Enrico Bertini (@FILWD) - Feb 27, 2013 Nashville, TN
Jonathan Corum, NYT
Jonathan’s portfolio: http://13pt.com/
Jonathan put up a recap of the talk at http://style.org/tapestry/ which includes most of the slides.
Think about who is your audience. Design for someone else instead of designing for yourself.
Axes:
Oversimplification <---> Too many details
Explanation <---> Decoration
Narrative <---> Exploration
Don't be your own audience
[Similar to Alberto Cairo’s classification in his Functional Art book]
Many examples on communication of scientific ideas taken from scientific journals. Amazing one of how a flea flies.
[People can actually have different starting points and it's important to support all of them, this is also true in visualization in general and related to my idea of selected first + expand.]
Add meaningful annotation, not just labels [Inspiring for Vis in general! How do we make labeling better? … Related to InfoVis 2012 paper on Just in Time Labeling]
Reduce tedium: interact with data not the interface [I think a lot of people still need to learn this]
Reveal patterns (make them apparent) [How can we do this (semi-)automatically in vis?]
How can I throw much of this away and still tell a story.
Apply common sense.
Quotes grabbed from twitter:
"Don't let the tool decide what the visualization is, make sure you use the tool,"
"Visualization does not equal explanation. There are patterns - explain them."
“Respect your audience. Respect the data. Apply common sense. Make/facilitate connections btwn points”
"Best way to learn something is to have a project for it." Big data set + need to share = new skills” [fully agree with this! this is very good advice]
Use common sense vigorously. Respect readers. Don't just visualize — tell stories. And do it for others.
Pat Hanrahan, Stanford
http://graphics.stanford.edu/~hanrahan/
Showing is not Explaining
My job is to teach things to people, that’s not different from what journalists do.Most of what I do is teach people how to think. Gist of the talk is: you cannot just show things you have to explain them.
Euclid’s algorithm (first known algorithm). The algorithm is more interestin1g than the data for a CS, e.g., “How does Google rank pages”. Greatest common divisor
One thing you have to learn is coding (Obama with Zuckenberg and Bill gates). Everybody should learn how to code - http://code.org video mentioned by Pat Hanrahan https://www.youtube.com/watch?v=dU1xS07N-FA … #tapestryconf. What Most Schools Don't Teach.
[Pat showing the algorithm on a python terminal live! How cool is this?]
Wikipedia has a nice visualization of the algorithm: http://en.wikipedia.org/wiki/Euclidean_algorithm
Problem with animation. Tversky (http://www-psych.stanford.edu/~bt/) studies it at Stanford
[Showing the steps of the algorithm with a visualization]
I’ve shown how it works but I didn’t explain WHY it works!
Recording and showing: I can show you the data but the data is not interesting. That’s only the first step! Recording and showing is getting really really easy with computers.
Q&A
Al lot of algo work is making it faster. Does visualization help making algos better? Some algos are very deterministic others are based on heuristics. You use heuristics because there’s no right answer. But in this way you can cheat and give the wrong answer. [Visualizing heuristic algorithms is a super interesting idea, we should try that].
What do you think is the role of algorithm visualization? The public should be aware of how we process these data and what we as CS people do.
Cheryl Phillips, Seattle Times
Choosing the right story
The first thing I want to focus on is story. The story is the starting point. What do you want to say? Is there anything surprising? What do you want to tell? Data without a theme is just a bunch of data not a story.
Click on a link showing the data and the result is a spreadsheet. How do you go about focussing on a story? “Interview the data” [I like the idea of “interviewing the data”].
(Story on methadone used as painkiller.) Tell the story from different perspectives. Tell multiple stories. Found through visualization higher density in areas with lower income, this triggered a new story to tell: “Methadone and the politics of pain”: http://video.seattletimes.com/1311237892001/.
Elephants dying in the zoo. Using an anecdote to tell a story.
Being open to possibilities. There are many elements to tell a story.
Q&A
Do you think people have lost the focus on story telling to to focus on technology? Not sure if it’s lost. I don’t know … Not sure.
Nigel Holmes
Favorite graphic: the queen mary measured against other famous monuments.
How big is the moon? Moon compared to Australia. Putting things into a context. Real person vs. sculpture.
Number 29. No idea Jonathan was going to show the long jump graphics.
In 1968 something amazing happened: a guy (Bob Beamon) failed his first two jumps. Then he jumped 29ft 2.5
[Nigel shows 29ft with a rope. Wow!!!]
The reason why you can see this now is because it is in a different context. Context is the key here.
Mike powell broke the record in 1991.
Numbers are wonderful. Data is wonderful but context is the key to understanding.
How much toothpaste people use. Did a calculation. Relating things people can immediately grasp.
Hannah Fairfield, NYT
The Art of Honest Theft: Evolution of a connected scatterplot
We are all building on each other’s work. Show how graphics evolve.
Oil price graph from a paper found by Amanda Cox. We are so used to see time series that we are not used to see lines going backward. So she created “Oil’s roller coaster ride”. See how this is related to how much people drive “Driving shifts into reverse”. The price of gas has a lot to do with how much people drive. But there’s no loop in the graphic.
Now, we have done a couple of these, what can we do next? Keep the same horizontal line (miles driven per hour) and map the mortality on the y-axis “Driving safety, in Fits and Starts”. At what points are things changing?
The most important question in journalism: “what’s next?”
Snow fall the avalanche and tunnel creek. Influence from Hugo Cabret, it opens with drawings, introduces a wonderful visual story. Then there is tect and then drawing again, etc. Maybe we can have the same immersive feeling?
One of the skiers survived because he had an airbag. As you read the story you can see all in a sudden the figure of the guy.
