Intersectional feminism isn't just about women nor even just about gender. Feminism is about power – who has it and who doesn’t. And in a world in which data is power, and that power is wielded unequally, data feminism can help us understand how it can be challenged and changed.
Contributors (2)
Nov 06, 2018

This chapter is a draft. The final version of Data Feminism will be published by the MIT Press in 2019. Please email Catherine and/or Lauren for permission to cite this manuscript draft.


Christine Mann Darden first passed through the gates of NASA’s Langley Research Center, in Hampton, Virginia, in 1967. In the city of Hampton, and across the United States, tensions were running high. In Los Angeles that June, a massive protest against the Vietnam War ended in violence when over a thousand armed police officers attacked the peaceful protestors, sparking national outrage. One month later, in July of that year, even more violence engulfed the city of Detroit after a police raid gone awry. The 1967 Detroit Riot, or the 1967 Detroit Rebellion, as it is increasingly known, resulted in over 40 deaths and 1000 injuries--most of which were sustained by the city’s predominantly Black residents. The rebellion ended only after the governor, with the support of the President of the United States, called in the Army’s Airborne Division and the Michigan National Guard.

The gates of Langley might have shielded Darden from those physical confrontations, but her work there was no less politically engaged. By 1967, the Space Race was well underway, and the United States was losing. The Soviet Union had already sent a man into space and a rocket to the moon. The only thing standing in the way of a Soviet victory was to put those two pieces together. Meanwhile, the U.S. had suffered a series of defeats--and, in January of that year, an honest-to-goodness tragedy, when a fire during a launch test of the Apollo 1 spacecraft killed all three astronauts on board.

While the nation was in mourning, everyone at NASA threw themselves back into their work-- including Darden, who held a master’s degree in applied math and was employed as a data analyst at the time. Two years later, it would be Darden’s point-perfect analysis of the physics of rocket reentry that would help to ensure the successful return of the Apollo 11 mission from the moon, effectively winning the Space Race for the United States. But it would be Darden herself, as a Black woman with technical expertise, working at a federal agency in which sexism and racism openly prevailed, who demonstrated that the ideological mission of the United States was far from accomplished.

The 1960s, after all, were years of social protest and transformation as well as exploration into outer space. Darden herself had participated in several lunch-counter sit-ins at Hampton Institute, the historically Black college that she attended for her undergraduate studies. The Hampton sit-ins were the first in the state of Virginia, and contributed significantly to the dismantling of the Jim Crow era policies of segregation that were still in place at the time. By the time that Darden joined NASA, its Virginia facility had been fully desegregated for several years. But it had yet to reckon with another issue of equality of opportunity: the status of its human computers.

Darden’s arrival at Langley coincided with the early days of digital computing. While Langley could claim one of the most advanced computing systems of the time--an IBM 704, the first computer to support floating-point math--its resources were still limited. For most data analysis tasks, Langley’s Advanced Computing Division relied upon human “computers” like Darden instead. These computers were all women, trained in math but treated like secretaries. They were brought into research groups on a project-by-project basis, often without even being told anything about the source of the data they were asked to analyze. Most of the male engineers never even bothered to learn the female computers’ names.

But Darden had a sense of justice in addition to her advanced degree. So after several years of working as a computer, she decided to ask her boss why men with her credentials were placed in engineering positions, where they could be promoted through the ranks of the civil service, while women like herself were sent to the computing pools, where they languished until they retired or quit. As Darden, now 75, told Margot Lee Shetterly, who interviewed Darden for her book, Hidden Figures: The American Dream and the Untold Story of the Black Women Who Helped Win the Space Race, her boss’s response was sobering: “Well, nobody’s ever complained,” he told her. “The women seem to be happy doing that, so that’s just what they do.”1

Today, a response like that would get a boss fired (or, at the least, served with a Title IX complaint). But at the time, stereotypical remarks about “what women do” were par for the course. In fact, assumptions about what women could or couldn’t do--especially in the workplace--was the central subject of Betty Friedan’s best-selling book, The Feminine Mystique. Published in 1963, The Feminine Mystique is often credited with starting feminism’s so-called “second wave.”2 Fed up with the enforced return to domesticity following the end of World War II, and inspired by the national conversation about equality of opportunity prompted by the Civil Rights Movement, women across the United States began to organize around a wide range of issues, including reproductive rights and domestic violence, as well as the workplace inequality and restrictive gender roles that Darden faced at Langley.

That being said, Darden’s experience as a Black woman with a fulltime job was quite different than that of the white suburban housewife--the presumed audience of The Feminine Mystique. And when critics called out Friedan, rightly, for failing to acknowledge how no single person could claim to speak on behalf of all women, everywhere, it was women like Darden, among all sorts of different folks, whom they had in mind. In Feminist Theory: From Margin to Center, a pioneering Black feminist text, bell hooks puts it plainly: “[Friedan] did not discuss who would be called in to take care of the children and maintain the home if more women like herself were freed from their house labor and given equal access with white men to the professions. She did not speak of the needs of women without men, without children, without homes. She ignored the existence of all non-white women and poor white women. She did not tell readers whether it was more fulfilling to be a maid, a babysitter, a factory worker, a clerk, or a prostitute than to be a leisure-class housewife.”

