Home

The Proof

By Nahum MarkovPublished Updated

This brand makes two claims: eat the food humans are built for, and train hard for twenty minutes a week. This page is where we show our work. The first five categories below test what we say about food. The sixth tests the training.

Each card is one claim, with the figure it rests on and a badge saying how strong the evidence is. Open a card for the full explanation, what we concede, and links to every study, so you can check us yourself.

The Science

What your body can actually absorb, measured.

There is a vitamin you cannot get from plants at all.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Zeroactive vitamin B12 in any plant food

Go deeper

Vitamin B12 does not exist in an active form in any plant food. The B12 in spirulina and similar supplements is largely pseudovitamin B12, which is inactive in humans. Any diet that needs a pill to prevent neurological damage is telling you something about itself.

Sources

The vitamin A in a carrot is not vitamin A yet.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Not yeta plant carries beta-carotene, which your body still has to convert

Go deeper

Plants contain beta-carotene, which your body has to convert. Conversion from real food is far worse than the old textbook assumption, and two common gene variants cut it further. A large share of people are poor converters and have no idea. Liver, egg yolk, butter and cream contain retinol, the finished form, with no conversion step.

Sources

You absorb two to three times more iron from a diet that includes meat.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

14 to 18%iron absorbed on a mixed diet, against 5 to 12% on a vegetarian one

Go deeper

Those are whole-diet figures from the paper the reference values are built on. Phytate, polyphenols and calcium block the plant form of iron, and they are usually in the same meal. Adding meat to a plant meal rescues the minerals in the plant meal. The reverse does not happen.

Sources

Protein quality is not the number on the label.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Grams liewhat counts is the amino acid profile and how much you digest

Go deeper

The measure that matters is the amino acid profile and how much survives digestion. The UN replaced the old scoring method precisely because it flattered plant protein. On the current standard, milk and whey score above the bar and wheat scores lowest, with soy and pea in between. Gram for gram, animal protein also produces a greater muscle-building response.

Sources

Almost every pesticide you eat was made by the plant, not sprayed on it.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

99.99%share of dietary pesticides, by weight, that plants make themselves

Go deeper

A plant cannot run away, so it defends itself chemically: oxalates, lectins, phytates, goitrogens, glycoalkaloids. The scientist who invented the standard laboratory test for chemical mutagenicity calculated that 99.99 percent by weight of the pesticides in the human diet are produced by plants themselves. This is ordinary plant biology. It is just missing from the conversation about what you should eat.

What we concede

Most people handle normal amounts of vegetables perfectly well, and some of these compounds have genuine benefits at low doses. We do not say vegetables are poison. We say they are optional, they are not free of cost, and the only way to know how your body handles them is to remove them for two weeks and watch.

Sources

Mainstream medicine already treats gut trouble by removing plants.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Already standardthe accepted treatment for IBS removes specific plant carbohydrates

Go deeper

The accepted treatment for irritable bowel syndrome is a diet that removes specific plant carbohydrates, demonstrated in a controlled feeding trial. In a separate study, patients with long-term constipation improved when fiber was reduced, in proportion to how much was removed. Almost nobody notices what that implies.

What we concede

The fiber evidence overall is better than we would like it to be: large reviews find consistent associations between higher fiber intake and lower disease risk. That is why our guidance is a personal two-week test rather than a rule for everybody.

Sources

The History

How humans actually ate, for two million years.

Test the bones of ancient humans and they read like a predator.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Like a wolfwhere bone chemistry places Neanderthals on the food chain

Go deeper

Nitrogen isotopes concentrate as you move up a food chain, so bone chemistry places its owner on that chain without any guesswork. Every Neanderthal sample tested across their whole range came back at the top trophic level, alongside wolves and hyenas, taking all or most dietary protein from large herbivores. A later study using single amino acids confirmed it and ruled out the objection that an unusual environment could fake the signal.

Sources

Nearly three quarters of hunter-gatherer societies lived mostly on animals.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

73 of 100societies taking most of their food from animals, out of 229 studied

Go deeper

Researchers went through the ethnographic records of 229 hunter-gatherer societies. 73 percent got more than half their food from animals. Fewer than 14 percent got more than half from plants. The pattern inside the data is the important part: plant intake rose near the equator and fell near the poles, tracking what was available rather than what was preferred. Plants were the fallback.

