Impact cannot be measured, and other sad half-truths about impact measurement

Abstract

• Not everything that can be counted counts, and not everything that counts can be counted. In this talk, I look at bringing algorithmic fairness to impact measurement, from web-scale attention tracking to computer-assisted data story-telling. Drawing on my experience with altmetrics, I argue that many proxies for impact correlate not only with impact but also with privilege, and that all stakeholders should therefore disclose the inner workings and the limitations of their information systems to pay off (some of) their fairness debt.