In other words, Friedan had failed to consider how additional factors like race and class, not to mention sexuality, ability, age, religion, and geography, among many others, intersect with each other in order to determine any particular woman’s personal experience in the world. Although this concept did not have a name when hooks described it--the term intersectionality would be coined by legal theorist Kimberlé Crenshaw in the late 1980s--its necessity was already clearly borne out by, well, everything. In the face of the racism embedded into U.S. culture, coupled with the many other forms of oppression experienced by minoritized groups, it would be impossible to claim a common experience--or a common movement--for all women, everywhere. Instead, what was needed was what the Combahee River Collective, the famed Black lesbian activist group out of Boston, advocated for back in 1977: “the development of integrated analysis and practice based upon the fact that the major systems of oppression are interlocking.”   

We’ll have more to say about the importance of intersectionality in a few pages, but first let’s find out what happened with Christine Darden and her goal of becoming an aerospace engineer. As Shetterly tells it, Darden heard nothing from her boss but radio silence. But two weeks later, she was indeed promoted and transferred to an aerospace engineering group. Darden would go on to conduct ground-breaking research on sonic boom minimization techniques, author more than sixty scientific papers in the field of computational fluid dynamics, and earn her PhD in mechanical engineering-- all while “juggling the duties of Girl Scout mom, Sunday school teacher, trips to music lessons, and homemaker,” she recalls.

For over a decade, this research required that Darden work overtime in the office as well as at home. But she could tell that her scientific accomplishments were still not being recognized at the same level as her male colleagues. Once again, data analysis opened doors for Darden. But this time, Darden wasn’t responsible for the math. Instead, her technical expertise provided a key datapoint for a larger advocacy project.

Over in Langley’s Equal Opportunity office, a white woman by the name of Gloria Champine had been compiling a set of internal statistics about gender and rank. The data showed that men and women with identical academic credentials, publication records, and performance reviews, were still promoted at vastly different rates. Champine then visualized the data—in the form of a bar chart—and presented her findings to the head of her Directorate. He was “shocked at the disparity,” Shetterly reports, and Darden received the promotion she had long deserved.

Darden would advance to the top rank in the federal civil service, the first Black woman at Langley to do so. By the time that she retired from NASA, in 2007, Darden was the head of a Directorate herself.

<p>Christine Darden in the control room of the Unitary Plan Wind Tunnel at NASA’s Langley Research Center in 1975.</p><p>Credit: NASA.</p><p>Source: Wikipedia,</p>

Christine Darden in the control room of the Unitary Plan Wind Tunnel at NASA’s Langley Research Center in 1975.

Credit: NASA.

Source: Wikipedia,

Christine Darden’s rise into the leadership ranks at NASA was largely the result of her own knowledge, experience, and grit. But Darden’s story is one we can only tell as a result of the past several decades of feminist activism and critical thought. For it was a national feminist movement that brought issues of women in the workplace to the forefront of U.S. cultural politics, and it was a local feminist advocate in the form of Gloria Champine who enabled Darden to continue her own professional rise. It was also, presumably, the work of many unnamed colleagues and friends, who may or may not have considered themselves feminists, who provided Darden with community and support--and likely a significant number of casseroles--as she ascended the ranks of NASA. And it was the work of feminist scholars and activists that allows us to recognize that labor, emotional as much as physical, as such today.

Now it’s probably time to clarify the relationship between the dictionary definition of a feminist and what is described, more generally, as feminist thought. The Merriam-Webster dictionary (and also, for the record, Beyoncé) define a feminist as “a person who believes in the political, social, and economic equality of the sexes.”3 Feminist thought has its basis in this theory of the equality of the sexes, but it is much more expansive. It includes the work of activists like Champine, or bell hooks, or--however problematically-- Betty Friedan, who have taken direct action to achieve the equality of the sexes. It also includes the work of scholars and cultural critics--again like hooks, or like Kimberlé Crenshaw, or like Margot Lee Shetterly--who have explored the social, political, historical, and conceptual reasons behind the inequality of the sexes that we face today.

In the process, these scholars and activists have given voice to many of the additional ways in which the status quo is unjust. These include the power differentials between men and women, as well as those between--for instance--white women and Black women, academic researchers and indigenous communities, and people in the Global North and the Global South. These feminist thinkers arrived at their emphasis on power, rather than gender alone, because of their insistence on intersectionality, the concept we started to get at just a few pages ago. So let’s get a little more specific in our explanation of intersectionality. The concept doesn’t simply describe the intersecting aspects of any particular person’s identity that shape their experience in the world. Rather, it describes the intersecting systems of power--the systems of privilege, on the one hand; and systems of oppression, on the other--that determine that particular person’s experiences. When you stop to think about it, many people experience at least a little of both.

In fact, it was an example of how oppression and privilege themselves intersect that prompted Crenshaw to name the concept that she’d seen play out over the course of her legal career. In law school, Crenshaw came across the anti-discrimination case of DeGraffenreid v. General Motors. Emma DeGraffenreid was a Black working mom who had sought a job at a General Motors factory in her town. She was not hired. The factory did have a history of hiring Black people: many Black men worked in industrial and maintenance jobs there. They also also had a history of hiring women: many white women worked there as secretaries. These two pieces of evidence provided the rationale for the judge to throw out the case. The company did hire Black people and did hire women, so it could not be discriminating on the basis of race or gender. But what about discrimination on the basis of race and gender together, Crenshaw wanted to know? This was something different, it was real, and it needed to be named.