Sources

The academic case is that we sat at the top of the food chain that whole time.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

2 million yearshow long our genus is argued to have eaten as a top predator

Go deeper

A peer-reviewed synthesis assembled roughly 25 independent lines of evidence, from stomach acidity and fat storage to gut shape, bile acids, tooth wear, isotopes and archaeology. Its conclusion: our genus occupied a hypercarnivorous trophic level for around two million years, with plant food rising only late, as the large animals disappeared. The first tools humans ever made were for cutting meat off bone and cracking bone open for the fat inside.

Sources

In 1928 an explorer ate nothing but meat for a year, under doctors.

Suggestive: A small study, a survey, a historical account, or reasoning from first principles.

One full yearon meat, fat, organs and marrow alone, under hospital supervision

Go deeper

Vilhjalmur Stefansson had lived for years on an all-animal diet in the Arctic and was not believed, so he volunteered for a supervised experiment. He and a colleague ate only meat, fat, organs and marrow for a full year, beginning as hospital inpatients at Bellevue. No scurvy, no deficiency disease, normal kidney function. His own account insists the diet only worked with enough fat, and that lean meat alone made you ill.

What we concede

Two people. It is a landmark case study under medical supervision, not a trial, and we never present it as one.

Sources

When farming replaced hunting, human skeletons got worse.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Shorter and sickerwhat the skeletal record shows across the shift to grain farming

Go deeper

The skeletal record across the shift to grain agriculture shows reduced height, more dental disease and more markers of nutritional stress. Farming won because it fed more people per acre, not because it fed them better. That is one of the most useful facts in this whole library, and it is not controversial among the people who study bones.

Sources

Humans ate plants too, and the Inuit story is weaker than people think.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Honest notetwo places where we correct our own side of the argument

Go deeper

We are not claiming nobody ever ate a plant. There are starch grains on ancient grinding stones and charred tubers in ancient hearths, and the Hadza today eat a great deal of honey and tubers alongside meat. Our claim is about proportion and priority. We have also retired the usual Inuit talking point: the claim that Greenland Inuit had little heart disease does not survive the mortality records, and they carry genetic adaptations to that diet that most people do not.

Sources

The Last 100 Years

Two curves, and they are the same curve.

Adult obesity roughly tripled in sixty years.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

13% to 42%American adults with obesity, early 1960s against today

Go deeper

About 13 percent of American adults were obese in the early 1960s. Today it is around 42 percent. Diagnosed diabetes went from under 1 percent of the population in 1958 to roughly 38 million people, with another 98 million prediabetic. The human genome did not change in sixty years.

What we concede

Diabetes diagnostic criteria changed over that period. The change in criteria does not come close to explaining a curve that shape, but it is real and we say so.

Sources

Only about one adult in eight is metabolically healthy.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

About 12%share of US adults passing all five basic metabolic criteria

Go deeper

Measured against the standard five criteria in a national survey, roughly seven adults in eight fail a basic metabolic screen. That is not a population with some sick people in it. That is a population where health has become the exception, and a heart attack was a case report worth publishing in 1912.

Sources

Given processed food, people ate 500 more calories a day without noticing.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

500 calorieseaten extra per day on processed food, in a sealed metabolic ward

Go deeper

This is the strongest single study in this library, and it is not a correlation. Researchers kept adults in a metabolic ward for four weeks and fed them ultra-processed food against unprocessed food, matched for calories on offer, sugar, fat, fiber and macronutrients. On the processed diet people spontaneously ate about 500 extra calories a day and gained weight. On the real food they lost it. Same people. Same calories available.

Sources

Seed oil consumption rose more than a thousand-fold in one century.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

1,000-foldrise in soybean oil available per person across the 20th century

Go deeper

Before 1911 almost nobody on earth ate extracted seed oil, because making it requires industrial presses and chemical solvent. It is not a food you can make in a kitchen. Estimated soybean oil availability then rose more than a thousand-fold across the twentieth century. When trials finally tested replacing saturated fat with seed oil, they lowered cholesterol and did not save lives; in one recovered trial the intervention group died at a higher rate.

What we concede

The consumption figures come from food disappearance data, so they measure availability rather than what went into mouths. We label them that way.

