An idea for new Handicap games for dragon

[Uri Blass]

I see that the average error of smerdon was the same as komodo and it is clear that komodo is stronger than smerdon.

It supports my theory that the winner tend to do less mistakes so if you give the human material odd that is big enough for him to usually win then he is going to do less mistakes.

chessqueen is not a GM but he played with odds that are bigger than knight odds.
I do not claim that chessqueen did not cheat but only that you can expect lower average for odd matches when the human win every game(in the case of smerdon it was almost every game because smerdon lost the first game).

My point is that if you want to prove cheating then you need to show that humans who are at a level when they win in similiar conditions have a bigger average.

I do not think that you need to be a GM to win all games in similiar conditions(more than knight odd when you do not play against dragon but only against weaker komodo)

Edit:Note that I see that smerdon won the centi-pawn race in every game when he won.
The average is the same only because of the single game that he lost because the handicap was not enough for 6-0 score for him.

It does not support your theory. A score of 15 or higher is expected from a GM player. You just ignored everything.
And this was also the same positions repeating. And mistakes were made in the games. Unlike chessqueen who made ZERO!

So now you want to throw out data of the match, and cherry pick.....

This supports Centipawn analysis.

And you are clueless here. This is not a comparison of players of this match. One has nothing to do with the other. This is determining if a players precision is above expected human norms. Based on the stats of the best players in the world. Everything else is just a bonus. Who played better, and who is the stronger and more accurate player.

GM Smerdon played within humans norms as shown by the results. Chessqueen did not and played way outside expected human norms of a GM Player. And Chessqueen is not a GM.

More results coming.

I wonder what is the average error rate in the following game of both opponents
I took extreme case when I knew I can win easily.

If my error rate is smaller than stockfish when I did not find the fastest mates then it proves my point

[pgn][Event "Computer chess game"]
[Site "DESKTOP-7QE6S12"]
[Date "2020.12.09"]
[Round "?"]
[White "àåøé"]
[Black "Stockfish_20112916_x64_bmi2"]
[Result "1-0"]
[BlackElo "2200"]
[Time "08:56:25"]
[WhiteElo "2400"]
[TimeControl "120+1"]
[SetUp "1"]
[FEN "1nb1kbn1/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1"]
[Termination "normal"]
[PlyCount "49"]
[WhiteType "human"]
[BlackType "program"]

1. e4 g6 {(g7-g6 Ng1-e2 b7-b6 d2-d3 Nb8-c6 g2-g4 e7-e6 Bf1-g2 Bf8-g7 O-O
Ng8-e7 g4-g5 Ke8-f8 Qd1-d2 Bc8-a6 c2-c3 Nc6-e5 Ne2-f4 d7-d5 Rf1-d1 d5xe4
d3xe4 Kf8-g8 h2-h4) -22.78/29 12} 2. d4 e6 {(e7-e6 c2-c3 b7-b6 Bc1-f4
Bf8-d6 Bf4-e3 f7-f5 e4-e5 Bd6-f8 Qd1-a4 a7-a5 Nb1-d2 Ng8-e7 c3-c4 Ne7-c6
Nd2-f3 Bc8-b7 a2-a3 Nc6-e7 O-O-O Bb7-c6 Qa4-b3 Bc6xf3 Ng1xf3 Ne7-c6 h2-h4
Bf8-g7 h4-h5 Ke8-f7 Nf3-d2 Kf7-g8 h5xg6 h7xg6) -22.67/27 4} 3. Nf3 d6
{(d7-d6 Bc1-g5 f7-f5 e4xf5 g6xf5 Nb1-c3 Bf8-e7 Bg5-c1 a7-a6 d4-d5 Ng8-f6
Qd1-d4 c7-c5 Qd4-a4+ b7-b5 Qa4-b3 c5-c4 Bf1xc4 Nf6-e4 Nc3xb5 a6xb5 Qb3xb5+
Nb8-d7) -22.63/23 2} 4. Bc4 Nc6 {(Nb8-c6 c2-c3 Bf8-h6 Nb1-d2 Ng8-f6 Bc4-d3
Ke8-f8 O-O Kf8-g7 Rf1-e1 Bh6xd2 Bc1xd2 a7-a6 Qd1-a4 Nc6-e7 h2-h3 h7-h5
Qa4-b3 Ne7-c6 Re1-e2 Nc6-e7 a2-a3 Ne7-c6 c3-c4 e6-e5 d4-d5 Nc6-e7)
-22.37/26 1} 5. O-O Bh6 {(Bf8-h6 Nb1-d2 Ke8-f8 c2-c3 Kf8-g7 Bc4-d3 Bh6xd2
Bc1xd2 b7-b6 Qd1-e2 Bc8-b7 g2-g4 Nc6-e7 g4-g5 d6-d5 e4-e5 a7-a6 a2-a4
Ne7-c6 b2-b4 Ng8-e7 b4-b5 a6xb5 a4xb5 Nc6-d8) -22.44/25 2 Black resigns} 6.
Bxh6 Nxh6 {(Ng8xh6 Qd1-d2 Nh6-g8 Qd2-c3 h7-h6 d4-d5 Nc6-e7 d5xe6 f7xe6
Bc4-b3 c7-c6 Rf1-d1 d6-d5 Qc3-g7 b7-b5 Nf3-e5 Bc8-d7 Nb1-c3 a7-a5 Qg7-f7+
Ke8-d8 a2-a4 b5-b4 Nc3-e2 Kd8-c7 f2-f3 g6-g5 Qf7-h5 Kc7-d8 c2-c3 b4xc3
Ne2xc3 Kd8-c8) -22.59/28 3 Black resigns} 7. d5 Ne5 {(Nc6-e5 Nf3xe5)
-23.99/29 18 Black resigns} 8. Nxe5 a6 {(a7-a6 Ne5-f3 e6-e5 Qd1-d2 Nh6-g8
Qd2-a5 b7-b6 Qa5-c3 f7-f6 Nf3-d2 Ke8-d8 Qc3-e3 Kd8-e7 Nb1-c3 a6-a5 Bc4-b3
Bc8-d7 a2-a4 h7-h5 Qe3-e2 Ke7-d8 Qe2-a6 Bd7-c8 Qa6-c4 Bc8-d7 Ra1-e1 Kd8-c8
f2-f3 Ng8-e7) -24.80/27 7 Black resigns} 9. dxe6 Bxe6 {(Bc8xe6 Bc4xe6 f7xe6
Ne5-d3 Nh6-f7 Qd1-g4 Ke8-d7 Qg4-h4 h7-h6 Qh4-f6 Nf7-d8 Qf6-g7+ Kd7-c8
Qg7xh6 b7-b5 Qh6xg6 Kc8-b7 Qg6-g8 Nd8-c6 Qg8xe6 Nc6-a5 Qe6-d5+ Kb7-b6
Nb1-c3 Na5-c6 Qd5-f7 Kb6-b7 a2-a3) -24.40/20 1 Black resigns} 10. Bxe6 fxe6
{(f7xe6 Ne5-d3 Nh6-f7 Qd1-g4 Nf7-d8 Qg4-h3 e6-e5 Qh3xh7 Nd8-e6 Qh7xg6+
Ke8-e7 Qg6xe6+ Ke7xe6 g2-g3 a6-a5 Nb1-c3 c7-c6 Kg1-g2 Ke6-f7 a2-a4 Kf7-f6
Kg2-f3 Kf6-g7 h2-h3 Kg7-g6 h3-h4 b7-b6) -25.03/24 4 Black resigns} 11. Nf3
Nf7 {(Nh6-f7 Qd1-d4) -26.02/24 16 Black resigns} 12. Qd2 c5 {(c7-c5 Qd2-f4
b7-b5 Qf4-f6 b5-b4 Qf6xe6+ Ke8-f8 Qe6xf7+ Kf8xf7 Nb1-d2 Kf7-e6 Nf3-g5+
Ke6-e7 Ng5xh7 d6-d5 e4xd5 Ke7-d6 Nh7-f6 Kd6-e5 Nd2-e4 c5-c4 h2-h4 c4-c3
b2xc3 b4xc3 g2-g4) -24.94/22 3 Black resigns} 13. Ng5 a5 {(a6-a5 Ng5xf7
d6-d5 e4xd5 Ke8xf7 d5xe6+ Kf7-e7 Rf1-e1 a5-a4 Qd2-f4 c5-c4 Re1-d1 Ke7xe6
Rd1-d6+ Ke6-e7 Qf4-f6+ Ke7-e8 Rd6-d8+) -M9/24 4 Black resigns} 14. Nxf7 d5
{(d6-d5 Qd2-f4 Ke8-e7 Qf4-c7+ Ke7-f6 f2-f4 g6-g5 Nf7xg5 Kf6-g6 Qc7xh7+
Kg6-f6 Qh7-f7+) -M6/51 1 Black resigns} 15. Ng5 d4 {(d5-d4 Qd2-f4 Ke8-d7
Nb1-c3 d4xc3 Rf1-d1+ Kd7-c6 Qf4-d6+ Kc6-b5 a2-a4+ Kb5-c4 Qd6xe6+ Kc4-b4
Qe6-b3+) -M7/48 1 Black resigns} 16. Nxe6 Kf7 {(Ke8-f7 Ne6-c7 g6-g5 Qd2xg5
h7-h6 Qg5-f5+ Kf7-e7 Qf5-e6+ Ke7-f8 Qe6-f6+ Kf8-g8 Nc7-e6 d4-d3 Qf6-g7+)
-M7/46 1 Black resigns} 17. Nxc5 b6 {(b7-b6 Qd2xd4 b6xc5 Qd4-e5 h7-h6
Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4 Qe5-f5+) -M6/54 1 Black resigns}
18. Qxd4 bxc5 {(b6xc5 Qd4-e5 h7-h6 Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4
Qe5-f5+) -M5/145 1 Black resigns} 19. Qd6 c4 {(c5-c4 Rf1-d1 a5-a4 Qd6-c7+
Kf7-e6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns} 20. Re1 h5 {(h7-h5
Re1-d1 h5-h4 Qd6-c7+ Kf7-f6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns}
21. Re3 Kg7 {(Kf7-g7 Re3-f3 Kg7-h7 Qd6-e7+ Kh7-h6 Qe7-f8+ Kh6-g5 Qf8-f4+)
-M4/245 0 Black resigns} 22. Rg3 h4 {(h5-h4 Rg3xg6+ Kg7-h7 Qd6-f6 a5-a4
Qf6-g7+) -M3/245 0 Black resigns} 23. Rxg6+ Kh8 {(Kg7-h8 Qd6-f8+ Kh8-h7
Rg6-h6+) -M2/245 0 Black resigns} 24. Qf8+ Kh7 {(Kh8-h7 Rg6-h6+) -M1/245 0
Black resigns} 25. Qg7# 1-0[/pgn]

