The pump handle
In 1854, John Snow removed the handle of the Broad Street pump. He did not conduct a focus group. He did not invite Soho residents to do their own research on waterborne disease. He had a map, a theory, and the authority to act. Six hundred and sixteen people had already died while the miasma theorists — who had done their own research and arrived at confident, deeply felt conclusions about bad air — argued with him in the Lancet.
We start here because the personal verification camp always opens with institutional failures. We would like to establish the alternative’s body count first. Before institutional medicine, people verified remedies personally. They bled fevers. They dosed infants with mercury. Personal verification, applied to questions requiring training the verifier lacks, has a name in the historical record. The name is the seventeenth century.
The method we defend is not blind deference. A class of people spend years acquiring specialized knowledge under structured adversarial scrutiny — qualifying exams, peer review, replication requirements. Their conclusions are not infallible. They are better than yours. A randomized controlled trial with 30,000 participants is not equivalent to a podcast host’s interview with a retired chiropractor.
The personal verification camp recites the litany: Iraq, opioids, Theranos. We know the litany. We helped write parts of it. A hospital that botches a surgery is evidence for better surgery, not for performing your own appendectomy with a YouTube tutorial. The Lancet retracted Wakefield. Name one podcast that has issued a retraction.
Do your own research
has become a permission slip for motivated reasoning with a search bar.
Where we concede ground: The opioid crisis is our Broad Street pump in reverse — every channel we call trustworthy amplified the fraud.
What would change our mind: Populations using personal verification outperforming institutional guidance on hard medical questions.
Read the full synthesis: How should a normal person decide what’s true?