Co-Author Question Dominates Ethics Panel Discussion
Last year's high-profile cases of scientific fraud may have been resolved, but the aftershocks are still rippling through the physics community, as became apparent during a panel discussion on scientific ethics at the APS March Meeting in Austin, Texas. The panelists provided a broad overview of the various issues involved, but it was the question of the responsibility of co-authors in cases of fraud that dominated the audience concerns and subsequent discussion.
Pierre Hohenberg (Yale University) distinguished between two types of ethical issues: those related to the applications or misuse of science, and those related to the process of scientific research (scientific misconduct). "Physicists have led the way in questioning the ethics of such scientific applications as nuclear weapons, for example," he said. But most of the speakers agreed that in the past, the physics community has felt overly secure in the fact that, because of its reliance on reproducibility of results, physics would remain largely unaffected by the type of blatant misconduct that plagued biomedicine in the 1980s and 1990s.
Then came allegations that Victor Ninov (Lawrence Berkeley National Laboratory) had fabricated data to support the discovery of Element 118. It was followed closely by similar allegations of data falsification against Henrik Sch"n, a materials scientist at Lucent/Bell Labs. Two separate investigative committees were formed to determine whether fraud had been committed.
Fortunately for the committee members, there were federal guidelines already in place to assist them in their deliberations. Formally issued in December 2000 by the White House Office of Science and Technology Policy (OSTP), the guidelines focus on fabrication, falsification and plagiarism. The alleged misconduct must have been done "intentionally, knowingly or recklessly," and be supported by a preponderance of the evidence.
Although the APS adopted guidelines for professional conduct as early as 1991, relatively little attention was paid to the matter in the physics community until recently. As a result of the Ninov and Schön cases, the Society revised its guidelines (See APS News, January 2003) and called for universal adoption of the OSTP guidelines. Arthur Bienenstock (Stanford University), who was instrumental in the development of those OSTP guidelines during his tenure as its associate director for science, believes the APS action was ;necessary, since a few federal agencies—most notably Health and Human Services and the DOE—have not yet published implementation plans called for in the guidelines. Also, there are rare institutions that don't receive federal funding and may not have policies in place for research misconduct.
LBL's George Trilling—a former APS president who was a member of the Ninov investigative committee—gave an overview of the facts surrounding the case, and speculated on possible motivating factors for the misconduct, including the presence of a highly competitive situation, with several labs vying to be the first to announce the discovery.
Malcolm Beasley (Stanford University), who chaired the Lucent investigative committee, reported that his committee found the guidelines particularly well-designed and useful. But the central issue that sparked the most heated debate during the subsequent discussion was the responsibilities of co-author. In both the Berkeley and Lucent cases, there were experienced, respected co-authors who nevertheless failed to detect the fabrications.
"Given the importance of the result, it was incredible that, prior to publication, no one had looked at the raw data for the particular events claimed to make sure that there had been no errors " said Trilling of the Element 118 case. "Extraordinary results demand extra ordinary supporting evidence, and the burden of proof for an unexpected or major discovery is much greater than for a routine measurement." He added that journal referees "could and should" help enforce this principle.
Yet determining co-author responsibility is a complex issue, due in large part to the different cultures of the various subfields of physics. For example, high energy and nuclear physics are typically characterized by massive projects with hundreds of collaborators. A typical paper may have as many as 500 co-author, each of whom has made a significant contribution to a small part of the overall project.
"We have to feel responsible and be held accountable," said Beasley. "But I don't think there's an easy checklist; the guidelines should not be overly prescriptive. We must leave room for discretion, because we can't define the issue in such a way that would apply to all individual cases." This is one reason why the Berkeley committee ruffled feathers with its sharp rebuke of Ninov's co-authors, while the Lucent committee received criticism from a few individuals for not chastising Schön's senior co-authors more directly.
Because of these differences, there was general agreement among the speakers that the OSTP guidelines should not be amended to address the co-author issue. It is an issue more appropriately left to the scientific community to resolve, and professional societies like the APS have a vital role to play by fostering further discussion and setting their own internal policies to address potential fraud.
"The federal policy represents the law and it carries with it legal repercussions for research misconduct," says Bienenstock. "Ethics go beyond the law. You don't want to limit things so much that you hinder good science from being performed."
However, despite the community's best efforts to guard against misconduct, "With a sufficiently motivated and capable hacker, fraud is always a possibility," said Trilling. He insisted that the independent confirmation of scientific results is still the best guarantee against misconduct, and independent confirmation within a collaboration provides the best guarantee that a fraudulent result is not published—"something that could have been done by the Element 118 collaboration, but was not." The same considerations apply not only to fraud but also to sloppy work or to errors in data analysis.
While the speakers agreed that, in the end, the scientific system worked, several emphasized that many younger colleagues of the perpetrators of the fraud were hurt by the misconduct. In the Schön case, for example, there were a good number of young condensed matter physicists who were establishing their careers by building on what they thought were favorable results, and they received a very rude jolt when it turned out to be a house of cards.
"For the most part, the system served science well, but there has been long-standing criticism of the process by which we conduct the business of science," said Beasley. "It's important to understand how much science has changed, and how those changes are demanding a re-examination of professional ethics. We have not become less ethical, but the circumstances under which we work have changed. We need to adapt accordingly."