Lynching in America (https://lynchinginamerica.eji.org/), a promotional website made by Google for the Equal Justice Initiative, a mass incarceration not-for-profit organization, contains an interactive map titled Racial Terror Lynchings (https://lynchinginamerica.eji.org/explore). Drawing on a dataset compiled by the EJI for their eponymous report of “4075 racial terror lynchings of African-Americans,” this interactive choropleth map of the United States purports to depict “Reported lynchings by county”, occurring between 1877 and 1950 ["Lynching in America"]. Overall, the map has a minimalist aesthetic, reminiscent of Google’s Material Design visual style (https://material.io/) that depicts topographical elevation as well as country, state, and county borders, but not cities, towns, roads, rivers, lakes, or landmarks (see Figure 1). The data visualized is also minimalist in presentation, and is tied directly to the state and county borders depicted, as is typical of choropleth maps.

The map is focused on the southeastern United States. On first page load in screen widths below 1500px, the map centers on the United States below the Mason-Dixon line, while in larger screens the entire contiguous United States is shown. However, in both cases, the American south grabs attention with many southern counties highlighted. Users can click on any county or state, and are taken to a zoomed-in view of the state, with the total reported lynchings in that state displayed in large letters. The user can then hover over individual counties to find out how many lynchings were reported in that county (see Figure 2). This zoomed-in view focuses on the number of lynchings per state and county — represented by polygons shaped according to county boundaries — creating a visual argument that encourages measurement of states and counties as more or less reprehensible in terms of number of lynchings. It is focused on lynchings in the context of political boundaries, and thereby presents a strong geopolitical argument about recorded lynchings that appears definitive and damning, particularly of the states with the most counties marked in bright red: Alabama, Florida, Louisiana, and Mississippi. There is a straightforwardness about this visual argument: racial terror was and is morally wrong, and its contours are plain to any competent observer.

In terms of color, the map features a dark, almost monochrome color palette, with a dark grey United States segmented by black state lines, placed in a dark blue sea. This mostly dark color scheme is dramatically interrupted with many counties highlighted in various shades of red, with bright red indicating 20 or more lynchings recorded in the county (see Figure 1). The color scheme of red on an otherwise dark, monochrome palette compounds the visual argument described above, as it references the brutality and violence inflicted upon African Americans in those locations, recalling the blood stains on United States history as it relates to racial violence and white supremacy.

The second example, Map of White Supremacy Mob Violence (http://www.monroeworktoday.org/explore/), is a far more complex visualization than the minimalist Racial Terror Lynchings. It is an interactive map within Monroe Work Today (http://www.monroeworktoday.org), a website dedicated to publicizing the research of sociologist Monroe Work, who systematically documented lynchings in the United States. Created by education-focused digital agency Auut Studio, this interactive map of the United States depicts lynching records in the context of historical racial violence and public discourses of white superiority, and is consequently subtitled “The lynchings and riots to enforce racial superiority in the US”. Upon first page load, Map of White Supremacy Mob Violence stands in stark contrast to Racial Terror Lynchings, presenting the entire contiguous United States center of screen, irrespective of screen size, and representing each recorded case of lynching as a single grey dot on a white national landmass background (see Figure 3). State borders are not visible in this view, and a single, bright, attention grabbing color is only found on instructions for users. This initial map view — including map framing, choice of colors, and the carefully worded title and subtitle — provides a strong and markedly different message than the Racial Terror Lynchings map. Collectively, these elements present a strong visual argument that discourses of white supremacy are a nationwide reality in the United States, one that has historically been enforced through lynchings and riots by mobs. The geopolitical framing of lynching as a nation-wide reality is balanced with the acknowledgement of a crucial aspect of the historically pervasive intellectual climate, one of white superiority, and an important aspect of the social behavioral climate, mob violence. At the center bottom of the page is a key annotation: a label that reads “Should I trust this? Find out.” If users click this message, they are taken to a plethora of information about the veracity of the data used, and a discussion of the importance of thinking critically about the visualized data.

