It’s rarely just a dichotomy.
An exploration of intersectionality in design
Disclaimer: I own the voice of someone new to the field of design, new to an awareness of intersectionality, and overall not particularly subjected to societal marginalization. However, as someone who aspires to design for large audiences, I acknowledge the value in exploring this topic, and welcome more informed voices to the conversation.
Quarto is one of my favorite games. It’s incredibly simple in form, yet surprisingly complex in thought.
It’s played with sixteen unique wooden pieces that are distinguished from one another by four binary characteristics: height, color, shape, and concavity. Your opponent chooses the pieces you place, and the object of the game is to align four pieces by at least one of their common attributes. More often than not, a player loses by overlooking one of the pieces’ more diminutive characteristics aligned in an unexpected place.
It can be challenging to consider a piece’s four characteristics at once, even when each of the characteristics has only two possible variations. This is especially true when they’re set against similar pieces with more visible attributes—when there’s one clear observable grouping, the others fall to the background, and we fail to judge each piece holistically.
Identifying such oversights—but in the judgment of people—with a mission of overcoming them, in essence, is the purpose of intersectionality. Kimberlé Crenshaw coined the term in 1989, and she explains its origins below:
While the term intersectionality was initially conceived to highlight the overlapping systems that oppressed black women, as Crenshaw explains, it can encompass any number of identities (including class, age, sexuality and physical ability) that, when co-existing, result in compounded marginalization.
So, how is design doing with combatting intersectionality?
If the current student body in the CMU School of Design is any indication, we have surely made modest strides in changing the face of the “average” designer—the next generation of designers are sure to be more diverse in background and experience than the previous. Nonetheless, we can do better: design higher education is still inherently classist, only accessible to those with some amount of financial confidence.
Even if issues of designer diversity were addressed, the foundational knowledge upon which we are expected to build our professional practice seems to still continually force us to lean on traditional, reductive methods of analysis, doing both us and the audiences for our work a disservice. We speak in generic terms of “the user” without much consideration for individual motivations. To combat this, we are sometimes encouraged to develop “personas,” which are essentially made up characters who are meant to represent “everyone.” We wield data perceived as objective, with little to no understanding of the biases under which it was collected.
In order to address issues of oppression with design, it is not enough to have diverse, mindful designers — it’s also necessary to develop new tools with which designers work.
As Margaret P. of Microsoft Design highlights in a Medium article from November, designing for the average “user” is inherently problematic. She cites a study by the US Air Force in 1950: hoping to address issues with cockpits designed for the average pilot in 1926, they measured 4063 pilots’ bodies in 140 ways, and calculated averages to wield when updating cockpit design.
Zero pilots measured fit into all 140 average measurements taken. Even when abbreviated to the ten most essential measures, zero pilots fit. If choosing three of those ten, a modest 3.5% could be considered “average overall.”
Acknowledging the lack of an average led to many more adjustable parameters in a cockpit, which led to a far more successful Air Force.
“Why was the military willing to make such a radical change so quickly? Because changing the system was not an intellectual exercise — it was a practical solution to an urgent problem.“ —Todd Rose
This all occurred during the formative era of human factors, a field sparked by a need to understand the physical, cognitive, and emotional circumstances that influence how people interact with objects. Human factors significantly shaped how mainstream products are designed: rather than designing for a narrow average, they are typically designed for a broad range of measurements, varying between 5% and 95% of “average.” When a product can’t serve such a range in static form, it is made adjustable.
In an ideal world, we’d be able to design to adjust for the other complex measures of personal identity. Instead, we frequently design for majority populations and predictable use cases.
“When you say ‘Edge Case,’ you’re really just defining the limits of what you care about.” —Eric Meyer
Some noteworthy examples in digital design have been collected by Miriam E. Sweeney—her chapter in The Intersectional Internet focuses primarily on conversational user interfaces and their AVAs (anthropomorphized virtual agents), which most frequently present as female, often in subordinate roles and using language that perpetuates female stereotypes. She also notes that users are most comfortable when voice user interfaces present as their own race. A more subversive example exists in the presentation of a digital workspace as a “desktop,” a representation of a white-collar, middle class workspace. Despite these representations likely not being designed to oppress, it’s difficult to deny that they inherently leave swaths of people unrepresented, or misrepresented.
