Why Data Poetry?

A few years ago I described the similarities I saw between poetry and data visualization, and a co-worker said: "No. Poetry is confusing and hard to understand! Data visualization is about being simple! Your analysis shouldn’t be a riddle” (We'll come back to this conversation in a bit, so stay tuned).

After this conversation, I stopped talking about the connection I saw between the two arts - but I didn't stop thinking about it, largely because I think my training in data visualization started with my training in poetry. 

Poetry is an important part of my journey. I discovered it during my senior year of college. My first three years were an almost doubled course load of psychology and economics with a heavy involvement in research and my own thesis. I think I burned myself out on math and science, so my senior year was all photography, dance (yes, there are embarrassing videos), and poetry workshops. I found writing poetry to be one of the most challenging and enjoyable forms of expression I’d ever tried. I took as many workshops as I could before I graduated, and then I decided that I was going to go to graduate school to become a poet!

This spontaneous decision lasted about 8 months – long enough to work up a nice chunk of debt on a semester and a half of courses. I learned there that strong passions don’t always make the right career move.

After stumbling through a few jobs, I discovered the world of data visualization while working for an IT consulting firm. I came to find that my poetry workshops had well prepared me for a career in data visualization. This wasn’t an immediate realization; rather, it’s an insight that came as I was re-reading a fantastic creative writing book: “Ordinary Genius”, by Kim Addonizio. I stumbled across a quote about poetry that mirrored the way I thought about data visualization. 

How is data visualization similar to poetry?

“Poetry isn’t what we think of as the ordinary, but what we feel and sense is underneath the ordinary, or inside it, or passing through it…a poem is about both the ordinary and the extraordinary at the same time.” Kim Addonizio

In this section of the book where this quote appears, Kim is trying to distinguish poetry from text that written in lines. As I re-read these lines, I tried switching out a couple of words:

Data visualization isn’t what we think of as the ordinary, but what we feel and sense is underneath the ordinary, or inside it, or passing through it…a visualization is about both the ordinary and the extraordinary at the same time.

At the time, I was the manager of a thirty-consultant data visualization practice, and the data storyteller for the data science practice. I’d worked with some pretty ordinary (read: boring) data: checking account transactions, hamburger cooking times, direct mail marketing responses, denied insurance claims and more. But in all these projects we discovered extraordinary things – maybe not incredibly fascinating to the average reader, but certainly things the stakeholders were happy to pay hundreds of thousands of dollars to see: patterns that could predict when a customer was leaving a bank, assembly line obstructions that slowed cooking times, demographics that affected how fast someone responded by mail, and potential intervention opportunities to ensure patients received critical medications on time. Our analysis would “feel and sense what was underneath the ordinary, or inside it, or passing through it”, and then we used visualizations to show it.

In fact, the comparison between the arts of poetry and data visualization are even more basic. Stephen Few said “Data visualization is the graphical display of abstract information...” This definition mirrors something I was taught in one of my workshops: the aim of poetry is to describe the abstract using imagery. I remember one particular instructor telling me: “Don’t tell me you were in love. Show me an image. Show me the time you were sitting in a diner together at 2 AM talking about matchbox cars, playing footsie under the table.” 

We do the same thing in data visualization. We find an image, whether a line or bars or circles or some other shapes, and we use those images to describe abstract statistics.

At a core level, the two arts seek to do very similar things. Kim Addonizio spends time talking about the importance of poetry, and notes: “Description is important because it’s evidence.” Another quote that could be about either art.

The art of being concise

One of the most common critiques I heard during poetry workshops is "that [line/word] isn't carrying its weight." Being concise is incredibly difficult, and that challenge is what continually attracts me to poetry - and it's why I can spend months editing just a few lines. I've literally seen a single word be the difference in whether a poem was published or not. 

My poetry workshops served as practice for being concise when designing data visualizations. I would later read about the data:ink ratio from Edward Tufte, and Cole Nussbaumer Knaflic would tell me to "remove clutter". I just attended a presentation by Michael Warling, who spoke at the 2017 Tableau Conference, and he pointed out that knowing what to exclude is just as important as knowing what to include. 

One of the things that can distinguish a new poet from an experienced poet as well as a new data visualization designer from an experienced designer is conciseness. Great artists in both of these fields know how to evaluate what each piece contributes to the whole, and how to cut the things that aren't holding their weight.

