Terms of inclusion: Data, discourse, violence

Abstract

Inclusion has emerged as an early cornerstone value for the emerging domain of “data ethics.” On the surface, appeals to inclusion appear to address the threat that biased data technologies making decisions or misrepresenting people in ways that reproduce longer standing patterns of oppression and violence. Far from a panacea for the threats of pervasive data collection and surveillance, however, these emerging discourses of inclusion merit critical consideration. Here, I use the lens of discursive violence to better theorize the relationship between inclusion and the violent potentials of data science and technology. In doing so, I aim to articulate the problematic and often perverse power relationships implicit in ideals of “inclusion” broadly, which—if not accompanied by dramatic upheavals in existing hierarchical power structures—too often work to diffuse the radical potential of difference and normalize otherwise oppressive structural conditions.

Representativeness in Statistics, Politics, and Machine Learning

“…Just as proclaiming Gallup polls "representative" could paper over methodological details and questions about who most benefited from them, ML "participationwashing" might confer legitimacy on a decision-making process even when that process tokenizes, extracts value from, or further disenfranchises the people ostensibly represented. Debates about representativeness might also distract attention from whether the technology should be deployed at all [41,74,144].…”

“…The comparative harms of inclusion and exclusion from datasets are complex, contested and contextual. We do not aim to resolve them here, and they are thoughtfully considered by other scholars [20,74]; but for our purposes, it is crucial to note how these tensions complicate the positive connotations often associated with representativeness.…”

Designing for All: Consumer Response to Inclusive Design

“…Previously, we discussed the barriers to inclusive design, barriers that may be more difficult to overcome in some industries than others. Consider, for example, the relatively low cost of adding a slew of inclusive emojis to an existing platform compared to the cost of developing inclusive artificial intelligence and machine learning algorithms (Hoffmann, 2020). Furthermore, some industries, by their very nature, are not inclusive.…”

Hormonal Advantage: Retracing Exploitative Histories of Workplace Menstrual Tracking

“…Why then would feminist figures of today promote similar modes of bodily accounting, where more intimate details of daily life are rendered visible to company representatives? The past can offer insight into how this data might be leveraged-toward discriminatory practices, rather than increased workplace access or equality (Hoffmann 2020). Are the proposed advantages of tracking (e.g., a more personalized work schedule or relevant tasks) worth the potential loss of data privacy?…”