To get more games and more variety of games in engine match tests, I have started using 3-move ballots. There are 658 of these, but one side immediately loses a man in 164 of them, so 494 are usable as start positions. In a typical match each ballot is played twice so that both sides get to play each ballot as black and as white. I load up the 8-core machine with 8 simultaneous matches of 988 games each, a total of 7904 games, playing a baseline version of kingsrow against a test version, and with the results from 7904 games I can sometimes know if the change I made in the test version is better.
A pdn file containing the 494 usable 3-move ballots can be found here: 3-move ballots.zip
-- Ed
3-move ballots
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: 3-move ballots
I ran 8 matches of 3-move ballots this past weekend to test a search change, with these results.
kingsrow vs. baseline:
+65 -39 =884
+50 -49 =889
+46 -46 =896
+69 -42 =877
+68 -43 =877
+53 -46 =889
+54 -49 =885
+51 -37 =900
Feeding these results into bayeselo I get
Ratingslos
I'm still trying to get familiar with bayeselo. If I'm reading these results correctly, it means that the change produced a 4 elo improvement with a probability of it being a real improvement of 0.87. I had to modify the bayeselo source to be able to input the WLD results directly, as normally it wants to get its input by parsing pgn files.
-- Ed
kingsrow vs. baseline:
+65 -39 =884
+50 -49 =889
+46 -46 =896
+69 -42 =877
+68 -43 =877
+53 -46 =889
+54 -49 =885
+51 -37 =900
Feeding these results into bayeselo I get
Ratings
Code: Select all
Rank Name Elo + - games score oppo. draws
1 kingsrow 2 4 4 7904 51% -2 90%
2 baseline -2 4 4 7904 49% 2 90%Code: Select all
ki ba
kingsrow 87
baseline 12-- Ed
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: 3-move ballots
When I created the list of 3-move openings for engine matches, I removed the ones where one side immediately loses a man. But there are others that might also be losses. For example, 34-29 20-25 29-23. After 18x29 33x24 19x30 35x24, the white voorpost has only 5 defenders vs. 6 black attackers and is probably lost. Does anyone see a way to draw this?
-
Rein Halbersma
- Posts: 1722
- Joined: Wed Apr 14, 2004 16:04
- Contact:
Re: 3-move ballots
What does Kingsrow think? What if you play a 1000 game self-play match from this position?Ed Gilbert wrote:When I created the list of 3-move openings for engine matches, I removed the ones where one side immediately loses a man. But there are others that might also be losses. For example, 34-29 20-25 29-23. After 18x29 33x24 19x30 35x24, the white voorpost has only 5 defenders vs. 6 black attackers and is probably lost. Does anyone see a way to draw this?
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: 3-move ballots
Kingsrow does not find a way to avoid losing the voorpost man. I will remove this one from list of playable openings.
Here is another one that is less clear: 35-30 18-23 32-28. After 23x32 38x27 20-25, white has a weak center and a cramped wing. Below are 8 recent match games in the opening played between kingsrow and flits. Black won 6 of them. These are relatively quick games so that certainly doesn't mean the opening is a white loss. Kingsrow even managed to win one of them as white :-)
-- Ed
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.01.27"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 17-21 4. 40-35 15-20 5. 31-26 10-15 6. 26x17 11x31 7. 36x27 5-10 8. 41-37 12-18 9. 46-41 6-11 10. 33-28 20-24 11. 41-36 7-12 12. 42-38 2-7 13. 39-33 11-17 14. 37-31 17-21 15. 47-42 1-6 16. 44-39 18-23 17. 31-26 7-11 18. 26x17 11x31 19. 36x27 12-17 20. 42-37 8-12 21. 37-31 13-18 22. 31-26 6-11 23. 50-44 15-20 24. 34-29 23x34 25. 27-22 18x27 26. 32x21 16x27 27. 28-23 19x28 28. 33x31 12-18 29. 30x19 14x23 30. 39x30 25x34 31. 44-39 20-25 32. 39x30 25x34 33. 49-44 17-22 34. 31-27 22x31 35. 26x37 10-14 36. 44-39 14-20 37. 39x30 20-25 38. 38-33 25x34 39. 33-29 34-39 40. 