The Best Diet Is Sustainable, Plant-Rich, and Minimally Processed

Daniel Lieberman does not think nutrition can yield a universally “optimal” diet. The practical objective is more modest: choose a pattern that supports a person’s goals, accounts for their risks and preferences, and can be sustained outside ideal conditions.

That matters because the outcomes people care about—heart disease, cancer, frailty, longevity—can take decades to emerge. Someone can adopt a Mediterranean diet after reading encouraging research, Lieberman says, but will not know for 30, 40, or 50 years what that decision meant for their own health. Everyone therefore makes some leap of faith, whether they rely on a community, a dietary doctrine, a clinician, personal experience, or a reading of the research.

Lieberman borrows Isaiah Berlin’s distinction between the hedgehog and the fox. Hedgehog diets have one large answer: protein is the answer; meat is the answer; cutting carbohydrates is the answer; never eating animal products is the answer. The fox sees substitutions and trade-offs. Eating more of one thing generally means eating less of something else. A dietary pattern may also travel with exercise, smoking, income, education, culture, or social habits that shape health independently of the food.

That is why he resists treating dietary advice as a dosage instruction with a single correct number. Protein needs can depend on age, physical activity, pregnancy, menstruation, muscle loss, and other circumstances. A diet that produces rapid weight loss may be useful for that purpose while being difficult to maintain. A restrictive diet may relieve a person’s symptoms while creating a different concern elsewhere.

An evolutionary perspective reinforces the objection to perfection. Humans were not designed around one nutritional protocol. They evolved in varied environments, using varied foods, and natural selection favored reproductive success rather than guaranteed health into old age. Lieberman’s shorthand is “food in equals babies out.” The fact that an ancestral population ate something is not, by itself, a demonstration that it is the right modern health intervention.

The modern problem is also historically unusual. Hunter-gatherers and farmers generally ate what they could find or grow. In the remote Kenyan region where Lieberman has worked, a household meal might reliably be ugali—a cornmeal porridge—with greens, beans, or both. Nobody debates whether to have Italian, Chinese, low-carb, or paleo for dinner; they eat what is available. Modern consumers face tens of thousands of products, many of them designed for convenience and appeal, and must repeatedly decide what to eat.

That abundance helps make diet a source of identity. Lieberman thinks people increasingly define themselves by what they eat, or what they refuse to eat. The result can be what Chris Williamson called “nutritional theology”: communities organized around a dietary belief that feels consequential enough to defend as a personal truth.

Lieberman’s reading of the evidence favors plant-rich dietary patterns over meat-heavy ones. That is not an argument that anything derived from plants is healthy. Sugar, refined flour, and much ultra-processed food are plant-derived. “Plant-based” is meaningful only when it refers to the quality and overall pattern of the diet.

Meat is nutritious, was important in human evolution, and is a complete protein source. But large meta-analyses, long-term observational studies, and mechanistic research generally point toward better outcomes with plant-forward patterns, Lieberman argues. He cites dose-response associations between meat consumption and cancer risk, alongside possible mechanisms involving carnitine, TMAO, and Neu5Gc in red meat.

Carnitine can be transformed by gut microbes into TMAO, a molecule Lieberman describes as inflammatory and mechanistically related to atherosclerosis. Humans also have an inflammatory response to Neu5Gc, a glycoprotein found in red meat. These pathways are part of the comparative case, not a reason to pretend the question is settled for every person or every dose.

Saturated fat introduces another layer of variation. Meat tends, though not invariably, to contain more saturated fat than many plant foods. Some people may show little apparent consequence, while others may be more susceptible to increases in lipoproteins associated with heart disease. Meat can be nutritious, yet a diet centered heavily on it may carry long-term costs that plant-rich alternatives avoid.

Carnivore diets can still make people feel better. One explanation is substitution: someone moving from soda, fast food, and a generally poor diet to an all-meat regimen has removed a great deal of junk food. Their improvement does not establish that carnivore eating would outperform a Mediterranean, DASH, or other whole-food pattern over the long term. Lieberman also sees a tribal element in the movement, with meat’s associations with masculinity helping explain some of its appeal.

