What the body actually eats

Nutrition is the only domain in human health where two qualified experts can tell you opposite things on the same Tuesday, and both can cite peer-reviewed research, and both can be partly right.

Cardiology can be confused, but it agrees on roughly what a heart attack is. Sleep medicine argues at the edges, but everyone agrees you need sleep. Nutrition is different. The field carries a century of bad incentives, food-industry money, badly designed observational studies, a culture of dietary tribalism, and a small number of unusually good experiments that get drowned out by everything else.

The reader of nutrition content ends up trying to triangulate between people who appear equally credentialed and who say incompatible things. After enough years of this, most people quietly give up and eat what their friends eat.

This piece is an attempt to do something different. There is a small set of things that almost every reasonable nutrition researcher agrees on, including ones from camps that fight about everything else. And there is a much larger set where the evidence is contested, weak, or has been corrupted by industry. The job is to act confidently on the agreed set and stay honestly uncertain on the rest.

This is part six of a series. The first five pieces were about the nervous system: stress, breath, the upward spiral, sleep, and a starter protocol. Food sits in the same place those do. It's not a moral category. It's an input to a biological system that has preferences. The body votes through digestion, sleep, mood, energy, and a long list of biomarkers. You can listen, or you can let the wellness industry shout into the silence.

Three things this piece will do:

1. Name the consensus. The small set of things that virtually no one serious disagrees with.
2. Map the contested middle. Inflammation, fats, fasting, the gut, alcohol. Where the evidence is strong, where it isn't, and where industry has gotten in the way.
3. Build the playbook. Concrete defaults you can run for the rest of your life without needing to read a nutrition article again.

There is a small set of things almost every nutrition expert agrees on. There is a much larger set where they don't. Operate confidently from the first set. Stay honestly uncertain about the second.

Section 1What almost everyone agrees on

The actual consensus, stripped of marketing, is short enough to fit on the back of a napkin. Vegans, keto people, Mediterranean researchers, and most cardiologists agree on almost all of it. If you do only the things on this list, you are eating better than 90% of the people around you.

That's the napkin. Notice what isn't on it. Not "avoid carbs," not "eat animal protein only," not "intermittent fast," not "drink celery juice." Those are positions people hold, sometimes with evidence in narrow contexts, but they are not consensus and the certainty with which they are sold should be a yellow flag.

The rest of this piece is mostly about the contested middle. But the agreed set is where the leverage is. If your eating is dominated by whole foods, with adequate protein and fiber, low ultra-processed intake, water as the default drink, and a clear-eyed relationship to alcohol, you have already won the part of the game that's winnable. The rest is marginal.

Section 2Ultra-processed food is the big one

If there is one finding from the last two decades of nutrition research that deserves to be widely known, it is this one.

In 2019, Kevin Hall and colleagues at the NIH published a small but unusually rigorous study in Cell Metabolism. They took 20 healthy adults, admitted them to a metabolic ward, and fed them two diets for two weeks each in random order. One was matched on calories, sugar, fat, fiber, and macronutrients to the other. The only difference was processing: one diet was ultra-processed, the other was unprocessed. Participants could eat as much as they wanted from each.

On the ultra-processed diet, they ate roughly 500 more calories per day than on the unprocessed diet, and gained nearly a kilogram in two weeks. On the unprocessed diet, they lost about a kilogram. Same person, same caloric availability, same macros, same fiber. Different physiology. Different outcome.

This is one of the cleanest causal demonstrations in modern nutrition. It is not an observational study, not a survey, not subject to the usual confounders. It is a randomized controlled feeding trial in a metabolic ward, and it shows that ultra-processing alone, independent of calories and macronutrients, causes overeating and weight gain.

The Nova classification

The framework the study uses comes from Carlos Monteiro's group in Brazil. The Nova classification sorts foods into four categories:

The useful distinction in real life: "processed" alone is a bad framing. Olive oil is processed. So is cheese. So is canned tuna. None of that is the problem. The problem is the Nova 4 category, where industrial reformulation produces foods that are hyperpalatable, calorie-dense, low in satiety per calorie, and engineered for repeat consumption.

