Nunn2 A-standings after 12 positions out of 20

[Heinz van Kempen]

Geschrieben von: / Posted by: Heinz van Kempen at 30 January 2004 07:10:20:



Hi all :-),
Mystically I stumbled across that "How to..." button [[smile]]

1 Movei 00_8_158               196.5 / 288   
2 Gothmog 0.4.6a               191.5 / 288   
3 The Baron 1.2.0b8a           174.0 / 288   
4 Queen 2.42a                  167.5 / 288   
5 Terra 3.3B3                  158.5 / 288   
6 Cerebro 1.23                 148.5 / 288   
7 Chispa 4.032                 146.0 / 288   
8 Dorky 4.0b                   139.5 / 288   
9 Snitch 0.3.5                 132.5 / 288   
10 Nullmover 0.23b             117.0 / 288   
11 Black Bishop 0.9.2d         115.0 / 288   
12 Waster 0.11                  96.0 / 288   
13 Chiron 0.36                  89.5 / 288 
So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
For The Baron there are stronger versions. I will experiment a bit. 
Concerning Cerebro I got a rating after 200 Nunn position games with 40 moves in 20 minutes that is 50 points better than that posted in the first tiny Blitz rating list some days ago. A had the same phenomenon when testing Ruffian, Deep Sjeng, Aristarch, Junior, Rebel and others. And of course there are also some that are weaker with longer time controls compared to Blitz, I do not want to mention here. 
Anyone who is interested can &#receive an Excel-sheet, where I compared ELO for top engines down to Knight Dreamer 3.2 with 272 games each (combined in a big tournament) and with 25 minutes each side to their Blitz performance, most with 1000 games. 
Games I already offered before to any author interested. Dann Corbit received them all (upload of different files took some hours). They are now available on his fine ftp site.
You can find the data here:
ftp://cap.connx.com/Contrib/Blitz1.rar
ftp://cap.connx.com/Contrib/Blitz2.rar
ftp://cap.connx.com/Contrib/Blitz3.rar
ftp://cap.connx.com/Contrib/Blitz4.rar
ftp://cap.connx.com/Contrib/Blitz6.rar
ftp://cap.connx.com/Contrib/25min.rar

Have a nice day
Heinz

[Olivier Deville]

Geschrieben von: / Posted by: Olivier Deville at 30 January 2004 07:35:20:
Als Antwort auf: / In reply to: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Heinz van Kempen at 30 January 2004 07:10:20:

>Hi all :-),
>Mystically I stumbled across that "How to..." button [[smile]]
>
>1 Movei 00_8_158               196.5 / 288   
>2 Gothmog 0.4.6a               191.5 / 288   
>3 The Baron 1.2.0b8a           174.0 / 288   
>4 Queen 2.42a                  167.5 / 288   
>5 Terra 3.3B3                  158.5 / 288   
>6 Cerebro 1.23                 148.5 / 288   
>7 Chispa 4.032                 146.0 / 288   
>8 Dorky 4.0b                   139.5 / 288   
>9 Snitch 0.3.5                 132.5 / 288   
>10 Nullmover 0.23b             117.0 / 288   
>11 Black Bishop 0.9.2d         115.0 / 288   
>12 Waster 0.11                  96.0 / 288   
>13 Chiron 0.36                  89.5 / 288 
>So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
>For The Baron there are stronger versions. I will experiment a bit. 
>Concerning Cerebro I got a rating after 200 Nunn position games with 40 moves in 20 minutes that is 50 points better than that posted in the first tiny Blitz rating list some days ago. A had the same phenomenon when testing Ruffian, Deep Sjeng, Aristarch, Junior, Rebel and others. And of course there are also some that are weaker with longer time controls compared to Blitz, I do not &#want to mention here. 
>Anyone who is interested can receive an Excel-sheet, where I compared ELO for top engines down to Knight Dreamer 3.2 with 272 games each (combined in a big tournament) and with 25 minutes each side to their Blitz performance, most with 1000 games. 
>Games I already offered before to any author interested. Dann Corbit received them all (upload of different files took some hours). They are now available on his fine ftp site.
>You can find the data here:
>ftp://cap.connx.com/Contrib/Blitz1.rar
>ftp://cap.connx.com/Contrib/Blitz2.rar
>ftp://cap.connx.com/Contrib/Blitz3.rar
>ftp://cap.connx.com/Contrib/Blitz4.rar
>ftp://cap.connx.com/Contrib/Blitz6.rar
>ftp://cap.connx.com/Contrib/25min.rar
>
>Have a nice day
>Heinz
>
Hi Heinz ! I am interested of course to receive this Excel sheet
Olivier


