The Digital Humanities has existed as an institutionalized field of research for more than a decade, drawing undergraduates, graduate students, and researchers from around the globe. This series of interviews spotlights leading research in the digital humanities, and asks key contributors to reflect on the expansion of the field, its culture, and the major misconceptions that remain.

Why has this field sparked so much public engagement in its projects and debates? How has the digital humanities changed what it means to be a more “traditional” humanist? And how is the field engaging new developments in technology like artificial intelligence?

In this interview, Interventions editor, Utsavi Singh, speaks with the director of the Data by Design Project at the Emory Digital Humanities Lab, Lauren Klein, and three co-authors on the project, Tanvi Sharma, Shiyao Li, and Margy Adams.

Utsavi Singh: Thank you all for taking the time to talk about your fascinating digital humanities project, Data by Design. Lauren, I would love to hear about the origins of this project. 

Lauren Klein: Data by Design is a history of data visualization, which tells the story of the emergence of so-called “modern” data visualization, alongside the histories of slavery and colonialism. It also has a technical goal, which is to tell this story using data visualization itself. Visualization has always been something that I’ve loved, and have used in my digital humanities work for a long time. Some of my historical work is set in the same time period as these early visualization examples, so I was broadly interested in what the known history of data visualization looked like. But I was also trying to think of what a fun project would look like that would engage the students who were then working in my lab with me, that could make use of all of their amazing skills. I conceived of the project back when I was working at Georgia Tech, and had just started the Digital Humanities Lab there. I had tried out a couple of projects in the first few years, but this one was the clear winner. Everyone loved working on this project—software developers, designers, humanities majors. We have some old projects and papers that document this early work. But by now it has involved three cohorts of students, both graduate students and undergrads, first at Georgia Tech and now at Emory, where we’re also working with Emory’s Center for Digital Scholarship (ECDS). The current team includes Jay Varner, from ECDS, and the people in this interview. And as they will tell you, everyone is coming from a different disciplinary background and bringing their own expertise to this project, which is another reason why the project is really fun, and also, why I think it is really good!

Utsavi: I did take a look at the project and it is fantastic. It seems to me that not only are you documenting a history of data visualization, but that you're also creating it. How has that contributed to digital humanities scholarship overall?

Lauren: Yes, it was really important to me that this be a “born digital” project, meaning it was imagined from the start as a project that lives online. That's why using visualization and interaction in order to tell the story is a fundamental part of the project. Very often, because of the constraints of academic publishing or because of institutional requirements, you need to write the book first, and any website is often an afterthought. But I was really lucky in that I had a few other book projects going on (which have since been published!), so was in a position where this project could be whatever I wanted it to be, and I wanted it be a born-digital project. It was also really important to me that it be intentionally collaborative. Everyone included here is a coauthor on the project, and there are additional coauthors as well, including Jay, as well as additional students who contributed to the project in its early phases but have since graduated. I did this really intentionally, because it is too often the case that faculty are pressured to claim sole authorship of projects that are in fact collaborative. We wanted to model born digital scholarship, and also collaborative digital scholarship, and put it forward as if to say: the best digital humanities scholarship requires people coming from different disciplines and from their own areas of expertise. They need to be valued and credited. And the work needs to be published in a public way. These are all essential components of what makes for field-defining digital work, at least according to me.

Utsavi: Yes, definitely. I would love to hear from Margy, Tanvi and Shiyao as well. How did you come to this project?

Margy Adams: I came to this project through the Emory Center for Digital Scholarship, which is where I work as a Digital Scholarship Associate. For a bit of context, I am a PhD candidate in the English department at Emory, and I study Black American and Afro-Caribbean sound and performance theory. One of the things that ECDS does is pair their graduate student workers with projects of interest that align with their research and professional goals. For instance, folks who want to enhance their videography skills might work on film or recording projects; folks who want to learn about podcasting might get paired with collaborators who create aural storytelling classroom assignments; and people who have an interest in interactive mapping might develop instructional guides for ArcGIS work—things like that. Because my research focuses on sound as a gateway to understanding genre in text, and how we can leverage sound to strengthen our interpretive skills when we read by “close listening,” I was asked if I would be interested in writing the alternative (alt) text, or image descriptions for blind or low-vision users, for Prof. Klein’s forthcoming project. I don’t think any of us could have guessed how well my work on sound would fit into a project about visualization, but it does!