Chris Williamson described the attraction from his own experience. After living in a house with toxic mold and experiencing inflammation and brain fog, Williamson adopted a highly restrictive carnivore-style diet. The simple rules—essentially meat, blueberries, or raspberries—made it easy to follow, and he said he felt “remotely human” for the first time in months. He also became extremely lean and felt physically better.

But Williamson reported LDL in the 400s and what he described as plaque progression on repeat imaging over roughly nine months. Lieberman’s response was not that the diet could not have helped Williamson feel better. It was that “everything is a trade-off.” Immediate symptom relief, body-composition changes, and a possible longer-run cardiovascular concern can coexist.

Protein is one reason the plants-versus-animals framing can become misleading. Humans must obtain essential amino acids from food. The body can make fat from other substrates and can transform fats and proteins into sugar, but it cannot manufacture every amino acid it needs. Small amounts of omega-3 and omega-6 fatty acids are also essential.

Protein is filling, takes energy to digest, and can be especially important for people who are highly active, pregnant, menstruating, or trying to preserve muscle as they age. But “protein-maxxing” is not a general solution. The body cannot store an unlimited protein surplus as protein; excess beyond what it can use is converted, Lieberman says, into fat. Very high intakes may eventually create kidney problems, but he has not seen strong evidence that massive intake confers massive benefits.

Nor does protein require meat. Beans, legumes, whole grains, nuts, quinoa, and soy all contribute. Soy is a complete protein, Lieberman notes. Across farming cultures, combinations such as rice and beans or corn and beans solved the practical problem of obtaining complementary amino acids.

The central question is not whether plants or animal foods win a cultural contest. It is whether the overall diet supplies adequate protein and is built around foods a person can continue eating without relying on a permanent act of willpower.

The Mediterranean diet has probably accumulated more evidence in its favor than any other named dietary pattern, Lieberman says. He also cautions that this may partly reflect where nutrition research is conducted: European and American researchers have studied Mediterranean eating far more than traditional Asian, African, or Mesoamerican dietary patterns.

The evidence does not come from one decisive experiment. It comes from several kinds of studies with different strengths and limitations.

Epidemiology looks for statistical relationships between what people eat and outcomes such as heart disease, cancer, hypertension, or arthritis. Its difficulty is co-variation. Vegetarians may be less likely to smoke. People who exercise may be more likely to eat well. A group’s food choices can be tangled with work, education, social life, and broader health behavior.

That is a major problem, but not a reason to discard epidemiological evidence. A perfect alternative is not available. Researchers cannot assign genetically matched people to radically different diets for 50 years, control exercise and smoking, and wait for health outcomes to appear. Rejecting observational evidence because it cannot prove causation would remove one of the few ways to study long-term patterns. Rejecting shorter trials because they cannot capture an entire lifetime would make the opposite mistake.

Lieberman points to the Spanish PREDIMED trial, which enrolled more than 7,500 people and assigned participants to Mediterranean-style dietary interventions supplemented with olive oil or nuts, or to a comparison diet. He says the study was stopped after roughly five years because the benefits in the Mediterranean groups were large enough that continuing the comparison was judged unethical. It is one part of a larger convergence: prospective trials, controlled studies of health markers, epidemiological work, and plausible biological mechanisms.

The Mediterranean pattern performs well because it combines several features rather than relying on a magic ingredient. It tends toward whole grains instead of refined flour; fruits, vegetables, legumes, and diverse plant foods; fish and omega-3 fatty acids; and monounsaturated fats such as olive oil. It also tends to include less added sugar, fewer refined carbohydrates, less saturated fat, and less red meat.

Lieberman distinguishes that formalized dietary pattern from everything people may actually eat around the Mediterranean. A carb-heavy Italian dinner is not automatically a Mediterranean-diet intervention. Nor is olive oil necessarily the irreplaceable source of the pattern’s benefits. Traditional Asian, African, Mesoamerican, and Mediterranean farming diets can share the underlying features: minimally refined staple foods, legumes, vegetables, dietary diversity, and relatively little ultra-processed food.

