Visualization Woes: Smart Software Creates Human Problems
December 10, 2012
I am not dependent on visualization to figure out what data imply or “mean.” I have been a critic of systems which insulate the professional from the source information and data. I read “Visualization Problem”. The article focuses on the system user’s inability to come up with a mental picture or a concept. I learned:
I know I am supposed to get better with time, but it feels that the whole visualization part shouldn’t be this hard, especially since I can picture my wonderland so easily. I tried picturing my tulpa in my wonderland, in black/white voids, without any background, even what FAQ_man guide says about your surroundings, but none has worked. And I really have been working on her form for a long time.
A “tulpa” is a construct. But the key point is that the software cannot do the work of an inspired human.
The somewhat plaintive lament trigger three thoughts about the mad rush to “smart software” which converts data into high impact visuals.
First, a user may not be able to conceptualize what the visualization system is supposed to deliver in the first place. If a person becomes dependent on what the software provides, the user is flying blind. In the case of the “tulpa” problem, the result may be a lousy output. In the case of a smart business intelligence system such as Palantir’s or Centrifuge Systems’, the result may be data which are not understood.
Second, the weak link in this shift from “getting one’s hands dirty” by reviewing data, looking at exceptions, and making decisions about the processes to be used to generate a chart or graph puts the vendor in control. My view is that users of smart software have to do more than get the McDonald’s or KFC’s version of a good meal.
Third, with numerical literacy and a preference for “I’m feeling lucky” interfaces, the likelihood of content and data manipulation increases dramatically.
I am not able to judge a good “tulpa” from a bad “tulpa.” I do know that as smart software diffuses, the problem software will solve is the human factor. I think that is not such a good thing. From the author’s pain learning will result. For a vendor, from the author’s pain motivation to deliver predictive outputs and more training wheel functions will be what research and develop focuses upon.
I prefer a system with balance like Digital Reasoning’s: Advanced technology, appropriate user controls, and an interface which permits closer looks at data.
Stephen E Arnold, December 10, 2012