Are your tournaments checked for duplicates?

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 11:25:04:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

[Alex Schmidt]

Geschrieben von:/Posted by: Alex Schmidt at 10 July 2004 12:03:19:
Als Antwort auf:/In reply to: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 11:25:04:

Hi,

Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

You can simply delete the result in the engtourn.ctg, then Fritz will repeat the game after restarting the tournament.
Best,
Alex

[Olivier Deville]

Geschrieben von:/Posted by: Olivier Deville at 10 July 2004 12:36:23:
Als Antwort auf:/In reply to: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 11:25:04:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

Hi Norm
Here engines are playing with their ownbook if available : this is the case for most WB and Uci engines.
If a ctg book is needed (CB native engines or engines coming with no book), I use the following settings :
- Tournament is off, of course
- Variety is set to 1 (scale : 0 to 5)
If variety is too high, the engines may play weak variations.
Olivier


ChessWar

[Günther Simon]

Geschrieben von:/Posted by: Günther Simon at 10 July 2004 12:48:54:
Als Antwort auf:/In reply to: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 11:25:04:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

This is in first place a Winboard Forum (ever noticed the name?),
therefore the people reporting tournaments here obviously shouldn't
have the problems you mentioned.
We dont use 'tournament books' or chessbase books or whatever.
The only way duplicated games might happen, would be that both programs
dont have a book at all _plus_ zero randomness(or cloning).
(BTW a duplicated game is only a duplicated game if _all_ moves
are the same until the end, but I hope you know this already.
Deleting games other than real duplicates I would consider a bad
intervention from the user, except he made a setup error.)
Of course deleting such duplicates would be an error...as both
programs simply cant play else, regardless what the user likes to see.
(happens very seldom naturally)
Regards,
Günther

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 15:05:30:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Alex Schmidt at 10 July 2004 12:03:19:

Hi,

Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

You can simply delete the result in the engtourn.ctg, then Fritz will repeat the game after restarting the tournament.
Best,
Alex

I just tried doing this on a few games (non-duplicates) in a tournament that just finished. Unfortunately, it did not work. I got "Tournament Completed" when I opened the tournament and restarted.
When I "deleted" the game in engtour, it did not remove it. It just put a line through it and dimmed it. Maybe there is a follow-up step to fully delete it?

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 15:32:52:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Günther Simon at 10 July 2004 12:48:54:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

This is in first place a Winboard Forum (ever noticed the name?),
therefore the people reporting tournaments here obviously shouldn't
have the problems you mentioned.
We dont use 'tournament books' or chessbase books or whatever.
The only way duplicated games might happen, would be that both programs
dont have a book at all _plus_ zero randomness(or cloning).
(BTW a duplicated game is only a duplicated game if _all_ moves
are the same until the end, but I hope you know this already.
Deleting games other than real duplicates I would consider a bad
intervention from the user, except he made a setup error.)
Of course deleting such duplicates would be an error...as both
programs simply cant play else, regardless what the user likes to see.
(happens very seldom naturally)
Regards,
Günther

There are tournaments mentioned in this forum that have already reported using the fritz8/cb gui. I'm not the first to mention it. The chessbase gui is a great gui and it runs tournaments that allow me to easily match ruffian, list, theking, fritz8, shredder8, for example. I look at it as progress.
And since that gui accepts winboard engines, as well as uci-only and proprietary cb engines, I think it's inclusion here is justified.
I do have reservations about whether the cb engines have an advantage though. I would prefer that the cb engines also be wb and uci compatible, but they are not. Maybe chessbase has a reason for that in terms of protecting the investment in their business. Suppose Fritz8 was wb or uci compatible, wouldn't then anyone who gets a pirated copy of the engine not be paying chessbase? But if you have to also use their interface, then chessbase is more likely to get some money. Anyone who knows business, knows you have to be very tough to succeed.
As for duplicates:
Duplicates mean identical games with identical moves and end at the same point. They do not have to be played by the same players. If the players are also the same, the games are "twins". In an engine-engine tournament, the duplicates are highly, highly likely to also be twins. pgn-extract detects duplicates, while Scid only detects twins.
Duplicates happen because of lack of enough randomness in the engines. Given an opening with book moves that eliminate a lot of pieces before the engines kick in, the chances for duplicate games increase if the 2 engines play the same opening against each other many times.
Duplicates even happen in grandmaster play with 2 different sets of players. I've seen that in culling duplicates from large databases. And then there are the near duplicates: A 50 move game ends in a draw. The next time the players play, they repeat the same 50 moves and then add a few more moves on.

