Justin Li

The Data Experience

2009-05-13

I've been thinking a lot about the task of rapidly conveying vast amounts of information in a short period of time recently. I'm not sure why that has been on my mind. I'm calling it the data experience, because (perhaps with the thing on synesthesia and all) I feel that it is more than just data visualization, but it involves the other senses as well.

Experiencing data, if done properly, can give the sensation of more than just the 3D world we live in. For example, stress analysis simulations put areas of high stress in red, and areas of low stress in blue - for a 3D object. The color actually gives the model a fourth dimension. Of course, it's not the entire 4D object - in the case of stress analysis there is no such thing - but it is a mapping from 3-space to 1-space. Just as a 3D surface can be represented on a 2D graph with color (a heat map; the two representations can be combined), a 3D model with color can express a 4D surface. Although less common, other orthogonal features of graphs can be used to express other information, taking the visualization into even higher dimensions. An animated heat map (for example, a stress analysis which shows the change as the stress is applied then relieved) is convey 5D data. I can imagine other things - for example, perhaps the thickness of the mesh outline - which can be used to show even more information at once.

On the other hand, sometimes the data can just be in 3D, but the user experiences higher dimensions. One of the authors of The Mathematical Experience talks about a 4D intuition. A project at Brown University allows people to manipulate a tesseract (also known as a hypercube - a 4D structure where all edges are the same length and are orthogonal to every edge it shares a vertice with). By using the controls to rotate the virtual tesseract in different ways, the author suddenly could feel and reason about the tesseract, just as normal people can mentally rotate a cube. I find the possibility of lesser dimensional beings able to have an intuition about higher dimensional objects fascinating.

Another use of data visualization is not so much about the data, but about the visualizer's partition and organization of the world. Randall Munroe, for example, does this rather often, as does Indexed.

I don't have much more to say, but I would like to share some cool data experiences:

Finally, a recent example in my life. In psychology we were talking about the implicit association test (IAT). Usually this is done on a computer, so the response time can be measured to millesecond precision. That could not be demonstrated easily in class though, and so the professor did a cool analog. Instead of pressing a button, we had to slap one of our thighs to indicate where category an object belongs. The point is not to get it right - obviously there is no feedback between our legs and the computer - but to make the disassociated concepts apparent. Since people are slower with harder associations, but different people are slower by different amounts, what actually happens is with harder associations it would take the the class longer as a whole to return to silence. A simple, but very effective, way of showing how the IAT works, and a great example of why I call it the data experience.