Why do we love data-driven badges?

Or how I painted Easter Eggs for the Data Visualization Society

The Data Visualization Society (DVS) has been around for three months and already has over 5000 members. One of the features of the organization is that each member is given a data-driven badge based on their responses to a survey gauging their proficiency in aspects of data visualization grouped into broad categories of DATA (such as data processing), VISUALIZATION (such as design) and SOCIETY (such as project management). People really responded positively to these badges — posting them on social media and using them as avatars. And as far as information encoding goes, they do a pretty good job, as Amy Cesal explains here.

Each of these badges represents a member and maps their responses across three categories (each triangle represents one of those categories). You get your badge as part of an update email and one of these updates happened to fall on Easter weekend, so I thought I’d go with a sillier take on it: I decided I was going to make a data-driven badge in the form of an Easter Egg.

How were these made? First, I have to confess to not being the originator of the idea. That was Martin Telefont, who posted a couple examples of an egg white and yolk with DVS colors. It was a great idea, and I thought I could make them based on the membership responses but paired with a more decorative subject: the patterns of an Easter egg. Each color represents one of nine skills (three skills in each of the three categories from above). In this case, the color of the egg yoke represents your strongest skill, the color of the egg white is your second strongest and the background color of the eggshell represents your third-strongest skill. Each stripe from left to right goes down the list from fourth-strongest to seventh. Not sure how to read that? Don’t worry, there’s a legend at the end.

But before I talk more about these eggs, I want to dig into what they are: data-driven badges. A data-driven badge is, formally, different than a chart in that it is more decorative and represents a single, atomic subject — whether a person or a city or even a variety of popcorn. It leverages the value of small multiples when placed in a large bunch (like the eggs above) but not insofar as it allows you to easily scan. Data-driven badges are far too intricate for scanning. Instead, you see the incredible diversity of a community along with perhaps a few themes related to the encoded data.

It also plays into the long-heralded and long-toothed notion of the Quantified Self, allowing us to play at knowing each other — and ourselves — as constructs in a system. The act of authorship of a badge is less about optimizing data for reading and more about offering a way to convert the messy realities of people (or popcorn) into comprehensible glyphs. The key aspect to the data is that it has many dimensions… too many to effectively encode and compare.

Weather portraits by Nicholas Rougeux, part of his series of visual experiments that straddle the line between traditional small multiples and data-driven badges.

By no means are these DVS badges the first data-driven badges ever made. Some, like the weather portraits of Nicholas Rougeux above, walk the razor’s edge between decorative badges and comprehensible small multiples. Others, like the beautiful work of Giorgia Lupi and Shirley Wu below, eschew readability for evocativeness. They seem one step away from the geometric decoration of Islamic architecture.

Badges (left) made for movies by Shirley Wu and (right) for TED talk participants by Giorgia Lupi

They even come in 3D (both the digital kind and the physical kind) though this particular variety seems dominated by Shirley Wu.

(left) physical data-driven badges and (right) 3D data-driven badges, both of female nobel prize winners and both by Shirley Wu.

I wasn’t kidding, they also sometimes have to do with popcorn.

Tanveer Jeddy’s data visualization of Stephen Curry’s ratings of popcorn at NBA arenas.

A relatively simple graphic like this might not be considered by its creator to be a data-driven badge. But, much like the Death of the Author in literature, we are free to interpret these as badges based on our earlier definition: primarily aesthetic and less optimized than common charts when it comes to small multiples comparisons based on precision and accuracy.

They also come in the form of little adventurers with pointed hats and swords so absurdly large that even a JRPG hero might think them to be a bit much. I made the adventurers on the left as another pass at encoding the membership of the Data Visualization Society, and just one of many attempts that can be found in this gallery.

In this case, each adventurer has a sword shaped on the self-assessed rating of a member in their skill at visualization, a shield based on how they rated themselves at dealing with data and a hat with a length based on their skill at social aspects of data visualization. The clothing is more playful, with pants and shirts based on latitude and longitude. For those who chose not to answer, they might be barehanded or bare-headed or dressed all in black (a lucky accident and not at all intentional effect that occurred when latitude and longitude were not set).

Data-driven badges continue the tradition of real badges that have symbols and encodings that are meant to impress more than explain. There’s no reason why an FBI agent should have a piece of metal in the shape of a shield with a personified symbol of Liberty and an eagle. These are symbols, like those found in medieval coats-of-arms, that encode thematic data much like their data-driven cousins.

Finally, there’s one aspect to badges you might not expect to be interesting: creating a legend and telling people how to read your badge. For me, often the most interesting part of any new badge is the explanation of how to read it.

Here’s my own legend for the Easter eggs above. It is, hopefully, more comprehensible than my earlier written explanation.

My legend, along with hopefully helping someone read their egg, indicates the conceptualization behind my encoding: When you rate yourself, you also rank yourself, and there’s a sense that ranked lists are not all equal. The top-ranked skill is your core (the yoke) and from there each skill grows progressively less important and (implicitly) more decorative. In its way, it’s an argument about how we evaluate ourselves and what it means unintentionally.

But should we?

People love data-driven badges but are they really useful? Well, they’re never going to win against traditional small multiples approaches when it comes to precision and accuracy. They’re not good for trends or clustering. They likely aren’t going to be the best at encoding topologies or hierarchies or geographic data, either.

Glyphs can become a sort of signature for data visualization practitioners and reinforce their brand. That may not be about communication or traditional purposes of data visualization, but it’s a real need and a real value.

But more than that, data-driven badges are useful simply because people love them. There is no more powerful act in design than reaching out to your audience and making them happy. Happiness and novelty draw audiences in and make them think of systems and how they can think of themselves and other things (like popcorn) as systems and expand the possibility space in ways they might not have expected.

Legends for badges described above including Giorgia Lupi (left), Amy Cesal (middle top) and Shirley Wu (middle bottom and right)

Take a look at these legends. They range from minimal to extensive, they describe an information model that can be hierarchical or egalitarian. Reading the legends is often more like traditional data visualization than reading the actual badges. But encoded in any of these legends is an almost idealistic optimism that says a human being created this based on rules and you can learn to read it just like you learned to read the written word.

Finally, maybe there’s a time in the future when our data visualization literacy is higher and data-driven badges will be simultaneously more common and more useful. One of our problems with evaluating and critiquing forms of data visualization is that it’s done as if we’ve achieved the peak of data visualization literacy. We need a conceptualization of data visualization literacy that is evolving. Maybe that comes with eggs and squiggles and flowers.