43x34 23-28 41. 35-30 11-17 42. 48-42 18-22 43. 29-24 22-27 44. 42-38 9-14 45. 38-33 28x39 46. 34x43 17-21 47. 30-25 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.01.27"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 43-38 14-20 6. 40-35 21-26 7. 41-37 7-12 8. 39-33 20-24 9. 29x20 15x24 10. 32-28 10-14 11. 49-43 5-10 12. 37-32 26x37 13. 42x31 11-17 14. 47-42 17-21 15. 31-26 1-7 16. 26x17 12x21 17. 46-41 8-12 18. 41-37 3-8 19. 44-40 21-26 20. 50-44 7-11 21. 37-31 26x37 22. 42x31 11-17 23. 27-22 18x27 24. 31x11 6x17 25. 32-27 13-18 26. 48-42 9-13 27. 38-32 10-15 28. 36-31 2-7 29. 42-37 4-9 30. 44-39 17-21 31. 31-26 18-23 32. 26x17 12x21 33. 43-38 7-12 34. 37-31 14-20 35. 28-22 21-26 36. 22-18 26x28 37. 33x22 12-17 38. 22x11 13x31 39. 11-6 31-36 40. 38-32 36-41 41. 39-33 41-46 42. 33-29 46x28 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.02"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "1/2-1/2"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 15-20 4. 40-35 12-18 5. 33-28 17-21 6. 39-33 20-24 7. 31-26 10-15 8. 26x17 11x31 9. 36x27 7-12 10. 41-37 6-11 11. 46-41 5-10 12. 43-38 11-17 13. 41-36 2-7 14. 37-31 17-21 15. 31-26 1-6 16. 26x17 12x21 17. 42-37 14-20 18. 47-42 9-14 19. 37-31 8-12 20. 27-22 18x27 21. 31x22 3-8 22. 44-39 21-26 23. 49-43 4-9 24. 50-44 6-11 25. 36-31 26x37 26. 32x41 12-18 27. 42-37 18x27 28. 28-23 19x28 29. 33x31 14-19 30. 39-33 13-18 31. 44-39 9-13 32. 33-28 11-17 33. 31-27 17-21 34. 37-32 10-14 35. 39-33 7-12 36. 41-37 21-26 37. 37-31 26x37 38. 32x41 24-29 39. 33x24 20x40 40. 45x34 14-20 41. 28-22 12-17 42. 22x11 16x7 43. 48-42 19-24 44. 30x19 13x24 45. 34-29 24x33 46. 38x29 20-24 47. 29x20 15x24 48. 41-37 18-23 49. 42-38 25-30 50. 37-32 30-34 51. 43-39 34x43 52. 38x49 23-29 53. 32-28 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.02"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 41-37 7-12 6. 40-35 21-26 7. 43-38 14-20 8. 39-33 10-14 9. 32-28 5-10 10. 37-32 26x37 11. 42x31 11-17 12. 46-41 17-21 13. 41-37 21-26 14. 47-42 20-24 15. 29x20 15x24 16. 49-43 1-7 17. 45-40 14-20 18. 43-39 9-14 19. 48-43 3-9 20. 50-45 18-22 21. 28x17 12x21 22. 33-29 24x33 23. 38x29 10-15 24. 30-24 19x30 25. 35x24 14-19 26. 40-35 19x30 27. 35x24 13-19 28. 24x13 8x19 29. 42-38 19-24 30. 38-33 6-11 31. 45-40 9-13 32. 33-28 24x22 33. 27x9 4x13 34. 39-33 11-17 35. 31-27 7-12 36. 33-28 12-18 37. 43-38 13-19 38. 44-39 17-22 39. 28x17 21x12 40. 32-28 18-23 41. 37-32 12-17 42. 27-22 16-21 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.08"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 17-21 4. 40-35 12-18 5. 31-26 14-20 6. 26x17 11x31 7. 36x27 6-11 8. 41-37 10-14 9. 33-28 20-24 10. 42-38 7-12 11. 46-41 11-17 12. 39-33 5-10 13. 41-36 17-21 14. 37-31 1-6 15. 31-26 2-7 16. 26x17 12x21 17. 47-42 15-20 18. 42-37 7-12 19. 37-31 21-26 20. 44-39 26x37 21. 32x41 18-23 22. 38-32 13-18 23. 43-38 9-13 24. 41-37 6-11 25. 36-31 11-17 26. 49-43 17-21 27. 31-26 4-9 28. 26x17 12x21 29. 34-29 23x34 30. 28-22 18-23 31. 33-28 10-15 32. 39-33 34-40 33. 45x34 24-29 34. 33x24 20x40 35. 35x44 25x34 36. 44-39 34-40 37. 50-45 14-20 38. 45x34 20-24 39. 39-33 9-14 40. 34-29 23x34 41. 43-39 34x43 42. 38x49 15-20 43. 37-31 21-26 44. 49-44 26x37 45. 32x41 8-12 46. 41-37 20-25 47. 37-31 24-30 48. 31-26 14-20 49. 33-29 20-24 50. 29x20 25x14 51. 26-21 14-20 52. 22-17 30-35 53. 17x8 3x12 54. 28-22 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.08"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 41-37 7-12 6. 43-38 14-20 7. 40-35 21-26 8. 39-33 10-14 9. 32-28 5-10 10. 37-32 26x37 11. 42x31 11-17 12. 47-42 20-24 13. 29x20 15x24 14. 46-41 17-21 15. 31-26 1-7 16. 26x17 12x21 17. 49-43 8-12 18. 41-37 3-8 19. 37-31 21-26 20. 42-37 7-11 21. 27-22 18x27 22. 31x22 13-18 23. 22x13 9x18 24. 32-27 12-17 25. 37-32 17-21 26. 48-42 4-9 27. 42-37 11-17 28. 44-39 18-23 29. 34-29 23x34 30. 50-44 8-12 31. 27-22 6-11 32. 44-40 14-20 33. 40x29 25x23 34. 22-18 2-8 35. 18x29 20-25 36. 29x20 25x14 37. 45-40 19-23 38. 28x19 14x23 39. 33-28 9-14 40. 28x19 14x23 41. 40-34 21-27 42. 32x21 16x27 43. 34-29 23x34 44. 39x30 17-22 45. 30-24 12-18 46. 37-32 27-31 47. 36x27 22x31 48. 32-27 31x22 49. 24-20 26-31 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.17"]