Sources

They did not just die young. Childhood was lethal; adulthood was long.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

68 to 78years lived by hunter-gatherers who survived childhood

Go deeper

This is the objection everybody raises, and it has a real answer. Historical life expectancy is dragged down almost entirely by infant and child death. Pooling hunter-gatherer demography, the typical lifespan for those who survived childhood runs to roughly 68 to 78 years. They had grandparents. What killed them was infection, injury, childbirth, predation and accident, which is exactly what sanitation, antibiotics and surgery fixed.

Sources

A living non-industrial population has the cleanest arteries ever measured.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Zero calciumarterial calcium in 65% of Tsimane adults over 75

Go deeper

We do not have to infer from bones. Researchers CT-scanned 705 adults of the Tsimane in Bolivia and found the lowest coronary artery disease ever recorded in any population. Among those over 75, 65 percent had zero arterial calcium, against about 5 percent of comparable Americans. Age is not what causes this.

What we concede

The Tsimane eat a high-carbohydrate, low-fat diet and are physically active all day, so this is not evidence for our plate. It is evidence that a population eating non-industrial food and moving their body does not get the disease we treat as an inevitable part of ageing. Separately, CT scans of 137 ancient mummies did find atherosclerosis in four preindustrial populations, so it is not purely modern either.

Sources

The Laws of Nature

Look at the equipment you were given, and work backwards.

A cow needs four stomach chambers to live on plants. You have one.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Four against onestomach chambers a cow needs to live on plants, against your one

Go deeper

A cow holds food for 40 to 70 hours, brings it back up and chews it again, and still cannot digest grass itself. No mammal on earth produces the enzyme that breaks down cellulose. The cow rents it from bacteria and grows them a fermentation tank, then digests the bacteria for protein and lets them make its B vitamins. A horse has a huge hindgut chamber. A rabbit eats its own droppings to run the food through twice. Every animal that lives on plants has enormous, expensive, obvious machinery for it. We have one chamber, a few hours, and none of that.

Sources

Our stomach acid puts us in with the scavengers.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

pH 1.5human stomach acidity, against pH 4 to 5 in herbivores

Go deeper

Researchers measured stomach acidity across 68 mammal and bird species and mapped it against diet. Herbivores sit around pH 4 to 5. Carnivores sit lower. Scavengers, facing the heaviest microbial load, are the most acidic of all, and human stomach pH of about 1.5 sits right there. Acid that strong is metabolically expensive and biologically risky, and nothing evolves it without buying something. It is a sterilizing filter for meat. It is the wrong tool for a leaf.

Sources

Our gut is shaped the opposite way from an ape that eats leaves.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Invertedsmall intestine is 56 to 60% of our gut, the colon 17 to 20%

Go deeper

Across mammals, leaf eaters are colon-dominant and animal eaters are small-intestine-dominant. In a gorilla the colon is the largest part of the gut. In a human, the small intestine is roughly 56 to 60 percent of gut volume and the colon is roughly 17 to 20 percent. The small intestine is where protein and fat are absorbed; the colon is where plant fiber is fermented. Our gut allocates its volume to the first job.

Sources

The human brain was paid for with a smaller digestive system.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Brain for gutacross primates, brain size and gut size trade off against each other

Go deeper

Your brain is about 2 percent of your body and burns about 20 percent of your resting energy. Something had to pay for it. Across primates, brain size and gut size trade off against each other, because both are expensive tissue on a fixed budget. Humans have an unusually large brain and an unusually small gut. A small gut can only be afforded on food that is dense and digestible without long fermentation.

What we concede

Cooking is part of this answer, not just meat, and cooking makes plant food far more digestible too. We think that supports the framework rather than undermining it: the human diet has always required processing that a human does.

Sources

Fruit is the one part of a plant that evolved to be eaten.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

It advertisesfruit colors, sweetens and smells when ripe, and no other part does

Go deeper

A plant cannot spread its own seeds, so it makes a deal: it wraps them in something so attractive an animal will carry them away. Fruit changes color when it is ready, sweetens when it is ready, and smells. Nothing else on a plant does that. Unripe fruit is loaded with bitter defenses that fall away as the seed matures, and the seed itself is either armored or outright toxic. The plant is telling you which part is lunch. A leaf that gets eaten loses; a fruit that gets eaten wins.

What we concede

Modern fruit has been bred for sweetness and size for centuries. A wild apple and a supermarket apple are not the same food. Whole fruit is still the intended package, and it is still a different object from extracted sugar.

