ASHFALL INSTITUTE | SUBDUCTION ZONE
WHAT WAS TAKEN OUT OF THE FOOD
On an Ultra-Processed Childhood, a Field in Hertfordshire, and Fifty Years of Measuring the Wrong Thing
P. A. Moore
Ashfall Institute | Subduction Zone
Written under the collaboration disclosed in Time Is a Wheel and We Are the Ground. Concept, argument and judgment: P. A. Moore. Research and composition: Claude Opus 5.
Fifty-five percent of the calories Americans eat come from ultra-processed food.
For children it is 61.9%.
Those are CDC figures, from the National Center for Health Statistics, covering August 2021 to August 2023. They are not estimates from an advocacy group. They are the government’s own count.
And here is the number that should stop you.
Among adults there is an income gradient, and it is small. Households under 130% of the poverty line take 54.7% of their calories from ultra-processed food. Households above 350% of it take 50.4%. A four-point spread across the entire income range of the country. The wealthiest Americans still eat half their calories from the same place as the poorest.
Among children, CDC found no gradient at all. Their words: the mean percentage of calories from ultra-processed foods did not differ by income level.
Rich children and poor children eat the same way.
This is not a story about poor families. It is a story about the food. If you were preparing to read this at a distance, there is no distance available.
The thing to hold before the numbers start
There are two changes running at once, and they are usually discussed by different people who do not cite each other.
The first is what we eat. Fewer vegetables, more packaged food.
The second is what the vegetables contain. Less than they used to — for several reasons at once, which this essay will take in turn.
These do not compete as explanations. They multiply.
If servings fell by twenty percent and density fell by fifteen, delivered nutrition did not fall by twenty or by fifteen. It fell by about thirty-two. That is two fractions applied in sequence, and it is a number almost nobody calculates, because the two halves belong to separate fields.
A population eating less produce needs each vegetable to be stronger, not weaker.
That is the whole argument. Everything below is evidence for the second half of it.
What it does
In 2019 Kevin Hall and colleagues at the National Institutes of Health did something nobody had managed before: they ran a controlled experiment.
Twenty adults lived at the NIH Clinical Center. For two weeks they ate an ultra-processed diet; for two weeks an unprocessed one; the order randomized. They could eat as much as they liked.
The two diets were matched for presented calories, sugar, fat, sodium, fiber and macronutrients.
On the ultra-processed diet, people ate 508 calories a day more, and gained weight.
Read the matching again, because it is what makes this devastating. It was not the sugar. It was not the fat or the salt, and it was not missing fiber — those were held equal. Something about the processing itself caused twenty people to eat five hundred extra calories a day without deciding to.
That is not a correlation. It is a cause, demonstrated under controlled conditions, published in Cell Metabolism.
Now the outcomes.
Childhood obesity in the United States stands at 21.1% — the highest on record. Seven percent of children are severely obese. Extremely severe obesity rose from 0.32% in 2008 to 1.13% in 2023, more than tripling in fifteen years.
CDC’s report this February found record highs among children and teenagers while adult rates slightly declined. The adults have drugs now. The children are still getting worse.
And there is a sentence that carries this whole section without any statistic at all.
Type 2 diabetes used to be called adult-onset diabetes.
The name had to be changed, because children started getting it.
Between 2002 and 2015 it rose 4.8% a year among the young. One in three new childhood diabetes diagnoses is now type 2. Twenty-eight thousand American children had it in 2017; the projection for 2060 is 220,000.
The response
In children aged eight to eleven, prescribing of GLP-1 drugs — liraglutide, semaglutide, tirzepatide — rose 310-fold between 2019 and the middle of this year.
FDA approval extends to age twelve and above. Every one of those prescriptions is off-label. Off-label tirzepatide has exceeded liraglutide since early 2025. Trials are now evaluating use down to age six.
I want to be careful here, because this is not an argument against the drugs. I take one myself. For a child who already has severe obesity and comorbidities, the injection may well be the right decision, and the physician making it is not the problem.
