Make Me Care: Ethical Visualization for Impact in the Sciences and Data Sciences

摘要

Scientists and data scientists have long aspired to eliminate bias from their visualizations. This paper argues that eliminating bias from visualizations is impossible, efforts to do so have negative real-world consequences for people, and that strategically emphasizing bias in visualizations is not only desirable, but also ethical. The growing public mistrust in science has not been helped by efforts to produce visualizations devoid of bias. This paper further argues that ethical visualization can only be achieved by acknowledging and embracing the treacherous nature of data visualization as a medium, committing to an ethics of care in visualization, investigating the potential for both benefit and harm when visualizing specific data, and then employing strategies to mitigate the harm involved in creating, using, and sharing visualizations. These strategies center around consciously crafting a visual frame (ie bias) for communicating data to a given audience. This paper offers 1) a critical lens on the rhetorical nature of visualizations, 2) the Hippocratic oath as a means of committing to maximizing the benefit, and mitigating the harm, done by visualizations, 3) ethical visualization for impact as a practical strategy for taming treacherous visualizations, and 4) compassionate visualizations as the end goal of following ethical visualization practices.