Sources

We stopped making our own vitamin C, and the gene is still there.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

A broken genethe gene for the final enzyme is still in our DNA, switched off

Go deeper

Almost every mammal manufactures vitamin C in its liver. We cannot. The gene for the final enzyme is still in our DNA, switched off. The same loss happened independently in other primates and in fruit-eating bats. A lineage can only afford to abandon a metabolic pathway when its diet reliably supplies the product.

What we concede

Fresh animal organs and blood also contain vitamin C, which is how populations eating almost no plants avoided scurvy. That belongs in the same paragraph if we are being honest, and it is.

Sources

Red Meat on Trial

What red meat gives you, and the case against it, study by study.

Measured nutrient by nutrient, animal foods top the list, with organ meat strongest across the board.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

11 of the top 12best food sources of the nutrients diets most commonly lack are animal foods

Go deeper

A 2022 study built a global food composition database and rated foods by how densely they supply six nutrients diets around the world commonly lack, especially in poorer countries: iron, zinc, folate, vitamin A, calcium and vitamin B12. Of the 12 foods it named as top sources, 11 are animal foods, beef among them, with organ meats the strongest across the board. USDA figures show what that means on a plate. 100 grams of pan-fried beef liver holds about 35 times the daily value of vitamin B12, more than 8 times the vitamin A, in its finished form, and 16 times the copper, with about 27 grams of complete protein. An ordinary 100-gram beef patty gives 25 grams of protein, about a full day of B12 and over half a day of zinc, in forms the body absorbs well (see The Science).

What we concede

The one plant food on that list is dark green leafy vegetables, and the scoring covers six nutrients, not every nutrient a body needs. Liver is rich enough to have a ceiling: its vitamin A builds up in the body, so once or twice a week is plenty, not every day. And muscle meat on its own is low in vitamin C and folate. We say the best animal foods, organ meat and red meat among them, are the richest sources of the nutrients people most often lack, not that they are the only foods worth eating.

Sources

Count hot dogs apart from steak, and most of the heart disease signal goes with the hot dogs.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

9% against 18%more heart disease per 50 grams a day, unprocessed red meat against processed

Go deeper

Most headlines about "red meat" pool a steak with hot dogs, bacon and deli slices. When researchers pull them apart, the picture changes. A Harvard meta-analysis of more than a million people found no clear link between unprocessed red meat and heart disease, and a 42 percent higher risk with processed meat, which per serving carried about four times the sodium. A larger 2023 analysis from Oxford found a small link for unprocessed red meat, 9 percent per 50 grams a day, against 18 percent for processed. The PURE study, 134,297 people in 21 countries, found no rise in deaths or heart disease with unprocessed red meat, and 51 percent more deaths at 150 grams or more of processed meat a week.

What we concede

Observational, and for diabetes, chicken shows almost the same link. Diabetes is less kind to us. Across 31 studies in 20 countries and nearly 2 million people, unprocessed red meat carried 10 percent more type 2 diabetes per 100 grams a day. Poultry carried a similar, if less robust, link (8 percent), which makes the cause harder to pin on red meat in particular, but the signal is consistent. And PURE mostly studied people eating little meat by American standards, in countries where more meat often means more money and better healthcare.

Sources

The top red meat eaters in these studies also smoked more, moved less and weighed more.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

3 times the smokersamong men eating the most red meat, against the least, in a 545,000-person study

Go deeper

In the NIH-AARP study, the source of many "red meat kills" headlines, the men eating the most red meat were three times as likely to smoke (14.8 against 4.9 percent) and half as likely to exercise hard. They were heavier, ate half the fruit and were less likely to have finished college. The women showed the same pattern. The study's "red meat" also counted bacon, hot dogs, cold cuts and the meat in pizza and lasagna. Researchers can adjust for the habits they measured, never for the ones they did not. That is the risk of healthy-user bias: for decades, eating less red meat was what health-conscious people did, so the people who ate less of it tended to be healthier in other ways too.

