I toast to LeelaqueenOdds and its creators. I raise my wine glass high and shout: Long live the queen. We have a queen!!

[lkaufman]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

[chrisw]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Concept of strongest move doesn’t require an absolute accuracy of evaluation function, it requires a relative one, the idea being to rank the moves in order (best on top), so there will be a “strongest” move in any position, winning or not. Queen down or not, an evaluation (even the simplest bean counter) is going to be able to perform relative ranking and thus propose “strongest move”.

[Father]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Concept of strongest move doesn’t require an absolute accuracy of evaluation function, it requires a relative one, the idea being to rank the moves in order (best on top), so there will be a “strongest” move in any position, winning or not. Queen down or not, an evaluation (even the simplest bean counter) is going to be able to perform relative ranking and thus propose “strongest move”.

...by playing against chess machines, little by little, step by step, I become more or less familiar with "the way machines think." Something like an adaptation to the algorithm's personality. And in this way, from my purely empirical perspective, I could say that there are more or less predictable personalities. And when considering "LeelaQueenOdds's thinking profile," I find her fundamentally unpredictable. When I have studied Bobbie Fischer's games, I find them perfectly rectilinear in their prodigious depth, but very predictable. On the contrary, the robot LeelaQueenOdds seems to have "hidden aces or cards" with a marked tendency to psychologically complicate us humans in what I have been learning about time controls of one minute without time increment.

[Father]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Concept of strongest move doesn’t require an absolute accuracy of evaluation function, it requires a relative one, the idea being to rank the moves in order (best on top), so there will be a “strongest” move in any position, winning or not. Queen down or not, an evaluation (even the simplest bean counter) is going to be able to perform relative ranking and thus propose “strongest move”.

...by playing against chess machines, little by little, step by step, I become more or less familiar with "the way machines think." Something like an adaptation to the algorithm's personality. And in this way, from my purely empirical perspective, I could say that there are more or less predictable personalities. And when considering "LeelaQueenOdds's thinking profile," I find her fundamentally unpredictable. When I have studied Bobbie Fischer's games, I find them perfectly rectilinear in their prodigious depth, but very predictable. On the contrary, the robot LeelaQueenOdds seems to have "hidden aces or cards" with a marked tendency to psychologically complicate us humans in what I have been learning about time controls of one minute without time increment.

Good evening, Mr. Larry Kaufman. I've been watching the leaderboard where the results of the top 100 humans facing the LeelaQueenOdds bot appear. It's truly evident that more and more strong competitors are emerging every day. The leaderboard is becoming something like a stable where only the best horses are present. The competition is getting tighter and tighter, and even the last few at the top of the leaderboard are often equally strong performers. For my part, I will continue to strive for a good position in the race results. Ratings of 2700 or higher are becoming somewhat more common. The leader, "SuoerEsay," has a steady, devastating stride, with great consistency. I still haven't ruled out the possibility of climbing the ranks by running and galloping on my little "wooden horse" in the quest to access privileged outposts. I perceive the psychological profile of the computer LeelaQueenOdds as flexible and mutant. I congratulate all the creators and workers who make this robot project a reality.

