Data Feminism Reading Group – Chapter 2

Once you know because He’s encouraging people to go back to work and some people are listening, but other people’s either or not or they can’t yeah way to get people unable to file from So that people will come in yeah, I mean, yeah, we can keep talking crazy My dress is really puffy. So it’s gonna like I just noticed that was him fading in and out That’s me The kid icarus one, right This one is from Georgia Lupe’s that’s right, but wasn’t it like an homage to Some sort of space thing. Oh You know, I should know that since I’m wearing it. You’re probably right. I Was so gone front that was close to not fit me I know yeah, I’m I like love them although they’re very puffy they’re very like Flimsy I feel like welcome to the welkins to regroup. I feel like you’re getting the full range of Feminist conversations in our and their little while we wait for everyone to log in last week. You guys kid combo and today you’re getting Fashion Yeah, yeah, what’s the can update for today? Lauren should be a mmm, who knows They may they may come down and make an appearance my younger daughter is wearing Fairy wings Batman underwear and hopefully a shirt if she makes it down here, but we’ll see older daughter has answered a Short but intense panda obsession like it just sort of came out of nowhere, but it’s all pandas all the time right now. So They may make an appearance too. I Guess whatever gets you through, right? whatever happens Great. Okay, so I’m gonna start Sorry Well, that’s great lots of international folks hi all international folks Serbia Wow Detroit hata wanna Denmark Vienna Warsaw. Oh my goodness. Oh They’re coming so fast, right, New Jersey, New Jersey fried here. There we go. I Just saw a question while you’re starting the a slide show about the book being out of stock and the 20% promotion So there’s actually a I actually am not sure about the book being I can we can find out when it will be back in stock. However, there’s a better deal for 35% off which is part of the MIT, press spring sale I actually off the top of my head don’t know the promo code, but I wonder scare you sit in back I’m doing it right now a spring sale 20 20 is the promo code and you actually get 35 percent and if it doesn’t work, let us know because then we’ll bug our publicist to make it work And that’s just via the MIT, press website it’s the not via secondary sellers, however You can also ask your local bookstore to order and they will get it for you Great Oh, shall we diamond Sure All right So hi everybody Oh to Answer one question when you see going by the book code does not work on Amazon You have to buy it at the MIT, press bookstore and into that discount code. It won’t work on Amazon Oh Kate ostrich, um email us if your promo code didn’t work Any any like problems email us? And then we’re gonna like forward all the things that they didn’t work to publicist Cool – hi everyone. Welcome to Our third week of the reading group this week. We’re working on chapter two of Data feminism. So thanks for joining us. It’s been just so incredible for us to see the participation to see all of your amazing geographic diversity that people are coming in with so Yeah, we’re just like thrilled to be able to have this conversation with you all every single week and get to go more in depth than our other Yeah, there are other venues and events and things like that so but we’ll just dive in this as before the way that the event goes is there’s plenty of

Feel free to do chat and the chat window, but then for any more formal questions that you want to address at the end Please you can start even now just putting those into the Q&A box. And so that’s where we’ll source the questions So as before we like to open up with an indigenous land acknowledgement Even though we’re all virtual. We are currently standing on unseated lands And so this is an indigenous land acknowledgment that comes from the MIT indigenous peoples Advocacy Committee So in exploring and writing about the imagined foundations of a society and a culture We feel it necessary to acknowledge the very real foundation of our own we therefore acknowledge indigenous peoples as the traditional stewards of the land where this event is taking place and The enduring relationship that exists between them and their traditional territories The land that we’re on today at least from Lauren and i’s perspective Is the traditional unceded territory of the Wampanoag nation and the Muscogee Creek people We acknowledge the painful history of genocide and forced removal from this territory and we honor and respect the many diverse Indigenous people connected to this land on which we gather and perform from time immemorial Okay, and so then yeah just a couple of logistical things One is that thank you to Kasich Whitney on Twitter They pointed out that we that our captions from the videos that we’ve been publishing after each of these sessions are They’re not terrible terrible, but they could definitely be improved So we found that the it was doing intersectional feminism was intersection feminism and just things like that So we’re reviewing those before we post them There’s again this ongoing Google Doc that we’ve linked to here from the slides where folks are putting their names and their Twitter handles and they’re posting resources So feel free to join in and post resources there We have been sending out past couple weeks we’ve been sending out a PDF of the chapter that we’ve been reading each week Just our we’ve been posting it on Twitter We will we did this today and then we’ll do it one more time for next week but then we would just say try to purchase because we’re not going to do it for the whole book and Remember that your money for purchasing the book? goes to indigenous women driving and Karis books who we did a feminist shout out to in the beginning, so it’s all going to a good cause Beta live set up a data feminism reading group list on Twitter So you might want to get in touch with him if you want to be connected there And then here we listed the artists who brought us the zoom backgrounds for today My outfit is from Georgia Lupe’s clothing line, which I just want to say. It is amazing. I’m Georgia. Loopy is amazing Our toilet area for target, but that oh, yeah. Lauren is my target Yeah, a big endorser of data feminism Spite the fact that we trash them in the books Um, and then our child care is brought to you by lawrence partner and my neighbors who are currently watching my children Yeah, okay So where are we in the principles today last week? We talked about examining power This week we’re talking about challenging power because of commitment to Feminist thought and feminist action means that we can’t only sit around and analyze how power works in the world We actually have to work to change and shift that balance of power So today we’re in the chapter collect analyze Imagine and teach and those are sort of the four Are four starting points that we recommend for how we can go about challenging power and so the shorthand for the principle of Challenging power is that data feminism commits to challenging unequal power structures and working towards justice And so I think it’s here Where I passed tomorrow? Yes, okay. Hold on. I just have to switch the slide for you Okay, hey super Hi everyone. Thanks so much for tuning and it’s really great to see everyone here And you know, I know there are a lot of demands on everyone’s time. So we appreciate you making the time for us So just to sort of refresh our memory so last week we in our discussion about examining power. We talked about the value of Collecting counter data, especially in order to fill in the gaps left by what we call it after me me

