OpenVis Conf 2017, Aerial View
Well, another OpenVis Conf has come and gone. The gathering was a great opportunity to discuss the state of the field with a variety of practitioners, from data scientists to information designers to social scientists to everything in between (and adjacent).
While my head is still buzzing from the breadth of perspectives and wealth of inspiration, I figured I’d jot down some notes to share. Rather than writing a general-purpose recap with summaries of all the talks, however, I’d like to focus on threads running through the field of data visualization as presented at OVC 2017.
I’ll add links to the videos as they become available online. Note also that, sadly, I had to leave early to catch a flight home, and so missed the last couple of talks. Luckily, Twitter gave me a decent insight into what happened after my departure (but you should definitely watch the videos!).
Education and Data Literacy
Data visualization is, fundamentally, a communication medium. We use it to explain and to educate, and so it’s no surprise that this was a common thread among the talks.
Mike Bostock
Mike kicked off the conference with a keynote in which he unveiled his work in progress, d3.express. It will offer a notebook environment analogous to Jupyter (for Python), with a focus on improving expressivity and reducing friction. In addition to offering a livecoding scratchpad in which the output of your program appears and updates instantly, it follows a reactive paradigm that means variables can be referenced before they’re declared, and as their values change via code, the output seamlessly updates to reflect the change.
You can sign up for a d3.express release notification here, at d3.express (TIL express is a TLD!).
Catherine D’Ignazio & Rahul Bhargava
Catherine and Rahul made a case for designing for learners, rather than users. As our tools become increasingly complex, sophistication does not necessarily follow. In order to ensure that a new generation of visualization authors will rise up to put our work to shame, we have to work to bring them on board. They offered four guidelines for offering affordances to newcomers:
- Focused
- Guided
- Inviting
- Expandable
Note that while learners !== users, we were all learners before we were users. In this light, as we create new tools, we should strive to welcome learners before we focus on advanced, experienced users.
Amanda Cox
Amanda focused on the knotty problem of uncertainty in data and visualization. As evidenced most viscerally by the gulf between 2016 election forecasts and results, we as practitioners have a lot to learn about how to convey uncertainty. And of course, that is a dual responsibility — we need to work to improve data literacy so that others can more fully grasp the concept of uncertainty as well.
Process
A number of talks were dedicated to examining process: how we create the things we create. These offered insights into how a range of visualization practices are structured, from personal work to small collaborations to 1000+ person organizations.
Shirley Wu & Nadieh Bremer
Many of us are familiar with this product of last year’s OpenVis Conf, and if you’re not, you should go dive in right now: datasketch.es is the tree from which hangs the fruits of the collaboration between Shirley and Nadieh. They each presented thorough, live examples documenting their experimentations along the way to each month’s final works, offering a detailed look at all the discoveries and dead-ends along the way.
Hadley Wickham
Unfortunately, Hadley’s was one of the presentations I had to miss, so my details are sparse. However, the Twitters tell me that he offered a compelling and detailed look at his own workflow, in the spirit of the Feltron Report. Introspection like this always offers takeaways that scale up to the field-at-large, so I’m looking forward to watching this one when it airs.
Eric Socolofsky
Hey, that’s me! I focused in on overlaps between data visualization and artistic practice, with an analysis of techniques that translate from the art world (specifically, generative art) to the world of vis. I wrapped up with a caveat that these techniques are not suitable for all use cases, but also that those of us toward the ends of Elijah’s spectrum should consider ways we can move toward the middle.
Amy Cesal
Scaling the process conversation up, Amy made a strong case for the use of visualization style guides in the work of medium-to-large organizations. In addition to cleaning up and standardizing look and feel, visualization style guides offer the added benefits of increasing production speed and accessibility.
Tools & Frameworks
This is easily the largest group of this writeup. It appears, not surprisingly, that much of our work as visualization practitioners is also tool-building. Perhaps this speaks to the relative youth of our field; perhaps to our desire to communicate and share; or perhaps just an offshoot of the open nature of modern technology.
Nico Belmonte
deck.gl is a framework for visualizing large-scale datasets in a performant way, with an emphasis on geospatial work, leveraging the speed offerings of modern GPUs. It grew out of internal work at Uber and is now robust at its second open-source release version.
Matt Brehmer
Timeline Storyteller addresses the perennial need to visually convey stories that elapse over time. It offers flexibility in the representation of time, and recognizes the need for non-linearity in many data stories.
Mike Bostock
d3.express, described above, looks like an amazing new way to express and analyze data. Can’t wait to try it.
Mikola Lysenko
re.gl offers a lightweight, expressive API for writing shaders and leveraging WebGL capabilities of modern browsers. Plus people have already made some jaw-dropping stuff with it.
John Alexis Guerra Gomez
The quest to “untangle the hairball”, to better understand the myriad relationships between people in social and professional networks, is a noble one. John offers two combs for the untangling: netclustering.js, a d3-force plugin; and networkcube.net, a network visualization authoring tool.
Connor C. Gramazio
Connor’s doctorate research into color and human perception has bequeathed us with at least these two tools (due to my flight, I missed any gems beyond these): d3-jnd, a micro-library to aid with color choice, and colorgorical, a smart hybrid between Kuler and ColorBrewer.
