The “Voinnet Case”: Lost Control
(January 13th, 2016) What was behind the multiple paper manipulations by Olivier Voinnet et al.? An attempted explanation...
In September 2014, a whole slew of suspicions publicly surfaced that a good deal of papers by the Zurich-based plant scientist and small RNA pioneer Olivier Voinnet might contain multiple figure manipulations. Until today, almost one-and-a-half years later, seven of his papers have been retracted and 18 – in part huge – corrections have been issued so far. Hence, the case has been crystal clear for quite a while now: The lion’s share of the suspected figures was indeed manipulated, and the name “Olivier Voinnet” now stands for one of the biggest and most spectacular cases of scientific misconduct ever.
Nevertheless, there is one thing about this whole story, which still remains perplexing to every observer: the extremely high citation counts of the manipulated papers. Just three examples: According to Google Scholar the 2003 paper of Voinnet et al. in The Plant Journal (vol. 33: 949-56) had already accumulated more than 1,200 citations until its retraction a couple of weeks ago; the likewise retracted 1998 article of Giannina Brigneti et al. in The EMBO Journal (vol. 17: 6739-46) had been cited 1,150 times; and the corrected 2002 publication of Andrew Hamilton et al. in The EMBO Journal (vol. 21: 4671-9) had also already received more than 1,000 citations at the time of its correction. Numbers, the vast majority of Voinnet’s research colleagues can only dream of.
What is so perplexing about this fact is that, apparently, a whole lot of colleagues successfully (!) built at least parts of their own subsequent experiments on the manipulated papers by Voinnet and Co. – and obviously none of them became suspicious about the integrity of the data. In other words, the observations, discoveries, insights and methods described in Voinnet’s papers did not seem to cause any problem when using them as the basis for further experiments – which basically means that they have subsequently been confirmed multiple times over, even if originally presented in considerable part by faked figures.
This point is, for example, nicely illustrated by the first article mentioned above – Voinnet et al. in The Plant Journal 33: 949-56 – which was published as a “Technical Advance” paper titled “An enhanced transient expression system in plants based on suppression of gene silencing by the p19 protein of tomato bushy stunt virus”. It is obvious that for the majority of the more than 1,200 articles citing this paper the respective authors indeed adopted the described “p19 method” themselves, to enhance the expression of certain gene constructs – and successfully so, as their publications finally proved. Hence, the method Voinnet et al. had developed was true and worked, even if they had presented it with faked figures to an extent that finally required the retraction of the article. No wonder, therefore, that Sophien Kamoun from The Sainsbury Laboratory in Norwich, UK, instantly commented on Twitter: “Voinnet retracts highly cited paper with >900 citations – P19 does work and we all use it. What a mess...”
The high citation counts of the faked Voinnet papers therefore bear two pressing questions: Are the discoveries described in the later retracted and corrected papers of Voinnet et al. perhaps overall correct? And if yes, why did he then have to manipulate figures when he and his co-workers were actually “on the right track” with their experiments?
Possible answers come from a conversation I recently had with someone who was involved in the formal investigation of at least a part of the “fishy” Voinnet papers. “It is a very weird case,” he stated. “Voinnet presented us with raw data of the described experiments – and, in many cases, they were even better than what he had published in the faked figures.”
Apparently, so the “investigator” continued, Voinnet had adopted a strategy to start writing the paper manuscripts right from the moment it was basically clear in which direction the respective study would develop – and which further results could therefore be expected with high certainty. Accordingly, Voinnet constructed “placeholder figures” to provisionally fit them in the places of the manuscripts where the appropriate real data were still missing because the corresponding experiments had not yet been completed or even performed at all. “And all of this only to publish as fast as possible in that highly competitive small RNA field,” the investigator surmised.
“I am sure Voinnet had wanted to replace the placeholder figures by true figures as soon as the data were at hand but somehow he must have lost control at a certain point. It seems as if he just ‘forgot’ it in certain cases. Unintentionally, by pure sloppiness? Perhaps. On the other hand, however, he might later well have decided to publish one or the other constructed ‘placeholder’ directly once he had made the experience that they didn’t make a difference to the ‘stories’ – and nobody noticed it anyway. But this point, of course, is speculation,” the investigator emphasised.
Indeed, the sheer mass of manipulations that has already led (and is still leading) to the high number of corrections and retractions makes it unlikely that all of them went into the journals unnoticed and without any intention. Rather, it “smells” like something systematic. And this systematic factor might well have been the creation of preliminary placeholder figures for experiments that hadn’t (yet) been done at the time but for which was already very clear what results they would yield.
The picture that therefore emerges is that, on the one hand, Voinnet’s brilliant brain led him on the right track towards several important and “true” discoveries. On the other hand, however, it seems that under the pressure of an almost absurdly competitive science system he nevertheless felt the necessity to speed up the publication of his and his team’s insights by implementing strategies that he soon lost control of – either this way or the other.
Does that mean you can put part of the blame on the system? No, not at all. The system and its rules are the way they are, and Voinnet has clearly been aware of them. Hence, there are no excuses for his manipulative actions – even if the insights and conclusions of the papers in question might prove and remain “true”.
Strange, therefore, that Voinnet’s current employer, the ETH Zürich, has not imposed any really substantial sanctions against him – stating as a reason that, in spite of his misconduct, the scientific conclusions of his papers have proven to be “substantiated”. Fortunately, however, this seems not to be in line with what the research community is thinking. When our “investigator”, himself a plant scientist like Voinnet, was asked whether he thought that Voinnet would now be able to somehow re-start his research career, he replied, “No, I can’t imagine that Voinnet will ever again get his foot in any door.”
Previous articles on the "Voinnet Case":
- Too Much to be Nothing? (March 24th, 2015)
- The "Voinnet Case": Background of a Retraction plus an Open Letter (April 13th, 2015)
- The "Voinnet Case": Correct Corrections? (July 7th, 2015)
- The "Voinnet Case": Incorrect Measures for Incorrect Figures? (July 12th, 2015)