[White "Kingsrow 1.52"]
[Black "Flits"]
[Result "2-0"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 15-20 4. 40-35 12-18 5. 33-28 17-21 6. 39-33 7-12 7. 31-26 1-7 8. 26x17 11x31 9. 36x27 12-17 10. 41-37 17-21 11. 46-41 20-24 12. 42-38 7-11 13. 47-42 10-15 14. 44-39 21-26 15. 28-22 8-12 16. 33-28 14-20 17. 41-36 2-8 18. 50-44 18-23 19. 44-40 12-18 20. 39-33 5-10 21. 49-44 10-14 22. 44-39 4-10 23. 37-31 26x37 24. 42x31 8-12 25. 48-42 11-17 26. 22x11 16x7 27. 31-26 7-11 28. 27-22 18x27 29. 32x21 23x32 30. 38x27 13-18 31. 33-28 11-16 32. 27-22 18x27 33. 21x32 6-11 34. 42-38 12-18 35. 36-31 24-29 36. 34x12 25x34 37. 39x30 11-17 38. 12x21 16x36 39. 26-21 36-41 40. 21-16 20-25 41. 40-34 19-23 42. 28x19 14x23 43. 16-11 41-46 44. 32-28 23x32 45. 38x27 46-19 46. 11-7 19-2 47. 7-1 2-16 48. 27-22 16x49 49. 22-17 10-14 50. 17-11 9-13 51. 11-6 3-8 52. 1-7 8-12 53. 7x20 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.17"]
[White "Flits"]
[Black "Kingsrow 1.52"]
[Result "0-2"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 40-35 21x32 5. 37x28 14-20 6. 39-33 20-24 7. 29x20 15x24 8. 41-37 10-14 9. 31-27 11-17 10. 37-32 12-18 11. 43-38 7-12 12. 36-31 5-10 13. 46-41 1-7 14. 41-36 17-21 15. 42-37 10-15 16. 47-42 21-26 17. 27-22 18x27 18. 31x22 13-18 19. 22x13 9x18 20. 49-43 4-9 21. 44-40 14-20 22. 37-31 26x37 23. 32x41 9-13 24. 38-32 12-17 25. 41-37 7-11 26. 43-38 17-21 27. 37-31 11-17 28. 31-27 3-9 29. 42-37 18-22 30. 27x18 13x22 31. 37-31 8-12 32. 31-27 22x31 33. 36x27 12-18 34. 48-42 2-8 35. 42-37 18-23 36. 50-44 21-26 37. 44-39 17-21 38. 34-29 23x43 39. 38x49 25x34 40. 40x29 20-25 41. 29x20 15x24 42. 49-44 9-13 43. 44-40 8-12 44. 40-34 12-18 45. 34-29 18-22 *
Here is another one that is less clear: 35-30 18-23 32-28. After 23x32 38x27 20-25, white has a weak center and a cramped wing. Below are 8 recent match games in the opening played between kingsrow and flits. Black won 6 of them. These are relatively quick games so that certainly doesn't mean the opening is a white loss. Kingsrow even managed to win one of them as white :-)
-- Ed
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.01.27"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 17-21 4. 40-35 15-20 5. 31-26 10-15 6. 26x17 11x31 7. 36x27 5-10 8. 41-37 12-18 9. 46-41 6-11 10. 33-28 20-24 11. 41-36 7-12 12. 42-38 2-7 13. 39-33 11-17 14. 37-31 17-21 15. 47-42 1-6 16. 44-39 18-23 17. 31-26 7-11 18. 26x17 11x31 19. 36x27 12-17 20. 42-37 8-12 21. 37-31 13-18 22. 31-26 6-11 23. 50-44 15-20 24. 34-29 23x34 25. 27-22 18x27 26. 32x21 16x27 27. 28-23 19x28 28. 33x31 12-18 29. 30x19 14x23 30. 39x30 25x34 31. 44-39 20-25 32. 39x30 25x34 33. 49-44 17-22 34. 31-27 22x31 35. 26x37 10-14 36. 44-39 14-20 37. 39x30 20-25 38. 38-33 25x34 39. 33-29 34-39 40. 43x34 23-28 41. 35-30 11-17 42. 48-42 18-22 43. 29-24 22-27 44. 42-38 9-14 45. 38-33 28x39 46. 34x43 17-21 47. 30-25 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.01.27"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 43-38 14-20 6. 40-35 21-26 7. 41-37 7-12 8. 39-33 20-24 9. 29x20 15x24 10. 32-28 10-14 11. 49-43 5-10 12. 37-32 26x37 13. 42x31 11-17 14. 47-42 17-21 15. 31-26 1-7 16. 26x17 12x21 17. 46-41 8-12 18. 41-37 3-8 19. 44-40 21-26 20. 50-44 7-11 21. 37-31 26x37 22. 42x31 11-17 23. 27-22 18x27 24. 31x11 6x17 25. 32-27 13-18 26. 48-42 9-13 27. 38-32 10-15 28. 36-31 2-7 29. 42-37 4-9 30. 44-39 17-21 31. 31-26 18-23 32. 26x17 12x21 33. 43-38 7-12 34. 37-31 14-20 35. 28-22 21-26 36. 22-18 26x28 37. 33x22 12-17 38. 22x11 13x31 39. 11-6 31-36 40. 38-32 36-41 41. 39-33 41-46 42. 33-29 46x28 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.02"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "1/2-1/2"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 15-20 4. 