Computer Olympiad 2000 Lines of Action Championship =================================================== +-------------+-------+-------+-------+-------+--------+ | Crosstable | YL | Mona | MIA | W-L | Medal | +-------------+-------+-------+-------+-------+--------+ | 1 YL | x | 3-1 | 3-1 | 6-2 | Gold | +-------------+-------+-------+-------+-------+--------+ | 2 Mona | 1-3 | x | 3-1 | 4-4 | Silver | +-------------+-------+-------+-------+-------+--------+ | 3 MIA | 1-3 | 1-3 | x | 2-6 | Bronze | +-------------+-------+-------+-------+-------+--------+ Background on the Computer Olympiad ----------------------------------- The Fifth Computer Olympiad was held in August 2000, in London, England, in conjunction with the Mind Sports Olympiad. After several inactive years, the computer competition returned with contests in Chess, Go, Shogi, Awari, Hex, Amazons, and Lines of Action (LoA). Our initial plan was to create a hybrid program called "Debby", combining YL's amazing search speed with Mona's better-informed evaluation function. We discussed ideas for adding new features within the constraints of YL's ultra-fast framework, and Yngvi spent about a week trying to produce a program that was better than either of its parents. Unfortunately, that wasn't an easy task, and we decided about a week before the Olympiad that both YL and Mona would compete in London. I hadn't worked on Mona for months, and now had to hastily prepare for the tournament. It required code for handling time control, a small opening book, and a number of other minor details. The time management system was rather crude, and I made an "improvement" that I would later regret, resulting in a loss on time (which never could have occurred with the basic algorithm). Naturally, it happened in a winning position, in the game that would decide the gold medal... The opening book was hacked together in a few hours the night before the competition, and didn't even handle symmetric moves. It was worthwhile though, producing several instant moves, and sometimes went deeper than Yngvi's much larger automated opening book. Meanwhile, Yngvi produced a well-tested time management system and also implemented pondering on the opponent's time. This gave YL a significant edge in real time over Mona and MIA, in addition to already having a much faster search. That could have resulted in some very lopsided victories, but it didn't turn out to be quite that easy... Gamescores (in chronological order) August 23, 2000 ----------------------------------- Click on the game number for the complete move-by-move analysis. Notes on how to interpret the output are given below. Olympiad-1 MIA vs YL 0-1 1. f8-f6 h4-f2 2. d8-b6 h2-e2 3. b6:f2 a2-d2 4. f1-f4 h7-f5 5. c8:f5 h3:f5 6. e8-g6 a6-c4 7. g8-g5 c4:f4 8. c1-c2 e2-d3 9. b1-f1 h6-h4 10. g5-g2 a3-c3 11. g6-g3 d3:g3 12. f6-d4 h4:d4 13. b8-e5 a7-e3 14. e1-a1 f5-e6 15. a1-e1 a5:e1 16. f1:f4 e6-d5 17. c2-d3 a4-c2 18. d3-e2 h5:e2 0-1 (mate in 2 moves) Olympiad-2 YL vs MIA 1-0 1. b1-b3 h3-e3 2. b8-b6 h5-f7 3. b3:f7 h6-e6 4. c1:e3 h4-f2 5. c8:e6 h2-e2 6. d1-b3 a7-a1 7. d8-e7 a6-d3 8. e8:e2 h7-f5 9. b6:f2 