On doing science

Mysteries vs puzzles


“Puzzles … are problems for which there is not enough information. … Mysteries are problems for which we have plenty of information, but no accurate analysis.“

Gladwell gives an example of WWII, spies vs. the “screwball division.“ Spies went to gather more information, screwball division analyzed already available data. Screwball division figured out the V-1 rocket had been developed, had been stalled, and accurately projected when it would be finished.

It seems like much of medical ML is this - we have SO MUCH data, but it hasn't been properly cleaned and analyzed. It's also often siloed and unavailable to the “screwball division.“

Fri 24 Apr 2020 08:13:53 AM CDT