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 15:38:07:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Olivier Deville at 10 July 2004 12:36:23:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

Hi Norm
Here engines are playing with their ownbook if available : this is the case for most WB and Uci engines.
If a ctg book is needed (CB native engines or engines coming with no book), I use the following settings :
- Tournament is off, of course
- Variety is set to 1 (scale : 0 to 5)
If variety is too high, the engines may play weak variations.
Olivier

Hi Olivier,
The advantage(?) of having them use their own copy of the same book is that the engines determine the winner, not the books. It levels the playing field, perhaps.
-Norm

[Günther Simon]

Geschrieben von:/Posted by: Günther Simon at 10 July 2004 15:58:36:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 15:32:52:

There are a lot of tournaments reported in this forum. I wonder if they all check for duplicate games? I found out that duplicate games are a big problem for the tournaments I run in the Fritz8 gui.
Detecting duplicate games in a pgn file can be done using utilities like Scid or the utility pgn-extract.
The problem is more acute with the Fritz8 gui because the default is to let all engines share the same opening book and the default values "tournament book" and "variety", in book settings, restrict opening variety. Btw, if the opening book is being shared, then book learning should be off because there will be conflicting "learning".
There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.
Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

This is in first place a Winboard Forum (ever noticed the name?),
therefore the people reporting tournaments here obviously shouldn't
have the problems you mentioned.
We dont use 'tournament books' or chessbase books or whatever.
The only way duplicated games might happen, would be that both programs
dont have a book at all _plus_ zero randomness(or cloning).
(BTW a duplicated game is only a duplicated game if _all_ moves
are the same until the end, but I hope you know this already.
Deleting games other than real duplicates I would consider a bad
intervention from the user, except he made a setup error.)
Of course deleting such duplicates would be an error...as both
programs simply cant play else, regardless what the user likes to see.
(happens very seldom naturally)
Regards,
Günther

There are tournaments mentioned in this forum that have already reported using the fritz8/cb gui. I'm not the first to mention it. The chessbase gui is a great gui and it runs tournaments that allow me to easily match ruffian, list, theking, fritz8, shredder8, for example. I look at it as progress.
And since that gui accepts winboard engines, as well as uci-only and proprietary cb engines, I think it's inclusion here is justified.
I do have reservations about whether the cb engines have an advantage though. I would prefer that the cb engines also be wb and uci compatible, but they are not. Maybe chessbase has a reason for that in terms of protecting the investment in their business. Suppose Fritz8 was wb or uci compatible, wouldn't then anyone who gets a pirated copy of the engine not be paying chessbase? But if you have to also use their interface, then chessbase is more likely to get some money. Anyone who knows business, knows you have to be very tough to succeed.
As for duplicates:
Duplicates mean identical games with identical moves and end at the same point. They do not have to be played by the same players. If the players are also the same, the games are "twins". In an engine-engine tournament, the duplicates are highly, highly likely to also be twins. pgn-extract detects duplicates, while Scid only detects twins.
Duplicates happen because of lack of enough randomness in the engines. Given an opening with book moves that eliminate a lot of pieces before the engines kick in, the chances for duplicate games increase if the 2 engines play the same opening against each other many times.
Duplicates even happen in grandmaster play with 2 different sets of players. I've seen that in culling duplicates from large databases. And then there are the near duplicates: A 50 move game ends in a draw. The next time the players play, they repeat the same 50 moves and then add a few more moves on.

I am simply not interested in other GUIs than WB, that's because I am here!
Like I said before, the problem with duplicates in WB is neglectable
because it has a completely different approach and doesn't want to do
all stuff for the programs...
No need therefore to explain us problems from other GUIs here.
I have (nearly) nothing against people posting their results including
WB programs in CB or Arena, but I admit I dont care much for results
in those GUIs, as I dont trust them, as much as WB for WB programs.
OTH posting problems which arise in those non topic GUIs only is
quite different and more than OT, you should respect that.
Günther

[Alex Schmidt]

Geschrieben von:/Posted by: Alex Schmidt at 10 July 2004 16:20:18:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 15:05:30:

Hi,

I just tried doing this on a few games (non-duplicates) in a tournament that just finished. Unfortunately, it did not work. I got "Tournament Completed" when I opened the tournament and restarted.