a1-b2 10. f2:b2 a5-c5 11. g1-d4 f5-d5 12. g8-g7 d5-b5 13. b2:b5 a4-c4 14. e6-d6 a3-a1 15. e1-e5 a1-c1 16. d6:d3 c1-b2 17. g7-c3 b2:e2 18. f1-f4 e2-d1 19. f4:c4 d1:b3 20. e7-f6++ Olympiad-3 MIA vs YL 1-0 1. d1-b3 a5-c7 2. b1-b4 a7-d7 3. g1-g3 h3-e6 4. g3:c7 h7-e7 5. e1-e5 h6:f8 6. c1-c4 f8:b4 7. b8-f4 a3-d6 8. f1-e1 a6:c4 9. e1:e6 a4:f4 10. e8-g6 b4-b6 11. b3-a3 b6-c5 12. g8:c4 h5:e5 13. g6:d6 h4-e4 14. a3-a5 e5-b5 15. d8-b8 f4-e3 16. b8-b6 a2-a4 17. a5-a7 e3-d2 18. c8-b7 h2-h1 19. a7-b8 e7-d8 20. e6:e4 d2:d6 21. c4-d5 c5:c7 22. e4-a8 d6:b6 23. a8-a6 h1-e4 24. d5:b5++ Olympiad-4 YL vs MIA 1-0 1. d1-b3 h3-e3 2. b1-b4 a7-d4 3. c8-c6 h6-e6 4. c1:e3 h7-g6 5. b8-d6 h4-f2 6. e8-b5 h5-e5 7. d8-d5 h2-e2 8. g1-h2 e5-c3 9. h2-h3 c3-c1 10. f8-d8 c1:e3 11. d8:d4 f2-c2 12. g8:g6 a2-d2 13. e1-f2 c2-c4 14. c6:c4 e2-f3 15. h3-f5 e6-e4 16. f2:d2 f3:b3 17. c4-c3 a6:d6 18. f5-e6 e3:e6 19. g6:e4 e6-c6 20. f1-e1++ Olympiad-5 YL vs Mona 0-1 1. b8-b6 h7-f7 2. e1-g3 a4-c2 3. g8-g5 a5-d5 4. g1-g4 h3:f1 5. f8:h6 h2-e2 6. e8-e6 a6-d3 7. g3:d3 f7-d7 8. b1-b3 a7-c7 9. c8-b7 h5-g6 10. b7:d5 g6:d3 11. g4:d7 c7-d6 12. g5-e5 f1-c4 13. c1:a3 h4-g3 14. b3-b5 g3-f2 15. b6-b4 f2-f3 16. a3:d3 d6:b4 17. d7-d2 b4:d2 18. e6-e3 a2-b1 19. d1:b1 c2-b3 20. e5:e2 f3:d5 21. b5:d5 d2-c3++ Olympiad-6 Mona vs YL 0-1 1. d1-b3 a5-c7 2. b1-b4 a7-d7 3. g1-g3 h3-e6 4. g3:c7 h7-e7 5. f1-d3 a6-d6 6. c1-e3 h4-e4 7. f8:h6 a3-a6 8. b4:e4 h5-f7 9. c8:a6 d7:d3 10. a6-c4 f7-f6 11. g8-c8 e7:c7 12. c8:e6 a2-c2 13. e8-c6 a4:c6 14. e1-e5 h2-h4 15. h6:h4 c6:e4 16. h4:e4 c7:c4 17. e3-d4 f6:d8 18. e5-d5 d8-c7 19. d5-c5 c7-d7 20. b8-c8 d7-c6 21. b3:d3 c2-b2 22. d4-b6 d6-e5 23. Resigns Olympiad-7 MIA vs Mona 0-1 1. f8-f6 a5-c7 2. d8-g5 a7-d7 3. c1-f4 a4-c2 4. e1-g3 a2-d2 5. f6-f3 h2-e2 6. g8-g4 a6-d3 7. b1-b3 h7-e7 8. e8-g6 h3:f1 9. g6:d3 d7:d3 10. g1-f2 h4-h1 11. c8-b7 h1-e1 12. g5-e5 a3-c5 13. b8-b5 e1:e5 14. b7-b4 e5-d4 15. b3-c4 c7:f4 16. g3:d3 h6-e3 17. b4:d2 e7-e4 18. g4-g3 h5-h4 19. f3-h1 c5-b4 20. h1-f3 b4:d2 21. b5-b4 h4-g5++ Olympiad-8 Mona vs MIA 1-0 1. d1-b3 h3-e3 2. e1-g3 a7-d4 3. b1-b4 a2-c2 4. g1-g4 a5-c5 5. g8-g5 a4-c6 6. e8:c6 a6-c4 7. b8-b5 c2-e2 8. f1-d1 d4-e5 9. g4:c4 a3:c1 10. f8:c5 c1:g5 11. b3:e3 h7-g6 12. d8:g5 h2-f2 13. c5:f2 h5-f7 14. g3:g6 e2-g4 15. f2-f4 h4-g3 16. b5:e5 g3:e3 17. d1-f3 e3:e5 18. f3-d5 h6-h5 19. g6-f5 h5-h6 20. f5-e4 h6-f6 21. c8-c5++ Olympiad-9 MIA vs Mona 1-0 1. d1-b3 h4-f2 2. g8-d5 a5:d5 3. b1-b4 h2-e2 4. f8-c5 a7-c7 5. b8-b5 h7-f7 6. c8:a6 f7:b3 7. a6:e2 f2:c5 8. f1-c4 h3-g4 9. e1-e4 a3:c1 10. g1-e1 c1:e1 11. e8-c8 h6-g6 12. e2-e5 c7:e5 13. b4-b7 a2-b1 14. b7:b3 b1:e4 15. d8-b8 c5-b6 16. c8-c6 g6-d6 17. b8-a7 g4:c4 18. b3:b6++ Olympiad-10 Mona vs MIA 1-0 1. g1-g3 a3-d3 2. d8-b6 a4-c2 3. c8-c5 a2-d2 4. c5:c2 a5-c5 5. c2:c5 a7-c7 6. b8-b5 a6-a7 7. f1-f3 a7-d7 8. g8-f7 d3-g6 9. e1:h4 c7-f4 10. f8:f4 d2-d5 11. f7:d5 d7-f7 12. d1-g4 h7-g8 13. e8:g8 h5:d5 14. f3:d5 h2-g2 15. b1-d3 g2-f2 16. c5-f5 f2-e2 17. d3:g6 e2:g4 18. b6-e6 h3-g2 19. g8-h7 g2-e4 20. b5-e5 f7:h7 21. e6-c4 h6-g7 22. c1-e3++ Olympiad-11 YL vs Mona 1-0 1. d1-b3 h4-f2 2. b1-b4 h2-e2 3. b8-b5 h7-f7 4. b3:f7 h3:f1 5. f8-c5 f1:b5 6. e8:b5 e2:b5 