Search Google for the phrase “ileal-lymphoidnodular hyperplasia,” and you are likely to find several free copies of a popular medical article hosted on public websites around the Internet. The problem is, this article was retracted in February 2010, the result of a investigation that ultimately found the paper fraudulent and stripped its author of the right to practice medicine.

If the medical terminology of this paper is still confusing you, this is the discredited study by Andrew Wakefield, who in 1998 published a paper which linked early-childhood vaccination with autism and began a decade of mass hysteria about the safety of immunization.

If you make it to the journal’s website, you will find the words “RETRACTED” printed boldly on the article; however, many readers will never get this far. There are many free copies of the earlier (unretracted) version sitting on public websites. For all intents and purposes, the article is still published.

The vaccine-autism scandal received international press coverage. Yet every year, thousands of articles are either retracted or corrected with little (if any) media attention, and there is little a publisher can do to alert readers if they don’t visit the journal’s website for a new copy.

In her article, “Distinguishing published scholarly content with CrossMark“ (Learned Publishing, April, 2010), Carol Anne Meyer, Manager of Business Development and Marketing for CrossRef, writes:

It stands to reason that if a publisher goes to the trouble of issuing a change notice for an article, a mechanism should exist to easily alert readers to those changes.

The issue of persistent, uncorrected errors in the scientific literature is not new. Errors promulgated easily in print journals because readers rarely became aware of editorial notices appearing in later journal issues. Electronic publishing made it much easier to update articles, but for those who routinely download PDF copies to archive on personal computers, getting updates to readers is much more difficult. Starting 1987, the National Library of Medicine began indexing retractions and correction when notified by the publisher.

Article repositories only add to the problem of versioning. While they are designed to archive and disseminate copies of research papers, few have any mechanism to correct errors, update the status of articles, or alert potential readers if the article becomes retracted. The problem is even worse for papers stored on personal websites.

For example, the 20o2 article, “Rule learning by cotton-top tamarins” was subsequently retracted by the journal Cognition; however, the in-press copy is still available from the Harvard lab’s website and shows up 1st in Google while the publisher’s website is listed 5th.

In spite of publisher guidelines on how to alert readers of retractions, the problem of versioning wasn’t being solved on its own. A new approach was needed to provide readers with authoritative source on the status of articles, and not only for high-profile retractions, but for the thousands of corrections and updates that are issued by publishers each year.

This is the rationale for CrossMark, a new service by CrossRef which will debut later this year.

According to Meyer, articles baring the CrossMark logo will be linked to metadata on the status of the article.  Clicking on the logo will retrieve that information. Most often, readers will see a message that the document is current, although occasionally they will alerted that the document has been updated, corrected, or retracted. In these cases, a CrossRef DOI link will point readers to the publisher’s website for the most current version.

CrossMark is not limited to journal articles, but can be used for any kind of document that has been issued a DOI, like books, book chapters, and conference proceedings. According to Meyer, CrossMark is secular when it comes to publishing model; a publisher, however, must commit to maintaining the content with any updates. She writes:

At CrossRef, where persistence is part of the mission statement, we believe that the act of publishing a document implies a commitment to maintaining stewardship of it for the long term. Whether a publisher chooses to adopt the NISO definition of a version of record or not, it commits to communicating changes, errata, revisions, or, in the worst case, retractions. Prepublished versions of scholarly content may be convenient and free, but they do not come with the level of commitment that publication entails.

CrossMark is the brainchild of Geoff Bilder, Director of Strategic Initiatives as CrossRef. Many know Bilder for his intensity, although he is not above a little good-natured ribbing from the Scholarly Kitchen (see CrossDress). The idea for CrossMark came to him several years ago when visiting a publisher and saw a retraction notice for a medical reference work tacked to the bulletin board in their lobby. “How could it be that after almost a decade of electronic scholarly publishing, we still had no standard, automated way to alert people to changes in the status of published scholarly documents?” he responds by interview.

Over the past several years, pressure from funders, universities, and governments have increased the number of article versions being placed in public archives, mostly in the form of author manuscripts. Little is known on the status of many of these documents and how many were subsequently retracted, corrected, or updated.

We have a solution. Now it’s important to document the extent of the problem.