17 June 2013

President's Column: Making Excellent Journals Even Better

David Helfand Columbia University

The Society's journals are its most important assets and serve as the primary and longest-lasting vehicles through which we fulfill our stated mission: to enhance and share humanity's scientific understanding of the universe. As with other aspects of our field, the times are interesting in science publishing. Various colors of open access are being mandated in different jurisdictions; with free access after one year, our journals are fully compliant with the recent US government dictates, though scientists in other nations face different constraints. Technological advances offer both cost savings for current activities and new opportunities for enhanced products. And the demographics of our authorship ― 40% of first authors in ApJ and AJ are now from non-US institutions ― present both challenges and opportunities.

This spring we conducted a survey of a sample of our members who publish in the Society's journals and held a workshop on the journals' future. We heard interesting presentations from creative publishers in other fields, reviewed the survey results, and explored a variety of scenarios ― both salutary and less so ― that might impact the future of our publications. The point was not to arrive at specific recommendations or an action plan, but to inform the deliberations of the Publications Board, the editors, and the Council as we navigate through the challenging times ahead.

The current state of our journals is excellent. The number of pages published by the AAS has increased 39% in the last five years; it could top 50,000 pages in 2014. Despite this growth, library subscription costs have increased more slowly than inflation over this period such that the number of libraries hosting our journals has increased over the past decade during which subscription cuts were the norm. The journal reserve funds are extremely healthy, allowing us to weather any anticipated storms with equanimity and to set aside funds to enhance our publications.

And the best news is that page charges (now called quantum charges) are declining. We reduced them by 12% in 2012; at the Council meeting in Indianapolis, we approved another 14% drop for 2014. I suspect very few of your other research expenses have dropped by one-quarter over the last three years. Furthermore, the Council also voted to offer AAS members a significant new benefit beginning next year: an additional 15% off the total charges for a single paper published in 2014 if membership dues are renewed by 31 December of this year. For most people the value of that discount will exceed the cost of membership ― yet another reason to join the Society and bring your recalcitrant colleagues along with you.

Being incapable of resisting the temptation to play the provocateur when given the platform, I'd like to follow all this good news with a proposition: when one publishes an article, one should publish the data that underlies one's conclusions simultaneously.

The greatest strengths of the enterprise we call science are the twin pillars of reproducibility and falsifiability. The latter ― the notion that a valid scientific hypothesis or model must admit the possibility that it could be confronted with data and proven incorrect ― is fundamental to our progress in crafting ever more refined and powerful descriptions of nature. Reproducibility is essential in this context. The presentation of any scientific result should be met with healthy skepticism, and for that result to take its place as a tile in the mosaic of science, its conclusions should be reproducible.

At the Journals Futures Workshop we were shown an example of a publication in which an astronomer had rotated and displaced an astronomical image in order to obscure its location from competitors. To me, this is not science. But without access to the original data, a referee or editor would be hard-pressed to uncover this subterfuge. Any reader interested in reproducing the result, however, would have quickly found the discrepancy if a link to the data had accompanied the article; indeed, I suspect if such a link had been required, the discrepancy would not have existed in the first place.

We are a data-rich ― and data-driven ― field. Our journals are already in the process of implementing a "data-behind-the-figures" feature that will forever obviate the need to trace plots against a window. But I am advocating something much greater: a link in articles to the data that underlies a paper's conclusions.

Yes, I know the objections: "Which data, raw or processed?" "Where will it be archived?" "That's unfair to my poor graduate student who hasn't finished wringing the last bit of science out of her data yet!" "I don't have the time or resources to make my data available." "I expended all this effort to get these data (build this instrument, fundraise for this telescope) ― it belongs to me!"

Space limitations preclude my rebuttals to each of these objections (though I am happy to provide them on request). There are, of course, a number of real issues that would have to be addressed were such a policy to be implemented. But I was encouraged by the results of our author survey that showed a solid majority of those authors now publishing in Society journals agree on the value of sharing data: 62% of AAS authors reported sharing datasets in the past two years (vs. 23% for plasma physicists publishing in AIP journals). A total of 86% of corresponding authors said they "probably or possibly" would be sharing datasets as a supplement to a published article in the near future, while only 2% said they definitely would not do so.

In my view, the time has come ― and the technological resources are available ― to make the conclusion of every ApJ or AJ article fully reproducible by publishing the data that underlie that conclusion. It would be an important step toward enhancing and sharing our scientific understanding of the universe.

With best wishes for a productive and/or enjoyable summer!