ASHFALL INSTITUTE | SUBDUCTION ZONE
THE COST OF A PARTIAL HIT
On Three Interventions That Worked for the Wrong Reason
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.
CASE ONE: THE AIR
Before germ theory there was miasma — the doctrine that disease arose from foul air emanating from filth, rot, and standing waste. It is now the standard example of prescientific error, invoked whenever someone wants a shorthand for how wrong the past was.
It was wrong. It also worked.
Sanitary reform built on miasma theory cleared refuse, built sewers, drained standing water, and moved waste away from where people lived. Cholera and typhoid deaths fell enormously. The mechanism was imaginary; the intervention was correct, because filth and pathogens occupy the same ground. Acting on the proxy moved the real agent.
This is what makes the case useful, and it is exactly the part the shorthand leaves out.
In 1854, two men independently identified the true cause.
In London, John Snow mapped cholera deaths around the Broad Street pump and concluded the disease traveled in water. In Florence, the anatomist Filippo Pacini examined the intestinal mucosa of cholera victims under a microscope, found the organism, described its relation to the disease, and published.
| Snow, Broad Street | 1854 — ignored in his lifetime |
| Pacini describes the vibrio | 1854 |
| Pacini publishes again | 1865, 1866, 1871, 1876, 1880 |
| Koch re-isolates it | 1883 |
| The organism is named Vibrio cholerae Pacini 1854 | 1965 |
The stated reason Pacini was ignored is not obscurity, incompetence, or bad writing. It is that his peers believed in miasma, and within that belief the finding had nowhere to go. Koch, working thirty years later, did not know Pacini’s discovery existed. It had been forgotten so completely that the man who repeated it had never heard of it.
It took 111 years for the organism to carry the name of the man who found it.
Nothing about this is a failure of capability. Snow’s method was sound and is still taught. Pacini’s observation was correct and is still correct. The field had the answer twice in one year and could not read it, because the theory it already held was producing sewers, and the sewers were saving lives.
CASE TWO: THE ACID
The obvious reply to case one is that it was long ago, before modern instruments and modern review. So take a case with both.
Through the twentieth century, peptic ulcer disease was understood as a disorder of acid, brought on by stress, temperament, diet, and constitution. The treatment followed from the theory: suppress the acid.
It worked. Patients improved.
An industry was built on it.
| Tagamet approved in the US | 1977 |
| Zantac launched | 1981 |
| Warren and Marshall’s work rejected by the Australian Gastroenterological Society | 1982 |
| Lancet publication, after delays finding reviewers | June 1984 |
| Tagamet exceeds $1 billion in annual sales | 1986 |
| Zantac becomes the world’s best-selling prescription drug, outselling Tagamet 3:1 | 1987 |
| Evidence becomes, in the field’s own account, impossible to ignore | early 1990s |
| Nobel Prize in Physiology or Medicine | 2005 |
The first billion-dollar drug in the history of pharmaceuticals treated the acid produced by a bacterial infection.
Robin Warren saw the organism in gastric biopsies. Barry Marshall connected it to the disease. Their society rejected the work in 1982. The Lancet paper was delayed, reportedly for difficulty finding reviewers willing to assess it. Clinicians remained unmoved for most of the following decade, by which time the acid-suppression market was the largest in medicine.
Twenty-three years from the finding to the Nobel. Roughly a decade from publication to practice.
And again: not a failure of capability. The laboratories were modern, the journals were peer-reviewed, the clinicians were competent and were watching their patients get better. Acid suppression genuinely relieved ulcers. It simply did not cure them, because it was not addressing what caused them — and the relief was sufficient to make the question look answered.
People took those drugs in good faith, for years, believing they were treating their disease. They were treating a symptom of an infection, accurately.
CASE THREE: THE SUGAR
The third case has not resolved, which is the reason for including it.