I had met Lauren a year prior because I was her teaching assistant in this amazing class called Eating in the Archive. I had also taken a fantastic course that she and Prof. Nicole Guidotti-Hernández taught on feminist approaches to understanding archives and race. So, the opportunity to practice the collaborative work that we learned about in that class was exciting to me. Though I had not served as an alt-text writer on a project before this point, in my research and in my classes I highlight the importance of considering alternative modes of composition, so the task was appealing to me. Over the course of my time collaborating on this project, I have also realized how crucial image descriptions are to a project about images. Eventually, my role as an alt-text composer became interwoven with and inseparable from my own research practices. I also would like to add that while we all have specific roles on this project, they are fluid and adaptable. Our opinions and expertise are always taken into account. Even though I'm not a designer, I am impacted by the predominance of sight in our world and continually analyze the ways meaning is made and exchanged visually. So, in addition to providing design feedback from my subjective standpoint, I also started thinking about what we do when images don’t load, or what might lie behind these pictures and how to translate that opacity into words. To think about what lies behind images, I turned to Tina Campt’s monumental Listening to Images, which opened up the ways I’ve read and looked at things ever since.  

Tanvi Sharma: I would also like to bring up the considerations of alt-text that addresses both “What if you can’t see” and “What if you don’t want to see.” We operate from a place of not wanting to harm or trigger the viewer. One of our design principles throughout the project is that you don’t have to depict and restage violence to acknowledge that violence occured. And that's where I think alt-text is important, for figuring out what parts of the narrative need to be visual or need to be explained in greater detail; and when we can step away from that, and give the user or reader more agency to not view certain things. I think this specifically comes into play with the “Description” chapter that's about the slave ship diagram, and where you’re given the option to not see the images altogether. We believe that you don't necessarily need to see those things to understand the theoretical function of those forms, especially if you’re already familiar with it.

Margy: Yeah, this was something that came up when we were trying to design our visualization using information from the Trans-Atlantic Slave Trade (TAST) database. We found that we needed to consider what is lost when information is presented via that kind of list or table setup, and also what kinds of categories appear; what types of variables or quantifiable trends did people decide to mark and label? Something that Data by Design brings to the fore is how visuals are, and have been, leveraged sociopolitically, how they can manipulate, or how they're used for very certain purposes. Often, they have an argument, especially when the data being visualized comes from a historical source that participated in rendering people into data, as the TAST database reveals. That gets to the central problematic of how data is abstracted away from who or what is being studied, and how data is translated into visual content. The choice not to see is as important to us as what we show and how we describe it. The alt-text practice for me, then, became a way to think more deeply about the contours of an image: why an image is displayed or what it is arguing underneath its “objective” layer of visuality. So, in the chapter, the choice to “toggle” the visibility of the archival material depicting the floor plans of enslaving vessels was one that we wanted to take into consideration, because if images can be manipulated to enhance or expose a particular argument, as they were during their creation, they can be leveraged to hurt. The question for us became: how can we make the choice to look more agential, and how can we make sure the reader is implicated in—and conscious of the consequences of—that decision?

Utsavi: Yes, I see that there's an ongoing critical reflection of not only data creation in the past, but also as you work on this project. I would love to hear more about that from Shiyao as well.

