The latest iteration of a colorful web-based computer program illustrates the “thought process” of a computer faced with making a move in a chess game. The Thinking Machine 4 (http://www.turbulence.org/spotlight/thinking/method.html) lets you see how it analyzes each possible move before it decides how to respond to yours.
Using threads of green for your moves and orange for its counter moves, it calculates many moves ahead before deciding, but so quickly that its human opponent cannot possibly keep up.
“The effect of a dizzying number of curves, far too many moves to keep track of, is exactly what I was shooting for,” says creator Martin Wattenberg, founding manager of IBM’s Visual Communication Lab. “Part of the point of the piece is to get across that idea—the inhuman scale of a computer’s calculations.” It gives one pause to realize that the human brain can actually compete against, and beat, such a machine.
The viewer does get some clues: the thicker the lines, the better the moves are perceived by the computer to be. It also radiates “waves of influence,” subtly showing which pieces are in particularly powerful positions on any given move.
The result is a beautiful picture shifting before your eyes, and a pretty good game of chess. Not a great game, however: the designers wanted it to represent an average player, not a grandmaster.
“The chess engine we built is simple and uses only basic algorithms from the 1950s (alpha-beta pruning and quiescence search)” notes the website. “The program’s unconventional initial moves may raise eyebrows among experts: we did not give it an ‘opening book’ of standard lines since we wanted it to think through every position.”
From a usability perspective, the choices of colors and movement were intentional. “I wanted to be able to play a standard game of chess, and the main issues were to make sure people knew what was going on: when the computer was thinking, when it was their turn, and so forth,” says Dr Wattenberg.
Dr. Wattenberg and his lab are now using social media to explore other forms of data visualization through their Many Eyes project (http://manyeyes.alphaworks.ibm.com/manyeyes/). Taking usability to the extreme, the project encourages people to upload data sets and play with different ways of visualizing the data. Others comment on the visualization, so one learns directly what types of visualization work best, at least for the sort of audience that is attracted to this site, (which, one suspects, may not be typical Internet users.)
“Many Eyes is a bet on the power of human visual intelligence to find patterns. Our goal is to ‘democratize’ visualization and to enable a new social kind of data analysis,” explains the Many Eyes website.
With over sixteen basic visualization approaches ready to play with, and more being created by users, it will be interesting to see if, over time, any of these new approaches becomes as common as pie charts and line graphs are today. We have already seen tag clouds emerging as an important way of visualizing word frequencies, but to judge from the amount of explanation the site needs to provide about the alternative forms of visualization, it doesn’t look like many new approaches are yet ready to be presented to the CEO.