Not as a cause, not as a protest, and not because anybody asked me to. I was sixty-six. I was heavy, exhausted, in constant pain, and doing exactly what I had been told to do. One man, one change, nobody watching and nobody informed.

It worked. That is the part I could not have known in advance, and the part nobody could have told me with enough authority to be worth believing. I had to find out in the only laboratory available to me.

Then I wrote about it, because by then I had something to report rather than a theory to float.

And what comes back now is mostly not questions. It is people telling me what happened when they changed something.

A man with type 1 diabetes who was taking more than a hundred units of insulin a day, now under forty. A woman who had spent most of her adult life with three overlapping autoimmune conditions, whose cardiologist cut her medications in half. A couple in their late seventies, five years in, seventy pounds each, off everything they had been taking.

None of that was solicited and none of it is mine. It arrives because something worked and the person wanted somebody to know. It is also a great deal harder to stay at the desk once people start reporting back, which is why I have lately been doing more than writing.

Notice the order, because it is the whole argument of this post and I did not plan it that way. One person acted. The result came back. Then it was worth saying out loud. Then other people acted, and reported. Nothing about that sequence could have started at the other end.

And when a question does come, it is some version of the same one: what am I supposed to do about it?

Here is the honest version of my answer, including the parts that do not flatter it.

It begins with you, and with nobody else.

Your first responsibility in this is to yourself. Not to a movement, not to a boycott, not to me. Before any of this becomes a public matter it is a private one — a question of what you are going to eat tomorrow morning, decided by you, for your own reasons, in your own kitchen, with no audience.

That is not modesty about the larger argument. It is the order the thing actually works in, and there are two reasons for it.

The first reason is that it is the only place the evidence works

Everything else that might fix this operates on a horizon of years and delivers a benefit nobody can feel.

I filed a comment on a federal rulemaking last week. If that rule is finalized it will not take effect before 2029, and if it works perfectly no individual will ever know it helped them. Industrial trans fats left the American food supply by federal determination in 2015, with a compliance deadline in 2018 — a real improvement affecting everyone who eats, and not one person felt it happen.

Changing what you eat is the only intervention in this entire subject that returns a verified result to the person who made it, in their own body, inside a year.

That is not a small advantage. It is the only place in this problem where the measurement actually works.

I have written here before that a cause present in nearly everybody cannot be found by studying variation across a population, because there is no unexposed group to compare against. That is true, and it is why the population-level evidence in this field will stay thin no matter how long anyone waits for it. But the same fact that makes the population unmeasurable makes the individual measurable. A person who changes one variable becomes their own control. You get a before and an after in the same body, and nobody has to run a trial to show it to you.

So the evidence you cannot get out of the literature, you can get out of yourself. In months. Without trusting me, or a guideline, or anyone with something to sell.

The second reason is that nothing else is available to you until you have done it

Nobody persuades anybody of something they have not done themselves. Not because it would be hypocritical — because it would be unconvincing, and because you would not know what you were talking about.

It is also why this travels when nothing else does. A person who got a result does not need to be convinced of anything, and tends to tell somebody. That is how every account at the top of this page reached me, none of them in response to a request. It is how the whole thing reached me five years ago, at a dinner where nobody was selling anything. And it is the only reason I am any use to you now — not because I read about it, but because I did it first and it happened to work.

What it is worth, in numbers

Forty million Americans have diabetes — about one in eight of us. A hundred and fifteen million adults have prediabetes, which is more than two in five. The annual cost of diagnosed diabetes runs around $640 billion, and in 2021 the condition accounted for roughly a quarter of all health care spending in the United States.

One condition. A quarter of everything we spend on being sick.

That is not a line item. It is prescriptions, clinic visits, test strips, specialist referrals, procedures, hospital stays, amputations and dialysis chairs. It is an enormous industry, and every dollar in it is downstream of something that was eaten.

Whether changing the input changes the output is answerable by a trial rather than by an opinion, and a good one exists.

It is called DiRECT — the Diabetes Remission Clinical Trial — run by Roy Taylor at Newcastle University and Mike Lean at Glasgow, and published in the Lancet in December 2017. Diabetes UK paid for it. That is a charity rather than a company, and it put more than £2.8 million into the question.

The trial enrolled 306 adults who already had type 2 diabetes and ran them through forty-nine ordinary primary care practices across Scotland and northern England — not a research unit, but the kind of surgery you would actually walk into. Half continued with their usual care. The other half went onto a managed weight-loss program, and — this is the part that matters — every one of their oral diabetes drugs was stopped on day one.

At twelve months, seventy-four percent of them were taking no diabetes medication at all. In the control group the figure was eighteen percent, and eight of those people started medication during the year.

Remission was defined strictly: blood sugar back in the normal range after at least two months off all diabetes drugs. Forty-six percent of the intervention group met it. Four percent of the controls did.

It was open-label, a quarter of participants dropped out, and it was one trial. I am not going to oversell it. But it is a randomized comparison in which three out of four people walked away from their medication and nearly half no longer had the disease, and nothing whatsoever was added to achieve it.

That is what a successful outcome looks like, and it is not measured on a stock ticker. It is measured in prescriptions that never get written and appointments nobody needs.

So here is the thing to stop

Not a brand and not a company. A category, and the category is refined carbohydrate.

Sugar, and the starches that have had their structure stripped out so that they arrive in the bloodstream at a speed nothing was built to handle. It has to go, or be cut hard, because the rate of arrival is the entire mechanism and there is no version of this that survives leaving it in place.

I want to be exact, because exactness here is what makes it actionable, not what softens it. The refined forms go — all the way out, no remaining quantity that counts as acceptable. Milling and refining take the brakes off. They perform the work of digestion in advance, in a factory, so that what arrives in you is glucose with nothing left to slow it down. There is no responsible serving size of that, and the industry’s entire defense rests on persuading you there is one.

The rest of the carbohydrate — the potato with its skin on, the fruit, the bowl of rice — is a different event in a body, and it is still a load. It does not get a pass. It gets cut hard, and how hard is the thing you are going to have to determine on yourself, because nobody has determined it for you. That is the whole structure: the refined goes, the remainder comes down sharply, and you find out where your own floor is.

Stop buying refined carbohydrate. Not one item off the shelf. Not most of it. The category.

I am not going to dress that up as something gradual, because it was not gradual for me and it was not gradual in the trial above, where every oral diabetes drug came out on day one. The rate of arrival is the mechanism. A smaller dose of the mechanism is still the mechanism, and a half-measure produces a half-answer you will not be able to read.

So it comes out of the cart. The cereal, the bread, the soda, the juice, the pasta, the crackers, the thing in the cupboard you reach for without deciding to. All of it, this week. Then find out what happens in your own body over the next few months, which is the only evidence in this entire subject that will ever actually belong to you.

One caution, and it is not a hedge. If you take medication for diabetes or blood pressure, taking the carbohydrate out while the dose stays where it is can put you too low. That is precisely why DiRECT managed the drugs and the food together, on day one, with a physician watching. Do it the same way. The change is the point — but making it without telling the person who writes your prescriptions is not boldness, it is just a worse experiment.

That is the whole of what I am asking anybody to do, and it is deliberately the smallest version. One person, one change, in one kitchen, with nobody watching. It is also the only part of this that returns an answer you can actually read.

What happens when enough people do it is a different argument, and a longer one. It involves a volume number that has already stopped growing, an industry that cannot be abandoned the way people assume, and four occasions on which somebody paid for the appearance of evidence. That is Saturday.

Five years ago, at a battery conference, I sat down to a dinner that changed what I ate for the rest of my life. That story is a book and it is not this post. And on the eleventh I turned seventy-two, which commemorates an event I had no part in arranging and cannot take credit for. My mother did the work. I have simply kept going, which for a while there was not the obvious outcome.

Then on Wednesday I wrote here that an entire aisle of products had been engineered to hit a number. That is a hard thing to say about somebody else’s business without saying something about my own.

So here is my own. There is no graceful way to volunteer your finances to several hundred people on a Saturday morning, and I am not going to attempt one.

A few weeks ago I gave you three questions for telling careful writing apart from merely confident writing, and said to use them on this page too. There is a fourth I left out, because it is the one you cannot answer from the outside. Where does the writer’s money come from, and what have they turned down?

The first half you can check in about a minute. The second half you cannot check at all, because refusals leave no trace. Nobody can see the offers I have declined, and no amount of reading me carefully will reveal them. So I am telling you now, while nobody has asked, rather than later when somebody does.

I sell books. That is the complete list.

I am aware that this is not a thrilling disclosure. It is the whole of it.

I want to be plain about it, because the usual version is dishonest in both directions — people who perform standing outside of business, and people who are in six businesses and mention one.

I am in business. I write books, I would like you to buy them, and I am not going to act reluctant about it. A book is a long piece of work — the reading, the drafting, the outside review, the revisions nobody sees — and it is the thing I actually make. Selling my own work, at a price anyone can decide about in a second, strikes me as a clean transaction.

Here is what I will not do:

— No supplements, ingredients, powders or products of any kind, mine or anybody else’s.

— No affiliate links, referral codes or revenue share.

— No sponsorship, paid placement or brand partnership, here or anywhere I appear.

— No coaching, program, course or certification.

Those four are permanent. If any of them ever change you will read it here first, stated plainly, before you see it anywhere else.

I am aware of how that list reads three days after a post about an aisle full of exactly those things. It is supposed to. An argument about what other people sell is worth very little from someone who has not said what he sells.

About this page specifically

I spent a career building things that ship, and I think about this publication the way I would think about any product after release. The books are the release. This page is the continuous improvement — corrections when I have gotten something wrong, extensions when new evidence arrives, the reasoning carried forward as the field moves under it. That work is already implied in the purchase. You bought the argument. Maintaining the argument comes with it.

So the reasoning here is free, and it stays free. Not as generosity. An argument you have to pay for cannot be checked by the people it concerns, and the checking is the entire point of how I write.

I will be honest that I do not know what this becomes. The work is about to be more public than it has been. If the advocacy grows into something substantially larger than maintaining what I have already published — original investigation, sustained effort that is not downstream of a book anyone bought — then that is a different product, and a different product may have to support itself. I am not going to promise you it never will.

If that day ever comes I would rather know in advance than guess. Substack has a way for a reader to signal that they would support the work. It is called a pledge. It charges nothing and it changes nothing about what is on this page, so what I have effectively done is install a button that takes no money and does not work, in order to find out whether anyone presses it. If nobody does, that is a useful answer and I will have lost nothing but a small amount of dignity. If people do, I will tell you what I decide to do about it before I do it.

What I will promise is the floor. The argument itself, and every correction to it, stays where you can read it without paying. Whatever else happens, that does not move.

One word about what more public actually means, since I raised it and then left it hanging. This week I filed a comment on a federal rulemaking — the first time I have put my name on the record in a proceeding rather than only writing about one. It took an afternoon. It cost nothing. The docket stays open to anyone until the ninth of December, and I wrote about it here separately (see recent SS Note post).

I expect there will be more of that. Not a campaign you are asked to join, and not instructions — notice of where the openings are, when I find them, with enough detail to act on if you want to. What you do with the notice is entirely your own business, and I would rather it stayed that way. An argument that needs followers is usually not much of an argument.

Why any of this matters

The moment my income depends on your believing something particular about what you should eat, you lose any way of telling whether I believe it myself. That is a real loss and it is not recoverable.

That dinner cost nobody anything and sold me nothing. I have thought about it a great deal since — that the most valuable thing I have ever been given about my own health was the one thing nobody was trying to sell me.

That is the standard I am trying to hold here, and it is the only reason any of this is worth reading.

The books are the long version of the argument. The reasoning is here whether you buy one or not.

That is worth stopping on, because it is the fastest evidence I know that something real happened. Enough people changed how they ate, for long enough, with results good enough to keep going, that a retail category formed around them. Nobody builds an aisle for a fashion. Aisles get built for volume that has already shown up and stayed.

So let me say the first part plainly, without hedging it: cutting the refined carbohydrate worked, and it worked for a reason you can write down.

What it answered

Refined starch and sugar arrive in the bloodstream faster than anything the body spent its evolutionary history metabolizing. Milling and refining do to grain what chewing was supposed to do slowly and incompletely — they strip the structure that made digestion take time. What is left is glucose with the brakes removed. The regulatory system that handles it was tuned for a delivery rate it no longer receives, and running it constantly at the top of its range for decades is not a neutral thing to do.

Take the refined carbohydrate out and that load comes off. It comes off immediately, it comes off measurably, and for a great many people the downstream consequences come off with it.

I know this from the inside and I know it from the people who write to me. Weight gone. Pain gone. Bloodwork moved. More than one person has told me their physician cut a medication in half while watching it happen. That is a correct answer to a real question, arrived at by people who were given no help finding it, and I am not here to complicate it.

It is one question.

The questions it does not touch

Everything on this list is true of a person in ketosis and a person eating bread, because none of it is about carbohydrate.

The minerals that came out in the milling. Removing the bran and the germ removes roughly eighty percent of the magnesium with them. The 1941 enrichment standard put five things back into white flour — thiamin, riboflavin, niacin, iron, folic acid — and magnesium was not among them. The Rothamsted archive has kept wheat samples since 1843; mineral content held steady for a century and then fell from the 1960s, while the soil minerals did not. Cutting carbohydrate removes the flour. It does not put back what the flour was supposed to be carrying.

The sodium and potassium inversion. The kidney was tuned by natural selection on a diet with under a thousand milligrams of sodium a day and considerably more potassium than that. So it hoards sodium and sheds potassium, which is exactly backwards for how anyone eats now. Both inputs got inverted. Ketosis does not touch the ratio. What touches the ratio is whether your food still has its cell contents in it.

What went into the food rather than what came out of it. Under the self-affirmed GRAS pathway, a company could conclude on its own that a substance was generally recognized as safe and begin using it, with no requirement to tell the FDA it had done so. Nobody outside the company was obliged to look at the evidence. The FDA proposed ending that in August and the comment period closes on 9 December. Your macros have no opinion about any of this.

Fat composition. The ratio of fats in the modern supply is not the ratio anything was selected on, and a diet can be very low in carbohydrate while being built almost entirely from industrially extracted oils.

Four questions. Carbohydrate load is a fifth. Answering the fifth correctly is genuinely worth doing, and it leaves four standing.

And then the aisle

Here is where it turns, and the turn is not a moral story.

Once enough people were asking for low carbohydrate, the food industry did the thing it is extremely good at: it formulated a product that hit the number. Pick up anything in that section and read the ingredient panel rather than the front of the package. Soluble corn fiber. Glycerin. Sugar alcohols. Soy protein isolate. Fiber syrups with names that take a second to pronounce. Natural flavors.

That is a manufactured object, engineered to land on a macro target while staying shelf-stable and palatable. It is the same category of thing as what it replaced, built by the same industry, under the same rules, and sold to people who changed how they ate specifically to get away from manufactured objects.

And it could be built, because the specification you handed the industry was a number.

What happens to a number

Isomalto-oligosaccharide is the cleanest example I know, and it has a paper trail.

IMO is made from hydrolyzed corn or tapioca starch and it was widely used in high-fiber and low-carbohydrate bars. Counted as fiber, it gets subtracted from the carbohydrate total, which is what produces a low net-carb figure on the front of a wrapper.

In a randomized crossover trial, fifty grams of IMO and fifty grams of dextrose produced no meaningful difference in blood glucose, insulin, or breath hydrogen. Fifty grams of table sugar’s closest relative, and fifty grams of the fiber, behaved the same way in the body. Earlier work had put IMO’s digestible energy at seventy to eighty percent of sugar’s.

In 2018 the FDA declined to include IMO in its list of recognized dietary fibers and denied the manufacturer’s petition, concluding the evidence did not demonstrate a beneficial physiological effect. One of the largest bar makers reformulated after litigation over its nutrition claims. Plenty of products never did.

Note what did and did not happen. Nobody in this story committed fraud as far as I can tell. A legal ingredient was used in a legal way under a labeling term — net carbohydrate — that the FDA does not define or regulate at all. The number on the front of the package was arithmetic performed on a category the agency had not yet ruled on.

This has already happened once, at national scale

In the 1980s the country was told to cut fat. People did. They did what they were told, which is the part everyone forgets when they are being condescending about it.

The industry reformulated to hit the number. Fat came out and sugar went in, because something has to carry the palatability, and the specification said nothing about what replaced what. The low-fat cookie met the standard exactly. We spent two decades finding out what it cost.

The mechanism was not that people were weak or that anyone lied. The mechanism was that a single measurable number is trivially gameable, and the thing removed in service of the number is never part of the specification. Give the industry one target and it will hit the target. That is not a flaw in the industry. That is what the industry is for.

Low fat was the number in 1985. Whole grain was the number later, which is how we got the same product with some whole grain flour added to it. Net carbohydrate is the number now. The machinery has not changed and there is no reason to expect a different outcome from it.

What I am not saying

I am not saying the aisle is poison. I cannot demonstrate harm from any particular product in it, and I have been clear for four posts running about why that kind of harm is so hard to measure — an exposure that reaches nearly everybody leaves no unexposed group to compare against, so the study that would settle it cannot be run.

I am not running a purity test. Some packaged things are fine. A bar in a glovebox on a long drive is not a moral failure and I have eaten my share.

And I am not telling anyone who cut their carbohydrate and got their life back that they were wrong. They were right. That is the whole point of writing this down.

The test that actually survives

What I am saying is that a macro target is not a definition of food, and if it is the only specification you carry into a store, the store will meet it and keep everything else exactly as it was.

The question that holds up better is older and duller. Did this exist as a thing before somebody formulated it? Could you, in principle, point at where it came from?

That question cannot be gamed, because it is not a number. It cannot be met by substituting one ingredient for another with a better label. It takes no arithmetic and no trust in a front-of-package claim, and it happens to answer all five questions at once rather than the one.

You already did the hard part. You changed what you ate, against the advice, and you held it long enough to find out. The aisle exists because of you.

Do not let it become the reason you stopped looking.

Cutting the carbohydrate answered one question correctly. There were several — and an industry has already formulated a product for the one you answered.

There is a keto section in my grocery store. There was not one when I started.

That is worth stopping on, because it is the fastest evidence I know that something real happened. Enough people changed how they ate, for long enough, with results good enough to keep going, that a retail category formed around them. Nobody builds an aisle for a fashion. Aisles get built for volume that has already shown up and stayed.

So let me say the first part plainly, without hedging it: cutting the refined carbohydrate worked, and it worked for a reason you can write down.

What it answered

Refined starch and sugar arrive in the bloodstream faster than anything the body spent its evolutionary history metabolizing. Milling and refining do to grain what chewing was supposed to do slowly and incompletely — they strip the structure that made digestion take time. What is left is glucose with the brakes removed. The regulatory system that handles it was tuned for a delivery rate it no longer receives, and running it constantly at the top of its range for decades is not a neutral thing to do.

Take the refined carbohydrate out and that load comes off. It comes off immediately, it comes off measurably, and for a great many people the downstream consequences come off with it.

I know this from the inside and I know it from the people who write to me. Weight gone. Pain gone. Bloodwork moved. More than one person has told me their physician cut a medication in half while watching it happen. That is a correct answer to a real question, arrived at by people who were given no help finding it, and I am not here to complicate it.

It is one question.

The questions it does not touch

Everything on this list is true of a person in ketosis and a person eating bread, because none of it is about carbohydrate.

The minerals that came out in the milling. Removing the bran and the germ removes roughly eighty percent of the magnesium with them. The 1941 enrichment standard put five things back into white flour — thiamin, riboflavin, niacin, iron, folic acid — and magnesium was not among them. The Rothamsted archive has kept wheat samples since 1843; mineral content held steady for a century and then fell from the 1960s, while the soil minerals did not. Cutting carbohydrate removes the flour. It does not put back what the flour was supposed to be carrying.

The sodium and potassium inversion. The kidney was tuned by natural selection on a diet with under a thousand milligrams of sodium a day and considerably more potassium than that. So it hoards sodium and sheds potassium, which is exactly backwards for how anyone eats now. Both inputs got inverted. Ketosis does not touch the ratio. What touches the ratio is whether your food still has its cell contents in it.

What went into the food rather than what came out of it. Under the self-affirmed GRAS pathway, a company could conclude on its own that a substance was generally recognized as safe and begin using it, with no requirement to tell the FDA it had done so. Nobody outside the company was obliged to look at the evidence. The FDA proposed ending that in August and the comment period closes on 9 December. Your macros have no opinion about any of this.

Fat composition. The ratio of fats in the modern supply is not the ratio anything was selected on, and a diet can be very low in carbohydrate while being built almost entirely from industrially extracted oils.

Four questions. Carbohydrate load is a fifth. Answering the fifth correctly is genuinely worth doing, and it leaves four standing.

