Advanced Virtues & Visualizations
This blog accompanies a paper to be presented at IEEE VIS 2026, “Charting the Moral Universe: Capturing Virtues and Values of Data Visualization Practice” written by Chloe Hudson Prock, Enrico Bertini, and Michael Correll. For more details, read the paper.
In 2019 I had a paper at CHI. There’s an extended Heidegger quote in it and the world “critical” gets tossed around a few times, but, more or less, that paper can be distilled down to a couple of (hopefully) relatively uncontroversial claims:
- Rather than offloading all the ethical considerations to the people collecting the data, and/or to the people making use of the data, those of us engaged in data communication have our own ethical issues to think about as well.
- You can use the language of virtues to pick out some of these ethical considerations that might otherwise be lost if you just think of data visualization as a sort of neutral task of putting numbers in peoples’ brains as efficiently as possible. For instance, you can design visualizations to support agency, and/or to support transparency, or in service of lots of other things other than efficiently communicating as many individual numbers as possible.
I included two caveats in the (as far as I know, unrecorded and so permanently lost to the ages) paper talk. I didn’t phrase them as bluntly as this at the time, but hopefully gists were gotten across. Namely:
- Please do something, rather than nothing. Just pointing out that ethical issues exist is not sufficient. There’s the hoary old Marx chestunut “Philosophers have only interpreted the world, in various ways; the point, however, is to change it.” Connected with my (hypocritical?) desire that people actually do things rather than just critique those things, is the related form of helplessness I see a lot in tech circles, the “sympathetic magic” that so long as somebody, somewhere is thinking about ethics, the rest of us are absolved from having to do anything about it in our own work. I want everybody, not just ethicists, to think about the moral aspects of their work, and try to better themselves.
- I’m just some asshole (hereafter, “JSA”). I included some virtue-based exhortations in that paper (e.g., “make the invisible visible” and “challenge structures of power”). But those were just the things that I happened to be worried about at the time (admittedly, I am still worried about them). What I, personally, am worried about is not a particularly useful field-wide diagnostic. I am highly limited by my perspective, background, and straight up ignorance about many, many things. The idea of somebody going “well, our project nicely addresses the ethical concerns of this one rando who wrote a 10 page pdf once, let’s go ahead and ship it” should fill you with dread, regardless of the identity of the rando in question.
There are a few ways of circumventing the objections above. For instance, you could actually go around talking to people and learn from them. Then you’d get more information about what it is that people actually do, as well as what they care about, when they are thinking about the ethical components of their own data visualization work. As you might predict, doing so produces way more (and way more diverse) sets of ethical considerations than you get from a single source, especially if, as discussed, that single source is JSA.
Get Michael Correll’s stories in your inbox
Join Medium for free to get updates from this writer.
We therefore interviewed twenty visualization practitioners, data journalists, researchers, and “thought leaders” (the scare quotes are to indicate my distaste for the word, but it unfortunately fits here). We report on these interviewees anonymously both to encourage candor and to fulfill the constraints of our IRB protocol, but know that we got much heavier hitters than I expected (or, really, thought we deserved): people who have been shaping the conception of the field, who have produced data visualizations that would be in anyone’s shortlist of the most consequential of the modern era, and who have been teaching and thinking about ethics both longer and more deeply than I have. I’m grateful for (and, frankly, undeserving of) the time they shared with us.
The resulting paper is our attempt to distill all of these conversations into a set of things people care about (cribbing from the design community, we give these things the much more grandiose title of “matters of concern”). Many of these matters of concern have ethical components to them: those things we call (admittedly, not unproblematically or unambigiously) “virtues.” We end up with a lot of data visualization virtues (68, if you’re curious; but I’m less interested in the precise count than in the coverage), crossing broad categories like those associated with professionalism (like doing the work on time, or having personal integrity), artistic qualities (like doing beautiful and impactful work), and then those more tightly connected with the kinds of “critical data visualization” efforts that have been bouncing around the field for a while now (things like collecting data with empathy and in ways that preserve agency).
