Visualizing for the rhizome: How do we design data visualizations that embrace bodies?
A critical examination of data visualization practices through the design of reflective community conversations.
The problem
A good data visualization opens up space for good conversations. Alberto Cairo asserts this much in a recent talk about his new book Why Charts Lie. In this sense, a data visualization is a starting point in a dialogue about a shared understanding of the world around us, ways for us to know together. Then, in creating models of data and assembling them into a visual form, we create (or reinforce) ontologies; ways of knowing about the world.
If the act of data visualization is so tied up in how we know about the world, or ontologies, we must reflect on how we come to know. This means data visualization designers should examine how we create ways of understanding as part of our practice.
There is an emerging sense among scholars of data visualization that our practices create rigid, harmful understandings of the world. Digital humanities scholars have addressed this (for example, Johanna Drucker’s work on capta). Further, a wave of artists, practitioners, and scholars have been building on this thread, dissecting the implications of practicing data science and visualization in a disembodied way: Catherine D’Ignazio and Lauren Klien’s Data Feminism project, Jenny Odell’s writing on data and Designing for the In-Between, Safiya Umoja Noble’s work on Algorithms of Oppression, and Mimi Onouha’s work on data collection and ethics among others.D’Ignazio and Klein argue in their book, Data Feminism, that current practices of data visualization obscure bodies, leading to the silencing, extraction, monitization, or invisibility of people (or bodies) that the data represent. As put by Jenny Odell in a 2018 talk at KIKK for “Designing for the In-Between”:
“A one size fits all system means we’re only going to see one size, both in ourself and in others. There’s so much that stands to be rendered invisible by any system of knowledge or identity simply because it doesn’t fall squarely into one bucket or the other.”
What happens to the practice of data visualization (and our conversations) when we consider alternative ontologies (such as feminist) ways of knowing?
How can we make data visualizations through embracing an ontology that attempts to de-obscure the body? Through creating a data visualization tool to be used within these reflective, in person, community conversations, I hoped to find out ways in which I can practice data visualization that embraces situated ways of knowing.
The theory and design principles
Before embarking on my project, I derived data visualization principles from theories that embraced the situated nature of knowledge.
First, I aimed to define what I mean by data in terms of the body. Object Oriented Ontology (OOO) provides a way of thinking about how to define data as situated within bodies. Here, I define data as the phenomena a body observes. I’ve borrowed this definition of data from Timothy Morton’s characterization of OOO, the idea that no object is fully knowable. For example, I can only know about a cup by drinking from it, looking at it; the phenomena my body permits me to observe. However, I will never know about the cup in its entirety, the gesamtkunst-cup.
In my view, OOO provides a way of knowing that both situates knowing in bodies, and doesn’t privilege any single body or way of knowing about the world; it’s plural.
Principle: Data is situated in bodies. Embrace a plurality of situated, embodied perspectives as important and key ways of knowing
Principle: Appreciate other bodies and ways of knowing. Thus, employ “ecological” metaphors to embrace other ways of knowing.
Now that we have a way of defining data, how do we approach bringing data together into understandings of the world, or an ontology? Deluze and Guattari, in A Thousand Plateaus, discuss their concept of rhizome, which provides a way to frame knowledge and data as a heterogeneous, non-hierarchical, non-linear process of assemblages. A rhizomatic approach to knowledge does not afford “universals.”
In their world, one could view a data visualization as a tracing, in that it is a model of information that is encoded visually. Guattari and Deluze characterize tracings, or structured representations like these, as dangerous: “it has generated, structuralized the rhizome, and when it thinks it is reproducing something else it is in fact only reproducing itself.” Data visualizations are structured. To encode information visually, a structure (despite possible flexibility) is required. The structure being, in part, a model of variables, or data types, that link data to visual encodings. Thus, to avoid the danger of the structure they portray, visualizations must be situated within a flexible model of information.
Simultaneously, if enough nuance is imbued within a visualization, it can serve as a pathway into a particular view into the world. Especially if the data model used contextualizes the visualization. This means multiple data visualizations, using different models of knowing, can sit together as different doorways into plural views of the world.
Principle: Data visualizations as temporary crystalizations of what we think we know. Frame a visualization as a pathway into a way of viewing the world
Principle: Show interrelatedness of data through visual ‘echoes.’ Becoming; visual indicators of interconnectivity
Given the situated nature of knowledge, what happens to the practice of designing data visualizations? In their book Data Feminism, scholars Lauren Klein and Catherine D’Ignazio define principles (among more) to consider in designing data visualizations that might mitigate the obscuring of bodies:
Principle: Consider context. Who and what counts as knowledge producers / knowledge? What are the power relations at play?
Principle: Embrace emotion as an important and key way of knowing.
With these theories in mind, I have a set of principles to start designing from.
The project
I am designing for a context that is close to my heart: critical community conversations. Designing for this context requires unique consideration for emotion, nuance, and embodied knowledge.
I’ve chosen this context because it specifically requires care for context and emotion. In my mind, if we don’t know how to design for critical community conversations (driven by bodies and emotion, place-based and community), then we can’t begin to enact a way of practicing data visualization that doesn’t obscure bodies (or the plural, situated nature of knowing).