And then the aisle

Here is where it turns, and the turn is not a moral story.

Once enough people were asking for low carbohydrate, the food industry did the thing it is extremely good at: it formulated a product that hit the number. Pick up anything in that section and read the ingredient panel rather than the front of the package. Soluble corn fiber. Glycerin. Sugar alcohols. Soy protein isolate. Fiber syrups with names that take a second to pronounce. Natural flavors.

That is a manufactured object, engineered to land on a macro target while staying shelf-stable and palatable. It is the same category of thing as what it replaced, built by the same industry, under the same rules, and sold to people who changed how they ate specifically to get away from manufactured objects.

And it could be built, because the specification you handed the industry was a number.

What happens to a number

Isomalto-oligosaccharide is the cleanest example I know, and it has a paper trail.

IMO is made from hydrolyzed corn or tapioca starch and it was widely used in high-fiber and low-carbohydrate bars. Counted as fiber, it gets subtracted from the carbohydrate total, which is what produces a low net-carb figure on the front of a wrapper.

In a randomized crossover trial, fifty grams of IMO and fifty grams of dextrose produced no meaningful difference in blood glucose, insulin, or breath hydrogen. Fifty grams of table sugar’s closest relative, and fifty grams of the fiber, behaved the same way in the body. Earlier work had put IMO’s digestible energy at seventy to eighty percent of sugar’s.

In 2018 the FDA declined to include IMO in its list of recognized dietary fibers and denied the manufacturer’s petition, concluding the evidence did not demonstrate a beneficial physiological effect. One of the largest bar makers reformulated after litigation over its nutrition claims. Plenty of products never did.

Note what did and did not happen. Nobody in this story committed fraud as far as I can tell. A legal ingredient was used in a legal way under a labeling term — net carbohydrate — that the FDA does not define or regulate at all. The number on the front of the package was arithmetic performed on a category the agency had not yet ruled on.

This has already happened once, at national scale

In the 1980s the country was told to cut fat. People did. They did what they were told, which is the part everyone forgets when they are being condescending about it.

The industry reformulated to hit the number. Fat came out and sugar went in, because something has to carry the palatability, and the specification said nothing about what replaced what. The low-fat cookie met the standard exactly. We spent two decades finding out what it cost.

The mechanism was not that people were weak or that anyone lied. The mechanism was that a single measurable number is trivially gameable, and the thing removed in service of the number is never part of the specification. Give the industry one target and it will hit the target. That is not a flaw in the industry. That is what the industry is for.

Low fat was the number in 1985. Whole grain was the number later, which is how we got the same product with some whole grain flour added to it. Net carbohydrate is the number now. The machinery has not changed and there is no reason to expect a different outcome from it.

What I am not saying

I am not saying the aisle is poison. I cannot demonstrate harm from any particular product in it, and I have been clear for four posts running about why that kind of harm is so hard to measure — an exposure that reaches nearly everybody leaves no unexposed group to compare against, so the study that would settle it cannot be run.

I am not running a purity test. Some packaged things are fine. A bar in a glovebox on a long drive is not a moral failure and I have eaten my share.

And I am not telling anyone who cut their carbohydrate and got their life back that they were wrong. They were right. That is the whole point of writing this down.

The test that actually survives

What I am saying is that a macro target is not a definition of food, and if it is the only specification you carry into a store, the store will meet it and keep everything else exactly as it was.

The question that holds up better is older and duller. Did this exist as a thing before somebody formulated it? Could you, in principle, point at where it came from?

That question cannot be gamed, because it is not a number. It cannot be met by substituting one ingredient for another with a better label. It takes no arithmetic and no trust in a front-of-package claim, and it happens to answer all five questions at once rather than the one.

You already did the hard part. You changed what you ate, against the advice, and you held it long enough to find out. The aisle exists because of you.

Do not let it become the reason you stopped looking.

Which leaves the practical question I had been putting off, and which I said at the start of the magnesium series this investigation was partly meant to settle. If you cannot measure it, what do you actually do?

What I did about it, which is not a recommendation

You cannot measure your way out of a problem like this, so some years ago I stopped trying to.

What I did instead was an engineering move rather than a nutritional one. When you cannot instrument a system, you reduce the number of inputs until you can at least reason about it. So I eat a narrow diet. Carbohydrate near zero, almost nothing from the middle of a grocery store, most nights the same thing cooked the same way.

I am not claiming that is optimal. I have no way to know that and neither does anyone telling you otherwise. The claim is much smaller: whatever is wrong with eating that way is probably wrong in ways that have been wrong for a very long time, rather than in ways invented in the last hundred years. I traded a large number of unmeasured modern variables for a small number of old ones.

That is not health. It is variable reduction. They are not the same thing and I try not to confuse them.

And then I went and did the exact opposite

I take magnesium. I take a B complex. I take a multivitamin. I put a potassium-blended salt in what I drink.

That sits badly against everything in the paragraph above, and when this series started I said that settling exactly this question was one of the things it was for. That was a stated outcome, not an afterthought. So here is the accounting.

Every one of them traces back to the same thing this piece is about — something that was in the input our regulatory systems were naturally selected against, and is not in the input now. I will take them in order of how well I can name the removal.

The potassium-blended salt. The ancestral input was high in potassium and low in sodium. Plants are potassium-rich; it is the dominant cation inside every leaf and tuber. Sodium was genuinely scarce inland — scarce enough that salt functioned as currency and trade routes were built to carry it. So the kidney was selected to hoard sodium and dump potassium, and it is very good at both.

We have inverted both inputs inside a century. Sodium is in everything manufactured. Potassium is in the foods people stopped eating. I am not adding something novel to my body — I am putting the ratio back toward the one the organ was built around. That is also the only one of the four carrying a real downside: potassium is not something to take casually, the kidney’s ability to clear it declines with age, and anyone with reduced renal function has no business doing this without a physician.

Magnesium. It is the central ion of chlorophyll. For several million years it arrived continuously, inside every green thing anybody ate, with no gaps to design around — which is precisely why the body never evolved a reserve for it.

Then it was removed twice over. Breeding for yield diluted what was in the grain. Milling took out roughly eighty percent of what remained, with the bran and the germ. The 1941 standard put five things back and this was not among them. And then I removed the rest myself: I took the plant foods out of my own diet and did not replace them with anything that carries it, because almost nothing else does.

The B complex, and really only part of it. Meat covers most of this group well — B12 abundantly, along with niacin, riboflavin and B6. I am not short of those and I will not pretend otherwise.

The ones that worry me are the ones that came from greens and organ meats in the ancestral input. Folate is named for foliage. It was first isolated from spinach. The modern delivery route for it is enriched flour, which I do not eat, and the ancestral route was leaves, which I also do not eat. Both channels closed, for different reasons, and I closed the second one myself.

The multivitamin, where the removal is real but cannot be itemized. This is the one I could not justify before and I think I had the reasoning backwards.

The ancestral input was not a list of named nutrients. It was a very large number of compounds arriving from a wide variety of plants and animals eaten whole, most of which nobody has catalogued, because the catalogue was never made. Modern food delivers a narrower set of them. My own diet is narrower still, deliberately.

So I cannot tell you which trace element I am missing. Not because there is no removal behind it, but because the thing that was removed was never written down — which is the entire subject of this piece, arriving at my own plate. The multivitamin is a hedge against a subtraction I can demonstrate in general and cannot enumerate in particular.

I will say what that costs, honestly. It is a weaker kind of justification than the other three — a decision made on the shape of the risks rather than on a named mechanism, with no harm I can find at ordinary amounts and an expense too small to weigh. That is a legitimate way to decide under uncertainty and it is not the same thing as evidence. It is also, almost word for word, how the entire supplement trade sells. The only things separating me from the aisle are that I will tell you which kind of reasoning I am using, and that I drew the line at anything with a genuine downside — which is why the potassium gets treated differently from the rest.

And the part of it I watch: the fat-soluble vitamins, and any iron, are the components that accumulate rather than clear. The body has no regulated route for getting rid of iron at all — that was the point of the third post in this series. No harm at ordinary amounts is a statement about ordinary amounts, and it stops being true somewhere I will not locate by feel.

Which is where careful reasoning actually runs out

Here is the part that matters, and it is not that I was sloppy.

Reasoning your way to a plausible deficit is not the same as having evidence that filling it does anything. I can give you a mechanism for three of those four. I cannot give you a single measurement showing that taking them has changed anything in my body — because the measurement does not exist. That is the subject of this entire piece.

So the honest position is this. Careful reasoning got me to a defensible guess and then stopped, because there was nothing left to check it against. I am not improvising. I am reasoning as well as the available information permits, and the available information runs out well before the decision does.

That is worse than carelessness, not better. A careless person can be corrected by evidence. I have done the work and arrived at the same unverifiable place, and so will you, and so will your doctor, because the instrument that would settle it was never built.

The one thing I can say for my four is that nobody sold them to me and I am not selling them to you. But that is a statement about my motives, not about my evidence. On the evidence I am in the same position as everybody else in the supplement aisle — better reasons, identical certainty, which is to say none.

So here is the verdict, since I said I would give one

The question I put to myself at the start of this was whether to keep taking any of it. The answer is that I am continuing all four, and I am putting the reasoning on the record so it can be held against me later.

The potassium salt and the magnesium I will defend in front of anybody. The B complex follows from a gap that is specific and real, created by how I eat rather than by anything general. The multivitamin I keep on a different basis entirely, and I would rather say so than have it absorbed quietly into the other three.

That basis is an asymmetry — nothing I can find on the downside at ordinary amounts, against a gap I cannot rule out, at a cost not worth the arithmetic. It is a real way to decide something when evidence is unavailable. It is also, almost word for word, the reasoning the entire supplement trade is built on. The only things separating me from the aisle are that I will tell you which of the two kinds of justification I am using, and that I drew the line at anything carrying a genuine downside — which is exactly why the potassium is treated differently from the rest.

Here is what would end the rest of it, stated now so that it is not something I get to reinterpret later:

A validated measure showing no shortfall, and I stop. Evidence of harm at these amounts, and I stop. The multivitamin goes the day the asymmetry stops holding — the moment somebody shows a cost on the other side of it, the only argument I had for it is gone. And the potassium goes first and goes immediately at any sign of kidney trouble, because it is the one of the four with a real capacity to hurt me.

One more commitment, and it is the one I think matters most. The list does not grow. Every addition past this point would require a better reason than the ones I already have, and I do not have better reasons available, because nobody does. A stack that only ever gets longer is not a considered position. It is a habit wearing the clothes of one.

I will add that my own eating is well down the list of things I find worth worrying about. What concerns me is not what one man at a desk in Florida has for dinner. It is that two hundred million people are making this same call, with less information than I had, against a baseline nobody wrote down.

And if something does change for you, pass it on.

Not as advice — as a case report. Say what you changed. Say what else you held still, which is the part almost nobody says. Say what you measured and what you only felt, and keep the two separate. Say what you would have expected to see if you were wrong.

That is the whole difference between evidence and a story, and it costs you nothing but the care. In the absence of anybody running the study, it is the only record that is ever going to exist.

I could document the removal in detail. Breeding for yield diluted the mineral content of the grain. Milling takes out roughly eighty percent of the magnesium with the bran and the germ. The 1941 enrichment standard put five things back into white flour and magnesium was not one of them. The Rothamsted archive, which has kept wheat samples since 1843, shows the mineral content holding steady for a century and then falling from the 1960s, while the soil itself did not change.

Every link in that is on the record. None of it is disputed.

And at the end of four posts I could not tell you whether any of it has harmed you. Not because I did not look. Because the measurement does not exist, and it does not exist for a reason that applies to far more than magnesium.

We left the reference condition and did not write it down

For most of the time there have been humans, people ate whatever the landscape produced. I am not going to claim that diet was optimal — I do not know that and neither does anybody else. What I will claim is narrower and harder to argue with: it was the condition the body’s regulatory systems were calibrated against. Every absorption mechanism, every storage decision, every hormonal setpoint was tuned by natural selection to that input, because there was no other input to be tuned to.

Then, inside roughly a century, we replaced it. Not gradually and not by accident — deliberately, industrially, for shelf life and yield and cost.

Nobody measured the old condition before it went. There was no reason to. You do not take a baseline of a thing you have no idea is about to change, and in 1900 nobody was running serum panels on people eating stone-ground flour in order to preserve a record for us.

So the reference is gone, and there is no copy of it.

Which means every number you are given is measured against the wrong zero

This is the part that took me four posts to see properly, and it is worse than it sounds.

When a laboratory tells you a value is normal, normal means typical of the population they sampled. That population eats the modern food supply. All of it does. So the reference range is a description of people already inside the condition — it cannot tell you whether the condition itself is a problem, any more than measuring the average temperature in a burning building tells you the building is not on fire.

The same applies to how requirements get set. A recommended intake is derived largely from studying people who are already eating this way and observing what keeps them from visible illness. That produces a number that is honest about what it measured and silent about what it did not.

None of this is fraud. Every one of those numbers was produced carefully by people doing the job properly. The problem is structural and it sits underneath all of them: we are estimating what is normal for a human using a sample in which no human is living under the conditions the species was built for.

Geoffrey Rose said this in 1985 and it is still the most important sentence in the field

A cause that is universally present cannot be detected by studying variation within a population.

If everyone is exposed, there is no unexposed group. The comparison that would reveal the effect does not exist to be run. You can study why one person in the exposed population fares worse than another — and the field does, endlessly — but you cannot study the exposure itself, because it has no contrast.

That is why the magnesium question has no answer. Not because the question is silly, and not because somebody is hiding something. Because everybody eats from the same food supply, and the experiment that would settle it would require a group of people who do not.

The dangerous part is what unmeasurability does to accountability

Here is the whole thing in one line: a change you cannot measure is a change nobody has to answer for.

In the twentieth century we fixed a specific set of nutritional deficiencies by national mandate — iodine, vitamin D, thiamin, niacin, riboflavin, iron, folate. Every one of them produced a visible, nameable disease you could photograph. Committees acted, and they acted quickly, and they were right to.

We left another set alone. Magnesium, zinc, potassium, copper, manganese. Every one of those produces only a shift in a distribution — no syndrome, no photograph, no name.

The rule that sorted them was not harm. It was visibility. And that rule is still running, right now, deciding which problems get programs and which get nothing, with no reference at all to how much damage is actually being done.

So the shortfalls that are measurable got fixed, and the shortfalls that are not stayed exactly where they were, and will, because there is no instrument that would make anyone accountable for them and nobody is funding one.

The reason any of this exists

I should say plainly why I started looking, because it is not an academic interest and it would be dishonest to present it as one.

At sixty-six I was obese and failing. Fatty liver disease. Arthritis. Chronic fatigue. Mental fog. Back pain I had carried since high school. Daily pills and painkillers. And I had not been careless about any of it — I had been disciplined, for five decades, about a rule I never once made defend itself.

Then I changed what I ate. Within a year I had lost seventy pounds. The fatty liver cleared. The back pain went. The fog lifted. The arthritis in my hands disappeared. The pills and the painkillers stopped.

Five years later, at seventy-one, I take no medications and have had no procedures. It has not come back.

That is the reason this page exists. Not a theory I went looking to confirm — a result that arrived first, before any of the reading did, and left me with a question I could not put down: if a single change could reverse that much, in a man who had done everything he was told, then what exactly had I been eating for fifty years?

Everything on this page is downstream of that question.

And it was not an experiment of one

It did not start with me, and the part that matters is not what happened to me.

It started at a dinner. An old friend told me that something I had said to him more than twenty years earlier had changed the way he ate, and that he had eaten that way ever since. I could not remember the conversation at all. He had been living the result for two decades before he sat down and handed it back to me.

He had not recovered from anything. He never got sick.

That is the part I keep returning to. Two men, the same information, in the same room, on the same evening. One of us acted on it and did not develop any of what came for me. The other said it out loud, forgot he had said it, went on following the official advice for another twenty years, and arrived at sixty-six obese and failing.

I was the source, and I was the one who did not take it.

So we are not two instances of one experiment. We are two different experiments pointing the same direction. He tested whether the condition could be avoided. I tested, two decades later and at considerably greater cost, whether it could be reversed. He is not a second case of my result. He is the closest thing to a counterfactual I will ever have — the man who got the same information at the same moment and used it.

And what I did with it was an experiment rather than a story.

The intervention was a single variable. I removed one category of food. I did not add a drug, add a supplement, start a program, take up an exercise regime, change my work or change anything else I could hold still. Most people who overhaul how they eat overhaul eight things at once and afterward cannot tell you which one moved the needle. I changed one and held the rest, because that is how you find out what a change did.

It was documented rather than recalled. Weight before and after, from records and not from memory. A laboratory panel, published in full, including the one value on it I would rather not show anybody.

And it was reasoned about rather than merely experienced. I spent a career as a physical chemist studying how systems hold together and how they come apart. I knew which explanations were available, which measurements would distinguish between them, and where to go and read. Most accounts like mine come from people with the outcome and no access to the mechanism — who cannot separate a cause from a coincidence and do not know what would have falsified their own story, which means nothing they report can be checked, including by them.

And then there are all the others

Since that dinner I have read a very large number of accounts of the same shape. So, probably, have you.

I want to be careful about what those are and are not worth, because this is where people in my position usually overreach.

They cannot give you a rate. There is no denominator. You never see the people it did not work for — they do not write it up, nobody is counting them, and the selection runs one direction and runs hard. A thousand accounts of success are compatible with a hundred thousand failures nobody recorded. Anyone citing volume as proof is doing arithmetic with a missing number.

What they are is a convergent signal. A large number of people, with little else in common, describing the same intervention and the same shape of outcome, with a mechanism that is at least coherent.

In any other field that is the thing that prompts somebody to run the study. It is how most clinical questions start — enough case reports accumulate that a trial becomes obviously worth funding.

No such study exists. Not one that isolates the variable, with a control arm, at a size that would settle it. The signal has been accumulating for years in plain view and it has nowhere to go, because nobody has built the instrument and nobody is paying for one.

Which is the same finding as the rest of this piece, arriving from the other direction.

But it is also how this actually travels

There is a second thing those accounts are, and it took me a while to see it.

When the formal channel is closed — no trial, no instrument, no program, nobody funding one — knowledge does not stop moving. It moves the old way. One person tells another, usually across a table, usually because something happened to them that they cannot keep quiet about.

That is exactly how it reached me — and how it reached him before me. A man I had known for decades, at a dinner, telling me that something I said twenty years earlier and completely forgot saying had changed what he ate for the rest of his life. He was not citing a study. He was handing back something I had passed on without knowing it.

Everything since is the same transaction at a larger scale. The book is that conversation written down. This page is the same thing, continued.

I am not going to pretend that channel is any good. It is slow and lossy, it has no error correction, and it carries nonsense exactly as efficiently as it carries anything true. The same dinner table that brought me this brings somebody else a supplement scheme and a testimonial from a man who does not exist. You cannot tell the two apart by the channel they arrived through.

That is the entire reason I document what I document and say where I stop. It is not fastidiousness. It is the only thing that distinguishes one of those conversations from the other.

But it is what is left. When nobody builds the instrument, the thing still working is people telling each other what happened to them — and taking the trouble to say how they know.

Where that leaves the evidence

One man who made the change early and did not get sick, followed for twenty years. One man who made it late and came back, run as a deliberate single-variable intervention with records kept. A large and growing body of reports I cannot put a denominator under. And no trial.

That is not proof and I will not call it proof. It is also not nothing, and the people who dismiss the whole category as anecdote are making the opposite error — treating the absence of a study as though it were a negative result, when no study was ever run.

Rose says the comparison cannot be run, because everybody is exposed and there is no unexposed group to measure against. Between the two of us we had one anyway, by pure accident of circumstance — one man who stepped outside the exposure early and one who did not step outside it for another twenty years. It is a sample of two and it was not designed by anybody. It is still more of a comparison than the literature contains.

That is the only reference point any of us has ever had. And having had one is how I noticed that everybody else is working without it.

What this means for you, and it is not comforting

You cannot look this up.

Not because the information is suppressed or hard to find. Because it was never recorded. The condition your body was built for ended before anyone thought to document it, and the people measuring you today are measuring against what is ordinary now.

That is the no-man’s-land. We are far enough from the original condition that it no longer constrains anything, and we have no record of it to navigate by — so every claim about what is good for you, mine included, is being argued in a space with no fixed point in it.

I am not going to end this by telling you what to eat. I do not have the evidence for that and neither does anybody who says it with confidence.

What I will say is that the next time you are told a value of yours is normal, it is worth remembering what that word is actually reporting. It means you resemble the people around you. It has never meant anything more than that, and in a population where everyone shares the same exposure, resembling each other is not the reassurance it sounds like.

