Distrust: Big Data, Data-Torturing, and the Assault on Science

Smith (economics, Pomona Coll.; The AI Delusion) argues that the credibility of science and scientific methods is today being undermined by three problems in scientific communication: disinformation, data torturing, and data mining. These three problems create the structure for the book. Smith offers the most accessible examples for disinformation, citing ESP tests, the Flat Earth Society, Illuminati, the Deep State, and the 2017 satirical conspiracy story that birds aren’t real. He uses these examples to demonstrate that false stories spread faster than real ones; much of the distrust this creates is directed towards the government. The data-torturing section examines recent United States presidential elections, p-hacking (selective data reporting) in scientific papers and journals, and stories from the British Medical Journal. Overall, this section of Smith’s book ranks as the hardest to understand. The data-mining chapters offer examples from Jeopardy, the stock market, and Zillow, arguing that algorithms create their own set of data accuracy problems. Smith concludes with a skeptical review of artificial intelligence and a list of suggestions to combat all three problems showcased in this book.