The editor of the Journal of Neuroscience, John Maunsell, has announced that the journal will no longer permit authors to add "supplementary" material to their papers
Sure, the idea is great -- the data and methods can hide somewhere that the interested can find them, while the marginally-interested can focus on the sparkling whitewash of the introduction and conclusions. Er...
Science published a monograph's worth of pages on Ardipithecus. Many of the papers were online-only. Some of the papers were accompanied by up to 70 pages of "supplement." What happens? In some cases, these failed to include standard measurements and essential details to replicate their comparisons.
The Neandertal genome paper was awesome, I'm glad that the journal ran it in the form it did. But the 200+ pages of supplement should have been divided into 10 separate papers! They deserved separate recognition and review. The science will be better when the "supplementary information" is treated seriously.
Maunsell's statement discusses many of the problems of supplements, as applied particularly to neuroscience. This part of Maunsell's statement is worth reading carefully:
We have carefully considered alternatives to removing supplemental material from the peer review process, but have found none acceptable. The idea of demanding that reviewers thoroughly examine supplemental material is impractical. Even if all reviewers could be coerced to review supplemental material with care, it is not clear that this should be encouraged when it would inescapably reduce scrutiny of the main article. Attempting to limit the amount of supplemental material authors can submit is not a solution. Any reasonable fixed limit on what authors can present (e.g., as many figures as are contained in the manuscript) would permit enough material that it would not address the issues of inadequate peer review and misuse by reviewers and authors. Attempting to police submissions so that only important supplemental material was included would leave editors and reviewers with a burden comparable to the one they face now, and one that they are unlikely to take on with greater enthusiasm. Allowing The Journal to host supplemental material that has not been peer reviewed is not an option that the Society for Neuroscience is willing to support.
To their great credit, some important journals in paleoanthropology do not rely on supplementary information.
One point that is made too rarely: Separating content in this way impedes accessibility. Are those Microsoft Word and Excel files accessible by screen reader? Do they get transmitted with preprints to readers without subscriptions? Are they even readable for colleagues in developing countries using older or open source versions of spreadsheet software?
PDF files are not perfect in these regards, either. Journals should really be looking for a better way to make their content accessible.