Dimension Graphique: Ethnicities

Du Bois Book Cover. W.E.B. DuBois’s Data Portraits Visualizing Black America: The Color Line At The Turn Of The Twentieth Century

Du Bois Plate 11. City And Rural Population. 1890.

Du Bois Plate 23. Negro Property In Two Cities Of Georgia.

Several colors were picked from pages in Du Bois’s Data Portraits, and used to create a custom color palette of red, blue, yellow, green, and brown.

Custom Palette

These colors were applied as follows to the ethnicities values outlined in the schema, under Ethnicity Tags.

Custom Palette Annotated

Care was taken to subvert stereotypes and challenge easy assumptions. An early design decision: don’t use the color white to represent whiteness, and don’t use the color black to represent blackness. White and black are stripped of all but editorial or descriptive purposes, such as: numbers, titles, text that instructs. In the above diagram, the specific ethnicities tag “asian” uses a bright blue color in generated graphics, but the label “asian” uses the color black.


Early experiments at representing individuals used color for ethnicities and grid shapes for human species, repurposing the spaces and grid deconstructions from Plate 23 above, and triangle or other randomly-generated shapes for different species (z-tag) for animals or ai.

These experiments led to the initial grid system, where each space in a 3×3 grid was used for a different identity dimension.

For example, the 3×3 grid might be filled in as follows:

Grid Key Diagram V

The next step was to apply this thinking to a particular media object and see what forms are generated. The Netflix series Altered Carbon was picked as a data set, and it was used to examine and evolve design rules for applying shapes and colors to individual glyphs. One of the first examples of using ethnicity and rule-based color to form shapes is below.


Using the same characters from the Netflix series Altered Carbon as above, and applying the generative design rules, the following are generated. This iteration uses simpler design rules than above for sexualities, and a later model of the custom gender glyphs.



At this point, it might be useful to step back, and look past individuals to a group of individuals. Given these glyphs for individuals, how would the whole cast of Alterered Carbon be depicted? Here’s a first thought at aggregating the individual shapes:


Now, looking past groups of individuals: how will this scale to casts of 20 media objects, or aggregates that are the sum of 10 shows on Netflix? We were stumped, and took inspiration from circular visualization techniques by Du Bois and Yayoi Kusama: unravel the linear grids to use circular and radial forms.

Revisiting Du Bois Plate 11. City And Rural Population. 1890.

This is an example of a radial graph form, much like a clock where time is navigated clockwise starting at noon. Instead of hours, this represents the values 1 to 8, where each value is sized proportionately, so that the change in value is apparent.


Using a radial form to compress data (and compressing further with video) leads into forms like: