In the June 2024 Trump-Biden debate, a notable adherence to rules and procedures was observed. Unlike their 2019 debate, where interruptions were frequent, the candidates this time waited their turn to speak. This change was attributed not to a shift in political culture but to an audio adjustment: microphones were only functional during designated speaking periods.
This example illustrates the concept of "technology-as-policy," where technology shapes social behaviors and outcomes through its design. According to Jenny L. Davis, Professor of Sociology at Vanderbilt University, "Written rules and regulations codify what can, should, must, and cannot be done. Technologies do the same through their material form—preventing, confining, persuading, and compelling."
However, this raises a transparency problem. As policy instruments, technologies are often imprecise and inscrutable. They regulate people subtly and implicitly without clear statements of what rules are in play or whose values they reflect.
To address this issue, Davis has developed a framework over nearly a decade that focuses on 'affordance,' which refers to how technologies enable or constrain possibilities. Her book "How Artifacts Afford: The Power and Politics of Everyday Things" presents the “mechanisms and conditions framework of affordances.” This framework describes how technical features exert varying degrees of force through mechanisms like request, demand, encourage, discourage, refuse, and allow.
Davis applied this framework to machine learning (ML) technologies in her recent work. She examined how ML systems set workplace policies in Amazon's fulfillment centers by processing data from various sources to dictate workers' movements and productivity metrics.
"The mechanisms and conditions framework clarifies how ML sets workplace policies," Davis explains. For instance, GPS tracking encourages worker surveillance while feeding into data-hungry ML models that demand compliance with automated directives.
Davis suggests that just as her framework reveals current practices, it can also propose alternatives. For example, warehouse conditions could be reimagined so that only products are tracked instead of people or work rates recalibrated to allow for bodily diversity.
Algorithmic governance extends beyond warehouse work to areas like dating apps and résumé sorting. Across these domains, Davis argues that her framework can enhance transparency by making techno-policies visible for scrutiny and potential reformation.
###
