About 100 attendees, three keynotes, five short talks, demos, discussions, food, music, and a fantastic atmosphere: the Tapestry conference for storytelling with data took place on February 27 in Nashville, TN. Here is a conference report with links to talk videos, as well as some first news on Tapestry 2014.
Setting and Format
Conference hotels tend to all look the same: nondescript, badly lit, depressing ballrooms, terrible acoustics, and just way too many rooms with names that all sound the same.
Not the Union Station Hotel Nashville, though. It’s a former train station has been turned into a hotel, and it’s simply beautiful. It has been renovated very well, with lots of nice little touches, like a time table from the 1930s, etc.
The event was organized by Tableau Software and Investigative Reports & Editors (IRE), emceed by Ellie Ms. Tableau Public Fields, and lasted one day. We picked the place and time to be close to the NICAR conference in Louisville, KY, but in a somewhat more intimate and interesting setting.
In order to foster discussions, not just talks, we had generous breaks, a long lunch, a demo session, and a reception the night before the conference. The talks consisted of three hour-long sessions for the keynotes and five short stories.
I’ve been a fan of Jonathan Corum’s work at The New York Times for a while, in particular since he won the Best of Show award at Malofiej last year for his Guantanamo Bay graphic. I was delighted that we were able to convince him to give the first keynote at a time when many of the details of the conference were still a bit undetermined.
Jonathan is actually a science graphics editor (which few papers besides the Times have), and he talked about some of his science graphics. There were many great insights and pieces of advice about audience, the reading of graphics, how labeling should work, etc. It was fascinating to see his thought process and how he turned mediocre figures from scientific papers into very clear, well focused graphics.
Perhaps my favorite part of the talk was when he talked about the importance of not just visualizing data, but figuring out what it means and explaining that. That is really the key difference between pure visualization and journalism. One is a tool, the other one is communication and explanation.
You can watch Jonathan Corum’s talk here, as well as check out an amazing write-up of his slides. In addition, he has written up a set of deleted slides, which is almost as interesting as the ones he included.
The five short stories (admittedly a bit of mixed metaphor) were 20 minutes each and meant to bring in a larger variety of speakers than we could have had with only hour-long talks.
Pat Hanrahan, Showing Is Not Explaining
Continuing one of the topics of Jonathan Corum’s talk, Pat Hanrahan talked about how showing something is not the same as explaining it. Pat is a professor of Computer Science at Stanford University, and one of Tableau’s founders. He showed Euclid’s algorithm for finding the greatest common divisor of two numbers, both as a Python program and a visualization of it running. It was an impressive visual proof, but perhaps a bit much for people not used to thinking in terms of proofs or code that way.
Watch the video of his talk if you didn’t quite follow.
Cheryl Philips, Choosing The Right Story
Seattle Times journalist Cheryl Phillips brought the conversation back to reporting, in particular when data is involved. Just throwing the data at the reader is not enough, a journalist needs to interrogate the data. She introduced a lot of people in the audience to the term nut graf, which is the key point of the story. What does the data mean, what does it tell us? Being clear about these points is important when creating a story, not just a visualization.
Nigel Holmes, Why 29 Is Such A Stunning Number
The grandmaster of explanation graphics, Nigel Holmes, gave perhaps the most entertaining and impressive short talk. His focus was on providing context and making the reader understand the magnitude of a number. He gave a fantastic demonstration of Bob Beamon’s long jump world record in 1968, of over 29ft (about 8.9m), which he held for 22 years. The number doesn’t make people understand the magnitude of that feat, however, so Nigel demonstrated just how far that is. It was a fun and impressive demonstration.
Hannah Fairfield, The Art of Honest Theft
New York Times (and formerly Washington Post) journalist Hannah Fairfield talked about the connected scatterplot, and how she and Amanda Cox borrowed the idea from a paper in a journal about oil and gas extraction. She then also briefly talked about the Times’ famous Snow Fall project, which she had been involved in, where the inspiration came from, etc.
Bryan Connor, Critics, Critiques and Critical Visualization
Bryan Connor runs the most cleverly named visualization blog, The Why Axis. He talked about the importance of understanding the intention behind a project when discussing and critiquing it, and argued that we need to know more about the creation process to really appreciate a particular piece.
While the lunch was a great opportunity to talk to people, and the food was fantastic, it would otherwise not be something I’d include in a conference report. But I’m making an exception to mention our special guest.
Country singer Stephanie Quayle entertained us with classic, Nashville-style country music. While I’m not a huge fan of that kind of music, I was very impressed by her humor. Between songs, she told jokes that all had data as the punchline, even though they obviously had nothing at all to do with data. But she made them work.
After lunch, it was my turn. I spoke about the differences between visualization and visual stories, in particular when it comes to narrative. But just because we’re telling a story doesn’t mean we can ignore what we have learned in visualization research, so I spoke about information scent and the use of interaction. I think the talk worked well to build a bridge between the other two keynotes and to provide a connection with the academic research done in information visualization. I got a good number of comments and people who disagreed with me, which is always a good sign.
Scott McCloud is the author of Understanding Comics and Making Comics, two great books on comics that contain a lot of useful information for any kind of visual storytelling. As odd of a choice as he might seem for a conference on storytelling with data, he provided the perfect closing to the day. Not only is he an incredibly energetic and interesting speaker, but he has thought about how we read images and sequences of images in a much deeper way than most other people.
There were lots of fascinating insights in his talk, like his thoughts on how timelines work and when they don’t work, how children start out as visual thinkers and are then made to read and write, how we perceive things and what that means for the “cognitive load time” of information from an image, etc.
Unfortunately, we were not able to record Scott’s talk.
I should be reluctant to praise the conference, since I was one of the organizers; but I think we really succeeded in bringing together a very diverse set of people, who nevertheless had a clear common interest and shared many ideas. Some of the topics that came up in many talks and in the conversations included the difference between visualization and explanation, the importance of context, focusing a story (the nut graf) rather than just providing all the data and letting the reader figure it out, etc.
Throughout the day, and already on the evening before, where we had a little reception, there was an amazing energy. I didn’t see anybody stand around by themselves, everybody was engaged in discussions, demos, etc. There was an effortless mixing of academics, journalists, designers, and others, that I haven’t seen happen quite that well elsewhere. It it wasn’t clear before, that really drove the point home: we’re onto something here.
Many people have asked us whether there will be another event next year. And the good news is: yes, there will be a Tapestry 2014. Details will be announced soon. It will take place around the same time as this year, and again on the East Coast. To stay posted, follow the Tapestry Blog and/or the Tapestry Twitter account. The Tapestry website, which embarrassingly still talks about this year’s conference, will also be updated soon.