On Saturday: what I did about all of it, the audit I promised at the start of this series, and the four things in my cabinet that contradict half of what I have just told you.

They were not disinterested. I did not read a newspaper article about magnesium and sleep and think, there is an interesting structural argument here about federal enrichment standards. I read it and thought: do I need to be taking this?

I had been, in a desultory way, without ever examining whether it was doing anything. I spend my working life deciding whether a material belongs in a system, and it seemed indefensible to keep swallowing something on no better basis than that it was already on the shelf.

So I went and found out. Three weeks produced the chlorophyll argument, the Rothamsted archive, the enrichment list, the photograph rule and the evolutionary reason the body never built an alarm for any of it.

What it did not produce was an answer to the question I started with. Here is the audit, and it requires a disclosure.

How I Eat

Readers here know most of this. For some years I have eaten a strict ketogenic diet that has drifted steadily toward carnivore. Carbohydrate is effectively zero. Over the past six months or so vegetables have gone from a few ounces used mainly as a vehicle for butter to essentially nothing — under five percent of calories.

Dinner most nights is a ribeye, ten to twelve ounces, cooked in butter. Occasionally baked cod instead. Coffee with butter and cream. A few strawberries with heavy cream now and then, which is close to the whole of my produce consumption.

I am not defending this here and I am not recommending it. It is simply the input condition for the audit, and it makes my accounting much shorter than most readers’ will be.

The Arithmetic

Work it out and it is not ambiguous.

Beef carries somewhere in the region of twenty to twenty-five milligrams of magnesium per hundred grams. A twelve-ounce ribeye is about three hundred and forty grams, so call it seventy or eighty milligrams. Butter is negligible — two milligrams per hundred grams, and I am not eating hundreds of grams of it. Heavy cream, single digits. Coffee contributes perhaps seven milligrams a cup, so several cups gets me twenty or thirty. Cod does better than beef, around thirty per hundred grams, on the nights I have it.

Total it honestly and I am somewhere around a hundred and thirty milligrams a day, give or take, on a typical day.

The recommended intake for a man my age is four hundred and twenty. The estimated average requirement — the lower figure, the one half the population falls below — is three hundred and fifty.

I am taking in roughly a third of the recommendation, structurally, by design, every day.

That is not a marginal shortfall identified by a population survey. It is arithmetic anyone can do with a food composition table in ten minutes, and it does not require me to believe anything contested.

And I Wrote the Explanation Three Weeks Ago

Here is the part that genuinely stopped me.

In the first of these posts I made two arguments without noticing they were about me.

The first: magnesium rides the whole food chain, and it rides it inside the green part. It is the atom at the center of chlorophyll. Every leaf, every stem, every seed.

The second, in the section on livestock: feeding cattle more magnesium does not raise the magnesium in your steak, because mammals defend magnesium homeostasis tightly and do not warehouse it in muscle. The animal buffers it out of the product before it reaches you.

I wrote both of those paragraphs as general chemistry. They are a precise description of my own dinner.

I eat an animal that ate the green part, and I wrote the explanation of why that does not transfer, and I did not connect it to myself until I sat down to do this audit.

That is worth saying out loud, because it is the ordinary way people miss things. Not through ignorance — I knew every fact involved and had just published them — but because the category “general argument about the food supply” and the category “what I had for dinner” were filed separately in my head.

What Cuts the Other Way

Now the honest counterweights, which are real and which I would be dishonest to leave out, because they all push against my conclusion.

No phytate. This matters more than people realize. Whole grains and legumes carry magnesium, but they also carry phytic acid, which chelates divalent cations and substantially reduces how much of that magnesium you actually absorb. The magnesium in animal food has no such competition. My absorbed fraction is probably at the favorable end of the range — which means my effective intake is better than my raw intake suggests.

Lower renal losses. Hyperglycemia drives magnesium out through the kidney — that is precisely why type 2 diabetes sits on the clinical risk list. I have essentially no glycemic excursions at all. Whatever I take in, I am plausibly losing less of it than a standard eater.

Adaptive absorption. The gut takes up a larger fraction when intake is low. That mechanism exists specifically for this situation, and mine has had years to settle into it.

So the honest statement is: low intake, probably better absorption, probably lower losses. Where that nets out, nobody can tell me, and that is not a rhetorical hedge. It is the actual state of the question.

I will add one thing from the ketogenic world generally, which is not evidence but is not nothing: electrolyte supplementation — sodium, potassium, magnesium — is near-universal practice among long-term low-carbohydrate eaters, and cramping is the classic complaint of those who skip it. That is a community’s accumulated experience rather than a trial, and I weight it accordingly, but I do not discard it.

Why I Still Cannot Measure It

I could have a serum magnesium drawn tomorrow for almost nothing. It would tell me essentially nothing.

Sixty percent of my magnesium is in bone and nearly all the rest is inside cells. Under one percent is in the blood being sampled, and the body defends that one percent ferociously, drawing on bone to hold it steady. A normal result would mean my regulatory system is working. It would not mean my reserves are intact. Those are different claims and only one of them is being tested.

There is no accepted routine test for the thing I actually want to know.

And there is a second failure in the audit that follows directly from my own first post. The food composition tables may be stale. If the Rothamsted archive is right that mineral concentration in grain fell from the 1960s onward with high-yield cultivars, the reference values we all count against were measured on different crops. I have less exposure to that than most — I am not eating grain — but the beef numbers come from the same kind of tables, built over decades, on animals raised differently.

The blood test reads the wrong compartment. The dietary calculation reads against an uncertain standard. I cannot determine my own magnesium status, and neither, as far as I can establish, can anyone.

So I Decided Without It

Which is ordinary. I spend my professional life making decisions without sufficient data, because the alternative is making none. You do not wait for certainty to specify a material. You look at the cost of being wrong in each direction and choose the error you can live with.

If I supplement and did not need to: a few dollars a month, loose stools if I overshoot, and a real if modest risk that the pill makes me less attentive to the food.

If I do not supplement and did need to: a marginal chronic insufficiency that produces no symptom, crosses no threshold, triggers no test, and draws quietly on a bone reserve I would like to keep for other purposes.

That is not a symmetric pair, and the asymmetry is manufactured by the argument of the previous three posts: the harm, if it exists, is specifically the kind that never announces itself. There is no readout, so there is no experiment to run on myself.

Against a measured intake around a third of the requirement, and being past seventy, which puts me on the clinical list on its own, I am continuing a modest daily dose.

The Rest of the Stack, and a Weaker Argument

Full disclosure, because half a disclosure is worse than none. I do not take only magnesium. I take magnesium, a B complex and a multivitamin, and I add potassium-blended salt to my fluid intake. I am continuing all four.

Magnesium has the argument above — a documented removal chain and an intake gap I can calculate. The others are not equivalent to it, and they are not equivalent to each other. One of them has a better case than magnesium does. One of them is the weakest thing I do.

The B complex has a real argument, and it is the same one. Look at which vitamins sit on the 1941 enrichment list: thiamin, riboflavin, niacin, and later folate. Those are precisely the nutrients the United States decided were being stripped out of the national flour supply badly enough to mandate putting back. I have exited that delivery system entirely. Whatever the enrichment program is doing for everybody else, it is doing nothing for me.

And one of them is pointed. Folate is named for foliage. It was isolated from spinach and it is concentrated in leaves. I eat no leaves. That is not a subtle inference; it is in the word.

The multivitamin is the weakest thing I do, and I am not going to dress it up. There is no specific removal story behind it. It is a broad hedge against not knowing, which is approximately the argument I spent a section of this piece criticizing when the supplement industry makes it.

The reason I take it anyway is the one I keep arriving at from every direction in this series: we are badly served on knowledge here. There is no routine test for subclinical status in most of these nutrients. Trials cannot select deficient subjects because we cannot identify them. The food composition tables are aging against crops that have changed. And research money goes where the patents are, which is not toward settling whether a seventy-year-old eating an unusual diet is short of riboflavin.

In that environment a cheap, broad, low-dose hedge is a defensible response to ignorance. It is also precisely what somebody rationalizing a habit would say, and I cannot fully separate the two from the inside.

The salt has the strongest argument of anything I take, and it is the one with a genuine downside.

In the third post I described the kidney as fighting the last war: built over millions of years to hoard scarce sodium and dump abundant potassium, now operating in a world where both inputs are inverted.

A near-zero-carbohydrate diet partially un-inverts that, and not by changing what I eat. It changes the hormonal signal that controls retention. Low circulating insulin reduces renal sodium reabsorption, so a ketogenic eater loses sodium continuously through the kidney — this is the established mechanism behind what low-carb people call keto flu, and it does not stop after the first fortnight. And because the body answers sodium loss with aldosterone, which trades sodium back in exchange for potassium, the sodium loss drags potassium out behind it.

So unlike the magnesium case, this is not a story about low intake. It is an active, continuous, mechanistically understood loss, running every day, caused by a decision I made on purpose. Potassium-blended salt replaces both sides of it in roughly the proportions they leave.

On intake as well: beef carries potassium reasonably well, around three hundred milligrams per hundred grams, so a ribeye supplies perhaps a thousand. The adequate intake for a man is three thousand four hundred. Much the same shortfall as magnesium, from a food that does rather better at delivering it.

And now the part where my own asymmetry argument breaks.

I have argued throughout this piece that the cost of supplementing unnecessarily is a few dollars and loose stools. For potassium that is simply untrue. Potassium is the one electrolyte where overshooting is genuinely dangerous — hyperkalemia disturbs cardiac conduction and it can kill. The risk concentrates in people with reduced kidney function, or taking ACE inhibitors, ARBs, or potassium-sparing diuretics. Renal function also declines with age on its own, without any disease, and I am past seventy.

So the reasoning I used for the other three items does not transfer to this one, and I should not pretend it does. For potassium the honest position is not cheap, probably helps, cannot hurt. It is: this one has a real mechanism behind it and a real ceiling, I take a modest amount rather than a heroic one, and it is the single item in my stack I would raise with a physician rather than settle alone.

If you are reading this and reaching for potassium salt because the internet told you to, that is the paragraph to take away.

One thing I do check, and it comes straight out of the third post in this series. I wrote there that the human body has no regulated excretion pathway for iron — we recycle it obsessively and control only absorption, because ancestral iron loss happened almost solely through bleeding. I eat a great deal of red meat, which is the most absorbable iron there is.

Adding supplemental iron on top of that is the one item I would actively avoid, and plenty of multivitamins contain it. If you are a man eating this way, read the label on yours. My own argument from three weeks ago is the reason.

Now the Part That Should Worry You

I have just made an argument I would not accept from somebody else.

Look at its shape: the benefit cannot be measured, the absence of evidence is explained away, the cost of being wrong is asserted to be small, and therefore the sensible course is to buy the product.

That is the structure of every supplement pitch ever written. It is how a fifty-billion-dollar industry sells things that do nothing. I have used it on myself, knowingly.

Three objections I owe you, because they are the ones I cannot fully answer.

First. My claim that supplementation trials come out null because they enroll people who were already replete is reasonable and unfalsifiable as I have stated it. Any null result can be waved away that way. The honest version requires a validated test for deficiency so you can select subjects, which is exactly what does not exist.

Second. The asymmetry argument proves too much. It licenses supplementing with nearly anything cheap and non-toxic, which is how people end up with thirty bottles. The only thing restraining it is a documented removal story, and I have to hold that boundary strictly or admit the principle is doing no work.

Third, and this one is specific to me. A man who has deliberately removed an entire food category from his diet has a standing motive to conclude that a pill can replace it. That is a conflict of interest, it is mine, and no amount of chemistry cancels it. The multivitamin is where it shows most plainly — I can construct an argument for the magnesium and a decent one for the B complex, and for the third bottle I have only the general feeling that it probably helps and certainly cannot hurt, which is the exact sentence I would mock in somebody else. I do not think I am rationalizing. I am aware that is what rationalizing feels like from the inside.

The Harder Admission

There is one more thing the audit turned up, and it is the one I have been circling.

These three posts were an argument that the industrial food supply strips minerals out — bred out by yield selection, milled out by refining, never restored by a federal list written in 1941, and crowded off the plate by what replaced the foods that carried them.

I have exited that supply almost entirely. I eat no refined flour, so the enrichment omission does not touch me. I eat no engineered products, so the displacement argument does not apply to me. By the terms of my own case I should be fine.

I am not fine. I have arrived at a magnesium gap that is larger than the one I spent three weeks criticizing, by an entirely different route, through decisions I made myself and would make again.

Leaving the system did not exempt me from the arithmetic. The system is not the only thing capable of subtraction, and a diet assembled on first principles can drop a required input just as thoroughly as an industrial process can — more quietly, in fact, because there is no manufacturer to blame and no label to read.

That is a chastening thing to find at the end of three weeks of work, and I think it is the most useful thing in the series.

What I Am Not Doing

● Not treating the capsule as a substitute for the food. Magnesium belongs in the leaf, packaged with everything else that was removed alongside it. A pill restores one line item from a list I have not seen in full.

● Not expanding it. Four items, with four different strengths of argument behind them. I am not adding a fifth because something sounded interesting, and the strength of the argument behind each one is not equal — I have tried to say which is which rather than let the strong case carry the weak ones.

● Not claiming I feel anything. I do not.

● Not recommending it to you. I have shown you how I reasoned and where the reasoning leaks.

What Would Change My Mind

A validated routine measure of magnesium status — something reading the intracellular compartment or the bone reserve rather than the one percent floating in serum. With that, trials could enroll people who are genuinely deficient, and the question becomes answerable in a decade.

Short of that we are all guessing, and the only honest difference between us is whether we say so.

What I actually want is simpler than a pill, and it is not available to me. I would like the magnesium to arrive the way it arrived for several million years — inside the green part, with everything that travels alongside it. I have chosen, for my own reasons, not to eat that way.

So I take the capsule, and I know exactly what it is: a patch, applied downstream, at my own expense, on a judgment I cannot defend rigorously. Which is more or less what I accused the food supply of forcing on everybody else.

A Last Thought, About the Species

For several million years the environment did the optimizing and we did the adapting. Every reserve we carry, every absorption mechanism, every appetite signal is a solution to a problem that actually happened, worked out slowly against conditions nobody designed.

In roughly a century we reversed the direction of that arrangement. We now engineer the environment ourselves, deliberately, to specification, against objectives we choose — yield per acre, shelf life, consumption rate, cost per calorie. The environment changes at the speed of a product cycle. Our biology changes at the speed of a species.

So every adaptation that was once correct has quietly become a liability. A kidney built to hoard scarce sodium, in a world where salt is unavoidable. A satiety system calibrated against food that fought back against the teeth. A magnesium economy that built no reserve because there had never once been a shortage.

None of those are defects. They are precise answers to questions we stopped asking.

And here is the part that should unsettle anyone: the body cannot report this. It signals only the shortages it has met before. We are the first organism able to re-engineer its own environment faster than it can possibly be informed about the result — which leaves the instrument that was supposed to warn us permanently reading the wrong century.

I did a small version of the same thing to myself, on purpose, for reasons I still think are sound. I assembled an input condition from first principles and then found I had no way to tell whether it was working.

That is the species’ predicament, running in one kitchen.

On Saturday I laid out the rule that governed which deficiencies got fixed in the twentieth century and which did not. Seven that produced a visible, nameable disease were addressed by national mandate — iodine, vitamin D, thiamin, niacin, riboflavin, iron, folate. The ones that produce only a shift in a distribution were not, and still have not been.

I closed on a question I had been quietly treating as settled.

Why does running short of niacin blacken a man’s skin and kill him, while running short of magnesium produces nothing you could point at across a whole lifetime? Both are essential. Both are needed by the same cells, every second. One shouts and one is silent, and that difference decided which received a federal mandate.

I had been treating a nutrient’s visibility as a fact about the nutrient, the way atomic weight is a fact about an element. It isn’t. It is a fact about us — and specifically about what our ancestors ran out of.

You Do Not Defend Against a Famine That Never Came

Here is the principle the whole thing rests on, and it is an engineering principle before it is a biological one.

Every capability an organism maintains costs something — energy, genome, machinery that has to be built and kept. Selection is relentless about removing capacity that is not being used. So the reserves, recycling systems and synthesis pathways a body carries are not a wish list. They are a record of which nutrients were actually intermittent in the environment that shaped it.

Test it against what the human body stores, and the pattern is unmistakable.

● Vitamin B12 — the liver holds years of it. B12 comes only from microbial synthesis, reaching us through animal foods, which arrived episodically. Selection built a deep reserve against a supply that could fail for months.

● Vitamin A — months of storage in the liver. Organ meats and seasonal carotenoids. Intermittent.

● Iron — recycled obsessively, and here is the striking part: the human body has no regulated excretion pathway for iron at all. We cannot dump it. The only control is at absorption. That is what you build when loss happens solely through bleeding, which was rare.

● Magnesium and potassium — no dedicated store. Bone buffers magnesium and that is the whole of it. Because they arrived continuously, inside every green thing eaten, every day, for several million years.

The body is not badly designed for magnesium. It is correctly designed for an environment in which magnesium never ran out.

There was no selection pressure to build a warehouse for a commodity that was never once in short supply.

We Watched Ourselves Throw One Away

Vitamin C is the proof, because the discarding is still visible in the genome.

Most mammals synthesize their own vitamin C. We cannot. The gene for the final enzyme in that pathway — gulonolactone oxidase — is still sitting in human DNA, broken. A pseudogene. The primate lineage lost it tens of millions of years ago.

Why? Because the ancestral primate diet was fruit-rich enough that making your own was redundant, and redundant machinery gets dropped.

That was an entirely sound trade for roughly sixty million years. Then we put men on ships for six months and invented scurvy.

The deficiency was never a flaw in the design. It was the environment changing after the redundancy had already been thrown away.

The Kidney Is Fighting the Last War

The sharpest case is sodium and potassium, and it is worth slowing down for.

The ancestral diet was high potassium and low sodium. Plants are potassium-rich — it is the dominant intracellular cation in every leaf and tuber. Sodium, by contrast, was genuinely scarce inland, scarce enough that salt functioned as currency and trade routes were built to move it.

So the human kidney evolved exactly what that environment demanded: formidable machinery for retaining sodium, and efficient machinery for dumping potassium. Aldosterone, the renin system, the whole apparatus is built around hoarding salt.

We have now inverted both inputs completely. Sodium is in everything. Potassium is in the foods people stopped eating.

Which means we are carrying a powerful, highly refined, metabolically expensive defense against precisely the deficiency we no longer have — and almost nothing against the one we do.

The control system is not broken. It is working perfectly, against the wrong problem, with great determination.

Two Different Kinds of Deficiency

It would be easy to read all of this as “modern life broke us,” and that would be too simple. There are two distinct categories and they should not be blurred.

Ancient and geological. Iodine, selenium, vitamin D. These are driven by soil chemistry and latitude, and they are old. Goiter is not a modern disease; inland populations have had it for as long as there have been inland populations. Vitamin D shortage at high latitude is old enough to have depigmented the skin of northern peoples — that is an adaptation to it. Nobody manufactured these.

Recent and manufactured. Magnesium, zinc, potassium, fiber. These were not problems. There is no ancient disease associated with them because there was no ancient shortage. We created them inside a century, by removing things.

And notice which were addressed first. Iodized salt, 1924. Vitamin D in milk, in the 1930s. Both ancient, both visible, both solved before the flour enrichment program existed. Humanity had been looking at goiter and rickets for thousands of years and recognized them instantly when the chemistry arrived.

Nobody has been looking at marginal magnesium insufficiency for thousands of years, because until roughly 1900 there was nothing to look at.

We Have Made This Exact Mistake Before

There is a precedent worth knowing, because it is the same error with the polarity reversed.

Maize is a nutritional trap in its raw state. Its niacin is chemically bound and biologically unavailable, and the grain is poor in tryptophan, from which the body can otherwise make niacin. Eat maize as a staple without further treatment and you get pellagra.

Mesoamerican cultures solved this somewhere around three thousand years ago. Nixtamalization — cooking the grain in an alkaline solution of wood ash or lime — liberates the bound niacin and makes it available. It is a precise piece of food chemistry arrived at empirically and transmitted as ordinary cooking practice.

Europe took the crop and left the method behind.

The result was three centuries of pellagra across southern Europe and later the American South — a disease that did not exist in the populations that had domesticated the plant, caused entirely by importing a food without the processing step that completed it.

Roller milling did the same thing to wheat four hundred years later, in the opposite direction: not a step omitted, but a step added. Either way the principle is identical. A food and the processing it requires are a single system, and separating them has consequences that take generations to appear.

Why One Shouts and One Is Silent

Now the question from Saturday can actually be answered, and the answer has two parts that work together.

