No Control Group
Your body is running 300,000-year-old software on hardware nobody tested it against.
There is no cohort of humans still eating the 1955 food supply. No control arm, no consent form, no one running the numbers as it happened. We ran the experiment on ourselves and published nothing.
This page is not a diet. It does not tell you what to eat, and it has no opinion about your cupboard. It explains a mechanism — why the thing you reach for is the thing you reach for — and shows every number it used to do it.
By Jesse Meria · September 23, 2026
Sweetness evolved as an honest signal. Fruit turns sweet at the moment its seeds are ready, so following the signal was correct every time it fired. You carry roughly 25 different bitter receptors and a single sweet one — evolution spent far more machinery on avoiding harm than on finding calories, because calories were never the thing that could kill you quickly. Then the signal got measured, and products were built to fire it. The receptor is not broken. The dose changed.
The Mismatch Gap
The whole story is proportion.
Drag the axis. It runs the full 300,000 years, linearly, with no compression — which is why you will not be able to reach the era that engineered your food.
300,000 years of anatomically modern humans
300k years ago
Anatomically modern humans
Every taste receptor you have was already in place. Sweetness meant ripe and safe. Fat meant you would survive the winter.
You cannot drag to the last era. On this axis it is about a fifth of one pixel wide. That is not a rendering problem — it is the finding.
What an hour of effort returns
89× — and every input is a government statistic.
These two bars are not the same measurement. Here is exactly why.
- $37.64 is average, not median, hourly earnings — median is materially lower, so this understates how much work a calorie really costs a typical worker.
- Gross, before tax, and before rent, transport, or any other claim on that hour.
- Excludes shopping, travel, and preparation time — real costs that the forager's number does include.
- A foraging return rate and a wage-purchasing rate are different measurements. They share an axis because they answer one question, not because they are the same thing.
U.S. Bureau of Labor Statistics, June 2026 series CES0500000003 / APU0000715211 / APU0000701111; USDA FoodData Central 169655 / 168894
Kraft et al., “The energetics of uniquely human subsistence strategies,” Science 374:eabf0130 (2021) — cross-cultural mean, n=14 hunter-gatherer societies. The paper defines both gross and net return rates; the sentence reporting 729 does not specify which. The arithmetic is most consistent with gross.
There was no single ancestral diet
A Hadza man eating honey on a foray takes in 614 kcal/hr. Eating berries, 21 kcal/hr. That is a 29× spread inside the ancestral state itself — wider than most of the arguments people have about it.
Berbesque et al., “Eat first, share later,” Evolution and Human Behavior (2016) — Table 3, kcal eaten per hour on foray; 146 person/day follows, 921 hours, 2001–2013. These are self-feeding rates — what a forager eats on the spot per hour — not total acquisition rates. They are not interchangeable with the 729 figure.
Then someone measured it
The signal was located, and priced.
Howard Moskowitz — a psychophysicist, not a physician — spent the 1970s optimizing rations for the U.S. Army and went on to define what the food industry calls the bliss point. In his own words, it is “just that sensory profile where you like food the most.” Add sugar and liking rises, peaks, then falls. The peak is the target.
It would be convenient for this page if Moskowitz agreed that he had built an addiction machine. He does not, and says so plainly:
“Does it make you addicted to food? Absolutely not. It just makes it delicious.”
Howard Moskowitz, Retro Report, “Bliss Point: How Food Companies Make Us Crave Their Products”
We are quoting him against ourselves on purpose. He is alive, he built the thing, and the honest version of this argument has to survive his objection rather than route around it.
What is not in dispute is who was holding the research. Philip Morris owned Kraft and General Foods; R.J. Reynolds owned Nabisco. Peer-reviewed work documents cigarette marketing personnel and playbooks moving directly into the food divisions — and foods sold by tobacco-owned brands between 1988 and 2001 were significantly more likely to be hyper-palatable than competitors’. The same paper finds that by 2018 hyper-palatability had saturated the food system regardless of who once owned what, which cuts against any story where one villain is uniquely to blame.
The part that is genuinely hard
Nobody designed this. That is what makes it difficult to fix.
The satisfying version of this story has an author — a room where someone decided. We looked for that room and did not find it, and we are not going to invent one, because the evidence supports something simpler and less comfortable.
A company optimizing for quarterly revenue, with health costs carried by individuals and taxpayers rather than by the balance sheet, produces exactly what we observe. No meeting required. No plan. Every participant behaving locally rationally, and the aggregate arriving somewhere nobody chose.
A villain would be better news. Villains can be exposed, sued, and legislated against. A system with no author cannot be arrested, and every part of it can pass an audit while the whole continues to do what it does.
Where the line actually falls
The line isn’t sugar. It’s made versus engineered.
Sweet-versus-not-sweet is the wrong axis, and it is the one nearly everyone uses. An all-butter croissant and a shelf-stable pastry engineered to a target and built to survive six months in a warehouse are both pastries. Only one was optimized against you.
The counting rule
One step per irreversible industrial transformation between the living organism and the thing you eat. Harvesting, washing, and transport do not count. Milling, extraction, refining, chemical conversion, and recombination each count once.
- Wild berryPicked. Eaten.0
- Modern applePicked. Stored. Eaten.0
- Whole-grain flourGrain → milled1
- White flourGrain → milled → bran and germ removed2
- Table sugarCane → pressed → clarified → crystallized3
- High-fructose corn syrupCorn → milled → starch separated → acid/enzyme hydrolysis → glucose isomerized to fructose → blended → refined6
A modern apple scores zero — the same as a wild berry. That is the correct answer and the useful one: this axis measures processing, not modernity. Nothing here is an argument that the past was better.
These counts are sensitive to how finely you slice a step — a hostile reader can make corn syrup 4 or 9. The rule is published so you can disagree with it precisely.
This is not a framing we invented for a web page. It is the classification our scoring already runs on — processing first, nutrients second. How we score sets out the whole method, including where we cap a number and why.
What we cut
The claims that did not survive.
A page about evidence should show its own discards. Each of these is repeated constantly in this subject, we wanted to use several of them, and each failed verification.
Seasonal availability, in weeks per year
No study reports weeks-per-year for wild food availability. Only ethnographic season description exists. It cannot be honestly rendered as a number, so it is not in the instrument.
“Farm subsidies made corn syrup the cheapest calorie”
False, and backwards. Farm commodities are ~20% of retail food cost and the corn in a bottle of HFCS is under 2% of its sales price. US sugar policy is trade protection that holds sugar prices UP — which is what made corn syrup competitive. Alston, Sumner & Vosti, Food Policy 33(6), 2008.
“Modern apples were bred to be far sweeter”
Not supported. Measured comparisons find no significant difference in total sugar between wild and cultivated apples. Domestication selected for lower acidity, not higher sugar — modern apples taste sweeter because they stopped being sour.
The 17th-century Stanchi watermelon painting
Cited constantly as evidence that fruit was less sweet. It is not evidence of sugar content — it depicts flesh morphology, and that reading is itself contested. Every source repeating it is popular media.
“Wild bananas were not sweet”
No measured comparison between genuinely wild seeded Musa and Cavendish exists. The claim rests entirely on morphology.
It isn’t poison. It’s a supernormal stimulus.
Tinbergen gave herring gull chicks a fake beak with more stripes than a real one and they pecked at it harder than at their own mother. The instinct was not malfunctioning. It was working exactly as designed, on an input that was never supposed to exist.
So the useful move is not abstinence and not vigilance. It is knowing which object you are holding. Eat it on purpose; just don’t eat it by accident.
That is the entire job of the app: it reads the back of the label and tells you which one this is.