What we concede

Observational, but consistent, rising with dose, and seen in never-smokers. The link held in people who never smoked, it rises steadily with the amount eaten, and it appears in studies on several continents, which confounding alone does not easily explain. The strongest study against unprocessed red meat, from Harvard's long-running studies of nurses and health professionals, found a 13 percent higher death rate per daily serving after adjustment. It is still observational, and its groups show the same pattern: among the men, 5 percent of the lowest red meat eaters smoked, against 14.5 percent of the highest. A steady rise with dose is also what healthy-user bias would produce, since the more red meat people ate, the less health-conscious they tended to be.

Sources

In Britain, vegetarians died at the same rate as meat eaters from the same background.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Same death rateUK vegetarians against meat eaters from the same circles

Go deeper

If meat were the cause, vegetarians should outlive meat eaters even when both come from the same walk of life. Two Oxford studies tested that across 60,310 British adults, recruiting meat eaters from the same circles as the vegetarians, often their friends and relatives. The vegetarians died at the same rate as the regular meat eaters. In Australia, a study of 243,096 adults found no clear difference for vegetarians either. Across eight Asian cohorts, red meat even tracked with lower cardiovascular death in men, though there eating meat signals prosperity, a confounder that runs in our favor, so we weigh it no more than the American studies.

What we concede

Observational: the same death rate, but less heart disease and more stroke in vegetarians. The UK vegetarians had fewer deaths from pancreatic cancer and blood cancers, and people eating a little meat did slightly better than regular meat eaters (about 7 percent, at the edge of statistical significance). The same Oxford cohort found 22 percent less ischemic heart disease in vegetarians, and more stroke. The best-known study the other way, Adventist Health Study-2, found 12 percent lower mortality in its vegetarians, in a community that also avoids smoking and alcohol. And vegetarian against meat eater compares whole diets, not doses of meat.

Sources

Short feeding trials found red meat did not worsen cholesterol, blood pressure or blood sugar.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

24 trialsfeeding people red meat, with no change in cholesterol or blood pressure

Go deeper

Everything so far compared people who chose their own diets. These trials assigned the diet. A meta-analysis of 24 randomized trials compared at least half a serving of red meat a day with less, and found no effect on cholesterol, triglycerides or blood pressure. A second analysis of 24 trials found no effect on blood sugar, insulin or a standard blood marker of inflammation. A review of 36 trials led by Harvard researchers found red meat no different overall from the diets it was compared with.

What we concede

Real trials, but short, measuring markers, and plant proteins did better on cholesterol. These trials ran for weeks to months and measured markers, not heart attacks. Compared with beans, soy or nuts, cholesterol fell less on red meat (against fish, it fell more). The senior author of two of these analyses has received research funding from beef and other food groups, though not for this work. And these trials say nothing about the colon. In a 15-day feeding trial, 21 people eating 420 grams of red meat a day, close to a pound, had more compounds in the gut that can damage DNA, and more of a marker of that damage in colon cells, than on a vegetarian diet. Heme iron is the likely driver, and this kind of evidence is what moved the cancer agency to call red meat probably carcinogenic. These are markers, not outcomes, but they are real. In another trial, a month of red meat more than doubled blood TMAO, a compound linked with heart disease, compared with white meat or no meat, and it dropped sharply within four weeks of stopping. TMAO itself is a shaky yardstick, though: a single fish meal raised it in blood and urine 46 to 62 times more than beef or eggs, and genetic studies have not found that it causes heart disease.

Sources

The two big diet trials that cut red meat saw no drop in heart disease, colon cancer or deaths.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

48,835 womenin the biggest trial that cut red meat: years later, no fewer deaths

Go deeper

In the Women's Health Initiative, women were assigned at random to a low-fat diet, and as part of it they ate about a fifth less red meat. Over eight years there was no change in heart disease, stroke or colorectal cancer, and with longer follow-up, none in total deaths. In the Polyp Prevention Trial, 2,079 people who had precancerous growths removed from the colon ate about 30 percent less red meat as part of a high-fiber, low-fat diet. Four years later the growths had come back at the same rate: 39.7 percent against 39.5. A review of all 12 eligible red meat trials concluded that eating less red meat may make little or no difference to heart disease, cancer or deaths.

What we concede

No trial has ever assigned people to a lot of red meat against very little and followed them for years. These trials cut it modestly and changed other things at the same time, and with so small a cut, neither could detect an effect of the size the observational studies suggest. So the honest wording is that the trials do not show harm, not that they prove safety. That gap runs both ways: the case against red meat has no trial behind it either.