LeelaQueenOdds Leaderboard
Next update in: 8m 47s
# Player Rating Games Last Game TC
1 SuperEasy 2942 263 2025-04-29 8+3
2 wateenellende 2824 154 2025-03-26 1+0
3 JoanFluvia 2775 363 2025-04-27 2+1
4 Former_Player 2751 288 2025-03-22 3+2
5 verystablegenius69 2732 154 2025-04-29 1+0
6 Orichess88 2704 88 2025-04-03 2+3
7 Alex_Muezersky 2683 179 2025-04-05 1+1
8 dauANHbac 2679 74 2025-04-27 1+1
9 smeets 2672 29 2025-04-02 2+1
10 Hissha 2626 65 2025-04-05 1+10
11 Catecan 2609 2770 2025-04-29 1+0
12 edgelorde 2607 784 2025-04-28 1+1
13 Iron_jack 2586 120 2025-04-22 6+2
14 pozvonochek 2583 187 2025-03-13 1+0
15 Phara0h1 2577 134 2025-04-23 4+0
16 jtcoach 2573 185 2025-03-22 3+2
17 pultis12 2551 259 2025-04-02 1+0
18 Mikulas 2550 124 2025-04-26 3+2
19 FinalSpot 2537 125 2025-03-29 3+2
20 arturchix 2530 38 2025-03-27 2+1
21 Paulewski 2526 159 2025-04-28 5+0
22 Andrey_Smaev_1998 2521 222 2025-04-24 7+7
23 eelke 2507 124 2025-04-17 1+1
24 monkeyking123 2501 126 2025-03-30 1+0
25 davidb2 2485 127 2025-03-15 1+0
26 gotschi 2476 835 2025-04-28 2+2
27 VTChess 2476 1130 2025-04-29 1+1
28 ITakeTrenPassant 2472 200 2025-03-27 3+1
29 BT_CHESS42 2469 274 2025-04-20 3+2
30 vlahi 2437 30 2025-03-30 3+0
31 shreckm8 2437 429 2025-04-04 3+0
32 kristie 2422 143 2025-03-28 1+0
33 quetag 2419 142 2025-04-28 6+3
34 Mykingislonely 2415 277 2025-04-03 2+0
35 CoolerMove 2414 37 2025-03-24 9+10
36 LeafTree 2412 2373 2025-04-28 10+0
37 tasknight 2410 469 2025-04-10 2+2
38 AsheliaXII 2408 147 2025-04-26 15+10
39 behemothbeer 2408 485 2025-04-20 5+3
40 Spinaltap 2408 52 2025-03-23 3+0
41 garyhat 2408 135 2025-03-27 3+0
42 zerture 2406 259 2025-04-28 7+3
43 turton 2404 63 2025-04-29 4+6
44 MiguelPetroff 2400 149 2025-04-27 3+2
45 Reuseup 2387 340 2025-04-18 5+5
46 Sersemlet 2383 138 2025-02-27 3+2
47 astrozilla 2383 340 2025-04-22 3+2
48 xmyy080909 2382 301 2025-04-02 1+1
49 ioste29 2381 61 2025-03-19 3+1
50 karloballa 2376 222 2025-03-30 5+2
51 Litigator94 2371 43 2025-03-08 3+2
52 White_Phoenix 2370 91 2025-03-30 5+3
53 ChessEducation 2359 137 2025-04-24 3+2
54 Switchitter 2357 127 2025-04-23 3+1
55 benkonian 2354 108 2025-03-30 1+1
56 Chillkroete77 2352 112 2025-04-04 3+2
57 vilius100 2345 87 2025-04-28 3+2
58 edgebottle 2344 1337 2025-04-29 5+5
59 jflup 2342 34 2025-04-28 3+1
60 Somelier 2340 30 2025-03-18 2+2
61 nauthnim 2338 471 2025-03-31 3+2
62 dritharashtrar 2338 42 2025-03-17 3+1
63 Dalnor 2337 431 2025-04-28 9+3
64 grandmastergauri 2333 74 2025-03-23 1+1
65 resb 2330 62 2025-04-26 5+3
66 Zaxar2010 2325 44 2025-04-28 5+3
67 Kelvandil 2324 84 2025-03-24 3+2
68 Fumitoks 2323 1166 2025-04-28 1+0
69 Chrisinc 2322 29 2025-04-14 15+10
70 Inventor_1 2319 35 2025-04-16 5+17
71 sergo786 2313 437 2025-04-18 5+3
72 Finnish_wall 2311 55 2025-04-23 8+2
73 KnightVision88 2309 90 2025-04-19 3+2
74 Srive 2303 62 2025-03-20 5+3
75 ritang 2301 425 2025-04-25 3+2
76 Carobee 2301 30 2025-04-10 1+2
77 Red_Aruba 2300 142 2025-04-18 3+5
78 so197371so 2294 263 2025-04-12 3+2
79 Serpent45 2293 111 2025-04-28 3+2
80 pressive 2291 114 2025-04-01 1+0
81 Drviak1 2287 109 2025-03-22 3+2
82 serg4000 2283 145 2025-03-22 4+3
83 JudahH 2275 612 2025-04-24 9+0
84 DerDickePole 2274 36 2025-04-26 5+0
85 King_Lenny 2274 341 2025-04-26 1+0
86 Ajktulm 2274 157 2025-04-16 5+5
87 Mascotechess 2273 55 2025-04-18 8+2
88 Thesphinx 2272 30 2025-04-28 10+5
89 RanchoWangdu 2270 587 2025-04-27 2+1
90 serg4001 2263 72 2025-04-28 5+3
91 AdmiralHabibi 2263 72 2025-04-01 1+0
92 rowrulz 2261 18 2025-04-11 5+2
93 Brutukseni 2259 243 2025-04-18 5+3
94 TheBigGreek 2258 90 2025-03-27 1+1
95 chesseb24 2253 17 2025-03-09 25+20
96 Replicadoe 2252 111 2025-03-28 2+2
97 socoban 2247 56 2025-04-13 7+0
98 Kaidan404 2247 113 2025-03-20 2+2
99 HVC135 2246 122 2025-04-28 1+1
100 andrewmoron 2239 22 2025-03-16 11+0

[Father]

Correction..." The leader, "SuperEasy,"..: GM JOEL BENJAMIN. !! Without a doubt, in addition to being a three-time United States champion, he's a true specialist at playing against chess computers. Congratulations, GM. Applause from me!

