Creating Visualizations that Situate Data and Amplify Activism
These @DefendBoyleHts line graphs and charts demonstrate how frequency in tweets visualized in graphs can change and correspond to very different kinds of encounters: from ad-hoc anti-gentrification media remix redefining gentrification in association with popular culture and white appropriation to grassroots activism using Twitter and other online digital platforms to unify publics, broadcast nondominant voices and actions, and coordinate resistance across both local and global expanses and related to intersectional issues of oppression. The design and appearance of these Tableau visualizations also generate a disembodied and quantitatively valued perspective that isn't adequate to visualize the nuance and impact of anti-gentrification encounters, especially those involving grassroots activism like those associated with @DefendBoyleHeights.
To better ground data in contexts, situated geographies, and the consequences of anti-gentrification resistance on the built environment and lived experience, I endeavored to visualize gentrification's encroachment into Boyle Heights through the perspective of @DefendBoyleHts tweets. To do so, I used grounded theory to create a coding structure. John Creswell (2014) described grounded theory as a "design of inquiry" whereby "a researcher derives a general, abstract theory of a process, action, or interaction grounded in the views of participants" (p. 14). In the case of this study, I understood DBH's perspective to be limited to the expression of their tweets. The process I followed used "multiple stages of data collection and the refinement and interrelationship of categories of information" (p. 14). For me, this process was informed through extensively researching Boyle Heights; developing familiarity with the geography, places, and figures currently central to Boyle Heights' gentrification and its resistance; and engaging in recursive coding processes, which included reading and rereading all of @DefendBoyleHts tweets, including every piece of attached media, linked URL, and place tag. First, I coded tweets for resistance actions, initially identifying over twenty actions that already condensed actions into more general terms; however, to visualize this resistance more clearly, I reduced the action codes down to three overall types: communication practices, direct actions, and legal actions. I retained the more granular codes, categories like celebration, mobilizing, surveilling, fundraising, art intervention, for instance, so that they’d appear in the visualization, but wouldn’t be one of the interactive features that control the points on the map. Adding these categories and codes to my data structure, I then coded for the targets of these actions and, again, went through the process of condensing over twenty different codes down six broader categories: art, business, community building, entertainment, government forces, and housing. I then coded each tweet, looking both at the visual media and tweet text.
Partly due to the time and labor required to manually read and code each tweet and the needs of visualization conventions to keep it simple, clear, and easy to use, I chose to apply only one primary action and theme code to each tweet. Although many tweets contain explicit or implicit language and media that might tie to another code, at least in this first iterative phase, I coded for what I determined was the primary subject/target of the tweet, the primary action (sometimes resulting in two codes when both seemed equally emphasized), and I identified any location information in the tweet or media. Although the majority of the tweets in the large dataset weren't geolocated, many of DBH's tweets included latitude and longitude coordinates, although usually to the general neighborhood of Boyle Heights, Los Angeles. To better visualize gentrification's movement into Boyle Heights and the resulting anti-gentrification encounters, I coded all @DefendBoyleHts tweets and attached media for the most specific location information that I could find in each tweet/metadata. Although some visual analytics exist, like Google Vision API, few are available for free, and even fewer will analyze a large dataset for free. After experimenting with a few options like Google Vision API, I found that applications couldn't perform location coding on tweets and images with the accuracy of human reading. For instance, a tweet might mention "bbq la," without indicating that bbq la is the name of an art gallery or providing any recognizable street address or entity. Additionally, location information was often only present in photos or videos, displaying or referencing building names or street signs. This coding process probably required more than 100 human hours, and it was certainly influenced by my perspective on the data; however, using grounded theory and human coding provide a way to depart from the data-structuring choices made by technologies and to structure data towards visualizing and amplifying activism.