The evidence supports a family of patterns more readily than it supports a branded formula.

Lieberman sees little objection to much of the Blue Zones eating advice. It emphasizes beans, vegetables, relatively little meat, moderate portions, social connection, and habits such as not always finishing one’s plate. Nutritionally, he sees it as broadly adjacent to the Mediterranean diet, often with even less meat.

His objection is to the marketing claim that following a Blue Zones regimen can make someone live to 100. Even accepting the premise that designated regions have a higher proportion of centenarians, he says the absolute numbers are tiny: roughly six people in 10,000 compared with roughly three in 10,000 in the United States. Those are statistical outliers, not a basis for promising exceptional longevity to the average person.

Lieberman also raises doubts about age validation. He cites the work of Oxford demographer Saul Newman, who examined remarkable-age records and argued that incomplete documentation can make extreme-age claims unreliable. Lieberman says only 18% of people in the Blue Zones had birth certificates, leaving substantial room for mistaken or self-reported ages.

Loma Linda, California, makes the dietary story still harder to isolate. Its long-lived population includes many Seventh-day Adventists, who tend not to smoke, value exercise, and often follow vegetarian diets as part of a religious commitment to caring for the body. Those habits cannot be reduced to a menu supposedly shared with centenarians elsewhere.

The Blue Zones case, as Lieberman presents it, is not evidence for a special diet that guarantees a special lifespan. It is an example of how reasonable advice can be attached to a claim far larger than the data can comfortably bear.

Lieberman accepts the basic biology behind low-carbohydrate diets. When carbohydrate is digested, some is broken down into glucose. Blood glucose must remain in a narrow range, so rising levels trigger insulin release. Insulin helps cells take up glucose and promotes energy storage.

Reducing carbohydrate can keep insulin lower and, in Lieberman’s account, make stored fat more available for use. He says low-carb diets tend to have somewhat greater efficacy for weight loss than low-fat diets under controlled conditions. He also says he co-authored a paper on the carbohydrate-insulin model.

But efficacy is not effectiveness. A diet can work when participants follow it perfectly and still have limited real-world impact if most people cannot remain on it. Lieberman tried Atkins and found the carbohydrate cravings difficult, particularly as a runner. The difference between low-carb and low-fat weight loss narrows, he says, once adherence becomes the constraint.

Low-fat diets can work as well. He points to DASH—Dietary Approaches to Stop Hypertension—as a well-studied, lower-fat diet originally developed for blood pressure. People who sustain it can lose weight. The meaningful distinction is not carbohydrates versus no carbohydrates; it is refined, rapidly absorbed carbohydrate versus fibrous, minimally processed carbohydrate.

Kale, beans, whole grains, ugali, greens, and lentils contain carbohydrate, but do not behave like sugary drinks or refined flour. Fiber slows digestion and changes the glucose response. Lieberman describes the communities where he works in Kenya and Rwanda as obtaining at least 65% of their energy from carbohydrates while having little visible overweight or obesity. Their diets are high in fiber and minimally refined, and physical activity is high.

Physical activity changes more than calories expended. Active muscles become more insulin-sensitive, Lieberman says, so the body can produce less insulin for a given amount of carbohydrate. That is one reason exercise may be particularly important for preventing weight gain, even if trying to lose substantial weight through exercise alone is difficult.

His view of paleo follows the same logic. Paleo advice to limit sugar and highly processed food can be sensible. Its categorical bans on legumes, grains, and dairy are less defensible. Lieberman says hunter-gatherers both do and did consume plant foods, including legumes and starches. He notes archaeological evidence of starch on Neanderthal teeth, and points to the marama bean as a major food among hunter-gatherers in the Kalahari.