In keeping with my JSA-dom, I will admit some bias here and say that one of the reasons I like talking about virtues in a visualization context (and I should point out that you don’t have to be a specifically “virtue ethics” person to use or talk about virtues in an ethical context) is that it makes ethical data visualization much more interesting to talk about. In a lot of ethical data visualization writing and teaching (my own included, I should point out), there’s this weird focus on the same set of obvious bad examples, your 3D pie charts and bar charts with truncated axes and what have you. The idea being that, well, if you avoid doing something from this set of obvious bad behaviors, then you’ve done your job and you don’t have to worry anymore. And I think that focus on avoiding (occasionally parodic, mustache-twirling) “vicious” behavior rather than attempting to be “virtuous” is both incredibly incomplete and incredibly limiting. I can fail to collect data with empathy, or fail make my visualizations appropriately accessible to my audience, or generally produce all sorts of harms that have nothing to do setting out to intentionally mislead people.
These observations didn’t make it into the paper, but these interviews did produce some patterns (or anti-patterns?) about ethics in data work that gave me some things to be anxious about. In particular:
Ethical thinking as a thing that only sometimes kicks in. I think people (especially those who’ve read something like Data Feminism) would look a project around visualizing sensitive private data about, say, representing some minoritized group currently targeted by a repressive government, and go, “ah, there are ethical concerns here, and real harms to getting this right, I should be very careful here.” But those same people would perhaps be on automatic/autopilot if it was “just” a dataset about the price of pizza in their neighborhood or something. I call this turning on your ethical brain (you could call it something else, like, if you didn’t get enough Heideigger from the first paper, you could say it would make ethical data visualization design “present-at-hand” as opposed to “ready-to-hand”). I get why this is the case. Like, if I had some big complex mental (or legal!) apparatus that had to be engaged to think critically about data visualization, presumably you’d think it a bit of a waste to have to fire it up just to look at a bar chart you made in five seconds for a powerpoint presentation that most of your audience will be asleep for. But, as per my “please do something” admonishment way up top, I think this is kind of a dangerous place to be. There’s a reason why we make pilots follow pre-flight checklists, after all, no matter how many hours they have in the cockpit, or how “important” the flight might be. It turns out all sorts of “unintended consequences” can crop up if you only start thinking about consequences in what you feel are extraordinary ethical circumstances!
Ethical thinking as only for experts. Again, this is a perfectly natural impulse. The world is complicated. Doing the right thing is also complicated. When things are complex, and we will be in big trouble if we get it wrong, it makes sense to rely on experts. If I’m sick, I will go to the doctor. If I am in legal trouble, I go to a lawyer. If something is flooding or sparking I’d go to the plumber and electrician, respectively. Their expertise is valuable here, and my lack of expertise immensely harmful except in simple cases (I know how to treat a cold or a skinned knee, for instance, but if I’m having a heart attack then I’m calling the ambulance no matter how many WebMD articles I’ve skimmed). Similarly, there was a common (and true!) refrain that ethics is this complicated thing that there is a lot of expertise around, and even those smart ethicists can (and do) disagree about things, so best to leave the ethical thinking to them. As with the prior pattern, I think there’s a bit of danger here. The (admittedly imperfect) metaphor I’ve landed on after some trial and error is nutrition. There are expert nutritionists, and people with many years of post-graduate education who know all sorts of stuff about the digestive system. But, we’ve all got to eat food sometimes, right? With, admittedly, some exceptions, we don’t sit around starving because “gee, it’s all too complex” or “I’m not a Michelin chef, what do I know about food?” You must have some procedure for deciding what to eat. In the same way, the people we talked to, ethicist or not, were out there making charts, often every working day. They had to have some way of thinking about what makes them good or bad.
Virtue-based thinking is, to me at least, a nice solution to both of these above concerns. It’s a nice compromise between only “activating” your critical faculties when you are pressured to (and it’s so easy to be convinced that it’s not your problem, and the borders between “trivial” and “non-trivial” ethical issues so fuzzy…) and a sort of Chidi Anagonye-style total ethical paralysis. It’s also a nice way that you, yes you, can work on improving yourself (and your work) without having to bust out some sort of utilitarian calculator to precisely determine how good your chart is, or memorize every single law of good chart design, or whatever your preferred expert-driven theory of right action is. You can develop virtuous habits, prioritize particular virtues in your own work, model your behavior after particularly virtuous role models: all sorts of virtue-flavored activities that get you metaphorically up and moving.
That’s the kind of data ethics I think I like to see. Okay, yes, smart and tech-savvy ethicists thinking about things at the policy and institutional level about establishing practices and guidelines and policies. But, down here in the data visualization trenches, working to better ourselves, our work, and our world, a little each day.