Intersectionality helps us name and recognize the interaction between categories of social difference, such as race and gender. It helps us see when people who embody two or more of those characteristics fall through the cracks, because they are doubly or triply marginalized. It also helps us unmask the privilege that comes with embodying the dominant dimensions of identity, and helps us understand how oppression and privilege can co-exist in the same body. For example, a white, gay, disabled, cisgendered man might reap the benefits of privilege for his race and gender, but experience oppression for his sexual orientation and disability. A straight, college-educated, cisgendered Muslim woman might experience certain privilege on account of her sexual orientation and level of education, but experience oppression on account of her gender and religion. The intersection of categories of social difference, and of the forces of privilege and oppression that are bound up in them, are what Crenshaw’s term names.

In the case of Christine Darden and her promotion, Gloria Champine was primarily concerned with the issue of gender. But she was also, like Crenshaw, intent on exposing a larger system of power and privilege. She knew that unless she confronted the systematic nature of the discrimination faced by women at NASA, she would continue to hear from individuals like Darden for the rest of her career. Her goal was to implement changes that would improve the lives of all women at NASA, and to achieve that goal, she required a complete picture--in the form of her bar chart-- of the problem at hand.  

Of course, when Champine created her bar chart, she also recognized in the 1980s what many of us are only now beginning to understand: that data visualizations, and the data science that underlies them, hold tremendous rhetorical force. Now, as then, a single data visualization can dazzle, inform, and persuade. Champine aligned her goal of challenging the systemic nature of the gender discrimination that plagued NASA at the time with the rhetorical power of a bar chart. She asked herself: “What is the source of the problem that my colleague is facing? What information do I need in order to bring this problem to light? And what is the format through which I can best advocate on her behalf, and effect structural change?” In doing so, Champine joined with Darden to model a key aspect of what we call in this book data feminism: a way of thinking about data and its communication that is informed by direct experience, by a commitment to action, and by the ideas associated with intersectional feminist thought.  

Data feminism can show us how images like Champine’s bar chart might seem neutral and objective, but are in fact the result of very human and necessarily imperfect design processes. Data feminism can also show us how the categories of data collection matter deeply, especially when dividing people into groups. Because Champine was able to collect data on both gender and rank, she was able to show the extent of the gender discrimination that was taking place at NASA at the time. But because she did not include any additional demographic information in her report--whether by circumstance, or by design--she was unable to show the effects of any additional forms discrimination that might also have been then taking place. The alliance between Champine and Darden, and the chart they together produced, also helps underscore the importance of listening to and learning from the data’s stewards: the people who serve as the source of knowledge about the issues the data purports to represent. Champine knew to crunch the numbers only because Darden shared her personal experience of gender discrimination with her. Without Darden’s first-hand knowledge of the problem, Champine might never have known that action was necessary.



It took five state-of-the-art IBM System/360 Model 75 machines to guide the Apollo 11 astronauts to the moon. Each was the size of a car and cost $3.5 million dollars. Fast forward to the present and we now have computers in the form of our phones that fit in our pockets, and-- in the case of the iPhone 6-- can operate 120 million times faster than a standard IBM System/360. We’ve also witnessed an equally remarkable growth in our capacity to collect information in digital form--and in the capacity to have data collected about us.

As it turns out, the IBM System/360 line was viewed as a major milestone in the history of data processing as well as rocket science. It was the first family of machines that could be scaled up to be used for aerospace-level operations-- and also scaled down to be used by a single data analyst in, for instance, an insurance company or a bank. But those banks and insurance companies today? They now collect data on our purchase histories and online behaviors, the times of day we’re most active on Facebook and the number of items we add to our Amazon cart. Our most trivial everyday actions – taking a single step, searching for a way around traffic, or liking a friend’s cat video – are now hot commodities. Not because our friends’ cats are exceptionally cute (they are cute, of course, but not exceptional), or because our step counts are exceptionally exciting (they are emphatically not), but because those tiny actions can be combined with other tiny actions in order to determine, for instance, whether we’re in a liking kind of mood, whether we tend to click on links when we’re also liking videos, whether we might also happen to be frustrated with our daily commute, and whether today might be the day for a well-placed ad for new sneakers, or a coupon for 20% off. An alternative to the daily grind and a way to increase our step counts in one fell (Nike) swoop.   

This is the data economy. And corporations and governments, often aided by scholars and researchers, are scrambling to see what consumer behaviors remain untapped and unrefined. Nothing is safe from datafication, the process of turning phenomena in the world into digital information. Not your cat, or your aspirational exercise goals, or-- more realistically-- the butt you are currently using to sit in your seat. Shigeomi Koshimizu, a Tokyo-based professor of engineering, has been designing matrices of sensors that collect data at 360 different positions around your rear end while it’s smushed in a chair. Those data are then analyzed by custom software that detects micropatterns in weight and pressure. The result is a data profile of your butt that is, according to Koshimizo’s research, as unique as your fingerprints. In the future, he suggests, our cars could be outfitted with butt-scanners instead of keys or car alarms. If Catherine sits down in her car of the future--self-driving, of course--this technology would scan her butt, crunch the data, and welcome her with a warm hello. But if Lauren sat in Catherine’s car, it would refuse to move--or it might even be programmed to call the police.

