Top engines with or without EGTB

[Chessqueen]

Hi Günther,

these messages created such impressions, yes!

On the other hand is every point of view interesting.
This creates new ideas for things we can do in the time we have.

Code: Select all

   1 ShashChess 21.1 NN         :  3345.35    357    62.7   92   264     1   224.0   73.9  19.48  3238.90  17.11   36.4
   2 Stockfish 15 NN            :  3341.51    360    62.4   89   271     0   224.5   75.3  18.64  3237.99  17.31   36.2
   3 Koivisto 8.6 NN            :  3324.00    487    62.8  132   348     7   306.0   71.5  15.91  3229.65  17.33   38.5
   4 Fire 8 MC.3 NN             :  3323.35    332    61.1   85   236    11   203.0   71.1  19.06  3232.12  17.37   36.7
   5 Dragon 2.6 NN (Komodo)     :  3319.53    428    59.1   80   346     2   253.0   80.8  15.95  3243.01  17.35   36.3
   6 Berserk 8.5.1 NN           :  3318.86    454    62.1  125   314    15   282.0   69.2  16.28  3230.38  17.28   38.3
   7 rofChade 2.321 NN dev      :  3305.90    412    59.7   98   296    18   246.0   71.8  16.52  3232.37  17.18   37.1
   8 Revenge 2.0 NN             :  3300.41    287    58.4   66   203    18   167.5   70.7  20.22  3235.88  16.97   35.3
   9 Caissa 0.5 NNSf            :  3299.69    408    60.7  107   281    20   247.5   68.9  17.35  3220.45  17.16   38.2
  10 Seer 2.4.0 NN              :  3291.05    356    58.8   93   233    30   209.5   65.4  17.84  3228.81  17.27   37.8

Endgame stats from Caissa 0.5.
The strength from Caissa is the endgame, no doubt about it ... to 100% clear!
If people like endgames Caissa is interesting.
Nothing to do with Stockfish Network or not.
Wit the same Network file Caissa made +200 Elo from version 0.4 to 0.5.
And Endgame stats from Caissa 0.4 are strong also.

RubiChess is not inside in TOP-10 endgame stats from my "run-4"
On 13 ... 15 Elo lesser.

SF is playing perfect chess for me.
But Caissa with SF NN is playing ... for me ... most boring chess I ever saw.
Other like endgames and thinking ... the opinion from Frank is wrong.
I am looking more for aggressives engines.

Chessqueen should have a look on Seer 2.5.0 ...

Endgame stats from "run-6" ...

Code: Select all

   # Player                     :      Elo  Games  Score%  won  draw  lost  Points  Draw%  Error   OppAvg   OppE   OppD
   1 Stockfish 020622 NN dev    :  3341.98    364    62.1   88   276     0   226.0   75.8  18.54  3240.17  17.12   36.1
   2 Seer 2.5.0 NN              :  3334.47    373    64.5  120   241    12   240.5   64.6  18.57  3224.73  17.13   37.8
   3 Fire 8 MC.3 NN             :  3332.14    358    63.0   99   253     6   225.5   70.7  19.43  3226.82  17.19   35.9
   4 Koivisto 8.9 NN            :  3320.92    464    62.3  118   342     4   289.0   73.7  16.06  3229.23  17.38   38.3
   5 Dragon 3 NN (Komodo)       :  3318.90    398    58.8   73   322     3   234.0   80.9  16.52  3240.93  17.30   36.5
   6 Berserk 9 NN dev3          :  3311.05    507    60.9  120   378     9   309.0   74.6  15.55  3227.61  17.20   38.1
   7 Revenge 2.0 NN             :  3306.92    289    59.3   68   207    14   171.5   71.6  20.06  3235.44  17.01   34.7
   8 rofChade 3.0 NN            :  3304.99    378    60.4   90   277    11   228.5   73.3  18.21  3227.98  17.20   38.3
   9 Arasan 23.3 NN             :  3294.76    340    59.1   75   252    13   201.0   74.1  17.52  3225.31  17.03   36.9
  10 Minic 3.22 NN              :  3292.49    440    60.8  114   307    19   267.5   69.8  16.47  3214.48  17.12   38.1

Not the topic I like ... the endgames.
But I have different material about it and here Seer is at the moment here most interesting.
Fact is fact ...

And Seer is not playing boring chess.
More defensive as others with same strength. In the late mid-games Seer is very danger!

I believe the Caissa programmer can do a lot and his main interest seems the endgames.
Example: Such a program Ktulu was ... for Rahman are endgames with rooks most important as all other things together.

Stats become more and more importance.
Often we can see only with stats.

What I like to say is:
Caissa is not such a program I like but for sure a program many others will like.
People have more interest on endgames.

Personal opinion:
Made more sense to looking in Seer as in Caissa for interesting endgame positions.

Best
Frank

Yes Seer 2.50 has good endgame knowledge, this is a forced Draw, since the h pawn will never be able to promotote, and if it did not take the Bh3! Bait, the Bishop would have taken the g2 pawn with same result a Draw !

