Semmelweis had the data. What he lacked was a world in which it counted as data.
In 1847, Ignaz Semmelweis worked out that doctors were killing new mothers by examining them with hands contaminated by cadaverous material from autopsies. He had his staff wash in chlorine, and mortality fell from 11.4% to 1.3%. The medical establishment ridiculed him. He died in an insane asylum in 1865, from the same kind of infection he had spent his life fighting.
The easy explanation is that 19th-century doctors were arrogant, stupid, or indifferent to their patients. They were not. They were intelligent, educated, and cared about the women they treated. So why did careful people reject a correct idea backed by an almost ninefold drop in deaths?
What the doctors could see
Thomas Kuhn’s answer is that scientific facts are never free-floating. They are understood only inside a paradigm, an accepted predictive framework, and the paradigm does more than organise knowledge. It decides which questions can be asked and what counts as legitimate evidence.
Run Semmelweis’s data through the doctors’ paradigm and it stops looking like proof. Their framework said diseases were too complex to have a single cause, so his theory seemed naive. They knew enough logic to see that his evidence resembled a correlation-causation fallacy: epidemics come and go, and this one happened to go when he started washing. Above all, they had no concept of invisible pathogens. Germ theory did not yet exist, and the explanations they had, bad air, seasonal epidemics, building conditions, could not hold his. Even the contagionists, who already believed doctors could spread disease, rejected him because his mechanism did not match their idea of how spreading worked.
That is why the lock-in is more than stubbornness. Consistency bias plays a part, because once you have announced a position your ego wraps around it. But the deeper mechanism is perceptual. Predictive processing describes it at the level of the brain: the mind runs top-down predictions and only updates when the error is too large to smooth away. The paradigm is the prediction, so evidence that fits it is noticed and recorded, while evidence that contradicts it is dismissed as noise, reinterpreted to fit, or not perceived at all. Two people working from different paradigms can look at the same numbers, see different things, and each conclude the other is ignorant, because there is no neutral ground from which both frameworks are equally visible.
Who converts
The first converts to handwashing were junior doctors and students who had watched it work firsthand. That evidence was visceral, and it could not be carried by a publication. They were also new enough to the field that the old paradigm had not set.
Kuhn’s sharpest observation is that most established practitioners never convert. Anomalies pile up and the accepted story needs more and more exceptions to survive, until newcomers suspect a better framework and eventually supply one. The old guard goes to its graves loyal to the old model, which is the point of Max Planck’s line, usually paraphrased as “science advances one funeral at a time”.
Age has little to do with it, except as a proxy for experience. Older newcomers to a field adopt new paradigms about as readily as younger ones. What locks you in is experience and repetition. The MD&A essay this note draws on makes the point about investors: you learn rules from noisy feedback, the ones that make you money become The Inviolable Laws of Investing, and anyone who contradicts them is subconsciously written off as an idiot. Your success is precisely what makes the alternative look obviously wrong.
The new paradigm does it again
None of this means experts should be dismissed. They are not wrong because they are experts. They are wrong on the specific points where their framework has become load-bearing in a way that hides anomalies, and an outsider who knows just enough to wave away inconvenient evidence is in a worse position than either.
Germ theory eventually made Semmelweis’s evidence legible. The same numbers that had read as noise now proved that doctors had been killing patients for decades. Then, armed with germ theory, doctors made the same mistake again. They insisted that deficiency diseases like scurvy could not possibly be cured by minuscule amounts of a simple chemical, because that did not fit the germ paradigm. The framework that rescued Semmelweis became the one that blinded his successors.
So the practical lesson is a prediction about yourself: the more successful your framework has been, the harder it will be to see when it stops working, because the evidence that could unseat it is exactly the evidence it makes hardest to see. Therapeutic etiology is one countermeasure: put a framework down once it has done its work, before it thickens into the paradigm that hides what it cannot accommodate. The harder question is which of your own Inviolable Laws you would recognise as a paradigm at all.
References:
- Thomas Kuhn, The Structure of Scientific Revolutions
- One Funeral at a Time — MD&A
- Max Planck on scientific truth and generational turnover
- Nick Bostrom, Dragon Tyrant fable