Why ultra-processed food works on the body the way it does

Several mechanisms running in parallel, none of them controversial:

• Hyperpalatability. Industrial formulations combine fat, salt, sugar, and texture in ratios that do not occur in nature. Almost nothing in the natural world is simultaneously very sweet and very fatty. Ice cream is. Donuts are. The reward system responds disproportionately.
• Low satiety per calorie. Whole foods activate stretch receptors, fiber-derived gut hormones (GLP-1, PYY, CCK), and slow digestion. Ultra-processed foods are typically energy-dense and low in fiber, so the same calories produce less satiety signal.
• Speed of eating. Hall's group later showed that participants ate ultra-processed meals faster, which gave the body less time to register satiety before the next bite. The eating-rate effect is large.
• Reward learning. The brain learns which foods produce the largest reward signal and prioritizes them. Ultra-processed foods, by design, hijack this learning.
• Additive effects. Emulsifiers and certain other additives have effects on the gut barrier and microbiome in animal models, and increasingly in humans. The evidence here is real but less mature than the calorie-and-reward story.

How to use this in real life

The Nova 4 category is not a rule. It is a slider. Reducing your share of calories from ultra-processed foods, even partially, is one of the highest-leverage dietary moves available. Going from 60% of calories (the modern Western average) to 30% is meaningful. Going to 10-15% is excellent.

Two practical heuristics work better than reading every label:

• The ingredient test. If the ingredient list contains items you would never use cooking at home (lecithin, dextrose, maltodextrin, modified starch, glucose-fructose syrup, mono- and diglycerides, artificial flavors, color codes), it's probably Nova 4.
• The shape test. If the food doesn't look like its source ingredients (a chicken nugget vs. a chicken breast, a fruit gummy vs. a fruit), it's probably Nova 4.

You don't need to eliminate ultra-processed food entirely. You need to know which category most of your calories come from. If the answer is "Nova 1 mostly, with some Nova 3, and occasional Nova 4," you're operating from a solid floor.

Section 3Inflammation, fats, and a hundred years of bad maps

Most of the contested terrain in nutrition is about fats. Saturated vs. unsaturated, animal vs. plant, omega-3 vs. omega-6, seed oils vs. olive oil. Each of these has a confident camp and a counter-camp. The honest read on the evidence is more boring than either side admits.

The saturated fat story, briefly

In the 1950s, Ancel Keys published the Seven Countries Study and argued that saturated fat raised cholesterol, and cholesterol caused heart disease. The full picture was more complicated than that, and Keys made choices about which countries to include that have been criticized since. But the policy machine moved on the simple version. By the late 1970s, U.S. dietary guidelines told people to cut saturated fat. The food industry replaced animal fats with vegetable oils, refined carbohydrates, and sugar. Heart disease did not drop.

From the 2000s onward, several large meta-analyses and observational studies (including the PURE study, which followed 135,000 people across 18 countries) found that the association between saturated fat and cardiovascular mortality was much weaker than the original story claimed, and that the food substitutions made in the name of cutting saturated fat were probably worse than the saturated fat itself.

A 2020 JACC state-of-the-art review by a group of researchers including Ronald Krauss concluded that the evidence does not support blanket recommendations to limit saturated fat. Whole-food sources of saturated fat (whole-fat dairy, eggs, fatty meat, dark chocolate) are not clearly associated with cardiovascular events. Refined carbohydrate substitutes for those foods are worse.

This is not the same as "saturated fat is good for you." A 2024 European Society of Cardiology study found that high-saturated-fat diets raise liver fat and LDL cholesterol within weeks even without weight gain. There is still a dose-response at high intakes, and people who eat enormous amounts of saturated fat (the carnivore-diet extreme) probably do incur some cardiovascular cost.

The honest summary:

• Eating saturated fat from whole foods, as part of an otherwise reasonable diet, is probably fine for most people.
• Eating very large amounts of saturated fat, especially while also having elevated LDL, is probably not optimal.
• The "saturated fat causes heart disease" simplification was wrong, and the corrective swing to "saturated fat is fine in any amount" is also wrong.
• Most reasonable adults can stop worrying about the saturated fat in eggs, whole-fat dairy, and unprocessed meat, and instead worry about Nova 4 intake, which matters more.