Oh, no ha missing datasets, right? So these are Datasets that you’d expect to exist on because they would seem to help address pressing social or political issues But they don’t exist because of a lack of institutional political Personal will or some combination of those. Um, but then we also discussed how sometimes collecting data or Offering up your experience to be collected as data isn’t the best path forward and actually at times this can lead to significant personal and even bodily harm right and so I forget if we cited this example already, but we’ll talk about it in a couple of weeks like a Really? Sort of good bad example from the present moment is the case of undocumented immigrants, right? You know people come to this country. It would be really nice to be able to Know where they are so that you could give them information about resources that might be available to them You know, like how do you sign up for school? Where how do you you know get access to various social services? Where is the local community center? Something like that? So knowing where these people are located On the one hand, you know this data Could be used to help but this exact same data Very clearly and obviously could be used to send like ice right to their door right the exact same record of an address And so this is just an example of how data is Very often a double-edged sword and we say this is something we say a lot in the book, right? So the very same data can be wielded two completely opposite ends. So Asking what we described last week as who questions if this matters a lot for data science So asking questions like data science for whom? Data science by whom data science with whose values and interests in mind So you can ask these questions about data collection and you can ask them about data analysis You can also ask them about the display of data. So we’re gonna start this week with a sort of a comparison and it has to do with these two maps that you see here and How they help to show how much like data sort of data technologies Are the methods of presenting data? In this case mapping like the data themselves can be used to vastly vastly different ends And so I’m gonna pass back to Catherine to talk about the first example of us So focus in here on this map which is by the Detroit geographical expedition and Institute led by Gwendolyn Warren And this is a map it’s it’s older now historic now, it’s 1971 was the year that this map was released It was I think important to note was a map is actually widely circulated But it’s important as part of this whole report that this DGE I did Which is about the geography of children in Detroit And it’s very provocative. If you look at this map, it’s called where commuters run over black children on the points downtown track and so it uses really sharp black dots to illustrate the places in the community where Children were being killed but basically by white commuters we were driving in from the suburbs and The people who lived along that route and in these communities in these neighborhoods They knew about this problem. They hadn’t known about it for a long time As well as you know, the profound impact it had when folks are losing children from their community Um, but Gwendolyn Warren describes how just gathering data and support like this thing that everybody in the community at least Knew was already going on turned out to be a really major challenge because no one from the no sort of official Organization was actually keeping detailed records at these deaths or making information publicly available. So going on or unexplained, but we just can’t even get that information how these Political connections basically to kind of get inside the institution for that get the data to make them out And so this is this is a really interesting map and it’s a really interesting example of academic and community collaboration That was not always harmonious Gwendolyn Warren was a young black organizer from Detroit And hooked up with these mostly white men academic geographers from Michigan State University among couple of other institutions and The youth the youth from the neighborhood learned these cutting-edge mapping techniques So these are like the most kartik advanced cartographic techniques of the time and then leverage their local knowledge to produce the series of reports and their reports covered things like

social and economic inequity They did this whole project on redistricting to propose more racially equitable school district boundaries and so on and We’ll come back to sort of how this project its legacy in geography because the way that the story was told Doesn’t correspond to reality but maybe for the moment I’ll leave it there and sort of say this is what Maps look like when communities that when they’re made by Communities like from which they came right? Uh, the leadership here is from bundling Okay, so this is also a map it’s also of Detroit it is from about thirty years earlier and It was not made by the community. It was made by the Detroit Boyd Board of Commerce which consisted of only white men and it was in collaboration with Federal Home Loan Bank board, which consisted mostly of white men And I think a lot of you might already be able to recognize this for what it is, which is an example of redlining But just for the non-us listeners here. So redlining is this term used to describe The process by which US banks rated the risk of granting loans to potential homeowners on the basis of neighborhood demographics rather than individual creditworthiness and actually one thing I learned well after I learned about the concept of redlining was that and maybe this Isis should have been It was a revelation to me. Maybe everyone else learned this first But the term comes from the fact that the practice first involved drawing literal red lines on a map So here the red areas are shaded, but it’s the same idea So in this map All of Detroit’s black neighborhoods fall into red areas on the map and this is because housing discrimination and other forms of structural oppression Predated the practice of redlining, right? so if you evaluate by neighborhood there were structural forces that made these neighborhoods lesser in the eyes of the the Chamber of Commerce and the Federal Home Loan Bank board But redlining made these forces of oppression even worse and so we’ll see this a lot this way in which they’re sort of existing structural Inequalities and then some sort