A review Lieberman conducted with an undergraduate examined 12 hunter-gatherer populations with high-quality dietary data. Their diets varied dramatically: some relied overwhelmingly on plant foods, others heavily on animal foods, with many combinations in between. Humans evolved, in his phrasing, as “the animal kingdom’s ultimate omnivores.” There is no single ancestral macro split to reproduce.

The same point applies to antinutrients. Lectins, phytates, and oxalates can bind minerals and vitamins and reduce absorption. Raw beans can cause real problems. But cooking and fermentation reduce those compounds. For most people, Lieberman says, cooked beans, grains, and spinach are not nutritional threats, although people with particular kidney conditions may need to treat high-oxalate foods differently.

The raw-food diet was the clearest failure in his own self-experimentation. Lieberman and his wife gave it up quickly. He was constantly hungry and preoccupied with food because raw foods delivered too little accessible energy without enormous amounts of eating. His wife stopped after three days; he lasted four.

Cooking is not an artificial departure from ancestral living. It is central to the human evolutionary story. Lieberman says cooking may have begun as long as a million years ago and was extremely common by 400,000 years ago. It breaks down cell walls, releases nutrients, increases digestible energy, and makes food safer. Those changes helped support the large brains, bodies, metabolic demands, and reproductive patterns of modern humans.

Not all processing does the same thing. Cooking and fermentation can make food safe and nutritious. Grinding and blending can make energy easier to absorb. Lieberman says that grinding peanuts into peanut butter makes more energy available than eating them whole, and that blended fruit yields more usable energy than whole fruit. Industrial processing can go further: removing fiber, concentrating calories, and making food unusually easy to consume.

Calorie labels do not fully capture those differences. They estimate energy from carbohydrate, protein, and fat using an old formula intended to describe average availability. They do not precisely measure how much a particular person absorbs, how much energy digestion costs, or what the microbiome contributes. Lieberman’s point is not that calories are imaginary. It is that “calories in” is less precise than the label suggests.

Lieberman identifies processed food as the leading candidate among several interacting causes of rising obesity. He also names physical inactivity, genetic predisposition, antibiotics, and other factors. The concern is that modern food is engineered to be hyper-palatable, convenient, inexpensive, and hard to stop eating.

Humans are adapted to value energy-rich food. Sugar by itself can become cloying, and fat by itself greasy. Mixed in the right proportions, they become especially rewarding. Milk is a rare natural food high in both fat and sugar, Lieberman notes; after weaning, people historically had few comparable foods available without processing.

Food manufacturers can now combine fat, sugar, salt, flavor, temperature, and texture deliberately. Lieberman uses the Big Mac as an example: hot and cold, crunchy and smooth, sweet and salty, with nearly every sensory contrast except spice. Doritos contain a long list of flavoring ingredients balanced so that no one note becomes overwhelming. He says even the multiple colors of M&Ms may encourage continued eating, although the candies taste the same.

These products are not simply accidental by-products of industrial food production. In Lieberman’s telling, they are formulated through testing to keep people reaching for more. That environment interacts with a biology that makes weight loss hard.

Fat cells generally shrink rather than disappear. As people lose fat, hormonal signals including leptin, ghrelin, and other appetite-related peptides can increase hunger and encourage weight regain. Lieberman agrees with Williamson that gaining substantial weight at one point in life can make maintaining a lower weight later more difficult. The body did not evolve to shed stored energy easily.

The situation is not reducible to a failure of personal responsibility. People face time pressure, stress, limited money, long commutes, family demands, and products tuned to exploit hunger and convenience. But individual choices are not irrelevant either. Learning how to choose food is now a necessary skill.

Some foods marketed as wholesome can obscure the choice. Fruit juice is “basically soda,” Lieberman says, because it contains concentrated sugar without much of the fiber in whole fruit. Dried fruit can function as candy when eaten casually in large quantities. Fruit roll-ups can advertise real fruit and vitamin C while amounting largely to fiber-stripped sugar.