While this redefinition of butt-dialing may still be a few years away, the datafication of our everyday lives is already a reality--and not only when we’re actively clicking. Decisions of social and civic importance, ranging from which products to stock in the grocery store before a hurricane, to which city buildings to inspect for the risk of fire, to which citizens to tag as pre-trial flight risks, are increasingly being made by automated systems sifting through large amounts of data. For example, Walmart's predictive analytics team has long combined consumer purchasing patterns with weather data in order to address the first of these scenarios. If a hurricane is coming, what goods should Walmart get on the shelves quickly? It turns out that obvious items like flashlights and generators are in high demand. But Walmart also always sends truckloads of strawberry Pop-Tarts to areas where there are hurricane warnings, because their data analysis detected spikes in Pop-Tart sales during episodes of severe weather. As VP of Information Systems Dan Phillips told Fortune magazine, "They are preserved until you open them, the whole family can eat them, and they taste good."

There are similar examples of data-driven decision making in place in the government sector. In a widely-cited example, a stats team employed by the City of New York helped to integrate data analysis into the building inspection process. In New York, there are upwards of 25,000 complaints per year about buildings that have been illegally converted into apartments. But there are only around two hundred inspectors on city payroll who can handle the complaints. These illegal apartments often pose fire hazards, and in fact, many firefighters have lost their lives trying to save the residents who live there. So the stats team brought together several datasets relating to property tax delinquency (an indicator of neglect), rat complaints (ditto), arrest locations (a proxy for poverty), and more, in order to rank the 25,000 complaints by fire risk. The inspectors began with the buildings that were determined to hold the highest risk, and postponed their inspections of lower-risk sites. To their surprise, the inspectors issued five times more "vacate orders" than they had without the data-assisted ranking system. They might have been responsible for causing short-term inconvenience on the part of the people who were required to move, but they were able to reduce the longer-term risk of fire and potential loss of life. Here, the power of data was wielded for civic good: to allocate scarce resources to address issues of public health and safety. 

But data-driven decision-making can be just as easily used to amplify the inequities already entrenched in public life. In Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy, Cathy O’Neill has drawn attention to how predictive policing models use data about property tax delinquency and arrest locations, the very same datasets used for good by the City of New York’s stats team, in order to determine which neighborhoods to patrol more heavily, and which neighborhoods to leave alone. Like city building inspectors, it turns out that police are also in short supply. But in the case of policing, data analysis does not always save lives. Because more police are sent into neighborhoods that have had more crimes already reported, the police are more likely to be there when a new crime is committed (or, as in most cases, simply suspected of taking place). And because the police are already there, the people involved--usually poor, and usually people of color-- are more likely to get ticketed, arrested, or even killed. This creates what O’Neill calls a “pernicious feedback loop,” amplifying the effects of the already pernicious criminalization of poverty that takes place in the United States. Meanwhile, in more affluent neighborhoods, the very same petty crimes--like jaywalking or littering, for instance-- are far less likely to be prosecuted because the police simply aren’t there to see those crimes take place. This disparity in law enforcement is what led to the creation of White Collar Crime Risk Zones, a satirical map of all of the white collar crime that goes uninvestigated because neighborhoods of color are so overpoliced.

<p>White Collar Crime Risk Zones uses machine learning to predict where financial crimes by white people may occur.</p><p>Credit: Brian Clifton, Sam Lavigne, and Francis Tseng for <em>The New Inquiry Magazine</em>, Vol. 59: ABOLISH.</p><p>Source:</p>

White Collar Crime Risk Zones uses machine learning to predict where financial crimes by white people may occur.

Credit: Brian Clifton, Sam Lavigne, and Francis Tseng for The New Inquiry Magazine, Vol. 59: ABOLISH.


The double-edged sword of data shows just how important it is to understand how structures of power and privilege operate in the world. The questions we might ask about these structures can relate to issues of gender in the workplace, as in the case of Christine Darden and her wrongly delayed promotion. Or they can relate to issues of broader social inequality, as in the case of predictive policing described just above. So one thing you will notice throughout this book is that not all of our examples are about women--and deliberately so. This is because data feminism is about more than women. It’s is about more than gender. Put simply: Data Feminism is a book about power in data science. Because feminism, ultimately, is about power too. It is about who has power and who doesn’t, about the consequences of those power differentials, and how those power differentials can be challenged and changed.

This paragraph deserves re-stating, because we want you to remember these points as you read this book:

  • Data feminism isn't only about women.
    It takes more than one gender to have gender inequality; and more than one gender to work towards justice.

  • Data feminism isn't only for women.
    Men, non-binary, and genderqueer people are proud to call themselves feminists and use feminist thought in their work.

  • Data feminism isn't only about gender.
    Intersectional feminists have keyed us into how race, class, sexuality, ability, age, religion, geography, and more, are factors that together influence each person’s experience and opportunities in the world. In this book, we choose to foreground many examples where racism and patriarchy intersect. This reflects our location in the United States, where the most entrenched issues of injustice have racism at their source.

  •  Feminism is about power - who has it and who doesn't.
    And in our contemporary world, data is power. Which is why we wrote this book.