40-35 12-18 5. 33-28 17-21 6. 39-33 20-24 7. 31-26 10-15 8. 26x17 11x31 9. 36x27 7-12 10. 41-37 6-11 11. 46-41 5-10 12. 43-38 11-17 13. 41-36 2-7 14. 37-31 17-21 15. 31-26 1-6 16. 26x17 12x21 17. 42-37 14-20 18. 47-42 9-14 19. 37-31 8-12 20. 27-22 18x27 21. 31x22 3-8 22. 44-39 21-26 23. 49-43 4-9 24. 50-44 6-11 25. 36-31 26x37 26. 32x41 12-18 27. 42-37 18x27 28. 28-23 19x28 29. 33x31 14-19 30. 39-33 13-18 31. 44-39 9-13 32. 33-28 11-17 33. 31-27 17-21 34. 37-32 10-14 35. 39-33 7-12 36. 41-37 21-26 37. 37-31 26x37 38. 32x41 24-29 39. 33x24 20x40 40. 45x34 14-20 41. 28-22 12-17 42. 22x11 16x7 43. 48-42 19-24 44. 30x19 13x24 45. 34-29 24x33 46. 38x29 20-24 47. 29x20 15x24 48. 41-37 18-23 49. 42-38 25-30 50. 37-32 30-34 51. 43-39 34x43 52. 38x49 23-29 53. 32-28 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.02"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 41-37 7-12 6. 40-35 21-26 7. 43-38 14-20 8. 39-33 10-14 9. 32-28 5-10 10. 37-32 26x37 11. 42x31 11-17 12. 46-41 17-21 13. 41-37 21-26 14. 47-42 20-24 15. 29x20 15x24 16. 49-43 1-7 17. 45-40 14-20 18. 43-39 9-14 19. 48-43 3-9 20. 50-45 18-22 21. 28x17 12x21 22. 33-29 24x33 23. 38x29 10-15 24. 30-24 19x30 25. 35x24 14-19 26. 40-35 19x30 27. 35x24 13-19 28. 24x13 8x19 29. 42-38 19-24 30. 38-33 6-11 31. 45-40 9-13 32. 33-28 24x22 33. 27x9 4x13 34. 39-33 11-17 35. 31-27 7-12 36. 33-28 12-18 37. 43-38 13-19 38. 44-39 17-22 39. 28x17 21x12 40. 32-28 18-23 41. 37-32 12-17 42. 27-22 16-21 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.08"]
[White "Kingsrow 1.51c"]
[Black "Flits"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 17-21 4. 40-35 12-18 5. 31-26 14-20 6. 26x17 11x31 7. 36x27 6-11 8. 41-37 10-14 9. 33-28 20-24 10. 42-38 7-12 11. 46-41 11-17 12. 39-33 5-10 13. 41-36 17-21 14. 37-31 1-6 15. 31-26 2-7 16. 26x17 12x21 17. 47-42 15-20 18. 42-37 7-12 19. 37-31 21-26 20. 44-39 26x37 21. 32x41 18-23 22. 38-32 13-18 23. 43-38 9-13 24. 41-37 6-11 25. 36-31 11-17 26. 49-43 17-21 27. 31-26 4-9 28. 26x17 12x21 29. 34-29 23x34 30. 28-22 18-23 31. 33-28 10-15 32. 39-33 34-40 33. 45x34 24-29 34. 33x24 20x40 35. 35x44 25x34 36. 44-39 34-40 37. 50-45 14-20 38. 45x34 20-24 39. 39-33 9-14 40. 34-29 23x34 41. 43-39 34x43 42. 38x49 15-20 43. 37-31 21-26 44. 49-44 26x37 45. 32x41 8-12 46. 41-37 20-25 47. 37-31 24-30 48. 31-26 14-20 49. 33-29 20-24 50. 29x20 25x14 51. 26-21 14-20 52. 22-17 30-35 53. 17x8 3x12 54. 28-22 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.08"]
[White "Flits"]
[Black "Kingsrow 1.51c"]
[Result "0-1"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 37-32 12-18 5. 41-37 7-12 6. 43-38 14-20 7. 40-35 21-26 8. 39-33 10-14 9. 32-28 5-10 10. 37-32 26x37 11. 42x31 11-17 12. 47-42 20-24 13. 29x20 15x24 14. 46-41 17-21 15. 31-26 1-7 16. 26x17 12x21 17. 49-43 8-12 18. 41-37 3-8 19. 37-31 21-26 20. 42-37 7-11 21. 27-22 18x27 22. 31x22 13-18 23. 22x13 9x18 24. 32-27 12-17 25. 37-32 17-21 26. 48-42 4-9 27. 42-37 11-17 28. 44-39 18-23 29. 34-29 23x34 30. 50-44 8-12 31. 27-22 6-11 32. 44-40 14-20 33. 40x29 25x23 34. 22-18 2-8 35. 18x29 20-25 36. 29x20 25x14 37. 45-40 19-23 38. 28x19 14x23 39. 33-28 9-14 40. 28x19 14x23 41. 40-34 21-27 42. 32x21 16x27 43. 34-29 23x34 44. 39x30 17-22 45. 30-24 12-18 46. 37-32 27-31 47. 36x27 22x31 48. 32-27 31x22 49. 24-20 26-31 *
[Event "Game 927, opening 35-30 18-23 32-28"]
[Date "2012.02.17"]
[White "Kingsrow 1.52"]
[Black "Flits"]
[Result "2-0"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 37-32 15-20 4. 40-35 12-18 5. 33-28 17-21 6. 39-33 7-12 7. 31-26 1-7 8. 26x17 11x31 9. 36x27 12-17 10. 41-37 17-21 11. 46-41 20-24 12. 42-38 7-11 13. 47-42 10-15 14. 44-39 21-26 15. 28-22 8-12 16. 33-28 14-20 17. 41-36 2-8 18. 50-44 18-23 19. 44-40 12-18 20. 39-33 5-10 21. 49-44 10-14 22. 44-39 4-10 23. 37-31 26x37 24. 42x31 8-12 25. 48-42 11-17 26. 22x11 16x7 27. 31-26 7-11 28. 27-22 18x27 29. 32x21 23x32 30. 38x27 13-18 31. 33-28 11-16 32. 27-22 18x27 33. 21x32 6-11 34. 42-38 12-18 35. 36-31 24-29 36. 34x12 25x34 37. 39x30 11-17 38. 12x21 16x36 39. 26-21 36-41 40. 21-16 20-25 41. 40-34 19-23 42. 28x19 14x23 43. 16-11 41-46 44. 32-28 23x32 45. 38x27 46-19 46. 11-7 19-2 47. 7-1 2-16 48. 27-22 16x49 49. 22-17 10-14 50. 17-11 9-13 51. 11-6 3-8 52. 1-7 8-12 53. 7x20 *
