Conclusion: Now Let’s Multiply
Less than 24 hours after the 2016 US Presidential Election, and the unexpected defeat of Hillary Clinton, a retired grandmother from Hawaii, Teresa Shook, took to Facebook to declare her intention to march on Washington in protest of the president-elect. Less than two days after that, Shook’s idea became a reality when a group of professional activists and organizers offered to help her plan an official event--what they decided to call, as Shook had, the “Million Women March.”
When Shook picked the name, she was thinking about the Million Man March, the 1995 event which brought a reported 1.5 to 2 million African American men to the National Mall in order to call attention to issues of civil rights and racial injustice. It’s unclear as to whether Shook knew that there had already been a protest called the Million Women March, which took place in Philadelphia, in 1997, on behalf of African American women. But in either case, the name of the 2016 event drew critics– who pointed out, rightly, that because the initial group of organizers consisted only of white women, they lacked the crucial perspectives that would allow them to develop a truly inclusive social and political platform.
The critics were right. And after adding additional women of color to the organizing committee– many of whom were professional activists and organizers– and also changing the name of the event, it expanded into a more powerful and inclusive movement, if not ever controversy-free. With more than 600 distinct locations across the US and an estimated three to five million participants, the Women's March on Washington became the largest single protest in U.S. history.
The story of the Women’s March serves as a reminder that while feminism can serve as a valuable starting point for identifying issues of inequity and injustice, it’s not the only position that a person might start from. As we hope we’ve made clear, a feminist approach– to data science, to visualization, or to anything else in the world– should always be accompanied by an awareness of the perspectives that it does not, or cannot, account for.
We are two white women with four advanced degrees and five kids between us, who work in the privileged world of higher education. While we have learned immensely from, for example, the Black feminist activists and organizers whose work we describe in this book, we can never speak directly from the life experiences that motivate their work. But imagine a book on data that takes Black feminism as its focus. What concepts would it introduce, what principles would it propose, and what examples would it cite?
Some of these possibilities are hinted at in the mission of the group Data for Black Lives, which we discuss in The Power Chapter. The group’s emphasis on liberation, rather than a generic form of social good, leads them to projects that actively work to overturn the discrimination and injustice experienced in Black communities as a result of data-driven systems like predictive policing, predatory lending, and risk-based prison sentencing. Or, for another example, consider the principles that guide the corpus creation work of the Colored Conventions Project, which we discuss in Teach Data Like an Intersectional Feminist. Like the nineteenth century organizing meetings that the project seeks to document, the CCP promotes the work of collectives over individuals, and insists on acknowledging the humanity of any person or group represented in their data set.
Or, as another starting point, consider what a queer approach to data science might entail. Queer data science might build off the concept of failure, as described by Maria Munir with respect to the lack of non-binary gender categories in What Gets Counted Counts. Amplifying the moments in the data processing pipeline when our work leads not to new knowledge, but to something we can’t ever know, a queer data science could help to call out the otherwise invisible assumptions embedded in our technical and social systems. This might take the form of visualizing the gaps in a data set--a sort of inverse of Daniel Cardozo Llach’s visualization of architectural data traces that we discuss in Show Your Work. Or it could lead to the design of an entire visualization that, rather than employing interaction to lead to increasing insight, instead becomes increasingly opaque over time--a sort of data science version of the “refusal of legibility,” to borrow Jack Halberstam’s term, that characterizes much of queer life.
Or how would a disability studies perspective, which shifts the focus from the individual body to the social structures that enable the capacities of certain bodies, while disabling the capacities of others, push us to rethink our approach to interface design, and of the larger frames in which we display the results of our data analyses? Or a postcolonial approach, which would challenge us to connect issues of power, politics, and geography, in the context of data? Or an explicitly indigenous approach that values cultural knowledge as sacred and has rigorous accountability structures for working ethically with outsiders? These are only a few of the possibilities that approaches to data science, informed by additional perspectives, might present.
Our goal with this book has been to provide a model of how feminist thinking might be applied to data science, and to plant the seeds for exploring how other modes of thinking that intersect with social and political concerns can help advance the field. In the examples in this chapter, we sketch out some of these possibilities. This conversation – about data, design, and justice – is one that’s only just begun.