Showing the skiers path. That was really challenging. There are so many names and you don’t know who is going to live or die. The graphics shows specific skiers. It gets very specific. We designed in many different ways. Always had the idea of the connected scatter plot in the back of my mind.
Q&A
How does the process work? More details? The snowfall piece was very collaborative. We built a lot of things you never saw because they never worked. Each piece had about 2 or 3 different people. On another piece we had real time simulation of the avalanche. An expert from Switzerland was able to estimate the size of the avalanche. We never work in a vacuum.
How much time you spend chasing the “right” data to be shown in something like a connected scatter plots? Part of it is following the news. In between the scatter plots I did hundreds of graphics. Have both the bread and butter and set aside some time for a playground: very important for all of us [true for me as well in research, little experimentation very important anf setting aside the time for it].
Bryan Connor
I came to vis from design school. People coming into this field could feel intimidated. Stefanie Posavec also mentioned to be intimidated by the field. The same sense of intimidation. There is a sense of right and wrong in visualization. Some people find it useful to point “wrong things” out.
There is so much history of visualization.
Explore, explain, entertain.
The loudest people: Tufte and Few
Example from branding world. It’s easy to judge visualization on a surface level. Goal of y-axis it to go deeper.
Critic, Designer, Audience [this is useful and important]
I am trying to look at visualization more closely. What is the decision making process. Why is this designed this way? Trying to get past this surface level judgment. Why? Why people created this visualization. Create an environment where critic is useful to anyone (audience and designer).
One of the most effective ways to create a good critic is to create a story about the visualization. You can come to a much deeper understanding and make it more useful.
[Visual literacy is a big theme and not sufficiently taken into consideration. We should all do more to teach people how to make but also how to read visualization.]
Robert Kosara
Visual Story Telling in the Age of Data
Research has great stuff but its stays in the circle of academics. Paper on storytelling: http://kosara.net/papers/2013/Kosara_Computer_2013.pdf
How can we communicate data more effectively. What can do to make a bar chart more effective? Monstrous costs from Nigel Holmes. That’s wrong! You are distorting the data.
But there are charts with no story. Lots of complexity and no narrative.
Of course there is a danger in telling stories.
There is a range in two dimensions: tell a story/doesn’t tell a story, story depth.
Simple charts: no story apth, nd no deInformation graphics: tell you something, Visualization: high depth no story.
Shifting waterways from Jonathan Corum: narrative, storytelling. It’s like a sequence in a comic.
Napoleon’s march chart
Is it an effective way to tell as story? It’s effective to show information.
These are ways to show you data points and see all of them at once. But there are examples where information is added incrementally. Walking through sequential steps can be very effective to tell a story [I have to remember that as a more general principle for visualization. We have been so obsessed with the idea of making overviews but that’s not the only way of showing data].
Showing gapminder: http://www.gapminder.org/
Showing all the data at the same time vs. going through steps
More depth …
Scatterplot from NYT: Man vs. Women weekly earnings. Many say people can’t read scatter plot. But just adding reference lines can make a scatter plots very readable.
Housing’s rise and fall in 20 cities: http://www.nytimes.com/interactive/2011/05/31/business/economy/case-shiller-index.html#city/IND20
You can add data and interaction and increase depth (several examples of reworked visualizations with Tableau). Adding small multiples on the chart.
Adding more data can be very effective. Exploding the bubbles in the gapminder chart you can tell a much richer story. Countries within a continent differ a lot!
But giving more data can also be dangerous: it could be overwhelming.
Focus …
Another, more complex, map of Napoleon’s march. There’s too much data.
Bush vs. Obama Administration (Job Loss) aka The Bikini Chart: http://eagereyes.org/blog/2012/bikini-chart
Information Scent …[a]
The Jobless Rate for People Like You: http://www.nytimes.com/interactive/2009/11/06/business/economy/unemployment-lines.html
You can see one data points but many others in the background. This tells you there is more if you interact with it.
Showing unemployment rates in time for different cities. Some show some interesting spikes.
Information scent is a very powerful tool and not used enough in storytelling.
Author … and Audience …
What’s in the top right quadrant of the chart? Robert proposes Visual Data Stories.
Q&A
What is Tableau doing to get into this quadrant? We are trying to see what is out there first and then see what else we could do. But we think of a bigger community. This is a new way of telling a story especially online.
Scott McCloud
Amplification through simplification. In comics we write with pictures. The language takes many forms … Facial expressions: there are only 6 basic facial expressions. But we can mix them the same way we mix primary colors.
Showing a cartoon with 4 mixed emotions.
The Grimace Project: http://grimace-project.net/
“All pictures are words”
Do comics count as reading for kids?
Cognitive load time: how long it takes to load it in our mind.
RSAnimate: http://www.thersa.org/events/rsaanimate. I receive visual stimuli only when I need it.
Showing Scott’s daughter giving a talk. The notion of synchronization is important.
I am seeing a lot of people knocking at the same door.
We don’t really have to fill up all the space. If it takes 10 slide
I had trouble thinking of what to note for this one, would greatly appreciate contributions here.
Side note: This seemed mostly aimed at journalists.
My sparse notes here don't really do the talk justice, probably because I was enthralled with the presentation. Especially Nigel attempting the long jump - I doubt any of us will forget seeing that!
See Driving Shifts Into Reverse and Driving Safety, in Fits and Starts
Story Depth | Tells a Story |
Facts | Narrative |
Information Scent | Focus |
Audience | Author |
Many thanks to the hosts for putting together a great conference. I'm looking forward to next year's!
[a]Remember to review the literature on Information Scent. Peter Pirolli did a lot in the past.