" LeelaQueenOdds Leaderboard
Next update in: 8m 47s
# Player Rating Games Last Game TC
1 SuperEasy 2942 263 2025-04-29 8+3
2 wateenellende 2824 154 2025-03-26 1+0
3 JoanFluvia 2775 363 2025-04-27 2+1
4 Former_Player 2751 288 2025-03-22 3+2
5 verystablegenius69 2732 154 2025-04-29 1+0
6 Orichess88 2704 88 2025-04-03 2+3
7 Alex_Muezersky 2683 179 2025-04-05 1+1
8 dauANHbac 2679 74 2025-04-27 1+1
9 smeets 2672 29 2025-04-02 2+1
10 Hissha 2626 65 2025-04-05 1+10
11 Catecan 2609 2770 2025-04-29 1+0
12 edgelorde 2607 784 2025-04-28 1+1
..."

[Father]

Correction..." The leader, "SuperEasy,"..: GM JOEL BENJAMIN. !! Without a doubt, in addition to being a three-time United States champion, he's a true specialist at playing against chess computers. Congratulations, GM. Applause from me!

" LeelaQueenOdds Leaderboard
Next update in: 8m 47s
# Player Rating Games Last Game TC
1 SuperEasy 2942 263 2025-04-29 8+3
2 wateenellende 2824 154 2025-03-26 1+0
3 JoanFluvia 2775 363 2025-04-27 2+1
4 Former_Player 2751 288 2025-03-22 3+2
5 verystablegenius69 2732 154 2025-04-29 1+0
6 Orichess88 2704 88 2025-04-03 2+3
7 Alex_Muezersky 2683 179 2025-04-05 1+1
8 dauANHbac 2679 74 2025-04-27 1+1
9 smeets 2672 29 2025-04-02 2+1
10 Hissha 2626 65 2025-04-05 1+10
11 Catecan 2609 2770 2025-04-29 1+0
12 edgelorde 2607 784 2025-04-28 1+1
..."

..Good evening, Mr. Larry Kaufman. This game I just played against the robot, which I'm about to publish, represents an important step on the ladder to heaven in my existence: In this ecosystem of man versus machine, I managed to climb to number 9 in the standings. At a time when there are quite a few very strong players, my little wooden horse has battled on the courts against top-tier national exponents. I thank my programmer from heaven and the Leela team again.

... [pgn][Event "Casual bullet game"]
[Site "https://lichess.org/CvPaeSHu"]
[Date "2025.05.03"]
[White "Catecan"]
[Black "LeelaQueenOdds"]
[Result "1/2-1/2"]
[GameId "CvPaeSHu"]
[UTCDate "2025.05.03"]
[UTCTime "01:31:47"]
[WhiteElo "2030"]
[BlackElo "2000"]
[BlackTitle "BOT"]
[Variant "From Position"]
[TimeControl "60+0"]
[ECO "?"]
[Opening "?"]
[Termination "Normal"]
[FEN "rnb1kbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1"]
[SetUp "1"]
[Annotator "lichess.org"]