While the feminisms of the 19th and 20th centuries accomplished a great deal, feminism remains an unfinished project, and an urgent one. Consider that, while Darden's story ended well, her accomplishments are just beginning to be recognized--largely due to the work of a single person, Margot Lee Shetterly, and her efforts to shine a light on the "hidden figures" of the history of computing. Or, consider that, in 1996, the field of computing held a celebration for the fiftieth anniversary of the first electronic computer, the ENIAC. But the organizers of the event did not think to invite most of the computer’s original programmers to the party! These programmers were all women, of course. So is it a surprise that the number of female graduates of information sciences programs continues to decline? In 2013, the numbers were bleak: the field was only 26% women, the same percentage as in 1974.

On a larger cultural scale, Kimberlé Crenshaw and her colleagues have started a campaign called #SayHerName to call attention to police brutality against Black women, whose stories of racialized and gendered violence are so often left out of public conversations. And yet, the US federal government still has no comprehensive database on people killed by police officers. The #MeToo movement has demonstrated the pervasiveness of sexual assault as well as the bravery of women from all backgrounds to come forward. And yet, the U.S. has placed accused sexual predators in the White House and nominated them to the Supreme Court, all while their elite white male colleagues rally angrily behind their innocence. Feminism is unfinished and urgent work, in data and technology as well as in our most powerful political institutions.  

In the chapters that follow, we draw from a wide range of examples of feminist data science in order to show how we can take steps towards a more just and equal world. In the examples we discuss, we are guided by our values around intersectionality, equity, and proximity, which we outline in Appendix 1. We do so in order to challenge a set of widely held assumptions, like the idea that "the numbers speak for themselves," and to explore projects that expand our ideas about what constitutes “data” in the first place. We call attention to the people and their bodies who are typically included in the data collection process, as well as to the people and their bodies who are typically left out. We question whose work gets recognized, and whose research questions should matter most. Along the way, we introduce you to some of the analysts, designers, journalists, scholars, and teachers who are already doing the transformative work we hope to see more of in the world. There are a lot of data feminists out there! You just might not know about them until you read this book.