For twenty-five years the United States conducted the largest voluntary dietary intervention in its history, without a tax, a ban, or a law — through guidelines, labeling, reformulation, and sustained public alarm about a single ingredient.
It succeeded.
| US adults, NHANES | |
|---|---|
| Added sugar, % of calories, 1999–2000 | 18.1% |
| Added sugar, % of calories, 2015–2016 | 13% |
| Adults 19–50, 2001 → 2018 | 16.2% → 12.7% |
| Mean intake, all ages, since 2000 | 24.3 → 16.9 tsp/day, −30% |
Carbonated soft drinks fell from a peak near 53 gallons per person in 1998 to about 42 gallons in 2025. Bottled water passed soda to become the country’s largest beverage by volume in 2016.
A thirty percent reduction in a population’s intake of a targeted nutrient, achieved without coercion, is not a failure by any standard ever applied to public health.
And the returns were partial in a specific, recognizable way.
New diabetes diagnoses peaked at 1.7 million a year in 2008 and fell to 1.3 million by 2017 — a 35% decline, arriving about a decade after soda peaked, which is the right interval for a disease reflecting exposure decades back.
Then it stopped. Prevalence rates show no statistically significant change from 2008 to 2021. And obesity did not follow sugar down at all: it climbed through the entire period of the sugar decline and flattened only recently, at a moment coinciding with a class of drugs rather than with anything on a plate.
A third of the new cases stopped. Nothing else moved.
WHAT THE SAME SURVEY WAS RECORDING AT THE SAME TIME
| US adults, 2001–02 → 2017–18 | |
|---|---|
| Added sugar | −30% |
| Ultra-processed food | 53.5% → 57.0% of calories (p < 0.001) |
| Processed, non-ultra | ~10%, unchanged |
| Minimally processed food | 32.7% → 27.4% |
Sugar fell. Ultra-processed food rose. What made room for it was whole food.
These are not two findings from two studies with two methods. They are the same forty thousand respondents, the same instrument, the same waves. Both columns existed throughout and were never read against each other, because the question under examination was sugar, and sugar was going the right way.
WHY THEY ARE ONE EVENT AND NOT TWO
There is a technical detail that joins them.
A food classified as ultra-processed does not change classification when it is reformulated. Remove the sugar from a sweetened beverage and substitute a non-caloric sweetener, and the product remains exactly what it was in category terms. Reduce the sodium; adjust the fat; the same. The classification tracks degree of industrial processing, and altering an ingredient panel is not a reduction in processing. It is processing.
So the thirty percent did not leave the food supply. It came out inside the products, which stayed on the same shelves, bought by the same people, who were told and correctly believed the products had improved.
Whether that improvement carries through is now directly studied, and reformulation does not remove the association between ultra-processed food and chronic disease.
WHETHER IT WAS EVER THE SUGAR
Sucrose is half fructose, and fructose has a glycemic index of about 19. It barely raises blood glucose; it is handled in the liver. The half of sugar that is distinctively sugar is not the half that does what sugar is accused of doing.
Refined starch is entirely glucose, and is not milder:
| Glycemic index | |
|---|---|
| Sucrose | ~65 |
| White rice | 64 ± 7 |
| White bread | ~75 |
| Boiled potato | ~78–85 |
A meta-analysis pooling 25,956 cases across 577,426 participants found white rice associated with type 2 diabetes at a relative risk of 1.16 between extreme intake categories, with 13% higher risk per additional 158 g daily serving above roughly 300 g.
But a starch hypothesis fails its own test. Traditional Japan and Okinawa ate white rice in quantities that would alarm any modern guideline, with low rates of the disease. A molecule cannot be the agent if it behaves differently in two populations.
Which leaves the form of the food — and there is a controlled experiment on exactly that.
In 2019 the National Institutes of Health admitted twenty adults to a metabolic ward for twenty-eight continuous days: a randomized crossover, two weeks on an ultra-processed diet and two on an unprocessed one, eating freely. The diets were matched for presented calories, energy density, macronutrients, sugar, sodium, and fiber.