1. d4 f5 2. f4 e6 3. e3 g5 4. Nf3 g4 5. Nh4 b6 6. g3 Nc6 7. Bg2 Ba6 8. Bf1 Bb7 9. Bg2 Ba6 10. Bf1 Bb7 11. Bg2 Be7 12. O-O Bxh4 13. gxh4 Nf6 14. Bd2 Rg8 15. Be1 Ba6 16. Bg3 d5 17. Re1 Kf7 18. Bf1 Bxf1 19. Rxf1 Ne7 20. Rf2 c5 21. Rg2 c4 22. c3 b5 23. a3 a5 24. Qe2 a4 25. Nd2 h5 26. Nf1 Ke8 27. Rc1 Ne4 28. Rc2 Kd7 29. Kh1 Rac8 30. Kg1 Nd6 31. Kh1 Ne4 32. Kg1 Nd6 33. Kh1 Rg7 34. Kg1 Rd8 35. Kh1 Rdg8 36. Kg1 Ra8 37. Kh1 Rag8 38. Kg1 Rd8 39. Kh1 Ra8 40. Kg1 Rh8 41. Kh1 Rb8 42. Kg1 Rc8 43. Kh1 Ra8 44. Kg1 Rh8 45. Kh1 Rb8 46. Kg1 Rc8 47. Kh1 Rh8 48. Kg1 Rb8 49. Kh1 Rc8 50. Kg1 Ra8 51. Kh1 Rh8 52. Kg1 Rb8 53. Kh1 Rh7 54. Kg1 Rd8 55. Kh1 Rdh8 56. Kg1 Re8 57. Kh1 Rb8 58. Kg1 Rg8 59. Kh1 Rf7 60. Kg1 Rh7 61. Kh1 Rd8 62. Kg1 Rdh8 63. Kh1 Kc8 64. Kg1 Rg8 65. Kh1 Kd7 66. Kg1 Kc7 67. Kh1 Kb7 68. Kg1 Kc8 69. Kh1 Kd7 70. Qe1 Kc6 71. Kg1 Re8 72. Kh1 Rg8 73. Kg1 Rf8 74. Kh1 Ndc8 75. Kg1 Nd6 { The game is a draw. } 1/2-1/2[/pgn]

... LeelaQueenOdds Leaderboard
Next update in: 9m 05s
# Player Rating Games Last Game TC
1 SuperEasy 2956 285 2025-05-02 8+3
2 verystablegenius69 2887 365 2025-05-01 1+0
3 wateenellende 2824 154 2025-03-26 1+0
4 JoanFluvia 2774 424 2025-05-02 2+1
5 Former_Player 2751 288 2025-03-22 3+2
6 Orichess88 2704 88 2025-04-03 2+3
7 Alex_Muezersky 2683 179 2025-04-05 1+1
8 dauANHbac 2679 74 2025-04-27 1+1
9 Catecan 2673 3077 2025-05-03 1+0
10 smeets 2672 29 2025-04-02 2+1
11 edgelorde 2638 924 2025-05-03 1+1
12 Hissha 2626 65 2025-04-05 1+10
13 Iron_jack 2586 120 2025-04-22 6+2
14 pozvonochek 2583 187 2025-03-13 1+0
15 Phara0h1 2577 134 2025-04-23 4+0
16 jtcoach 2573 185 2025-03-22 3+2
17 Mikulas 2562 125 2025-05-02 3+2
18 pultis12 2551 259 2025-04-02 1+0
19 FinalSpot 2537 125 2025-03-29 3+2
20 Paulewski 2532 177 2025-05-01 5+0
21 arturchix 2530 38 2025-03-27 2+1
22 VTChess 2521 1198 2025-05-03 1+1
23 Andrey_Smaev_1998 2521 222 2025-04-24 7+7
24 eelke 2507 124 2025-04-17 1+1
25 gotschi 2501 907 2025-05-02 2+3
26 monkeyking123 2501 126 2025-03-30 1+0
27 davidb2 2485 127 2025-03-15 1+0
28 ITakeTrenPassant 2472 200 2025-03-27 3+1
29 BT_CHESS42 2469 274 2025-04-20 3+2
30 MiguelPetroff 2466 163 2025-05-01 3+2
31 vlahi 2437 30 2025-03-30 3+0
32 shreckm8 2437 429 2025-04-04 3+0
33 kristie 2422 143 2025-03-28 1+0
34 LeafTree 2421 2476 2025-04-30 10+0
35 zerture 2416 282 2025-05-02 8+3
36 Mykingislonely 2415 277 2025-04-03 2+0
37 quetag 2414 148 2025-05-01 6+3
38 CoolerMove 2414 37 2025-03-24 9+10
39 AsheliaXII 2413 153 2025-05-02 15+10

[M ANSARI]

The bot is actually quite easy to beat at longer time controls ... from 10 minutes all the way to 3 minutes. But moving the time scale down to 1min + 2s ... and here it just becomes really frustrating. With time running out, it becomes less about Chess and more about speed of making a move ... the BOT plays instantly and while a biological human has a big reaction time. The tactical awareness of an engine is truly incredible and shows how far software AND hardware have improved. I will say that this BOT has really opened my eyes to how rich chess is and how much tactical resources there are in what look like innocuous positions. I certainly think this BOT should be a permanent feature in all the Chess servers. It still is a PIA sometimes to try and get a game with the BOT as it seems to be quite popular and always busy!

I absolutely can see how this BOT can be used as a useful training tool for aspiring chess players. It really makes you much more tactically aware of tricks and traps and in between moves! I think playing many fast games will also build tactical "intuition".