The question is not whether to treat that child. It is why there are so many of them.
And note what the disparity does. Children’s ultra-processed consumption does not differ by income. Prescribing does — upper-income communities show 55% higher rates.
The exposure is equal. The treatment is not.
A field in Hertfordshire
Now the second half, and it begins in the best-documented agricultural experiment on earth.
Rothamsted has grown wheat in the same field since 1843. It is called Broadbalk. For more than a century and a half they have archived samples of both the grain and the soil.
In 2008 a team analyzed the lot.
Zinc, iron, copper and magnesium in the grain held steady from 1845 until the mid-1960s. One hundred and twenty years, no decline.
Then they fell — and the fall coincides with the introduction of semi-dwarf, high-yielding cultivars.
Modern dwarf wheat from that field carries twenty to thirty percent less zinc, iron, copper and magnesium than the old long-straw varieties.
And here is the part that settles the argument that has run for fifty years.
The archived soil shows those same minerals increasing or holding stable across the same period. Magnesium from fertilizer, zinc and copper from manure, more zinc from atmospheric deposition.
The soil got richer while the grain got poorer, in the same ground, with physical samples of both.
In that field, soil depletion cannot be the explanation. The minerals were there. The wheat stopped picking them up in proportion.
The mechanism is unglamorous and is called the dilution effect. The new varieties produce more grain and bigger grain. The additional mass is largely starch. The minerals are spread through a larger denominator.
Nobody removed the zinc. The wheat was simply asked to be heavier, and it complied.
But I want to be careful with what that proves, because it is narrower than it first appears.
Broadbalk is one field under a hundred and eighty years of careful management, including farmyard manure. It is not a picture of commercial agriculture. What it establishes is that the cultivar effect is real and operates independently of soil — not that soil is fine.
Soil is not fine.
Over-tillage remains a genuine and continuing problem: it destroys structure, accelerates erosion, and burns off organic matter. And the practices that used to repair it — crop rotation, fallow periods, cover crops — worked. They were known to work. In much of commercial row-crop agriculture they have been replaced by continuous cropping with purchased chemical inputs.
Which does not substitute. Synthetic fertilizer supplies nitrogen, phosphorus and potassium. It does not supply the micronutrients, the organic matter, or the soil biology — including the mycorrhizal fungi that mediate uptake of zinc, magnesium and copper, and which are measurably less active in over-worked ground.
So there are three mechanisms running in the same direction: the plant was changed, the soil was degraded, and — a third, from the atmosphere — rising CO2 measurably reduces zinc, iron and protein in wheat, rice and legumes in open-field experiments.
They compound. Like everything else in this essay.
And notice the shape of the soil story, because we have seen it twice already. A practice that worked, cost nothing but time and land, and could not be sold — replaced by a product that could be. Rotation and fallow are the agricultural equivalent of a rhubarb crown. Nobody patents leaving a field alone.
The date
Hybrid corn went from one percent of American acreage in 1933 to more than ninety-five percent by 1960.
Semi-dwarf wheat entered US germplasm in the 1950s, reached the Pacific Northwest in 1961, and spread through the decade.
Broadbalk’s decline begins in the mid-1960s. These are the same event.
The tomato
The clearest case is one we can name at the level of the gene.
Almost every commercial tomato carries a mutation called uniform ripening. It was bred in because it makes fruit color evenly, which matters for mechanical harvesting and for how a tomato looks in a box. It was selected for appearance and logistics.
In 2012 researchers identified what the mutation actually does. It disables a transcription factor called SlGLK2, which governs chloroplast development in the fruit. Chloroplasts drive photosynthesis as the tomato ripens. Fewer chloroplasts mean less sugar and less lycopene — the flavor, and a carotenoid with established health relevance.
The lead researcher described these as traits that had been unknowingly bred out of modern tomatoes.
Unknowingly. Nobody decided to remove the lycopene. They selected for even color, got the loss for free, and it stayed invisible for roughly seventy years until somebody sequenced the gene.