of system comes in and amplifies them So in this case those who couldn’t buy in these neighborhoods because they were denied the loan But who still wanted to own their homes? They were required to look elsewhere so they deprived those neighborhoods of High-income individuals are people who wanted to invest in the neighborhood right people who are committed to the neighborhood’s a long-term growth And then fewer of these individuals committed to staying in the neighborhood Over the long term made those neighborhoods less desirable for location as locations for businesses and other long-term developments, right? And so there’s less investment from outside. And so the cycle continues and it really continues on and set the present and actually this leads to I’m not gonna cut my favorite footnote for the chapter because it doesn’t really describe a good thing but I at least it I think sort of one of my One of the footnotes in the chapter that I think raises more interesting things So this is about the forms of redlining that still exists in the present that are not necessarily tied to a physical map but still have to do with the legacies of Racial social class based inequities Sort of in relationship to geography so in the footnote. We talked about Sofia Nobles concept of what she calls technological redlining To explain how technologies like search engines and Google in particular reinforce oppression and engage innovational profiling There’s a more specific term that is being used increasingly a digital redlining Which describes how digital projects and services are similarly? distributed across different geographies and in the book actually in this footnote We cite the example of the district disproportionate number of Pokemon go stops in white neighborhoods versus black neighborhoods Which is not again. It’s not like a high-stakes Example, but it’s an example of how these kind of Geographic or geography based decisions sort of carry over into all aspects of life including Aspects of life that you just sort of wouldn’t think about as being racialized And then another example of this discursive redlining which is coined by this team that did this really interesting paper that looks at descriptions of local businesses and how they correspond in particular how Restaurants are described on Yelp and how they correspond to ideas about gentrification Right, um, but so back to this redlining map One of the things we wanted to try to call the girls attention to in the chapter

Is that the practice of redlining is actually very similar to Big Data approaches of today? I mean redlining was happening in like 30s and 40s in the United States but if you look at how redlining was undertaken it was a Very high-tech for the term the time they were using very advanced cartographic methods Surveying methods they were actually combining qualitative and quantitative data as they had folks on the ground walking around and making very discriminatory Judgments such as like how many immigrants they saw in the neighborhood and stuff like that But this was done at scale, so this is something that was done, you know a lot of US cities And sort of deployed as a tool as a particular kind of technology And but thinking about this is sort of like in comparison to the other map. It’s neat about well, like what did this map? Succeed in doing and one of the arguments that we make here in there. It’s based on many arguments many other people have made Is that redlining and practice and certainly is a scaled practice at the scale of the nation? This work to ensure that wealth remained attached to the racial category whiteness So basically if we think back to last week’s conversations about minoritized identities and dominant identities You know racial Identities in the United States are you know so much of oppression what happens in the u.s. Context is promise from Rage the dimension of race. And so it’s worth thinking about how whiteness is not only a Kind of signifier of race and we want to be really careful here to not say that whiteness is a Biological category right? It’s actually a social category as many people have said The category of whiteness has a long history and in particular whiteness was in a way first theorized by people of color so written about by James Weldon Johnson WEP to boys states Baldwin and other folks will be the case that you know in order to Survive black people had to understand whiteness as a category But to beyond just the social category to identify an individual person if we look at what whiteness does at scale It becomes the system for maintaining racialized economic dominance so there’s this tie between Whiteness as a kind of location of categories to note like a group of people or an individual persons racial identity but also a system in which the people who embody that identity remain on top and the system works to ensure that that status flow is maintained And at the same time so and we should probably say like redlining was just one of these sort of Big Data Technologies that work to ensure this kind of perpetuation of the status quo So the insurance industry was implementing similar data driven methods at the time Zoning laws were another legal tool and neighborhood covenants were another way so it’s kind of like reinforce both a time an Economic order which both hand-in-hand with the dominant racial order And so we know in the book Cedric Robinson terms this phenomenon vision of capitalism And so all that’s to say that like these two things are sort of inextricable and we quote Cheryl Harris who has a Very influential essay called whiteness as property Where again? She’s speaking about whiteness as this link to the the economic Effects of whiteness as embodied by all these different tools for maintaining white dominance So she says whiteness retains its value as a consolation prize It doesn’t mean that all whites will win but simply that they will not lose if losing is Defined as being on the bottom of the social and economic hierarchy the position to which blacks have been consigned Economy Lauren’s eyepatch that keeps you now I think so. Yes. Yeah. I just I feel like that idea of the consolation prizes I mean, it’s such a good encapsulation of how whiteness functions as sort of as the social category right this idea that like, you know if you are white It’s you get it regardless of what else you do And I just think that’s a really helpful way of thinking through what we’re talking about when we’re talking about this idea about whiteness In any case where I just started to come back to the Maps So to sort of sum up this comparison So the the redlining map is really one that tries to secure the power of its makers, right? So this is the people who made it are these white men on the Detroit Board of Commerce. It’s about their families

It’s about their communities and it reflects this desire not just secured their property values but also more broadly so protected preserve home ownership as this method of accumulating first wealth But then also status and power right and they want to protect it as a mechanism that’s available to white people only And so we see a lot of today’s are like data driven solutions Being deployed in very similar ways right in support of the interests of the people in institutions in positions of power Whose world views and value systems differ again? Really really vastly from the communities whose data the systems rely upon So like if you want to think of an example that’s being discussed right now like the prospect of these Co vid tracking apps right So if we open up the government to allow the surveillance of its citizens whose interests are going to be preserved And whose are not and it doesn’t really take much imagination Given the history of surveillance and the over surveillance of black and brown people in this country to see where the surveillance for lambda But the DGE I map is a little different right it Challenges the unequal distribution of both data and power and it does so in a couple of ways it does so by collecting its own Counter data it then uses those data to deliberately and explicitly visualize the forces of structural oppression that the neighborhood faced It did so under the leadership of a black woman who was an organizer in the community the was helping to make the maps Obviously it involve that community in the mapping process and then as we’ll talk about a little bit later It actually even involved them in some schooling actually actual college credit relating to the skills that they were learning to use and Then the academic geographers who were also involved in the collaboration, they provided support and access to their own institutional resources Right. So their access to powerful institutions was what enabled them for instance to give college credit for the mapping skills the the mostly Teenagers and young adults in the community we’re learning And then they employed their institutions printing and publishing resources to circulate the final report more broadly And things like this So, yeah Exactly these reasons we feel that the DGE I under the leadership of Gwendolyn warren Provides a model of the second principle of data feminism which is to challenge power So challenging power with data signs requires mobilizing data to push back Against existing and unequal power structures and to work towards more just and equitable features and so challenging power is really linked to Examining power because of course you can’t challenge power if you started to don’t understand power and inequality in the environment So we first have to work to examine power, but we really can’t stop there I think data feminism and feminism more generally implies this commitment to action. This is writing the balance of power and so The action that one takes to challenge power can take many forms And in a way, this was a hard chapter to write because it’s sort of like there’s so many ways You can challenge power. So these are for what we would call starting points But if this is not an exhaustive list And certainly there are many other things that one might add to this list for we that we can challenge powers of data science But briefly they’re collecting and analyzing counter data as we see in the example of the DGE I was trying to get the data that they’re not making publicly available Analyzing and auditing algorithm other kinds of systems imagining the goal of Collaboration we serve need to reframe what this work entails And then teaching the next generation So let’s just dive it there So we’re gonna dive right into collecting and analyzing counter data And just to say here you might think back to last week’s chapter where we were talking about Feminists. I’ve data collection the work that’s being done my by Maria Salguero in Mexico That was an example of again missing data data. That’s extremely important to people’s health and well-being And future possibility in the world and yet nobody’s collecting it. So stepping in to collect where others are not the DGE I also collected things like they went out into this playground study and they went out and I’d actually looked for broken glass on playgrounds and then extensively weighed Broken glass on playgrounds to show there are these like racial disparities and how playgrounds were maintained in communities of color versus white communities and then, you know we can even think about Collecting counter data right now in relationship to Cove it like where there is. There’s so much missing data

We’re missing data around race and ethnicity We’re missing data around sex and gender of both cases and deaths and this is a place where a group called data for black lives have actually stepped in when they’re trying to press the government for releasing collecting and releasing Race and ethnicity data in conjunction with Kovach data so this idea counter data can be given the right circumstances a powerful strategy for Change but now I’m going to take it some lorem to tell us about analyzing and auditing Robert Sure. So, yeah, so You know Once you have the data or once you’re you’ve found data from someone else the question is like what do you do with it? So One thing you can do is well Analyzing data, that’s a broad category but within that category a Specific thing that you can do is fought it some of the algorithms that exist out there in the world So I’m just going to talk through this via an exam recent example that does this really well and so Yeah, so this example it starts in 2016 when Julia Angwin led a team at Pro Publica The sort of data journalism outfit to investigate something called risk assessment algorithms So these are computational models that are packaged and then sold by for-profit companies that determine the quote risk score Of a person who has been accused of a crime and then they’re used by judges to inform decisions about a range of things but like the length of a particular prison sentence the amounts of bail, that should be set or Even whether bail should be set in the first place and again, so, you know, I think about these cumulative effects, right? So if you’re denied bail, even if you end up getting exonerated, you know If you’re hell, if you’re held in jail while you’re awaiting trial like if you have kids to watch, you know You can’t watch them. If you have a job to go to you lose it. There’s all sorts of compounding effects, right? And so the higher the risk score The you know quote riskier the person and therefore the longer prison sentence They might be issued or the higher bail amount if there’s a bill set at all And so on and then the lower the score the less risky this person According to the algorithm was determined to be and therefore the lower the bail amount the easier to get out of jail The shorter prison sentence and so on Sort of how are these decisions being made? No one knew Because it’s you know blackbox software and then was it possible that they were racist and the short answer is like, of course they were but The more interesting answer to the second question is how Pro public can I sort of determine this? because I was going to look at a couple of sort of phases in the process I’m so reporters for ProPublica