Harvard Professor: “I Tried Every Diet. This Is By Far The Worst.” - Daniel Lieberman

Lieberman describes a research method that combines personal experience with scientific comparison. While writing a book on the evolution of diet, he and his wife tried many major diets intermittently, using those experiences as participant observations rather than treating them as a substitute for research. His wider work focuses on the evolution of physical activity, but teaching nutrition, digestion, and physiology exposed him to a literature that he found deeply contradictory: articles in the New England Journal of Medicine could reach opposite conclusions about saturated fat and appropriate salt intake. He therefore wants to assess diet by combining an evolutionary perspective with biomedical, cultural, anthropological, and experiential evidence. The evolutionary lens is meant to answer why human biology and dietary behavior have their present form, which can change how researchers approach what and how questions. He grounds this rationale in Theodosius Dobzhansky's 1972 formulation that biology becomes intelligible through evolution, while acknowledging that the interview is presenting a framework for inquiry rather than resolving the nutritional disputes itself.

The conversation treats dietary conflict as more than a disagreement about nutrients. Lieberman suggests that food has historically reflected cultural belonging, whereas a more globally uniform food environment and weakening institutions may leave people looking for identity and meaning in what they eat or refuse to eat. That can turn online diet communities into highly cohesive groups and make contrary evidence feel personally threatening. If diet affects longevity, another person's confidence in a different diet can imply that one's own choices are wrong or may shorten one's life. The stakes therefore feel existential rather than abstract. A further problem is that the relevant evidence arrives slowly: someone following a Mediterranean diet may read favorable studies but cannot know for decades whether the choice will produce the hoped-for health outcome. Because long-term effects cannot be observed at the moment of choice, everyone must make some leap of faith. The exchange offers this as an explanation for dietary conviction, not as proof that any particular diet works or that all diet communities are equally extreme.

Lieberman resists treating every diet as equally well supported, while allowing that even fad diets may contain a grain of truth. His stated reading of meta-analyses, long-term studies, and mechanistic evidence favors plant-based diets for health outcomes, but he immediately qualifies the label: plants are not automatically healthy. Sugar, highly refined grains and carbohydrates, and many ultra-processed foods can be plant-based without providing the benefits associated with a varied plant diet. The same caution applies to meat. Meat is nutritious and played an important role in human evolutionary history, so the argument is not that it has no value. Rather, he points to epidemiological studies reporting a dose-response relationship between meat consumption and cancer risk. He recalls, with uncertainty, a figure of an 85-gram serving associated with roughly a 13% increase over some period, with the risk doubling when consumption doubles; the exchange does not specify the exact study or timeframe. He also points toward mechanistic evidence, including carnitine in meat, but this excerpt does not yet explain the mechanism. The conclusion is therefore a qualified preference for plants over animals, not a claim that all plant foods are beneficial or that meat is categorically harmful.

The proposed mechanisms add detail to the earlier epidemiological association without making the case simple. Lieberman says carnitine from meat can be transformed in the gut into TMAO, which he describes as an inflammatory molecule associated with, and mechanistically related to, atherosclerosis and higher blood pressure. He then distinguishes red meat from poultry and fish by discussing the red-meat glycoprotein N-EU-5GC. The interview says humans mount an inflammatory response to it and connects that response to the CMAH gene; the speakers present this as another possible mechanism behind the observed relationship between meat dose and cancer risk, not as a complete explanation. Saturated fat introduces another source of variation: people differ in their response, and genetic susceptibility may mean that it raises lipoproteins associated with heart disease for some people but not others. The exchange also identifies a basic limitation in nutrition research. A plant-based diet may show better outcomes because of beneficial plant components, because meat has been removed, or both. Every dietary substitution changes two sides of the comparison: eating one food means not eating another. This makes it difficult to isolate a single ingredient or conduct a perfectly controlled comparison, and it supports the broader conclusion that meat and diet effects involve trade-offs rather than one universal answer.

The interview explains why nutrition evidence rarely reaches the standard of a perfectly isolated experiment. Even a hypothetical twin study comparing an all-meat group with an all-plant group would still have to account for what each group was not eating, as well as exercise, smoking, and other influences. Lieberman sees the recent appeal of carnivore diets as partly cultural: meat can be associated with masculinity and tribal identity. He recounts James Blunt's reported all-meat experiment and scurvy as an example of how an identity claim can motivate a restrictive choice, while noting that this does not explain every follower. Another explanation is comparison with a poor baseline. Someone can feel better after replacing a diet dominated by fast food and soda with almost any diet that is not similarly poor, without that showing that an all-meat diet is better than Mediterranean, DASH, or blue-zones patterns. The answer becomes more goal-dependent when weight loss is the question. Lieberman cites the National Weight Loss Registry as an observational source tracking people who lost weight and maintained it for years, while acknowledging that it cannot perfectly randomize large groups. The participants' diets vary, but the shared features described here are limited junk food, sugar, and added fat, together with the ability to maintain the pattern. Adherence therefore matters, but the excerpt does not establish that adherence alone determines every health outcome.

The discussion of protein begins with a practical lesson about adherence. Low-fat diets can be difficult because fat makes food taste good, and major studies may show limited effectiveness when participants cannot tolerate the prescribed pattern. Protein has a stronger physiological case than a mere trend: the speakers say it is required every day, that some amino acids are essential, and that the body cannot make them from other macronutrients. The excerpt also says that only small amounts of two fatty acids, named unclearly in the transcript as omega-3 and 'amacic,' are separately required. Protein can increase fullness, costs more energy to digest, and supplies useful nutrition. Higher needs are identified for physically active people, people who are menstruating, people who are pregnant, and older adults who are losing muscle mass. Yet the argument stops short of protein maximization. The body cannot store extra protein as protein; the speaker says unused amounts are converted to fat. Very high intake is described as unlikely to be harmful except at extreme levels, where kidney damage is possible, but the excerpt emphasizes that massive intake has no demonstrated massive benefit. Lieberman also says he has not seen strong evidence for the one-gram-per-pound rule. These claims are presented as a broad nutritional explanation, not as an individualized intake prescription.

Lieberman declines to endorse one gram of protein per pound of body weight as a universal rule. He argues that protein needs depend on what a person does, age, muscle loss, and exercise, so treating nutrition like a fixed medical dose oversimplifies human variation. His preferred principle is enough but not too much, with each person having to work out what that means. This uncertainty reflects a broader problem: people want a simple diagnosis, a magic bullet, and a quick fix, while diet does not reliably provide any of them. The exchange then separates protein from meat. Although meat is rich in protein and provides complete protein, beans, legumes, whole grains, soy, quinoa, and nuts can also supply it. Soy is described as a complete protein, and the interview notes that farming cultures worldwide developed combinations such as rice and beans or corn and beans to provide adequate protein. The point is not that meat contains no nutritional value, but that equating meat with protein exaggerates the available choices. The transcript ends by introducing a separate fiber concern and a commercial reference, so it does not establish a quantified fiber recommendation or independently verify the claim that 95% of people fall short.