Two Books About Selling Math and Its Consequences for Inequality

Over at Consumer Law & Policy Blog, Jeff Sovern recently discussed James Kwak's new book, Economism: Bad Economics and the Rise of Inequality, which mounts a convincing case against the blind application of Economics 101 to important policy questions, such as healthcare, international trade, the minimum wage, mortgages and other financial products, and taxes. Kwak details the consequences of "economism," which he defines as "the belief that a few isolated Economics 101 lessons accurately describe the real world." Kwak analogizes using economics in this way to justify widening socioeconomic inequality to prior century's reliance on religion and applications of Darwinian evolution to justify the social order of those times. Part of the lure of economism, and how it can be used as an effective justification, is that it seemingly is grounded in math. And math appears to many as absolute, complicated, and scary. 

Which made me think of another relatively new book, Cathy O'Neil's Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. O'Neil chronicles the repercussions of relying on algorithms fed by big data to assess everything from grade school teachers' effectiveness to credit worthiness to which households politicians should target during election campaigns. When not used properly, these "weapons of math destruction" can entrench and perpetuate inequality.

For instance, e-scores, which take into account information from an individual's browsing history, browsing location (zip code), Facebook friends, etc. These e-scores allow companies to determine a person's potential value as a customer and hyper-segment people. Have a problem with Comcast? If your e-score is high, you are more likely to speak with a representative quickly. If your e-score is low, you are more likely to wait on hold and eventually be transferred to someone located out of the country. More perniciously, e-scores are used by for-profit colleges to identify and market to potential students, funneling desperate people to, as O'Neil writes, "what's usually a false road to prosperity."

One of the main problem she identifies with products like e-scores is that there is no feedback mechanism to correct for inappropriate assumptions and spurious correlations. The result is that these models often serve to reinforce inequality by predicting their own outcomes. O'Neil also discusses U.S. News & World Report rankings, the housing market, crime predictive programs used by police departments, and personality tests used to select employees. Across all her examples, to an outsider, the models seemingly must be characterizing and segmenting people "correctly" because they are based on math. This observation likely seems all the more true because outsiders are not privy to the models' inputs, and are told that it is all just too complex to be explained. And as with Kwak's economism, math again is sold without deeper consideration of its consequences for inequality.