In their 2017 CHI paper, Schlesinger, Edwards, and Grinter explore how distinct and overlapping notions of identity have been addressed historically in HCI. Although they examined a corpus limited in scope to CHI submissions, it seems to point to overall trends: Gender (specifically generic female identity) is addressed most frequently, with class as a far second, and ethnicity and race trailing significantly — but these factors are usually addressed independent of one another.
Of the 14,000 CHI submissions they examined, only 10% address the above factors of identity at all.
They go on to consider how intersectionality might be applied more broadly, as it “supports an effort to situate the relationship between technology and social systems.” They highlight three intersectional methodologies to consider incorporating into HCI research: Anticategorical complexity (Don’t reduce people into unnecessary categories when they oversimplify), intercategorical complexity (Leverage categories “provisionally” when they serve to make necessary comparisons across groups), and intracategorical complexity (Leverage categories when they serve to identify anomalous patterns within a larger group).
Lastly, they make suggestions for generally being mindful of identity in research, all of which focus on bringing identity information of researchers, research participants, and users forward whenever possible, and to overlap facets of identity in data analysis whenever possible.
Outside of the context of academic research, I also translate this to mean that designers should be explicit about who we are, with whom we conduct research, and for whom we intend to design. This also means we should be explicit about considerations we are not making and why. It can be uncomfortable, but acknowledging who our designs don’t serve might help point to the fact that they should.
But, what can be done about those tools we rely on to do our work?
Kill the “user.” Schlesinger et al. point to the previous work of Dourish and Mainwaring that calls for the abolishment of the term “user” because of its reductive nature. Not having read their work, I don’t know if they offer substitutes, but I am in full support of this. Perhaps we might consider identifying interactors by their essential role in the interaction: readers, listeners, correspondents?
Kill the “persona.” As for “personas,” Microsoft Design has recently offered an alternative in the form of “persona spectrums,” meant to highlight how context can shape motivation, and that identities can be dynamic.
Decolonize data. In order to subvert oppressive design practices, it is of course necessary to seek and uplift otherwise oppressed voices in research. In his work developing a tinnitus-tracking app, Bryce Peake exclusively tested with historically marginalized people, and designed the app in such a way that gave them accessible control to their data to prevent its future misuse.
This list is too short.
In any case, the point of this exploration was not to immediately generate solutions to address intersectionality in design, but to challenge designers to reject reductive practices and experiment with ways they might be expansive instead. Schlesinger et al. point out that this is not easily done, thanks to industry pressures to produce as much and as quickly as possible:
Previous work finds that the culture of “publish or perish” in the U.S. has increased the pressure to report statistically significant results. We worry that reporting statistically significant results based on gender differences alone is
pursuing easily attainable results at the expense of a more reasoned analysis. When using identity categories for quantitative analysis, we must ask what our motivations and reasons are for stratifying our participants in particular
ways. Why gender, rather than race, class, or a combination of the three?Incorporating intersectionality in statistical analysis by investigating more identity variables at a time may mean it is more difficult to get significant results.
However, this also means we are developing fine-grain, well-reasoned statistical inferences about our surrounding contexts. While statistically significant results will be harder to come by, we should embrace this challenge as it allows us to interpret the world in more nuanced and accurate ways.
This is a challenge I raise to myself as well. My hope is that, in the coming semester, I can spend time exploring ways to make digital interfaces more culturally inclusive and less reductive, and identify more of the factors that limit the ability to do this.
I also hope to play Quarto more.
Sources Cited
- Rinku Sen (2017), How To Do Intersectionality.
- Kimberlé Crenshaw (1989), Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics.
- Kimberlé Crenshaw (2016), The Urgency of Intersectionality.
- Margaret P (2018), Kill Your Personas.
- Todd Rose (2016), When US Air Force Discovered the Flaw of Averages.
- Margaret Gould Stewart (2018), Able, Allowed, Should; Navigating Modern Tech Ethics.
- Mimi Onuoha (2016), The Point of Collection.
- Ari Schlesinger, W. Keith Edwards, Rebecca E. Grinter (2017), Intersectional HCI: Engaging Identity through Gender, Race, and Class.
- Miriam E. Sweeney (2016), The Intersectional Interface. From The Intersectional Internet: Race, Sex, Class, and Culture Online (edited by Safiya Umoja Noble and Brendesha M. Tynes).
- Bryce Peake (2015). Decolonizing Design Anthropology with Tinn.