Omega-3 to omega-6 ratio: what's actually known

The story you have read on the internet: modern diets are way too high in omega-6 (industrial seed oils) relative to omega-3 (fish), the ratio used to be 1:1 and is now 15:1, and this drives inflammation and chronic disease.

The actual evidence is messier. The original ratio framing, popularized by Artemis Simopoulos in the 1990s, was based on the idea that linoleic acid (the main omega-6 in seed oils) converts to arachidonic acid and drives pro-inflammatory eicosanoids. More recent research has shown that, in adults consuming Western diets, increased linoleic acid intake does not increase tissue arachidonic acid much, and does not reliably raise inflammatory markers in controlled studies. A 2025 study examining red blood cell membrane levels of these fatty acids found no clear association with most inflammation biomarkers.

The current consensus is shifting. Many serious lipid researchers now argue that the ratio itself is a less useful frame than the absolute levels of omega-3. The thing that matters is that most people are too low on long-chain omega-3 (EPA and DHA, from fatty fish), independent of how much omega-6 they eat.

The seed oil panic that has become loud online overstates the direct inflammation evidence. The kernel of truth is that most omega-6 in modern diets is consumed inside ultra-processed foods (Section 2), and those foods have many problems, of which the oil is one component. The cleaner intervention is to eat less Nova 4, which automatically lowers seed oil intake along with everything else, rather than to fixate on the oil itself.

The Mediterranean diet evidence (and the PREDIMED retraction)

The single largest randomized trial of a whole dietary pattern is the PREDIMED study, published in NEJM in 2013. About 7,400 high-risk adults in Spain were randomized to one of three diets: Mediterranean plus extra-virgin olive oil, Mediterranean plus nuts, or a low-fat control. The Mediterranean arms showed roughly a 30% reduction in major cardiovascular events over five years.

In 2018, the original paper was retracted after investigators found that some participants had been assigned to diets by household or clinic rather than truly individual randomization, which is a real methodological issue. The same authors immediately republished a reanalyzed version that handled the clustering properly. The conclusions did not change. Mediterranean diets with olive oil or nuts still reduced cardiovascular events by approximately the same magnitude. The retraction-and-republication looks bad in headline form and is honestly a model of how science is supposed to self-correct.

The takeaway: the Mediterranean pattern has the strongest randomized evidence of any dietary pattern for cardiovascular outcomes. Olive oil and nuts are part of why. So is what it lacks (low Nova 4 share, modest red meat, low refined grain).

Section 4The microbiome eats what you eat

The third piece in this series ("The upward spiral") made a point that's worth re-stating here. The vagus nerve is mostly afferent. Eighty percent of the fibers carry signals from the body to the brain, not the other way around. The largest single source of that signal is the gut, and the gut signal is partly determined by what lives in it.

You have roughly as many bacterial cells in your gut as human cells in your body, give or take depending on which count you use. They have their own metabolism. They produce short-chain fatty acids, neurotransmitter precursors, immune signals, and metabolic byproducts that interact with your physiology continuously. The composition of that community is shaped, on a time scale of days to weeks, by what you eat.

What actually shifts the microbiome

The two highest-leverage inputs, both supported by reasonably strong evidence:

1. Fiber diversity. Different bacteria eat different fibers. A diet centered on the same three plants produces a less diverse microbiome than a diet that rotates through thirty. The "thirty plants a week" target, popularized by Tim Spector and the American Gut Project, has actual data behind it. People who hit it have measurably more diverse gut communities and better digestive and inflammatory markers.
2. Fermented foods. A 2021 Stanford study by Justin and Erica Sonnenburg, published in Cell, gave 36 healthy adults either a high-fiber diet or a fermented-foods diet for 10 weeks. The fermented-foods group ate yogurt, kefir, kimchi, sauerkraut, kombucha, and other fermented vegetables daily. After 10 weeks, the fermented group showed increased microbial diversity and significant reductions in four different inflammatory cytokines, including interleukin-6. The high-fiber group did not show the same diversity increase in the same time window (longer-term studies suggest fiber does shift the microbiome, just slower).

