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

[Father]

LeelaQueenOdds is amazingly strong. I always thought that being a queen ahead would be enough to win against god. But I am 1400 Elo and so far I have lost all my games (playing white at TC 15+3). It is amazing how Leela manages to create and exploit all kinds of vulnerabilities, while at the same time maintaining an impenetrable position herself.

…Michel. This paragraph you've written defines, in some way, the reasoned and reasonable criteria with which we chess players of the 18th, 19th, and 20th centuries were born and raised. That is to say, literally, a first-class player, something like 2000 FIDE rating in the 1980s, would bet with absolute certainty that he would win a match with a queen advantage against the then-reigning world champions. But not only in classical mining, even at blitz levels and similar levels with increased time per move. But today, science and scientists have shown us the incredible. The truth is, a single draw by this robot against a master would be "the jewel in the crown." But no! We are itinerantly, and even regularly, observing the punishment that LeelaQueenOdds is inflicting on us all!

[SirSlemp]

I was surprised Former_Player has a 60% score in total. I was spectating his games and he was scoring 20% in the last 10 games. Some games he blundered, but often he got outplayed. Maybe this session he was playing late at night or in a distracted environment.

Edit: must be because in his last 20 games he scored 14-6 (70%). Maybe he was experimenting when I spectated.

[lkaufman]

Today IM Levy Rozman, the most popular chess streamer, played a five game match with LeelaQueenOdds at 5'3", winning by 3 to 2. Yesterday and today LQO is playing about 7000 games per day, maybe comparable to all the other bots on LiChess combined?! Just incredible interest. Meanwhile the knight odds bot defeated another top player, Alexei Sarana, by 12.5 to 0.5 in blitz (3'1"). Levy plans to play rook odds next.

[Father]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

[lkaufman]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

The bot was updated a couple days ago in two small ways: the opening book was updated, and the use of a tiny temperature setting in the endgame (for variety) was eliminated as it sometimes caused missed checkmates. A larger update (for queen odds, queen for knight, and multipiece) may happen in a week or so if tests confirm that it is an upgrade. Meanwhile I did a review of Rapid (10'5" minimum so Rapid by FIDE rules) games by Leela Queen Odds since the Feb27 upgrade, and the break-even Rapid elo seems to be about 2520, which suppedly equates to 2355 FIDE! Even if we allow a bit for some players playing too fast or resigning too soon, it looks like Leela Queen Odds is a fair match for a par FM at FIDE Rapid time controls. Quite incredible.
Just to make sure you don't misunderstand, there is no penalty for draws on the Leaderboard, they are not scored as 0.25 but as 0.5. But the draws are only half-weighted. This is not at all the same. If you are winning and losing equally often, draws won't affect your rating at all. But if you are usually losing, then draws don't help as much as winning half of those games would. I think the effect is to favor using long enough time controls that allow you to make a plus score, where draws won't cost as much as normally. So while the time-adjustment formula may be about right with normal rating rules, it is probably a bit out of line with the draws count for half rule. Basically, the leader will be the player who can consistently win at the lowest time control, which so far is Joel Benjamin at 8' +3". If another player can win almost all the games at a faster time control, he will be the leader. So far only "former_player" has come close, but he seems to play some silly openings sometimes just to make it more challenging, and so doesn't win consistently.

[Father]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

The bot was updated a couple days ago in two small ways: the opening book was updated, and the use of a tiny temperature setting in the endgame (for variety) was eliminated as it sometimes caused missed checkmates. A larger update (for queen odds, queen for knight, and multipiece) may happen in a week or so if tests confirm that it is an upgrade. Meanwhile I did a review of Rapid (10'5" minimum so Rapid by FIDE rules) games by Leela Queen Odds since the Feb27 upgrade, and the break-even Rapid elo seems to be about 2520, which suppedly equates to 2355 FIDE! Even if we allow a bit for some players playing too fast or resigning too soon, it looks like Leela Queen Odds is a fair match for a par FM at FIDE Rapid time controls. Quite incredible.
Just to make sure you don't misunderstand, there is no penalty for draws on the Leaderboard, they are not scored as 0.25 but as 0.5. But the draws are only half-weighted. This is not at all the same. If you are winning and losing equally often, draws won't affect your rating at all. But if you are usually losing, then draws don't help as much as winning half of those games would. I think the effect is to favor using long enough time controls that allow you to make a plus score, where draws won't cost as much as normally. So while the time-adjustment formula may be about right with normal rating rules, it is probably a bit out of line with the draws count for half rule. Basically, the leader will be the player who can consistently win at the lowest time control, which so far is Joel Benjamin at 8' +3". If another player can win almost all the games at a faster time control, he will be the leader. So far only "former_player" has come close, but he seems to play some silly openings sometimes just to make it more challenging, and so doesn't win consistently.