I will look at the game later today. And post the results. I hope you realize this is a statistical measurement. And any one game even without odds can show anything. That is why you can not cherry pick data. All the matches we are looking at have many games. And the match as a whole is what you look at for the data.

Meaning it is not really the game that is the most important. It is a meaningful number of moves.

And remember we are not comparing the players in the match.

We are comparing the strong side only. Or the player that starts with all his material. In the odds games. To see if your theory stands. Will that human player ever look like a computer play. As chessqueen.

I have analyzed most of the odds games on Larry's site, and so far the answer is no.

I do NOT know why you are accusing me of using an engine when the advantage is clearly close to a Rook, and my online trainer who constantly tell me that I think very precise when I am ahead in material even to nickname me Capa and always telling me that I am advancing fast, but anyway I decided to set upu another position where the advantage is close to a Rook Anyway I used Komodo Free Version this time. mwyoung, If you live in the USA I would pay for 1/3 of the plane ticket if you can bring your computer and I will replay some of these Odds, but I would like to bet $300.00 for each game using some of the Odds that I have posted here. On this game Komodo resigned because after exchanging the pawns White bishop can NOT take because I would fork both Bishops by placing my Queen on d5 !

[pgn][Event "Blitz:20'+10""]
[Site "MyTown"]
[Date "2020.09.12"]
[Round "7"]
[White "Komodo 12.1.1 64-bit"]
[Black "ChessQueen"]
[Result "0-1"]
[SetUp "1"]
[FEN "3qkbnr/pppppppp/8/8/8/8/PP1PPPPP/1NB1KBNR w Kk - 0 1"]
[PlyCount "55"]
[TimeControl "1200+10"]

1. e4 d5 2. exd5 Qxd5 3. Nc3 Qf5 4. d3 Nf6 5. Nf3 g6 6. Be2 Bg7 7. O-O O-O 8.
d4 Nd5 9. Nxd5 Qxd5 10. b3 Bxd4 11. Nxd4 Qxd4 12. a4 Rd8 13. Bc4 Qe4 14. Be3
Qc2 15. h3 Rd1 16. Rxd1 Qxd1+ 17. Kh2 Kg7 18. Kg3 a6 19. Kh2 g5 20. Bxg5 b5 0-1[/pgn]

You were tricking us. You only cheated yourself.

Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung[/size] ?

[pgn][Event "Blitz:20'+10""]
[Site " mwyoung "]
[Date "????.??.??"]
[Round "?"]
[White "Komodo 12.1.1 64-bit"]
[Black "Chess, Queen"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1"]
[PlyCount "1"]

1. d4 1-0[/pgn]

Uri is not correct. And Uri has done no testing. Uri is not even clear what or how centipawn analysis works. And the same can be said for you.

I have tested it, and in no case did it show a human as playing like a computer. Because the games was played with odds.

Now if you guys have data show it, and stop making up fact with motive.

I know you already tried but you are embarrassed to post your losing game,I know if you try hard you can beat this position in less than 30 moves
Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung

[d]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1']

I will post the unbiased data from Larry's site. Since you and Uri are good at making up facts. I want to see you produce any data without the program.... I suggest that if you have data then show it.
[/quote]

Based on your own data:
PlayKomodo: 6/14/21/27/14/10 => Average=0.15
smurfo: 37/7/15/15/3/6 => Average=0.15


1)I checked the average of these numbers and I see the 0.15 is not correct and komodo has bigger error rate than the opponent.
komodo's average (6+14+21+27+14+10)/6=92/6=0.153333...
smurfo's average (37+7+15+15+3+6)/6=83/6=0.1383333...

2)komodo had higher average in the games that it lost.
14-7 in game 2
21-15 in game 3
27-15 in game 4
14-3 in game 5
10-6 in game 6

The data supports my opinion that if the human wins the game then usually he has a smaller error rate.
I do not know how to calculate the error but if you compare between evaluation before the player's move and after the player's move the calculation
is biased against the loser because you will find mistakes of the loser when the evaluation goes down after the move(otherwise the loser does not lose).

[mwyoung]

I see that the average error of smerdon was the same as komodo and it is clear that komodo is stronger than smerdon.

It supports my theory that the winner tend to do less mistakes so if you give the human material odd that is big enough for him to usually win then he is going to do less mistakes.

chessqueen is not a GM but he played with odds that are bigger than knight odds.
I do not claim that chessqueen did not cheat but only that you can expect lower average for odd matches when the human win every game(in the case of smerdon it was almost every game because smerdon lost the first game).

My point is that if you want to prove cheating then you need to show that humans who are at a level when they win in similiar conditions have a bigger average.