[Event "Game 928, opening 35-30 18-23 32-28"]
[Date "2012.02.17"]
[White "Flits"]
[Black "Kingsrow 1.52"]
[Result "0-2"]
[FEN "B:W28,30,31,33,34,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50:B1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,23"]
1. ... 23x32 2. 38x27 20-25 3. 33-29 17-21 4. 40-35 21x32 5. 37x28 14-20 6. 39-33 20-24 7. 29x20 15x24 8. 41-37 10-14 9. 31-27 11-17 10. 37-32 12-18 11. 43-38 7-12 12. 36-31 5-10 13. 46-41 1-7 14. 41-36 17-21 15. 42-37 10-15 16. 47-42 21-26 17. 27-22 18x27 18. 31x22 13-18 19. 22x13 9x18 20. 49-43 4-9 21. 44-40 14-20 22. 37-31 26x37 23. 32x41 9-13 24. 38-32 12-17 25. 41-37 7-11 26. 43-38 17-21 27. 37-31 11-17 28. 31-27 3-9 29. 42-37 18-22 30. 27x18 13x22 31. 37-31 8-12 32. 31-27 22x31 33. 36x27 12-18 34. 48-42 2-8 35. 42-37 18-23 36. 50-44 21-26 37. 44-39 17-21 38. 34-29 23x43 39. 38x49 25x34 40. 40x29 20-25 41. 29x20 15x24 42. 49-44 9-13 43. 44-40 8-12 44. 40-34 12-18 45. 34-29 18-22 *
-
Rein Halbersma
- Posts: 1722
- Joined: Wed Apr 14, 2004 16:04
- Contact:
Re: 3-move ballots
It's funny that the variation in Kingsrow's opening book is:Ed Gilbert wrote:Kingsrow does not find a way to avoid losing the voorpost man. I will remove this one from list of playable openings.
Here is another one that is less clear: 35-30 18-23 32-28. After 23x32 38x27 20-25, white has a weak center and a cramped wing. Below are 8 recent match games in the opening played between kingsrow and flits. Black won 6 of them. These are relatively quick games so that certainly doesn't mean the opening is a white loss. Kingsrow even managed to win one of them as white
-- Ed
1. 34-29 20-25 2. 29-23 18x29 3. 33x24 19x30 4. 35x24 14-19 5. 40-35 19x30 6. 35x24 9-14 7. 45-40 14-19 8. 40-35 19x30 9. 35x24 3-9 10. 50-45 9-14 with a propagated eval of +26 for black. However, the score at the end of that line rises to +85 for black within a few seconds of searching. How long were leaf nodes searched when you built the opening book? I agree BTW that the opening is lost.
However, it got me thinking: almost every 2-move ballot can be followed by such a ridiculous looking white advance to the 6th row. E.g. what about 33-28 17-21 28-23? How many of these ballots are actually sound? At first glance, only the example that you gave seems to be an exception to the rule that it is generally playable.
The 35-30 18-23 32-28x27 opening looks funny but it's very unlikely to be lost (perhaps it's not even bad for white). I would doubt if there is any losing 3-ballot opening that does not forcibly lose a man (as with the lost voorpost after 34-29 20-25 29-23?). As long as you keep the material even, any opening should be playable.
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: 3-move ballots
It is difficult to know how that happened. The book searches are for several minutes, but they are "allscores" searches which are much less efficient, and until recently they were not MTD-f searches but something else. Some of the book was built several years ago using older versions of the kingsrow engine. The book generator usually runs 24/7 for months at a time, and sometimes unexpected things happen, like another family member starts using the computer while the book generator is running and does something that is CPU-intensive or RAM-intensive...It's funny that the variation in Kingsrow's opening book is:
1. 34-29 20-25 2. 29-23 18x29 3. 33x24 19x30 4. 35x24 14-19 5. 40-35 19x30 6. 35x24 9-14 7. 45-40 14-19 8. 40-35 19x30 9. 35x24 3-9 10. 50-45 9-14 with a propagated eval of +26 for black. However, the score at the end of that line rises to +85 for black within a few seconds of searching. How long were leaf nodes searched when you built the opening book? I agree BTW that the opening is lost.