First: the body has no alarm for a shortage it never encountered. You do not build a warning system against a condition that never once occurred. There is no magnesium-deficiency syndrome in the way there is a scurvy, because there was never a magnesium famine for a syndrome to evolve against. The silence is not evidence that the shortfall is harmless. It is evidence that it is unprecedented.

Second, and this is the part I had been missing: the modern deficiencies are partial, not total.

Pellagra, beriberi and scurvy appear when a nutrient is essentially absent — bound maize and nothing else, polished rice and nothing else, no fresh food for six months. Total deprivation. The body hits a wall, systems fail in sequence, and a syndrome results.

Refined flour is not zero-magnesium. It is low-magnesium, by about eighty percent, permanently, across an entire population. Nobody is deprived. Everybody is thinned.

And a partial shortfall, sustained for decades, in a system with adaptive absorption and a bone reservoir and a kidney that tightens its grip, produces exactly what we observe: compensation, not collapse. The body spends its reserves slowly and quietly and the blood test stays normal the whole way.

Total deprivation of something you never learned to store makes a picture.

Partial deprivation of the same thing makes a statistic.

We industrialized our way into the second kind, exclusively and comprehensively, in about a hundred years.

What This Means for the Rule

So the sorting rule from Saturday is not arbitrary and it is not merely political. It runs on physiology.

A deficiency becomes visible when it crosses a threshold the body evolved a response to. Our ancestors ran out of vitamin C, and of B12, and of iodine, and we carry machinery and syndromes that reflect those encounters. They never ran out of magnesium, zinc or potassium, so we carry nothing — no store, no alarm, no syndrome, no name.

Which means the deficiencies we manufactured in the last century are, by construction, the ones our biology is least equipped to report. Not by coincidence. Because they are the ones that never happened before, and biology only reports what it has seen.

And a political system that acts only on what biology reports will act on exactly the wrong set.

That is the whole chain, and there is nobody in it who did anything obviously wrong. Breeders optimized yield. Millers optimized shelf life. A committee in 1941 fixed the diseases that were killing people. Legislatures responded to the harms that could be counted. Every one of those was a reasonable decision.

The gap they left is real anyway, and it is still open, and the body that is supposed to warn us about it was built in a world where the problem could not occur.

In 1941 the United States determined which nutrients would be restored, by law, to the national flour supply. Five made the list. Magnesium did not, and has not been reconsidered since.

I said that was strange and that it turns out not to be. Here is why.

The Ones We Fixed

Start with the successes, because they are real and they are among the best public health work this country has ever done.

● Iodine. Deficiency produces goiter and, in the developing child, cretinism. Iodized salt arrived in 1924, beginning in the Michigan goiter belt. Essentially solved.

● Vitamin D. Deficiency produces rickets — bowed legs, deformed ribs, a visibly misshapen child. Milk fortification in the 1930s. Essentially solved.

● Thiamin. Deficiency produces beriberi — neuropathy, heart failure, wasting. Flour enrichment, 1941.

● Niacin. Deficiency produces pellagra — dermatitis, diarrhea, dementia, death. Enrichment, 1941.

● Riboflavin. Deficiency cracks the corners of the mouth and inflames the tongue. Enrichment, 1941.

● Iron. Deficiency produces anemia, diagnosable with a cheap blood test and a threshold value. Enrichment, 1941.

● Folate. Deficiency during a pregnancy produces neural tube defects — spina bifida, anencephaly. Mandatory fortification, 1998. Neural tube defect rates fell by roughly a third.

Seven interventions. National, mandatory, industrial, cheap, and effective.

The Ones We Did Not

● Magnesium. Diluted by yield breeding, stripped by milling, never added to the enrichment standard. About half of American adults consume below the estimated average requirement.

● Zinc. The same story and arguably worse. Milling removes seventy to eighty percent. Declining in grain alongside magnesium in the Rothamsted archive. Falling further under elevated atmospheric CO2. Absorption inhibited by the phytate in the very whole grains that carry it. Not in the standard.

● Potassium. Most Americans consume well under the adequate intake — less a milling story than a displacement story, with the same result.

● Copper and manganese. Same milling losses, lower stakes, same silence.

Now hold the two lists side by side and look for what separates them.

It Is Not What You Would Guess

It is not severity. Magnesium participates in every energy transaction in your body — the active form of ATP is the magnesium complex. Zinc is structural in hundreds of enzymes and in the transcription factors that read your genome. These are not peripheral nutrients.

It is not prevalence. Half the country sits below the requirement for magnesium. That is not a rare exposure.

It is not cost or difficulty. Magnesium and zinc salts are cheap and stable. They could have been added to the same flour, in the same mill, on the same day, by amending the same sentence of the same federal standard. There is no technical obstacle and there never was one.

It is not the state of the science. We have understood what these minerals do since the middle of the last century.

What separates the two lists is this: every deficiency on the first list produced a named disease you could look at.

Goiter is a swelling at the throat. Rickets bows a child’s legs. Pellagra blackens the skin of the forearms in a pattern with a name. Beriberi puts a man on the floor. Anemia is a number under a line on a chart. Spina bifida is an infant on an operating table.

Each is an event — discrete, visible, countable, assignable to a single cause.

Nothing on the second list produces an event. Insufficient magnesium produces, at most, a slight rightward shift in a population’s blood pressure distribution and a slightly worse night’s sleep. Insufficient zinc in a well-fed adult produces nothing a physician would name. There is no patient to photograph, no syndrome to title, no parent to testify.

Testing the Rule Where It Should Break

A rule is only worth keeping if it survives the cases built to destroy it. Two try.

Folate is the hard one. The deficient person — the mother — has no symptoms whatsoever. She feels perfectly well. By my rule she should have been ignored alongside magnesium.

But look at what the deficiency produces downstream. An infant with anencephaly or spina bifida is as visible and as countable as anything in medicine. And in 1991 and 1992 two randomized controlled trials demonstrated that folic acid prevented it — not an association, a randomized trial with a hard endpoint.

So the rule survives, and it gets sharper. It was never about whether the deficient person shows symptoms. It is about whether the harm is visible and assignable to one cause. Folate had both. Magnesium has neither.

Calcium is the second test, and it is the one that convinced me, because it sits in the middle.

Calcium deficiency does produce a visible outcome — fractures, collapsed vertebrae, the stooped posture of advanced osteoporosis. But it arrives after four decades, it is multifactorial, and no individual fracture can be assigned to it. Half visible. Half attributable.

And calcium received exactly half a measure. In the federal enrichment standard it is listed — as optional. Permitted, encouraged, not required. Eighty-five years of a middle verdict for a middle case.

I did not go looking for that. The rule predicted a gradation and the gradation was sitting in the regulation.

Zinc Is the Controlled Experiment

If you want the cleanest demonstration available, set zinc against iron.

Both are trace metals. Both concentrate in the bran and the germ. Both are removed by the same roller mill in the same pass. Both are cheap to add. Both were well understood by the 1930s. Both were on the table in 1941.

Iron is in the standard. Zinc is not.

The only material difference between them is that iron deficiency produces anemia — which has a name, a clinical picture, a cheap blood test and a numerical threshold. Zinc deficiency in a well-fed adult produces slightly impaired immunity and slightly slower wound healing: measurable in a study, invisible in a clinic.

Same mineral class, same removal, same cost, same decade, same committee. One got fixed.

The difference is diagnostics, not biology.

Why the Rule Holds

None of this requires anyone to have behaved badly, and I do not believe they did. The people who built the enrichment program were doing serious work against the problems in front of them, and they succeeded.

The rule holds because of what an institution is physically able to grip.

To mandate a national intervention you need, roughly: an identifiable event, attributable to a single cause, that can be counted, with somebody prepared to come and testify about it.

A statistical drift has none of those. There is no event. The attribution is spread across a dozen causes. The counting requires a cohort study and an epidemiologist. And there is no constituency — nobody’s child has magnesium insufficiency, nobody forms an association for it, nobody writes to a senator about a distribution curve.

So the system does not ignore invisible harms out of malice. It ignores them because there is no handle on them.

I made this observation once before about thalidomide, which produced photographs and therefore produced the most sweeping drug reform in American history, while the food half of the same agency kept an exemption written for vinegar. It is the same rule. Reform follows the picture. It does not follow the injury.

What the Rule Predicts

This is the part that bothers me, because a rule that explains the past also forecasts.

Essentially everything we now call chronic disease sits on the wrong side of it. Insulin resistance, hypertension, metabolic dysfunction, fatty liver, the slow accumulation of visceral fat — every one is a distribution shift, multifactorial, with no acute presentation. There is no moment. There is no photograph. There is a curve.

Which is why the fight over defining ultraprocessed food, which I wrote about recently, is not a semantic sideshow. It is the entire contest.

A definition is the closest thing available to a photograph. Name the category and it can be counted. Count it and it can be studied against outcomes. Study it and it can be labeled, taxed, restricted, or named in a complaint. Until then it is a curve, and curves do not move legislatures.

That is also why the resistance concentrates precisely there, and why it has held.

The Short Version

We have never decided which nutrient deficiencies to fix on the basis of how much harm they cause. We decided on the basis of whether the harm had a face.

Seven that did were fixed, thoroughly and permanently, at trivial cost. The ones that did not are still waiting, and the remedy is still one additional line in the same federal standard.

Nobody will write that line, because nobody can produce the photograph that would require it.

But Something Is Still Missing

I have been treating the visibility of a deficiency as a given — as though it were simply a fact about each nutrient, like its atomic weight, that determined its political fate.

That is not an explanation. It is a restatement.

Why does running short of niacin blacken a man’s skin and kill him within a few years, while running short of magnesium produces nothing you could point at across an entire lifetime? Both are essential. Both are required by the same cells. One shouts and one is silent, and the difference decided which got a federal mandate.

That question has an answer, and it is not about politics or regulation at all. It is about which shortages our ancestors actually encountered, and what the body was therefore built to notice.

Next Wednesday.

That is a fair verdict on the sleep claim and I have no quarrel with it.

But the article asks whether magnesium helps you sleep, and never asks the question sitting underneath it.

Why is half the country short of magnesium in the first place?

This Is Not a Rare Element

Start with what magnesium is, because the word “supplement” makes it sound exotic.

Magnesium is roughly the eighth most abundant element in the Earth’s crust. It is the third most abundant element dissolved in seawater. There are mountains made substantially of it.

And it is the atom at the center of chlorophyll.

Every chlorophyll molecule is a porphyrin ring with a single magnesium ion held in the middle of it. That ion is where photosynthesis happens. Every green thing on this planet is built around magnesium, which means it is present in every leaf, every stem, every seed, and every unrefined plant food that has ever existed.

And because every green plant carries it, so does everything that eats green plants. A grazing animal takes in magnesium with every mouthful of forage. That is the delivery route, and it has been the delivery route for as long as there have been grazing animals — rock to soil, soil to chlorophyll, chlorophyll to the animal that eats the leaf, and then to whatever eats the animal. Magnesium rides the whole chain, and it rides it inside the green part.

It is not scarce in the ground, not scarce in the ocean, not scarce in plants, and not scarce in the animals that live on plants.

It is scarce in the American diet. That is a different kind of fact, and it requires a different kind of explanation.

What It Actually Does

A short primer, because the rest of this depends on it and almost nobody is taught any of it.

Start with the one fact that reframes everything else. ATP does not work without magnesium.

Adenosine triphosphate is the energy currency of every cell you own, and the biologically active form of it is not ATP — it is Mg-ATP, the molecule with a magnesium ion chelated to its phosphate chain. Bare ATP is essentially inert. Every enzyme that spends energy is handed the magnesium complex, not the nucleotide alone.

So magnesium is not a minor cofactor in a few reactions. It is a required participant in every energy transaction in your body. Several hundred enzymes need it, including every kinase that moves a phosphate group, and the polymerases that copy your DNA and build your proteins.

Its second job is electrical. Magnesium is calcium’s antagonist. Calcium drives contraction and excitation; magnesium opposes it, blocking calcium channels and damping the NMDA receptor. Calcium fires the gun, magnesium is the safety. That balance sets the excitability of nerve and muscle, which is why shortage announces itself as twitching, cramp, tremor and — at the extreme — tetany, and why a plausible sleep mechanism exists at all.

Now where it sits. An adult carries roughly twenty-five grams. About sixty percent is locked in bone, nearly all the rest is inside cells, and less than one percent is circulating in your blood.

And how it is managed. Absorption is adaptive — your gut takes up a much larger fraction when intake is low than when it is high. The kidney is the real regulator, filtering magnesium and reabsorbing the great majority of it, with the reabsorption concentrated in one stretch of the nephron called the thick ascending limb. Excretion, not absorption, is the control knob. And bone functions as the reservoir the body draws on when the books do not balance.

Hold onto that arrangement, because it explains something later that otherwise looks like a contradiction: a person can take in too little magnesium for years while every blood test comes back normal. Three compensating mechanisms are working to make that happen.

Four Subtractions

Here is how a mineral that is abundant everywhere becomes insufficient in a wealthy country. Four steps, and none of them was anybody’s stated intention.

One. The grain carries less than it used to. Rothamsted Research in England has grown wheat on the same plots since 1843 and archived the grain and the soil the entire time. When that archive was analyzed, zinc, iron, copper and magnesium concentrations in the wheat had held steady for over a century — and then fell sharply from the 1960s onward, tracking the arrival of semi-dwarf high-yield cultivars.

The crucial control: the soil minerals did not decline. The magnesium stayed in the ground. It stopped arriving in the grain.

The mechanism is dilution. Breed and fertilize for yield and you get more starch and water per plant; mineral uptake does not keep pace, so concentration falls. Nobody degraded anything. The objective function was bushels per acre, mineral density was not in it and was not measured, and it drifted. You get what you optimize for.

A chemist’s footnote worth adding. NPK fertilizer supplies nitrogen, phosphorus and potassium — the three inputs that limit yield and the three that get measured. It supplies no magnesium. And potassium and magnesium compete for uptake at the root, so heavy potassium application actively suppresses magnesium absorption. We are not merely failing to add it. One of the things we do add works against it.

Two. Milling removes most of what survived. Magnesium concentrates in the bran and the germ. Refining wheat into white flour strips roughly eighty percent of it out with them.

Three. Enrichment does not put it back. This is the step that should stop you.

Since 1941 the United States has required that refined flour be enriched. The federal standard specifies thiamin, riboflavin, niacin, iron, and folic acid. Calcium is optional.

Magnesium is not on the list. It was not on the list in 1941 and it has never been added.

Four. The other sources were displaced. Legumes, nuts, seeds, whole grains, leafy greens — the foods carrying the rest of the dietary magnesium are precisely the categories that lost shelf space to the engineered supply over the same period.

Four removals. Then a seventeen-percent-growth retail market selling it back in a capsule.

What About Meat and Milk?

Since I write about butter and beef, the obvious question: if the plants lost magnesium, what happened to the animals that eat them?

Two answers, and the first is better than you would expect.

The livestock industry solved this decades ago.

There is a disease called grass tetany — hypomagnesemic tetany, grass staggers. Cattle and sheep on lush spring pasture go down with tremors and convulsions, and frequently die. It has been recognized for roughly a century and it costs real money.

The mechanism will sound familiar. Rapidly growing pasture runs high in potassium and nitrogen and low in magnesium. And in the rumen, potassium directly antagonizes magnesium absorption across the wall — the same competition I described at the plant root, one organism further along the chain.

So fertilize pasture hard with nitrogen and potassium to drive yield, and you produce forage that carries less magnesium and blocks the uptake of what it still carries.

The response to this was not a debate. It was supplementation, and it has been routine for decades: high-magnesium mineral blocks, magnesium oxide in the ration, dosing the water through tetany season. Any cattleman knows what high-mag mineral is and when to put it out.

We identified the mechanism, named the disease, and built supplementation into standard agricultural practice.

For cows.

But it will not rescue you, and this is the part that cuts against the version of this argument I would prefer to make.

Feeding cattle more magnesium does not meaningfully raise the magnesium in your steak.

Mammals defend magnesium homeostasis tightly. It goes to bone and into cells, and it is not warehoused in muscle in proportion to intake. Beef muscle carries somewhere around twenty-odd milligrams per hundred grams and milk roughly ten per hundred milliliters — both modest against a four-hundred-milligram daily requirement, and neither moves much with what the animal was fed.

That is worth understanding generally, because it is not true of every nutrient. Transfer from feed to animal product depends on whether the nutrient has somewhere to accumulate. Selenium transfers efficiently because it gets built into selenoproteins, which is exactly why selenium-enriched feed works. Iodine in milk tracks feed supplementation closely. Fatty acid profile responds strongly to diet, which is the entire basis of the grass-fed argument. Vitamin E and carotenoids accumulate in fat, which is why grass-fed beef has yellower fat and more of both.

Magnesium has no such sink. It is regulated rather than stored, so the animal buffers it out of the product before it reaches you.

Which means you cannot eat your way out of this through meat and dairy, and I am not going to suggest otherwise because it would sit well with my other arguments. The magnesium is in seeds, nuts, legumes, greens and whole grains — in the bran, the germ and the leaf. Exactly where it was before anybody milled anything.

So — Is Supplementation Actually Needed?

I want to answer this honestly rather than usefully, because the honest answer has three parts and only one of them is satisfying.

For a specific and substantial group, yes, and this is not controversial. Long-term proton pump inhibitor use causes magnesium depletion — the FDA issued a warning about it in 2011. Loop and thiazide diuretics waste it, acting on exactly the stretch of nephron that reabsorbs it. Type 2 diabetes wastes it through the kidney. Chronic alcohol use depletes it. Malabsorptive conditions and bariatric surgery reduce uptake. Older adults absorb less and excrete more. Add those populations together and you have a great many people with a real clinical indication, and their physicians should be looking at it.

For the general population, the picture is murkier than either side admits. Around half of American adults consume less than the estimated average requirement from food. But frank, clinically obvious magnesium deficiency is uncommon in otherwise healthy people, and there is a reason both of those can be true.

Serum magnesium is a poor instrument. Sixty percent of your magnesium is in bone and nearly all the rest is inside cells. Barely one percent is circulating, and the body defends that one percent tightly — drawing on bone to hold it steady. So a normal serum magnesium does not establish that you are replete. It establishes that your regulatory system is still working.

Whether the gap between chronically low intake and normal serum represents meaningful subclinical depletion is genuinely unresolved. There is no accepted routine test for it. Anyone who tells you confidently in either direction is filling a hole in the literature with their own opinion.

And the supplement itself is a mediocre instrument. Magnesium oxide, the cheapest and most common form, is absorbed at a few percent. The citrate form is sold in the next aisle as a laxative, which is the practical ceiling on dosing. The article notes this almost in passing: the thing limiting how much you can take is that your gut will not tolerate it.

Which is itself informative. The body’s magnesium handling evolved for a dilute, continuous supply arriving inside food, all day, for several million years. It is not built to receive it as a bolus in a pill, and it does not handle it well.

What I Would Actually Say

Should you take magnesium for sleep? The evidence is thin and the Duke specialist is right. If you want to try it, it is cheap and unlikely to hurt you at sensible doses, and that is about as strong a recommendation as the data supports.

Should you take magnesium at all? If you are on a proton pump inhibitor, a diuretic, have type 2 diabetes, drink heavily, have a malabsorptive condition, or are elderly — raise it with a physician, because that is a real question with a real answer.

But for everyone else I think the question is miscast, and the capsule is a correct response to a situation that should not exist.

You are not short of magnesium because magnesium is hard to come by. You are short of it because it was bred out of the grain, milled out of the flour, left off the list that put the other nutrients back, and crowded off the plate by the foods that replaced the ones that carried it.

The pill is the fourth patch on a chain of three removals, sold at retail, unregulated for efficacy, absorbed at a few percent.

One Line That Was Never Written

What I keep returning to is the third subtraction, because it is the only one that was a decision.

Dilution was a side effect of breeding for yield. Milling was a choice about shelf life and texture. Displacement was a thousand separate commercial decisions nobody coordinated.

But enrichment was deliberate. In 1941 a group of people sat down and determined which nutrients would be restored to the national flour supply, mandatorily, forever. They put five on the list. It is one of the most effective public health measures this country has ever undertaken and I am not criticizing it.

Magnesium was simply not considered worth including — and it has not been reconsidered in eighty-five years, through every revision, while the other three subtractions compounded underneath.

And we were plainly capable of acting on this mineral in that same period, through that same potassium mechanism, when the consequence was visible. Grass tetany got identified, explained, and answered with routine supplementation, because a cow convulsing in a field is an event. Somebody loses an animal, counts the loss, and buys the mineral.

The cattle got their magnesium. We did not.