and I should say there’s a whole team involved in this and you can find that all the team’s names on the So they began by looking at Northpoint, which is the company that offers one particular Type of software called compass which is used in I actually don’t know the statistic off the top of my head but it’s used in a lot a lot a lot of Local municipalities across the They used the Freedom of Information Act I believe or it might have been it’s based in, Florida so it might have been like the state version of that to obtain the risk scores that were assigned to more than 7,000 people who have been arrested in Broward County, Florida And then they actually checked on those same people to see how many of them in real life Um actually went on to be charged with a new crime over the next two years And what they found was the algorithm was incredibly unreliable So only 20% of the people who were predicted to commit violent crimes actually went on to do so And they also found that the formula was more or less was more likely to false flag black defendants as future criminals Whereas it was more likely to rate white defendants as less risky than they would actually turn out to be And then they ran a regression analysis that allowed them to isolate variables like race and age gender And they were able to show that among their data. The black defendants were 77 percent more likely To be rated a higher risk than the white defendants and What’s interesting is that the results of this analysis has in turn entered into several legal cases and really helped change the conversation About the use of risk assessment algorithms. So in response to this piece of journalism the New York City Council proposed and then passed the four nation’s first algorithmic discrimination bill and Formed something called an equity and fairness and City algorithms task force the piece also prompted more than a hundred civil rights groups to come together to write a statement against the use of These pretrial risk assessment algorithms and this piece has actually been extremely influential on academia So it’s been cited in almost a thousand academic papers

In a social sciences, so you always need to like scale the citation count for various fields but this is not normal for like a piece of Journalism right to be cited among academic Papers and so just briefly what you’re looking at. So on the top left is part of the form that Defendants are required to fill out in order to generate the data for the algorithm and note that it doesn’t ask about race directly But it does ask questions that sort of research has shown our proxies for a race like, you know, I actually can’t even read They don’t have okay so they have a question like You know has your wife husband partner ever been arrested that you know have your brothers and sisters ever been arrested as you know and so compare that to like what we know about how Black and brown people in this country are overly arrested in relation to white people, right? So already you’re getting sort of racially skewed answers to questions. They ask questions about like did you get in trouble in primary school? Similarly, we know that black kids are punished more severely and more often as young as prior, you know as long as nursery school kindergarten In comparison to white people in this country. So any so there are these questions that sort of serve as these proxy surveys Okay, so that’s one part of the piece On the top, right? You see some of the analysis performed by ProPublica as part of their reportage. So this is all available on their website It’s a Jupiter notebook. You can just go take a look at it And then that bar chart that you can sort of see I don’t know how large your screen is But it shows the distribution of risk scores for black defendants versus white one So you get ranked on a score of one to ten with one being less risky and ten being? The most risky and you can see on the left if you are black. There’s like an even distribution between Zero or one and ten of all of the scores but if your white looks like You know a third if not more of the white defendants were given the score of one the lowest risk and it tails off Dramatically after that with very very few people being given the highest risk for recidivism And then oh and then on the bottom I just she I put in another paper that actually we didn’t talk about in the book, but I think it’s a really interesting example So it’s a Paper by Ben green and Yi Ling Qin. It’s an experiment serving the social science sense based on the initial work of the ProPublica reportage that attempted to determine how just sort of the knowledge of these risk scores affect human decision-making so they ran an experiment where they gave they did a run group was given like a textual description of the What was it presented about the defendant and the crime? Ahmud then said like how would you evaluate their risk? And then the second group was given the textual description? Plus it said like and the algorithm determined their risks to be X and then they asked that group. How would you rate that? How would you rate their risk store and what they found was that if the risk score was? lower than the human determined score It didn’t ultimately affect the score very much like people stuck to their principles If after they saw the if they were like, oh I already decided the risk score is lower they mostly continued on with the same risk score, but for black Defendants if the risk score the algorithm score was higher Than the one that the people came up with it caused the ultimate score to go up Meaning that the higher algorithmic score gave humans sort of more of an excuse to serve up their assessment of the defendants risk So anyway, um, you know, I wanted to call this out because you know, this was done by two grad students I believe you know It took a lot of creativity to imagine this experiment but the actual experiment like used existing data Did not require a significant money in order to go do it You know, it’s just a really good example of a thing that you can do If you have these skills either of like designing experiments or a data analysis and you want to sort of push this work forward, okay? Great So we want to do like a quick little pause For a second here Because so far we have talked about collecting and analyzing counter data and then analyzing and auditing algorithms You’re like auditing these data driven systems. Um, but we just so it’s so like you might think okay Well, yeah like this maybe this is like about like more data is better. Let’s always get some more data And let’s always use that data to unmask the question. So like Demonstrate quantitatively that looks racism attack inator look sexism is happening or both of them are happening but we just want to do a little pause here and again, like if you think back to these questions from the last meeting and we want to introduce a feminist to a question, which is Who is it exact that needs to be shown the harms of these? Differentials of power and what kind of proof do they need to believe that oppression is real?