[lkaufman]

The bot is actually quite easy to beat at longer time controls ... from 10 minutes all the way to 3 minutes. But moving the time scale down to 1min + 2s ... and here it just becomes really frustrating. With time running out, it becomes less about Chess and more about speed of making a move ... the BOT plays instantly and while a biological human has a big reaction time. The tactical awareness of an engine is truly incredible and shows how far software AND hardware have improved. I will say that this BOT has really opened my eyes to how rich chess is and how much tactical resources there are in what look like innocuous positions. I certainly think this BOT should be a permanent feature in all the Chess servers. It still is a PIA sometimes to try and get a game with the BOT as it seems to be quite popular and always busy!

I absolutely can see how this BOT can be used as a useful training tool for aspiring chess players. It really makes you much more tactically aware of tricks and traps and in between moves! I think playing many fast games will also build tactical "intuition".

When you say it's easy to beat LeelaQueenOdds at 3 min, I suppose you mean plus increment, since 3'0" is almost as fast as 1'+2". In fact the results since April 1 for LeelaQueenOdds (playing White) at 3'0" are clearly plus for Leela (27 to 19 with 7 draws) against players in the 2600 to 2800 range on LiChess, which roughly corresponds to IM level, and I don't think you claim to be anywhere near IM level. At 3'2" I have to move down a class (to 2400-2600) to get a positive result for Leela (328 to 141 with 59 draws), but that still about CM level on average, so you need at least par FM level to make a plus score at 3'2". Are you FM level, or perhaps you mean something like 3'5", which is of course much easier to win at. Bullet specialist GM Andrew Tang recently scored 2.5-1.5 vs LQO at 1'0", but had lost 3 to 0 earlier so he is still two games down, and even Hikaru is only plus one thanks to taking the White pieces in 3 out of 5 games, just even in the two where Leela played White. So still no one has an overall plus score vs. LQO playing White at 1'0", out of a huge number of GMs who have tried. Tang also tried 1'0" with Rook odds, but lost 21 to 0, not even a draw; he remarked on how the gap between queen odds and rook odds was so enormous. We do have intermediate odds on LeelaPieceOdds, but the bot isn't trained for them yet so it doesn't play them quite as well as queen odds or rook odds.

[Father]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

[lkaufman]

Oh I was very aware that Knight and Rook odds BOTS were out there! What I really didn't quite appreciate is that these BOTS had NN specifically designed to take advantage of human weaknesses when playing Knight or Rook odds. I had thought that this was just a super powerful engine that could play with this handicap. So in essence these BOTS were NOT playing the strongest chess move in the position, but rather the chess move that had the best chance of succeeding against a human. That is something I just did not appreciate the effectiveness of.

Well, there are no "strongest chess moves" in a lost position,

engines with non-infinite search and an evaluation function would beg to differ

unless your definition is the move that postpones the checkmate as long as possible with perfect play. The human definition is the move that gives the best practical chances of saving the game against an imperfect opponent, and this is precisely what the bots do. The imperfect opponent is a bot that is a simulation of a human GM (for the smaller odds) or master (for queen odds), but it is still a bot, not an actual human, so perhaps it's best to think of the Leela bots as optimized against both strong humans and weak engines, ones it can actually draw or beat at the odds.

If "strongest move" depends on an evaluation function, then you are just saying that the strongest move is a matter of opinion, depending on which evaluation function is used. Most evaluation functions are not designed to give accurate evals when someone is a queen down. so these "opinions" will vary greatly and often be worthless.

Good morning, Mr. Larry Kaufman. I would like to know, and therefore I ask, in the list of the 100 humans at the top of the LeelaQueenOdds Robot leaderboard, how many of these humans hold the title of Chess Master in any of its denominations. I would also like to know how many of these non-titled players hold a Master level. I appreciate your kindness in helping me with these questions. I would also like to know what you consider to be the "estimated average" of the 100 humans in the leaderboard. Thank you in advance. I wish you a happy day.

"Chess Master" is not a well-defined term; for FIDE it means FM (Normally reaching 2300 elo), for US it means reaching 2200 national rating, other countries may have varying standards. I think that the only practical definition we can use here is to say that anyone with a LiChess blitz rating over 2400 is at least as strong in blitz as a borderline US national master. Many LiChess players are anonymous, so that's the only practical definition we might use. You would then just look up the LIChess blitz ratings of each of them and call them masters if over 2400. As for average strength, if you average the Lichess blitz ratings and then subtract 200, that is a good estimate of the FIDE rating needed for equivalent skill in blitz.