ChessWar

[Heinz van Kempen]

Geschrieben von: / Posted by: Heinz van Kempen at 30 January 2004 07:51:59:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Olivier Deville at 30 January 2004 07:35:20:

Hi Heinz ! I am interested of course to receive this Excel sheet
Olivier

Hi Olivier,
you will receive it of course this evening. I will collect all requests where there are no special questions and send to all at once together with the tables for the huge tournament where the progress can be seen after each round. Tournament took some months and of course there are not for all the latest in as it was held a few months ago. So Ruffian 1.0.1 for example and Shredder 7.04. But most were not updated since then. To be more precise there are the following for the big tournament with 272 games each and 25 min.:
Fritz 8
Shredder 7.04
Junior 8
CM 9000 Milan 2
Ruffian 1.0.1
Deep Sjeng 1.5
Rebel 12 PR
Chess Tiger 15.0
Hiarcs 8
Aristarch 4.21
List 504
SOS 3 for Arena
Nimzo 7.32
Gandalf 4.32
Crafty 19.03 Capa
Yace Paderborn
Little Goliath Nemesis
Green Light Chess 3.00
Smar Think 0.16b++
Pharaon 2.62
Anaconda 1.0
Little Goliath 2000 v.3.9
Delfi 4.1
Delfi 4.2
Pepito v1.59
Wild Cat 2.79
Knight Dreamer 3.2
Abrok 5.0
Tao 5.4
Ktulu 3.6
Amyan 1.59
I think that newer versions like Ruffian 2.0.0 will still be much better with longer time controls.
Best Regards
Heinz

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 30 January 2004 07:59:23:
Als Antwort auf: / In reply to: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Heinz van Kempen at 30 January 2004 07:10:20:

>Hi all :-),
>Mystically I stumbled across that "How to..." button [[smile]]
>
>1 Movei 00_8_158               196.5 / 288   
>2 Gothmog 0.4.6a               191.5 / 288   
>3 The Baron 1.2.0b8a           174.0 / 288   
>4 Queen 2.42a                  167.5 / 288   
>5 Terra 3.3B3                  158.5 / 288   
>6 Cerebro 1.23                 148.5 / 288   
>7 Chispa 4.032                 146.0 / 288   
>8 Dorky 4.0b                   139.5 / 288   
>9 Snitch 0.3.5                 132.5 / 288   
>10 Nullmover 0.23b             117.0 / 288   
>11 Black Bishop 0.9.2d         115.0 / 288   
>12 Waster 0.11                  96.0 / 288   
>13 Chiron 0.36                  89.5 / 288 
>So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
>For The Baron there are stronger versions. I will experiment a bit. 
It is fair to add that not only The Baron, but also the versions of
Queen and Terra which play in this tournament seem to have some nasty
bugs.  I also suspect there is something wrong with Dorky, which 
should normally perform much better in this field.
Tord
>

[Heinz van Kempen]

Geschrieben von: / Posted by: Heinz van Kempen at 30 January 2004 08:06:55:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Tord Romstad at 30 January 2004 07:59:23:

It is fair to add that not only The Baron, but also the versions of
Queen and Terra which play in this tournament seem to have some nasty
bugs.  I also suspect there is something wrong with Dorky, which
should normally perform much better in this field.
Tord

Hi Tord ,
thanks for adding this, we already discussed it a bit. For Terra I mentioned this time managment bug concerning the Fischer clock in an earlier post.
Matt anyway seems to be satisfied. Like I am seeing Dorky improved over the last Nunn positions and you can be sure, that I won´t run tournaments with 480 games for each engine, if I am not convinced from experiences over the last years, that there might still be enormous changes even after 300 games.
Best Regards
Heinz

[Sergio Martinez]

Geschrieben von: / Posted by: Sergio Martinez at 30 January 2004 09:50:33:
Als Antwort auf: / In reply to: ELO comparison Blitz and longer time control geschrieben von: / posted by: Heinz van Kempen at 30 January 2004 07:51:59:

Hi Heinz ! I am interested of course to receive this Excel sheet
Olivier

Hi Olivier,
you will receive it of course this evening. I will collect all requests where there are no special questions and send to all at once together with the tables for the huge tournament where the progress can be seen after each round. Tournament took some months and of course there are not for all the latest in as it was held a few months ago. So Ruffian 1.0.1 for example and Shredder 7.04. But most were not updated since then. To be more precise there are the following for the big tournament with 272 games each and 25 min.:
Fritz 8
Shredder 7.04
Junior 8
CM 9000 Milan 2
Ruffian 1.0.1
Deep Sjeng 1.5
Rebel 12 PR
Chess Tiger 15.0
Hiarcs 8
Aristarch 4.21
List 504
SOS 3 for Arena
Nimzo 7.32
Gandalf 4.32
Crafty 19.03 Capa
Yace Paderborn
Little Goliath Nemesis
Green Light Chess 3.00
Smar Think 0.16b++
Pharaon 2.62
Anaconda 1.0
Little Goliath 2000 v.3.9
Delfi 4.1
Delfi 4.2
Pepito v1.59
Wild Cat 2.79
Knight Dreamer 3.2
Abrok 5.0
Tao 5.4
Ktulu 3.6
Amyan 1.59
I think that newer versions like Ruffian 2.0.0 will still be much better with longer time controls.
Best Regards
Heinz

Hi Heinz. Please send me the excel too
Next tournament will be equal.I dont see a winner (maybe Shredder,Ruffian,Deep Sjeng,Chess Tiger...)
Best Regards





http://usuarios.lycos.es/stalewb/index.htm

[Heinz van Kempen]

Geschrieben von: / Posted by: Heinz van Kempen at 30 January 2004 10:02:51:
Als Antwort auf: / In reply to: Re: ELO comparison Blitz and longer time control geschrieben von: / posted by: Sergio Martinez at 30 January 2004 09:50:33:

Hi Heinz. Please send me the excel too
Next tournament will be equal.I dont see a winner (maybe Shredder,Ruffian,Deep Sjeng,Chess Tiger...)
Best Regards

Hola Sergio ,
you are added. Robert Allgeuer, too.
With all those rating, please let us not forget to watch the games, especially with more time they are already on a level, that I hardly understand all what is going on. I mean there are engines, playing a daring style like Gothmog. I like it, although it might cost rating points as not all sacrifices prove to be sound.
Best Regards
Heinz

[Andreas Herrmann]

Geschrieben von: / Posted by: Andreas Herrmann at 30 January 2004 11:27:12:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Tord Romstad at 30 January 2004 07:59:23:

>>Hi all :-),
>>Mystically I stumbled across that "How to..." button [[smile]]
>>
>>1 Movei 00_8_158               196.5 / 288   
>>2 Gothmog 0.4.6a               191.5 / 288   
>>3 The Baron 1.2.0b8a           174.0 / 288   
>>4 Queen 2.42a                  167.5 / 288   
>>5 Terra 3.3B3                  158.5 / 288   
>>6 Cerebro 1.23                 148.5 / 288   
>>7 Chispa 4.032                 146.0 / 288   
>>8 Dorky 4.0b                   139.5 / 288   
>>9 Snitch 0.3.5                 132.5 / 288   
>>10 Nullmover 0.23b             117.0 / 288   
>>11 Black Bishop 0.9.2d         115.0 / 288   
>>12 Waster 0.11                  96.0 / 288   
>>13 Chiron 0.36                  89.5 / 288 
>>So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
>>For The Baron there are stronger versions. I will experiment a bit. 
>It is fair to add that not only The Baron, but also the versions of
>Queen and Terra which play in this tournament seem to have some nasty
>bugs.  I also suspect there is something wrong with Dorky, which 
>should normally perform much better in this field.
>Tord
>>
Hi Tord,
which chess prog has no nasty bugs . In BlackBishop, i have found in the last days two heavy bugs in the search. One of these bugs, i have had since about 2 years in my engine. So after the IPCCC i will write again more debug code for the search, although about 80% of my source is already code for debugging.
best wishes
Andreas

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 30 January 2004 12:36:31:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Andreas Herrmann at 30 January 2004 11:27:12:

Hi Tord,
which chess prog has no nasty bugs . In BlackBishop, i have found in the last days two heavy bugs in the search. One of these bugs, i have had since about 2 years in my engine. So after the IPCCC i will write again more debug code for the search, although about 80% of my source is already code for debugging.