Although it is not entirely clear from this initial page view what the grey dots represent, the user is presented with instructions to zoom in and click on individual points. After the user follows these instructions, the map gains more color, and individual dots become more visible, along with state and county boundaries (see Figure 4). In this view, state and county lines are visible, as are the boundaries of Native American reservations and areas that historically were Spanish colonies. However, these geographic lines are indicated with subtlety, using pale colors. In contrast, the dots representing each individual lynching are highlighted using bright colors and an onclick interaction effect. Marking geopolitical boundaries but deemphasizing them, combined with emphasizing each individual lynching with bright colors, makes the visual argument that lynchings occurred in the context of geopolitical boundaries, but that the individual deaths are of greater significance than the boundaries themselves.

Along the bottom of the screen, supplemental information and interactive features appear, including a timeline, a label stating the time span represented in the current view, and a color-coded legend of dot colors. Six different dot colors are used: five to represent races of lynched people, and one to represent “other”, where the race recorded in the historical records does not fall into one of the main five. Notably, the “other” category includes lynchings of white abolitionists, thereby demonstrating the complexity of the history of lynching in the United States, and that white people were also, however rarely, victims of lynching in the name of white supremacy.

Both the timeline and the legend serve two purposes: one functional, the other persuasive. Users can select a time period using the timeline, which then alters data presented on the map, so they can see how many lynchings occurred in an area over a specific timeframe. Secondly, the timeline provides a persuasive visual cue that racial superiority-motivated lynchings occured continually over a long timespan, with some time periods seeing evidence of significantly more racial superiority-motivated lynchings. The legend also works on these two levels. It firstly allows the user to identify the race of a lynched person based on the color of the dot used to represent them, and secondly provides a strong visual counterargument to the widespread public assumption that lynchings were perpetrated exclusively on African Americans. These two elements, the timeline and the legend, confront the user with the temporal and racial extent of white superiority-motivated lynchings, both qualities that are absent from the Racial Terror Lynchings map.

Most strikingly, the zoomed-in view of Map of White Supremacy Mob Violence contains a list, in the bottom left hand corner of the screen, of the name (where available) of every lynched person represented by a dot in the current map view, and the year they were lynched. Clicking on any individual dot brings up a callout box containing extra details of the lynching available in the historical record, including the county in which it occurred, the details of mob violence in which the lynching occurred, the accusation made before the lynching, and links to every available historical source that verifies the record (see Figure 5). Naming individuals who were lynched, and providing circumstances surrounding their death, focuses attention on the humanity of victims of lynching and on the social circumstances in which lynching was a viable possibility. In the present day context of many white supremacists denying their racism, it is worth noting that these historical records rarely mention race as a motivator for lynching. For example, William B. Willis of Richmond County, Georgia, was accused of murder before being lynched. The Map of White Supremacy Mob Violence does an exemplary job of demonstrating that racism was indeed the motivating factor in lynchings, and also that racism was largely hidden in the official historical record by documenting other, non-racial reasons. Providing links to multiple historical sources within the interactive map increases trust in the veracity of data, while at the same time giving users the opportunity to investigate the historical evidence themselves. Map of White Supremacy Mob Violence uses multiple compelling strategies to both humanize the data it represents, and to contextualize it in the societal racism and discourses of white supremacy in the United States.

The contrast between these two maps is even more striking when looking at the west coast of the United States. For example, Figure 6 shows a marked difference between the recorded lynchings in California in the Racial Terror Lynchings map and the Map of White Supremacy Mob Violence respectively. The difference in the representation can be accounted for by the fact that the former only depicts lynchings of African Americans, whereas the latter depicts lynchings of African Americans, Native Americans, Latinos, Italians, and other races. The view of the American West depicted in Map of White Supremacy Mob Violence in Figure 6b provides a compelling visual narrative that lynchings were common across California in the name of enforcing white superiority. The view of Racial Terror Lynchings Map depicted in Figure 6a, by its lack of clarity about exactly which data is being represented (i.e. historical records of lynchings of African Americans only), its use of a bold and expansive title that suggests comprehensive coverage of lynchings of all races (i.e. Racial Terror Lynchings), and its tonal emphasis on the south (California appearing grey, while visual attention is drawn to the large amounts of red in the bottom right hand corner of the map), makes a visual argument that California had few instances of racial terror or lynchings. This is particularly problematic because the Equal Justice Initiative’s stated goal is to challenge black incarceration, and a key part of their organizational message is that this goal is urgent and directly related to the history of lynchings motivated by white supremacy. Black incarceration rates in California are among the highest in the nation today: according to the U.S. Bureau of Justice, California imprisons blacks 8.8 times more frequently than whites, well above the national average of 5.5:1 ["The Sentencing Project"]. Consequently EJI’s visual argument in the Racial Terror Lynchings Maps inadvertently breaks down in advocacy regarding California prisons.