Yes, you are right, if the tournament is completed it don't work. I didn't use Fritz now for a long time because of this 'simple' problems...
What you can do is:
Start the tournament once more. Before that give in the timecontrol once more so Fritz will recognize it's a new tournament.
In the engtourn.ctg all games appear a second time with the extension "-1". Delete the results of the games you want to replay of the first tournament, delete all games of the second tournament. Then continue the tournament, Fritz will ask for the number of the first game, give in the number of the first game of the first tournament in the engtourn.ctg.
A little bit confusing but it should work
Best,
Alex

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 16:53:13:
Als Antwort auf:/In reply to: unknown WB subject... geschrieben von:/posted by: Günther Simon at 10 July 2004 15:58:36:

I am simply not interested in other GUIs than WB, that's because I am here!
Like I said before, the problem with duplicates in WB is neglectable
because it has a completely different approach and doesn't want to do
all stuff for the programs...
No need therefore to explain us problems from other GUIs here.
I have (nearly) nothing against people posting their results including
WB programs in CB or Arena, but I admit I dont care much for results
in those GUIs, as I dont trust them, as much as WB for WB programs.
OTH posting problems which arise in those non topic GUIs only is
quite different and more than OT, you should respect that.
Günther

Understood.
I wish WB gui would have an update that enables use of other types of engines. Unfortunately many engine makers- amateur and professional, are the ones that are bypassing the wb gui. That sort of forces me as an enthusiast/hobbyist to go with the flow. What is the future of the WB gui? I hope it can maintain and improve its "market share".
Duplicates occur in WB gui. They occur in all gui's. I found them when I used Jori's TM. The thing is that you have to check for them, otherwise they are invisible.

[Olivier Deville]

Geschrieben von:/Posted by: Olivier Deville at 10 July 2004 19:41:53:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 15:05:30:

Hi,

Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

You can simply delete the result in the engtourn.ctg, then Fritz will repeat the game after restarting the tournament.
Best,
Alex

I just tried doing this on a few games (non-duplicates) in a tournament that just finished. Unfortunately, it did not work. I got "Tournament Completed" when I opened the tournament and restarted.
When I "deleted" the game in engtour, it did not remove it. It just put a line through it and dimmed it. Maybe there is a follow-up step to fully delete it?

Another good trick, Norm :
Open the .trn file with notepad, and set variable Crated to 2 and Complete to 1. So the GUI will replay the games that have no results yet.
Olivier



ChessWar

[Norm Pollock]

Geschrieben von:/Posted by: Norm Pollock at 10 July 2004 20:38:02:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Olivier Deville at 10 July 2004 19:41:53:

Hi,

Unfortunately if a duplicate is found, I do not know how to go back and eliminate the duplicate and replace it with another game.

You can simply delete the result in the engtourn.ctg, then Fritz will repeat the game after restarting the tournament.
Best,
Alex

I just tried doing this on a few games (non-duplicates) in a tournament that just finished. Unfortunately, it did not work. I got "Tournament Completed" when I opened the tournament and restarted.
When I "deleted" the game in engtour, it did not remove it. It just put a line through it and dimmed it. Maybe there is a follow-up step to fully delete it?

Another good trick, Norm :
Open the .trn file with notepad, and set variable Crated to 2 and Complete to 1. So the GUI will replay the games that have no results yet.
Olivier

Thanks. I know you meant "Created" above. I also had to go into edit games of engtour in order to erase the results of played games. You have to click on the blank choice in order for the result to disappear. Then when I continued the tournament, the games without results were replayed.
-Norm

[Dieter Bürßner]

Geschrieben von:/Posted by: Dieter Bürßner at 10 July 2004 23:06:49:
Als Antwort auf:/In reply to: unknown WB subject... geschrieben von:/posted by: Günther Simon at 10 July 2004 15:58:36:

Like I said before, the problem with duplicates in WB is neglectable
because it has a completely different approach and doesn't want to do
all stuff for the programs...
No need therefore to explain us problems from other GUIs here.
I have (nearly) nothing against people posting their results including
WB programs in CB or Arena, but I admit I dont care much for results
in those GUIs, as I dont trust them, as much as WB for WB programs.
OTH posting problems which arise in those non topic GUIs only is
quite different and more than OT, you should respect that.