I do not think that you need to be a GM to win all games in similiar conditions(more than knight odd when you do not play against dragon but only against weaker komodo)

Edit:Note that I see that smerdon won the centi-pawn race in every game when he won.
The average is the same only because of the single game that he lost because the handicap was not enough for 6-0 score for him.

It does not support your theory. A score of 15 or higher is expected from a GM player. You just ignored everything.
And this was also the same positions repeating. And mistakes were made in the games. Unlike chessqueen who made ZERO!

So now you want to throw out data of the match, and cherry pick.....

This supports Centipawn analysis.

And you are clueless here. This is not a comparison of players of this match. One has nothing to do with the other. This is determining if a players precision is above expected human norms. Based on the stats of the best players in the world. Everything else is just a bonus. Who played better, and who is the stronger and more accurate player.

GM Smerdon played within humans norms as shown by the results. Chessqueen did not and played way outside expected human norms of a GM Player. And Chessqueen is not a GM.

More results coming.

I wonder what is the average error rate in the following game of both opponents
I took extreme case when I knew I can win easily.

If my error rate is smaller than stockfish when I did not find the fastest mates then it proves my point

[pgn][Event "Computer chess game"]
[Site "DESKTOP-7QE6S12"]
[Date "2020.12.09"]
[Round "?"]
[White "àåøé"]
[Black "Stockfish_20112916_x64_bmi2"]
[Result "1-0"]
[BlackElo "2200"]
[Time "08:56:25"]
[WhiteElo "2400"]
[TimeControl "120+1"]
[SetUp "1"]
[FEN "1nb1kbn1/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1"]
[Termination "normal"]
[PlyCount "49"]
[WhiteType "human"]
[BlackType "program"]

1. e4 g6 {(g7-g6 Ng1-e2 b7-b6 d2-d3 Nb8-c6 g2-g4 e7-e6 Bf1-g2 Bf8-g7 O-O
Ng8-e7 g4-g5 Ke8-f8 Qd1-d2 Bc8-a6 c2-c3 Nc6-e5 Ne2-f4 d7-d5 Rf1-d1 d5xe4
d3xe4 Kf8-g8 h2-h4) -22.78/29 12} 2. d4 e6 {(e7-e6 c2-c3 b7-b6 Bc1-f4
Bf8-d6 Bf4-e3 f7-f5 e4-e5 Bd6-f8 Qd1-a4 a7-a5 Nb1-d2 Ng8-e7 c3-c4 Ne7-c6
Nd2-f3 Bc8-b7 a2-a3 Nc6-e7 O-O-O Bb7-c6 Qa4-b3 Bc6xf3 Ng1xf3 Ne7-c6 h2-h4
Bf8-g7 h4-h5 Ke8-f7 Nf3-d2 Kf7-g8 h5xg6 h7xg6) -22.67/27 4} 3. Nf3 d6
{(d7-d6 Bc1-g5 f7-f5 e4xf5 g6xf5 Nb1-c3 Bf8-e7 Bg5-c1 a7-a6 d4-d5 Ng8-f6
Qd1-d4 c7-c5 Qd4-a4+ b7-b5 Qa4-b3 c5-c4 Bf1xc4 Nf6-e4 Nc3xb5 a6xb5 Qb3xb5+
Nb8-d7) -22.63/23 2} 4. Bc4 Nc6 {(Nb8-c6 c2-c3 Bf8-h6 Nb1-d2 Ng8-f6 Bc4-d3
Ke8-f8 O-O Kf8-g7 Rf1-e1 Bh6xd2 Bc1xd2 a7-a6 Qd1-a4 Nc6-e7 h2-h3 h7-h5
Qa4-b3 Ne7-c6 Re1-e2 Nc6-e7 a2-a3 Ne7-c6 c3-c4 e6-e5 d4-d5 Nc6-e7)
-22.37/26 1} 5. O-O Bh6 {(Bf8-h6 Nb1-d2 Ke8-f8 c2-c3 Kf8-g7 Bc4-d3 Bh6xd2
Bc1xd2 b7-b6 Qd1-e2 Bc8-b7 g2-g4 Nc6-e7 g4-g5 d6-d5 e4-e5 a7-a6 a2-a4
Ne7-c6 b2-b4 Ng8-e7 b4-b5 a6xb5 a4xb5 Nc6-d8) -22.44/25 2 Black resigns} 6.
Bxh6 Nxh6 {(Ng8xh6 Qd1-d2 Nh6-g8 Qd2-c3 h7-h6 d4-d5 Nc6-e7 d5xe6 f7xe6
Bc4-b3 c7-c6 Rf1-d1 d6-d5 Qc3-g7 b7-b5 Nf3-e5 Bc8-d7 Nb1-c3 a7-a5 Qg7-f7+
Ke8-d8 a2-a4 b5-b4 Nc3-e2 Kd8-c7 f2-f3 g6-g5 Qf7-h5 Kc7-d8 c2-c3 b4xc3
Ne2xc3 Kd8-c8) -22.59/28 3 Black resigns} 7. d5 Ne5 {(Nc6-e5 Nf3xe5)
-23.99/29 18 Black resigns} 8. Nxe5 a6 {(a7-a6 Ne5-f3 e6-e5 Qd1-d2 Nh6-g8
Qd2-a5 b7-b6 Qa5-c3 f7-f6 Nf3-d2 Ke8-d8 Qc3-e3 Kd8-e7 Nb1-c3 a6-a5 Bc4-b3
Bc8-d7 a2-a4 h7-h5 Qe3-e2 Ke7-d8 Qe2-a6 Bd7-c8 Qa6-c4 Bc8-d7 Ra1-e1 Kd8-c8
f2-f3 Ng8-e7) -24.80/27 7 Black resigns} 9. dxe6 Bxe6 {(Bc8xe6 Bc4xe6 f7xe6
Ne5-d3 Nh6-f7 Qd1-g4 Ke8-d7 Qg4-h4 h7-h6 Qh4-f6 Nf7-d8 Qf6-g7+ Kd7-c8
Qg7xh6 b7-b5 Qh6xg6 Kc8-b7 Qg6-g8 Nd8-c6 Qg8xe6 Nc6-a5 Qe6-d5+ Kb7-b6
Nb1-c3 Na5-c6 Qd5-f7 Kb6-b7 a2-a3) -24.40/20 1 Black resigns} 10. Bxe6 fxe6
{(f7xe6 Ne5-d3 Nh6-f7 Qd1-g4 Nf7-d8 Qg4-h3 e6-e5 Qh3xh7 Nd8-e6 Qh7xg6+
Ke8-e7 Qg6xe6+ Ke7xe6 g2-g3 a6-a5 Nb1-c3 c7-c6 Kg1-g2 Ke6-f7 a2-a4 Kf7-f6
Kg2-f3 Kf6-g7 h2-h3 Kg7-g6 h3-h4 b7-b6) -25.03/24 4 Black resigns} 11. Nf3
Nf7 {(Nh6-f7 Qd1-d4) -26.02/24 16 Black resigns} 12. Qd2 c5 {(c7-c5 Qd2-f4
b7-b5 Qf4-f6 b5-b4 Qf6xe6+ Ke8-f8 Qe6xf7+ Kf8xf7 Nb1-d2 Kf7-e6 Nf3-g5+
Ke6-e7 Ng5xh7 d6-d5 e4xd5 Ke7-d6 Nh7-f6 Kd6-e5 Nd2-e4 c5-c4 h2-h4 c4-c3
b2xc3 b4xc3 g2-g4) -24.94/22 3 Black resigns} 13. Ng5 a5 {(a6-a5 Ng5xf7
d6-d5 e4xd5 Ke8xf7 d5xe6+ Kf7-e7 Rf1-e1 a5-a4 Qd2-f4 c5-c4 Re1-d1 Ke7xe6
Rd1-d6+ Ke6-e7 Qf4-f6+ Ke7-e8 Rd6-d8+) -M9/24 4 Black resigns} 14. Nxf7 d5
{(d6-d5 Qd2-f4 Ke8-e7 Qf4-c7+ Ke7-f6 f2-f4 g6-g5 Nf7xg5 Kf6-g6 Qc7xh7+
Kg6-f6 Qh7-f7+) -M6/51 1 Black resigns} 15. Ng5 d4 {(d5-d4 Qd2-f4 Ke8-d7
Nb1-c3 d4xc3 Rf1-d1+ Kd7-c6 Qf4-d6+ Kc6-b5 a2-a4+ Kb5-c4 Qd6xe6+ Kc4-b4
Qe6-b3+) -M7/48 1 Black resigns} 16. Nxe6 Kf7 {(Ke8-f7 Ne6-c7 g6-g5 Qd2xg5
h7-h6 Qg5-f5+ Kf7-e7 Qf5-e6+ Ke7-f8 Qe6-f6+ Kf8-g8 Nc7-e6 d4-d3 Qf6-g7+)
-M7/46 1 Black resigns} 17. Nxc5 b6 {(b7-b6 Qd2xd4 b6xc5 Qd4-e5 h7-h6
Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4 Qe5-f5+) -M6/54 1 Black resigns}
18. Qxd4 bxc5 {(b6xc5 Qd4-e5 h7-h6 Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4
Qe5-f5+) -M5/145 1 Black resigns} 19. Qd6 c4 {(c5-c4 Rf1-d1 a5-a4 Qd6-c7+
Kf7-e6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns} 20. Re1 h5 {(h7-h5
Re1-d1 h5-h4 Qd6-c7+ Kf7-f6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns}
21. Re3 Kg7 {(Kf7-g7 Re3-f3 Kg7-h7 Qd6-e7+ Kh7-h6 Qe7-f8+ Kh6-g5 Qf8-f4+)
-M4/245 0 Black resigns} 22. Rg3 h4 {(h5-h4 Rg3xg6+ Kg7-h7 Qd6-f6 a5-a4
Qf6-g7+) -M3/245 0 Black resigns} 23. Rxg6+ Kh8 {(Kg7-h8 Qd6-f8+ Kh8-h7
Rg6-h6+) -M2/245 0 Black resigns} 24. Qf8+ Kh7 {(Kh8-h7 Rg6-h6+) -M1/245 0
Black resigns} 25. Qg7# 1-0[/pgn]