[pgn][Event "Endgame knowledge"]
[Site "DESKTOP-OFQ3C0P"]
[Date "2022.07.12"]
[Round "?"]
[White "Seer_v2.5_x64_avx2_popcnt"]
[Black "Seer_v2.5_x64_avx2_popcnt"]
[Result "*"]
[BlackElo "?"]
[Time "09:02:14"]
[WhiteElo "?"]
[TimeControl "360+3"]
[SetUp "1"]
[FEN "r7/8/8/5bk1/8/5B2/5RPP/6K1 b - - 0 18"]
[Termination "unterminated"]
[PlyCount "6"]


18. ... Ra1+ {(Ra8-a1+ Rf2-f1 Ra1xf1+ Kg1xf1 Kg5-h4 Kf1-e2 Bf5-h3 g2-g3+
Kh4-g5 Ke2-e3 Kg5-f5 Ke3-f2 Kf5-g6 Bf3-g2 Bh3-d7 h2-h4 Kg6-h7 Bg2-e4+
Kh7-g7 Kf2-f3 Kg7-h6 Kf3-e3 Kh6-h5 Ke3-f4 Bd7-c8 Be4-f3+ Kh5-h6 Bf3-g2
Bc8-d7 Bg2-e4 Kh6-h5 Be4-d3 Bd7-c8 Bd3-f1 Kh5-h6 Kf4-e3 Kh6-g6 Bf1-e2
Kg6-h6 Ke3-e4 Bc8-b7+ Ke4-e3) -3.21/32 42} 19. Rf1 {(Rf2-f1 Ra1xf1+ Kg1xf1
Kg5-h4 Kf1-e2 Bf5-h3 g2-g3+ Kh4-g5 Ke2-e3 Kg5-f5 Ke3-f2 Kf5-g6 Bf3-g2
Bh3-d7 h2-h4 Kg6-h7 Bg2-e4+ Kh7-g7 Kf2-f3 Kg7-h6 Kf3-e3 Kh6-h5 Ke3-f4
Bd7-c8 Be4-f3+ Kh5-h6 Bf3-d5 Bc8-d7 Kf4-e3 Kh6-g7 Bd5-f3 Kg7-h6 Bf3-e2
Kh6-h7 Be2-f1 Kh7-g6 Ke3-f4 Kg6-h5 Bf1-e2+) +3.21/34 26} Rxf1+ {(Ra1xf1+
Kg1xf1 Kg5-h4 Kf1-e2 Bf5-h3 g2-g3+ Kh4-g5 Ke2-f2 Kg5-f5 Bf3-g2 Bh3-g4 h2-h4
Kf5-f6 Bg2-f3 Bg4-d7 Kf2-e3 Kf6-g6 Ke3-f4 Kg6-h6 Bf3-e4 Kh6-h5 Be4-f5
Bd7-b5 Bf5-c2 Bb5-d7 Bc2-e4 Bd7-c8 Be4-f5 Bc8-a6 Bf5-g4+ Kh5-h6 Kf4-e3
Ba6-f1 Bg4-e2 Bf1-h3 Ke3-f3 Bh3-e6 Kf3-f2 Kh6-g6 Kf2-e3 Be6-d7) -3.21/28
22} 20. Kxf1 {(Kg1xf1 Kg5-h4 Kf1-e2 Bf5-h3 g2-g3+ Kh4-g5 Ke2-e3 Kg5-f5
Bf3-e2 Kf5-g5 Ke3-e4 Bh3-c8 h2-h4+ Kg5-h6 Ke4-e3 Kh6-g7 Be2-d3 Kg7-h6
Ke3-f4 Kh6-h5 Bd3-f5 Bc8-a6 Bf5-g4+ Kh5-h6 Bg4-f3 Ba6-c8 Bf3-g2 Bc8-d7
Bg2-e4 Kh6-h5 Be4-d5 Kh5-h6 Kf4-e3 Kh6-g7 Bd5-g2 Kg7-g6 Bg2-e4+ Kg6-h6
Be4-f3 Bd7-f5 Ke3-f4) +3.21/31 13} Bh3 21. gxh3 {(g2xh3 Kg5-h6 Bf3-g4
Kh6-g7 Kf1-f2 Kg7-h7 Bg4-f5+ Kh7-h8 Kf2-f3 Kh8-g7 Kf3-g4 Kg7-f6 Kg4-f4
Kf6-e7 Kf4-g5 Ke7-f7 Bf5-g4 Kf7-g8 Kg5-g6 Kg8-f8 Bg4-f5 Kf8-g8 Bf5-e4
Kg8-h8 Be4-f3 Kh8-g8 h3-h4 Kg8-h8 h4-h5) +15.13/34 19} *[/pgn]

[Cardoso]

and now he hates RubiChess and Ethereal for telling him the truth.

Sometimes people don't accept the truth if they don't like the truth, I've experienced this many many times in my life.
It's frequent human behavior. A public figure in my country said people sometimes behave like they want to be fooled and I can confirm that.
Someone should write a book about that.

[Chessqueen]

and now he hates RubiChess and Ethereal for telling him the truth.

Sometimes people don't accept the truth if they don't like the truth, I've experienced this many many times in my life.
It's frequent human behavior. A public figure in my country said people sometimes behave like they want to be fooled and I can confirm that.
Someone should write a book about that.

Escuse me but you took his statement as legit, what truth did the programmers of Rubichess stated to me? At the contrary he stated that based on what I showed him that he will work to improve the KpB and other easy to fix endgame algorithm that do not require extensive endgame knowkedge