February 17th, 2009
In In Praise of Pattern, Stephen Ramsay makes much the same point that I made last week, that one of the most effective ways that computers can benefit qualitative, non-binary research is by breaking down texts (in the broader definition of the word) and presenting them in a way that a human can understand them in a way not possible before.
When looking for inspiration for visualisations, Ramsay went fishing, finding that nature naturally forms into its own graphs in many places, if care enough to pay attention. In this spirit of breaking down visual communication, I’d like to go through a thought experiement on visualising a basic piece of information, represented as a node. This is based on some old sketches that I found in my journal, and the node idea is influenced by the Mandala browser.

Okay, so say we have our information organized in nodes. Really, it could just as easily be a screenshot or box of information, but for now, I’ll start with a simple point. Like on a graph, it’s free in it’s own space; that is to say, the node exists in a larger, two- (or more) dimensional plane, unlike the one dimension that textual information usually follows.

Now, if have multiple nodes that need to be differentiated, we can easily show this visually through shape or fill (color, shade, texture). Say you’re plotting nodes of hit pop songs from the past fifty years. You could quickly identifier their most important characteristics by the look of the node. The gender of the lead singer could be visualized with shape (female=circle, male=square, and none=triangle). Each decade could be assigned a different color, so songs from the sixties could be blue, or songs from the eighties could be pink. Length of songs could be quickly shown by the size of the the node (i.e. longer songs could have larger nodes).

Once you have a number of nodes, you can show relationships between them. Since they exist in a multidimensional plane, distance is probably the most aparent way of showing relationship between items. Other ways include branching, which shows a flow, and orbit, which can use distance (from center) to show relevance to the main node, but also show relationships between the satellite nodes (by virtue of how close they exist in orbit).

The last form of visualising that I considered is longitudinal analysis, or showing relationships and changes over time. Traditionally, nodes are graphed with time as the independent variable. However, in computing, animation is also a useful and effective way to show temporal change. I’ve found an increasing number of visual communication relies on animation not for novelty, but to emphasize a point. This 2008 Democratic primary breakdown is a prime example (for example, click between “Whites” and “Blacks” and note how the animation brings home the point). Animation can even be combined with graphing, if there’s a different dependant-independant relationship that you hope to show. Google’s motion charts are an example of this.
—
These concepts just scratch the surface. If you have any of your own ideas, feel free to share them below.