I will look at the game later today. And post the results. I hope you realize this is a statistical measurement. And any one game even without odds can show anything. That is why you can not cherry pick data. All the matches we are looking at have many games. And the match as a whole is what you look at for the data.

Meaning it is not really the game that is the most important. It is a meaningful number of moves.

And remember we are not comparing the players in the match.

We are comparing the strong side only. Or the player that starts with all his material. In the odds games. To see if your theory stands. Will that human player ever look like a computer play. As chessqueen.

I have analyzed most of the odds games on Larry's site, and so far the answer is no.

I do NOT know why you are accusing me of using an engine when the advantage is clearly close to a Rook, and my online trainer who constantly tell me that I think very precise when I am ahead in material even to nickname me Capa and always telling me that I am advancing fast, but anyway I decided to set upu another position where the advantage is close to a Rook Anyway I used Komodo Free Version this time. mwyoung, If you live in the USA I would pay for 1/3 of the plane ticket if you can bring your computer and I will replay some of these Odds, but I would like to bet $300.00 for each game using some of the Odds that I have posted here. On this game Komodo resigned because after exchanging the pawns White bishop can NOT take because I would fork both Bishops by placing my Queen on d5 !

[pgn][Event "Blitz:20'+10""]
[Site "MyTown"]
[Date "2020.09.12"]
[Round "7"]
[White "Komodo 12.1.1 64-bit"]
[Black "ChessQueen"]
[Result "0-1"]
[SetUp "1"]
[FEN "3qkbnr/pppppppp/8/8/8/8/PP1PPPPP/1NB1KBNR w Kk - 0 1"]
[PlyCount "55"]
[TimeControl "1200+10"]

1. e4 d5 2. exd5 Qxd5 3. Nc3 Qf5 4. d3 Nf6 5. Nf3 g6 6. Be2 Bg7 7. O-O O-O 8.
d4 Nd5 9. Nxd5 Qxd5 10. b3 Bxd4 11. Nxd4 Qxd4 12. a4 Rd8 13. Bc4 Qe4 14. Be3
Qc2 15. h3 Rd1 16. Rxd1 Qxd1+ 17. Kh2 Kg7 18. Kg3 a6 19. Kh2 g5 20. Bxg5 b5 0-1[/pgn]

You were tricking us. You only cheated yourself.

Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung[/size] ?

[pgn][Event "Blitz:20'+10""]
[Site " mwyoung "]
[Date "????.??.??"]
[Round "?"]
[White "Komodo 12.1.1 64-bit"]
[Black "Chess, Queen"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1"]
[PlyCount "1"]

1. d4 1-0[/pgn]

Uri is not correct. And Uri has done no testing. Uri is not even clear what or how centipawn analysis works. And the same can be said for you.

I have tested it, and in no case did it show a human as playing like a computer. Because the games was played with odds.

Now if you guys have data show it, and stop making up fact with motive.

I know you already tried but you are embarrassed to post your losing game,I know if you try hard you can beat this position in less than 30 moves
Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung

[d]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1']

I will post the unbiased data from Larry's site. Since you and Uri are good at making up facts. I want to see you produce any data without the program.... I suggest that if you have data then show it.

Based on your own data:
PlayKomodo: 6/14/21/27/14/10 => Average=0.15
smurfo: 37/7/15/15/3/6 => Average=0.15


1)I checked the average of these numbers and I see the 0.15 is not correct and komodo has bigger error rate than the opponent.
komodo's average (6+14+21+27+14+10)/6=92/6=0.153333...
smurfo's average (37+7+15+15+3+6)/6=83/6=0.1383333...

2)komodo had higher average in the games that it lost.
14-7 in game 2
21-15 in game 3
27-15 in game 4
14-3 in game 5
10-6 in game 6

The data supports my opinion that if the human wins the game then usually he has a smaller error rate.
I do not know how to calculate the error but if you compare between evaluation before the player's move and after the player's move the calculation
is biased against the loser because you will find mistakes of the loser when the evaluation goes down after the move(otherwise the loser does not lose).
[/quote]


Uri stop being stupid. Or do you just ignored everything posted. This is not a comparison of the 2 players in the match. This is about your claim of a false positive. We are only concerned about the human playing the strong side of the odds game. That was your claim, and the data shows you like to make up facts.

All humans tested show that they play like a human player. Even in odds games.

[Chessqueen]

I see that the average error of smerdon was the same as komodo and it is clear that komodo is stronger than smerdon.

It supports my theory that the winner tend to do less mistakes so if you give the human material odd that is big enough for him to usually win then he is going to do less mistakes.

chessqueen is not a GM but he played with odds that are bigger than knight odds.
I do not claim that chessqueen did not cheat but only that you can expect lower average for odd matches when the human win every game(in the case of smerdon it was almost every game because smerdon lost the first game).

My point is that if you want to prove cheating then you need to show that humans who are at a level when they win in similiar conditions have a bigger average.

I do not think that you need to be a GM to win all games in similiar conditions(more than knight odd when you do not play against dragon but only against weaker komodo)

Edit:Note that I see that smerdon won the centi-pawn race in every game when he won.
The average is the same only because of the single game that he lost because the handicap was not enough for 6-0 score for him.