That is a strange outcome, and once I started pulling on it I found it is not strange at all — it is the predictable result of a rule that decided every one of these questions, for every nutrient, across the whole twentieth century.

I will lay that rule out on Saturday. It is not a rule about chemistry.

I read that as a triumph for about a day. A tax code and a deadline pulled the sugar out of a national beverage supply, and no parent had to say no in a supermarket aisle to make it happen.

Then I looked at what replaced it, which was aspartame and sucralose and acesulfame K and steviol glycosides. And I had to decide whether I thought that was a victory or a trade.

Where I landed, and why it took me a while

I cannot endorse a policy whose measured success is a substitution nobody has shown to be safe.

I want to be careful about the strength of that, because it would be easy to overrun it and I have been wrong in this direction before. The evidence against non-nutritive sweeteners is not remotely the evidence against sugar. The World Health Organization’s 2023 guidance advising against them for weight control was conditional and rated low-certainty. The observational data is badly confounded, because people who are already gaining weight are the ones who switch to diet drinks, which makes the diet drink look like the cause of a problem it arrived after.

So this is not a claim that sweeteners are as bad as sugar. It is a narrower thing: the one real-world success in this whole field consists of moving an entire population onto a compound class whose long-term effects are genuinely unsettled, and calling that a public health achievement. I would not sign that, and my reason is precaution about an unknown rather than certainty about a known.

But the deeper problem is what the substitution preserved

Nobody needs sugar. There is no dietary requirement for it, and no deficiency you can develop from never consuming any. I have written about that before.

What I had not followed through is that the same is true of sweetness. Sweetness is not a nutrient. It is a signal — a report filed by the tongue about what is arriving, generated before anything has been absorbed and before anything can be verified. It has no requirement because it is not a substance. It is information.

Which means the reformulation did not solve the problem. It removed the substance and preserved the signal. The drink still tastes the way it did, so it is still wanted, so the category survives, so the industry survives. The tax took out the part that could be measured and left intact the part that maintains the demand.

The comparison I wanted to make, and why it failed

I had a line ready. Nobody ever invented a safe cigarette, I was going to say. Tobacco use collapsed by people not doing it, because nobody thought the habit was owed a replacement.

That is wrong, and it is wrong because of vaping.

Somebody did invent the substitute. It removes the combustion and keeps the nicotine, and it is now a very large industry with a serious argument in its favour for people already smoking and a serious problem attached to everybody else it reached.

So the parallel holds after all, and it holds harder than the version I was going to write. Look at what each substitute keeps.

Vaping removes combustion and keeps the nicotine. Diet soda removes the sugar and keeps the sweetness. In both cases the thing removed is the carrier — the measurable, indictable, chemically obvious part. In both cases the thing preserved is the signal that maintains the habit.

Combustion is to sugar as nicotine is to sweetness. That is not a metaphor. It is the same engineering decision made twice by two industries facing the same regulatory threat, and it produced the same result: a product that answers the specific charge and retains the customer.

The difference that actually matters

Two of them, and they are the whole argument.

The first is timing. Vaping arrived in the two-thousands — forty years after the Surgeon General’s report, thirty after cigarette advertising left the airwaves, long after smoking had already fallen from a majority habit to a minority one. The substitute came after the collapse. Sweeteners arrived instead of one. One followed a reduction; the other is the mechanism by which a category avoided being reduced at all.

The second is that the substitute got regulated in turn. Flavour restrictions, age limits, advertising rules, taxes on the liquid. When the industry moved, the apparatus moved after it.

Nothing of the kind has happened with sweeteners. They are not taxed, not restricted, not labelled as anything in particular, and the entire reformulation was treated as compliance rather than evasion. The substitute was not merely permitted. It was the point.

So what would I actually do

Tax the sweetness, not the sugar.

Any added sweetener, nutritive or not, by sweetness intensity rather than by grams of carbohydrate. That closes the route the British levy left open, and it is no harder to draft — sweetener content is on the label and relative intensity is a published figure.

Nobody proposes this, as far as I can find, and I think the reason is that everybody has quietly accepted that the sweetness has to stay and only the sugar is negotiable. That is the concession underneath the whole debate and it has never been argued for. It was simply assumed, by people with an interest in it being assumed.

I am not certain I am right about this. It is an opinion held by a citizen who has read the file, not a finding, and I have kept those two things apart for five years and intend to go on doing it.

But I know what it looks like when an industry answers a charge by changing everything except the thing that keeps you coming back. I watched it happen to my own body for fifty years, one product at a time, and every one of them was compliant with whatever the rule was that year.

WARNING: A graphic photograph appears partway down this page. It ran in the British Medical Journal in September 1962 and it shows a living child. Almost no American has ever seen one, and that is not an accident — it is the subject of this essay. The image sits below the opening so that arriving at it is your decision and not mine.

NPR ran a piece Friday about American thalidomide survivors. Before anything else in this essay, it is worth being specific about what that word is covering, because the softened version is part of how this stayed buried.

Thalidomide was a sedative. In West Germany it was sold over the counter and marketed as so safe that a person could not overdose on it. It was recommended for morning sickness. That is the entire shape of the thing: a drug handed to pregnant women, for a symptom of pregnancy.

It cured nothing and it prevented nothing. At its absolute best it was a comfort — a way to feel less wretched before lunch. Hold onto that, because by the end of this essay it turns out to be the part that matters most.

It crossed into the embryo and it worked on a schedule. Exposure around the twentieth day after conception damaged the brain. Day twenty-one, the eyes. Day twenty-two, the ears and the face. Day twenty-four, the arms. Through day twenty-eight, the legs. After about six weeks, nothing at all. What determined a given child’s injury was not how much the mother took. It was which morning she took it.

Children were born with hands growing from their shoulders and feet from their hips. Children were born with no arms. Children were born deaf, or blind, or with facial paralysis, or with malformed hearts, kidneys, intestines and genitals — frequently several at once, because a woman who took the tablets across several days collected several days of damage.

More than ten thousand of them worldwide. Roughly four in ten died at or shortly after birth. In Britain, of about two thousand affected babies, half were dead within months. The miscarriages and stillbirths were never counted and never can be.

The ones who lived are in their sixties now. In the United States they have spent those sixty years being told by their own physicians that this did not happen here.

That denial was never ignorance. In September 1962 the British Medical Journal published a paper titled “Thalidomide babies: management of limb defects” — a clinical article on fitting these children with artificial limbs. In English, in a journal American orthopedists and pediatricians subscribed to. While mothers here were being told they had not been given the drug, the literature sitting in their doctors’ own reading rooms was already several papers deep into what to do about the children.

One detail in the reporting is why I am writing about it instead of simply reading it. The American women were handed the drug with no label on it. No prescription slip, no bottle with a name, nothing entered in the chart. When the children were born and the questions started, there was nothing to check.

That is not a footnote to the tragedy. That is the shape of the thing, and it has repeated for sixty years.

Everything That Came Later Is Already in This One

Thalidomide is worth studying not because it was the worst but because it was the clearest. Every element that recurs later is visible in it, uncomplicated:

● The product moved ahead of the evidence. While the application sat unapproved at the FDA, the manufacturer distributed thousands of doses to American physicians anyway, as investigational material.

● The record was never created. No tracking requirement. No obligation to report back. No consent. The absence of documentation was not an oversight in the system; it was the system.

● The profession carried the denial. When the defects appeared here too, doctors did not come forward. Some told patients outright they had not been given the drug. Sixty years later, survivors still get told by new physicians that America was spared.

● The official number became the settlement. The FDA formally acknowledges seventeen American survivors. People who have worked the records think it is closer to a hundred. It cannot be resolved, because — see above — nobody wrote it down.

● The reform was bounded to the failure that got photographed. The 1962 Kefauver-Harris Amendments fixed the drug-approval channel and touched nothing else.

Hold those five. They are the template.

They Did It Again, and This Time There Are Guilty Pleas

What follows is not allegation. Each of these has a court record, a guilty plea, or a published forensic reanalysis behind it.

Vioxx. Merck’s arthritis drug, approved 1999, withdrawn 2004. The pivotal VIGOR trial was published in the New England Journal of Medicine in 2000; in 2005 the journal’s editors issued an Expression of Concern stating that three myocardial infarctions had been deleted from the manuscript before submission. A 2008 analysis in JAMA, built from documents produced in litigation, documented systematic ghostwriting — papers drafted by or for the company and then fitted with academic authors. The FDA’s own David Graham estimated 88,000 to 139,000 excess cases of serious coronary heart disease in the United States. Merck paid $4.85 billion in 2007 to settle roughly 27,000 claims, and in 2011 pled guilty to a criminal misbranding charge and paid another $950 million.

Paroxetine, Study 329. A trial of Paxil in adolescents, published in 2001, reported the drug as safe and effective for that population. It was ghostwritten. In 2015, under the Restoring Invisible and Abandoned Trials initiative, independent researchers obtained the underlying data and republished the analysis in The BMJ: the drug was neither effective nor safe, and harms had been undercounted. The original paper stood uncorrected for fourteen years while the profession cited it. In 2012 GlaxoSmithKline paid $3 billion — then the largest health care fraud settlement in U.S. history — covering off-label promotion of Paxil and Wellbutrin and failure to report safety data on Avandia.

OxyContin. Purdue Frederick pled guilty to felony misbranding in 2007 and paid $600 million. In 2020, Purdue Pharma pled guilty to three federal felonies, including conspiracy to defraud the United States and violation of the anti-kickback statute, in a resolution valued at $8.3 billion.

That last one contains the single most instructive artifact in modern medicine. The claim that addiction was rare in patients treated with opioids traced back to a five-sentence letter to the editor published in 1980 — not a study, a letter, describing hospitalized patients under direct supervision. A 2017 bibliometric analysis in the New England Journal of Medicine found it had been cited 608 times. Seventy-two percent of those citations used it as evidence that addiction was rare during long-term opioid treatment. Eighty-one percent did not mention that the patients had been hospitalized.

Five sentences. Six hundred and eight citations. A generation of prescribing built on a paragraph that never said what it was cited as saying.

Fifty-One Percent Became Ninety-Four Percent

The objection to everything above is that they are anecdotes — four famous failures pulled from a century of practice.

So here is the measurement. In 2008 the New England Journal of Medicine published a study comparing the antidepressant trials registered with the FDA against the trials that reached the published literature. There were 74 registered trials. By the FDA’s own assessment, 51 percent were positive.

In the published literature, 94 percent appeared positive.

Nearly every positive trial was published. Of the thirty-six negative or questionable ones, all but three either never appeared or appeared written to read as positive.

That is not four bad actors. That is the evidence base a conscientious physician consults, differing from reality by forty-three points, in a single therapeutic class, in the open.

The Product Is Not the Pill

Not the drugs. Most of them work, and I will come to that.

The indictment is that the industry’s most reliable product is not a molecule. It is a literature. The recurring moves:

● Author the evidence base rather than argue with it. Ghostwriting, selective publication, and trial designs built to produce a publishable result.

● Convert a citation into a fact by repetition. Nobody has to lie if the original document is thin enough and cited often enough.

● Move the product ahead of the record. Unlabeled doses in 1961; off-label promotion in 2005.

● Let the practitioner carry the denial. The manufacturer is never in the room. The physician is, and is the one who has to tell the patient something.

● Price the penalty as a line item. A settlement assessed a decade later, after the revenue has been banked, booked and distributed, is not a deterrent. It is an installment.

None of this requires anyone to be a villain, and I would distrust any account that needed them to be. It requires only that the party with the largest financial interest in a question also be the party that funds, designs, analyzes, writes and submits the study that answers it. Put any industry in that position and you get this outcome. We ran the experiment.

And Now There Are 1.1 Million People in the Experiment

Everything above is retrospective, which is the comfortable direction to run an argument like this. Here is the uncomfortable one.

The GLP-1 drugs work. In the populations they were approved for, the evidence is strong and still strengthening — weight loss nothing else has matched, reductions in cardiovascular events in trial populations with obesity, improvement in sleep apnea, protection of kidney function in type 2 diabetes. I am not going to pretend otherwise and I would be suspicious of anyone who did.

That is not the open question.

An analysis published this month, drawn from the medical records of 92 million American adults, identified more than 1.1 million people taking these drugs with no record of any condition that would qualify them for one. Off-label prescribing rose fifteen-fold between 2021 and 2025. More than a third of those patients had a normal body mass index. The authors expect the true number is higher, because their data cannot see telehealth prescribing or compounded supply.

Those people were largely excluded from the trials. That is not an accusation — it is how the trials were designed, and designing them that way was correct. They enrolled patients with obesity or diabetes, because that is who the drug was for.

Which means that for the fastest-growing group of users in the country, the long-term safety data does not exist. Not is contested. Does not exist.

One signal is already being argued over. NAION — a sudden and usually permanent loss of vision — has been associated with semaglutide in some analyses and not in others. A company-funded study found no increased risk. A specialized eye center reported nearly an eightfold increase. In 2025 the European Medicines Agency concluded it is a very rare side effect and required a warning on the European label. The United Kingdom, Japan and Australia followed. The United States has no such warning. The FDA has been investigating since late 2024 and will not say when it intends to decide.

Sit with that for a second. Same molecule, same published evidence, a warning in four jurisdictions and not in this one.

But the argument that actually settles this is one the coverage almost never makes, and it is simple enough to state in a line.

What risk a drug may carry is set by what the drug is for.

Early antiretrovirals were brutal. They damaged livers, redistributed body fat, caused neuropathy, and had to be taken on a punishing schedule. Nobody seriously argued they were not worth it, because the alternative was dying at thirty-five. The same reasoning covers chemotherapy, and immunosuppressants after a transplant, and every drug whose side effects would be intolerable if the disease were not worse. A high risk against a high stake is a rational trade, and patients make it knowingly every day.

Thalidomide sat at the other end of that scale. It was a sedative for insomnia and nausea. The benefit on offer was comfort, and comfort buys almost no risk at all. The entire catastrophe consists of a comfort-grade drug being distributed with no testing, because nobody imagined a comfort drug required any.

Now put the 1.1 million on that scale.

For a patient with type 2 diabetes, or with obesity and established cardiovascular disease, a GLP-1 sits at the antiretroviral end. Real stakes, real trial evidence, a rational trade. For a person with a normal body mass index who would like to drop ten pounds, it sits at the other end. Not because the molecule is thalidomide — it is not, and I am not saying it is — but because what is being purchased is comfort, and comfort has never been able to pay for much risk.

A Mayo Clinic neuro-ophthalmologist studying the vision question put it in one sentence: for patients who need the drug to control diabetes or heart disease, the benefits probably outweigh the risk; if they want to lose five pounds, that is probably not worth it.

The difficulty is that both patients are taking the same molecule, reading the same coverage, and receiving the same reassurance. The scale that separates them appears nowhere on the label, in the advertising, or in the telehealth intake form.

Then the questions nobody can answer yet, because the time simply has not passed. What does sustained appetite suppression do to lean muscle mass in someone who had no weight to lose? What does it do across twenty years in a person who started at thirty? What happens at population scale on discontinuation? One of the researchers studying this put it as plainly as it can be put: it is like a big experiment.

I do not know how it turns out. Neither does anyone else, and that is the whole of my point. It may well turn out fine. Most drugs do.

But look at the shape of it, against the five moves above. A compound validated in one population is being consumed by a different one. The people taking it are generating the safety data by taking it. And whatever the finding turns out to be, it will arrive years after the revenue has been banked.

That is not an allegation about anyone’s conduct. It is a description of how we have decided to find out.

Now the Part That Ruins My Own Argument

The same industry took HIV from a death sentence to a managed condition. It cures hepatitis C at better than ninety-five percent. Childhood leukemia survival went from almost nothing to roughly ninety percent within a working lifetime. Antibiotics. Insulin. Vaccines, which remain among the most thoroughly validated interventions in the history of medicine — I said so here last week and the evidence has not changed since Friday.

Those all sit at the high-stakes end of the scale, and at that end the record is extraordinary. Now notice where the failures in this essay sit. A sedative for morning sickness. An arthritis pill for ordinary aches. An antidepressant pushed into adolescents who were never the population studied. A painkiller sold for chronic conditions it had never been tested in. Weight loss in people who were not overweight.

Every one of them is at the comfort end.

That is not a coincidence, and I do not think it is even mainly about greed. Where the stake is survival, everyone is paying attention — the regulator, the investigator, the patient, the press. Where the stake is comfort, the scrutiny relaxes at exactly the moment the tolerable risk approaches zero. The gap between how hard we look and how little risk is actually acceptable is widest for the drugs we take least seriously, and that gap is where every scandal in this piece lives.

The practicing physician is not the industry. In most cases the physician is the last person downstream, reading a literature that someone else curated and being told it is the state of the science.

And the reforms worked where they were tried. Mandatory prospective trial registration — required by medical journal editors from 2005, by federal law from 2007 — is why selective publication can now be measured instead of merely suspected. Open Payments, since 2010, is why anyone can look up what a given physician was paid and by whom. We know about most of the cases above because of reforms of exactly this kind, which is the strongest available argument for more of them.

An indictment that cannot hold both halves at once is not an indictment. It is a mood.

Your Food Never Even Needed the Fake Study

Because the method transferred.

In 2016, researchers publishing in JAMA Internal Medicine worked through the internal archives of the Sugar Research Foundation and documented that in the 1960s the trade group paid Harvard nutrition researchers to produce a literature review on diet and heart disease. The review, published in the New England Journal of Medicine in 1967, minimized sugar and directed attention to dietary fat. The funding was not disclosed. It did not have to be; the journal would not require disclosure for another seventeen years.

That is the pharmaceutical playbook, run earlier, in a sector with less oversight. And it worked better than any drug campaign ever did, because it did not merely sell a product — it set the national research agenda for thirty years and pointed it at the wrong macronutrient.

Here is the asymmetry that ought to end the argument. When a drug company distorts the literature, it is distorting the literature about a compound that at least went through premarket review. A food company does not need to bother. Under the self-affirmed GRAS pathway, a company can convene its own panel, conclude its own ingredient is safe, and put it into the food supply without notifying the FDA at all.

There is no study to ghostwrite, because there is no review to survive.

The Fix Is Boring, Which Is Why We Don’t Do It

Every scandal in this piece resolves into the same short list, and none of it is glamorous: write it down. Register the trial before you run it. Publish it whether or not it worked. Disclose who paid. Make the penalty exceed the revenue, and assess it against the people who made the decision rather than the shareholders who inherited it.

We know this works because in the one place we implemented it — trial registration — it immediately made the distortion visible and measurable.

And the last thing worth saying is the first thing. Those women in 1961 were not lied to about what the pills contained. They were handed something with nothing written on it, by someone who had also written nothing down.

The industry’s durable failure is not that it tells us things that are false. It is that it keeps arranging for there to be nothing to check.

I kept the rule. I kept it through decades of running, through injuries, through a career that ran on early mornings and long weeks. I was not lazy, I was not undisciplined, and I was not looking for an easy answer. By my mid-sixties I was obese and failing anyway — fatty liver, arthritis, chronic fatigue, mental fog, and back pain I had carried since high school.

I wrote a book about why the advice was wrong. Butter in My Coffee follows the guidelines and the money behind them. Then I wrote a second one, More Butter, about how a man gets there while doing everything right. This piece is about the thing sitting underneath both of them, which took me the longest to see.

Satiety is a signal, not a virtue

Your body has a mechanism for telling you to stop eating. It is not willpower and it has nothing to do with character. Stretch receptors in the stomach report volume. Protein provokes a satiety response more reliably than either of the other two macronutrients. Fat slows gastric emptying and triggers the release of cholecystokinin. These are switches, and when they are thrown you stop eating because you are no longer hungry.

Notice what that means. When the mechanism is working, restraint costs nothing. You are not resisting the third helping. You do not want the third helping.

So what is willpower for?

Willpower is what you spend when the signal does not fire. It is the manual override on a system that is supposed to be automatic. Every ounce of it you are forced to expend at a meal is a measurement of how thoroughly that meal defeated the mechanism.

Which turns the usual question inside out. We ask how much willpower a person has, as though it were a fixed property of the person, like height. The better question is how much willpower a food demands. That number is a property of the food.

Engineered to not fire

Now consider what has happened to the food.

I wrote about soda a few weeks ago: sugar in liquid form arrives without chewing, without bulk, without the slow signaling that solid food produces, and you do not eat less at the next meal to make up for it. That is not an accident of the product. It is the product. Every design decision — the acid that cuts the sweetness, the carbonation, the cold — exists to permit a dose your mouth would otherwise refuse.

Extend that across the shelf. Refined carbohydrate stripped of fiber. Protein diluted. Fat and sugar combined in ratios that occur nowhere in nature. Texture engineered so the food disappears before the stomach has registered that anything arrived. A product designed to be eaten in quantity is, by definition, a product designed not to trigger the mechanism that ends eating.