The fermented foods finding was a surprise. The standing assumption was that fiber would dominate. It didn't, at least not on a 10-week scale. The mechanism is partly the live microbes in the food and partly the metabolic byproducts of fermentation. Both probably contribute.

What doesn't shift the microbiome much

• Most probiotic supplements. The marketing is far ahead of the evidence. Most over-the-counter probiotic capsules contain a small number of strains, in doses that don't reliably colonize the gut, and the bacteria they do contain are often dead on arrival. There are specific clinical contexts (post-antibiotic, IBS subtypes, C. diff prevention) where targeted probiotic strains have real evidence. As a general health practice, they are mostly a way to spend money.
• "Cleanses" and detoxes. These do nothing for the microbiome other than temporarily disrupt it, which is the opposite of what you want. Your liver and kidneys handle detoxification continuously. The wellness-industry version is theater.
• One-time interventions. The microbiome responds to consistent inputs over weeks. A single week of perfect eating won't do much; a year of mediocre-but-diverse eating will.

The fiber question

Fiber is the single most under-consumed component of modern diets. The average adult in industrialized countries gets 15 grams per day. The likely-optimal range is 25-40 grams per day. Hitting that range, from a variety of sources, is one of the more consistent predictors of cardiometabolic health across the literature.

Fiber matters for three reasons, beyond microbiome diversity:

• Satiety. Fiber slows digestion, activates stretch receptors, and triggers gut hormones (GLP-1, PYY, CCK) that reduce appetite. Fiber is the original GLP-1 agonist, just slower.
• Blood sugar. Fiber blunts glucose spikes after meals. This matters across the metabolic spectrum (Section 5).
• Short-chain fatty acids. Bacteria ferment fiber into butyrate, propionate, and acetate, which feed colon cells, regulate inflammation, and have systemic metabolic effects.

Practical sources: legumes (very high), oats, barley, whole grains, berries, apples with skin, broccoli, artichokes, avocados, nuts, seeds. Fiber supplements (psyllium, inulin) work in a pinch but a varied whole-food approach feeds a wider range of bacteria.

Section 5Blood sugar, timing, and the CGM hype

Continuous glucose monitors (CGMs) are everywhere now. The marketing message is that everyone should be obsessing over their post-meal glucose response, that "spikes" are damaging, and that you should be eating to keep your glucose flat. The evidence does not particularly support this for people without diabetes.

What's true

• For people with diabetes or prediabetes, glucose control is genuinely important and CGMs are a real clinical tool.
• For everyone, large repeated glucose spikes (from sugar-sweetened beverages, refined carbs without fiber or protein) over years probably contribute to insulin resistance.
• Postprandial glucose responses vary enormously by individual, even for the same meal. This is real and was documented by Eran Segal's Weizmann team in 2015. Genetics, microbiome, sleep, and stress all change how you respond to a given meal.
• Stable glucose tends to come with stable energy, less afternoon crash, and easier appetite regulation, partly through mechanisms separate from glucose itself.

What's overclaimed

• Healthy non-diabetics have glucose spikes after meals. That is normal physiology. A 130-150 mg/dL spike after eating fruit, bread, or rice is not pathological.
• The CGM industry markets continuous monitoring as a wellness tool. The evidence that this changes outcomes in healthy people is thin. It mostly creates anxiety and a new food-policing behavior.
• "Glucose hacking" advice (drink vinegar before meals, eat vegetables first, walk after eating) has real but small effects. Useful at the margins, not a meaningful health intervention compared to the basics.

The summary: if you have diabetes or prediabetes, listen to your endocrinologist. If you don't, you probably don't need a CGM, but you can use the general principle. Eating refined carbs alone produces the biggest spikes. Eating the same carbs with protein, fat, and fiber produces smaller, more useful spikes. You don't need a sensor to apply this; just eat real food with mixed macros.

Time-restricted eating: the actual evidence

Intermittent fasting and time-restricted eating (TRE) are some of the most-marketed wellness practices of the last decade. The evidence is more modest than the marketing.

The largest current synthesis is a 2025 BMJ network meta-analysis of 56 randomized trials. The summary: intermittent fasting in various forms reduces body weight, blood pressure, LDL, and fasting glucose. But when compared directly to calorie restriction without fasting, the differences are small to nil. Modified alternate-day fasting comes out as the most effective format, but most of the weight-loss effect appears to be calorie restriction in different clothing.