And so for example, like in the case of dge, I the community knew this was happening right like they didn’t need convincing The Wunderland worried did not need to make a map To show him/her community that this is the thing that is going on like they all knew that was happening But it was to the folks who are in power warn To the doctor dominantly like male and white dominated certain institutional power structures in Detroit that they are making this data-driven case So it was the dominant groups that both were in a way like cause of the problem and also in a way that ones that were Ignoring the problem and we talked a little bit about that too. And which is this idea of the privilege hazard. So when Institutions are controlled by dominant groups. There’s this sort of collective ignorance And unlikely to see these problems in the first place So thinking about like it to whom do we need to show the proof and then? When is the proof ever enough? And so like thinking about how sometimes we actually in the book there’s a footnote which we would encourage folks to look at Candice la Gnaeus wrote this very widely shared blog post called your demand for statistical proof is racist where she basically says how Folks in positions of power will accept anecdotal evidence from people like them But then they will demand endless statistics from minority groups so it’ll always be like Oh We need to do more data collection or we need another study or we need to like Really prove that this thing is happening in the community before we can take any action about it and so we just want to introduce this idea that like More data doesn’t always lead to more action precisely because of these differentials of power because ultimately in many cases like who can actually take Like systemic action are the institution’s controlled by by dominant groups Okay, and then this a related something that is sort of related to I guess this caveat Is this idea of the deficit narrative, which actually I was starting to see that come up a little bit I’ve been trying to sort of monitor the question, but I’ve been seeing that come up a little bit in the questions where When we are unmasking racism or unmasking oppression with data and data science maps and things like that How do we avoid? reinforcing stereotypes about various communities And these are these kind of deficit narratives are narratives that reduce the whole group or a culture to its problems Rather than portraying it with strengths creativity or the agency that folks actually possess And so Kimberly steals a lurch and we talked about her work last week and we quoted her in chapter one around the maternal mortality Statistics and so she does the statistics that are rightfully creating awareness around the black maternal mortality Crisis are also contributing to this doom and gloom narrative Where white people are like, how do we save black women? And that’s not the solution that we need the data to produce so many and if you go back and kind of look look at the minute much of the coverage of the maternal mortality Inequities. It really does portray black women as victims They fail to amplify the efforts of black led organizations who’ve been working on the issue for decades and so on And this is likewise along the lines in lines and Indigenous folks as well. So there’s this amazing book indigenous statistics where the author’s talk about how When in digit when there has been data collected about indigenous communities it often functions to document what they call difference dysfunction and deficits so this is I mean I think it’s a limit of a quandary that all of us need to reflect on a little bit because Those of us who work with data may think we are helping when we are providing proof of inequality But we also need to be really sensitive about how that proof plays into existing stereotypes of mine As people as passive as victims as having no agency and then one like sidebar a Grumpy note of my own is like I feel like a lot of research about women and stem falls into this category Where it’s always like boom owning the fact that there’s no women, you know like the women are leaving. Where did the women go? What’s wrong with the women? How can we get the women back? When actually what we need to be thinking about is why are the men not making space for? Binary folks, right and so like we’re sort of focused on like fixing the wrong group So that’s one way to turn over the deficit narrative it was like focus on the right group

And then secondly is it’s also thinking about how do we really work with? Communities directly and authentically and we’ll discuss that in a later chapter and that just leads to my Favorite footnote from this book which has to do again with dge. I And the way that the work of DGI has been historicized in the Academy which is not corresponding to reality so the work of DG I Actually inspired a generation of critical cartographers. I know there’s a couple of them out there today like Whitney mogul and Tim stomping I’m pretty sure our reading audience today At the bay know I miss this history. Well I’m talking about And so it became kind of pretty famous, but it was almost always credited for the work of one white male academic geographer named Bill Bunge a lot actually a lot of social capital and a lot of prestige in his field for Collaborating with and transferring knowledge too as they wrote it the disadvantaged blacks and so even early on if you reach the old notes three, you can see that Warren herself pushes back against Right. Well the projects happening. She’s like no, that’s not how it works. And that’s that’s not how it’s working But this white male save your narrative has persisted in academic geography So much so that actually very recently Wendell and Warren and Cindy count themselves collaborated on paper To kind of retell the story and said the narrative straight and center the work back on getting Lauren’s leadership So sort of like even with folks and I’m saying this is coming from academic geographers would fancy themselves Anti-racist and anti-sexist and so on but there was a way in which the the deficit narrative is convenient for people to then have a savior narrative like oh I Saved the poor people or whatever and so we just need to be really sensitive to when we start to sense those Deficit narrative and senior narratives and sign into the equation All right smart Turn it to you Okay so I’m just There’s always too much to say Okay, so we’ll move a little bit more to get through. So we’ll keep on going um, so the third way that we talk about challenging power in the book is just imagining this goal of Collaboration. I don’t know – offering do you want to talk about the chart? Do you sure? Yes, totally? Yeah Yeah, I love the church, um, so so yeah one of these things to then thinking about how do we how to think about the work that we’re doing and so there’s