Plagiarism and the role of publishers

Anyone who reads Just Like Cooking or Chemjobber is by now probably aware of the freshly-discovered plagiarism case wherein a 2013 Chem. Eur. J. article by Professor Xi Yan lifts, verbatim or nearly so, entire portions of a three-years-prior (2009) JACS paper by Professor Valérie Pierre. Reading the commentary, the plagiarism is pretty egregious. Additionally, comments and a further blog post by See Arr Oh have brought up more instances of plagiarism like this.

Of course, #spacedino is immediately brought to mind (for the uninitiated, that particular hashtag refers to the 2012 controversy where esteemed chemist Breslow, of Columbia University, was found to have submitted what amounted to essentially the same publication to multiple journals; it was only really caught because biologists took his lame ending joke seriously). After lengthy conversation on various media and social media outlets, the offending papers were retracted at Breslow's (and others') request (though initially he denied wrongdoing).

While the Breslow violation was aggravating and arguably wrong, the Yan violation is worse. I don't think that's a very controversial point to make, but it's an important one to recognize. The Breslow affair fell into what some categorized as an ethical grey zone; the Yan violation is in clear things-they-tell-you-from-day-one-in-university territory.

Rapid retraction is vital, I'd wager; thought many retracted papers still get heavily cited or believed, early correction is probably the best way to prevent such propagation. Once an article has been believed as true for a substantial period of time and gets included in dissertation of paper-introduction citation-vomits, less care is taken to check the original source and see the glaring retraction notice. Better to stop the train before it leaves the station.

I think this highlights the role of open and quick dialogue among scientists, along the same lines as Blog Syn. The ability to rapidly disseminate these issues when they are discovered can, ultimately, improve the quality of scientific communication. In a pre-social-media era, it was certainly easier to just assume a paper would slip by unnoticed. Not so much anymore; via Twitter, for instance, a wide audience can be quickly reached. See Arr Oh wrote that he has notified the two publishers (Wiley and ACS); it will be very interesting to see what action occurs.

Of course, a point could be made that such post-publication watchdog action shouldn't be necessary. Aren't these things peer reviewed? Alas, I think anyone with a healthy dose of realism is aware that a good bulk of reviewers put in the minimum conceivable effort in reviewing (consider the sheer quantity of bad Supplementary Information files). Usually, this is justified by a variety of factors, namely that reviewers are (1) busy and (2) unpaid. (And they aren't held accountable if the article is later retracted).

But hey, you know who are paid? Publisher staff. I'm a bit surprised that more journal editors haven't made a practice of submitting manuscripts to plagiarism detection software. There's many available; universities often have TAs use them to check undergraduate papers. One comment on Chemjobber's blog points out that Elsevier provides access to one such service, iThenticate, to its reviewers. One readily available service, eTBLAST, compares blocks of text to databases of scientific papers using the BLAST algorithms commonly employed in bioinformatics for sequence alignment. In short, many tools exist for plagiarism detection. So what's the deal, journals? A common claim of open-access opponents is that paywall publishers "add value" to papers by the publication process.

So why aren't all papers routinely submitted to these checks after peer review and before publication? Ideally, reviewers should do a thorough job vetting submitted articles and should be supplied with tools to do so. But publishers share that responsibility. If you can Cantrill an article, it shouldn't slip by.

(One caveat: I'm not sure about the journal coverage of many of the anti-plagiarism packages. eTBLAST, for instance, appears to have access to Medline and PMC, but doesn't seem to have many of the synthetic journals in full-text. That's from an initial assessment; I may be wrong. CrossCheck, the service powered by iThenticate, has a very wide list of included content, but interestingly, I don't immediately see ACS on the list). Is it possible that some of these illustrated issues just don't get picked up by the software?

Lastly, something that would be very interesting: how much scientific plagiarism occurred in the pre-software-detection era? That is, how prevalent were issues of plagiarism in the 1960s? If one took journals from the 1960s, for instance, and ran them through this software--what would they find? Is it more prevalent now? Is it less prevalent? When did scientific plagiarism mature? I suspect the issue is a very old one.