Using Carto, I mapped all of the locations associated with @DefendBoyleHts tweets, which I'd manually coded for in my close reading process. The resulting map (Figure 11) visualizes @DefendBoyleHts gentrification encounters on a global scale as well as the time-lapse of this activity. In another Carto map (Figure 12, below), I mapped @DefendBoyleHts tweets with local locations to visualize resistance actions and gentrifying forces moving into Boyle Heights, Los Angeles. The resulting visualization orients the gaze towards agents of gentrification in Boyle Heights and amplifies activist resistance efforts, identifying direct actions, communication practices, and legal actions taken against these agents (Figure 13, below).
Interacting with these maps shows just how close Boyle Heights, with an average household income of $33,235, is to greater Los Angeles (Chiland, 2018), an area with some of the highest real estate prices and income levels in the country. The blue dots between the Los Angeles River and Highway 101 in Boyle Heights represent the art galleries DBH identifies and protests, including PSSST, the gallery in the previously discussed video. Although the dots aren't adequate to represent the embodied and digital encounters related to resisting these agents of gentrification, each dot surveils the gentrification agents, rather than activists. By choosing to display Google Street View images of locations, rather than the text and media of @DefendBoyleHts, the map orients the visual perspective towards surveilling gentrification agents in the neighborhood and amplifies resistance by retaining the action code, such as protest, march, or art intervention. Further interacting with this visualization, users can explore gentrification agents by type, such as art, business, and government forces. Layers can be toggled on and off to provide an understanding of where certain agents occur in Boyle Heights and what actions have been taken to resist these agents.
These custom maps are in the early inventive phases, and I although I have plans for adding additional visual elements like symbology and more visual texture, I haven't yet found a way to completely depart from the omniscient, disembodied perspective maps afford; however, by coding and mapping from the perspective of DBH's tweets, the view of Boyle Heights is one that surveils incoming gentrifying forces and illuminates activist tactics used to oppose these forces. In future iterations of the maps, I also want to amplify stories of activism by visualizing the consequences of anti-gentrification activism, like showing the galleries and businesses that have closed, the rent strikes that resulted in tenant victories, and the retention of cultural spaces and art. Even in these early phases, the data collection and visualizations provide a record of anti-gentrification resistance and nondominant perspectives vying to retain the Latinx character of the neighborhood and protect existing residents from displacement. From early victories like the closing of PSSST to the recent closing of 365 Mission gallery, future iterations of this map will further attempt to trace the consequentiality of DBH as a vitality that stretches beyond the moment of its tweets to what Laurie Gries (2015) termed consequentiality, or the "consequences [that] emerge before, during, and after a thing's initial physical production and delivery" (p. 87). By mapping from DBH's tweet perspective, the map surveils urban change from the perspective of existing residents and challenges the racial shorthand often present in dominant narratives circulated through mainstream media, urban planning documents, and tourism strategies created by city elites.
Organizing tweets by the distinct categories required by spreadsheets and most visualizations requires delineation that elides the ways in which categories of themes and actions bleed together. Everything I've coded could be labeled a communication practice because each one corresponds to a tweet. Similarly, some art interventions lean more towards communication practices, like handmade zines and eviction maps, while others like graffiti and splashed paint outside galleries might be considered more akin to direct actions. Future iterations of this map will aim to better represent this nuance. In addition to the struggle over distinct categorization, some actions and practices are combined under broader umbrella terms. As I noted previously, this broader categorization achieves a greater level of usability and clarity, but it loses meaningful distinction. For instance, "government forces" includes those posts that reference policing, Immigration and Custom Enforcement (ICE), and transportation, as well as white supremacy, imperialism, and patriarchy. To limit the number of categories and the visual attention required to understand the map, I combined these terms under the category "government forces," because each subcategory relates in some way to state power structures. This is just a brief discussion of the many ways in which I chose to structure the data within spreadsheet structures that visualizations rely upon. The resulting visualizations are far from ideal representations of this data, in no small part because they reduces visually rich, complex interactions to points whose symbols only communicate a fraction of the depth and nuance of the rhetorical encounter and doubtlessly even less so of the embodied moment. Despite these drawbacks, these DBH maps come closer to situating data in embodied, geographical situations than the word frequency and data frequency charts of the entire "gentrification" dataset.