You need just one contradicting observation to invalidate something, while a million consistent observations can never fully prove it.
(Paraphrasing Karl Popper, mentioned in the article).
(Paraphrasing Karl Popper, mentioned in the article).
What epistemological techniques do you use to distinguish intervention A from intervention B and how do you distinguish your experience from the paper that shows the opposite result?
The number of variables are large: including, distressingly, your skill at technique A and technique B. Is the difference in the outcome merely your lack of skill in one? Is the difference merely the lack of skill of the guys observed by the paper?
My parents are also experienced surgeons and the normalizing method they use is to share their findings with fellow surgeons. Patients where they work are much more amenable to having photographs of their surgeries used as discussion material.
Medical science is a primitive science and it is practised with primitive statistics, so most evidence is just missing. Lots of surgical techniques are discussed based on too little evidence with massive true error bars.
The experimentalist world is messy and full of hidden dependencies.
(And to be fair it's usually the people with less experimental experience that take him the most seriously)
No, that's wrong. There are several reasons why a negative result doesn't necessarily disprove a theory or a hypothesis: measurement uncertainties/noise, experimental errors, software bugs, bit flips, ...