Semmelweis
On Changing People's Minds or Why Being Right Is Not a Strategy
There is a story managers tell each other for comfort, usually late in a chat thread, usually after a decision has gone the wrong way. It goes like this. There was once a doctor who discovered that washing your hands stopped women from dying. The establishment mocked him. He died in an asylum. Years later he was proven correct. The moral, delivered with the satisfaction of a man who has just lost an architecture review, is that the world hates truth-tellers.
That is a lovely story. It is also the wrong lesson, told by the wrong people, for the wrong reason.
Ignaz Semmelweis did not fail because the world hates the truth. He failed because he treated truth as a deliverable rather than a starting position. He had one of the most spectacular results in the history of medicine and he handled it so poorly that women continued dying, in the same hospitals, for decades, while he was correct at them.
The numbers were not subtle
Vienna General Hospital in the 1840s ran two maternity clinics. The First Clinic was staffed by doctors and medical students. The Second was staffed by midwives. Admission alternated, which meant the assignment of a labouring woman to one clinic or the other was close to arbitrary.
Death rates in the First Clinic ran around ten percent of mothers, and in bad years considerably worse. In the Second, around four percent. This was not a subtle signal buried in a p-value. It was a difference so large that the women of Vienna had worked it out for themselves without a single chart. They begged to be sent to the midwives. Some, when told they were bound for the First Clinic, chose to give birth in the street instead, correctly calculating that a gutter was safer than a teaching hospital.
Reread that. The patients had already performed the analysis. The profession had not.
In 1847 Semmelweis's colleague Jakob Kolletschka died after a scalpel nick during an autopsy, and his post-mortem findings looked uncannily like those of the women dying on the wards. Semmelweis put it together. Doctors and students went from dissecting corpses to examining living women, carrying something on their hands. He ordered washing with chlorinated lime.
Mortality in the First Clinic collapsed, in some months to zero.
He had the intervention, the mechanism sketch, the before-and-after, and a natural experiment sitting right there in the admissions rota. By any modern standard he had shipped. And it did not take.
A fact with no story attached is just an anecdote wearing a lab coat
Germ theory did not exist yet. Pasteur was more than a decade away, Lister two. The reigning explanations for disease were miasma, atmospheric conditions, and individual constitution - frameworks in which a specific transmissible particle carried on a specific pair of hands had no place to live.
So when Semmelweis said the cause was cadaverous particles, the honest follow-up question from a colleague was - what particles? And the answer was that he did not know. He could not name them, isolate them, show them, or explain why some exposures killed and others did not. He was describing a ghost with excellent attendance records.
This is the part managers skip. We like to imagine his critics as vain men swatting away a graph. Some were. But some were doing the thing we claim to admire, which is asking for a causal account rather than accepting a correlation on faith. A finding without a mechanism is not self-evidently true. It is a claim that requires trust to accept, and trust is a social resource, not a logical one.
His theory also moved. First it was corpses. Then, after an infected patient spread disease without any autopsy involved, it became any putrid organic matter. That revision was scientifically correct and rhetorically catastrophic. To an opponent it looked like a man patching his hypothesis to survive contact with data, which is exactly what good science looks like from the outside and exactly what bad faith also looks like from the outside.
Thirteen years of silence, then a war
Here is where the martyrdom narrative gets uncomfortable.
Semmelweis had powerful allies early. Senior Viennese figures championed him, lectured on his behalf, wrote him up in journals. He had, in effect, sponsorship from people whose reputations could have converted his numbers into consensus.
He then spent roughly thirteen years not publishing a proper account of his own work. He let others describe it, sometimes inaccurately. He left Vienna abruptly for Pest (now part of Budapest, in case you are wondering) without telling the men who had staked their credibility on him. In Pest he drove mortality below one percent, which is a magnificent result and which he communicated to the wider profession approximately as effectively as if he had written it on a leaf.
When he finally published his book in 1861, it was long, repetitive, defensive, and structured like a man arguing with the voices in the room rather than explaining something to a reader who had not been present. Then came the open letters, in which he informed named professors that they were participating in mass murder.
He was, in the deepest sense, right. He was also asking a profession to adopt a practice whose stated premise was that its members had personally killed their patients. There is no version of that message where a middle-aged professor of obstetrics reads it and thinks - fair point, I shall reform. Adoption required confession, and he had priced confession at the destruction of the adopter's entire self-image.
One doctor who did adopt the practice, Gustav Adolf Michaelis in Kiel, confirmed the effect, understood that he had likely killed his own niece, and killed himself. Semmelweis's framing was not merely tactless. It was, for the honest, unbearable. He built a truth that only cowards could survive hearing.
The wound
In 1865 his behaviour had deteriorated badly enough that colleagues, including one of his oldest allies, brought him to an asylum outside Vienna under a pretext. Whether the cause was early dementia, syphilis, or something else is still argued. He was restrained and beaten by attendants, and died about two weeks later, aged forty-seven, of (queue the irony) an infected wound.
He died of sepsis. The thing he spent his life describing killed him in a building his friends put him in, and the profession he was trying to save barely noted his passing.
