Extreme Chess Update
Today the Extreme Chess Team faced its most formidable challenge yet–a 1480, ranked # 13.
We eagerly invited our opponent into a 10 minute match playing as black. The first match was opened with Petrov’s Defense (1. e4, e5, 2. Nf3, Nf6). We tried to move ourselves into a better position by forcing black to retreat his knight and allowing us to castle queenside. On move 23, our probability of winning dropped from +0.27 to -3.22. The engine was not looking ahead to the position of black’s bishop. White resigned on move 26.
During the second match, the chess engine failed to execute a viable strategy. The opponent would counter faster than the Extreme Chess Team could input and output moves. After 92 moves, white lost on time and gave up 12 Elo points to the higher ranked black. White’s probability of winning on move 92 was +0.03. Had the game continued, it would have resulted in a draw.
What went wrong here? First, the Rybka chess engine was being used on an outdated platform. The Extreme Chess Team is currently in the process of building a new computer with a quad-core processor that should increase the engine’s rate of calculation and allow it to search at a greater depth within the time constraints. It is likely that our opponent was using a more powerful computer. Even if our opponent was using a lesser algorithm, it was able to perform significantly more calculations per second of CPU time, leading to a victory in match one.
As a result of poor performance in match one, the Extreme Chess Team increased the engine clock to allow for greater search depth in the first few moves after book play. This put us at a time disadvantage of about one minute. During the last three minutes of our game clock, we forced the computer to move within three seconds of input, keeping our total move time at less than ten seconds. Our opponent did the same and because we gave up time early in the game, we lost on time by 53 seconds. The lesson? Modifying the engine clock during a timed match can only be considered under favorable circumstances, meaning the engine shows a probability of winning greater than +1 and/or our opponent is down on time. It may have helped us avoid a checkmate but also caused us to lose on time. Despite our opponents CPU advantage, our chess engine was clearly a more selective algorithm than our opponents and would have resulted in a draw during match two.
Although our opponent claimed to be a human player, he would need to be thinking 10 to 12 moves in advance to counter every move made by the CPU in match two. In the human mind, our short term memory is limited relative to the computer. Therefore it is extremely unlikely that this individual was a human player.
Our current Elo rating is 1353. Our leaderboard standing is 20 out of 19,734 players.
If you intend to buy the latest computer technology, be prepared to drain your wallet. Intel’s flagship quad-core processor, the