Nazareno Andrade: Unclear to those not in the US
Pratyusha Kalluri: I’d say if you’re including this sentence it needs more unpacking
Pratyusha Kalluri: recommend rewording. hard to read sentence.
Pratyusha Kalluri: Maybe “Critics called out Friedan, rightly, for….; it was multiply marginalized folks like Darden who the critics had in mind”. Maybe with a footnote to define multiply marginalized for folks.
Pratyusha Kalluri: as in “specifically Darden’s” or “Darden’s team’s”? Currently unclear to me
Julee Burdekin: Why are you calling it “feminism”, then? (And does movement in one sector of social justice necessarily impact another?) Or should we be retiring “feminism” for the more inclusive, exacting term “gender equality”? Not willing to do that? Why not? <— The answer to that is what differentiates data feminism from data gender-equity. (Notwithstanding the subsequent bullet points…) So looking forward to this book, but hesitate to read anything “feminism” in lieu of “gender equality”.
Momin M. Malik: Is the negative phrasing of all these points deliberate, supported by some theory? I would imagine positive framings are preferable, defining feminism by what it is rather than what it isn’t. If it was a rhetorical choice in order to emphasize “power” in the last point, I’m not sure if it’s worth it.
Elizabeth Losh: Because this paragraph is about metadata (and how hashtags highlight issues), it might be helpful to say more about how metadata and data visualization are potentially related strategies for making information more visible (which as you point out in the section on policing can be a double-edged sword).
Elizabeth Losh: It might be helpful to know something about how literacy practices at NASA (and the use of the bar chart specifically for other purposes) might have made this particularly persuasive information.
Yanni Loukissas: I really like this way of beginning the book. But I think a broad audience would be well served by a section, early on, that gives an explicit account of what you intend to do in the book and the stakes in doing so. For as I understand it, the book isn’t only about what data looks like from a feminist perspective; but rather what we risk societally by ignoring that perspective.
Yanni Loukissas: I really like this way of beginning the book. But I think a broad audience would be well served by a section very early on that gives an explicit account of what you intend to do in the book and the stakes in doing so. For as I understand it, the book isn’t only about what data looks like from a feminist perspective; but rather what we risk societally by ignoring that perspective.
Momin M. Malik: There’s a beautiful blog post by Candice Lanius, “Your demand for statistical proof is racist.” overarching critique I have is that, in some ways, you cede too much ground to data. Was it really “data” through which Champine knew of discrimination? Or was that the only language that was respected? My partner (and former SAFElab member!) Maya Randolph connected this point to a famous quote from Toni Morrison, in a 1975 address at Portland State University:“…the function of racism is distraction. It keeps you from doing your work. It keeps you from explaining over and over again, your reason for being. Someone says you have no language so you spend twenty years proving that you do. Somebody says your head isn’t shaped properly so you have scientists working on the fact that it is. Somebody says that you have no art so you dredge that up. Somebody says that you have no kingdoms, again you dredge that up. None of that is necessary; there will always be one more thing”. I would say that for Champine and Darden, the function of data, and visualization, was a distraction. Alongside any discussion of redemptive uses of data, I think we should recognize that things need not be this way, and perhaps shouldn’t be this way. We should believe racism exists because people experience it. If there were any data that showed there was no inequality in some situation where people told me they experience it, I would sooner doubt the data and the modeling, because I know those are imperfect tools to get to what matters, human dignity. Fortunately, there’s a simple way to get to that: asking people and believing them.
Catherine D'Ignazio: Beautiful point, Momin - thank you.
Momin M. Malik: An overarching critique I have (which is why I’m putting it here, and not in the chapters I’m actually reviewing) is that this book implicitly has a strong thesis, but I don't see it explicitly anywhere; stating it explicitly, and returning to it frequently, would help make the book feel more focused. I would phrase it as: "Feminism reveals the politics of data." An expanded version might add, specifically intersectional feminism, and "…and feminism tells us the implications of those politics, both current and future, and points to alternatives." This comes from how there is a fair amount of defense of why feminism is uniquely equipped to reveal these politics, as well as a fair amount of argument about why data has politics. Explicitly tying them together makes the two points seem less haphazard, although it is still a challenge to introduce feminism in its generality alongside using it to make critiques.
Catherine D'Ignazio: Thanks Momin - this is a super helpful comment.
Alison Booth: appointed. Clarence Thomas…
Alison Booth: Your introduction is engaging and winds up well. I do wonder if the last sentence or two lead into a book about personalities, as does the focus on Darden. Won’t your reader want to know more about “data” and “visualization” and “feminism”? Maybe invite your reader to become a data feminist?
+ 2 more...
Alison Booth: a rare moment where your narrative seems to invite unclear thinking about causality. So many causes of decline in enrollments by gender, besides open behaviors of exclusion
Alison Booth: Yes, your bullet points work well! I would welcome you to connect to “knowledge is power.” Data are not knowledge, or even information. There might be experts on information, data theory who are concerned if you don’t signal that these are not simple concepts.
Alison Booth: be more exact—many?
Alison Booth: the shift to data surveillance could be more explicitly tied to the mission of your introduction. It now seems like a new topic of interest, almost unrelated to intersectionality, etc. above
Alison Booth: This is key, and I like it.
Alison Booth: =person
Alison Booth: Good explanation of intersc. concept. This may go too far, but I also see the idea as helpful for “reading” each identity as overdetermined by the others: hetero masculinity in US is troped as working class and Black, e.g.
Alison Booth: In my view, this shouldn’t be capitalized as an adjective for the communities. But could stand corrected, of course!
Alison Booth: In this paragraph, I am getting a strong sense that you are reaching an undergraduate or nonspecialist audience.
Alison Booth: nice move.
Alison Booth: Already clear. Yes, even Virginia Woolf in 1928 and 1937 did understand the intersection of race/empire, class, and the individualist self-fulfillment of middle class women. She wrote about servants. In other words, many people writing about male or female rights would not do a good job in 1963 of seeing these intersections, yet socialist/materialist analysis here and there as long ago as the late eighteenth century had gone quite far toward what Crenshaw highlighted. I don’t object strongly to the simplified history here, but maybe add “Read More” or notes to good synopses of feminist thought?
Alison Booth: As an introduction, I could use by now some signal by now why you are retelling this history, interesting and relevant as it is. This note suggests you WILL need to invoke a history of feminism but not claim that it has clear waves. OK, other than the title, does your reader know where we’re going? Why is the book focused on US? The history of women, feminism, and data, is not just a Black/White US history.
