Slashdot Mirror
Calling Out GE's Misleading Data Visualizations
theodp writes "Stephen Few never did suffer data visualization fools gladly. After seeing an oil exec (mis)use data viz to put a positive spin on Gulf Oil Spill cleanup efforts, Few felt compelled to call out BP. And now it's General Electric that's got Few's dander up: 'The series of interactive data visualizations that have appeared on GE's website over the last two years,' writes Few, 'has provided a growing pool of silly examples. They attempt to give the superficial impression that GE cares about data while in fact providing almost useless content. They look fun, but communicate little. As such, they suggest that GE does not in fact care about the information and has little respect for the intelligence and interests of its audience. This is a shame, because the stories contained in these data sets are important.' Concerned about his strong reactions to poorly designed data visualizations, Few asked his neuropsychologist wife whether he might be overreacting. She, too, agrees that GE's natural gas visualizations are maddening, which one might be tempted to dismiss as predictable, although Eyeo Festival presenter Michal Migurski also declares GE's effort 'one terrible, terrible bit of nonsense.'"
"never did suffer data visualization fools gladly" == "Few despises bad charts"
"got Few's dander up" == "Few lost patience."
Recommendation: steer clear of the writings of William F. Buckley, Jr. There is a difference between business English and literary English.
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
What's good for GE shows up as legislation in America.
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
After looking at the various visualizations the only one that is worth anything is the third one that shows the years of remaining reserves for each of the fossil fuels. Even that one isn't that impressive. Also I don't get the use of the Sierpinski triangle, Apollonian gasket, and Sierpinski carpet style shapes for representing each fuel source. I haven't looked at much data visualization, but it doesn't seem the use of these doesn't add anything.
Time to offend someone
Since the summary is so difficult to understand, the jist of the article is that GE's visualisations (I will not grace them with the title "graph") are completely useless in comparing datasets, and are completely confusing to use. This seems to indicate that GE (like many companies) like to fiddle with the presentation of data to push their agenda. (Shock, Horror!) Sadly, this case is an insult to good design principles and statisticians everywhere.
What's good for GE is good for India
FTFY
ACs don't waste your time replying, your posts are never seen by me.
At first I was thinking that this story was about GE trying to push its agenda or doing something evil. But I RTFA and this is actually about this guy complaining that people are using the wrong type of chart and making poor design decisions. The big punch is that his wife agrees with him.
I am so shaken up by this story, I know I will get all nervous the next time I insert SmartArt charts in Powerpoint - I would be so ashamed to end up publicly flogged on this guy's strongly-worded blog...
Reminds me of a former coworker who is spending his evenings writing blog entries about companies that dare use Arial instead of Helvetica on their websites.
lucm, indeed.
...then perhaps we'd better not even get started on the quality of the underlying data.
The sole purpose of corporations providing information is to convince the public of something that will benefit the company. From inception to design through data collection, analysis and reporting there is a defined marketing objective and it never involved "let's find out". Yet we treat with less scepticism than reports from an independent academic that at least in theory has survived a thorough peer review - though even that tends to assumes the technical parameters operated were correct. How often do you see reference to, say, questionnaires in the methodology which then goes on to even let you see a copy of the questionnaire?
My how the fallen have fallen.
Nerd rage is the funniest rage.
.. I reach for my revolver, because it means that they want to impose their view on me and consider themselves justified in doing so.
Justice equals law equals the desire to use force. Otherwise they would talk about 'social suggestions' instead.
I think we should make a distinction between GE, the company hosting the site, and Stephen McCandless, the rather famous data visualization specialist who created the figures. (Here's his website: http://www.informationisbeautiful.net/ )
The problem is not that the data presented are not useful, or that they're deliberately intended to deceive, which we could fault GE for. As I see it, the problem is that the graphs themselves are crap. They hide useful information, and they use shape and color in ways that seem to provide information but don't, and in general they focus on the aesthetic appeal of the charts at the expense of the data.
When I first encountered McCandless's site a few years ago, I really loved it, but as time goes on it's begun to piss me off. For example, his chart on relative radiation risks:
http://www.informationisbeautiful.net/visualizations/radiation-dosage-chart/
Logarithmic charts are always difficult to explain to the public, but the triangular shape of his graph makes it even worse, suggesting a linear increase in dose. He compares it to XKCD's chart, but his version is inferior in every way. XKCD uses color and shape to provide information; in McCandless's version color and shape have negative information content.
Another example: a graph of time travel plots in film and TV (minus Dr. Who):
http://www.informationisbeautiful.net/visualizations/timelines/
The curvy lines look nice, but all anyone can make out of this is a confusing snarl of lines too tangled to parse. Once again, shape has negative information content in this image.
But the king of the bad visualizations is probably another graph McCandless did for GE:
http://visualization.geblogs.com/visualization/co2/#/flights_London_Tokyo
Here, there's no way to intercompare various quantities, and figure out which of two choices is bigger. Shape, color and position are once again meaningless or misleading (things are shown the same size even when they're 8x different), quantities are in incompatible units, and worst of all some of the numbers are flat-out wrong (for instance, fuel usage of aircraft).
But the one thing these all have in common is McCandless, not GE. So let's not fault megacorporations who're trying to communicate a message: let's fault information presentation gurus who care more about appearances than on information presentation.
Didn't stop them from losing tens of billions of dollars in the financial meltdown, but GE is a big fan of Forced Ranking: "Jack Welch, General Electric's former CEO, is often associated with a 20-70-10 distribution: the top 20 percent is rewarded for best performance, the middle 70 percent is rated 'average' and the bottom 10 percent is coached for improvement. The 'rank-and-yank' system, also associated with Jack Welch, automatically terminates employees in the bottom category, allowing organizations to purge the worst performers."
Maybe the summary could actually try to summarize what is going on. For example it could let us know:
* Who is Stephen Few? Is he an expert in this filed or some random blogger?
* What data on the GE website are they talking about? If there are many different sets of data give an example of one.
* What is 'Eyeo Festival'?
Providing some basic context lets us know whether TFA is worth reading - as is stands it just bunch of meaningless drivel.
Limited Liability doesn't mean what you think it does. It provides a wall of separation between owners/shareholders and employees, but it does nothing to limit the liability of individuals doing bad acts. If someone does something bad, doing so under the guise of an LLC doesn't shield them from legal liability. Merely from financial liability if the actors are acting within the confines of their fiduciary responsibility.
>"never did suffer data visualization fools gladly"
suffer: verb - to put up with, to endure, to tolerate
data visualization fools - collective noun - idiots that don't know how to draw a graph.
gladly: adverb - enjoy with a smile. this word modifies the verb "suffer"
To rewrite the sentence fragment:
"never did gladly tolerate idiots that can't draw a good graph"
I like the original better.
> and "got Few's dander up"
If you've ever angered an animal enough to provoke a fight or flight response, its fur stands on end and its dander (skin flakes and dried saliva from grooming) is disturbed in a cloud as it moves. This can be seen if backlit. It's pissed.
It's good imagery.
--
BMO