Shiyao Li: Yes, I think Margy and Tanvi both bring up a very important point, that data visualization could also serve as a platform for arguments. I am a computer science student, and I study visualization, and I mainly focus on how people perceive data visualization. What types of visualization design could bias people’s interpretation of the visualization? Data visualization is actually a channel to accelerate the exploration of the data, and it should not bias people's interpretation. The major aspect of my research is to identify the bias and mitigate the bias in visualization. So when I was brought on to the project, I was provided that giant Slave Voyages database. And since I have a Master's degree in data science, I was not that uncomfortable with dealing with complex data and cleaning the data. But I had never worked with sensitive data like this before, let alone create a visualization to support an argument, rather than merely presenting statistical insights from a complex dataset. So at the very first stage, my purpose was to try to accurately represent as much information as I could. Our first draft was actually a statistical chart, showing the duration and the people on board for each voyage. But then the importance of the collaboration really came into play, especially as we got Margy’s and Tanvi’s perspectives. They said, “we are not only focusing on accuracy, we also want the visualization to reflect the concept of resistance.” So then I explored an alternative visual metaphor. I first tried to find something interactive that looked like a flowing river. But as we tried to map each voyage to each river, we realized that the visualization was becoming too beautiful, and the dataset was not a beautiful dataset. So we finally chose the river diagram for our visual metaphor. We used a curved line to represent each voyage, and we also hoped that this would convey the resistance that happened on those voyages. This process helped me transition from being a visualization researcher who focused on accuracy and lack of bias, to understanding that it is not only about representing accurate data—that we also wanted the visualization to reflect the argument of the chapter and emphasize the resistance on the voyages.

Utsavi: Yes, and this is the beauty of digital humanities—that it allows for certain arguments to emerge that might not necessarily emerge from text. But what I am also hearing is that we are also constantly introducing our own biases into data. And I imagine that that is even more pronounced in digital humanities projects such as this, which is very much about making choices—visual choices, textual choices. I also saw the toggle feature for the footnotes, and in a sense those choices go back to the organization of the project itself. For instance, the layout of the website allows viewers to go to certain pages first, and make decisions about where to go next. 

Lauren: This is all true, but one thing that’s important to recognize is that it’s not only in digital humanities projects that people are making choices. This happens in all projects, right? Even when you're manipulating a dataframe and you think, “Let me just drop this column, this seems not useful,” that's a choice. It might seem like data cleaning or like a technical choice, but it is actually connected to what you, the human, think is important about the dataset. And so a large part of our project overall is to emphasize that you as a viewer have a lot of responsibility to ask yourself: what choices did the designer make? Why am I seeing this presentation of the data and not another one? And the goal is not to reject the visualization out of hand, but rather, to come to a better understanding of what it is that’s being communicated. And sometimes you need to do that by also considering what is not included in the chart. 

Tanvi: At various stages in this project, while we were visualizing certain things, we would look at the historical databases and kind of take them as the starting point. But often, something unexpected would emerge from the data that we had to pay closer attention to. For example, when we were looking at the student directories from Atlanta University for the chapter on Du Bois, at how many black students were graduating and what other professions they were going to take on as their lives unfolded, something weird that I noticed was that as I was transcribing the directories from year to year, I would lose track of some of the students. And then I realized that, oh! It is because the women are changing their last names. So it’s not like they don't exist, but they sort of get folded into their spouse’s name. And that was interesting, because if I had just taken the data and visualized it without probing it further, I would have made the mistake of missing those names, without even realizing that I'm perpetuating the biases of partiarchy that apply to the original dataset. 

Lauren: As another example—and this is from a different chapter where it's a series of charts visualizing each century of U.S. history from the perspective of Elizabeth Palmer Peabody, who was (among other things) a white woman from New England—there was a thing that Dan Jutan, one of our previous software engineers, observed as he was transcribing the data from a 19th century textbook. In the beginning, the colonists and the Native Americans were treated as the same category because the dominant group they were both opposing were the British people who lived in England. The distinction there was “where, geographically, did you live?” But at a certain point when “colonist” or later, “citizen,” became an identity group, they claim the color that had previously been used to represent the group they shared with the Indigenous inhabitants of Turtle Island. There's almost this colonization of the color scheme, in addition to the colonization of the land. I don't think this was an intentional choice on behalf of Peabody. But in her colonial mindset, that color had always been hers, and so she took it. So it's a really interesting place in which even a seemingly innocuous choice of “Should this set of datapoints be red or orange?” is overloaded with colonial implications.