-- Ed
Re: 3-move ballots
Based on multiple matches against Kingsrow (and frequently losing the same game, and/or similar positions), I'm now quite sure that I need to improve my Outpost routine.
Especially the outpost on 24 for White or 27 for Black is a programming nightmare.
The base seems to be simple (number of defenders and/or attackers), but then you need to include tempo (so race conditions for defending/attacking), and side combinations are also possible, and blocking an attack and/or defense.
And last but not least sometimes defending an outpost is costing so much material, that in the end your wing is destroyed, yielding breakthrough possibilities.
For me a reason to re-program this routine.
Unfortunately I (most likely) need to code myself, and self play (and or Kingsrow matches), will provide new clues.
And compared with other evaluation routines, this Outpost routine (at least in my case) is really-big, and a nightmare in terms of debugging and/or open for (small) changes.
What I was questioning if there are not better approaches these days.
Computer chess was started as the drosophila for artificial intelligence, unfortunately brute-force and alpha-beta were extremely successful.
So I hope that some new computer learning techniques are applicable, but I doubt.
How do the other programmers approach this dilemma.
From Gerard I understood that he also split dynamic and static evaluation elements.
But I still think this is based on hard-wired coding, and not pattern learning....
Bert
Especially the outpost on 24 for White or 27 for Black is a programming nightmare.
The base seems to be simple (number of defenders and/or attackers), but then you need to include tempo (so race conditions for defending/attacking), and side combinations are also possible, and blocking an attack and/or defense.
And last but not least sometimes defending an outpost is costing so much material, that in the end your wing is destroyed, yielding breakthrough possibilities.
For me a reason to re-program this routine.
Unfortunately I (most likely) need to code myself, and self play (and or Kingsrow matches), will provide new clues.
And compared with other evaluation routines, this Outpost routine (at least in my case) is really-big, and a nightmare in terms of debugging and/or open for (small) changes.
What I was questioning if there are not better approaches these days.
Computer chess was started as the drosophila for artificial intelligence, unfortunately brute-force and alpha-beta were extremely successful.
So I hope that some new computer learning techniques are applicable, but I doubt.
How do the other programmers approach this dilemma.
From Gerard I understood that he also split dynamic and static evaluation elements.
But I still think this is based on hard-wired coding, and not pattern learning....
Bert
-
Rein Halbersma
- Posts: 1722
- Joined: Wed Apr 14, 2004 16:04
- Contact:
Re: 3-move ballots
Bert,BertTuyt wrote:Based on multiple matches against Kingsrow (and frequently losing the same game, and/or similar positions), I'm now quite sure that I need to improve my Outpost routine.
Especially the outpost on 24 for White or 27 for Black is a programming nightmare.
The base seems to be simple (number of defenders and/or attackers), but then you need to include tempo (so race conditions for defending/attacking), and side combinations are also possible, and blocking an attack and/or defense.
And last but not least sometimes defending an outpost is costing so much material, that in the end your wing is destroyed, yielding breakthrough possibilities.
For me a reason to re-program this routine.
Unfortunately I (most likely) need to code myself, and self play (and or Kingsrow matches), will provide new clues.
And compared with other evaluation routines, this Outpost routine (at least in my case) is really-big, and a nightmare in terms of debugging and/or open for (small) changes.
What I was questioning if there are not better approaches these days.
Computer chess was started as the drosophila for artificial intelligence, unfortunately brute-force and alpha-beta were extremely successful.
So I hope that some new computer learning techniques are applicable, but I doubt.
How do the other programmers approach this dilemma.
From Gerard I understood that he also split dynamic and static evaluation elements.
But I still think this is based on hard-wired coding, and not pattern learning....
Bert
Just to illustrate both the necessity and difficulty of outposts, here's an example from a game by Schwarzman (black):
26... 18-23! 27. 28x19 14x34 28. 40x29 10-14 29. 45-40 14-19 30. 40-35 19x30 31. 35x24 9-14 32.
44-40 14-19 33. 40-35 19x30 34. 35x24 5-10 35. 49-44 10-14 36. 44-40
14-19 37. 40-35 19x30 38. 35x24 3-9 39. 50-44 9-14 40. 44-40 14-19 41.
40-35 19x30 42. 35x24 13-19 43. 24x13 8x19 with a db win for black.
It is probably infeasible to come up with search extensions that will play the 14 move (28 ply!) forced sequence from black move 28 till black move 42 before concluding that it was a loss.
But perhaps some sort of static exchange mechanism can be made? E.g. how humans think after 28.40x29 is "Remove pieces 3,5,9 and 10 for black, and pieces 44,45,49 and 50 for white. This doesn't change the tempo difference, so it's black to move then, and play 42...13-19x19". This sort of thinking saves 28 ply of search. But a lot of things could happen in 28 ply to mess up this schematic evaluation!
Rein
-
Ed Gilbert
- Posts: 860
- Joined: Sat Apr 28, 2007 14:53
- Real name: Ed Gilbert
- Location: Morristown, NJ USA
- Contact:
Re: 3-move ballots
Yes, there are several like that one with a white man looking very vulnernable on the 6th rank. The stats from the last 8 games on this one are 2 black wins, 5 draws, and 1 white win. I think you're right that they should all be played unless it is really obvious that they lose a man.However, it got me thinking: almost every 2-move ballot can be followed by such a ridiculous looking white advance to the 6th row. E.g. what about 33-28 17-21 28-23? How many of these ballots are actually sound? At first glance, only the example that you gave seems to be an exception to the rule that it is generally playable.
