Short abstract:

What happens when data is harnessed to explicate rather than obfuscate values, expose rather than perpetuate discrimination, and pluralize rather than homogenize quantified parsings of the social world? I present lessons learned and questions raised by efforts to do “good” with data.

Long abstract:

Data-intensive technologies have been thoroughly critiqued for a number of negative outcomes: they can obfuscate contentious values through the automation of decision-making; they can discriminate against marginalized people through the reification of biases held by their designers and users; they can homogenize sociocultural landscapes through the hegemonic adoption of quantification. But done with care and intention, data-intensive technologies may also be leveraged to the opposite effect. After spending four years ethnographically studying a program called “Data Science for Social Good” and another five years as a lead organizer of said program, I will share stories of projects designed to explicate values rather than obfuscate them, expose discrimination rather than perpetuate it, and pluralize quantified parsings of the social world rather than homogenize them. Drawing on nine years of research and personal experience, plus new data from a recently completed study evaluating the longer-term impacts of these well-intentioned projects, I will present lessons learned and questions raised by efforts to do “good” with data.