Further, I’ve chosen to focus on a community I’m embedded in; the cohort of graduate design students at Carnegie Mellon University. We share a studio, where many of us spend the majority of our waking hours collaborating in the shared space, eating in the kitchen, or working at our desks.
In particular, I aimed to design a visualization tool for my community to use while we have weekly, in-person conversations with one another. These conversations focus on reflecting upon and sharing our experiences as graduate design students. I’ve been holding space for these meetings for the last year, reading and learning through how to have life-affirming conversations with my fellow students.
These meetings have become dear to my heart. And, most importantly, I’ve chosen this context because it would benefit from a prosthetic that would help us have good conversations, to know about ourselves together in a healing way, and to remember these special moments. I wondered if a data visualization could serve that purpose.
As I worked through deciding what data to visualize, and how, I responded to each of the principles in a certain way. As I created this visualization, several practices emerged.
Consider context. Who and what counts?
I sought guidance from community and leadership scholars such as Margaret Wheatley, Peter Block, and adrienne maree brown. I asked myself: how should a community know, feel, and talk about itself? I defined the possibility of a caring conversation is what counts; the type that is life-affirming for this group of people. Designing in a visual mode of care would be key, which means only data that fits this paradigm of care exists.
Practice: Articulate who and what count as knowledge holders and producers, and why that matters.
Because I decided to design in a visual mode of care, that is the reality I’m imposing. As a student who is embedded, my view is perhaps helpful. Yet, this is still an exertion of power.
Practice: Think about what mode of reality you are imposing on your audience, and why you are justified in exerting this power.
Embrace emotion and situated perspectives as an important and key ways of knowing.
According to Wheatley, and brown, the act of reflecting and listening to people tell their story of their experience in the community is the next critical aspect of conversations. These reflections, whether they take the form of text or otherwise, would be crucial.
Practice: Participants (if possible) should decide how their experiences are measured.
Appreciate other bodies and ways of knowing. Thus, employ “ecological” metaphors to embrace other ways of knowing.
This principle centers important beings of the landscape as worthy of knowing. Critical old growth Eastern Hemlock (a kind of conifer) forests exist in Pennsylvania, where my community lives. What can we learn about Eastern Hemlock Trees? Peter Wohleben, a forester and author, asserts that, in old growth forests, trees only survive through generously sharing nutrients through connected roots. I decided to focus on the roots of old growth Eastern Hemlocks as a visual metaphor.
Practice: Look at place you are designing your visualization in / for in terms of the bioregion. What can we learn from other earthlings? Consider those learnings as metaphors to create visual displays.
Show interrelatedness of data through visual ‘echoes.’ Becoming; visual indicators of interconnectivity
Root structures of hemlock trees are interesting because they are a tangible, physical way of visualizing not only vitality, but the possibility of interconnectivity. To experiment with this idea, I decided to generate these roots with cellular automata, such that they take interlock in a more natural form that demonstrates this interconnectivity.
Practice: Use a visual form that allows data to interact and intersect with one another, such as cellular automata
At this point, I had a rough model of what my visualization would look like:
Data visualizations as temporary crystalizations of what we think we know. Frame a visualization as a pathway into a way of viewing the world
There was a moment during the project, when I realized my data visualization bounded by the way in which I see the graduate studio. I wonder how co-design would further improve this process. Nevertheless, making this interpretation clear to my audience is key.
Practice: Frame your act of making this visualization as your own view of reality, a necessarily bounded view.
Further, how might I show uncertainty? Using a generative system (cellular automata) might imbue a sense of and emergence and complexity in the design, allowing the exposure of the relationships community members have with one other (the “echoes”).
Practice: Show uncertainty in the visual form the data visualization takes, possibly through using generative systems.
At this point, I had a visualization paradigm in the form of a generative system, ready to be tested with real data:
Where do we go from here?
At the end of this project, I find myself with a collection of theories, principles, and practices that may serve as a first step in designing visualizations that embrace the body (or situated ways of knowing). Plus, I have several new threads of inquiry for data visualization as a practice:
Generative systems as a way of creating a new grammar of graphics. As an alternative to a standard grammar of graphics, designing systems with emergent properties could be explored. This requires the design of new conventions of mapping data to visual outcomes via system characteristics.
Noticing and generative systems. Generative systems offer us a way in which to behold other life forms; to feel data instead of decoding it. What other ways can we design for encountering data?
Practices for prescribing a mode of visualizing data. What are the ways designers can think through and communicate their models of reality used to produce data visualizations? In which cases are co-design practices necessary?
Yet, the most profound discovery from this project has come from within; my sense of what data and what it means to create visual representations of it has expanded. I see data visualizations as a possible source of life-affirming conversations. A starting point to talk about, to begin to imagine the kinds of futures we can have together as people.
Sources
- Alberto Cairo, talk on How Charts Lie. https://www.youtube.com/watch?v=Low28hx4wyk
- D’Ignazio and Klien, Data Feminism. https://bookbook.pubpub.org/data-feminism
- Timothy Morton, Being Ecological. https://bookbook.pubpub.org/data-feminism
- Deluze, Guattari, A Thousand Plateaus: Capitalism and Schizophrenia.
- Jenny Odell, How to Do Nothing