Tanvi: Yeah. Noticing these things in the design of the original artifacts made us stop and think “What are we doing?” Are we perpetuating our own biases, and how can we check that throughout the process?

Lauren: Right. And there’s another example of this too. We have a chapter about Indigenous map-making. We really struggled with the question of how we could visualize the data from the original maps we study in a way that wasn’t extractive. And in the end, we came to the conclusion that there just wasn't a way to do it that was aligned with the goals of the project. I mean, the question running through the whole project is: "Now that we've returned to this history and thought about these particular visualizations, how do they change our visualization practice today? How do they make us ask new questions, or do things differently?" And what we realized that we ourselves would do differently after immersing ourselves in the history of settler-colonialism is, put plainly, not to restage colonial violence through data visualization. So we visualize other information in that chapter, but we do not visualize the original cartographic data.

Margy: A lot of what we're trying to do ultimately is to make explicit where the power lies underneath all of these forms of composition. The project itself is almost like an exercise in redirection and translation. There are different linguistic practices spread out throughout this whole project, from drawings to mistakes made—both accidentally or purposefully—and some of these things we really don't know comprehensively, and we won't ever know in totality. But having the room for that kind of margin of error is really crucial to then thinking about what might arise as a possibility for the lives that were living through these really tumultuous, terrific circumstances. And so the drawings, the numerical data, and even the alt-text are all interpretive strategies, and something is gained and lost as we move from one form to another. But I also think seeing how things cohere, as well as what is inconsistent amid these various modes of expression, is how we learn about power and personhood.

Something else that Shiyao reminded me of, regarding the power of visualization is that the visualizations we are trying to make are not representations of say, the Middle Passage, but a perspective on how the Middle Passage was conceived. So resistance became a variable we wanted to think about more deeply, because it was a category that was specifically labeled in the TAST data. As Shiyao, who we call the “team wizard,” said, every time he would come up with this amazing and often beautiful visualization about resistance, we asked, “How do you visualize resistance?” This kind of fluid, weird category. But also, “How do you quantify resistance? What does that even mean?” To approach these questions, we looked to narratives by formerly enslaved people, especially Olaudah Equiano. His narrative is the example that we use in the chapter to think about the kind of interstitial space that resistance took on board, and after disembarking the ship. For example, in the “Description of a Slave Ship,” the actual document from 1789, there's this (abhorrent) footnoted text on the bottom half of the page that gives a clue as to what the seafaring enslavers considered “resistance” which, by their accounts, was quantifiable: unsanctioned, insurrectionist, male movement. Something that guided us through reconceptualizing resistance as unquantifiable is Kevin Quashie’s Sovereignty of Quiet. This book really gave us the language and the methodology to think about how life can be resistant in its quotidian daily acts, rather than the overt, pitchfork-holding picture that you often see or associate with revolution.

Utsavi: Yes, that really brings out the resistance that also underpins the project overall. I also would like to hear about the long-term life of the project. Data by Design seems to be quite an iterative project, and it also says on the website that it is a working draft. What direction will the project go in in the future?

Lauren: Funny you should ask! I know I said at the beginning that I had conceived of this as a born digital project, and I was very insistent that it be imagined first for the web. But now, as I near the end of this project, having gone through two different web frameworks, and mentored three different cohorts of student researchers, and shifted hosting providers more times than I could count,  I've come back around to the belief that what will last in the longest term is a book. So I'm very happy that MIT Press is going to produce a book out of the digital project so that we can have the digital project for as long as it lasts, and also have a print version in case it ceases functioning at some date in the future. The book is being designed by Silas Munro at Polymode, by the way, and we’re very excited about what it will look like. 

Utsavi: That's fantastic, congratulations! 

Lauren: Thank you. Yeah, it's been nice to feel like I could sort of put my broad knowledge of the digital to use, and to do so in a way that would lead to the team that I wanted to build, and the rewards for each member of the project team, and then a project and product that we all believe in—I think that doesn't just happen. You know, you need to build the project you want to be a part of, to adapt a phrase we have heard in other contexts.