That is not a story about greed. It is a story about a trade that could not be seen at the moment it was made, because nothing was measuring the other side of it.
Why nothing was measuring
There is an institution. USDA FoodData Central, with a laboratory analysis program running since 1997.
Three facts about it.
Sampling a single food nationwide and analyzing it costs seventeen to twenty thousand dollars. Per food.
More than half of that laboratory’s food-analysis funding comes from external sources rather than appropriation. USDA’s own published assessment says internal and external funding cannot keep pace with the demands of the food system.
And it is a composition database, not a surveillance system. It is built to answer what does a tomato contain. It cannot answer does a tomato contain less than it used to, because it does not sample the same varieties on a schedule and track them across time.
The data is not hidden. The instrument was never built.
And it sits inside a department holding two statutory mandates at once: promote American agriculture, and advise the public on diet. The measuring function is housed inside the interested party.
Whether that is why it was never built is an inference, and I will mark it as one. What is not an inference is that Cochrane has quantified the funding effect across the medical literature and found it survives every standard check for bias — when analysis was restricted to the best-designed studies, the relationship between sponsorship and conclusions grew stronger. It does not require fraud. It lives in which question gets asked.
Who is paying
Federal agribusiness lobbying ran $181.7 million in 2024 and $158.2 million in 2025. In the first three months of 2026 alone it was $58.5 million.
Look at where it comes from. In 2024: agricultural services and products, $45.4 million. Food processing and sales, $38.1 million. Crop production and basic processing, $32.8 million.
Now the sectors that actually farm. Livestock, $5.4 million. Dairy, $7.8 million. Poultry and eggs, $1.55 million.
Roughly seven dollars from the people who sell to farmers and buy from farmers for every one dollar from the people who farm.
Tobacco spent $32.4 million — more than dairy, livestock, poultry and miscellaneous agriculture combined. A crop nobody eats outspends every food-producing sector in the country.
And the most prominent farm organization in America calls itself the Voice of Agriculture. It claims about six million members. The United States has roughly two million farmers. In many states, buying Farm Bureau insurance makes you a member automatically. In Missouri, fewer than a third of members farm, and all 113,000 pay dues.
Fifty-four Farm Bureau-affiliated insurance companies earn more than $6.5 billion a year in net premiums. Those affiliates have bought stock in ConAgra, Dow Chemical, Kraft Heinz, Kroger and Tyson.
The organization that speaks for farmers owns equity in the companies that buy from farmers. The price a farmer receives is a cost to Tyson. An alliance of a hundred and eighty farm groups has said so in writing.
Meanwhile row-crop producers are down $34.6 billion this year before insurance, farm debt stands at $624.7 billion, and the debt-to-asset ratio is the highest in over a decade.
The sector that is going bankrupt is not the sector doing the spending.
And what happened to the plants nobody could own
There is a second failure, and it is quieter.
Every garden used to have a patch of rhubarb. You almost cannot buy it now.
Rhubarb did not fail in the market. It never entered it. It is a perennial that returns for thirty years, needs almost nothing, and produces more than one family can use. You propagate it by digging up a crown and splitting it. Nobody bought it because everybody had it, so no industry ever formed to supply it — and when the gardens went, it simply left.
And you cannot patent a plant that people divide with a spade and hand over the fence. The Supreme Court case that prohibits replanting saved seed has nothing to say about a crown.
So the same system produces two different deaths. What enters the market is selected for yield and appearance and loses its contents. What cannot be enclosed is never selected at all, and loses its existence. The tomato is on every shelf, missing its lycopene. Rhubarb kept whatever it had and disappeared.
This is not sentiment about old gardens. A 2020 meta-analysis covering 613 cultivated perennial vegetables across 107 botanical families found that perennials represent between a third and a half of all cultivated vegetable species — and occupy six percent of the world’s vegetable cropland. Many carry higher levels of exactly the nutrients populations are short of. The paper notes they receive little attention in the scientific literature.
Half the species. Six percent of the ground. More nutritious. That is not consumer preference. It is a plant that cannot generate a recurring transaction.