a whole section in Chapter three, that’s like Questioning whether or not we are using the right concepts to describe And so folks may be familiar with this chart that we have here and One of the reasons why we made this chart is because on the one hand There’s a lot of really interesting work that is happening right now Especially in technical communities where as issues like algorithmic bias and discrimination Happen starting to come out in the press This has really pushed folks to think about ethics and technology and to consider what like, okay. Well, why why are we producing? You know racist risk assessment algorithms or facial recognition systems And so on the one hand all the concepts on the left, they are circulating right now It’s very wonderful that these conversations are happening literally just like three or four years ago. There were no conversations like this So the fact that we’re talking about this is great but one of the things we feel like an intersectional feminist approach can add here is a Kind of a reframing of some of the terms of engagement of these concepts Specifically because the concepts on the left for the most part use is not for everybody but for the most part and When we look at these concepts, they’re locating the source of the problem. So say we’re talking about the risk assessment how They’re locating the source of the problem either an individual’s so like You know racist individuals or else in technical systems Like oh if we could just improve the data then the risk assessment algorithm would be more fair Right. So there’s this idea of like the algorithm is biased or You know, there’s an individual person whose intent was to do harm and their system of bias But those nice other people who didn’t intend there sexism Their thing is fun And we just feel like there’s a deeper conversation to be had here Which is more of a conversation about root causes because without understanding the root causes of why we keep introducing data products that are sexist and racist

We can’t actually we can’t like mitigate We can’t take steps to mitigate the influence of inequality in our systems at any level We don’t know what to look for and then we can’t also even like evaluate them properly So on the right hand, we propose concept that other folks have been talking about not only us but these are concepts that challenge power because they’re addressing this idea that there is a root cause to all of these things and that root cause is You know what, we would locate the root cause as is the matrix of domination as we as we talked about last week And so how do we challenge the matrix of domination and well instead of talking about ethics? we talk about justice and to me look at how can I lose be just across many different social groups and instead of thing about individual bias we think about structural oppression so systems like the redlining system along with all host of other leg tools and technologies that kind of Propagate and reinforce whiteness. These are structural considerations that reinforce structural oppression so that in that way again understanding that power we human challenges how it really fairness again, an interesting concept of why be starting to be used right now, but Where does that leave us in terms of history often? People’s concepts of fairness and certainly the kind of fairness that we’re trying to bake into our AI systems right now Starts from one particular time and that time is the present So as soon as we all arrive in the present with equal opportunities and equal sort of Histories and our ancestors have equal histories and they did not So there’s these huge inequalities historically that have brought us to this point and equity is the concept that actually can Sort of hold those and help us develop ways of addressing them And then so I’m not gonna go through all these cuz anyway we’ll but the one last one that I want to talk about is collaboration because I you know, it’s sort of like why why don’t we just want to build systems that are Can be hot like hold people in power accountable. What if we can build systems that actually aspired to collaboration? And here we wanted to just point out some of the work that inspired us in thinking about this value of collaboration But it’s basically this idea that oppressive systems hiring Polybus So if not, if none of us are free, unless all of us are free And there’s a great first Concept of this that I came across was the same on the left from the aboriginal activist group And they say if you have come here to help me you are wasting your time But if you have come because your liberation and bound up with mine then let us work together So again, it’s sort of like fighting back against that like deficit narrative save your narrative thing I’m thinking about how do we work together? And how is our liberation bound up with each other and then a couple of other? books and projects here just to do a shout out to these groups that are also thinking very much about data design and how we can actually use those things for the purposes of collaboration But here’s the point I passed away so I’m going to try to do this quickly and we keep on Eliminating slides and examples every time so maybe next week will be the time that we actually leave time for a lot of questions So the final way of challenging power that we talk about in this chapter Is teaching and we observe there and I’ll observe now, you know This is often overlooked and sort of undervalued not the least because women are Historically the ones who have done most of it And you know, this is something that we see this in this sort of emphasis on teaching and sort of the ability to empower And serve transit or enact transformation through the next generation of in this case data scientists This is something that Gwendolyn Warren it recognized when she insisted that The academic geographers not just use their skills to help the community but transfer their skills to the community teach these courses on mapping and data collection and survey design that the people who took them in the community actually received college credit for And we actually in the book we talked about this other example, which is more contemporary Which you have a picture with yeah, and this is the local lot of a project So this is a project of the city digits design team which includes a math education researcher Lori, Lori rebel It includes Sara Williams and the Civic data design lab at MIT along with the Center for urban pedagogy, who which is a group based in there and It was designed to foregrounds sort of teach