>>>Hi all :-),
>>>Mystically I stumbled across that "How to..." button [[smile]]
>>>
>>>1 Movei 00_8_158               196.5 / 288   
>>>2 Gothmog 0.4.6a               191.5 / 288   
>>>3 The Baron 1.2.0b8a           174.0 / 288   
>>>4 Queen 2.42a                  167.5 / 288   
>>>5 Terra 3.3B3                  158.5 / 288   
>>>6 Cerebro 1.23                 148.5 / 288   
>>>7 Chispa 4.032                 146.0 / 288   
>>>8 Dorky 4.0b                   139.5 / 288   
>>>9 Snitch 0.3.5                 132.5 / 288   
>>>10 Nullmover 0.23b             117.0 / 288   
>>>11 Black Bishop 0.9.2d         115.0 / 288   
>>>12 Waster 0.11                  96.0 / 288   
>>>13 Chiron 0.36                  89.5 / 288 
>>>So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
>>>For The Baron there are stronger versions. I will experiment a bit. 
>>It is fair to add that not only The Baron, but also the versions of
>>Queen and Terra which play in this tournament seem to have some nasty
>>bugs.  I also suspect there is something wrong with Dorky, which 
>>should normally perform much better in this field.
>>Tord
>>>
Yes, of course we all have bugs.
What I meant to say was that Baron, Queen, Terra and Dorky all seem tore buggy than usual, and therefore don't score as well as they
should usually do in such a field. It is possible that this has happened
to some of the other participants as well, but I don't know them well
enough to decide.
In particular, it is clear that something is very, very wrong with The
Baron. Normally Gothmog scores exactly 0% against Baron, but in the 24
games played so far in this tournament Gothmog has a +1 score!
Tord

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 30 January 2004 13:02:34:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Tord Romstad at 30 January 2004 12:36:31:

Hi Tord,
which chess prog has no nasty bugs . In BlackBishop, i have found in the last days two heavy bugs in the search. One of these bugs, i have had since about 2 years in my engine. So after the IPCCC i will write again more debug code for the search, although about 80% of my source is already code for debugging.

Yes, of course we all have bugs.
What I meant to say was that Baron, Queen, Terra and Dorky all seem to
be more buggy than usual, and therefore don't score as well as they
should usually do in such a field. It is possible that this has happened
to some of the other participants as well, but I don't know them well
enough to decide.
In particular, it is clear that something is very, very wrong with The
Baron. Normally Gothmog scores exactly 0% against Baron, but in the 24
games played so far in this tournament Gothmog has a +1 score!
Tord

>>>>Hi all :-),
>>>>Mystically I stumbled across that "How to..." button [[smile]]
>>>>
>>>>1 Movei 00_8_158               196.5 / 288   
>>>>2 Gothmog 0.4.6a               191.5 / 288   
>>>>3 The Baron 1.2.0b8a           174.0 / 288   
>>>>4 Queen 2.42a                  167.5 / 288   
>>>>5 Terra 3.3B3                  158.5 / 288   
>>>>6 Cerebro 1.23                 148.5 / 288   
>>>>7 Chispa 4.032                 146.0 / 288   
>>>>8 Dorky 4.0b                   139.5 / 288   
>>>>9 Snitch 0.3.5                 132.5 / 288   
>>>>10 Nullmover 0.23b             117.0 / 288   
>>>>11 Black Bishop 0.9.2d         115.0 / 288   
>>>>12 Waster 0.11                  96.0 / 288   
>>>>13 Chiron 0.36                  89.5 / 288 
>>>>So a hard fight between Movei and Gothmog for the pole position in the first Nunn tournament.
>>>>For The Baron there are stronger versions. I will experiment a bit. 
>>>It is fair to add that not only The Baron, but also the versions of
>>>Queen and Terra which play in this tournament seem to have some nasty
>>>bugs.  I also suspect there is something wrong with Dorky, which 
>>>should normally perform much better in this field.
>>>Tord
>>>>
It may be results of different positions.
I am surprised to read that Gothmog normally scores 0% against the baron.
Gothmog even scores better against crafty or smarthink when they are stronger than the baron.
I also do not think that queen is more buggy than usual and the result of the queen is completely normal.
Uri