Good morning, Mr. Larry Kaufman. Thank you very much for your important information. I am ignorant of mathematics. Please excuse my mistake. As for the ascent to the summit of the chess Mount Everest: "The Great Mountain Leela Queen Odds," I'm sheltering in place at the "advanced camp" waiting for the "update" and watching the other climbers at "Hillary Pass." Then I'll continue on my rocking horse, "Catecan," struggling to plant the flag at the top of the highest mountain... we are ascending without an oxygen tank. My horse and I are determined to be the first to win the top prize.

[Uri Blass]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

The bot was updated a couple days ago in two small ways: the opening book was updated, and the use of a tiny temperature setting in the endgame (for variety) was eliminated as it sometimes caused missed checkmates. A larger update (for queen odds, queen for knight, and multipiece) may happen in a week or so if tests confirm that it is an upgrade. Meanwhile I did a review of Rapid (10'5" minimum so Rapid by FIDE rules) games by Leela Queen Odds since the Feb27 upgrade, and the break-even Rapid elo seems to be about 2520, which suppedly equates to 2355 FIDE! Even if we allow a bit for some players playing too fast or resigning too soon, it looks like Leela Queen Odds is a fair match for a par FM at FIDE Rapid time controls. Quite incredible.
Just to make sure you don't misunderstand, there is no penalty for draws on the Leaderboard, they are not scored as 0.25 but as 0.5. But the draws are only half-weighted. This is not at all the same. If you are winning and losing equally often, draws won't affect your rating at all. But if you are usually losing, then draws don't help as much as winning half of those games would. I think the effect is to favor using long enough time controls that allow you to make a plus score, where draws won't cost as much as normally. So while the time-adjustment formula may be about right with normal rating rules, it is probably a bit out of line with the draws count for half rule. Basically, the leader will be the player who can consistently win at the lowest time control, which so far is Joel Benjamin at 8' +3". If another player can win almost all the games at a faster time control, he will be the leader. So far only "former_player" has come close, but he seems to play some silly openings sometimes just to make it more challenging, and so doesn't win consistently.

Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black rewgardless of the improvement in Leela.

FM Ori Taichman expect to win 10-0 with black against LeelaQueenOdds in a serious match at 10+5 when he get money for every game.
Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black.

Itay Sitbon(fide rating 2195) agree with Ori Taichman.

I consider if to offer the same players who played against KomodoDragon to play a match of 10 games against LeelaQueenOdds at time control 10+5 in one day when they are black to see if they can win 10-0 or something close to it in these conditions with the idea of high prize only in case of 10-0 or at least 9-1 result.

[Father]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

The bot was updated a couple days ago in two small ways: the opening book was updated, and the use of a tiny temperature setting in the endgame (for variety) was eliminated as it sometimes caused missed checkmates. A larger update (for queen odds, queen for knight, and multipiece) may happen in a week or so if tests confirm that it is an upgrade. Meanwhile I did a review of Rapid (10'5" minimum so Rapid by FIDE rules) games by Leela Queen Odds since the Feb27 upgrade, and the break-even Rapid elo seems to be about 2520, which suppedly equates to 2355 FIDE! Even if we allow a bit for some players playing too fast or resigning too soon, it looks like Leela Queen Odds is a fair match for a par FM at FIDE Rapid time controls. Quite incredible.
Just to make sure you don't misunderstand, there is no penalty for draws on the Leaderboard, they are not scored as 0.25 but as 0.5. But the draws are only half-weighted. This is not at all the same. If you are winning and losing equally often, draws won't affect your rating at all. But if you are usually losing, then draws don't help as much as winning half of those games would. I think the effect is to favor using long enough time controls that allow you to make a plus score, where draws won't cost as much as normally. So while the time-adjustment formula may be about right with normal rating rules, it is probably a bit out of line with the draws count for half rule. Basically, the leader will be the player who can consistently win at the lowest time control, which so far is Joel Benjamin at 8' +3". If another player can win almost all the games at a faster time control, he will be the leader. So far only "former_player" has come close, but he seems to play some silly openings sometimes just to make it more challenging, and so doesn't win consistently.

Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black rewgardless of the improvement in Leela.

FM Ori Taichman expect to win 10-0 with black against LeelaQueenOdds in a serious match at 10+5 when he get money for every game.
Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black.

Itay Sitbon(fide rating 2195) agree with Ori Taichman.

I consider if to offer the same players who played against KomodoDragon to play a match of 10 games against LeelaQueenOdds at time control 10+5 in one day when they are black to see if they can win 10-0 or something close to it in these conditions with the idea of high prize only in case of 10-0 or at least 9-1 result.