It does not support your theory. A score of 15 or higher is expected from a GM player. You just ignored everything.
And this was also the same positions repeating. And mistakes were made in the games. Unlike chessqueen who made ZERO!

So now you want to throw out data of the match, and cherry pick.....

This supports Centipawn analysis.

And you are clueless here. This is not a comparison of players of this match. One has nothing to do with the other. This is determining if a players precision is above expected human norms. Based on the stats of the best players in the world. Everything else is just a bonus. Who played better, and who is the stronger and more accurate player.

GM Smerdon played within humans norms as shown by the results. Chessqueen did not and played way outside expected human norms of a GM Player. And Chessqueen is not a GM.

More results coming.

I wonder what is the average error rate in the following game of both opponents
I took extreme case when I knew I can win easily.

If my error rate is smaller than stockfish when I did not find the fastest mates then it proves my point

[pgn][Event "Computer chess game"]
[Site "DESKTOP-7QE6S12"]
[Date "2020.12.09"]
[Round "?"]
[White "àåøé"]
[Black "Stockfish_20112916_x64_bmi2"]
[Result "1-0"]
[BlackElo "2200"]
[Time "08:56:25"]
[WhiteElo "2400"]
[TimeControl "120+1"]
[SetUp "1"]
[FEN "1nb1kbn1/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1"]
[Termination "normal"]
[PlyCount "49"]
[WhiteType "human"]
[BlackType "program"]

1. e4 g6 {(g7-g6 Ng1-e2 b7-b6 d2-d3 Nb8-c6 g2-g4 e7-e6 Bf1-g2 Bf8-g7 O-O
Ng8-e7 g4-g5 Ke8-f8 Qd1-d2 Bc8-a6 c2-c3 Nc6-e5 Ne2-f4 d7-d5 Rf1-d1 d5xe4
d3xe4 Kf8-g8 h2-h4) -22.78/29 12} 2. d4 e6 {(e7-e6 c2-c3 b7-b6 Bc1-f4
Bf8-d6 Bf4-e3 f7-f5 e4-e5 Bd6-f8 Qd1-a4 a7-a5 Nb1-d2 Ng8-e7 c3-c4 Ne7-c6
Nd2-f3 Bc8-b7 a2-a3 Nc6-e7 O-O-O Bb7-c6 Qa4-b3 Bc6xf3 Ng1xf3 Ne7-c6 h2-h4
Bf8-g7 h4-h5 Ke8-f7 Nf3-d2 Kf7-g8 h5xg6 h7xg6) -22.67/27 4} 3. Nf3 d6
{(d7-d6 Bc1-g5 f7-f5 e4xf5 g6xf5 Nb1-c3 Bf8-e7 Bg5-c1 a7-a6 d4-d5 Ng8-f6
Qd1-d4 c7-c5 Qd4-a4+ b7-b5 Qa4-b3 c5-c4 Bf1xc4 Nf6-e4 Nc3xb5 a6xb5 Qb3xb5+
Nb8-d7) -22.63/23 2} 4. Bc4 Nc6 {(Nb8-c6 c2-c3 Bf8-h6 Nb1-d2 Ng8-f6 Bc4-d3
Ke8-f8 O-O Kf8-g7 Rf1-e1 Bh6xd2 Bc1xd2 a7-a6 Qd1-a4 Nc6-e7 h2-h3 h7-h5
Qa4-b3 Ne7-c6 Re1-e2 Nc6-e7 a2-a3 Ne7-c6 c3-c4 e6-e5 d4-d5 Nc6-e7)
-22.37/26 1} 5. O-O Bh6 {(Bf8-h6 Nb1-d2 Ke8-f8 c2-c3 Kf8-g7 Bc4-d3 Bh6xd2
Bc1xd2 b7-b6 Qd1-e2 Bc8-b7 g2-g4 Nc6-e7 g4-g5 d6-d5 e4-e5 a7-a6 a2-a4
Ne7-c6 b2-b4 Ng8-e7 b4-b5 a6xb5 a4xb5 Nc6-d8) -22.44/25 2 Black resigns} 6.
Bxh6 Nxh6 {(Ng8xh6 Qd1-d2 Nh6-g8 Qd2-c3 h7-h6 d4-d5 Nc6-e7 d5xe6 f7xe6
Bc4-b3 c7-c6 Rf1-d1 d6-d5 Qc3-g7 b7-b5 Nf3-e5 Bc8-d7 Nb1-c3 a7-a5 Qg7-f7+
Ke8-d8 a2-a4 b5-b4 Nc3-e2 Kd8-c7 f2-f3 g6-g5 Qf7-h5 Kc7-d8 c2-c3 b4xc3
Ne2xc3 Kd8-c8) -22.59/28 3 Black resigns} 7. d5 Ne5 {(Nc6-e5 Nf3xe5)
-23.99/29 18 Black resigns} 8. Nxe5 a6 {(a7-a6 Ne5-f3 e6-e5 Qd1-d2 Nh6-g8
Qd2-a5 b7-b6 Qa5-c3 f7-f6 Nf3-d2 Ke8-d8 Qc3-e3 Kd8-e7 Nb1-c3 a6-a5 Bc4-b3
Bc8-d7 a2-a4 h7-h5 Qe3-e2 Ke7-d8 Qe2-a6 Bd7-c8 Qa6-c4 Bc8-d7 Ra1-e1 Kd8-c8
f2-f3 Ng8-e7) -24.80/27 7 Black resigns} 9. dxe6 Bxe6 {(Bc8xe6 Bc4xe6 f7xe6
Ne5-d3 Nh6-f7 Qd1-g4 Ke8-d7 Qg4-h4 h7-h6 Qh4-f6 Nf7-d8 Qf6-g7+ Kd7-c8
Qg7xh6 b7-b5 Qh6xg6 Kc8-b7 Qg6-g8 Nd8-c6 Qg8xe6 Nc6-a5 Qe6-d5+ Kb7-b6
Nb1-c3 Na5-c6 Qd5-f7 Kb6-b7 a2-a3) -24.40/20 1 Black resigns} 10. Bxe6 fxe6
{(f7xe6 Ne5-d3 Nh6-f7 Qd1-g4 Nf7-d8 Qg4-h3 e6-e5 Qh3xh7 Nd8-e6 Qh7xg6+
Ke8-e7 Qg6xe6+ Ke7xe6 g2-g3 a6-a5 Nb1-c3 c7-c6 Kg1-g2 Ke6-f7 a2-a4 Kf7-f6
Kg2-f3 Kf6-g7 h2-h3 Kg7-g6 h3-h4 b7-b6) -25.03/24 4 Black resigns} 11. Nf3
Nf7 {(Nh6-f7 Qd1-d4) -26.02/24 16 Black resigns} 12. Qd2 c5 {(c7-c5 Qd2-f4
b7-b5 Qf4-f6 b5-b4 Qf6xe6+ Ke8-f8 Qe6xf7+ Kf8xf7 Nb1-d2 Kf7-e6 Nf3-g5+
Ke6-e7 Ng5xh7 d6-d5 e4xd5 Ke7-d6 Nh7-f6 Kd6-e5 Nd2-e4 c5-c4 h2-h4 c4-c3
b2xc3 b4xc3 g2-g4) -24.94/22 3 Black resigns} 13. Ng5 a5 {(a6-a5 Ng5xf7
d6-d5 e4xd5 Ke8xf7 d5xe6+ Kf7-e7 Rf1-e1 a5-a4 Qd2-f4 c5-c4 Re1-d1 Ke7xe6
Rd1-d6+ Ke6-e7 Qf4-f6+ Ke7-e8 Rd6-d8+) -M9/24 4 Black resigns} 14. Nxf7 d5
{(d6-d5 Qd2-f4 Ke8-e7 Qf4-c7+ Ke7-f6 f2-f4 g6-g5 Nf7xg5 Kf6-g6 Qc7xh7+
Kg6-f6 Qh7-f7+) -M6/51 1 Black resigns} 15. Ng5 d4 {(d5-d4 Qd2-f4 Ke8-d7
Nb1-c3 d4xc3 Rf1-d1+ Kd7-c6 Qf4-d6+ Kc6-b5 a2-a4+ Kb5-c4 Qd6xe6+ Kc4-b4
Qe6-b3+) -M7/48 1 Black resigns} 16. Nxe6 Kf7 {(Ke8-f7 Ne6-c7 g6-g5 Qd2xg5
h7-h6 Qg5-f5+ Kf7-e7 Qf5-e6+ Ke7-f8 Qe6-f6+ Kf8-g8 Nc7-e6 d4-d3 Qf6-g7+)
-M7/46 1 Black resigns} 17. Nxc5 b6 {(b7-b6 Qd2xd4 b6xc5 Qd4-e5 h7-h6
Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4 Qe5-f5+) -M6/54 1 Black resigns}
18. Qxd4 bxc5 {(b6xc5 Qd4-e5 h7-h6 Rf1-d1 g6-g5 Rd1-d7+ Kf7-g6 g2-g4 c5-c4
Qe5-f5+) -M5/145 1 Black resigns} 19. Qd6 c4 {(c5-c4 Rf1-d1 a5-a4 Qd6-c7+
Kf7-e6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns} 20. Re1 h5 {(h7-h5
Re1-d1 h5-h4 Qd6-c7+ Kf7-f6 f2-f4 g6-g5 Rd1-d6+) -M4/245 0 Black resigns}
21. Re3 Kg7 {(Kf7-g7 Re3-f3 Kg7-h7 Qd6-e7+ Kh7-h6 Qe7-f8+ Kh6-g5 Qf8-f4+)
-M4/245 0 Black resigns} 22. Rg3 h4 {(h5-h4 Rg3xg6+ Kg7-h7 Qd6-f6 a5-a4
Qf6-g7+) -M3/245 0 Black resigns} 23. Rxg6+ Kh8 {(Kg7-h8 Qd6-f8+ Kh8-h7
Rg6-h6+) -M2/245 0 Black resigns} 24. Qf8+ Kh7 {(Kh8-h7 Rg6-h6+) -M1/245 0
Black resigns} 25. Qg7# 1-0[/pgn]