And then, for fifty years, we watched people eat it and could not stop, and we concluded that the people were weak.

The proof arrived in an injection

Here is the part I find hard to get past.

The drugs everyone is talking about work by activating a satiety pathway pharmacologically. That is the whole mechanism. They fire the signal.

Which is the strongest confirmation anyone could ask for that the pathway exists, that it can be switched on, and that switching it on ends the struggle. Fifteen million Americans are now paying to have a signal restored chemically.

Restored from what? Something had to have switched it off, or there would be nothing for the drug to switch back on.

That question has an answer and nobody is asking it. I read a long piece this week about why men with obesity do not seek treatment. The first man in it describes what he was eating — twelve to twenty-four beers a week, constantly ordering pizza — in his own words, in the third paragraph. It is never mentioned again. Not by him, not by any of the six physicians quoted, not in the account of why he got sick and not in the account of what fixed him.

The doctors in that article are right about one thing and I want to give them full credit for it: they say obesity is not a willpower failure. They are correct. But their explanation of how it happened is still entirely behavioral — he drank, he ordered in, he stopped exercising. That is the same causal story with the moral judgment removed. If it were truly a disease, someone would ask what the exposure was.

What I actually did

I am not going to tell you what to eat. That has never been the argument here and it is not going to start now.

But I will describe the mechanism honestly, because it is the part that surprised me. When I changed what I ate at sixty-six, the thing I expected was a fight. Forty years of running had taught me what sustained deprivation feels like, and I assumed I was signing up for more of it.

That is not what happened. Within a matter of weeks I stopped thinking about food between meals. Not because I was resisting anything. Because the question stopped coming up.

I am one man and one man proves nothing. I have been careful about that through two books and I am not going to get sloppy about it here. What I can tell you is what it felt like from inside: the discipline I had spent fifty years admiring myself for turned out to be almost entirely unnecessary. It had never been strength. It had been the cost of a bad input.

The measurement

So here is what I would offer instead of a rule.

If eating the way you eat requires vigilance — if you are counting, resisting, white-knuckling through an afternoon, congratulating yourself on getting through a day — that is information. It is not information about your character. It is a reading on what you are eating.

A body that is being fed what it can read does not need to be argued with.

And if you find that idea uncomfortable, consider who benefits from the other one. As long as the failure is yours, nothing on the shelf has to change. Fifty years of that arrangement produced a country where forty percent of adults have obesity and the emerging solution is a weekly injection that does chemically what a meal used to do for free.

I kept the rule perfectly for fifty years. It nearly finished me. The rule was never the problem and neither was my grip on it.

It was what I was gripping.

Five years ago I was a physical scientist and engineer working in batteries, with a career spent leading teams that brought new chemistries into the market. I was sick and I was functioning. Those two things coexisted for a long time, which is the part that is easy to miss from the outside. Nothing about my life looked like a crisis. I got up, I worked, I traveled, I delivered. I was also, by every measure I later bothered to look at, in trouble.

Then a dinner, when I was sixty-six. I have written about that evening twice now and I will not repeat it here. What followed took about two years, removed seventy pounds, and gave back most of what I had been told was age.

Then a fortuitous change in my work took me, for a significant part of my life, to a small room far from home. A room, and time.

That is the part nobody expects, and it took me a while to see it for what it was. The room gave me hours. The change that came before it had already given me something I needed more than hours, which was a working head. The fog had lifted. I cannot tell you when it went, only that I could hold a problem in mind for a full evening again and I had not been able to do that in years. Everything I have written since is downstream of that. It is also the plainest evidence I have that the change did something real. Not a lab value. Four books that the man who sat down to that dinner could not have written.

So for the first time in a working life I had hours to think about something that was not on a schedule, and a head clear enough to use them. What I thought about first had nothing to do with food. It was a physics question. Why do some things persist and others do not, and what does feedback have to do with it? I worked on it the way I would have worked on any problem at the bench, and eventually I wrote an academic paper.

Then I looked at the paper and asked two questions about it. Who is going to read this? And of what use is it? Almost nobody, and not much. An idea that no one outside the field can reach is not doing any work.

So I wrote it again. Patterns of Persistence was an exercise: take the ideas in the paper and put them into lay language a generally educated person could follow and perhaps apply to their own life, without the jargon and without the defensive hedging that academic writing requires. It was my first attempt at that translation and it taught me most of what I know about doing it.

The ideas would not stay put. Persistence and feedback turned out to have a great deal to say about the origin of life and about what selection is actually selecting for, and that became a second book, The First Natural Selection. I was still inside my own metabolic change while I wrote it. Somewhere in there I began to suspect I had a story worth telling, though I had no idea what shape it should take.

The shape arrived when I finished the evolution work. I could suddenly see the whole line: chemistry to origins to us, and then forward through the last century to what is in the grocery store. It was one continuous path, and it was the path my own body had been walking without my consent. I understood then that my story could not start with me. It had to start there.

So I did the investigation first. What happened to the food supply, who decided it, what they were paid, and what the evidence actually shows about sugar. Then I fit my own history on top of that structure, because a case needs something to hang on and because I happened to have a body that had run the experiment. That is Butter in My Coffee. The argument carries the weight. The story carries the reader.

And then, finishing it, I ran into the thing I had not looked at.

The reason I spent decades getting sicker was not that I was careless. It was the opposite. I did what I was told, thoroughly, without interruption, for most of my adult life. I was rigorous about the wrong instructions and I never once asked who wrote them. The discipline that made me good at my work was the same discipline that kept me aimed at the wrong target, and it did that precisely because I did not question the source.

That is a different book, and it is the one that is out today. More Butter is the memoir. It is about the rule that built me and then trapped me, and about what it cost to find that out.

Something else happened along the way that I did not anticipate. Patterns of Persistence and The First Natural Selection have drawn modest attention, which is about what I expected; they ask a reader to care about persistence and origins, and most people have other things to do. Butter in My Coffee did something else entirely, and I am frankly amazed by it. Readers write to me. They tell other people. They describe their own histories in terms I recognize immediately, because the terms were mine.

Nearly every one of you reading this today read that book first. I am truly humbled by that interest, and I thank you for it. All of you.

I have thought about why the book landed, and I do not think the answer is the book. I think it is that the outcome in it is one a great many people want badly and have been told is not available to them. They are living inside the same conditions I was living inside, and in a good many cases worse ones. They are carrying the same explanation I carried, which is that this is what getting older looks like. When somebody learns that the explanation might be wrong, they become very interested.

That has changed what I intend to do with the time I have. I did not set out to advocate for anything. I was a scientist with a question, and then a man with a story. But what actually moved me was not a treatment, a product, or a program. It was finding out how something worked and then acting on what I found. Education and investigation are the mechanism, and personal change is what they are for. I have become serious about arguing for that, and I will admit to some passion about it.

I want to be careful about how that sounds. I have no medical training. My doctorate is in physical chemistry, not medicine, and I am not anyone’s physician. What I was trained to do, and what I have spent a working life doing, is investigate something thoroughly and draw conclusions that hold up under examination. That is the qualification I bring and it is the only one I claim. The whole of my personal evidence amounts to one body, which is the weakest kind of evidence there is. What I have beyond that is feedback. People write to tell me that something here changed how they eat, or what they asked their doctor, or what they had stopped expecting of themselves. I cannot verify most of it, and I do not know how long any of it holds. But it is a signal, and it says that this work might be doing some good.

I think that is general. I do not believe anyone sustains effort at anything without some return signal that the effort is working. We move toward evidence that we have had a positive effect, and we go slack without it, and whatever else motivation is made of, that is the engine underneath it. I am no exception and I claim no discipline here. I keep doing this because the signal keeps coming back.

That signal is also the mechanism. It returns, it reinforces the thing that produced it, and it makes the next round larger. A book found readers. The readers returned evidence that the book was useful. That evidence is why there is a second book and why there is this page. It is also precisely what the paper in that small room was about. I do not have a better name for the capacity to amplify a positive effect than purpose.

Which is why this publication exists. A book is finished; the questions are not. New evidence arrives. I get things wrong and have to say so. The news produces something that needs an answer, and readers ask what I did not think to answer. Twice a week, this is where that goes. It is the part of the work that does not close.

What it becomes I do not know yet, and I would rather not decide that by myself. Tell me what you want more of and what you can do without. Ask what you did not get answered. Tell me where I am wrong, which is the most useful thing anyone sends me. And if something here is worth someone else’s time, send it to them. That is the amplification I just described, and it is the only way any of this reaches past the people already reading. We can work out the shape of it together.

I did not set out to do any of it. I set out to answer a physics question in a small room far from home, having recently gotten my life back and having no idea yet what to do with the time.

More Butter is available today in ebook, paperback, and hardcover.

They also did better on memory, on movement, on glucose control. And the researchers did the thing you want researchers to do: they ran a second group that simply ate 24 percent less of the same food, matched to what the drugged mice were eating. That group got most of the benefit — but not all of it. On spatial memory and blood sugar the drug did better than eating less, and it did not lower metabolic rate the way restriction did.

So there is something going on beyond appetite. I take that seriously, and if your first instinct was to wave it off as a mouse study, I would ask you to resist it.

My question is a different one, and it is not about the drug.

Both arms ate the same food

Look at the design. One group got the drug. One group ate less. A third ate normally. All three ate the same chow.

The groups differed in quantity. They never differed in composition. Which means the study is silent about composition by construction — not because the researchers were careless, but because that was not the question they asked.

What it can tell you is that the drug beats eating less of that particular food. What it cannot tell you is whether the drug beats eating something else, because that arm does not exist.

What is in the control diet

Standard rodent chow is a grain-based commodity feed. Corn, soy, wheat byproducts, fish meal. It was formulated to grow animals quickly, breed them reliably, and cost as little as possible — perfectly good goals for running a colony, and nothing whatever to do with how long anything lives.

It is also not a fixed thing. Composition varies between vendors and between batches, because the inputs are agricultural commodities that vary with the harvest. It carries compounds nobody is measuring. This is not a secret or a scandal; it is well known, and it is exactly why purified reference diets with defined ingredients had to be invented — because ordinary chow is not reproducible enough to serve as a scientific control.

And yet ad libitum chow is what almost everything gets measured against. Not because anyone argues it is the right diet for a mouse. Because it is what is in the hopper.

Which means the reference condition in all of this work is an animal already losing ground. Hold on to that.

The missing arm was published nine years ago

In 2017 a group at UC Davis put the same strain of mouse on a ketogenic diet — ten percent protein, one percent carbohydrate, the rest fat — starting at twelve months. Median lifespan rose 13.6 percent. And in the aged animals, only the ketogenic group held on to physiological function.

Thirteen point six against twelve. Same strain. Nobody has ever run the two against each other.

Before that turns into a talking point, three things cut the other way, and I would rather say them myself. The Davis mice started at twelve months and the Berkeley mice at twenty — middle age against old age, so not a clean comparison. A companion paper published the same month found that a cyclic ketogenic diet reduced midlife mortality but did not move maximum lifespan at all. And a much larger study from 2014, which fed 858 mice across twenty-five different diets, concluded that the longest-lived animals ate low protein and high carbohydrate — close to the opposite. The Davis authors say plainly in their own paper that their results do not fit that hypothesis.

Three candidate answers about what a mouse should eat. All three published. None ever tested against each other, and now a drug has been tested against none of them.

Ninety interventions

It is worth knowing how crowded this field already is. Rapamycin extends the life expectancy of middle-aged mice by as much as sixty percent. Metformin does it. Acarbose does it. Extreme calorie restriction does it. By one biotech investor’s count, about ninety separate interventions keep laboratory mice alive past their usual expiration date.

Not one of them has produced a human longevity therapy.

That reframes the number in the headline. Twelve percent is not an outlier. It is a modest entry on a long list, in a field whose defining feature is that the list does not translate.

Why this particular result may not survive

The Berkeley study used female mice, one strain, at one site. That is precisely the design the National Institute on Aging built an entire program to correct, after years of single-laboratory results that failed to replicate. That program uses genetically varied mice, tests at three separate sites, and reports the sexes separately — and it reports them separately because the differences are enormous. One compound in that program extended male median lifespan by 22 percent and female lifespan by 5 percent, same protocol, same doses. Another extended male lifespan by 12 percent and did nothing at all for females.

Then there is median against maximum. Median lifespan moves when fewer animals die early. Maximum lifespan moves when aging itself slows down. Those are different claims, and the cyclic ketogenic study found the first without the second. As far as I can tell nobody has reported a maximum lifespan figure for the semaglutide mice, and until somebody does, the honest description is that fewer of them died young.

The one time somebody ran the comparison in people

In 2002 the New England Journal published the Diabetes Prevention Program. Three thousand two hundred and thirty-four people at high risk of type 2 diabetes were randomized to placebo, to metformin, or to a lifestyle program. Average follow-up, 2.8 years.

The lifestyle arm reduced progression to diabetes by 58 percent. Metformin reduced it by 31. The non-drug arm beat the drug by nearly double, and it did so on a modest intervention — seven percent weight loss and a hundred and fifty minutes of walking a week. Ten years out, the gap held: 34 percent against 18.

Now look at what the winning arm actually ate. A low-calorie, low-fat diet. The composition recommended by the dietary guidelines of the day.

So the one large trial that put diet against a drug used the diet the guidelines already endorsed, won by a factor of nearly two, and still never asked the composition question. Twenty-four years later, that question has not been asked at scale in humans by anyone. The money went the other way.

A test I have used before

I wrote in July about a scene in Breaking Bad — Mike, the old fixer, explaining to Walter why he does not do things partway. Years earlier, as a cop, he had answered a domestic call. He wanted to kill the man. Instead he leaned on him, warned him, let him live. Two weeks later the man killed his wife.

The test I took from that scene was simple. Ask what happens if you stop. If the danger comes straight back the moment you lower your guard, you never dealt with it. You were only holding it down.

The Berkeley mice were injected for the rest of their lives. That is the design, and the paper is explicit about it: continued treatment extended lifespan. The effect is the treatment. By the test, that is a half measure, and I said as much in July.

But this study shows me something I did not say in July, and it is worse than what I did say.

In July I wrote that we know what the real threat is and are choosing the half measure anyway. In the mouse room, nobody established what the threat was. The reference diet was never characterized. The arm that would have told them was never run. So this is not a half measure chosen in place of a known cause. It is a half measure standing in for a cause nobody went looking for — which is how a field ends up with ninety interventions and no answer.

Mike’s rule assumes you have already worked out what you are dealing with. He knew exactly what the threat was; that was the whole horror of it. The step that got skipped here is the one before his.

Set the Diabetes Prevention Program against the same test. Ten years after randomization, the lifestyle group was still ahead — 34 percent against metformin’s 18. Whatever that arm did, it was not merely holding something down.

What I am actually arguing

Not that these drugs do nothing. The evidence says they do something, including something not explained by eating less, and I have no interest in pretending otherwise. If you are sick and a drug helps you, take the drug.

What I notice is the order in which the work got done.

I have written before that you cannot extend a life you are still losing. I meant it about a person. It describes the mouse room exactly.

The control animals at Berkeley were not healthy creatures aging on their own terms. They were declining on a commodity feed nobody designed for their health, in a room where the reference condition is whatever costs least to pour into a hopper. The drug slowed the rate at which they lost ground. Twelve percent is a real number, and that is what it is a number about — a deceleration of a loss, measured against a loss nobody thought to examine.

That is most of what longevity research currently is. Not addition. Deceleration.

The Diabetes Prevention Program found the same thing, and nobody described it that way. The lifestyle arm did not add anything to anyone. It stopped something. Fifty-eight percent fewer people progressed to a disease they were already progressing toward. Longevity in that trial was exactly what I have said it is — what remains once the decline stops.

Drugs have been introduced to mitigate the effects of bad food. That is a reasonable thing to attempt. But the baseline they are measured against has never been characterized — not in the mouse room, where the reference diet is chosen for cost, and not in people, where the control arm eats whatever the country happens to be eating that decade. Almost no money goes toward asking what the baseline should be. Enormous money goes toward interventions measured against it.

You cannot know what a medicine adds if nobody established what was being lost, or why it was being lost.

This is not an argument for delay. Run the drug trials. But somebody should be funding the other question, and at present almost nobody is — which is a fact about incentives rather than about science, and worth saying out loud.

The mice cannot tell us what they should have been eating. Neither, so far, can we. We have simply stopped asking.

She is right, and the supplement aisle deserves worse than she gave it. But the sentence has a second half, and I have spent enough years around catalysts to know what it is.

What the thing under your car actually is

A catalytic converter is a ceramic honeycomb, coated with a rough oxide layer to maximize surface area, onto which a very small quantity of platinum, palladium and rhodium is dispersed. Total precious metal in the whole device: a few grams.

Exhaust gas passes over that surface and three reactions happen at once. Carbon monoxide oxidizes to carbon dioxide. Unburned hydrocarbons oxidize to carbon dioxide and water. Nitrogen oxides are reduced back to nitrogen.

None of those reactions would proceed at exhaust temperatures without the metal. The metal is not consumed. It lowers the energy required for the reaction and then it is still there for the next molecule. That is what a catalyst is.

It is a genuine detoxification device, it is remarkably durable, and it asks nothing of you for a hundred thousand miles.

The four ways it dies

Catalysts fail in ways that are well understood, and the distinctions matter more than the fact of failure.

Fouling is deposition. Carbon or oil residue physically covers the surface and blocks access. It is a nuisance and it is often reversible — run the engine hot and much of it burns away.

Sintering is heat damage. Run the catalyst too hot for too long and the tiny metal particles migrate and coalesce into larger ones. The same mass of platinum is present with a fraction of the surface area. Partially recoverable, mostly not.

Mechanical failure is the honeycomb cracking. Rare, and obvious when it happens.

And then there is poisoning, which is the interesting one.

Poisoning

A poison does not cover the catalyst. It binds to it.

Lead is the classic case. Put leaded gasoline through a converter and the lead chemically attaches to the active sites where the reactions occur, and it does not come off. Not with heat, not with solvent, not with time. Sulfur and phosphorus do versions of the same thing.

When you examine a poisoned converter, everything is still there. The honeycomb is intact. The washcoat is intact. The platinum is present in the same quantity it started with. Nothing is missing and nothing is blocked.

It simply no longer catalyzes.

This is not a marginal engineering detail. It is the reason leaded gasoline had to be removed from the American fuel supply before catalytic converters could be fitted to cars at all. You could not have both. The device could not survive the feedstock, so the feedstock had to change first.

Nobody proposed selling drivers a bottle of converter cleaner instead.

Where the analogy holds

I want to be careful here, because a good analogy invites you to push it further than it goes.

The liver is not a catalytic converter. Hepatic injury is not chemisorption onto platinum sites. The enzymes are replaced continuously, the organ regenerates in a way no ceramic honeycomb can, and anyone who tells you the biochemistry maps cleanly onto the surface chemistry is selling you something.

What holds is one thing, and it is the thing that matters.

A detoxification device is defined by its feedstock. Give it the input it was built for and it runs for decades without attention. Give it the wrong input and no downstream intervention repairs it, because the failure is upstream of everything you could add.

And the second thing that holds is the distinction between fouling and poisoning — between the reversible failure and the permanent one.

Three percent and ten percent

The same article that quoted the dietitian gave two numbers, a few lines apart, without naming what separates them.

A three percent reduction in body weight improves liver markers. Reversing scarring takes ten percent or more.

Those are two different failures. Fat accumulating in the liver is the fouling case: the working tissue is intact, the deposit is on top of it, and it drains. Fibrosis is the other kind — the working tissue has been replaced by scar, and nothing drains scar.

The interval between them is the only quantity in this whole subject that actually matters to a person, and nobody can tell you where yours is. It differs between people, there is no routine test that locates it, and no one has funded the study that would map it. You find out afterward, from which side you were on.

I was on the reversible side. I did not know that at the time and I could not have known it.

What changing the fuel actually meant

Here is the part of the converter story that gets left out, and it is the part that matters most.

Nobody solved leaded gasoline by asking drivers to be more careful. There was no campaign urging motorists to choose responsibly at the pump, no pamphlet about willpower at the filling station, no suggestion that a man whose converter failed had brought it on himself.

The lead was taken out of the supply. The Environmental Protection Agency began the phase-out in 1973, converters were required on cars from the 1975 model year, and leaded fuel for road use was finally banned outright in 1996. It was a supply-side removal, imposed, over industry objection, across two decades.

And it worked twice over. The converters survived, and blood lead levels in American children fell by roughly three quarters. Nobody’s character improved. The input changed.

Notice what that decision did not do. It did not tell anyone which fuel to buy. It removed one thing that was destroying the device, and left every other choice alone.