It can be a problem under WB, too. For example at least earlier versions of Crafty played by default with almost no book randomness under newer versions WB (which tells the engine, that the opponent is a computer - and this switches off all/most the randomness in the book of Crafty. It was btw. not trivial to switch this behavior of Crafty off, settings in crafty.rc will not help, IIRC).
If you now face Crafty with an opponent, that does similar, you will get duplicate games, or games that are duplicated until decided. This happened to me for example, when testing a (non public) "tournament book" for Yace under WB. So, I think, it is an issue under WB, too. Recent versions of Yace have options, to switch randomness in the book off, so it can be an issue there, too (I don't use this for my tests, so I am only guessing).
So, IMHO duplicate games can be an issue under WB. Even more, when you include practical duplicate games, that diversed at some point, where already decided. (For example one engine may have positional learning, and chose in the next game to lose in a slightly different way. Or some randomnoss, that maybe all engines have, may make them to chose a slightly different path in a dead drawn position. For example when there are perpetual checks or unavoidable repetitions.)
Well, wouldn't it be easy, to just ignore those posts. I read:
"Please post messages about Winboard and related topics only."
at the top of this forum. Running a WB engine under any GUI looks related to me, but I can understand, that it doesn't look too related to other people (the above does not mention Winboard engines at all). And I surely will respect this. I glided off topic rather often I fear. I'll think about it.
BTW. I also do not trust other GUIs as much as WB. Still it seems on topic to me, to discuss some aspects. There could be various reasons to not trust the other GUIs. Is the protocol implemented in accordance to or in the spirit the spec (remember old WB adapter)? Where does it differ from the implementation of WB/XB (the GUIs)?
With kind regards,
Dieter

[Günther Simon]

Geschrieben von:/Posted by: Günther Simon at 11 July 2004 01:35:36:
Als Antwort auf:/In reply to: Re: unknown WB subject... geschrieben von:/posted by: Dieter Bürßner at 10 July 2004 23:06:49:

Like I said before, the problem with duplicates in WB is neglectable
because it has a completely different approach and doesn't want to do
all stuff for the programs...
No need therefore to explain us problems from other GUIs here.
I have (nearly) nothing against people posting their results including
WB programs in CB or Arena, but I admit I dont care much for results
in those GUIs, as I dont trust them, as much as WB for WB programs.
OTH posting problems which arise in those non topic GUIs only is
quite different and more than OT, you should respect that.

It can be a problem under WB, too. For example at least earlier versions of Crafty played by default with almost no book randomness under newer versions WB (which tells the engine, that the opponent is a computer - and this switches off all/most the randomness in the book of Crafty. It was btw. not trivial to switch this behavior of Crafty off, settings in crafty.rc will not help, IIRC).
If you now face Crafty with an opponent, that does similar, you will get duplicate games, or games that are duplicated until decided. This happened to me for example, when testing a (non public) "tournament book" for Yace under WB. So, I think, it is an issue under WB, too. Recent versions of Yace have options, to switch randomness in the book off, so it can be an issue there, too (I don't use this for my tests, so I am only guessing).
So, IMHO duplicate games can be an issue under WB. Even more, when you include practical duplicate games, that diversed at some point, where already decided. (For example one engine may have positional learning, and chose in the next game to lose in a slightly different way. Or some randomnoss, that maybe all engines have, may make them to chose a slightly different path in a dead drawn position. For example when there are perpetual checks or unavoidable repetitions.)
Well, wouldn't it be easy, to just ignore those posts. I read:
"Please post messages about Winboard and related topics only."
at the top of this forum. Running a WB engine under any GUI looks related to me, but I can understand, that it doesn't look too related to other people (the above does not mention Winboard engines at all).
BTW. I also do not trust other GUIs as much as WB. Still it seems on topic to me, to discuss some aspects. There could be various reasons to not trust the other GUIs. Is the protocol implemented in accordance to or in the spirit the spec (remember old WB adapter)? Where does it differ from the implementation of WB/XB (the GUIs)?
With kind regards,
Dieter