I will look at the game later today. And post the results. I hope you realize this is a statistical measurement. And any one game even without odds can show anything. That is why you can not cherry pick data. All the matches we are looking at have many games. And the match as a whole is what you look at for the data.

Meaning it is not really the game that is the most important. It is a meaningful number of moves.

And remember we are not comparing the players in the match.

We are comparing the strong side only. Or the player that starts with all his material. In the odds games. To see if your theory stands. Will that human player ever look like a computer play. As chessqueen.

I have analyzed most of the odds games on Larry's site, and so far the answer is no.

I do NOT know why you are accusing me of using an engine when the advantage is clearly close to a Rook, and my online trainer who constantly tell me that I think very precise when I am ahead in material even to nickname me Capa and always telling me that I am advancing fast, but anyway I decided to set upu another position where the advantage is close to a Rook Anyway I used Komodo Free Version this time. mwyoung, If you live in the USA I would pay for 1/3 of the plane ticket if you can bring your computer and I will replay some of these Odds, but I would like to bet $300.00 for each game using some of the Odds that I have posted here. On this game Komodo resigned because after exchanging the pawns White bishop can NOT take because I would fork both Bishops by placing my Queen on d5 !

[pgn][Event "Blitz:20'+10""]
[Site "MyTown"]
[Date "2020.09.12"]
[Round "7"]
[White "Komodo 12.1.1 64-bit"]
[Black "ChessQueen"]
[Result "0-1"]
[SetUp "1"]
[FEN "3qkbnr/pppppppp/8/8/8/8/PP1PPPPP/1NB1KBNR w Kk - 0 1"]
[PlyCount "55"]
[TimeControl "1200+10"]

1. e4 d5 2. exd5 Qxd5 3. Nc3 Qf5 4. d3 Nf6 5. Nf3 g6 6. Be2 Bg7 7. O-O O-O 8.
d4 Nd5 9. Nxd5 Qxd5 10. b3 Bxd4 11. Nxd4 Qxd4 12. a4 Rd8 13. Bc4 Qe4 14. Be3
Qc2 15. h3 Rd1 16. Rxd1 Qxd1+ 17. Kh2 Kg7 18. Kg3 a6 19. Kh2 g5 20. Bxg5 b5 0-1[/pgn]

You were tricking us. You only cheated yourself.

Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung[/size] ?

[pgn][Event "Blitz:20'+10""]
[Site " mwyoung "]
[Date "????.??.??"]
[Round "?"]
[White "Komodo 12.1.1 64-bit"]
[Black "Chess, Queen"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1"]
[PlyCount "1"]

1. d4 1-0[/pgn]

Uri is not correct. And Uri has done no testing. Uri is not even clear what or how centipawn analysis works. And the same can be said for you.

I have tested it, and in no case did it show a human as playing like a computer. Because the games was played with odds.

Now if you guys have data show it, and stop making up fact with motive.

I know you already tried but you are embarrassed to post your losing game,I know if you try hard you can beat this position in less than 30 moves
Uri is correct the more pieced you take the less you can blunder, I wonder what is the average error rate in the following position setup, so can you beat Komodo 9 with Black, mwyoung

[d]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQK3 w Qkq - 0 1']

I will post the unbiased data from Larry's site. Since you and Uri are good at making up facts. I want to see you produce any data without the program.... I suggest that if you have data then show it.

Based on your own data:
PlayKomodo: 6/14/21/27/14/10 => Average=0.15
smurfo: 37/7/15/15/3/6 => Average=0.15


1)I checked the average of these numbers and I see the 0.15 is not correct and komodo has bigger error rate than the opponent.
komodo's average (6+14+21+27+14+10)/6=92/6=0.153333...
smurfo's average (37+7+15+15+3+6)/6=83/6=0.1383333...

2)komodo had higher average in the games that it lost.
14-7 in game 2
21-15 in game 3
27-15 in game 4
14-3 in game 5
10-6 in game 6

The data supports my opinion that if the human wins the game then usually he has a smaller error rate.
I do not know how to calculate the error but if you compare between evaluation before the player's move and after the player's move the calculation
is biased against the loser because you will find mistakes of the loser when the evaluation goes down after the move(otherwise the loser does not lose).

Uri stop being stupid. Or do you just ignored everything posted. This is not a comparison of the 2 players in the match. This is about your claim of a false positive. We are only concerned about the human playing the strong side of the odds game. That was your claim, and the data shows you like to make up facts.

All humans tested show that they play like a human player. Even in odds games.
[/quote]


This is NOT my fight to defend myself, but NOT only that you accused me of using an engine to play against Komodo with Odds suitable for my level, and instead of analyzing the Bullet games of Andrew Tang which was the original discussion to see if he could really be a match against Komodo at Bullet and when I mentioned that most humans GM playing bullet games blunder an average of 3 moves in critical tactical position which engine will NOT blunder you came in and started analyzing my games accusing of using some engines because the Odds that I chose was slightly less than what Mr. Kaufman recommended me to take which was a full Rook. Another thing Now you are calling Mr. Uri Blass Stupid, which is a disrespect to another member of this forum we are all here trying to post our point of view NOT to insult other Talkchess member of this forum. As a matter of fact by looking at all the post of Uri blass, he and Mr. Kaufman are among the most knowledgeable persons on this forum. So please STOP insulting and discuss like a civilized person.