Alison Booth: Maybe other readers as well as I begin to wonder about the sources of information? Will the book make notes/sources easy to find? I see the first note below.
Alison Booth: Maybe worth noting that desegregation was probably technical and not “full.” Gender and class kept people apart in different parts of the facility.
Momin M. Malik: agreed!
Patricio Davila: I don’t know if this is the convention in the USA but in Canada “Indigenous” is capitalized when referring to a people
Catherine D'Ignazio: Thanks for this heads-up.
Aristea Fotopoulou: This is great!
Aristea Fotopoulou: This is great, but I’d also like to see what it is about. Is it about activism involving all genders and that aims at social justice in relation to the applications of data science?
Aristea Fotopoulou: It’s great to see what isn’t but I’d really love to see a definition about what it is too. Is data feminism about activism involving all genders and aiming at social justice in relation to data science?
Heather Krause: Interesting juxtaposition of “organizations” to “people”. Is this intentional? Do we want to align with the idea that the actors are corporations and governments rather than the people that make up these institutions? This is what “often aided by” implies to me.
Jacque Wernimont: I suspect that this could be even stronger, if you’d like. For example, rather than saying “ideas associated with intersectional feminist thought”, why not outline the political commitments of intersectionality as those of data feminism? So, for example, “a commitment to action, and attention to the networks of power that constrain the lives of black, brown, queer, GNC, and disabled people in particular ways.” Or: “…to action, and dismantling racism, seixism, ableism, and homophobia as interlocking systems of oppression” — I’ll also note that much of the text is situated in/draws from spaces that are Anglo-American dominated and it might be worth acknowledging that.
Jacque Wernimont: both the “stronger” and the positionality show up later in the bulleted section - I don’t know if it’s worth flagging up here…but it might be. I tend to restate things several times (signposting ftw) but that’s a stylistic choice.
+ 1 more...
Jacque Wernimont: Some readers will find this easy to dismiss
Aristea Fotopoulou: Would we say ‘digitization’ then? It may be useful here to differentiate between the two terms.
Jacque Wernimont: perhaps its worth making it a two step process (as I’d argue it is historically), first the datafication and then the digitization. Or, a remediation and acceleration with modern digital datafication…
Aristea Fotopoulou: It would be great to see an expanded version of this definition. Also - if this is a key aspect of data feminism, I am wondering: are there other aspects (not mentioned in this definition) that you will come back with at a later point, with a different story?
Aristea Fotopoulou: It is great to read the story behind the concept.
Aristea Fotopoulou: To me the transition from the previous paragraph is slightly abrupt - I also am not sure that this clarification is absolutely necessary for the readers that I imagine this title will have.
Yanni Loukissas: Is there a reason you are not using this year’s numbers?
Yanni Loukissas: At this point I’m still not quite sure what you mean by “power.” You say this earlier: “the systems of privilege, on the one hand; and systems of oppression, on the other” But since the book is evidently focused on power, perhaps you could take a bit more space to explain what you mean by the term and how you see it working in relationship to data.
Jacque Wernimont: just a suggestion here - this would be relatively easy to address within the discussion of feminist intersectionality…
Yanni Loukissas: Echoing the point above: what it meant to be “treated like a secretary” in the 1960s might not be easily imaginable for undergrads today.
Marian Dörk: i really appreciate the time jump into the present here, but i wonder if/how the quantitative leaps in data collection and the qualitative changes in terms of automatic analysis could be fleshed out a bit more.
Momin M. Malik: I object more. There is a missing link here, from computation to data. Yes, data and computation are inextricably linked modern history, but we had (quantitative) data without computation (e.g., Mesopotamian farming records that elites used to keep control), and computation without data (like simulation modeling; that does have output data, but doesn’t have to have input data, only parameters). The “data analysis” that you talked about earlier would have likely been of simulation outputs, which I would say is very different from personal data. I would need to be convinced that the issues around data on purchase histories and online behaviors has the same material threats as the lack of credit and respect for people analyzing the outputs of computed simulations. I could more easily be convinced by a thematic link (e.g., the same institutions who held back Darden and Champine are the ones creating the infrastructure of data), although that would be weaker, and I would still like to see made explicitly.
Marian Dörk: to me it is not entirely clear here whether and how Darden and Champine collaborated on producing the data / chart
Marian Dörk: so far the mentioned chart did not read as a collaborative effort
Marian Dörk: is it feasible to have Gloria Champine’s bar chart included here as a figure? in the cited interview it is only mentioned by her, but not depicted
Marian Dörk: succinct and to the point - quoteworthy!
Marian Dörk: i would really appreciate a few more references here – even if necessarily incomplete
Marian Dörk: this reads as if she continued her work in a similar fashion as before, yet the promotion did represent a significant career change, didn’t it?
Marian Dörk: i wonder whether this could be phrased a bit more empathically - considering that intersectionality was introduced more than 20 years later and that we’re now +50 years later looking back with our ‘intersectional’ glasses on
Marian Dörk: it is not necessarily clear what that ideological mission was
Zara Rahman: I love this! Really snappy/easy-to-understand description of the whole concept.
Marian Dörk: Fully agree! Wonder whether you want this line to also start with Data feminism… or maybe change the 3 points as they apply more generally to Feminism ?
Rebecca Michelson: I’m curious if this relates to the creator(s) of the data visualizations or their subject(s).
Megan Adams: I question the use of “short-term inconvenience” here. Although the program may have ultimately resulted in important improvements for health and safety, losing a place to live can have devastating consequences for already vulnerable people (e.g., job loss,exposure to violence, forfeiture of personal property).
Ingrid Marx: I agree.
+ 3 more...
Amanda Makulec: Will you define what you mean by data science? It’s sometimes used as a familiar catch all term for stats, math, ML/AI, etc.,, but also can have a more narrow technical definition. Even wikipedia’s definition seems to fall somewhere in between - Personally, I wouldn’t generalize that all of the data visualizations I create required data science, but they do require analysis/stats. And data feminism is very relevant to even the simple charts and graphs that surface insight (but may require nothing more than plotting points with a pen and paper).
Zara Rahman: Agreed - it seems to me that this book (and the way that you use the term ‘data feminism’) covers topics that might not be considered as ‘data science’ necessarily..