[Matt McKnight]

Geschrieben von: / Posted by: Matt McKnight at 30 January 2004 15:14:53:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Heinz van Kempen at 30 January 2004 08:06:55:

It is fair to add that not only The Baron, but also the versions of
Queen and Terra which play in this tournament seem to have some nasty
bugs.  I also suspect there is something wrong with Dorky, which
should normally perform much better in this field.
Tord

Hi Tord ,
thanks for adding this, we already discussed it a bit. For Terra I mentioned this time managment bug concerning the Fischer clock in an earlier post.
Matt anyway seems to be satisfied. Like I am seeing Dorky improved over the last Nunn positions and you can be sure, that I won´t run tournaments with 480 games for each engine, if I am not convinced from experiences over the last years, that there might still be enormous changes even after 300 games.
Best Regards
Heinz

I am satisfied that you haven't had any crashes or time losses, but the strength does seem to be lower. I have made many many changes to version 4.0, and I'm sure that not all of them were for the best. It was hoped that by CCT, I would have my engine at optimal strength, but I guess that won't be the case. I really hate to release a version that is weaker than previous versions, so it will probably remain as 4.0 _beta_
Matt

[Heinz van Kempen]

Geschrieben von: / Posted by: Heinz van Kempen at 30 January 2004 15:45:18:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Matt McKnight at 30 January 2004 15:14:53:

I am satisfied that you haven't had any crashes or time losses, but the strength does seem to be lower. I have made many many changes to version 4.0, and I'm sure that not all of them were for the best. It was hoped that by CCT, I would have my engine at optimal strength, but I guess that won't be the case. I really hate to release a version that is weaker than previous versions, so it will probably remain as 4.0 _beta_
Matt

Hello Matt,
there are eight Nunn positions still to come. Scoring very good in three of them and averagely the others would bring Dorky at least to the level of version 3.48.
I had such things before, when I wondered why all were having so fine results with Smar Think 0.17a, as I had it rated lower than version 0.16b after 250 games. Now after 800 games it is the second best amateur engine in my list, improving by 57 points.
I will have such things again, as statistics is weird, Just look at those +- values and there are even 5% that drop out of this probable margins. We will see, (of course I can´t know that it will the case for Dorky), when we have more of those Nunn tournaments that there will be engines that are either 50 points stronger or weaker after 480 games than after 240. I am very certain that this will happen and not only once. I believe that all testers who are testing with hundreds of games can confirm this with other examples. It just happens that way.
Best Regards
Heinz

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 30 January 2004 19:15:01:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Uri Blass at 30 January 2004 13:02:34:

I am surprised to read that Gothmog normally scores 0% against the baron.
Gothmog even scores better against crafty or smarthink when they are stronger than the baron.
I also do not think that queen is more buggy than usual and the result of the queen is completely normal.

Yes. The Baron is a nightmare opponent for Gothmog, mainly because of its
style. It plays sound and solid chess, and plays the endgame rather well.
It defends easily against Gothmog's primitive, clumsy and unsound attacks,
and then easily picks apart my engine by exploiting a material advantage or
superior pawn structure in the endgame. The games are ugly to watch.
I'm not so sure, but you probably know Queen better than me, so perhaps
you're right.
Tord

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 30 January 2004 19:16:56:
Als Antwort auf: / In reply to: Re: Nunn2 A-standings after 12 positions out of 20 geschrieben von: / posted by: Matt McKnight at 30 January 2004 15:14:53:

I am satisfied that you haven't had any crashes or time losses, but the strength does seem to be lower. I have made many many changes to version 4.0, and I'm sure that not all of them were for the best. It was hoped that by CCT, I would have my engine at optimal strength, but I guess that won't be the case. I really hate to release a version that is weaker than previous versions, so it will probably remain as 4.0 _beta_

Some of the earlier 4.0 betas seemed very strong, or at least much stronger
than Gothmog. What happened? Perhaps it's time to go back to the older
sources again?
Tord