptically alluded to!) in titles and abstracts. The fact that they are easier to find would, one might have thought, be enormously helpful as readers are overwhelmed with the ever-increasing amount of scholarly content (both paid-for and free) to which they have access;11,12 yet, curiously, it appears from their citations that they actually use fewer articles, and from fewer journals, than was the case in the print world13 – perhaps serendipity, coming across an article by chance when looking for something else, was more valuable than we realized! Some publishers have added even more enhancements to their online articles, such as open review and ongoing discussion (although these do not seem to be widely used – the Nature open-peer-review experiment was abandoned).14 If the article is presented in HTML or XML, as opposed to the more common PDF, navigation around the article can be aided by tagging the different sections; the data underlying graphs and charts can be made accessible and actionable; terminology can be linked to specialist ontologies, as for example in the Royal Society of Chemistry’s Project Prospect.15 All of this should make articles more useful to readers, though at the cost of considerable extra work by author, publisher, or both. But the article itself is much as it has always been. The structure of research articles has become virtually standardized over time (introduction–methods–results–discussion), and as mentioned earlier, they are actually getting longer: from 7.4 pages in 1975 to 12.4 in 2001.1 Yet readers are spending less time reading each article,2 and their reading is not necessarily a matter of starting at the beginning and reading through to the end – they look at the abstract, the conclusions, the illustrations, then perhaps the methods, and lastly they may or may not delve deeper into the results and discussion.16 So why aren’t we presenting article information in the way in which people actually seem to consume it? What is more, journal articles are not, in reality, set in stone the moment they are published. This may be heresy to some, but journal Editorial 3