The honest read on TRE specifically:

• An 8-10 hour eating window per day is a reasonable default and is sustainable for most people without much effort.
• It often produces modest weight loss, mostly through unintentional calorie reduction.
• Eating earlier in the day (an 8 AM to 4 PM window) tends to be better than later (12 PM to 8 PM) for blood sugar and cardiometabolic markers, because human metabolism is itself circadian. Insulin sensitivity is higher in the morning. This is consistent with the sleep-and-rhythm story (see "The master reset").
• The "skip breakfast, eat late" pattern works for some people but is not the optimal version of TRE physiologically.
• Long fasts (24+ hours, multi-day) have stronger autophagy and metabolic effects but also stronger costs (muscle loss, sleep disruption, cortisol increase). The cost-benefit is unclear for most people outside specific clinical contexts.

Late-night eating

The cleanest near-universal timing finding: eating large meals close to bedtime is metabolically worse than eating the same meal earlier. Late eating raises core temperature (which needs to drop for sleep), triggers digestive cortisol at the wrong time, blunts the next morning's insulin sensitivity, and reduces deep sleep that night. Cross-reference "The master reset" Section 5 for the sleep-side mechanism.

The practical heuristic: finish your last large meal three hours before bed when possible. Light snacks are fine. A full dinner at 10 PM, eaten before an 11 PM bedtime, is one of the more easily fixable bad habits.

Most "blood sugar" advice is just a re-packaging of "eat whole food, mostly plants, finish dinner before late." You can buy a sensor to learn this, or you can just do it.

Section 6Protein, the under-consumed nutrient

This is the section where the consensus has shifted most clearly over the last decade, and where most people are still operating from old information.

The RDA is too low for optimal

The official Recommended Daily Allowance for protein is 0.8 grams per kilogram of body weight per day. This number was set decades ago, with the goal of preventing deficiency in sedentary adults. It is the minimum below which nitrogen balance becomes negative for most people. It is not, and was never intended to be, an optimization target.

Current consensus from researchers who actually study protein metabolism (Stuart Phillips at McMaster, the PROT-AGE group, the International Society of Sports Nutrition) puts the optimal range higher:

• Healthy active adults: 1.2-1.6 g/kg per day.
• Adults doing resistance training: 1.6-2.0 g/kg per day.
• Older adults (65+): 1.0-1.5 g/kg per day, higher with illness or recovery. Older muscle is more anabolically resistant; it needs more protein to produce the same muscle protein synthesis response.
• Weight-loss contexts: 1.6-2.4 g/kg per day, to preserve lean mass while in a calorie deficit.

For a 70 kg adult, that's roughly 85-110 grams per day for an active person, 110-140 grams for someone training, and 70-105 grams for an older adult. Most adults eating typical modern diets are at the low end of this range or below it.

Why protein matters more than most macros

• Muscle. Adequate protein, especially combined with resistance training, is the primary input to preserving muscle mass through life. Muscle loss (sarcopenia) accelerates after age 40 and is one of the strongest predictors of disability and mortality in old age. You don't get to skip this part.
• Satiety. Protein is the most satiating macronutrient per calorie. People who eat sufficient protein eat less of everything else without effort. The "protein leverage hypothesis" (Raubenheimer and Simpson) says that humans regulate intake partly to hit a protein target, and if protein is diluted (Nova 4 foods are usually protein-low), we overeat total calories trying to compensate.
• Blood sugar. Adding protein to a meal blunts the glucose response, smooths post-meal energy, and reduces the rebound hunger from refined-carb meals.
• Recovery. Any kind of physical training (resistance, endurance, mobility) requires adequate protein for tissue repair.

The leucine threshold and meal distribution

For muscle protein synthesis to be activated, a meal needs to provide a minimum dose of leucine, an essential amino acid that triggers the mTOR pathway. The threshold is roughly 2.5-3 grams of leucine per meal, which corresponds to roughly 25-40 grams of high-quality protein in a single sitting.