Kanit “Ham” Wongsuphasawat, Dominik Moritz & Arvind Satyanarayan
Vega-Lite, from the UW Interactive Data Lab, offers visualization-via-config: a relatively simple yet broad set of JSON expressions allow for Tableau-like experimentation, but with much more room for expansion. Data Voyager takes this one step further, offering a chart creation GUI that generates the JSON that generates the charts.
Social Good
Visualization offers insight into the world that can be transformative. Many of us in the community are compelled to harness this quality not only to observe, but to actively improve the world.
Amelia McNamara
On the surface, Amelia’s talk was a relatively benign study of how we move from points to polygons, and what we can do with polygons from there. However, she quickly moved to show a consequential application of these ideas: by explaining gerrymandering as a manipulation of these rules, she offered us insight into how to fix it.
NB: If fixing our electoral system is a passion, or even a passing interest, of yours, please drop me a line! We have top men and women working on it right now.
Lisa Charlotte Rost
One of the most powerful and difficult talks of the conference came from Lisa. She investigated techniques for bestowing our visualizations with empathy: to lift ourselves away from faceless dots on maps and toward real human stories. A must-watch for anyone invested in using their work to improve the human condition, but really for anyone practicing data visualization — we all have a responsibility to be ethical and empathetic with the ways we communicate our data.
Matt Daniels
Matt taught me the word decarceration, for which I am forever thankful. He chronicled his process trying to understand, visually, why nearly 0.5% of America’s population is imprisoned. He came across a trove of incredibly detailed data about Florida convicts, which opened a bit of a Pandora’s box and set off controversy at the conference.
Thankfully, he displayed work in progress, and I believe the valuable feedback he received from attendees will improve the quality and sensitivity of the final work. It’s a pressing issue and he deserves kudos for investigating.
Alan McLean
Another of the talks I was disappointed to miss completely, but judging from the response on Twitter a lovely one. As I understand it, Alan dissected visualizations of personal health data in an effort to make these charts both more personal and effective. Having a fitness app nag you and call you “average” is hardly motivational; watch his talk for insights on what is.
Analysis
It goes without saying that visualization originated as an analytical tool. For many people, it’s still indispensable for visually identifying signal in data, and goes hand-in-hand with the other tools they use daily for the work of data analysis and data science.
Julia Silge
The gamut of human emotion and expression can be found in the texts we’ve written. Pulling these signals out from a birds-eye, corpus-wide view requires sophisticated analysis techniques. Julia reviewed best practices for structuring text for analysis and demonstrated techniques for extracting frequency and sentiment from text.
Kanit “Ham” Wongsuphasawat, Dominik Moritz & Arvind Satyanarayan
I’m mentioning Vega-Lite and Data Voyager here again because it was presented primarily as an analysis tool. The speed at which it can create visual representations of data will make these valuable tools for rapid iteration in the process of analysis and data science.
Personal quests
When you’re passionate about your work and your tools, you work overtime. For fun. And occasionally, but…not that often…for profit. The following talks are the natural outcome of a field filled with people who love the field.
Kai Chang
Kai took us on a detailed tour of the JavaScript Canvas API, with suggestions for squeezing the most out of it in terms of performance and interaction. He bookended the talk with context-setting about human perception and the relevance of our work.
Noah Veltman
Noah probably knows more about polygons than any non-officially-a-mathematician out there. He’s also, as it turns out, a total raconteur. His journey toward The Perfect Polygon Transition is full of horse-chairs, horse-elephants, and other equine oddities that actually illustrate the challenges and value of his quest exceedingly well. Easily the most entertaining talk of the conference and one of the most edifying as well.
Robert Simmon
Also in the realm of mathematics, Rob explored the famous GDAL tool at a level that even commandline-phobic folks can appreciate. My favorite takeaway: viewing the world from one million miles away is essentially just another map projection. Clever.
Shirley Wu & Nadieh Bremer
DataSketch.es offered a perfect framework for experimentation and, as described earlier, process documentation for Nadieh and Shirley. But we can’t forget the lengths they went through to turn out piece after piece for months. That’s the power of collaboration — your quest is no longer personal. You can’t let the other down. Amirite Elijah Meeks?
Talking outside of the talks
We go to conferences to share ideas, to hear new perspectives, to be inspired. Much of this comes from the formal presentations, but the discussions in the margins are just as valuable. I’m not going to outline all the conversations I had with friends and colleagues here — I’ve gone on long enough already.
However, there was one conversation in particular over dinner on Monday night that is worth a recap here. I mentioned this publication and how we’re trying to position it as an ongoing discussion about how we define our field and how we do our work. Seated at the table were seasoned veterans and relative newcomers, so we had a range of perspectives on the importance of this discussion that I wanted to capture here. Among the values we considered:
- being effective at your work
- moving forward in your profession
- conveying the value of your work to stakeholders
- conveying the value of your work to the public
- searching for a new job
- marketing yourself
- finding a school
- finding a focus in school
- being motivated to do your work
I’m sure there are plenty more reasons to keep this conversation open; these are just the cross-section that appeared in a 20-minute conversation among six different people. I’m looking forward to seeing even more surface as we collect more points of view here.
Onward
I caught up with a lot of bright people at OVC and met many more. At the close of my talk, I advertised this publication as a place for these conversations to continue, so expect followups from a range of perspectives to follow in the coming weeks!
