Guest post from Frank Van Ham, Master Inventor, Information Visualization and Visual Interaction Expert
As Many Eyes is being brought out of hibernation. I thought it would be appropriate to reflect on how it came to be and some of the paths the original research team followed.
As the title says, these are my personal views and recollections and they should be taken as such. Some of these I had to piece back from memory, and some from old presentation decks still on my drive.
I joined IBM in February 2006, having been asked by Martin Wattenberg to join IBM Research’s newly formed Visual Communication Lab (although I guess it wasn’t called that yet, at that time). Fernanda Viégas and Jesse Kriss had joined shortly before me, making for a nimble four person team.
At that time, IBM Research had a wonderful Explorative Research program, which came with four years of no-strings-attached (ok, fewer strings attached) funding goodness. That sounded awfully enticing, so the team set out to come up with a sufficiently forward thinking idea in the visualization space.
One of the initial ideas that came up was a computer program that could look at a visualization and then interpret the pixels like a human could, so we could use it for investigation and evaluation of the visualization process itself. In my mind, this sounded like a lot like “build part of an artificial brain,” which was certainly ambitious, but unfortunately none of us had even the slightest idea how this could be accomplished.
At the time, Martin had just completed the baby name voyager and had received a lot of online response to it; and, Fernanda had just finished a PhD on the communicative aspects of visualization. I guess this combination triggered the idea of “a site where people could upload their own data and analyze data together.”
Jesse and I immediately loved it. Over the next six months, the proto-idea was called “Societal Intelligence” and we set out to create a prototype (see photo) with the help of an IBM designer and extra hired programmer. The latter coined the hilarious phrase “it works in my head,”, when asked if we could see any of the code he’d been working on for a month or so.
Unfortunately, we couldn’t cross compile the code in his head to Java and ended up doing most of the work ourselves. The talented Jeff Heer also joined as a summer intern that year, and did a nice prototype on integrating fluid commenting with interactive visual exploration which confirmed some of our thinking.
Meanwhile, the Societal Intelligence prototype did the rounds at IBM customer conventions (the largest being Information on Demand in 2006 where Jesse and I ended up demonstrating it non-stop for four days). Matt McKeon had joined the team and we set out to create a scalable internet version of it, intending to launch in early 2007. Martin ended up churning out a lot of the basic visualization components, Fernanda did the designs, Jesse and Matt wrote most of the backend and I worked on the pluggability architecture and some of the advanced visualizations.
Somewhere along the way Martin and/or Fernanda came up with naming it Many Eyes (after the Open Source proverb, “Many eyes make all bugs shallow”). Looking back, the hardest and most frustrating part of this development exercise turned out to convince IBM legal to approve a public (!) website where people could upload any data they want (!) for anybody else to see (!).
As a tip, if a non-technical legal person asks you if a client machine is downloading IBM code, do not argue that technically users would be downloading Java byte code. And speaking of Java byte code, a lot of people have asked us “why are you running these visualizations using Java in the browser?”
In December 2006, another website called Swivel (now defunct) launched that offered many of the same capabilities we envisioned. This forced our hand and we moved up the release schedule.
The first release of Many Eyes went live in January 2007. This was a fun and exciting time, as you could check the website every hour and see what new visualizations users had created. Over 2007, the site received a lot of press.
Martin and Fernanda racked up a lot of frequent flyer miles that year during the 2007 Many Eyes tour, and we did a prototype version for the New York Times and started to add more social features (topic oriented groups that people could join, better commenting and user profiles).
In the fall of 2007, we presented the first paper on Many Eyes at IEEE InfoVis 2007. It energized the conference, and “Visualization for The Masses” became a thankful new banner for a research community that, at the time, was struggling with its relevance.
Although the site was having great pickup and usage, we did not observe as much discussion around visualization as we had initially expected. Instead of talking about visualizations they created on Many Eyes, our users tended to embed them in their personal webspace.
Discussions around specific visualizations were mostly happening on webpages outside of Many Eyes. In a way, Many Eyes was being used more like an online tool than an online community. By tracking these embeddings we could at least see where and in what context our visualizations were used, which made for some interesting ethnographic work. One of the other advantages of having a relatively well-visited public website with pluggable visualizations was that we could author new visualization types and deploy them online relatively easily, which made for instant massive testing.
Over the course of 2008 and a good portion of 2009 we saw a lot of work on textual visualization, with the word-tree and phrase net visualizations. The backend was rewritten in Ruby on Rails with the help of Irene Ros, who joined us midway through the project. In the meantime, the amazing Jonathan Feinberg created the Wordle app, which was integrated in Many Eyes as well.
As funding for Many Eyes drew to an end, the technology behind the site was transferred to the IBM Business Analytics division. The Visual Communication Lab disbanded between 2009 and 2010, with all of the original team going their separate ways.
Overall, working on the early instantiation(s) of Many Eyes was amazing. I was able to work closely with an extraordinary set of talented people and interns, helped deploy a site that was used by tens of thousands of people in the public and that generated a wave of enthusiasm for interactive visualization.
Ironically, six years later I now find myself working for the Business Analytics division, which is dusting off Many Eyes, integrating it with its core infrastructure and putting it forward as a the public face of our investment in the visualization area.
Many Eyes v2 launches at the end of March with several new enhancements that will continue the heritage of the site to advance visualization on the web, including:
· A comprehensive site redesign with updated layout and presentation. Plus, new affinity areas to find and navigate visualizations by industry or topic, such as finance, healthcare and risk.
· Addition of the Expert Eyes blog dedicated to helping you learn how to create effective and engaging visualizations that provide maximum insight and tell a story. IBM visualization luminaries and IBM Researchers from the Center for Advanced Visualization will contribute their perspectives regularly.
· Two new visualization types, including a heatmap and view-in-context visualization built on IBM’s Rapidly Adaptive Visualization Engine (RAVE). RAVE, a declarative language based on the Grammar of Graphics, provides a flexible way to create visual mappings by describing what the visualization should look like, without having to write any tedious rendering code.
· New options to share visualizations across the web and on social networks, including popular visual social networks, such as Pinterest.
Discover the newest version of Many Eyes beginning March 25 by visiting www.ibm.com/manyeyes (yes, a new URL as well).