Note: My games are NOT bullet games, otherwise I would blunder at least every other move, so out of an average 50 moves games probably 18 blunders or more.

[mwyoung]

1)I checked the average of these numbers and I see the 0.15 is not correct and komodo has bigger error rate than the opponent.
komodo's average (6+14+21+27+14+10)/6=92/6=0.153333...
smurfo's average (37+7+15+15+3+6)/6=83/6=0.1383333...

2)komodo had higher average in the games that it lost.
14-7 in game 2
21-15 in game 3
27-15 in game 4
14-3 in game 5
10-6 in game 6

1.You are correct. And again showing you now nothing about how Centipawn Analysis works.
As you do not have the program, and you are just shooting from the hip.

Each score of the game is a raw score. The final average is the WEIGHTED SCORE!

2. So what! We know that player is the computer. And this is not meant for comparing the two players. This is for detecting cheating. How do you think the online sites catch the GM players and other player so fast when they are using a computer! This is a proven method and works very well. Here is the latest public example.

Press release by Chess.com
The Saint Louis Arch Bishops are the winners of the 2020 PRO Chess League championship. The Armenia Eagles have been disqualified from the 2020 season due to fair play violations.

After a thorough investigation, Chess.com's Fair Play team determined that GM Tigran L. Petrosian, who played for the Armenia Eagles, violated fair play regulations during games in both the semifinal and final matches that took place on September 25 and 27, respectively.

Chess.com and the PRO Chess League have issued a lifetime ban against Petrosian for his actions, and per section F of the league's regulations, the Armenia Eagles have been temporarily banned from participation in future PRO Chess League seasons.

PRO Chess League Commissioner IM Greg Shahade released the following statement:

"It's always unfortunate when the league is presented with evidence of fair play violations, but we stand behind the evidence presented from Chess.com's Fair Play team."

The 2020 title is the third title overall for the Saint Louis Arch Bishops and their second consecutive championship. Both the Canada Chessbrahs and China Pandas will receive $10,000, half of the sum of the second, third, and fourth place prizes.

[mwyoung]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

[lkaufman]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

This is quite interesting. In all of the matches won by Komodo, which is to say all of the matches below knight odds except for one match at f2 + c2, the human error rate was much higher than the engine error rate. Error rates for the humans were in the .23 to .42 range, for Komodo in the .06 to .15 range. So for those matches we can say it worked perfectly.
For the f2+c2 match (first match with Neuman), if we discard the final game as it was played at just one pawn (f7) handicap, the human error rate was just 0.14, Komodo 0.17. Strange indeed, although no cheating is suspected as the game was played live with our webmaster. The computer just failed to pose problems for him at that handicap back in 2015.
At knight odds on the above list, the human error rate was about 0.16 for 12 games. Komodo's error rate was about 0.18. But these were much weaker players than the ones taking smaller handicaps.
So the identification of players as human vs. engine seems to work fine at handicaps below knight odds, but it doesn't seem to be able to distinguish well at knight odds, although even at knight odds the human error rate wasn't low enough to trigger a cheating alarm. Whether that is due to the initial evaluation being too one-sided or to Uri's theory that there is a strong bias in favor of the winner isn't clear to me. I would need to know more about how the numbers are derived.

[Chessqueen]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

What ?

[mwyoung]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

This is quite interesting. In all of the matches won by Komodo, which is to say all of the matches below knight odds except for one match at f2 + c2, the human error rate was much higher than the engine error rate. Error rates for the humans were in the .23 to .42 range, for Komodo in the .06 to .15 range. So for those matches we can say it worked perfectly.
For the f2+c2 match (first match with Neuman), if we discard the final game as it was played at just one pawn (f7) handicap, the human error rate was just 0.14, Komodo 0.17. Strange indeed, although no cheating is suspected as the game was played live with our webmaster. The computer just failed to pose problems for him at that handicap back in 2015.
At knight odds on the above list, the human error rate was about 0.16 for 12 games. Komodo's error rate was about 0.18. But these were much weaker players than the ones taking smaller handicaps.
So the identification of players as human vs. engine seems to work fine at handicaps below knight odds, but it doesn't seem to be able to distinguish well at knight odds, although even at knight odds the human error rate wasn't low enough to trigger a cheating alarm. Whether that is due to the initial evaluation being too one-sided or to Uri's theory that there is a strong bias in favor of the winner isn't clear to me. I would need to know more about how the numbers are derived.

Just remember again you are using this to compare the players of the match. In a odds match.

This is not how to think about the results. The results are not for the match itself. Or for the players in the match to compare.

Games are how the chucks of data are presented for analysis. But like any statistic, the more information. The more you expose who you really are in the match. And each players are independent, and the other player can be considered nothing more then a foil.

I will agree, and this is used in non odds games. And the data has more to offer. I am just not sure in the odds format.

As you are starting with a bias positions for one side, and then comparing the results. This is a error. Even if one side is a known computer player.

But it is clear that the analysis worked. By giving 0 false positive results even in odds game matches.

Where it did fail is in detecting Komodo in all games, but this is most likely because Komodo was playing from the weaker side in all of the odds games. But it did detect Komodo in most cases. But this is not a false positive.

[Chessqueen]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

This is quite interesting. In all of the matches won by Komodo, which is to say all of the matches below knight odds except for one match at f2 + c2, the human error rate was much higher than the engine error rate. Error rates for the humans were in the .23 to .42 range, for Komodo in the .06 to .15 range. So for those matches we can say it worked perfectly.
For the f2+c2 match (first match with Neuman), if we discard the final game as it was played at just one pawn (f7) handicap, the human error rate was just 0.14, Komodo 0.17. Strange indeed, although no cheating is suspected as the game was played live with our webmaster. The computer just failed to pose problems for him at that handicap back in 2015.
At knight odds on the above list, the human error rate was about 0.16 for 12 games. Komodo's error rate was about 0.18. But these were much weaker players than the ones taking smaller handicaps.
So the identification of players as human vs. engine seems to work fine at handicaps below knight odds, but it doesn't seem to be able to distinguish well at knight odds, although even at knight odds the human error rate wasn't low enough to trigger a cheating alarm. Whether that is due to the initial evaluation being too one-sided or to Uri's theory that there is a strong bias in favor of the winner isn't clear to me. I would need to know more about how the numbers are derived.

Just remember again you are using this to compare the players of the match. In a odds match.

This is not how to think about the results. The results are not for the match itself. Or for the players in the match to compare.

Games are how the chucks of data are presented for analysis. But like any statistic, the more information. The more you expose who you really are in the match. And each players are independent, and the other player can be considered nothing more then a foil.

I will agree, and this is used in non odds games. And the data has more to offer. I am just not sure in the odds format.

As you are starting with a bias positions for one side, and then comparing the results. This is a error. Even if one side is a known computer player.

But it is clear that the analysis worked. By giving 0 false positive results even in odds game matches.

What it did fail in is detecting Komodo in all games, but this is most likely because Komodo was playing from the weaker side in all of the odds games. But it did detect Komodo in most cases. But this is not a false positive.