Sources

Taken at face value, the red meat risk per person is small, and the evidence is rated low certainty.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

4 in 1,000fewer cardiovascular deaths over 11 years from 3 fewer servings a week, at best

Go deeper

An international panel graded the observational record (55 cohorts and more than 4 million people for heart and metabolic outcomes, plus a separate review for cancer) with GRADE, the system used to rate evidence for clinical guidelines. Eating three fewer servings of unprocessed red meat a week was associated with about 4 fewer cardiovascular deaths per 1,000 people over 11 years, and 7 fewer cancer deaths per 1,000 over a lifetime, at low to very low certainty. A 2022 review in Nature Medicine rated every unprocessed red meat link weak and found no evidence at all for stroke. The World Health Organization's cancer agency does put processed meat in the same group as tobacco, but that group says how sure the agency is that something can cause cancer, not how much risk it adds. For a daily 50-gram serving of processed meat, that risk is under one percentage point of lifetime colorectal cancer.

What we concede

Pooled observational studies, which the reviews themselves rate weak. Small per person is not nothing across a population, and GRADE starts every observational study at low certainty, so that rating partly reflects the method, not only the data. The 2022 review still found risk at least 6 percent higher for colorectal cancer, 3 percent for breast cancer and 1 percent for heart disease and type 2 diabetes, and its lowest-risk intake was zero (with a very wide range). The panel's advice to keep eating meat as usual was heavily criticized, so we cite its numbers, not its recommendation. Unprocessed red meat sits in the cancer agency's Group 2A, "probably carcinogenic."

Sources

Hunter-gatherers, most of them heavy eaters of animal foods, almost never become obese.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Under 5%obesity among hunter-gatherers, against about 42% of American adults

Go deeper

A review of hunter-gatherer health puts obesity at under 5 percent, and finds heart disease and metabolic disease rare. These are the same kinds of societies that, in the ethnographic record, mostly took more than half their food from animals. The Hadza of Tanzania, among the best measured, average a body mass index of about 20, and in a small sample of 23 adults, cholesterol and other heart markers were in the healthy range. The disease rates we treat as normal are not normal for human beings. They are normal for human beings eating industrial food.

What we concede

Hunter-gatherer diets vary. Some are heavy in meat, and some, like the Tsimane (see The Last 100 Years), eat mostly starch and are just as free of heart disease; a well-known critique of the animal-food estimates makes exactly that point. Some older Hadza do develop high blood pressure, about a fifth to a quarter of those over 40, though far fewer than in the US. The Masai, on meat and milk, had little heart disease in life but real artery disease at autopsy. So we do not claim meat makes people lean, or that it protects arteries. We claim that the diseases we treat as normal are rare wherever industrial food is absent, however much meat is eaten.

Sources

Beef per American is down over a third since 1976, and Americans got sicker anyway.

Suggestive: A small study, a survey, a historical account, or reasoning from first principles.

Down 37%beef available per American since the 1976 peak

Go deeper

USDA records show boneless beef available per person peaked in 1976 at 88.8 pounds a year and fell to 56.2 pounds by 2021. Over the same decades adult obesity climbed to about 42 percent and diabetes became ordinary. If beef were driving those curves, they should have bent the other way. What rose instead was ultra-processed food and seed oil (see The Last 100 Years).

What we concede

These are availability figures, not measured intake. Chicken more than doubled over the same period, so total meat went up, not down. A trend line proves nothing by itself, which is exactly the rule we hold the other side to. What it rules out is the simple story that Americans got sick by eating more red meat. It cannot test the real case against red meat, a modest added risk of colorectal cancer and diabetes, which is far too small to see in a national trend.

Sources

The Training

Why twenty minutes a week is the dose, not a compromise.

The largest analysis found the lowest death rate at under an hour a week.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

30 to 60 minstrength training a week, where all-cause mortality is lowest

Go deeper

A systematic review and meta-analysis of strength training and mortality found 10 to 17 percent lower risk of all-cause mortality, heart disease, cancer and diabetes, with the lowest risk at roughly 30 to 60 minutes of muscle-strengthening activity per week and no established further benefit beyond about an hour. Our method prescribes 20 minutes of maximum effort a week. That is not below the evidence. It is the efficient end of the window the evidence points at.