...man-versus-machine matches, and in this case, humans versus LeelaQueenOdds, don't have a single, unified approach. For me, the king among humans today is GM Joel Benjamin. I simply base my assessment on facts, results, and his ranking in the top 100 humans playing against Leela. Any human who considers themselves the best at playing the machine should come and see for themselves. I have a lot of faith in LeelaQueenOdds, but my faith in the computer is also corroborated by the daily experience I frequently and repeatedly witness as LeelaQueenOdds crushes and turns into "work clothes" many strong, renowned, amateur, and a multitude of human opponents. I raise a glass to LeelaQueenOdds, the "undisputed queen of 21st-century cyber chess."

[Father]

Good evening, Mr. Larry Kaufman. I was evaluating the synergy and direction that the wonderful LeelaQueenOdds project and Odds in general have been taking. I myself find myself surprised by the focus on "master moves" by "AI bots" and the "leaderboard." The system of updating the board every few minutes has catapulted the process and the enthusiasm of chess players. The world has been changing at a speed greater than that of light, that is, "at the speed of thought." "Science and faith" complement each other; "they are neither mutually exclusive nor opposed." I am a man of faith, and although my faith sometimes fails me, I cling to it in order to fly. I know that the seats at the dinner that has been served for the 20 strongest diners in the world will sooner rather than later be occupied by those 20 comets of light that illuminate our skies. From the "Mozart" of chess to many other flamingos, I hope, I have hope, and I have faith, that soon they will be here playing against LeelaQueenOdds…

Well, a big step in that direction happened today, with a video by IM Levy Rozman, the most popular chess streamer I believe, about the 5 to 0 victory of Leela Queen Odds from the GM in blitz. It resulted in a flood of games, peaking at 36 games at once (the limit for the server now). One new GM ("former_player") has already leaped to third place on the leaderboard in a few hours! It is likely that Levy will play against the bot himself soon, which should generate even more interest. So far, since the Feb. 27 upgrade, the bot has a positive score (93 to 92 plus some draws) at 3'2" against players rated over 2600 Lichess blitz, which is a typical rating for a "par" IM. So I think it is fair to say that it is upper IM level at this time control. Which is really incredible. Plus another upgrade may be coming soon!

I think that a new update can help mainly in games with no increment.

Watching the games of LeelaQueenOdds against the best players at 1+0 I see a lot of blunders of not considering the opponent time.

For example here the bot blundered by allowing a draw by repetition when the oppoent had only 15.2 seconds on the clock.
The correct strategy is not to allow repetition and to push f6 in the last move to avoid the 50 move rule.

If the opponent start premoving then some stupid sacrifice may practically win and I think no human opponent can draw at 1+0.


https://lichess.org/Dw8VcdAH#204

Here the bot blundered by draw by repetition when every different move probably could win on time because the opponent had 0 seconds on the clock in the final position.
https://lichess.org/6RdlThrA#87

Good evening, Mr. Uri Blass.
Mr. Larry Kaufman has informed us that a new "update" of LeelaQueenOdds is imminent.
What an interesting outcome this all is for me.
I never cease to be amazed at the progress and evolution of LeelaQueenOdds.
The new generations of humans who have been interacting with or against LeelaQueenOdds for the first time will be seeing the "computational miracle" as "normal." The truth is that "a simple draw between a GM and a computer playing without its queen is worth a thousand words." And we have these examples in abundance, along with the victories of the star computer "LeelaQueenOdds." I have always believed that when one path closes, by the law of causality, a new one always opens. And in terms of economic forces, this is verifiable. Markets are like water we try to catch with both hands... they look for outlets and promotion channels. Chess machines are the workbench; the programmers and the humans who face them are the players in charge of each side. Sometimes, human players are also experts in design, programming, and algorithm creation, and these humans have a few extra cards compared to ordinary mortals. So, programming LeelaQueenOdds to refuse to repeat moves would have its consequences, and if you're also going to force the computer to sacrifice or exchange pieces and values ​​on move 50, it will have its strategic consequences. Obviously, the strategic approach I've imagined in these hypotheses doesn't make sense to publish it here. Just as an example... having given a rating of 0.25 instead of 0.5 for each tie entailed general consequences for each player and the computer itself, and the rules of the game covered the entire human community that faces it under the same climate. If you trap humans in the closed "win-win" path, it will likely create the "lose-lose" ecosystem, given the evolutionary force of the computer. In short, I want to bring up a phrase my mother taught me as a child that applies well to computational guidelines: "We call 'evil' the 'good' that we cannot understand at the moment." Since March 14th, I have stopped my battles against the computer... I am simply waiting to observe brand new changes in the leaderboard and the emergence or resurgence of new or already proven candidates called to occupy the top positions... but it turns out that just like in the battle of men versus machines, "Many firsts will be last, and many last first." The battle between "the human algorithm" versus "the computer algorithm" is just beginning...

