SciArt Magazine - All Issues | Page 29

So when we speak of defining aesthetics, creativity, computational intelligence, and so on, it’s not a question of creating dictionary entries. It’s really about creating fleshed-out theories. And if we understand something, then we should know how to, at least in principle, build it. And if we don’t know how to build it, then it’s questionable whether we really understand it. So this art making as experimental philosophy is both a way to generate new theories and concepts as well as a kind of empirical attempt to verify theories and concepts through physical instantiation. So I wondered two things. First, in what specific ways did they all look alike? And then second, why in the world should solutions to an optimization problem have any kind of common aesthetic at all? For the first question, I did a statistical analysis of angles, lengths, areas, and so on. It turned out that those distributions were very consistent across a number of different problem solutions. For the second question further analysis showed that optimizing the path for minimization had inevitable formal effects. Indeed, these were the same effects that showed up as being statistically significant. PS: You express your thoughts on emergence—new, unforeseen ‘products’, so to speak, of a system that emerge following interactions between smaller components of that system—in The Traveling Salesman (2008). Can you walk me through the process of devisFor example, I devised a simple geometric ing and eventually executing the piece? proof that given four points, and connecting two pairs of points with a line segment each, PG: The left-brain versus right-brain model of the total length of the two line segments would cognition is an exaggerated oversimplification, always be longer when they crossed than when but it’s a useful metaphor in this case. One day they didn’t. Thus, crossed lines always indicated I was studying evolutionary computing, and one a suboptimal path, and the optimal path would of the sample problems was the traveling sales- not have crossed lines, and thus it would have man problem. This is a canonical problem in to create one single large shape. optimization theory. The goal is to start with a set of cities and then plan a route that visits evThe statistical analysis also showed that angles ery city once and only once such that the total close to 60-degree angles were favored. This distance traveled is minimized. This presumably makes sense from the point of view of optimisaves the salesman time and gas money. This zation because when connecting three points problem gets exponentially more difficult as the with two line segments, sharper angles result number of cities increases linearly. in longer paths than connecting those points a different way. I was looking at a diagram of cities as points connected by an optimally minimized path. At The traveling salesman aesthetic became, for some point, the right-brain aesthetics of the me, a variable icon representing the way form plot somehow pushed aside my left-brain techcan emerge from optimization. To present this nical interest. The solution was, in fact, a very result, I decided to paint the resulting shapes as interesting-looking line. And as I looked across murals. Scale always matters, and at a scale that a number of solved problems, they all had a envelopes the body, the figure-ground reversal common aesthetic despite the fact that the of the image is emphasized. initial points were randomly placed and spread over areas of different shapes. The murals are created in a site-specific manner in that an entire wall is used. In the comSo I scanned the page from the programming puter, the wall is populated with random points, book, pulled it into Adobe Photoshop, and then and then those points are connected using a filled in the enclosed areas with color. Aestheti- traveling salesman problem solver program. The cally, I liked what I saw. There was a very strong resulting image is then projected back on the figure-ground reversal effect. And oddly enough, real-world wall, the borders are outlined with none of the lines crossed. A single maze-like masking tape, and the shape is then painted figure was always created. with house paint. SciArt in America December 2014 29