Sources

One minute of hard work matched fifty minutes of cycling.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

60 secondsall-out effort a session, matching 50 minutes of moderate cycling

Go deeper

Over twelve weeks, a randomized trial had one group perform three 20-second all-out sprints, one minute of hard work inside a ten-minute session, three times a week. They matched a group doing 50 minutes of continuous moderate cycling on peak oxygen uptake, insulin sensitivity and muscle mitochondrial content, at one fifth the training volume. An earlier study found the same thing in two weeks.

Sources

A few minutes of vigorous daily movement, measured by device, tracked far lower mortality.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

3 to 4 min a dayvigorous movement, measured in 25,000 adults who never exercised

Go deeper

Over 25,000 adults who did no formal exercise wore activity monitors. As little as three to four minutes a day of vigorous movement, the kind produced by stairs or carrying shopping, was associated with substantially lower all-cause, cardiovascular and cancer mortality, in a near-linear dose-response. The measurement was a device rather than a questionnaire, which is a major strength. Separately, 15 minutes a day of moderate activity tracked three extra years of life expectancy, with each further 15 minutes adding very little.

Sources

For building muscle, how hard you go matters and how heavy does not.

Trial: A controlled trial, a direct measurement, or a mechanism nobody disputes.

Effort, not weightgrowth is similar across loads when sets are taken close to failure

Go deeper

A meta-analysis found that muscle growth is similar across a wide range of loads when sets are taken close to failure. Maximum strength still favours heavy weight, as you would expect, but growth does not care. This is the whole permission structure for bodyweight training: a push-up taken to genuine failure is doing the same job as a bench press taken to genuine failure. In a direct comparison, progressive push-up training matched bench press training for upper-body strength.

Sources

Researchers have already described our method in a review.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Twice a weeklow-volume, high-effort, full-body training, in a published review

Go deeper

A review written by the people who ran many of the underlying trials set out how to train for strength and size in the least time possible. Their conclusions support low-volume, high-effort, multi-joint, full-body training performed twice a week. Two sessions per muscle group per week is also consistently at or near optimal for growth in the wider literature. And once fitness is built, it is maintained on a small fraction of the training that built it.

Sources

More is not better, which is why we cap it.

Strong: A large study, a meta-analysis, or several independent lines of evidence agreeing.

Rest is the pointmuscle protein synthesis stays elevated 24 to 48 hours after a session

Go deeper

Muscle protein synthesis stays elevated for roughly 24 to 48 hours after a hard session and then returns to normal. Growth happens in that window, not during the session. Training the same tissue again before it has finished adding the adaptation interrupts the thing you paid for. The session is the request. The recovery is the delivery. Two requests a week, properly made, with time to fill them.

What we concede

Competitive athletes train for hours, and for them that is a rational trade for the last few percent of performance. We are not claiming 20 minutes makes you an athlete. We are claiming it buys nearly all of the health.

Sources

What we are not claiming

This matters as much as the rest. A page that only holds supporting evidence is a brochure, not a proof.

  • We are not claiming this is the only way anyone has ever got healthy. People have got healthy on many different diets, and the thing most of them have in common is that they removed processed food.
  • We are not claiming vegetables are poison. We are claiming they are optional, that they are not free of cost, and that the honest way to find out how your own body handles them is to remove them for two weeks and watch.
  • We are not hiding the evidence that runs against us. Saturated fat does raise LDL cholesterol. The fiber data is better than we would like. The healthy-user problem we raise against the red meat studies applies just as much to the exercise studies in The Training, one of which pools the same NIH-AARP study. The strongest objections are linked on this page, next to the claims they challenge.
  • We are not claiming red meat has been proven safe by a long-term trial. That trial has never been run, in either direction. Nor is the case against unprocessed red meat empty: its links with colorectal cancer and type 2 diabetes are small but consistent across observational studies, and the colorectal link has a plausible mechanism. We read that risk as small and uncertain, not zero, and Red Meat on Trial says exactly where the evidence stops.
  • We are not doctors. Nothing here diagnoses or treats anything, and if your bloodwork moves, that is a conversation for your physician, not for us.

Nobody argues with an experiment they ran on themselves. Two ten-minute sessions a week for a month is eighty minutes. If nothing changes, you have lost eighty minutes.

Outcomes belong to God. The effort is your part.

Apply for coaching

This page is education, not medical advice. Read the Health Disclaimer.