data, science and statistics But centered on an issue of social inequality and more specifically on the lottery, right? um so it’s widely known that the lottery takes more from poor neighborhoods in the form of lottery ticket sales and then redistributes the proceeds of the lottery to cities and towns and municipalities around the state which means that more rich communities end up receiving the money And then like as we all know your chances of winning the lottery are your chances of winning lottery, right? So the curriculum involves students interviewing people in their neighborhoods about the lottery What they think about it do they buy tickets? How many do they buy stuff like this? Mapping stories in the neighborhood that sell the tickets. I’m analyzing the advertisements for it like if you’re in New York and you look at the set like you ride the subway like You know that there is lottery ads everywhere And also crunching the numbers so like actually learning and employing what in your I guess their Common Core math standards or whatever And then in the end the students create infographics like the ones that you see on the right over there Along with and we have pictures of this in the books or like data-driven opinion pieces that show their conclusions And so we like this project. It’s really inspiring in part because of its successes, right? So there was a formal educational assessment that was conducted In the beginning in the end of the project and students did learn more about in math and stats than other students in more conventional Classes and it’s also really inspiring because it tries to route data science education in a real-world Ethical question, right? And then also it values lived experience alongside Sort of these data collection efforts and even sort of shows the students how lived experience can be transformed into data themselves But the other reason why we like it is because of the ways that Sort the project didn’t quite meet all of its goals around social justice, but then how the project team made these shortcomings public so one of the things that the team talks about in the paper the documents this project is that the team itself was mostly White and Asian and their students who are mostly Latin X and black and so when the students brought up issues of race Which apparently the students wanted to do all the time the teachers felt unprepared with how to engage them? in their write-up of the project they vowed to Incorporate this explicit discussions about race into future iterations of the curriculum Another interesting issue that came up and actually this is like this is super to think about is the issue of the lottery itself Right. Um many of the students pushed back on the central focus on income inequality and really felt like the it was sort of passing judgment on their communities and The students sort of like they say and which is true is that most people don’t don’t expect that They’re gonna win the lottery when they buy a ticket right like part of it is sort of Indulging in the fantasy that if you buy a ticket like it might it’s a way of escapism right and the course designers didn’t really capture the nuance of the lottery in these communities and They also sort of didn’t bring in community members who have been working on broader issues of income inequality At the same time and so anytime there was like a lesson about income inequality. It was sort of top-down rather than Sort of add an equal plane with the students themselves and so a second version of the course was proposed that would incorporate the experience of People in community groups in that very location that have been working on these issues for a long time And so I hear my younger daughter having a minor meltdown, but we’re almost done so We want it to end on this example because to us it really encapsulates what it means to try to Challenge power and privilege through data science, right? Um, so those of us who like us me who hold relative positions of power we’re constantly learning how to be better allies and accomplices right across areas of difference and One of the biggest strengths of look like and you make mistakes I make mistakes all the time but we thought that one of the biggest strengths of the local auto project was the courage of the creators to sort of publicly and reflexively interrogate themselves in their process and to really detail where they went wrong and Then specify how they plan to do better in the future Okay, I might just mute myself Okay, so in the interest of time we are gonna try to do better We want to leave time for questions. I realize we really didn’t do time for questions today. So we will try to do better I think we’re very enthusiastic to share a lot of things very well And so that’s I think doing better next time So I’m gonna skip the summary, but I did want to do the feminist shout out

to two really important groups one the trans lifeline Which is a really interesting group that got in touch with us So notice how there’s this thing here that says get in touch with us if you want to shout out to your station They got in touch They are a super interesting organization trans LED Connects trans people to community support and resources that they need to survive and thrive even a hotline And it’s all staffed by chains folks on the hotline and then they also have a micro grants program that offers low barrier grants to trans people for things like name changes and updated government identification documents and then even after support for undocumented trans folks and Incarcerated folks so a really good organization to go out and follow monitor their work And then somebody from our community just posted on Twitter today You all might remember Maria South Carroll, who is the person who’s not being feminist signs in Mexico? Whose work we feature in Chapter one? she has gotten in touch and with Twitter and Said I need help. Basically. She her mom has had an accident She ran out of money and her she was so tired from working, but she spilled something on our computers You have a computer. So she’s basically like at the end of her room and we have a paid mouth here Where you could send money to her address that I just is Q are you cute at I at amazon.com? Anything really amazing, there’s folks on Twitter who are already saying I’m sending money and sending money. Don’t worry But she does like just I think amazing work And it’s work that’s often kind of Alone, and so it’d be a great response from the feminist community that wants to support her and see this work go on Into the future that because it is quite sort of precarious to be mapping and monitoring some mini sites in this context So if you have like 10 or 20 bucks, please send it to her At that one. Oh, thanks Alice for posting that posting her email there And then that’s basically it for today and we will see you all next week for reading chapter three of data feminism and We are really going to have time for questions What about our failings – Sam, yeah We we do save all the questions and read them was there now when super inches on your question we would love to have like a three hour long conversation, but Thank you, bye You