So let me say plainly the thing I have been circling, because the article that prompted this would not say it and I do not want to leave it as an inference.

If a detoxification organ is failing at population scale, the feedstock is the variable. Not the willpower of the people eating it, not the supplement they take afterward, and not the drug that manages what the failure produces. The input is upstream of all of it, which means every intervention downstream is working against a supply that has not changed.

I am still not going to tell you what to eat. I have not done that in two books and I am not starting now. But there is a difference between prescribing a diet and naming the variable, and I have been too careful about the second one.

The lead came out. Nothing has come out of the food.

The second half

So, with respect to a dietitian who answered the question she was asked, and answered it correctly:

Your liver is already your body’s detoxification system. It does not need a cleanse.

And it will not survive the wrong feedstock indefinitely, no matter what you pour in afterward.

There is no product for that. Which may be exactly why it is the half that goes unsaid.

In my book I gave flavor a single sentence. I wrote that people raised on industrial food have never tasted food that wasn’t engineered — that their palates and appetites were shaped by substances bearing no relationship to what sustained human beings for two million years. Then I moved on to seed oils.

That was a mistake. Not because the sentence was wrong, but because it was load-bearing and I left it standing alone. Flavor is not a footnote to the argument about what industrial food did to us. It is one of the cleanest pieces of evidence in the case, and it is the piece that decides who gets blamed.

So let me do the work I skipped.

What flavor was for

Flavor is not a pleasure system. It is an identification system — a chemical assay your body performs on a substance before committing to it.

Consider what each channel was reporting. Sweetness meant carbohydrate: ripe fruit, honey, available energy. Umami meant free glutamate and amino acids, which is to say protein. Saltiness meant sodium and the electrolytes your blood chemistry requires. Sourness meant acid — underripe, or fermenting, or spoiled, depending on what came with it. Bitterness meant alkaloids, and alkaloids meant a plant defending itself with poison.

There is a detail here that took me a while to appreciate. Humans detect sweetness through essentially one receptor. We detect bitterness through roughly twenty-five distinct receptor types, tuned to different classes of toxic compound. Evolution invested twenty-five times more machinery in not being poisoned than in finding sugar. That asymmetry tells you what this system was built for. It was not built for enjoyment. It was a safety instrument.

And taste is only the crude part. What we call flavor is mostly smell — volatile compounds traveling up the back of the throat to an olfactory epithelium carrying on the order of four hundred functional receptor types. Taste gives you five or six coarse categories. Retronasal olfaction gives you a high-dimensional chemical signature, specific enough to distinguish one plant from a closely related poisonous one. That is the real bandwidth of the system, and it is not incidental to what follows. It is precisely the channel the flavor industry learned to write on.

The loop that made it reliable

None of those signals were guesses. They were forecasts, and forecasts are only worth something if they get checked.

Your body checks them. Flavor arrives first; the metabolic consequence arrives minutes to hours later; and the two get compared and stored. Eat something bitter and become violently ill, and you will not go near that taste again — often after a single exposure, often with hours between the meal and the sickness, in flat violation of how learning normally works. Psychologists have studied this for decades under the name conditioned taste aversion, and the reason it breaks the usual rules is that it is not ordinary learning. It is a dedicated circuit, hardwired, because the cost of a slow update on this particular question was death.

It runs the other direction too. Animals develop preferences for flavors that were followed by nutrients they needed, and they do it without being taught. A sodium-depleted animal will seek out salt immediately and without prior experience — the appetite is specific, and it is aimed at the deficiency. Grazing herbivores offered a varied landscape will construct an adequate diet by flavor alone, correcting toward what their bodies are short of. This is not folklore. It is a substantial literature, and its finding is consistent: given honest signals, appetite is a competent instrument.

That last clause is the entire argument of this piece. Given honest signals.

The system worked for one reason and one reason only: flavor molecules physically co-occurred with the nutrients they announced. Sweetness came attached to sugar because the compound and the calorie grew in the same fruit. The correlation was not a convention or an agreement. It was a fact of chemistry, and it held for the whole of our evolutionary history.

• • •

The severing

Then we learned to make flavor separately.

The moment is datable. In 1874 two German chemists synthesized vanillin, the principal aroma compound of vanilla, from a precursor in pine bark. For the first time the signal existed without the thing. Vanilla flavor no longer required a vanilla orchid; it required a reaction vessel.

Everything since has been an elaboration of that afternoon. There is now a global flavor and fragrance industry — a handful of large houses supplying compounds that make industrial food taste like food. On a label their work appears as “natural flavors” or “artificial flavors,” a distinction that concerns the origin of the molecule and not its function, and which is permitted to conceal essentially everything. A short phrase covers a proprietary formulation of dozens of compounds engineered to produce a specific sensory impression.

Here is what I want to be precise about, because the rest follows from it. Industrial food does not overwhelm your sense of taste. It lies to it.

Sweetness arrives without the energy it promised. Savory depth arrives without protein. The aroma of something roasted and nourishing arrives on a scaffold of refined starch and industrial oil that carries almost nothing your body was looking for. The forecast is issued at full strength. The payload does not come.

And the loop cannot close. There is no consequence corresponding to the signal, so there is nothing to learn from — only a promise, then nothing, then another promise. The body does the only sensible thing available to it under those conditions. It keeps eating. It is still looking for what it was told was in there.

Engineering against the off switch

It is worse than passive falsification, because the satiety machinery was targeted deliberately.

Under normal conditions a flavor becomes less pleasant as you consume it. That decline is a mechanism — it is how a meal ends, and how a varied diet gets built instead of a monotonous one. Industrial formulation defeats it. Products are engineered with enough sensory complexity that the decline never arrives, and with textures that dissolve fast enough that the brain never fully registers what was swallowed. The industry has its own vocabulary for this work. The optimization target that maximizes consumption is called the bliss point, and food scientists have mapped the combinations of sugar, fat, and salt that maximize palatability while minimizing satiety. None of this is a conspiracy theory; it is the stated content of the field, described at length by its own practitioners.

If you want the cleanest experimental evidence, it exists. In a controlled inpatient trial, subjects were fed ultra-processed and unprocessed diets matched for calories, sugar, fat, sodium, and fiber, and allowed to eat as much as they wanted. On the ultra-processed diet they consumed roughly five hundred additional calories a day and gained weight; on the unprocessed diet they lost it. Same nutrients on paper. The difference was in the delivery — in the form, the texture, the speed, the signal. That result is very hard to explain if appetite is a character trait. It is trivial to explain if appetite is a sensor reading forged inputs.

• • •

Why the orthodoxy needs you to be the problem

There are two ways to tell this story, and they arrive in opposite places.

In the first, flavor is a reward system and engineered food hijacks it — floods it, overstimulates it, hooks you. Under that account something inside you is responding wrongly. You are a person with a disordered appetite, and the vocabulary follows automatically: compulsive, addicted, out of control, lacking discipline. The problem is your response to food.

In the second, flavor is an identification system and engineered food falsifies its inputs. Under that account nothing in you is malfunctioning. Your equipment is accurate and your body is doing exactly what it evolved to do, at full working order, on forged information. The problem is the food.

Notice which story the official position requires.

Dietary guidance in the industrialized world is built on quantity and personal responsibility. Eat less. Move more. Balance your calories. Every one of those instructions presupposes that the eater is the variable and the food supply is the constant — that the product is inert and the person is the site of the failure. Under that frame, signal falsification is not a policy question at all. It never has to be named, because the eater has already been assigned the fault.

This is why I have stopped thinking of the blame framing as a byproduct of bad nutritional advice. It is closer to the reverse. Locating the disease in the consumer is what makes the engineering safe to continue. A regulator who accepted that flavor is an identification system being deliberately falsified would have to regulate the falsification. A regulator who believes the problem is willpower has nothing to regulate and can issue a pamphlet instead.

The same arrangement funds the research, staffs the advisory committees, and supplies the language. The clinical vocabulary now available to a person who cannot stop eating engineered food describes a defect in that person. There is no comparably common word for a product that reports one thing and delivers another. That absence is not an accident of language. It is the shape of an industry’s interests pressed into the culture’s vocabulary.

• • •

I know how all of this sounds coming from a man who spent fifteen years fat and sick and would very much like it not to have been his fault. I would be skeptical too. So here is the only argument I actually trust, which is that the account made a prediction and the prediction came out right.

If the defect were mine — some flaw of appetite or will — then removing the food would leave the defect in place. I would be permanently hungry and permanently holding on. That is exactly what every previous attempt felt like: white knuckles, arithmetic, and a countdown to the moment I gave in.

That is not what happened. When I took out the food that was lying to me, the hunger left. Not through effort — there was no effort, which is the part I still find remarkable. Appetite came back down to the size of an actual meal and stayed there, and it has stayed there for years. The instrument had been reading forged inputs. Handed accurate ones, it read correctly, and it turned out to have been trustworthy the whole time.

Appetite is not a character trait. It is a readout.

• • •

One last thing, about a phrase I carried around for fifteen years like a diagnosis: that I needed to learn to eat like a normal person.

Normal is a statistical claim. It describes the middle of a distribution, and this distribution is one in which most adults are overweight, better than half of average intake arrives ultra-processed, and metabolic disease drives the mortality and most of the spending. The normal person is the sick person. I had set my target at the center of a collapsing population and called it recovery.

The phrase sounds like health and denotes the middle of a wrecked distribution. It announces one thing and delivers another — which is to say the phrase does precisely what the flavors do.

I have no interest in eating like a normal person. I would like to eat like an intact one.

Those used to be the same word.

I want to write about personal responsibility, and I have to begin by admitting that the phrase makes me uneasy, because I know whose phrase it is.

Personal responsibility is the load-bearing term of every piece of dietary guidance this country has issued. Eat less. Move more. Balance your calories. Every one of those instructions presupposes that the eater is the variable and the food supply is the constant — that the product is inert and the person is the site of the failure. I have spent a book and a good deal of this newsletter arguing that this is backwards: that appetite is a readout rather than a character trait, and that most people’s equipment has been working correctly on forged inputs.

So when someone tells me the answer is personal responsibility, my first thought is that I have heard this before, from people with an interest in my hearing it.

And yet. If I refuse the phrase outright I am left with nothing to say to the person actually living inside the problem. Which means the phrase has to be taken back rather than abandoned — and taking it back requires separating two things that are almost always run together.

Fault and agency are different questions

Whose fault is it? and who can do something about it? feel like the same question. They are not, and nearly all the confusion in this subject lives in the gap between them.

You did not design the food supply. You did not commission the research that told your parents what to feed you. You did not sit on the advisory committee, fund the review, or set the incentives that made the cheap calorie the profitable one. You were handed a map at an age when you could not evaluate it, by people who believed it themselves, and you followed it. On the question of fault, the answer is that almost none of it is yours.

On the question of agency, the answer is entirely different, and it is not comfortable. You are the only party in this arrangement who can act on your own behalf. Not because you deserve the job. Because there is nobody else in the room.

Those two facts do not cancel. A man whose house is on fire because of faulty wiring he did not install still has to leave the building. His innocence is complete and it is also irrelevant to the next ten minutes. This is the part where the political argument and the practical one come apart, and where I think most writing on this subject fails: it picks one and pretends the other is not there. The blame side tells you it is your fault, which is false and useless. The systemic side tells you the system must change, which is true and slow. Neither of them says the thing that is actually true and actually actionable, which is that it is not your fault and you are still the only one who can move.

What you owe yourself is not discipline

If you expect the next paragraph to be about willpower, it is not. I have written elsewhere that discipline was very nearly the thing that killed me — not because I lacked it, but because I had an unusual quantity of it and pointed it at a bad rule for forty years. Discipline is an engine. It does not evaluate the map.

What I owed myself was not more resolve. It was an audit. Once, at any point in four decades, I might have asked of my own breakfast the question I asked reflexively of every technical claim that crossed my desk: how do we actually know this?

I never did, and the reason is worth stating precisely, because it is the same reason you have not. The belief had been installed early, uniformly, by everyone, and it had therefore stopped registering as a belief. It had become terrain. You do not audit the terrain; you walk on it.

So the duty to yourself is epistemic before it is behavioural. It is not eat this and avoid that. It is: find the rules you are keeping that you have never once examined, and examine one of them. Preferably one you are proud of keeping, because those are the ones that have gone furthest without inspection.

That duty has a second half, and it is the harder one: check the result. Not against how you feel about the change, and not against what the authority says should have happened, but against numbers that were drifting before and are not drifting now. I am not asking anyone to believe me. I am asking them to measure.

• • •

The hardest case, taken seriously

The argument I have just made assumes an adult who can, in principle, examine his own assumptions. There is a case where that assumption fails entirely, and it deserves more than a footnote.

A child raised on engineered food never had a map to audit. His preferences were formed by products, his metabolic set points were established before he could read, and by the time he is old enough to ask the question, the answer is already installed in his tissue. Whatever we mean by personal responsibility, it cannot reach backwards into a childhood he did not choose.

I do not think this defeats the argument. I think it relocates it. The responsibility in that case does not belong to the child; it belongs to whoever is feeding him, and to whoever built the environment that made feeding him well the expensive and inconvenient option. But it does mean the thing I am describing is a duty that arrives with capacity, and that capacity is unequally distributed — by age, by income, by geography, by how much attention a person has left at the end of a day. Anyone who states this obligation without stating that limit is selling something.

What you owe everyone else, and what you do not

Now the second half, and I want to be careful here, because the road from social obligation to moralising about other people’s bodies is short and well travelled and I am not going down it.

You do not owe anyone a particular body. You do not owe your neighbours thinness, and nobody is entitled to inspect your plate. Any argument that ends in judging strangers by appearance has smuggled in exactly the premise I spend most of my time attacking — that what you can see is a reliable readout of what you did.

But there is a collective dimension, and pretending otherwise would be dishonest.

The CDC reports that three in four American adults have at least one chronic condition and more than half have two or more; among adults over sixty-five, more than ninety percent have at least one. On cost I want to use the number correctly, because the version in wide circulation is wrong. The CDC’s actual statement is that ninety percent of the nation’s $5.3 trillion in annual health care spending is for people with chronic and mental health conditions — which is not the same as saying ninety percent is spent on those conditions. Estimates of spending on the conditions themselves land closer to forty percent. Forty percent of $5.3 trillion is still more than two trillion dollars a year.

That is a pooled cost. The insurance pool is collective, the hospital capacity is collective, the tax base is collective, and so is the caregiving burden that falls on families rather than institutions. So no, diet is not purely private, and I am not going to pretend I believe it is in order to sound tolerant.

The question is what obligation that actually generates. And I think the answer is narrower and stranger than the usual one.

The obligation is to find out, and then to tell someone

You do not owe society a body. You owe it two things it can actually use.

The first is that you find out. Not that you succeed — that you check. Every person who examines an inherited rule and reports honestly on what happened, including when nothing happened, adds a data point to a public record that is currently supplied almost entirely by parties with something to sell. That is a civic contribution, and it costs you nothing but the willingness to be wrong in public.

The second is that you tell someone. This is the part I have come to think is the whole mechanism, and I did not expect to.

This information does not travel through institutions. Institutions are the slowest and most captured node in the system, and by the time a guideline changes, the change is ratifying a fact rather than creating one. What actually moves is person to person, one at a time, across a table. My own turn came from a dinner with an old friend who was handing me back something I had said to him twenty years earlier and had entirely forgotten saying. That is not a charming anecdote I put at the end of a book for warmth. It is the transmission mechanism, and it is the only one I have seen work.

Which makes the social duty concrete and small. Not advocacy. Not converting anyone. Say what happened to you, accurately, once, to a person who might use it — and then let it go. You are not responsible for what they do with it. You are responsible for the fact that they had it.

• • •

What this does not license

Three limits, stated plainly, because an argument like this decays into a cudgel the moment they are dropped.

It does not license judging anyone by appearance. You cannot tell from the outside what someone is doing, what they have tried, or what they are working against. Most of the people you would describe as looking healthy are not metabolically healthy, and the reverse holds too.

It does not license ignoring access. The change is largely subtractive, and subtraction is cheaper than substitution — but subtraction is not free where the alternative is not stocked. The lever is available to most people and not to all, and it is least available to the people being harmed most. That is a real limit on the claim, not a footnote to it.

And it does not license certainty. I am one man with one sequence and no control group. What I have is an existence proof — evidence that something is possible, not that it is typical. I have tried throughout to keep the line visible between what I observed and what I merely believe explains it, and the moment I stop doing that I become the thing I have been arguing against.

The goal is a smaller machine, not a better-funded one

There is an assumption buried in almost every conversation about American health care, and it is worth dragging into the open: that a better outcome means more of the apparatus. More coverage, more access, more capacity, better financing. Argue about who pays and how much, and you are still inside that assumption.

Look at what we would be financing. An industry whose revenue base is chronic disease, assembled from three parties with an identical structural problem. Physicians are paid for interventions and carry no billing code for prevention. Pharmaceutical companies are paid per prescription, and the most valuable product in the catalogue is one taken daily, indefinitely, for a condition that is managed rather than resolved. Insurers take a percentage of a pool that grows when costs grow. None of those three is villainous. Every one of them is behaving rationally given what it is paid for. And not one of them can shrink the thing, because shrinking it is the one outcome none of them is compensated to produce.

I have watched this exact arrangement run in another species, with the clock compressed and the paperwork public. A third of American feedlot cattle carry liver abscesses caused by a grain ration their digestive systems did not evolve to handle. The veterinary reference manual says plainly that the primary control is management of the diet. The industry standard is a daily antibiotic instead, fed continuously for the whole finishing period, which cuts the abscesses without touching the underlying lesions. Everybody involved knows the cause. The cause is treated pharmaceutically rather than corrected, because the ration that produces the disease is the ration that produces the margin.

That is not a metaphor for our situation. It is the same structure, in a system honest enough to write it down.

And we know what this country can do when the incentives point the other way. Bovine tuberculosis was eradicated from American herds by a federal campaign that inspected essentially every herd and condemned four million animals, because a tubercular cow is an unprofitable cow. Health and profit agreed, and the result was an intervention at a scale human medicine has never come close to matching. The lesson is not that the agencies are incapable. It is that alignment determines everything, and on this question the alignment runs the wrong way.

So the question I care about is not how to build a better sickness system. It is how to need less of one. And the only variable that reduces demand sits outside the system entirely, with the people who are currently its inventory.

Let me say what I am not claiming, because this argument gets recruited into fights I am not having. I am not making a point about insurance, or coverage, or who ought to pay. I am not suggesting anyone be denied care. Trauma, infection, cancer, genetic disease, the accidents of a life — these are not diets and never were, and for them I want the most capable and best-funded medical apparatus that can be built. My claim is narrower and structural: a large fraction of what currently occupies that apparatus is metabolic, that fraction is substantially preventable, and it is the one part the system is constitutionally incapable of reducing on its own.

A smaller machine is not austerity. It is what success would look like. If the chronic caseload fell by half, the correct response would not be alarm about an industry in decline. It would be to notice that we had finally stopped manufacturing the demand.

Nobody is coming

I would like to end on the fact that made all of this urgent rather than merely interesting.

The agencies are not coming. The advice has not been substantially revised through forty years of contrary outcomes, and the incentive structure that produced it is intact. The industry is not coming; it responds to shelf velocity and nothing else, and it will reformulate when a category stops moving and not one day sooner. The medical apparatus is not coming, for the reason just described — it is not built to arrive. And your physician is not coming, which is not a criticism of your physician: he has under twenty hours of nutrition instruction across four years of training, twelve minutes with you, and a vocabulary that closes the file before the question is asked.

That is the situation. Not a conspiracy — an equilibrium, in which every party is behaving rationally given what they are paid for, and the accumulating cost lands on the one party who was never consulted.

So: it is not your fault. You were handed a bad map by people who believed it, in a food supply engineered to defeat the instrument you would have used to check. Nothing about you failed.

And you are still the only one who can act. Not because you deserve the job, but because you are the only one who has it — and because, unlike almost everything else you are told to care about, this one pays you back in weeks and requires nobody’s permission.

Find one rule you have never examined. Ask how you know it. Measure what happens.

Then tell one person.

Which is why most people conclude that individual action is theater. They are right about the arithmetic.

There is one exception, and it is not a small one.

You can change what you eat. No permission, no coordination, no institution in the path. It takes effect the day you do it. And if enough people do the same thing independently, without ever organizing, it forces structural change on the largest industries in the country.

I want to be precise about why this case is different, because “vote with your fork” has been repeated so often it has stopped carrying meaning.

The collective action problem that is not one

Most large problems fail at the individual level for the same structural reason. The cost of acting falls entirely on you, immediately. The benefit is spread across everyone, arrives late, and is imperceptible unless nearly everyone participates. A rational person defects. Most people do. That is not a moral failing; it is the correct response to the incentive.

