Great article by Steve Lohr and Natasha Singer in the New York Times: “How Data Failed Us in Calling an Election.”
Data science is solidly behind many technology advances in the last decade, we’re certainly not giving it up. But the massive failure of the presidential polls reminds me that a) it’s not a mature science as yet and b) it may never be — human beings build their personal assumptions into the predictive models and interpretations. Fascinating. Sobering.
P.S. Call it an over-reaction, but I am never trusting a political poll again.