This week we picked up on a thread that we first touched on in our readings on mapping and spatial narrative-building; in Béatrice Joyeux-Prunel’s article, she states, “[…] when you publish a map, be careful about what it might seem to say.” The operative concept here is “seem,” because as we have seen, the ways in which we present data heavily influence our perceptions of it. Data visualization at its core is a narrative expression of information with all its subjectivity and ambiguity, but the word “data” offers a connotation of emotionless truth that simply isn’t present. Many of us in the class seemed affected by the emotionally charged visualization of gun violence created by Periscopic (as shown in chapter 3 of Data Feminism by Catherine D’Ignazio and Lauren F. Klein), especially when compared directly with the Washington Post bar graph depicting the number of active shooter incidents occurring between 2000 and 2015. Even though they were created for completely different reasons, they ostensibly depict the same phenomena: gun violence in America. And yet, one offers a visceral reaction while the other only offers a vague impression.
Of course, I feel (or at least hope) that most people understand on some level that data can be represented differently in order to make it fit a narrative. However, it seems rare to me that the deliberate evocation of emotion is not only encouraged, but is the main purpose for the gathering of the data in the first place. This is of course not to say that I disagree with how the information is presented! On the contrary, I think that human impact is something we should be able to focus more on in a world deluged by data. In the example of gun violence, the scale of suffering we are experiencing is terrifying, and it grows every year. It is so easy to let the numbers be numbers rather than let them represent real, human lives, because comprehending suffering on that scale is incredibly difficult! But rebuilding that bridge is central to bringing an ethic of care to how we represent the world and our impact on it. As D’Ignazio and Klein argue, the traditional model of data visualization in which the emphasis is on “just the facts,” writes off ways of communicating information in a way that genuinely resonates with people. Data feminism as a philosophy seeks to challenge this false binary of truth and emotion, offering new ways of looking at and understanding how we represent and communicate information truthfully. Ultimately, it argues that “clarity without persuasion” (as cited by a workshop at the Guardian) is impossible.
I was particularly struck by the example D’Ignazio and Klein gave for the concept of “data visceralization” with the performance A Sort of Joy. There is a lot you can do visually to represent the gap between male and female artists in the MoMA, but it is easy to look once and not really comprehend what it means. By stretching that data spatially, aurally, and temporally, viewers are forced to confront the disparity in a highly uncomfortable way. I had never even considered a data visualization that is not purely visual and two dimensional, but by complicating the “look at a graph to understand the data” experience by removing looking and graphs, it forces a visceral presence in that moment that allows for emotion to surface. It is a highly persuasive representation, but at the same time, there is no untruth to it. Its point is clear and simple without supposition as to the “why” of the phenomenon; that is all on the viewer to determine.
In the vein of outside-the-box representations of data, we also took a look at AI images as a form of data visualization. I won’t lie, as both a digital artist and an information professional I do not have a high opinion of AI in general. Because of the impact on artists and the environment I think you would need to make an absolutely unequivocally solid case for the use of AI in general, and I just have never seen anything that to me justified its use. I’m happy to explore use cases and not write it off completely in case that unicorn use case comes along, but I truly think there is no good purpose for AI-generated images specifically. As Victoria Turk showed in her article How AI Reduces the World to Stereotypes, an image that is the average of everything on the internet is just going to give you the average of images created by the types of people who can access the internet. Although those numbers grow every day, the internet is largely still stuck in a white, Eurocentric, patriarchal world. Describing why this is a problem is preaching to the choir here, but the perpetuating of this singular worldview is an unfortunate and inextricable effect of the technology.
I also took a particular umbrage with Kaylee P. Alexander and Emily Lawhead’s article Data And Discontent: A Converging History Of Photography and AI Art and its conflations with the history of emerging art technologies. While certain technologies may look like magic and there have been artists who use technology to advance their work, they still require an eye for design and executive decision-making by the artist. But, I digress. AI generated images, which in a technical sense fall under data visualizations, at most evoke a curiosity in the current digital social landscape (as in Salvaggio’s Reading AI Art) and at worst are unethically produced visualizations that can cause real harm. I don’t think that ethics are worth setting aside for societal insights.
Keeping everything we learned about what makes a good visualization in mind, we were tasked with creating our own simple visualizations. I will admit that I struggled a bit with getting Palladio to work with the Tate dataset we input into it. I think this may be because I use Firefox rather than Chrome (to ideally disentangle myself from Google a little bit), but it crashed the site for me multiple times. I would love to explore it further with a different dataset at some point, but for the ease of sharing my results, I took a look at WordClouds instead. I input a portion of the first chapter from Skin Shows: Gothic Horror and the Technology of Monsters by Judith Halberstam since I’ve been reading it this month.

It definitely picked out a lot of themes of the chapter (and the book overall), and I suppose it’s kind of a fun graphic, but I’m honestly not quite sure what I would do with this. I think you could probably glean as much information from the description on the back of the book. Then again, I’ve never been entirely sure what effect word clouds are supposed to have- we made plenty of them in high school, and I never felt like picking out single words did more than fragment my understanding of a text. Maybe it’s just me and my experience though, so I’ll be curious to see if anyone else feels differently and had some breakthroughs for their texts.



︎