Lieberman is more tentative about other products. He worries about meal replacements such as Huel and Soylent because they contain emulsifiers and other ingredients whose long-term effects he says are poorly understood. His own experience trying them was mostly boredom: he missed food and grew to dread the shakes. He does not claim every meal-replacement product is harmful.

He also personally avoids maltodextrin, a sugar-based additive used to keep products smooth and stable. He describes it as a warning sign that a food is ultra-processed and says it can disrupt metabolism. That is one of his personal rules rather than a complete taxonomy of healthy and unhealthy ingredients.

His view of seed oils runs against another popular dietary fear. The concern centers on omega-6 fatty acids, which can enter pathways that produce inflammatory compounds. But Lieberman says controlled studies have not shown that increasing omega-6 intake causes inflammation, while large observational studies associate higher intake with lower risks of heart disease and cancer. On the evidence he describes, omega-6 should not be treated as inherently dangerous.

Calorie restriction has a stronger reputation than its human evidence can yet support, Lieberman argues. It extends lifespan in some mouse lines, as well as in yeast and fruit flies, but he says the evidence is mixed in humans and not decisive even in primates.

He recounts two long macaque studies. In the Wisconsin experiment, calorie-restricted monkeys appeared healthier than controls. But Lieberman says the unrestricted group had been fed a sugary, poor-quality diet and became obese. The National Institutes of Health repeated the experiment with both groups eating the same monkey chow, while one received 30% less food. In Lieberman’s account, that study found no difference in health or longevity.

The comparison does not make calorie restriction biologically implausible. Fasting or lower energy intake can create “metabolic switching.” After eating, the body enters a growth-oriented state associated with mTOR, which supports protein synthesis and growth but can also generate damage, including reactive oxygen species and misfolded proteins. During fasting or energy restriction, AMPK-associated repair processes become more active.

The useful feature may be alternation between growth and repair, not permanent deprivation. Modern eating often looks like constant grazing: snacks are available everywhere, and many people rarely become hungry. Intermittent fasting can create periods of negative energy balance without the continuous hunger of conventional daily restriction.

But Lieberman does not claim that a 16:8 schedule, a 5:2 plan, or one meal a day will make someone live to 100. That promise outruns the evidence. He also does not know of a study directly comparing people who consume the same calorie-restricted intake through constant grazing with people who consume it through one daily meal.

GLP-1 medications are, in Lieberman’s view, genuinely consequential tools. Some people lose as much as 25% of their body weight. The drugs can suppress appetite and cravings that conventional diets struggle to overcome, and that can be transformative for people whose biology strongly defends their existing weight.

The concern is not that they have no use. Rapid weight loss can include substantial muscle loss as well as fat loss. That raises the risk of sarcopenia—loss of muscle mass and function—particularly as people age. Frailty can become self-reinforcing: getting up, shopping, and carrying out ordinary tasks become more difficult; activity falls; weakness worsens.

Lieberman’s practical advice is direct: people using GLP-1s should strength train to help preserve muscle. Losing weight without preserving function is not a complete health outcome.

He also worries about discontinuation. Lieberman cites an estimate that around 60% of people who try GLP-1s stop after a year or two. When treatment ends, he says, appetite can return forcefully and weight can return largely as fat. Someone may lose muscle on the way down and regain fat on the way back up.

That possibility makes GLP-1s closer to a long-term management tool than a temporary reset. They also do not answer the prior question: what should people eat, and how should a society prevent obesity in the first place?

Lieberman calls the broader dynamic “dysevolution”: treating symptoms of a mismatch between evolved appetites and a modern environment while leaving the causes intact. A society can continue producing frozen pizza, fried foods, doughnuts, and other hyper-palatable products; people gain weight; drugs reduce appetite; the food system remains unchanged.

He favors measures that help people make informed choices without removing choice altogether: clearer labels, more nutrition education in schools and medical training, and public discussion of taxes on junk food. Britain’s green, yellow, and red food labels appeal to him because they make a judgment quickly visible. A person can still buy the red-labeled product; they are simply less likely to mistake it for a health food.