The practical implication: protein distribution across the day matters. 100 grams of protein eaten as 20-30-30-20 across four meals stimulates muscle protein synthesis four times. The same 100 grams eaten as 10-10-80 (small breakfast and lunch, huge dinner) stimulates it once, with most of the dinner protein over the ceiling and used for other purposes. Older adults benefit even more from per-meal targets because they need slightly higher leucine doses to overcome anabolic resistance.

You don't need to obsess over this. A reasonable target is 25-40 grams of protein per meal, three or four meals per day, ideally with one of those meals near a training session.

Plant vs animal protein, honestly

The protein quality debate has two real components and a lot of tribal noise.

Per-gram quality is genuinely different. Animal proteins (whey, eggs, meat, dairy) tend to be "complete" (contain all essential amino acids in good ratios) and have higher digestibility (DIAAS scores). Plant proteins tend to be lower in one or more essential amino acids (rice is low in lysine, beans are low in methionine) and slightly less well absorbed.

Practically, this matters less than it sounds. Mixing plant proteins through the day (rice plus beans, lentils plus grains, tofu plus most things) covers the amino acid profile. You just need slightly more total grams of plant protein to get the same effect, roughly 10-20% more. Vegan athletes do fine on 1.6-2.0 g/kg with mixed sources; the literature has repeatedly shown this.

The honest summary:

• Animal protein is the most efficient per gram. If you eat animal foods, eggs, fish, poultry, and dairy hit the target with minimal effort.
• Plant protein works fine if total intake is sufficient and sources are varied. Legumes are the cornerstone, with grains, nuts, seeds, and soy as supporting actors.
• Whey protein is the most-studied supplemental protein and works well as a convenience source when whole foods are inconvenient.
• The "you need to eat meat to get enough protein" claim is wrong. The "plant protein is just as easy to get enough of" claim is also slightly wrong, in the sense that it requires more attention. Both can work.

The category-defining mistake most adults make is not "wrong type of protein," it's "not enough total protein, period."

Section 7The honest alcohol section

This section will not tell you to quit. It will tell you what alcohol actually does, so that if you choose to drink, you are choosing knowingly.

The "no safe level" claim and what it means

In 2018, The Lancet published a Global Burden of Disease analysis concluding that "the safest level of drinking is none." The headline was widely reported. The claim was based on a meta-analysis of mostly observational studies, and it was specifically about all-cause mortality and disease burden across populations.

The claim is approximately right at the population level. The "J-curve" (light drinkers having lower mortality than non-drinkers) that older epidemiology had identified was substantially confounded by the "sick-quitter" problem (people who stop drinking due to health problems get coded as non-drinkers, dragging up the non-drinker mortality rate). Better-designed analyses, including 2023-2024 reanalyses that adjusted for these confounders, mostly eliminated the J-curve. Light drinking is not protective. The protective signal in older studies was a methodological artifact.

This is not the same as "one drink is going to kill you." The dose-response is real. The cost of one or two drinks a week is small. The cost of one or two drinks a day is much larger. And the cost of binge drinking is its own category, with effects on cardiovascular risk, accidents, and cancer that are not seen with the same total weekly intake spread out.

What alcohol actually does to the body

Worth naming explicitly, because most people drink without naming any of it:

• Sleep destruction. Alcohol is sedating on the way in and stimulating on the way out. It selectively destroys REM sleep, the part of the night that integrates emotional content from the day. Two drinks reduces total REM by ~25%. Cross-reference "The master reset" for the full mechanism.
• HRV reduction. Even one drink measurably lowers next-day heart rate variability. Two or three drinks is unmistakable. Habitual drinkers often don't notice because their baseline has shifted; if you stop drinking for two weeks and then track HRV, the number often jumps 15-25%. Cross-reference "Your body decided before you did" Section 7 for HRV measurement.
• Liver fat accumulation. The earliest pathology, often present in regular drinkers before any other sign. Reversible if drinking stops.
• Cancer risk. Alcohol is a Group 1 carcinogen (same category as tobacco and asbestos). The cancers most clearly linked are breast, colorectal, esophageal, head and neck, and liver. The dose-response is mostly linear, not flat-with-a-threshold.
• Anxiety the next day. The "hangxiety" effect. Alcohol modulates GABA in the short term; when it wears off, the rebound is sympathetic activation, often felt as anxiety. People with anxiety patterns often drink to manage them and then have worse anxiety the next day.
• Calories. Pure alcohol is 7 kcal/gram, between carbohydrate (4) and fat (9), and contributes nothing else. Most alcoholic drinks add sugar on top. A nightly habit of two drinks is 200-300 daily calories of essentially empty intake.