The part that proves it was a choice
Breeding food for higher nutrient density is not speculative. It has been done, at national scale, and it works.
HarvestPlus has delivered biofortified rice and wheat for zinc, maize for zinc and provitamin A, beans and pearl millet for iron. More than 640 million people across thirteen countries are eating the results. In Pakistan, over 198 million people — seventy-six percent of the population — eat zinc-enriched wheat, including 31 million children under five.
It was paid for by the United Kingdom, which provided 41% of the funding, and the Gates Foundation, which provided 30%. Governments and philanthropy.
Not by the four companies that control roughly 56% of the global seed market and own around 90% of commercial varieties.
They have been busy. You can look up the patents: US 9,518,256 and US 9,725,712, both titled “Tomato with improved shelf-life.”
Unknowingly, in 1950. There are patents for shelf life now.
And nobody is obliged to breed for a trait nobody pays for. Shelf life appears on a balance sheet. Yield is measured per acre. Zinc is invisible at the point of sale, carries no premium, and appears on no invoice.
The system optimized what it could count. What it could not count was allowed to fall.
One thing that went right
I want to put something hopeful here, and it is real.
Peanut allergy in American children quadrupled between 1997 and 2010. The cause turned out not to be the food system at all. It was official advice — guidance to avoid peanuts in early childhood, which we now know was backwards.
Somebody measured it. The LEAP trial reported in 2015, the guidance reversed, and among children born since: a 43% reduction in peanut allergy and a 36% reduction in food allergy overall. An estimated forty thousand children spared in a decade.
Official dietary guidance can be wrong at population scale, with a body count. And when the instrument exists, it gets corrected.
That is the entire ask of this essay in a single example.
What you can do tomorrow
Some of this has to be structural, and I am not going to pretend otherwise. But you have to feed someone in the morning.
Free, or nearly:
Eat the parts that get thrown away. Beet greens, carrot tops, broccoli leaves, chard stems, the outer leaves of cabbage. Concentration is frequently higher there than in the part you paid for, and they arrive attached.
Buy frozen. This one runs against intuition. Frozen vegetables are picked ripe and processed within hours. Fresh produce is picked early to survive shipping and loses nutrients across days of transit and shelf time. Frozen is often more nutrient-dense and usually cheaper.
Vary what you buy rather than buying more of it. The spread between varieties is enormous — in many cases larger than the historical decline this essay describes.
A few dollars, once:
A rhubarb crown costs between five and fifteen dollars and produces for thirty years. Asparagus likewise. Sorrel. Lovage. Perennials are the cheapest food per year that exists, and they are the category that is both more nutrient-dense and almost entirely absent from commerce.
This is the one part of the system a person can exit completely. Not as a gesture — as arithmetic. One purchase, three decades of yield, no annual seed, nobody’s patent.
Open-pollinated seed reproduces itself. Hybrid seed does not. A single packet becomes a permanent supply, and many public libraries now lend them free.
Einkorn, emmer, spelt and rye predate the semi-dwarf transition entirely. They are the long-straw varieties Broadbalk was measuring before the decline began.
What this does not do
None of it solves the problem.
Four companies, most of the world’s seed, and a measuring function housed inside the department that promotes the industry. A family growing rhubarb does not touch any of that, and the standard failure of writing like this is to end with individual action dressed up as a solution — which is false, and which quietly relocates responsibility onto the person with the least power in the arrangement.
What individual action does do is smaller and still real. It improves that household’s intake. It keeps varieties and knowledge alive that would otherwise go. And in the case of seed saving and perennials, it maintains a stock of genetics outside the enclosure, which matters at a scale beyond any kitchen.
The advice at the end of this essay is the same advice anyone would have given in 1950. Eat more vegetables. Eat more kinds of them. Grow something if you can.
What has changed is the reason.
It is no longer only that vegetables are good for you. It is that each one is carrying less than it did, at the exact moment we are eating fewer of them — and that the shortfall is the product of the two, not the sum.
