CALLING ONE'S SHOTS, CONVINCINGNESS, ADVERSARIAL COLLABORATION, STANDARDS OF EVIDENCE, AND GUT MICROBIOTA.

My new puppy, Peanut, seemed set on sharing some of her gut microbiota with me during a Thanksgiving hike with my oldest daughter, Sasha Allison . Perhaps Peanut has a strong belief that transfer of gut microbiota from her lean, young, vibrant body to mine will have long-term beneficial effects on me. It is clear that dogs do seem to share some microbiota with their human housemates ?(https://www.mdpi.com/2076-2607/9/3/557), though whether this is beneficial, harmful, or neutral is unknown (at least to me). In a recent presentation at the Royal Society (https://tinyurl.com/bzd9zp6p), Matthew Dalby surgically dissected the evidence for some claims about long-term gut microbiota effects on human health.

?Clearly, some long-term health effects of gut microbiota are established, such as the negative effects of C. difficile infection. But are more recently conjectured effects involving transplanted elements of gut microbiota from one animal to another well established? Dr. Dalby clearly thinks not. I too am skeptical. Yet, others have opined that there is compelling evidence for some clinically meaningful effects.

Purportedly, a devotee of Karl Popper once asked J.B.S. Haldane what it would take for Haldane to change his mind about accepting Darwin’s thesis that selection was a driver of evolution and the origin of species. Haldane is said to have retorted "Fossilized rabbits in the Precambrian." Sarcasm aside, Haldane specified a condition, the occurrence of which would cause him to change his belief. In my opinion, we specify such conditions too rarely in science. As Gary Taubes has commented, a seeming problem in nutrition and obesity-related research is that no one ever gets to be wrong, implying that the methods and clarity of the hypotheses are insufficient to permit determination that a hypothesis is refuted. The search for the opportunity to be wrong in a constructive, and meaningful way, is behind what Daniel Kahneman and others have called adversarial collaboration (https://tinyurl.com/m7fh9e7).

?In an adversarial collaboration, one lays out the circumstances that would cause one to change their mind. If Peanut and I were to have an adversarial collaboration on the effects of gut microbiota, what would I ask for? I might ask to see a fecal transplant experiment in which the statistical analysts and technicians were blinded to the experimental assignments, where the experimental assignments were randomized, where the studies were large enough to permit stable inferences, where the data analysis and experimental plan were specified publicly and in detail with a timestamp prior to beginning the experiment, in which all statistical analyses within the experiment were verified by an independent statistician, where the raw data and code were made publicly available for inspection. Then, I would want to see the entire experiment repeated, in a separate lab, by separate personnel under the same conditions of blinding, pre-specification, large sample size, etc. If each of those two studies produced statistically significant results in the same direction, on the same pre-specified outcomes, with the same pre-specified analyses, I would be prepared to say that, at least on that one type of gut microbiota transplant with that one type or set of outcomes, in that one species, I am now convinced that something has unequivocally been demonstrated.

?Would the converse of that change Peanut's mind or the minds of some of my colleagues who think profound effects have already been demonstrated beyond a reasonable doubt? Questions such as this can be addressed through the new American Society for Nutrition (https://nutrition.org/) initiative on adversarial collaborations (https://nutrition.org/nutrition-2022-live-online/adversarial-collaboration/).