Are Leo's ratings comparable to SSDF's?

Archive of the old Parsimony forum. Some messages couldn't be restored. Limitations: Search for authors does not work, Parsimony specific formats do not work, threaded view does not work properly. Posting is disabled.

Are Leo's ratings comparable to SSDF's?

Postby Dimitris Poulos » 29 May 2004, 22:41

Geschrieben von:/Posted by: Dimitris Poulos at 29 May 2004 23:41:20:

Hi to everybody,
I shall put it simple:
1) The SKR personality Leo uses is at least 30 points stronger than the default used by SSDF,
2) Leo seems to count the games played by previous( older) versions of the engines too, to calculate their ratings.
On this basis I think that we can't compare Leo's ratings to those by SSDF.
Some simple calculations drive to the conclusion that the new Aristarch is about 2700-2710 in the SSDF rating for example. Crafty seems to be around 2700 too. (Just count SKR for 2750 and Yace for 2600 to find the analogy).
Then it seems that the strongest free engines have long reached 2700.
Dimitris
Dimitris Poulos
 

Re: Are Leo's ratings comparable to SSDF's?

Postby Roger Brown » 30 May 2004, 03:52

Geschrieben von:/Posted by: Roger Brown at 30 May 2004 04:52:04:
Als Antwort auf:/In reply to: Are Leo's ratings comparable to SSDF's? geschrieben von:/posted by: Dimitris Poulos at 29 May 2004 23:41:20:
Hi to everybody,
I shall put it simple:

I shall put it even simpler:
Leo's list is NOT comparable and he has never sold it that way either.
A rating list can only rate what goes into it. Leo's list rates games between engines that are Winboard compatible, are playing games at 40 moves in 40 minutes etc.
The usefulness of a list is not whether it rates all engines you are interested in. It is whether it provides a reasonable basis for comparing items within the list. The SSDF does not rate the engines that Leo does - in the main. His list may be more than useful to users here than the SSDF which is the standard comp reference work.
It is a list for the purposes of his competition only.
Your statement that his list does not compare has an element of the strange about it as the comaparison was never made - certainly not by Leo.
Later.
Roger Brown
 

Re: Are Leo's ratings comparable to SSDF's?

Postby Norm Pollock » 30 May 2004, 07:11

Geschrieben von:/Posted by: Norm Pollock at 30 May 2004 08:11:53:
Als Antwort auf:/In reply to: Are Leo's ratings comparable to SSDF's? geschrieben von:/posted by: Dimitris Poulos at 29 May 2004 23:41:20:
Hi to everybody,
I shall put it simple:
1) The SKR personality Leo uses is at least 30 points stronger than the default used by SSDF,
2) Leo seems to count the games played by previous( older) versions of the engines too, to calculate their ratings.
On this basis I think that we can't compare Leo's ratings to those by SSDF.
Some simple calculations drive to the conclusion that the new Aristarch is about 2700-2710 in the SSDF rating for example. Crafty seems to be around 2700 too. (Just count SKR for 2750 and Yace for 2600 to find the analogy).
Then it seems that the strongest free engines have long reached 2700.
Dimitris
It is my understanding that the SSDF ratings are based solely on games using a chessbase gui (fritz/shredder) and with a chessbase book. Therefore non-native cb programs are at disadvantage(s).
The chessbase gui handles cb and uci format. Winboard engines require a chessbase converter to UCI format so that they can be used by the chessbase gui. It is likely some features are lost in translation, and the winboard engine in a cb gui is not as strong as a winboard engine in the winboard gui.
Leo's tournament on the other hand, uses the winboard gui. The cb engines cannot be converted to use this gui. Leo's tournaments use native winboard engines and a few uci engines with the polyglot uci to wb converter.
So I wouldn't worry if Leo's elos are different from SSDF's.
Norm Pollock
 

Re: Are Leo's ratings comparable to SSDF's?

Postby Dimitris Poulos » 30 May 2004, 09:50

Geschrieben von:/Posted by: Dimitris Poulos at 30 May 2004 10:50:36:
Als Antwort auf:/In reply to: Re: Are Leo's ratings comparable to SSDF's? geschrieben von:/posted by: Roger Brown at 30 May 2004 04:52:04:
Hi to everybody,
I shall put it simple:

I shall put it even simpler:
Leo's list is NOT comparable and he has never sold it that way either.
A rating list can only rate what goes into it. Leo's list rates games between engines that are Winboard compatible, are playing games at 40 moves in 40 minutes etc.
The usefulness of a list is not whether it rates all engines you are interested in. It is whether it provides a reasonable basis for comparing items within the list. The SSDF does not rate the engines that Leo does - in the main. His list may be more than useful to users here than the SSDF which is the standard comp reference work.
It is a list for the purposes of his competition only.
Your statement that his list does not compare has an element of the strange about it as the comaparison was never made - certainly not by Leo.
Later.
No, Leo didn't make such a comparison,
I made it to hear some opinions to estimate the actual strenght of the free engines compared to pros.
Dimitris
Dimitris Poulos
 

Re: Are Leo's ratings comparable to SSDF's?

Postby Roger Brown » 30 May 2004, 13:14

Geschrieben von:/Posted by: Roger Brown at 30 May 2004 14:14:18:
Als Antwort auf:/In reply to: Re: Are Leo's ratings comparable to SSDF's? geschrieben von:/posted by: Dimitris Poulos at 30 May 2004 10:50:36:
No, Leo didn't make such a comparison,
I made it to hear some opinions to estimate the actual strenght of the free engines compared to pros.
Dimitris


Ahhhh, you were provoking comment. For my part I feel on solid ground in saying that in the main, the amateur engines are freely available - for some even the source code is out there - so I would expect that commercial engines would be stronger if for no other reason than the fact that the commercials can test their code against the best amateur (not equal to second class!!).
Certainly Shredder (in its various incarnations), Fritz, the King, Chesstiger, Ruffian (wheeeeeee), Hiarcs, Deep Sjeng (wheeeeeee) and Junior are stronger than the best amateur free engines.
That would be expected in my opinion.
Then too, there is the gui issue. The King ships a genuine WB engine that kicks ass. Rebel unfortunately had some issues in WB. Chessbase stuff only works in the gui. The UCI engines for Shredder can be ported to WB but there are minor issues. The killer books are of course left behind in WB. Most other commercial products ship in their own interfaces.
It makes comp-comp games on one machine difficult if not impossible.

Having said that, isn't it wonderful to have all this free stuff that actually play and analyse at matster level?
I certainly cannot get enough of them!!
Later.
Roger Brown
 


Return to Archive (Old Parsimony Forum)

Who is online

Users browsing this forum: No registered users and 28 guests