7. f7-c4 a7-b7 8. d8-b6 h6-f4 9. c1-f1 h5-h4 10. b4-e7 b5-d7 11. b6:f2 h4-f6 12. g1-d4 f6:d4 13. c5:a5 a6-b6 14. c8-c6 b7-b5 15. g8-d5 d7-b7 16. e1-e3 f4-e5 17. a5-b4 b6-a5 18. f1-e1 a2-a6 19. c6-f3 a5-b6 20. f3-c6 b6-a5 21. c6-f3 a5-b6 22. f3-c6 b6-a5 23. c6-f3 b7:b4 24. e1:b4 b5-b7 25. f2-e2 a5:d5 Mona loses on time (White mate in 6 moves) Olympiad-12 Mona vs YL 0-1 1. g1-g3 a4-c2 2. b8-b6 h2-e2 3. c8-c5 a2-d2 4. c5:c2 a7-c7 5. e1:h4 a5-c3 6. h4-g4 a6-d3 7. g8-g5 a3:c1 8. g5-e5 e2:e5 9. b6-c5 h3:f1 10. d8-d4 f1-f3 11. b1-a2 c1-f4 12. g3-g5 c7-e7 13. f8-d8 e7:c5 14. g4-g6 f4:d4 15. a2:d2 f3-f4 16. Resigns How to interpret the output --------------------------- Each game was analyzed by both YL and Mona for 15 minutes per move. In each case, the principle variation (PV) and other interesting lines for the last five or six iterations of the search are given. Some of these lines may be slightly in error, or incomplete, due to the way the programs keep track of the PV. The last four moves of each line have been deleted (unless it is mate), since they are usually not very meaningful. In competition, Mona iterates only on odd plies. These moves tend to be more reliable, more aggressive, and benefit from the special analysis for connection threats. This commentary gives both even and odd plies, plus each improvement to the PV, in the hope that some interesting side lines will also be shown. A program might not agree with the move it actually played, because the tournament time control was 30 minutes each for the entire game. This resulted in search times from 20 seconds to 3 minutes, instead of the 15 minutes allocated for analysis of each move. The moves given here will generally be better than those actually played. Regarding the first move of the game, the current versions of Mona and YL believe 1. d1-b3 and 1. b1-b3 are the two best opening moves (along with symmetrical moves, of course), followed by 1. c1-c3. YL tends to do better with b1-b3, because it leads to sharper positions where YL's deeper search is advantageous. Conversely, Mona tends to do better with d1-b3, due to her better understanding of blockading positions. MIA prefers c1-c3, and more centrally-oriented moves in general, as did Mona and YL prior to the addition of blocking knowledge. Other opening moves are not considered to be as strong, but are still quite playable, and can lead to interesting positions. A note about the minimax scores: Mona and YL use different scales, so their scores should not be compared directly. A score of about +100 by YL or +300 by Mona usually indicates a clear advantage. There is a significant odd-even effect, so odd-ply values are optimistic (for Mona, too high by at least 50 points). Mona indicates a mate threat with a +1000 bonus, so a score of +3150 indicates a small positional advantage and a triple threat to connect. Only certain types of mate threats are detected statically, so there could be more in a given position. Mona reserves 0 (zero) for a draw and +32000 for a win. YL uses +30000 for a win, minus the number of moves to mate, so a score of -29988 means a loss in 12-ply.