It turns out that 35-30 18-23 32-28 is also reachable from 32-28 18-23 35-30, as I did not remove openings that are identical through transpositions. So I actually have 16 games of recent history on this. Total stats: black wins 10, draws 4, white wins 2. These stats are quite different from the normal distribution of kingsrow vs. flits where there are typically 20 kingsrow wins for each loss.
-- Ed
Re: 3-move ballots
Rein, you are right.
For whatever reason these positions seems to be logical from a human perspective, but extremely difficult for a computer.
You are right search-extensions might be difficult to implement to find the whole move-sequence.
Another option (which I thought of and experimented long ago), is the super move.
Do the sequence in 2 pseudo moves from white and black, and start calculations from there.
I once tested it for breakthrough race conditions.
For example white could promote in 4 ply, and black a ply later.
For these cases the computer did a super move for both (where to destination square was all the possible squares one could reach in 4 plies), and hereafter started all further calculations.
Sometimes it even seems to work, but as computers got faster, I now rely on a normal search.
Curious how others handle this, these days.....
Bert
For whatever reason these positions seems to be logical from a human perspective, but extremely difficult for a computer.
You are right search-extensions might be difficult to implement to find the whole move-sequence.
Another option (which I thought of and experimented long ago), is the super move.
Do the sequence in 2 pseudo moves from white and black, and start calculations from there.
I once tested it for breakthrough race conditions.
For example white could promote in 4 ply, and black a ply later.
For these cases the computer did a super move for both (where to destination square was all the possible squares one could reach in 4 plies), and hereafter started all further calculations.
Sometimes it even seems to work, but as computers got faster, I now rely on a normal search.
Curious how others handle this, these days.....
Bert
Re: 3-move ballots
So after a hard days night , and 50.000 seconds later at ply 29 my program finds the right move (already somewhat earlier 
) , but still with a relatively small positive score.
Are there other programs out there, who see immediate the 18-23 move with a huge score?
At least after the move sequence , as described by Rein, Damage sees the DB-win.
Bert
) , but still with a relatively small positive score.Are there other programs out there, who see immediate the 18-23 move with a huge score?
At least after the move sequence , as described by Rein, Damage sees the DB-win.
Bert
-
Rein Halbersma
- Posts: 1722
- Joined: Wed Apr 14, 2004 16:04
- Contact:
Re: 3-move ballots
Bert, I am very skeptical if one could ever make patterns to accurately assess such situations. Take first and last positions from the 28 ply forcing of the Wong-Schwarzman game:BertTuyt wrote:So after a hard days night , and 50.000 seconds later at ply 29 my program finds the right move (already somewhat earlier
Are there other programs out there, who see immediate the 18-23 move with a huge score?
At least after the move sequence , as described by Rein, Damage sees the DB-win.
Bert
However, the black win in the second diagram is far from easy. Without databases, it takes another 28 ply to proof it:
1. 43-39 20-24 2. 29x20 15x24 3. 38-32 25-30 4. 41-37 30-34 5. 39x30 24x35 6. 33-29 2-8 7. 29-23 19x28 8. 32x23 8-13 9. 36-31 35-40 10. 31-27 11-17 11. 37-31 40-44 12. 23-18 13x22 13. 27x18 17-22 14. 18x27 44-50 *
But suppose you code a pattern that captures the white right wing weakness, and you indeed test it on the above position that it is a win. Then my question is: how is your program going to judge the following almost identical positions:
Surprise: they are both draws! (This was discovered by Ton Sijbrands in a newspaper column 2 years ago: http://www.volkskrant.nl/vk/nl/2698/Spo ... llen.dhtml)
-
Rein Halbersma
- Posts: 1722
- Joined: Wed Apr 14, 2004 16:04
- Contact:
Re: 3-move ballots
Bert, just another example of how hard outpost pieces are to program:
Raichenbach-Springer, WCh 1937
1. 34-29 17-22 2. 40-34 11-17 3. 45-40 6-11 4. 31-26 20-25 5. 50-45 1-6 6. 37-31 14-20 7. 41-37 20-24 8. 29x20 25x14 9. 46-41 19-23 10. 35-30 22-27 11. 31x22 18x27 12. 32x21 16x27 13. 30-25 14-20 14. 25x14 10x19 15. 33-29 17-22 16. 29x18 12x23 17. 34-29 23x34 18. 40x29 13-18 19. 37-31 9-13
Simple pattern-based program: 4 defenders (2,6,7,11) vs 4 attackers (41,42,47,48): equal?
20.41-37 5-10 21.37-32 11-16 22.32x21 16x27 23.42-37 7-11 24.48-42 2-7 25.37-32 11-16 26.32x21 16x27 27.42-37 7-11 28.47-42 15-20 29.37-32 11-16 30.32x21 16x27 31.42-37 6-11 32.37-32 11-16 33.32x21 16x27
After this 28-ply almost forced move sequence, white came with a completely original move: 34.38-33!! 8-12 35.29-23 18x38 36.43x21 13-18 37.31-27 22x31 38.36x27 3-8 39.39-33 19-23 40.21-16 12-17 41.26-21 17x26 42.16-11 18-22 43.27x29 26-31 with a winning endgame.