If you want a fable about the cost of being right, that is the ending. If you want a fable about how being right propagates, note the more brutal detail - the practice did not stick after he left. Wards reverted. Women kept dying for years, in the specific hospitals where the specific proof had already been produced, because the proof had been attached to a man rather than built into an institution.
Compare and contrast, painfully
Joseph Lister faced comparable resistance and won. Not because he was more right. Because he had Pasteur's mechanism to point at, published steadily in the journal his opponents actually read, ran demonstrations, toured, trained disciples who carried the method into other hospitals, and argued with critics as colleagues to be recruited rather than defendants to be sentenced. He built a coalition and an explanation, and the coalition survived him.
John Snow (with an H for all your GOT fans out there), meanwhile, is a useful check on our smugness. He did the careful mapping, the elegant natural experiment, the whole heroic epidemiology, and the authorities of the day still rejected his conclusion. Vindication arrived decades later with a microbe. So no, good methodology alone does not save you either. This is not a story with a clean formula. It is a story about what raises your odds.
What Semmelweis would have had to do instead
Not what would have been fair. What would have worked.
Name the invisible thing and give it a handle. Not knowing the mechanism is survivable. Refusing to offer a usable model is not. He needed a named entity, a testable prediction, and a bridge to the vocabulary his audience already trusted, even if the bridge was later demolished. In engineering terms - if you cannot explain the root cause, you can still hand over a coherent hypothesis with falsifiable consequences. A theory people can attack is more useful to you than a fact they can ignore.
Publish early, in their venue, in their format. Own the record before someone else summarises you badly. Thirteen years of letting friends paraphrase your work is thirteen years of accumulating a distorted canonical version that you will then spend your remaining energy fighting.
Make replication trivially cheap and pre-empt the botched attempt. A failed replication reads as disconfirmation to everyone who was not there. He needed an exact protocol, the concentration, the duration, the scope, the common failure modes, and the negative controls. Ship the runbook, not the result.
Attack the confounders before your critics do. The clinics differed in more than hand hygiene - ventilation, teaching load, the sheer number of examinations per patient. Those objections were not stupid. He should have measured them, ruled them out publicly, and made the alternate-day admission rota the centrepiece of his argument rather than a background detail.
Recruit a sponsor and then do not set them on fire. Change propagates along status gradients. He had the sponsors. He abandoned them, undercut them, and eventually insulted the neighbourhood they lived in.
Design a face-saving exit for the people you need. Frame the past as collective ignorance, not individual guilt. Make the villain the system, the era, the absence of knowledge. Let the professor adopt your protocol and feel like a reformer rather than a confessed killer. This is not dishonesty. It is the difference between a truth people can act on and a truth people must first survive.
Find the constituencies whose incentives already align. The midwives were already winning on outcomes. The mothers had already reached the conclusion. Administrators counted beds and deaths. Newspapers loved a scandal about hospitals killing mothers. He aimed his entire case at the one group that had the most to lose and the most authority to refuse.
Build the practice into the institution, not into yourself. Written standing orders, supply chains for the chlorine, checks at shift handover, mortality posted monthly on the wall where everyone including patients could see it. If your improvement dies when you leave the room, you built a personality, not a system.
Stay functional. This one is not tactical, it is moral. Semmelweis was ground into powder by the gap between what he knew and what he could make happen. Plenty of managers are doing a slow-motion version of the same thing right now, burning themselves down over a decision that went the wrong way in 2023. Martyrdom is not a deliverable. Nobody's mortality curve improves because you were destroyed correctly.
The uncomfortable part
You knew this section was coming.
Everyone who invokes Semmelweis casts themselves as Semmelweis. This is statistically implausible. There is even a term, the Semmelweis reflex, for the dismissal of evidence that contradicts orthodoxy, and it is used far more often by people with a grievance and a theory than by people with a hospital ward and a ten-point drop in deaths.
So before you reach for the analogy, three questions.
Do you have the mortality curve, or do you have a conviction? Semmelweis had numbers so large that his patients had independently detected them. Most rejected proposals are not that. Most rejected proposals are a preference wearing the costume of a finding.
Have you made it cheap for someone to try your thing and easy for them to admit they were wrong? Or have you made adoption expensive and humiliating, and then interpreted the resulting silence as proof of institutional rot?
Who benefits from agreeing with you, and have you actually talked to them? If your entire strategy is convincing the one person most invested in the status quo, you do not have a strategy. You have a grudge with a spreadsheet.
Coda
The lesson of Semmelweis is not that authority is stupid, though it often is. It is not that the truth wins in the end, because it took decades and he was not there for it, and the women who died in the interval were not consoled by the eventual arrival of germ theory.
The lesson is that correctness is an input. It does not execute itself. It has no marketing department, no rollout plan, no change advisory board, and no patience for your dignity. Somebody has to carry it, name it, package it, sponsor it, and wire it into the walls of an institution so it keeps running after the person who found it has left or been broken.
Being right just means you have earned the privilege of doing all of that work. Semmelweis paid for the ticket and then refused to take the trip.
Before changing people's minds, wash your hands. Then go build the coalition.