For this second part of our investigation, we interview Paul Gibson, RIS at Institut Polytechnique de Paris.
A fascinating conversation that takes us to the heart of scientific integrity, and the temptation of misconduct.
Laurence Moss – How long have you been working as a RIS?
Paul Gibson – 10 months. My colleague, Henri Maitre, suggested I take the position
Laurence Moss - Why were you interested in the job?
Paul Gibson – I worked on a European project about digital ethics, and I found this very interesting.
Laurence Moss - What is your background?
Paul Gibson – I have a Ph.D. in math and computing.
Laurence Moss - In your opinion, what are the pillars of scientific integrity? Paul Gibson – Being honest, avoid bias, especially with math and statistics as it is always possible to present things a certain way.
Respecting others, colleagues, and society is also extremely important. It’s not “Can I?” we should ask ourselves, but rather “Should I?” We really need to think about the impact on society. For example, blockchain had a very negative impact on society but there is little debate about this.
It is essential we ask ourselves the right question before doing anything, and not act to win a race. Same for AI.
Always think about the risk, the impact.
Laurence Moss -The Office Français de l’integrité scientifique (French Office for Scientific Integrity states that « In France, scientific integrity is now defined in the Research Code (article L.211-2) as 'the set of rules and values that must govern research activities to ensure that they are honest and rigorous'”. » Anything else you would like to add?
Paul Gibson – Again, to think about the impact on society. Most people are honest but do not ask themselves the right questions. We shouldn’t go so fast. The level of rigor has declined in the past twenty-five years, especially where publishing is concerned. Many scientists, including myself, believe that many articles are being published prematurely; but the practice continues. Often, if an article is rejected by one journal or conference then it is just sent on to another, until it gets accepted somewhere.
Laurence Moss - Have you ever witnessed any form of fraud during your career? Can you talk about it?
Paul Gibson – I did, in Ireland, 3 times. In one of the cases, a Ph.D. student was asked to manipulate data in an unscientific manner. I was informed, we took legal advice, and it went on for eighteen months. In the end, no one confessed so no sanctions were imposed.
This is why I am interested in this grey area between the law and ethics. And most of all, I want to support these young researchers who have to deal with these difficult situations.
Laurence Moss - What are the consequences for misconduct?
Paul Gibson – The researchers can be fired. I used to know a sexist researcher. He would avoid putting female colleagues as first authors, even if they had done all the work. He regularly made unpleasant comments about women. But then, it wasn’t clear whether it was illegal or not.
Laurence Moss - Publishing articles is often mentioned as leading to misconduct. Is this a consequence of “Publish or perish"?
Paul Gibson – Absolutely.
Laurence Moss - Why so much pressure?
Paul Gibson – You know, I didn’t publish any article for my doctorate. But now, you’re supposed to have published articles even before writing your thesis! It has even become a recruitment tool. Students are asked how many articles they have already published! It’s only the numbers that matters; in most instances, no one even reads the articles.
Plus, there is no quality monitoring. How can one publish twenty-five papers a year?! There must be time for researching! It is much more reasonable to think in terms of two or three publications a year.
Laurence Moss - What would you say is the influence of Paper Mills (companies that sell false of plagiarized articles to authors) on global publication?
Paul Gibson – It certainly has an influence. Especially in the fields of math and computing. And it’s even worse with AI. I strongly suspect some articles are written by Chat GPT! The main goal nowadays isn’t to do some research but rather to publish at all costs. The whole publishing system needs to be brought into question. Peer-reviews have also lost in quality because no one has time anymore. But I don’t know have to fix the system.
Laurence Moss - Pierre Corvol, the author of a report on scientific integrity, (Bilan et propositions de mise en œuvre de la charte nationale d’intégrité scientifique-2016), in his interview with Colligere (L’intégrité scientifique selon Pierre Corvol : épisode 2, par Pierre Corvol, Laure Léveillé - 24/01/2022 speaks about « salami slicing » (“the act of dividing an article into several parts and produce as many publications counted as different”) and “cherry-picking”( giving priority to data and/or results that tend to corroborate the initial hypothesis). What would you say about this?
Paul Gibson – Indeed, “salami-slicing” increases the number of publications. An example: for simulations, we want to optimize things, costs, etc. so it is possible to use the same scenario, change a value, and bam, a new article is published. It is the same as the previous one though; all you need to do is change one parameter.
In the health sector, it happens frequently that the end results of the research are not published but rather the end results of each step. This allows for more publications
Regarding “cherry-picking”, it is absolutely right that there is a will to minimize risk otherwise you could not publish for three years. In that case, data is cooked, filtered, and that is dishonest.
In the last few years, I have noticed serious mistakes in statistics, as the data was wrong right from the start.
I personally think that negative results are as important as the positive ones because it helps research. But that’s not everyone’s opinion. If a result is negative then there is no publication, and the team will never agree to this. Rewriting a hypothesis to have positive results is so easy. Yet risk is part of research!
But as far as Ph.D. students are concerned, I give them work that does not entail too much risk. It’s only natural.
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.
Laurence Moss - Who would you consider a model of scientific integrity?
Paul Gibson – Barbara Liskov (an American computer scientist). She never puts her name on a publication she didn’t work on. She thinks it belongs to the researchers (often Ph.D. students) who have actually worked on it. If she does add her name, it’s always in the appropriate position. She has very good publishing ethics. And she also helped women a lot in her field.
Laurence Moss - What would you like to change being a RIS?
Paul Gibson – I would like to have more power to convince that having integrity is very important, that classes on integrity should be an integral part of the cursus, and as important as all the other classes. Students mostly feel obligated to attend workshops on integrity whereas they are fundamental. Integrity is part of life.
My objective as a RIS would be to convince that asking oneself questions is a life principle, which applies every day, and in all fields.
Image by Freepik
Envie d’être formé à l’éthique de la recherche scientifique ?