Making Random Moves

Most human players and chess computer engines alike select their moves through a combination of two processes: searching through the consequences of possible moves, and evaluating current and future positions. Each of us have our own talents and capabilities in performing these two tasks. But what if chess was played randomly, with no bias, ideas, experience or thought put into any moves?

Well, somebody took the time to answer this question, and did so in an impressive fashion. This report by @billautomata, generated approximately 100,000 random games, finding that a vast majority ended in a draw (85%) and lasted an average of 342 moves. This data is presented with some very neat visualizations.

While the high draw is not necessarily surprising (after all, these computer players are identically matched in skill), it would be interesting to know if White or Black had a greater share of the few wins that did occur. Unfortunately, the data from the above analysis is not freely available (as far as I know).

In light of this, I have utilized the chess.js JavaScript library to create an accessible way to generate and store random computer games. You can access this simple tool from my other site. While this may not serve any useful purpose, it could nonetheless be interesting to see which side (White or Black) ends up with a better score after random moves. I should note that others have created random moving chess programs with better visuals, such as developed by the chessboard.js team.

That's all for this week! Stay tuned for more computer analysis and links. Please share your thoughts, questions, comments and consternations below.