TalkChess.com

carldaman wrote: ↑Sun Oct 13, 2019 1:53 am Thanks +1 for the new release. The style is there as before, but contempt seems to be broken. I use centipawn score, and the scores don't reflect any sort of activated contempt, no matter what it's set to.

Ah.

Also, there seems to be two Dynamic Contempts, Dynamic_Contempt and Dynamic Contempt, I found that weird.

If you are using my version, I turn off every form of contempt by default. You can turn it back on with the UCI options.

I turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

Dann Corbit wrote: ↑Sun Oct 13, 2019 9:26 am If you are using my version, I turn off every form of contempt by default. You can turn it back on with the UCI options.

I turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

+3 Dann !
I compiled your version for my Mac - - compiled perfectly.
Many thanks.
One question - I use for analysis only - would you suggest all default options ?
Please !

Keep getting illegal move on all games i play both in Winboard and Arena
The centpawn setting and uci sleep option make no diffirence

Uci Settings are

Hash=128
Score_Output=CentiPawn
UCI_Sleep=true
UCI_LimitStrength=true
UCI_Elo=1200

[Event "Computer Chess Game"]
[Site "JOE3"]
[Date "2019.10.13"]
[Round "1"]
[White "NITRA"]
[Black "Honey X6"]
[Result "1-0"]
[TimeControl "40/60"]
[Annotator "1. +0.50 12... -2.84"]

1. d4 {+0.50/7} e6 {+0.10/2 1.6} 2. Bf4 {+0.16/6 0.7} d5 {+0.87/1 1.6} 3.
e3 {+0.04/6 0.7} c6 {+0.88/1 1.6} 4. Nc3 {-0.21/6 0.7} h5 {+0.01/1 1.6} 5.
Bd3 {+0.45/6 0.7} Nf6 {+0.05/1 1.6} 6. Qf3 {+0.64/6 0.7} Qb6 {+0.27/1 1.6}
7. O-O-O {+0.97/6 0.7} Be7 {+0.04/1 1.6} 8. Qg3 {+0.82/6 0.8} e5
{-1.23/1 1.6} 9. Bxe5 {+2.26/5 0.7} Nbd7 {-2.02/1 1.6} 10. Bc7
{+4.65/6 0.7} h4 {-2.64/1 1.6} 11. Qxg7 {+4.64/6 0.7} Rg8 {-0.66/1 1.6} 12.
Qxg8+ {+4.55/6 0.7} Nxg8 {-2.84/3 1.6} 13. Bxb6 {+4.48/7 0.7} axb6
{-2.84/1 1.6} 14. Kb1 {+4.52/6 0.6} b5 {-2.54/1 1.6} 15. Nf3 {+4.53/6 0.7}
Nb6 {-2.94/1 1.6} 16. Rde1 {+4.92/6 0.8} Bg4 {-2.84/1 1.6} 17. Ne5
{+5.03/6 1.1} Bh5 {-2.82/1 1.6} 18. g4 {+5.38/6 0.8} hxg3 {-2.81/1 1.6} 19.
hxg3 {+5.15/5 0.6} b4 {-3.35/1 1.6} 20. Nb5 {+6.35/6 0.6} cxb5
{-3.83/1 1.6} 21. Bxb5+ {+5.84/6 0.6} Kd8 {-4.11/4 1.6} 22. Rxh5
{+6.61/6 0.6} f6 {-4.21/1 1.6} 23. Nf7+ {+7.94/7 0.6}
{polyglot: resign (illegal engine move by black: a1h5)} 1-0


Problem signature:
Problem Event Name: APPCRASH
Application Name: Honey-X6_x64_modern.exe
Application Version: 0.0.0.0
Application Timestamp: 002ee8a0
Fault Module Name: StackHash_a807
Fault Module Version: 6.1.7601.24291
Fault Module Timestamp: 5be7857f
Exception Code: c0000374
Exception Offset: 00000000000bf302
OS Version: 6.1.7601.2.1.0.256.1
Locale ID: 2057
Additional Information 1: a807
Additional Information 2: a807dae0bead45355a1c1c94937c0f56
Additional Information 3: f532
Additional Information 4: f532aa66a326bdd58385dd941c6eb514

maxdeg wrote: ↑Sun Oct 13, 2019 11:21 am

Dann Corbit wrote: ↑Sun Oct 13, 2019 9:26 am If you are using my version, I turn off every form of contempt by default. You can turn it back on with the UCI options.

I turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

+3 Dann !
I compiled your version for my Mac - - compiled perfectly.
Many thanks.
One question - I use for analysis only - would you suggest all default options ?
Please !

You probably don't want the logging data ( but even in your case it does leave a permanent record for study ). If you decide to keep it the default file name is searchlog.txt. Printing human readable analysis to disk is controlled by the logging option.

Dann Corbit wrote: ↑Sun Oct 13, 2019 9:26 amI turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

Do you analyze games to find the best moves against opponents you're playing? In those cases what is the "right answer" depends on the opponent (Contempt 0 might be best against a superior opponent), but in general the way to win games is to get into positions where the opponent needs some interactive work to find the "right answer", and it seems Contempt really helps with that, as long as you hit its nonsense suggestions like nails. 24 seems about optimal in default Stockfish, decreasing it has only led to Stockfish being more... timid.

Dann Corbit wrote: ↑Sun Oct 13, 2019 6:22 pm

maxdeg wrote: ↑Sun Oct 13, 2019 11:21 am

Dann Corbit wrote: ↑Sun Oct 13, 2019 9:26 am If you are using my version, I turn off every form of contempt by default. You can turn it back on with the UCI options.

I turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

+3 Dann !
I compiled your version for my Mac - - compiled perfectly.
Many thanks.
One question - I use for analysis only - would you suggest all default options ?
Please !

You probably don't want the logging data ( but even in your case it does leave a permanent record for study ). If you decide to keep it the default file name is searchlog.txt. Printing human readable analysis to disk is controlled by the logging option.

Thanks Dann ..

MikeB wrote: ↑Tue Oct 08, 2019 9:42 pm all versions have all bell and whistles, including play by FIDE Elo (or CCRL Elo) and improved adaptive and variety play

thanks for this release Mike. I've played against during this week-end and I think that yours is one of the best implementation of Elo levels. I think the official Stockfish should use it too.

dannyb wrote: ↑Sun Oct 13, 2019 9:05 pm

MikeB wrote: ↑Tue Oct 08, 2019 9:42 pm all versions have all bell and whistles, including play by FIDE Elo (or CCRL Elo) and improved adaptive and variety play

thanks for this release Mike. I've played against during this week-end and I think that yours is one of the best implementation of Elo levels. I think the official Stockfish should use it too.

thanks - but Stockfish has a very good implementation with much better (simpler) code that what I wrote.

Dann Corbit wrote: ↑Sun Oct 13, 2019 9:26 am If you are using my version, I turn off every form of contempt by default. You can turn it back on with the UCI options.

I turn it off because I use chess engines for analysis and rarely for game play. And when I play them, the contempt is not important anyway. And when I analyze, I want the right answer.

I can confirm that contempt does work for Polygon, when activated thru UCI options, but does not for Honey/Bluefish.