Korva is a scroll of the internet’s best ideas.
Experiments built to kill ideas
Scientists can only imagine something and then try to prove it is not true. An experiment that kills the idea when the idea is wrong is called a killer experiment, and it is the better kind: when it turns out the way the scientist hopes, the idea has survived its hardest test.
Intuition is a poor guide to how a system is wired. A wish that something should happen because of age is lovely, and the whole system could still be configured in a baroque way that makes no intuitive sense. That is why some discoveries land so hard: a system that seemed simple turns out to be complicated.
The payoff is uneven. If an experiment has a 10 percent chance of yielding anything interesting, it has to be done at least 10 times, and really more. Five hours in does not return five units of knowledge. A lot of the time the return is zero, and sometimes five in gets 500.
Why do scientists try to disprove their own ideas?
In 1934, Karl Popper argued in Logik der Forschung that no number of confirming results can prove a general claim, but a single clear counterexample can refute it. On that view, the strength of an idea lies in the tests it could fail. An idea that survives a severe test earns more trust than one that has only collected agreeing observations. Popper called this criterion falsifiability.
What makes an experiment a killer experiment?
In 1964, John Platt described a method he called strong inference in Science. A researcher lists several competing explanations and designs one experiment whose possible outcomes would rule out at least one of them. The result cuts the list down, and the cycle repeats with the survivors. Platt held that fields advancing quickly were the ones using this method.
How did one experiment settle how DNA copies itself?
In 1958, Matthew Meselson and Franklin Stahl grew E. coli in a heavy form of nitrogen, then moved them to normal nitrogen and let them divide. They separated the DNA by density in a spinning cesium chloride gradient. After one generation, all the DNA sat at a single intermediate density, which ruled out the idea that each copy keeps an intact original. After the second, two bands appeared, which ruled out a mixed-up version and left the semiconservative model.
More from Andrew Huberman