The emphases of these two maps are necessarily different because of the different purposes of the sites in which they are situated. Lynching in America is a promotional and advocacy tool for the Equal Justice Initiative, primarily created to visualize data within (and thereby promote) the report “Lynching in America”, which records lynchings of African Americans and frames lynching as a societal tool — enabled through mob violence and discourses of white superiority — to subjugate African Americans between slavery and mass incarceration. It is in the Equal Justice Initiative’s interests to visualize historical lynching data in a way that draws attention to geopolitical divides, so that clear links can be made between historical lynching events and present-day constituencies of sitting politicians as well as county and state local governments. The map provides compelling visual evidence for the organization’s present-day advocacy work regarding the inequitable mass incarceration of black Americans, the case of California notwithstanding. The problematic aspect of this is that the Equal Justice Initiative’s website, report, and Racial Terror Lynchings map are unfortunately named to suggest that they cover all historical records of lynchings in the United States. The Lynching in America report includes instructions for educators who wish to use it as a teaching resource. In this context, the geopolitical emphasis, use of color, use of summary data, and lack of links to sources give a concerning impression that African-American lynchings were the complete record of lynchings in the United States for the purposes of racial terror.

The majority of both scholarship and public attention regarding lynchings centers on the experience of African Americans in the Southern United States, and for good measure: for blacks in the Jim Crow South, lynchings represented a terrifying aspect of everyday life. In the grand scheme of racial violence, however, lynchings represented one small piece in a complex puzzle of individual, institutional, and structural racism. The conflation of lynching with the full extent of racialized violence in United States history obscures the historic depth and breadth of the oppression of people of color. A black individual was far more likely to suffer public humiliation, assault, rape, and murder than a public lynching. While lynchings do not represent the totality of racial violence in America, they come to the fore because they were highly symbolic affairs: gruesome spectacles of white supremacy, racial violence, and bodily mutilation meant to suppress and intimidate as much as they were meant to kill [Wood 2009, 1–4]. Due to their highly symbolic nature and the lasting implications of racist attitudes, policies, and actions for African Americans today, lynchings have become synonymous with racial hatred in the postbellum American south.

However, mob murder historically extended well beyond Dixie, representing a form of prejudicial frontier “justice” in the Midwest, West, and Southwest against minorities and immigrants of various backgrounds [Pfeifer 2006] [Pfeifer 2013]. Outside of the latest scholarship, such victims, whom Carrigan and Webb describe as the “forgotten dead,” are largely overshadowed or overlooked in the public sphere, missing an opportunity to explore the structural, cross-regional, and transethnic dimensions of American lynching. In fact, lynch mobs murdered hundreds of Mexicans between 1848 and 1928 in the American Southwest [Carrigan and Webb 2017].

Moving beyond EJI’s limited focus on African-American populations in the American South, Auut Studio acknowledges the historic violence committed against Native American populations, which is noticeably absent from the vast majority of lynching data sets. This general oversight plays into present-day blind spots regarding violence against native populations, who, despite suffering more state violence and community disruption than any other minority group in 2016 according to data collected by the Centers for Disease Control and Prevention, rarely garner the public spotlight ["The Counted"]. To counteract the public oversight of this “forgotten minority,” Auut Studio included the boundaries of Native American reservations as “sovereignt[ies] deserving of equal visual treatment on the map”  [Ramey 2017]. Similarly, Auut Studio included lynchings of the Chinese along the frontier, namely in California, as well as Mexicans in the Southwest, keeping in line with recent historical scholarship.

In order to ethically represent historical subjects, data visualization techniques, particularly those geared toward public consumption, must remain abreast of the insights made in two areas of the scholarly literature: debates on the subjects they depict, and debates on design and representational considerations in the ethical visualization literature. Present-day academic debates on lynching challenge the widely accepted notion that lynching was exclusively a Southern phenomenon, excusing those regions of the United States outside of the South of their own racist heritage. The small but growing ethical visualization literature emphasizes the need for acknowledging and mitigating the potential for harm inherent in visualizing data, particularly when it comes to selection of design elements, visual style, and selections of data to annotate and visually emphasize [Cairo 2014] [Hepworth 2016] [Kostelnick 2016] [Skau, Harrison, and Kosara 2015].