On the ultra-processed diet, participants ate 508 more calories a day and gained 0.9 kilograms. They lost the same amount on the unprocessed diet.
The sugar was held constant. The effect appeared anyway.
One small trial, measuring two weeks of intake rather than a disease. It is also the only place anyone has separated the variable from the vehicle and watched what happened. When they did, the vehicle carried the effect.
THE LAW
Three cases, two of them closed and one running.
In each, an intervention was built on a mechanism that was wrong. In each, the intervention produced real and measurable benefit, because the thing targeted sat adjacent to the thing responsible — filth beside the pathogen, acid beside the infection, sugar beside the processing.
And in each, the benefit is what made the error durable.
Miasma held for thirty years past Snow and Pacini because the sewers worked. Acid theory held for a decade past Warren and Marshall because the drugs worked. A near miss that produces visible improvement is far harder to abandon than a clean failure, because there is nothing to abandon it for — the intervention is delivering, the patients are better, the numbers are moving in the right direction, and the only thing wrong is the reason.
There was never a moving target. Cholera was always waterborne. Ulcers were always bacterial. Whatever drives metabolic disease has been driving it the whole time.
What happened in all three is that we aimed slightly off, hit something adjacent, and the adjacent hit looked enough like success to keep us aiming there.
WHY MEASUREMENT DOES NOT PROTECT YOU
The reflexive remedy for scientific error is better data. It would not have helped in any of these cases.
The sanitary statistics of Victorian Britain were good, and they showed exactly what they should have shown: clear the filth, the deaths fall. Accurate numbers, correctly collected, confirming a false mechanism.
The clinical evidence for acid suppression was good. Patients on Tagamet improved, and the improvement was real and reproducible and honestly reported.
The American sugar data are excellent. NHANES has measured added sugar continuously for decades, and the thirty percent decline is not in question. The same survey was simultaneously recording the ultra-processed share and the collapse of minimally processed food — in the same waves, from the same respondents.
Both columns were on the table the entire time. Nobody was hiding anything.
A good instrument pointed at an adjacent variable returns good numbers. The numbers are not wrong. They are simply not about the disease, and no amount of precision in the measurement will tell you that, because precision is a property of the instrument and relevance is a property of the question.
Diligence on the wrong variable is not protection. It is only diligence.
WHAT THIS DOES NOT LICENSE
This is not an argument against intervening, and the history refuses that reading outright.
The sewers should have been built. Miasma theory was false and London’s dead fell anyway, in numbers that dwarf almost anything achieved since. The men who built those sewers were wrong about why and right about what, and hundreds of thousands of people lived who would not have.
The acid drugs should have been made. They relieved genuine suffering for millions of people, and they still have uses.
The sugar reduction should have happened. Dental disease alone justifies it, and a third of new diabetes diagnoses stopped.
Nor is it an argument that the people involved were careless. They were the opposite of careless. They identified a plausible agent, built an instrument for it, sustained the effort for decades, and moved outcomes. Carelessness does not produce that.
The claim is narrower and harder: a working intervention is not evidence that the mechanism is right. Those are two questions, and the benefit delivered by the first is the single greatest obstacle to anyone asking the second.
THE TEST
The uncomfortable implication is that partial success is the most dangerous epistemic position available — more dangerous than failure, which prompts revision, and more dangerous than ignorance, which prompts inquiry.
Two questions follow, and they are not difficult:
Is the benefit we are seeing large enough to be the whole of the effect, or only the part that came from standing near the right answer?
And what evidence do we have that the thing we removed is the thing that was causing harm — evidence independent of the fact that removing it helped?
Both are cheap to ask. Neither requires a new instrument. The reason they go unasked is not cost.
It is that an intervention which is working does not feel like a question.
P. A. Moore is the pen name of Pamela King, philosopher and artist. Available through the Ashfall Institute.