Saturday, May 25

Data viz - it is taking shape

A summary of the post "A Conversation with Data Visualization Experts" - simply the subheadings, as a jumping off point (and not intended a full repost as IFTTT did - the risks of automation!)

  • Which came first, the question or the data set?
  • Visualization as tool vs. visualization as art
  • Data visualization is an iterative process; seek feedback
  • The right tools for the job
  • Good questions, iteration, and feedback

Bostock confirmed that his team at the Times does this as well, using a git repository to track all of the iterations the visualization has been through…sometimes as many as 250 changes to a single piece. 

Love the idea of using git to track iterations during testing and ideation - that way you also have a backlog to play around with next time a similar dataset or question comes up. Been using svn and saving major iterations (ie when something works...) but taking it a step further makes sense. Just need to automate and simplify the flow a bit more than "right click, type, enter, close"

Then onto the ideas or discussion points themselves;

It seems that some of the better works come from the question "how can we do this differently" - taking a trope to the next level, like the 512 paths interactive. That piece is a functional artwork (ht A. Cairo) - it is great to look at even before you dig into the options and understand the underlying data. But it is most of all a storytelling tool - because all the paths were not created equal, and that (in my mind) was the main point of the story. You could dream up all possible iterations, but only a very limited set lead to something other than the actual outcome.

The time to iterate, and getting the culture in place for it seems like one of the most challenging parts of getting data viz going. At least in a smaller organization where there isn't room for a dedicated team, you need to get people comfortable doing a crit on something they don't really understand how came together. In that sense D3 is a great tool, because it is relatively easy to "live change" a sample; 
  • Could it be green to pink? Swap the colors, reload and discuss. 
  • How about bar vs column? Swap X and Y and you'll have an idea at least. 


And now with tools like Vega and the playground you can do a bit of up-front mocking in minutes. Personally I still love doing ideas on a whiteboard, but it does take a bit of 'translation' to get new people up to speed - that's where a digital first makes more sense.