Marie Léger-St-Jean: the podcast Reply All has an excellent two-part episode (#127 and #128) with interviews with the guy who started designing them: The Crime Machine. Maybe worth adding as a footnote?
Marie Léger-St-Jean: in French, this means “pots”, as in pots and pans, and they have been used by women in demonstrations in Chile in the 1970s (and in Québec during the 2012 student strike)... maybe use an expression like "pre-cooked meals" or something like that? it will prevent francophones like me to pause and have to re-read the sentence a few times because it would make sense to talk about women and protests and casseroles in the same sentence, =P
Marian Dörk: is this an inside joke? do casseroles stand for friendship or camaraderie in US-American culture?
+ 1 more...
Summer Hirtzel: would love for this to be defined a little more — it’s not quite clear what you mean here.
Rahul Bhargava: This feature selection process is, of course, fraught with power dynamics too. Who made those choices? Would the model be different with different features? Not sure if it is too early in the book to dig into that, but it gets at the _power_ theme you introduce in the book’s purpose just after.
Derya Akbaba: unclear sentence
Erik Simpson: This story of Darden (and then Champine) is fascinating and powerful. What a great way into this book! I wonder whether it would be worth bringing to the surface the way Darden is telling a feminist data story to her boss: a pool of workers has two starkly different sets of outcomes, with the more desirable outcomes correlating with gender rather than qualifications. (I hope I’m paraphrasing her case correctly.) Perhaps making that connection would help the reader see how Darden’s data-analytical habits of mind relate to her workplace advocacy: she is telling her own story not as an anecdote but as a consolidation and synthesis of the many stories she sees unfolding around her.
Lauren Klein: Thanks for this observation. We try to draw this out a little later but I think you’re right that it should be foregrounded here too.
Erik Simpson: I like this way of contextualizing Darden’s work. Small point: I hesitated on “no less politically engaged” because I wasn’t sure what Darden’s work is being compared to. The comparison seems to point to the actions of the military (the most immediate referent of “those physical confrontations”), and political engagement might not be the best frame for that. Is there another way to say that these different activities share political consciousness, or to specify the politically engaged actors in the Detroit half of the comparison?
Gabriela Rodriguez Beron: Not sure why Bell Hooks is written in all lower case when all the other names are not.
James Scott-Brown: bell hooks (like danah boyd) chose a name for herself that is entirely lowercase
Jaron Heard: The Nike icon is a swoosh, not a swoop. Still almost works.
James Malazita: Shannon already mentioned reading Eubanks’s book—I think it could be useful here too. Eubanks does a great job of tackling this claim as a “middle class, liberal myth”, and that even “better data” and “better data practices” by “good” actors fail in the face of power. So maybe it’s more that data intersects with practices of the powerful, and not that data, itself, is power.
Shannon Mattern: Maybe a nod to Virginia Eubanks’s work here, too?
Lauren Klein: A good idea. We have her in Chapter One, but might want to include a reference here too.
Maya Wagoner: May need more explanation of what the job of “computer” was, since most people will not be familiar
Lauren Klein: Thanks for this feedback. We thought we were clear enough on this, but what additional information would you want to see?
+ 3 more...
Sarah Yerima: I would resist referring to feminism as an “unfinished” project because I wouldn’t want readers who lack familiarity with feminist theories to think that the feminist project will ever be a “finished” or “completed” one. Feminist thought (at least radical, liberation feminist thought), compels us to always be critical of power and attentive to the mechanisms that sustain it. That will always be necessary work. “Ongoing” isn’t the best substitute, but I’d recommend using a term in that vain to describe feminist thinking, action, and work. It is never ending, and it will always be. I should also mention that when I use the term “liberation feminist thought,” I’m borrowing from Imani Perry’s ‘Vexy Thing: On Gender and Liberation.’ It’s an excellent text and if you haven’t read it yet, I would suggest it!
Catherine D'Ignazio: Thank you for this comment and the reference. I definitely agree about the idea that one could never complete or achieve feminism. I have not read that text (tho Lauren may have), but I’m going to go get it immediately!
+ 1 more...
Sarah Yerima: Very well stated! It is important to link discussions of feminism directly to power so that those who have little orientation to feminist thought and practice don’t come away from the text believing that feminism is apolitical or about those who are not cis men attaining *equality* to cis men by accessing the privileges that have traditionally only been allocated to cis men. You have to tie any and all conversations about feminism to power and the dismantling of patriarchy.
Momin M. Malik: This relates to a comment I made above, that I see the central thesis of this book is that “feminism reveals the politics of data.” This sentence would say how it is doing so: through an analysis of power. However, I’m not sure if “data science” is the right target. An earlier comment talks about the need to define data science; as somebody who is in that world, I would say most of the definitions are aspirational and political, descriptively data science is mostly applied machine learning (and machine learning I might quickly described as “megalomaniacal statistics”: same underlying machinery, but used in different ways). Just as applied statistics focuses on exploratory data analysis and visualization and working on serving a domain partner/client more than statistics without that “applied” qualifier, and will usually re-use existing models rather than developing custom ones, data science involves the same orientation compared to machine learning: they will seldom build their own models or even do much fine-tuning of existing model (unless they are machine learning people who turned to applied work), but will know what sorts of existing techniques to use. If you seek out a definition of data science, I would suggest only using something consistent with this (at a minimum!). I unfortunately don’t have good things offhand, but I trust Rayid Ghani as one person to look to (, although even things from the program he started, Data Science for Social Good (e.g.,, see what makes a good data scientist for social good) and which both I and Ben Green did in different years, are aspirational. But much of the actual target of your critique is the data economy and institutions that collect data and deploy machine learning to use it. There are critiques to be made about how modeling frames the world, and the implications of certain modeling mechanisms (which I can share a whole literature on), but that’s not what’s in this book. There’s also a critique to be made about the complicity of people who are “data scientists” in larger institutional structures, which I one thing I see in Ben Green’s work, and I’m not sure that appears later in the book.
Nick Lally: I agree with the concerns listed below, but to the best of my knowledge, there are no place-based predictive policing systems that are known to use arrest locations.
Shannon Mattern: I’ve spoken with a couple folks who’ve written about — or are writing about — predictive policing, and they agree that it’s not clear if arrest location are used, or are likely to be used in future platforms. So, it’s probably best to substitute another variable here.
+ 1 more...