Netflix (or should I say N3tfl1×) got the first season of Numb3rs this week, and I watched the first four episodes with mixed reactions. On one hand, any popular presentation of mathematics in the real world is a treat. But would it be so hard to cut some of the tritest Television Drama Moments in favor of a fuller explanation of the math and some acknowledgement of its limits?
The pilot episode is the best of these. Charlie the mathematician sees a map of attacks by a serial rapist and says, I cannot predict where the next attack will be but I can tell you where he lives. He takes a crash course in criminal psychology from his brother Don the detective, generates a model of the rapist’s behavior, and announces with 96% confidence that the rapist lives in the yellow zone on his map. (Later he insists that 96% is equivalent to certainty.) The map is the output of a single “equation”. In his place I would try several different models and see where their conclusions overlap, but there’s no sign that Charlie even tries varying the weights in his model.
Episode 4 “Structural Corruption” is nonsense. Charlie takes some measurements of a skyscraper’s movement in wind, and announces that the structure is flawed (without saying why he thinks so) but he doesn’t know how. He whips up a software model of the building — it would have to be a very simple model — and extrapolates what would happen in a major earthquake; and from this extrapolation, if I understand right, he infers that the flaw is in the foundation. Huh?
According to Keith Devlin, that first episode is based on a true story.
Or as the youth of today would say, based off of a true story.