The other side, honestly

Alcohol is also one of the most reliable social lubricants humans have. It lowers inhibition, deepens conversation, marks ritual occasions, and produces genuine positive experience for many people. The pleasure is real. The connection it can facilitate is real. The pretense that there's nothing being gained when you drink is dishonest in the other direction.

The honest position is therefore not "alcohol is bad, quit." The honest position is "alcohol has real costs that scale with dose, and there are no health benefits that outweigh them, so any drinking you do is for the social or pleasure value, and worth pricing accordingly."

Section 8A short list of supplements worth taking

The supplement industry is roughly $150 billion a year, and roughly 95% of what it sells is unnecessary for most people. The few exceptions are worth knowing.

The rule of thumb: a supplement is worth taking if (a) the evidence for it is solid, (b) you are likely to be insufficient on it from diet alone, and (c) the cost-benefit is favorable. Below is the short list that clears all three bars for most adults.

Everything else, briefly

Most things that aren't on the short list above are not worth taking for most people:

• Multivitamins: trivial benefit in well-fed adults. The VITAL trial did show a small cognitive aging benefit from a daily multivitamin in older adults (the COSMOS-Mind sub-study, 2023), which is the closest thing to a defensible reason to take one. Cheap, low downside if you do.
• Probiotics: see Section 4. Skip in favor of fermented foods.
• Greens powders: roughly the price of vegetables, without most of what vegetables actually provide. Skip.
• Collagen, BCAAs, glutamine, etc.: almost all of the supposed effects are achieved more reliably by hitting your overall protein target with whole food or whey.
• Adaptogens, nootropics, "stack" supplements: almost none have evidence holding up under rigorous trial. The placebo effect is real but expensive.
• Iron, B12, calcium, zinc, etc.: only if you have a documented deficiency or specific risk factor (heavy menstruation, vegan diet for B12, certain medications). Iron supplementation without need is actively harmful.

Section 9The practical playbook

This is the part you can run for the rest of your life without needing another nutrition article. Like the playbook in "The master reset," it is structured by leverage. Do the first parts consistently and the later parts are mostly small refinements.

Section 10Eat as if the body were listening

Here's what's worth remembering.

The body is not a calorimeter and you are not a chemical equation. You are an organism with a microbiome, a nervous system, a circadian rhythm, and a long evolutionary history of eating whole foods in social settings, at predictable times, with people you knew. The closer your eating gets to that pattern, the easier almost everything downstream becomes. The further from it, the more work you have to do to compensate, and the less of that work actually compensates.

The nutrition industry, including most of its honest parts, has spent fifty years trying to reduce eating to nutrients. Protein, carbs, fats, omega-3s, glycemic index, polyphenols, methylated this and acetylated that. Each frame produces useful information in narrow contexts. None of them produce a complete picture, and the people who insist on one frame loudest tend to be selling something.

The complete picture, as far as it can be drawn from the current evidence, is closer to common sense than the wellness industry would like you to believe. Eat whole foods. Eat enough protein. Eat lots of plants, varied through the week. Include fermented foods often. Avoid ultra-processed foods most of the time. Drink water. Be honest about alcohol. Take a small number of well-evidenced supplements if you're plausibly deficient. Cook your meals. Eat with people when you can.

The body, given those inputs, mostly takes care of itself. Not perfectly, not magically, but well enough that most chronic-disease risk that's modifiable through diet gets modified. The marginal returns above this floor are real but small. The marginal returns below this floor are large and most of modern medicine spends its energy patching them.

You don't need to optimize anything exotic. You need to do the boring obvious thing consistently, which almost no one in modern life does. The same conclusion as the previous pieces in this series, applied to a different system. The body is listening, the way it always has been. Feed it like that's true.

Most of what matters about food fits on a napkin.
The wellness industry charges by the gigabyte.