Collectively, we tend to visualize old arguments, as visualization practices have not kept pace with dominant arguments in the digital humanities literature about the importance of critical practices. Visualization practice in the digital humanities runs the risk of following a functionalist methodological approach that assumes visualization to be an impartial medium. This illusory functionalism has led others to charge that data collection, processing, and visualization practices constitute mere “janitorial work” in the service of “real” humanities scholarship, ignoring the important decisions made during such processes that critically shape historical narratives. Construing digital humanities practice as a “support field” has led to further accusations that the digital humanities simply show to us what we already know, rather than challenging us to think critically about historical topics in new, interesting, and socially responsible ways [Allington et al. 2016].

Following Alan Liu’s charge that digital humanists have ignored cultural criticism, which in turn has blocked “the digital humanities from becoming a full partner of the humanities,” data visualization practitioners need to critically engage with the ways in which digital tools can “communicate humanity” rather than relegating it to the margins, or worse, obscuring the human stories essential to understanding structural racism today [Liu 2012].

Visualizing data that exclusively focuses on the African-American experience in the Southern United States provides an important argument about the nature of Jim Crow racism. However, purporting such data to be an inclusive representation constitutes a harm in the sense that it perpetuates common narratives of racial violence as a southern exception to an otherwise inclusive nation. After publishing their interactive map, the Equal Justice Initiative itself recognized this oversight, acknowledging the 300 lynchings of African Americans outside of the American south, though notably leaving aside other ethnicities like Native Americans, Mexicans, and the Chinese that fell victim to much of Western and Southwestern mob violence ["EJI Releases New Data" 2017]. This overlooks the depth of structural racism and its support of white supremacy, thereby denying the experience of millions of present-day Americans. Historical information has the capacity to legitimize or delegitimize present-day experience, and visualizations of historical data are a particularly compelling and resonant medium through which such information can either harm or help.

In the pre-data collection phase, the first step involves clearly defining the subject area that the data visualization will depict, while the second step involves reviewing the latest secondary literature on the topic at hand. Reviewing subject area literature would inform the remaining steps in the ethical data visualization workflow, inviting the researcher to compensate for the data set’s shortcomings by seeking out and including new information, or to limit the scope of the visual argument to be produced with said data. Doing so would avoid the glaring oversights and interpretive overreach that plagued the EJI’s Racial Terror Lynchings Map.

The second phase, data collection and curation, is perhaps most crucial in producing an ethical data visualization, precisely because it is so frequently overlooked. The third step in the ethical visualization workflow involves collecting primary documents, artifacts, and datasets, as well as secondary datasets of potential relevance, while the fourth step involves checking the appropriateness, authenticity, veracity, and feasibility of use of collected primary and secondary materials, and pruning those that don’t hold up under scrutiny. Once these two critical data collection steps have been finished, the researcher completes the fifth step, describing, by creating their own dataset that combines the collected materials.

Creating a custom dataset for the researcher’s visualization in this way eliminates other people’s and institutions’ biases from the data, ensuring erroneous arguments are not unintentionally added through using unaltered historical datasets. This process of collecting, pruning, and describing data sets was undertaken by Auut Studios for ten years before visualization, contributing to the particularly considerate treatment of ethical factors in the Map of White Supremacy Mob Violence. Similarly, EJI created an extensive dataset of over 4,000 public lynchings based on work done by Tuskegee University and the research of E.M. Beck and Stewart E. Tolnay that compellingly shows the long legacy of terroristic violence in the American south. Nevertheless, decisions made about what counts or does not count as “racial terror violence” made during the data collection phase — namely to exclude the American frontier — ultimately shaped EJI’s visual argument, making it overreach in its claims and thereby creating a narrative around racial violence that excludes other minorities and other geographic locales.

The third and final phase of the visualization workflow, data visualization and argumentation, is the main one associated with the ethics of data visualization, and this phase involves four steps: surveying, pre-visualizing, visualizing, and publishing.