Nobody set out to take the zinc out of the wheat. There was no meeting. It was simply never on the list of things anyone was counting, and for fifty years the absence of a measurement was mistaken for the absence of a loss.
We changed the plant. We worked the soil harder and bought back the part of it that came in a bag. We did both for reasons that made sense on a balance sheet, and nothing was watching the other side of the trade.
Every garden used to have a patch of rhubarb.
Nobody decided to stop growing it. There was simply never a reason on anyone’s books to keep it, and one spring it wasn’t there.
Food is our most basic need other than breathing.
We celebrate it. We build family traditions around it. We take pleasure in being the breadwinner and the cook. Our hearts swell when we watch our children smile with delight as they take their first bites — and we don’t want to worry about what is in those bites.
Our lives are busy. Too busy. And all most of us want at the end of a day is to share a meal with the people we love.
We can get there again, but it will take effort.
I believe we can come together over this one issue.
Sources: Ultra-processed food consumption figures from NCHS Data Brief No. 536, CDC, August 2025, covering August 2021–August 2023, including the finding that consumption among youth did not differ by income level. Hall, K. D., et al., “Ultra-Processed Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled Trial of Ad Libitum Food Intake,” Cell Metabolism, 2 July 2019. Childhood obesity prevalence and severe obesity trends, CDC, February 2026. Youth type 2 diabetes incidence and projections, CDC and SEARCH for Diabetes in Youth. GLP-1 prescribing in children aged 8–11 and adolescent prescribing rates, published analyses of claims data, 2025–2026. Fan, M.-S., et al., “Evidence of decreasing mineral density in wheat grain over the last 160 years,” Journal of Trace Elements in Medicine and Biology 22:315, 2008, on the Broadbalk Wheat Experiment at Rothamsted. Hybrid corn adoption figures and semi-dwarf wheat timeline, USDA and contemporaneous agricultural histories. Powell, A. L. T., et al., “Uniform ripening Encodes a Golden 2-like Transcription Factor Regulating Tomato Fruit Chloroplast Development,” Science, 29 June 2012. Davis, D. R., Epp, M. D., & Riordan, H. D., “Changes in USDA Food Composition Data for 43 Garden Crops, 1950 to 1999,” Journal of the American College of Nutrition 23(6), 2004. USDA FoodData Central and the National Food and Nutrient Analysis Program; USDA’s own 2012 assessment of the database’s viability. Lundh, A., et al., “Industry sponsorship and research outcome,” Cochrane Database of Systematic Reviews, Methodology Review MR000033. Federal lobbying figures and sector breakdown from OpenSecrets, 2024–2026 cycles. American Farm Bureau Federation membership, insurance revenue and equity holdings as reported by Investigate Midwest, The Nation, and the Institute for Agriculture and Trade Policy; the “Amber Waves of Gain” report. Farm income, debt and bailout figures from USDA, Farm Bureau, Farmer Mac and the American Soybean Association. Toensmeier, E., et al., “Perennial vegetables: A neglected resource for biodiversity, carbon sequestration, and nutrition,” PLOS One, 2020. HarvestPlus delivery figures and funding composition, IFPRI and CGIAR. US Patents 9,518,256 and 9,725,712. Peanut allergy prevalence, the LEAP trial (2015) and subsequent reductions in peanut and food allergy among children born after the guidance change.
A note on what is left out. I have not claimed that declining nutrient density causes any specific disease. The individual links are documented — deficiencies impair health, the deficiencies are prevalent, and the food delivers less than it did — but the chain connecting them cannot be separated from a dozen other things that changed across the same fifty years, and an essay that asserts it would be answerable in one line. I have also left out gastrointestinal illness, which belongs in this argument and which I have not researched properly enough to include. And I have stated the methodological objection to the 1950–1999 comparison in the body rather than burying it, because the Rothamsted data does not depend on it and the argument is stronger without the weaker evidence propping it up.
P. A. Moore is the pen name of Pamela King, philosopher and artist. Available through the Ashfall Institute.