Just like the game that you lost with a Rook Odds which you were too embarrassed to post So the game that I won it showed a low centipawns score for me, what about the 2 games that I lost my concentration like this one ? I did NOT post my last game since I was even more embarrassed of how Komodo destroyed me, if you want to see it I will post it tomorrow since I have to go to work now. Komodo squeeze the crap out of me this time the space was too much and my position was hopeless. Instead of 7.....Qb6, after I made the move I thought that 7......Qd8 could have been better and later push my pawn to f6. I am NOT using my best computer since Komodo is only playing me, but it could be defective since it is a refurbished
https://www.officedepot.com/a/products/ ... lsrc=aw.ds

[pgn][Event "Computer chess game"]
[Site "MyTown"]
[Date "2020.12.08"]
[Round "6"]
[White "Komodo-13.3-64bit"]
[Black "ChessQueen"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbqkbnr/pp1ppppp/8/8/8/8/PPPPPPPP/1NBQKBNR w Kk - 0 1"]
[PlyCount "65"]
[TimeControl "1200+10"]

1. e4 Nc6 2. Nf3 e6 3. d4 Qa5+ 4. Nc3 Bb4 5. Bd2 d5 6. a3 Bxc3 7. Bxc3 Qb6 8.
e5 Nh6 9. Qa1 Nf5 10. b4 Bd7 11. Bd3 Nce7 12. a4 a6 13. Bd2 g6 14. O-O Qc7 15.
g4 Ng7 16. Bg5 h5 17. h3 Ra7 18. Kg2 Ng8 19. Rb1 Ra8 20. Qa3 hxg4 21. hxg4 b6
22. b5 1-0[/pgn]

[mwyoung]

The results are ready. And did the odds games trick centipawn analysis into thinking a human player was a computer player as Uri claims, with no data, or evidence. Let's find out. We have a wide range of odds games, and some really strange. We also have a wide range of players. From GM's to much lower rated players.

All odds matches were looked at with Centipawn Analysis on Larry's Komodo site.
Link here. https://komodochess.com/store/pages.php?cmsid=17

As we know the players in these matches. We know who was the humans, and how was the computers.

Here is the results. In all 11 odds matches. Centipawn analysis gave 0 false positive results. Regardless if the human player in the odds match. Won the match, or lost the match. Or the type of odds that were given.

And if you add the GM Smerdon odds match. Centipawn Analysis was 12 matches with 0 false positives.

The general indication of a computer players is a score of below 15. And scores below 10 are highly likely a computer. And in most cases even with the odds game. It flagged Komodo as the computer player.

Code: Select all

Bolzoni V:   7/23/8/9/30/48  => Average=0.21
Komodo:   17/7/24/56/14/17  => Average=0.22

Code: Select all

Komodo 9.2:   10/25  => Average=0.14
Mark Gray:   22/7  => Average=0.17

Code: Select all

Komodo 1502.00 64-bit:   10/14/16/7/25/16/29/8/21/11/5/18/21/16/13/10/10/22/27/3/30/15/15/15/15/4  => Average=0.15
GM Alex Lenderman:   41/67/21/31/18/48/23/73/27/58/51/7/17/31/27/7/36/13/29/13/10/91/7/6/7/4  => Average=0.33
:   61  => Average=0.61
New game:   78  => Average=0.78

Code: Select all

Komodo 1502.00 64-bit:   3  => Average=0.03
Komodo:   4/17/37/11/5  => Average=0.15
GM Alex Lenderman:   58/50/5/31/16/12  => Average=0.31

Code: Select all

Komodo 1494.00 64-bit:   14/2/15/11  => Average=0.10
Komodo 9.2:   14/13/3/17  => Average=0.11
John Meyer:   60/42/22/23/21/12/3/5   => Average=0.24

Code: Select all

Komodo 1538.00 64-bit:   7/12/18/3/4/14  => Average=0.09
Victor Mikhalevski:   56/58/56/35/9/15  => Average=0.41

Code: Select all

Komodo 9.2 (24 cores internet optimized):   13/10/5/12/7/2  => Average=0.08
Sergei Movsesian:   78/58/46/30/17/9  => Average=0.43

Code: Select all

GM Petr Neuman:   22/15/13/7/11/32  => Average=0.17
Komodo:   5/26/5/25/25/5  => Average=0.15

Code: Select all

Komodo 10.1:   8/4/6  => Average=0.06
Komodo 10.1 (23 cores):   7  => Average=0.07
GM Petr Neuman:   34/58/54/27  => Average=0.44

Code: Select all

Komodo:   6/7/10  => Average=0.07
Martin Petr:   13/13/45  => Average=0.23

Code: Select all

Komodo 9.2:   4/12/7/2  => Average=0.06
Larry Gilden:   12/56/53/45  => Average=0.38

This is quite interesting. In all of the matches won by Komodo, which is to say all of the matches below knight odds except for one match at f2 + c2, the human error rate was much higher than the engine error rate. Error rates for the humans were in the .23 to .42 range, for Komodo in the .06 to .15 range. So for those matches we can say it worked perfectly.
For the f2+c2 match (first match with Neuman), if we discard the final game as it was played at just one pawn (f7) handicap, the human error rate was just 0.14, Komodo 0.17. Strange indeed, although no cheating is suspected as the game was played live with our webmaster. The computer just failed to pose problems for him at that handicap back in 2015.
At knight odds on the above list, the human error rate was about 0.16 for 12 games. Komodo's error rate was about 0.18. But these were much weaker players than the ones taking smaller handicaps.
So the identification of players as human vs. engine seems to work fine at handicaps below knight odds, but it doesn't seem to be able to distinguish well at knight odds, although even at knight odds the human error rate wasn't low enough to trigger a cheating alarm. Whether that is due to the initial evaluation being too one-sided or to Uri's theory that there is a strong bias in favor of the winner isn't clear to me. I would need to know more about how the numbers are derived.

Just remember again you are using this to compare the players of the match. In a odds match.

This is not how to think about the results. The results are not for the match itself. Or for the players in the match to compare.

Games are how the chucks of data are presented for analysis. But like any statistic, the more information. The more you expose who you really are in the match. And each players are independent, and the other player can be considered nothing more then a foil.

I will agree, and this is used in non odds games. And the data has more to offer. I am just not sure in the odds format.

As you are starting with a bias positions for one side, and then comparing the results. This is a error. Even if one side is a known computer player.

But it is clear that the analysis worked. By giving 0 false positive results even in odds game matches.

What it did fail in is detecting Komodo in all games, but this is most likely because Komodo was playing from the weaker side in all of the odds games. But it did detect Komodo in most cases. But this is not a false positive.

Just like the game that you lost with a Rook Odds which you were too embarrassed to post So the game that I won it showed a low centipawns score for me, what about the 2 games that I lost my concentration like this one ? I did NOT post my last game since I was even more embarrassed of how Komodo destroyed me, if you want to see it I will post it tomorrow since I have to go to work now. Komodo squeeze the crap out of me this time the space was too much and my position was hopeless. Instead of 7.....Qb6, after I made the move I thought that 7......Qd8 could have been better and later push my pawn to f6. I am NOT using my best computer since Komodo is only playing me, but it could be defective since it is a refurbished
https://www.officedepot.com/a/products/ ... lsrc=aw.ds

[pgn][Event "Computer chess game"]
[Site "MyTown"]
[Date "2020.12.08"]
[Round "6"]
[White "Komodo-13.3-64bit"]
[Black "ChessQueen"]
[Result "1-0"]
[SetUp "1"]
[FEN "rnbqkbnr/pp1ppppp/8/8/8/8/PPPPPPPP/1NBQKBNR w Kk - 0 1"]
[PlyCount "65"]
[TimeControl "1200+10"]

1. e4 Nc6 2. Nf3 e6 3. d4 Qa5+ 4. Nc3 Bb4 5. Bd2 d5 6. a3 Bxc3 7. Bxc3 Qb6 8.
e5 Nh6 9. Qa1 Nf5 10. b4 Bd7 11. Bd3 Nce7 12. a4 a6 13. Bd2 g6 14. O-O Qc7 15.
g4 Ng7 16. Bg5 h5 17. h3 Ra7 18. Kg2 Ng8 19. Rb1 Ra8 20. Qa3 hxg4 21. hxg4 b6
22. b5 1-0[/pgn]

I am not interested in the games you present after you know I was looking at them with centipawn analysis...

Just make up your own facts as usual. And present your own data.