The bot was updated a couple days ago in two small ways: the opening book was updated, and the use of a tiny temperature setting in the endgame (for variety) was eliminated as it sometimes caused missed checkmates. A larger update (for queen odds, queen for knight, and multipiece) may happen in a week or so if tests confirm that it is an upgrade. Meanwhile I did a review of Rapid (10'5" minimum so Rapid by FIDE rules) games by Leela Queen Odds since the Feb27 upgrade, and the break-even Rapid elo seems to be about 2520, which suppedly equates to 2355 FIDE! Even if we allow a bit for some players playing too fast or resigning too soon, it looks like Leela Queen Odds is a fair match for a par FM at FIDE Rapid time controls. Quite incredible.
Just to make sure you don't misunderstand, there is no penalty for draws on the Leaderboard, they are not scored as 0.25 but as 0.5. But the draws are only half-weighted. This is not at all the same. If you are winning and losing equally often, draws won't affect your rating at all. But if you are usually losing, then draws don't help as much as winning half of those games would. I think the effect is to favor using long enough time controls that allow you to make a plus score, where draws won't cost as much as normally. So while the time-adjustment formula may be about right with normal rating rules, it is probably a bit out of line with the draws count for half rule. Basically, the leader will be the player who can consistently win at the lowest time control, which so far is Joel Benjamin at 8' +3". If another player can win almost all the games at a faster time control, he will be the leader. So far only "former_player" has come close, but he seems to play some silly openings sometimes just to make it more challenging, and so doesn't win consistently.

Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black rewgardless of the improvement in Leela.

FM Ori Taichman expect to win 10-0 with black against LeelaQueenOdds in a serious match at 10+5 when he get money for every game.
Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black.

Itay Sitbon(fide rating 2195) agree with Ori Taichman.

I consider if to offer the same players who played against KomodoDragon to play a match of 10 games against LeelaQueenOdds at time control 10+5 in one day when they are black to see if they can win 10-0 or something close to it in these conditions with the idea of high prize only in case of 10-0 or at least 9-1 result.

...man-versus-machine matches, and in this case, humans versus LeelaQueenOdds, don't have a single, unified approach. For me, the king among humans today is GM Joel Benjamin. I simply base my assessment on facts, results, and his ranking in the top 100 humans playing against Leela. Any human who considers themselves the best at playing the machine should come and see for themselves. I have a lot of faith in LeelaQueenOdds, but my faith in the computer is also corroborated by the daily experience I frequently and repeatedly witness as LeelaQueenOdds crushes and turns into "work clothes" many strong, renowned, amateur, and a multitude of human opponents. I raise a glass to LeelaQueenOdds, the "undisputed queen of 21st-century cyber chess."

...it also produces sadness and dismay to see great dancers who invited the Queen to dance, only to leave the party, never to return... perhaps it's the songs or melodies they danced to... I hope they return to the dance, including the dancers from Asia. The presence of the best dancers at the dance is sorely needed to liven up the Three Kings' Day celebration.

[lkaufman]

[
Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black rewgardless of the improvement in Leela.

FM Ori Taichman expect to win 10-0 with black against LeelaQueenOdds in a serious match at 10+5 when he get money for every game.
Oded Ross expect every player with fide rating above 2000 to win at 10+5 against LeelaQueen Odds with black.

Itay Sitbon(fide rating 2195) agree with Ori Taichman.

I consider if to offer the same players who played against KomodoDragon to play a match of 10 games against LeelaQueenOdds at time control 10+5 in one day when they are black to see if they can win 10-0 or something close to it in these conditions with the idea of high prize only in case of 10-0 or at least 9-1 result.

Based on actual results of Rapid games at 10'+5" minimum played since Feb 27 (Leela White), the break-even Lichess Rapid rating is about 2520, which is supposed to equate to 2355 FIDE. I know this sounds crazy high, presumably some of the players were playing too fast or resigning too early or just didn't try hard. But I doubt that the true figure is below 2200, since most players do play to win such games. Ori is a pretty strong player, he might win most of the games but probably not all ten. Lately in blitz even players rated over 2800 haven't made a plus score, crazy as this sounds. But Joel Benjamin does win nearly every game at 8'3", time does make a big difference. Anyway it would be very interesting if you do this, as it will help settle the question of how important motivation and incentive are. You might have to pay out for Ori or anyone else who is of his level, but I wouldn't expect you to pay out for 2200 rated players and below. Frankly I can't understand how so many strong players keep losing to LQO even with decent time limits, but it's hard to dispute the facts. When titled players with known identities play it, they have some incentive to do well, they don't want to hurt their reputations. It is also possible that FIDE ratings in Israel are lower than they would be elsewhere, I wouldn't know about that.