When Data Don’t Compute to Equity

It is implored, by one side of America’s political divide especially, that we “follow the science.” And that is good guidance, but is it consistent politics? Perhaps not. Many on the same left side have a problem with, for example, data science. This emerges in a New York Times Magazine interview with Colin Koopman, an Oregon philosophy professor who is publishing books questioning the social impact of algorithms that increasingly guide the digital economy. (https://www.nytimes.com/interactive/2023/03/20/magazine/colin-koopman-interview.html) Those codings, and indeed the data that underlie their results, get in the way of equity, Koopman suggests. “We need to be doing more work on this,” he says, by which he ultimately means legislation or regulation (“engage democratically”) that would curb their application in the various marketplaces that use them for product development and pricing. As the window quote from the piece puts it, “We’re still in this wild west, highly unregulated terrain where inequality is just piling up.” Anyone who follows the financial sector, with its challenged use of credit scoring and insurance rating, will recognize where this larger impulse to tame measurable and predictive data is going. Such use categorizes people in ways that offend widespread fairness and justice sensibilities because it condemns individuals to demographic damnation. Now, a libertarian might respond that unless a business has an irrational desire to sacrifice potential profit in pursuit of racial or other bias, it would seek to refine data as precisely as possible to weed out false signals (as, for example, some banks have done in order to extend loans to worthy applicants who don’t check all the standard boxes). But Koopman has a better idea. “A fuller approach would be reparative with respect to the ongoing reproduction of historical inequalities,” he academically puts it. So: “…systems that would take into account ways in which people are differently situated and what we can do to create a more equal playing field… .” Affirmative action, meet data entry (withhold some identifiers) or data analysis (thumb on algorithm). Maybe the zeitgeist needs to be supplemented: “Follow the social science.”