Good.is makes pretty pictures and neat videos. One came across my virtual desk today— this cute little video entitled “The Astronomical Amount of Data We Deal With” and its sister visualization “The World of Data We’re creating on the Internet”. Looking at them, you’re amazed at how much data we’re consuming. 24 Petabytes per day of data processed by Google! Wowza. And 1.3 exabytes of mobile data? That sounds like a lot.
So does 7.5x10^24 molecules of H20, until you figure out it’s just one glass of water. The problem is, as pretty as these videos and pictures are, the facts in them are rarely provided with any context. A lot of great visualizers and good visualizations fall into this trap. Good’s visualizations dazzle us with big numbers, but you can break just about anything down into large numbers and make it sound impressive. I’d be more impressed if there were some context around the stats— how have they changed? How many petabytes per day of data was being processed by Google in 2004?
I’m burying my face in hands in shame at the lack of context in this product that ultimately I was responsible for at the Sunlight Foundation. So Barbara Lee got $12,000 from Gilead Sciences. That stat alone is meaningless. What’s really important (and often times more important) is how much she’s taken from Gilead Sciences compared to everyone else, or how many individual contributions led to that sum? Politiwidgets is providing no context.Lots of visualizations suffer from this same problem. I’m guilty of helping to produce them. Let’s talk about Politiwidgets. It’s a site I helped to create at the Sunlight Foundation that tries to make it easy for bloggers and journalists to put information about members of Congress and their districts in their articles. Here’s one about Barbara Lee’s campaign contributions from Gilead Sciences:
In fact, the context is so sparse that the visualization tells a pretty horrific lie: that companies can contribute directly to a candidate’s campaign.[1] Barbara Lee has received 0 dollars from Gilead Sciences, Inc. They can’t. No, not even with the Citizens United Supreme Court Decision. They can buy ads and publicly endorse a candidate, but a company cannot give money directly to a candidate, only individuals and political action committees can. So that number isn’t based on corporate money, but rather based upon the employees of that corporation. Shame on me for not catching this whilst I was there.[2]
According to this lie, Verizon was the 4th largest contributor to Cynthia McKinney’s 2008 Presidential Campaign with $1600, all coming from this guy— hardly a corporate conspiracy by Verizon to send representative cum boxer McKinney to the oval office.
Data visualizations represent a great opportunity to distill truth into something easily consumable. But they’re not very infovegany. They’re highly processed nuggets of information, designed more to be visually stimulating than they are to give you any concept of depth or nuance about what you’re looking at. It isn’t to say that visualizations are bad— sometimes they’re wonderful and can explain things better than raw data can. It’s to say that at best, you should treat them as a gateway drug to questions you’re looking at answering rather than things that can contain answers themselves.
[1] The number in this particular transaction’s count is actually pretty true, though. The transaction includes two 5,000 contributions from Gilead’s PAC, and two thousand bucks from their head lobbyist. The point is, even if it is basically true, it still isn’t wholly true that Gilead Sciences, Inc. gave this money to Barbara Lee— there’s an important piece of context missing. Sunlight does a much better job at providing context around Gilead in its new site, IE.
[2] Eric Mill, the developer who led the politiwidgets project sent me a note saying that not only did I not catch it, but I also insisted upon it being this way. I don’t recall that, but the idea of me insisting on something stupid seems pretty reasonable. So double shame on me.