Diet inverts both terms.

The benefit of changing what you eat falls entirely on you. It arrives quickly. And it is large enough that you notice it without being told to look. It does not depend on anyone else. If you were the only person in the country who changed, you would still receive the whole of it. There is no free-rider problem here because there is nothing to free-ride on.

That structure is rare. I cannot name another lever available to ordinary people where the individually rational action and the collectively transformative action are the same action.

What “immediately” means

I have to be careful here, because I am one person and one person is not evidence.

At sixty-six I was obese, with fatty liver disease, arthritis, digestive trouble, chronic fatigue, mental fog, and back pain I had carried since high school. I had followed the advice I was given. Then I changed what I ate. Within a year I had lost seventy pounds and most of that list went with it. At seventy-one I take no medication and have had no procedures.

That is not a study. It is an existence proof — evidence that the thing is possible, not that it is typical, and I would not build an argument on it alone. What makes it worth stating is the timescale. Not thirty years. Not a generation. Weeks to months for the first changes, and they are not subtle.

No other civic act pays you back in weeks.

How one person becomes a market

The mechanism by which individual choices aggregate is not persuasion. It is demand.

Food manufacturers are indifferent to argument and exquisitely sensitive to shelf velocity. They do not need to be convinced of anything, and they will not be. They need a category to move. When a category stops moving, they reformulate or exit, and they do it without conceding a single point of the debate.

This is the part that gets missed. You are not trying to win an argument with the food industry. You cannot win an argument with the food industry, because it is not the kind of entity that holds positions. You are changing the only input it responds to, and you are doing it in a way it cannot contest.

The order in which things change

Most advocacy starts at the policy end, which is the slowest and most captured node in the system. The actual sequence runs the other way.

Individuals change what they buy. Retailers notice within a quarter, because shelf space is allocated on velocity. Manufacturers reformulate or launch alternatives within a year or two. Commodity agriculture follows manufacturing demand over several seasons, because acreage follows contracts. Healthcare economics shift last among the industries, because their revenue base is the accumulated damage of decades and that inventory takes a long time to clear. Policy changes after all of it, ratifying a fact rather than creating one.

Policy is the caboose. We keep trying to drive the train from it.

The size of the thing

The CDC reports that three in four American adults have at least one chronic condition, and more than half have two or more. Among adults sixty-five and older, more than ninety percent have at least one.

On cost I want to use the number correctly, because the version in wide circulation is wrong and I would rather be accurate than loud. The CDC’s actual statement is that ninety percent of the nation’s $5.3 trillion in annual health care spending is for people with chronic and mental health conditions. That is not the same as saying ninety percent is spent on those conditions. It describes spending attributable to people who have them — including everything else those people are treated for. Estimates of spending on the chronic conditions themselves land closer to forty percent.

Forty percent of $5.3 trillion is still more than two trillion dollars a year. The CDC lists poor nutrition among the short set of behaviors driving the preventable portion of it.

I use the smaller figure deliberately. The argument does not need the inflated one, and anything built on a misused statistic falls over the first time someone checks.

The objections, taken seriously

Cost and access. Real, and unevenly distributed. The change is largely subtractive — it is mostly about what you stop buying, and subtraction is cheaper than substitution. But subtraction is not free where the alternative is not stocked. In a place with one store and no fresh produce in it, this argument has less purchase, and I am not going to pretend otherwise. The lever is available to most people, not to all, and it is least available to the people being harmed most. That is a genuine limit on the claim, not a footnote to it.

Blame. The argument is not that people did this to themselves. Taste is not a reward system to be hijacked; it evolved as an identification system, a way of telling which things in the environment were worth eating. Industrial formulation does not overpower that system. It falsifies its inputs. You were not overriding your biology through weakness — your biology was being given false information, competently and at scale. Having latitude to act is not the same as being at fault.

Feasibility. Large populations do not change entrenched consumption habits, the argument goes, and certainly not without coercion. Except that they have, once, within living memory, on exactly this scale.

The precedent

In 1965, 42.4 percent of American adults smoked cigarettes. In 2022, 11.6 percent did.

That is a seventy-three percent reduction in a chemically reinforced, socially normal, heavily defended behavior, achieved across sixty years. It took decades. It took information reaching people faster than an industry could suppress it. It took that industry losing an argument it had spent a fortune to win. But nobody voted it into existence. It was hundreds of millions of individual decisions, each made for private reasons, aggregating into a structural change that reshaped an industry, a body of law, and the interior of every restaurant in the country.

Notice the order there too. The behavior moved first. The policy followed.

That is the template, and there is no reason in principle it cannot happen a second time.

What this actually asks of you

Nothing that requires anyone else’s cooperation. Nothing that requires an institution to agree. Nothing that requires you to win a debate with a corporation, an agency, or your brother-in-law.

You cannot personally decarbonize a grid. You cannot personally reform an agency. You cannot personally make a food company honest. You can change what you buy this afternoon, be measurably better in ninety days, and — if enough people independently do the same — take the largest industries in the country with you as a side effect of getting your own life back.

It is the only lever I know of that works whether or not anyone else pulls it.

Walk into a grocery store now and it runs the length of the building — cases of it, brands of it, a wall of plastic containing the one substance that used to arrive at your house through a pipe you had already paid for. Nobody voted on this. A market appeared where a public good had been, and we walked into it one bottle at a time.

That aisle is recent. The mechanism that produced it is not. It has been running since 1954, and it has now been applied to three separate industries in sequence. I want to describe it, because once you can see the shape of it you cannot stop seeing it.

One

Coca-Cola launched Dasani in 1999. PepsiCo launched Aquafina in 1994. Both are municipal water — the local supply, pushed through a reverse osmosis membrane, with a proprietary blend of minerals added back in, because the membrane removes the ones that make water taste like anything.

In July 2007, after months of pressure from Corporate Accountability International, PepsiCo agreed to spell out the letters on the Aquafina label. It had said P.W.S. It would now say public water source. Coca-Cola declined to make the equivalent change; its spokeswoman said she did not believe consumers were confused.

Dasani’s British launch is the cleaner case. February 2004, a seven-million-pound campaign, water drawn from the municipal supply at Sidcup in Kent. The press worked out where it came from within days. Then testing found bromate above the British limit, Coca-Cola recalled five hundred thousand bottles, and the brand left the country after five weeks.

None of that is the interesting part. The interesting part is that it worked everywhere else. Bottled water is now the largest beverage category in the United States, ahead of soda, and a substantial share of it is sold by the two companies that spent the previous century selling sugar water.

Which is the first thing to notice about this system. It did not lose the argument about soda. It sold you the alternative.

Two

Soda was never food.

Pour one into a glass, leave it on the counter for an hour, then drink it. You will not finish it. Warm and flat it is syrup — thick, sickly, obviously wrong. Nothing about the liquid changed. The engineering stopped working.

There are roughly ten teaspoons of sugar in a twelve-ounce can. You would not eat ten teaspoons of sugar off a spoon. So the product is built to get it past you: phosphoric acid to cut the sweetness into something tart, carbonation for bite, refrigeration to blunt what is left. Every one of those is a decision, and every one exists to permit the dose. Then there is the delivery. Sugar in liquid arrives without chewing, without bulk, without the slow signaling that solid food produces, and you do not eat less at the next meal to make up for it. The result is a load the body does not register as food.

American soda consumption has fallen for twenty-five years, from over fifty gallons a head at the turn of the century to under thirty-five now. That sounds like a correction. Run the arithmetic. Thirty-five gallons is about four thousand four hundred ounces a year, which is twelve ounces a day, for every person in the country — infants, the elderly, and everyone who never touches the stuff included.

Set that beside the decline everyone likes to cite. Smoking went from about forty-two percent of American adults to about eleven. Three quarters of the smokers stopped. Soda lost a third of its volume and stayed universal.

Three

By 2015 the association between sugar-sweetened beverages and metabolic disease had become difficult to argue with. So it was not argued with. It was reframed.

That year the New York Times reported that Coca-Cola had funded the Global Energy Balance Network, a nonprofit of credentialed academics whose message was that obesity is a problem of energy balance — that Americans worry too much about what they eat and not enough about how little they move. Internal emails obtained afterward under open-records law showed that Coca-Cola had chosen the group’s leadership, drafted its mission statement, and built its website. One participant likened the effort to a political campaign.

When the story broke, Coca-Cola disclosed that it had spent $118.6 million on health research and partnerships in the United States over five years. Among the recipients was the American Diabetes Association.

The University of Colorado returned a million dollars. The network dissolved within months. The message outlived both, because by then it had been installed. Calories in, calories out. Move more. It is not the food. It is you.

If that sequence feels familiar, it should. It was designed in 1954.

Four

In December of that year the tobacco industry ran a full-page advertisement in more than four hundred American newspapers, promising that the public’s health was its paramount concern and announcing the formation of a research committee to settle the open scientific questions. The advertisement was written by a public relations firm. The committee it announced would operate for the next forty-four years, and its function was never to settle anything. It was to keep the questions open.

That is the invention. Not lying about the product — anyone can do that. The invention was funding the appearance of unresolved science, so that the burden fell on the person deciding what to buy. Doubt as a product. Personal responsibility as a defense.

Seven years later the American Heart Association issued its first formal position on dietary fat, and the country’s guidance began to move in a direction it would hold for the next half-century. I am not claiming the two events were connected. I am pointing out that the technique for managing an inconvenient body of evidence was already built and demonstrated by the time nutrition needed one.

And then the same companies bought the food.

From 1980 to the mid-2000s the American processed food supply was owned by the tobacco firms. Philip Morris bought General Foods in 1985 and Kraft in 1988 and merged them into what was then the largest food company in the world. R. J. Reynolds merged with Nabisco. Oreo, Ritz, Kraft Macaroni and Cheese, Lunchables, Kool-Aid, Oscar Mayer — all of it under the same corporate roofs as Marlboro and Camel.

Researchers at Kansas published the empirical test in Addiction in 2023. Foods from tobacco-owned brands between 1988 and 2001 were twenty-nine percent more likely to be fat-and-sodium hyper-palatable and eighty percent more likely to be carbohydrate-and-sodium hyper-palatable than foods that were not tobacco-owned. That window is also when the availability of such foods rose fastest. The authors say plainly that this is an association across an ownership period and does not establish intent, and I will say it too.

Laura Schmidt’s group in San Francisco found the sharper thing. Executives at both companies had developed colors and flavors as cigarette additives, then used the same capability to build children’s beverage lines — Hawaiian Punch, Kool-Aid, Tang, Capri Sun. A former Philip Morris chief executive is on record observing that ultra-processed food and cigarettes were fundamentally similar businesses.

I am not going to tell you a soda is as dangerous to an individual as a cigarette. It is not, and anyone claiming otherwise can be taken apart on the relative risks. Roughly half of long-term smokers die of a smoking-related disease. Nothing in the diet literature comes near that at the individual level.

The comparison is not about the substance. It is about the conduct, and about who was conducting it. This country spent seventy years reckoning with tobacco — the 1954 advertisement, the 1964 Surgeon General’s report, the litigation, the 1998 settlement, the decades of unwinding since. We concluded that an industry will fund science against its own customers for as long as it is permitted to. We wrote the lesson down. And we let the same corporate parents run a version of it on the food supply while everyone was still watching the ashtray.

We learned the tobacco lesson about tobacco.

Five

Now the third act.

There is a class of drugs that restores the satiety signal the food was engineered to defeat. They work. Some people need them, and I am not going to argue otherwise.

But notice what nobody proposes to change. In 2024 Nestlé launched a frozen line built as a companion for people on these drugs, portioned — in its own description — to a medicated appetite. Conagra added a GLP-1 Friendly badge to twenty-six Healthy Choice meals and stated plainly that nothing about the meals themselves had been changed. The badge was the product.

If you want it stated plainly, read this week’s New York Times profile of Novo Nordisk’s chief executive. One section of it is headed customers, not patients. He says the company should serve anyone who wants the drug, without judgment about why they want it. Asked whether Novo is turning into a cosmetics company, he calls the question a distraction. He is not being cynical. He is being accurate about what the company now is.

Then look at why he has to be. Novo’s share price sits about seventy percent below its 2024 peak, when it was the most valuable public company in Europe. It laid off nine thousand people last year, the largest corporate restructuring in Danish history. Nearly all of its revenue comes from diabetes and obesity drugs, and the key American patent on its active ingredient expires early in the next decade. One fund manager quoted in the piece puts the weight-loss market at a hundred and fifty billion dollars.

A company in that position cannot wait for people to get sick. It has to widen the definition of who needs the drug, and it is doing precisely that — a daily pill instead of a weekly injection, a million and a half new users since January, prescriptions written without an in-person visit to a physician, the whole thing reframed around shame rather than disease.

Nobody in that chain is breaking a rule. That is the point. The rules reward exactly this.

So the sequence closes. The food supply produces the disease. The drug manages the appetite. The food companies sell the accompaniment to the drug. Three revenue streams, one problem, and no participant whose earnings depend on the problem ending.

Why the evidence looks thin

There is an objection to all of this that sounds devastating and is not. The evidence linking sugar to disease, people say, is weak — associations, small effect sizes, confounded cohorts. Nothing like the smoking data.

Geoffrey Rose answered that in 1985, in a paper called Sick Individuals and Sick Populations that is foundational in his field and has nobody’s money behind it. A cause that is universally present in a population cannot be detected by studying variation within that population. His own example: if every adult smoked twenty a day, epidemiology would have concluded that lung cancer was genetic. There would have been no contrast to measure, and contrast is the only thing observational research can see.

Tobacco research had non-smokers. Sugar research compares heavy consumers with slightly-less-heavy consumers, and that comparison compresses the measured effect toward nothing regardless of how large the real burden is.

So the universality of the exposure does two things at once. It makes the harm larger and it makes the harm harder to measure, in a predictable direction. Seventy years after the technique was invented, the industry is still pointing at the resulting weak numbers as proof that there is nothing there.

Rose’s second finding is the one that matters now. A large number of people at slightly elevated risk generate more disease than a small number at high risk. Which means treating the tail cannot fix a distribution. You can medicate the sickest and leave the population exactly where it was.

The corner

Here is what I think it adds up to, and it is not a claim about any one product.

The conditions a human body developed under are no longer available by default. Not available at a premium — not available at all without deliberate effort. The water needs treatment. The soil is depleted. The animals are fed on grain. The default calorie is engineered to defeat the mechanism that tells you to stop. There is no longer a version of eating normally that puts a person where their physiology expects to be, because normal has been redefined by the supply.

That redefinition took seventy years and it was not an accident of progress. At every stage there was a company that gained from the change and a body of funded work explaining why the change was fine.

Which means everyone is running a configuration now, whether they know it or not. The only question is whether they chose it.

I am not going to tell you what to eat or what to buy. This has never been that kind of argument, and the moment it becomes one it stops being falsifiable and starts being a regimen. What I will say is that the effort is now mandatory — that a person has to work, and spend, and read, in order to arrive at a baseline that was once simply the world they were born into.

That requirement is the indictment. Not any single company and not any single molecule. The fact that getting back to ordinary is now a project.

You were not the problem. The default was.

That framing contains an assumption nobody states: that what you already have is intact. That your baseline is stable, and the question is how to hold it or extend it.

For a large share of the people reading that advice, it is not stable. It is a process, and the process is running. And addition applied to an active subtraction does not produce a gain. It produces a smaller loss.

This is not a semantic point. Slowing a decline and stopping one are different operations, with different mechanisms and different outcomes, and the literature aimed at ordinary readers almost never distinguishes them.

Two things wearing the same word

What we call aging is at least two processes bundled under one label.

The first is intrinsic. Telomere attrition, cellular senescence, accumulated mitochondrial mutation, the slow degradation of repair machinery. It is real, it is measurable, and at present nobody can stop it. Most of the serious longevity science is aimed here, which is appropriate, because this is the hard problem.

The second is accumulated metabolic damage. Insulin resistance. Hepatic steatosis. Visceral adiposity. Chronic low-grade inflammation and the cluster of conditions that trail behind it. This is also real and also measurable — and unlike the first, a substantial part of it is reversible.

Both arrive on roughly the same schedule, which is why they get the same name. You are fifty-five, something hurts, something has slowed, a number on a lab report has drifted. Whether that is category one or category two is not obvious from the inside, and almost nothing in the culture encourages you to ask.

But the answer determines everything. If what is happening to you is category one, the correct posture is management and grace. If it is category two, the correct posture is to stop it, and management is a slow surrender to something that did not have to continue.

Why the advice defaults to prevention

There are honest reasons, and none of them require anyone to be acting in bad faith.

Healthy cohorts make cleaner studies. Prevention is easier to sell, because a person who never gets sick can never determine whether the thing worked, and will keep buying it. The physicians writing the popular books are mostly in their forties and fifties and in good health; they write from where they stand, and where they stand is upstream of the problem.

And reversal is professionally dangerous in a way that slowing is not. A clinician who says a condition can be slowed is saying something unfalsifiable and safe. A clinician who says it can be reversed has made a claim that can be checked, and has stepped toward territory the licensing boards watch closely. The incentive gradient runs away from the word.

So the reader who most needs to hear that the process can be stopped is the reader least likely to be told.

The evidence that it stops

Type 2 diabetes was described as chronic and progressive for most of my life. The DiRECT trial, published in The Lancet in 2018, put that description under a randomised test. The intervention was dietary. Antidiabetic and antihypertensive drugs were withdrawn at the start.

At twelve months, 46 percent of the intervention group were in remission — HbA1c under 6.5 percent, off all antidiabetic medication — against 4 percent of controls. At twenty-four months, 36 percent remained in remission. Among those who had maintained a loss of more than fifteen kilograms, 70 percent were still in remission.

That is not a supplement study or a mouse study. It is a chronic progressive disease going away in more than a third of a randomised primary-care population, two years out, on diet.

Now the detail I keep returning to. DiRECT recruited participants aged twenty to sixty-five, with diabetes of less than six years’ duration, not on insulin.

Sixty-five. The landmark trial demonstrating that this decline can be reversed excluded everyone older than sixty-five.

I was sixty-six.

I do not think that cutoff reflects a belief that reversal stops working on a birthday. It reflects trial design — comorbidity, confounding, recruitment practicality. But the effect on the evidence base is the same either way. The research establishing that the process can be stopped systematically excludes the age at which most people are living inside it. And clinical practice, quite reasonably, follows the evidence base it has.

What one case can and cannot do

At sixty-six I was obese and failing: fatty liver disease, arthritis, digestive trouble, chronic fatigue, mental fog, and back pain I had carried since high school. Six problems, treated by different specialties, understood as separate.

I changed what I ate. Within a year I had lost seventy pounds and most of that list went with it. At seventy-one I take no medication and have had no procedures. The decline has not resumed.

One case establishes nothing about frequency. It cannot tell you how often this happens or in whom. What it can do is contradict a universal claim, and the universal claim in circulation — that this is aging, that it is one-directional, that management is the ceiling — is exactly the kind of statement a single counterexample damages.

It also raises a question that six separate diagnoses had obscured. When one change reverses six conditions that were being treated independently, either you have had six simultaneous coincidences or they were never six conditions. A chemist finds the second explanation considerably more parsimonious.

The honest limits

Not everything reverses. Cartilage that is gone is gone. Calcified plaque does not decalcify. Lost neurons do not return, and there are cancers and autoimmune conditions where none of this applies at all. Anyone telling you the whole of aging is a dietary artefact is selling something, and I am not.

The claim is narrower and, I think, harder to dismiss: the reversible fraction is larger than the framing admits, and most people never discover the size of their own because nobody suggests they look. The default is management. Management is offered immediately, universally, and without anyone first establishing how much of the problem was ever irreversible.

The question is not answered. It is skipped.

What changes if you take this seriously

The first question stops being what to add. It becomes what is still running — and what would happen if it stopped.

That is a different orientation from almost everything on the longevity shelf, and it reorders the sequence. There is no point optimising the rate of decline of a system that is being actively damaged. You address the damage, and then you find out what your actual baseline is. Only then does the extension conversation mean anything, because only then is there something stable to extend.

Longevity is not something you add to a life. It is what remains once you stop losing it.

The part I did not expect

I would rather have known this at forty. I would have avoided twenty-five years of accumulating something I could not see, and I would not have spent my sixties finding out how far it had gone.

But I learned it at sixty-six, and it was enough. That is the part I did not expect and the reason I write any of this down. The window is wider than the advice implies, and it does not close on a birthday — whatever the inclusion criteria say.

If you are young enough for this to be prevention, it is prevention. If you are past that, it may still not be too late, and you will not find out by waiting to be told.