However, white managed to spoil the win with 44.33-28? 31-37 45.28-22 37-42 46.22-17 8-12 47.17x8 20-24 48.29x20 10-14 49.20x9 4x2 50.11-6 2-7 51.6-1 7-11 52.44-39 42-47 53.1-12 11-16 54.45-40 47-20 55.12-3 20-42 56.39-34 42-33 57.40-35 33-42 58.34-30 42-33 59.30-25 33-29 60.3-17 29-24 61.49-43 24-29 62.35-30 29-47 63.17-50 16-21 64.43-39 47-41 65.30-24 21-27 66.25-20 27-32 67.20-15 32-37 1-1
The best one could do is to add a pattern that recognizes the lack of the 3/8/12 formation in the first diagram, but it would be almost impossible to extend the 28-ply sequence based on such patterns.
You can read more about this Raichenbach-move in a series of articles by Ton Sijbrands (subscription might be required, but perhaps they are still available on the Media thread in the general forum):
http://www.volkskrant.nl/vk/nl/2844/Arc ... heid.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... bach.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... spel.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... uvre.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... imte.dhtml
http://www.volkskrant.nl/vk/nl/2698/Spo ... nijn.dhtml
Raichenbach-Springer, WCh 1937
1. 34-29 17-22 2. 40-34 11-17 3. 45-40 6-11 4. 31-26 20-25 5. 50-45 1-6 6. 37-31 14-20 7. 41-37 20-24 8. 29x20 25x14 9. 46-41 19-23 10. 35-30 22-27 11. 31x22 18x27 12. 32x21 16x27 13. 30-25 14-20 14. 25x14 10x19 15. 33-29 17-22 16. 29x18 12x23 17. 34-29 23x34 18. 40x29 13-18 19. 37-31 9-13
Simple pattern-based program: 4 defenders (2,6,7,11) vs 4 attackers (41,42,47,48): equal?
20.41-37 5-10 21.37-32 11-16 22.32x21 16x27 23.42-37 7-11 24.48-42 2-7 25.37-32 11-16 26.32x21 16x27 27.42-37 7-11 28.47-42 15-20 29.37-32 11-16 30.32x21 16x27 31.42-37 6-11 32.37-32 11-16 33.32x21 16x27
After this 28-ply almost forced move sequence, white came with a completely original move: 34.38-33!! 8-12 35.29-23 18x38 36.43x21 13-18 37.31-27 22x31 38.36x27 3-8 39.39-33 19-23 40.21-16 12-17 41.26-21 17x26 42.16-11 18-22 43.27x29 26-31 with a winning endgame.
However, white managed to spoil the win with 44.33-28? 31-37 45.28-22 37-42 46.22-17 8-12 47.17x8 20-24 48.29x20 10-14 49.20x9 4x2 50.11-6 2-7 51.6-1 7-11 52.44-39 42-47 53.1-12 11-16 54.45-40 47-20 55.12-3 20-42 56.39-34 42-33 57.40-35 33-42 58.34-30 42-33 59.30-25 33-29 60.3-17 29-24 61.49-43 24-29 62.35-30 29-47 63.17-50 16-21 64.43-39 47-41 65.30-24 21-27 66.25-20 27-32 67.20-15 32-37 1-1
The best one could do is to add a pattern that recognizes the lack of the 3/8/12 formation in the first diagram, but it would be almost impossible to extend the 28-ply sequence based on such patterns.
You can read more about this Raichenbach-move in a series of articles by Ton Sijbrands (subscription might be required, but perhaps they are still available on the Media thread in the general forum):
http://www.volkskrant.nl/vk/nl/2844/Arc ... heid.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... bach.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... spel.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... uvre.dhtml
http://www.volkskrant.nl/vk/nl/2844/Arc ... imte.dhtml
http://www.volkskrant.nl/vk/nl/2698/Spo ... nijn.dhtml
Last edited by Rein Halbersma on Mon Feb 20, 2012 22:45, edited 1 time in total.
Re: 3-move ballots
Rein,
you can take the move sequence as an integral plan from the perspective of black. If next tot that you execute these plan moves first, without changing remaining search-depth, then you will immediately start with the 3 positions as you described.
Then the search will see a win for the first position, so confirming the game plan, for the other 2 positions it will find a draw.
If the initial score is negative from the perspective of black (for position 2 and 3), reason to stick to this plan (as it will give a sure draw).
In case blacks position is slightly better, the program will look for alternatives.
I guess there is some old literature available for computer chess to work with (min)plans.
As shared in an earlier post, all these ideas were abandoned due to the straightforward suc6 of alpha-beta
.
But I might consider it in a next Damage Engine....
Do other engines work with mini-plans ??
Bert
you can take the move sequence as an integral plan from the perspective of black. If next tot that you execute these plan moves first, without changing remaining search-depth, then you will immediately start with the 3 positions as you described.
Then the search will see a win for the first position, so confirming the game plan, for the other 2 positions it will find a draw.
If the initial score is negative from the perspective of black (for position 2 and 3), reason to stick to this plan (as it will give a sure draw).
In case blacks position is slightly better, the program will look for alternatives.
I guess there is some old literature available for computer chess to work with (min)plans.
As shared in an earlier post, all these ideas were abandoned due to the straightforward suc6 of alpha-beta
.But I might consider it in a next Damage Engine....
Do other engines work with mini-plans ??
Bert