This is the charta! Please read it again.
It was about book learning stuff exclusively in Chessbase
independent from WB and WB engines at all...nothing on topic
remained in that post.
And I surely will respect this. I glided off topic rather often I fear. I'll think about it.
I am really not interested in discussing this all over now,
as the original posting lacked so much logic, that I would
need more time than I want to spend on this!
Moreover I see already the Chessbase or CSS guys (ex moderators)
coming in and trying to put their imagination of what a WB forum
has to be, there!
Therefore only one thing: Duplicates are no problem at all,
if all is set up correctly by the user. (according to the
authors config wishes)
If duplicates arise under proper conditions they are simply there,
no need to delete no need to cry, they all have to be counted!
If it was a setup error, delete what you want, but still no
problem, which is of much interest here....
WB has nothing to do with lacking randomness in the programs,
the book stuff is done by the programs. If authors decide best
play has no randomness and they trust their learning, why should
anyone delete the duplicates??
Moreover I am really not interested in which way we can set up the
book learning stuff wrong in Chessbase to produce more duplicates
than necessary.
I simply dont understand what it has to do with WB or may be any
GUI at all?? (except faulty book learning stuff in Chessbase)
If someone wants to flaw his tests and delete all duplicates,
even if they were produced under proper conditions, well he
could do what he wants...
Regards,
Günther
P.S. If I would match Yace vs. Crafty and Bob wants no randomness
and it would produce 100 duplicated won games vs. Yace, would
you demand I should delete 99 won games??

Goodnight and please leave me out of that trash...

[Tom Likens]

Geschrieben von:/Posted by: Tom Likens at 11 July 2004 14:50:27:
Als Antwort auf:/In reply to: OT geschrieben von:/posted by: Norm Pollock at 10 July 2004 16:53:13:

I am simply not interested in other GUIs than WB, that's because I am here!
Like I said before, the problem with duplicates in WB is neglectable
because it has a completely different approach and doesn't want to do
all stuff for the programs...
No need therefore to explain us problems from other GUIs here.
I have (nearly) nothing against people posting their results including
WB programs in CB or Arena, but I admit I dont care much for results
in those GUIs, as I dont trust them, as much as WB for WB programs.
OTH posting problems which arise in those non topic GUIs only is
quite different and more than OT, you should respect that.
Günther

Understood.
I wish WB gui would have an update that enables use of other types of engines. Unfortunately many engine makers- amateur and professional, are the ones that are bypassing the wb gui. That sort of forces me as an enthusiast/hobbyist to go with the flow. What is the future of the WB gui? I hope it can maintain and improve its "market share".
Duplicates occur in WB gui. They occur in all gui's. I found them when I used Jori's TM. The thing is that you have to check for them, otherwise they are invisible.

This is so true. A semi-amusing story, regarding duplicate games. When I first
started computer chess programming, one of Djinn's earliest sparring partners
was Arasan, (who in the beginning, *always* stomped all over Djinn). After a
ton of work, the engine eventually got good enough that it would tie or even
occasionally win a 100-game match. I couldn't have been prouder, because now
(I thought) I'm really starting to make some progress- next stop Fritz (these
were the early naive days, when anything was possible).
Except, that one day I got the bright idea to write a Perl script that checked
all my PGN matches for duplicates. As you can guess, the Djinn/Arasan matches
were *full* of duplicate games. Arasan was still thrashing Djinn!! And it
suddenly dawned on me, that I still had a ton of work ahead of me.
It's interesting at such moments, one either gives up computer chess programming
forever (never looking back) or such incidents fortify and strengthen the
resolve to create the world's strongest silicon grandmaster. Luckily, (I think )
it had the latter effect on me.
regards,
--tom
--tom

[Niyaz Khasanov]

Geschrieben von:/Posted by: Niyaz Khasanov at 12 July 2004 08:42:33:
Als Antwort auf:/In reply to: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 11:25:04:

There are some techniques to minimize duplicate games in the fritz8 gui. I use the following:
(1) set tournament book off
(2) set variety to maximum
(3) give each engine its own copy of the opening book and turn learning on. This takes up a lot of hd space though.
(4) get the pgn of the tournament and check it using Scid or pgn-extract.

All Fritz book are bad for engine-engine games.
A good engine-engine book should have:
(1) only short lines (not longer than 10 moves),
(2) no bad lines (holes),
(3) maximal variety.
There are two good solutions:
(1) use engines' own books and learn=on (this solution is preferred by authors),
(2) create a special book and learn=off.

[Niyaz Khasanov]

Geschrieben von:/Posted by: Niyaz Khasanov at 12 July 2004 08:48:57:
Als Antwort auf:/In reply to: Re: Are your tournaments checked for duplicates? geschrieben von:/posted by: Norm Pollock at 10 July 2004 15:38:07:

Hi Olivier,
The advantage(?) of having them use their own copy of the same book is that the engines determine the winner, not the books. It levels the playing field, perhaps.
-Norm

Yes if the book is wide, short-lined and proof.
Otherwise the book holes determine the winner in a short tournament.