The art of structure

A key feature of poetry is the focus on structure. Poets use things like line breaks, indentation, and poetry to control the cadence of your reading. This is how poets can enforce rhymes, or get you to linger on a particular word, or even surprise you with a word on the next line that changes the interpretation of the previous line! Here’s a fun example of structure I found on a terrifying hand-written note a work:

We are out of coffee
cups. Please use mugs.

When I read that first line, I immediately started thinking of the nearby coffee places I could get to before my next meeting. The line break that the author chose (intentionally?) carried a power with it – but the next line turned into a pleasant surprise.

This is the kind of structure that poetry cares about, and data visualization cares about it too. We may not talk about pacing, but we painstakingly adjust our annotation placement, graphic colors, and labels to ensure that we are guiding the user’s eyes in the right direction at the right time. We’ll agonize over the font size of the filter menu vs. titles to make sure the user doesn’t misinterpret or miss a functionality.

We don’t call it cadence, but at the end of the day we are messing with a visual cadence, instructing the user on how to engage with our art as they engage with it.

The art of play

The final comparison I’ll make here (although certainly not the last comparison to be made!) is that poetry is, by nature, playful. Even when covering dark or sad topics, playing with the medium, language, is a core function of poetry that really distinguishes it from other spoken or written arts. The poets I’ve learned from and written with love to read and try new things. In poetry you’ll often find words slightly misspelled or incorrect punctuation. Poets often use puns, alliteration or rhyme to connect to very disparate concepts. Often, the topic of the poem itself is fun.

This often describes data visualization, too. As data visualization artists, we’re playing with our medium. Take a gander through the Tableau Public Gallery, and much of what you’ll find is the result of artists playing with their medium. The ability to play is made much easier with a renewed focus on user experience by data visualization tools.

However, there is still a common perception that sometimes limits how much we play with these arts...

"But poetry is so...cryptic"

Back to the conversation with my co-worker: “No. Poetry is confusing and hard to understand! Data visualization is about being simple! Your analysis shouldn’t be a riddle.” 

At the time I didn't have a good way to respond to that statement, but a couple years later I read an amazing article in the New York Times: Understanding Poetry Is More Straightforward Than You Think, by Matthew Zapruder. In this article, Matthew notes that we’ve been taught that good poets deliberately obscure their work. In fact, he notes, this attitude is so pervasive that new poets must unlearn this. The problem is that poetry shouldn't be cryptic. With the exception of a few poets in a specific movement, good poets agonize over the wording of their poems to make it as engaging as possible.

Nonetheless, the poetry section in my local bookstore is down to just two shelves.

Ironically, when my co-worker was trying to give me a reason why poetry is a bad metaphor for data visualization, he actually strengthened the analogy - because I've heard people say very similar things about data. There's a number of people that start to shut down at the slightest hint of anything related to data. They'll say "I'm not a [math/computer/statistics] person". And while rigorous statistics and computational engineering do require rigorous education, most of the analytics I've performed and reviewed don't need a special education, so long as the analyst puts in a little bit of effort to make the work approachable. However, there seems to be a strong perception that the average person rolled too low during their character formation to understand anything about data.

The truth is that neither of these should be cryptic, and there are people fighting to take back both of these arts. Matthew Zapruder and others are fighting the “poetry is cryptic” mindset, and there are people and even entire businesses (like Tableau) aiming to fight the “data is cryptic” mindset to make the art of analysis more approachable to everyone. 

Both of these arts should be approachable and fun. They should start conversations, not stop them. They should invite questions, hypotheses, opinions, and they should encourage people to change their minds. 

Ultimately, both of these arts should encourage anyone to feel and sense what was underneath the ordinary, or inside it, or passing through it. 


  1. Love this!! Thanks so much for sharing, Josh. You make a clear, concise and eloquent case for data visualization being like poetry. How fitting.

    1. Thanks! Glad you enjoyed it. Stay tuned for some more direct comparisons.

  2. Leave it to you to connect poetry and visualizations. Nothing you do should surprise me anymore, yet it still does. Thanks for this great read!


Post a Comment

Popular posts from this blog

Visual Cadence

But isn't art subjective? (Refrigerator-note visualizations)