Schachcomputer.info Community

Schachcomputer.info Community (https://www.schachcomputer.info/forum/index.php)
-   Teststellungen und Elo Listen / Test positions and Elo lists (https://www.schachcomputer.info/forum/forumdisplay.php?f=10)
-   -   Info: Aktivschachliste 2025 (https://www.schachcomputer.info/forum/showthread.php?t=7318)

spacious_mind 19.01.2026 23:59

Re: Aktivschachliste 2025
 
Zitieren:

Zitat von tsmiller1 (Beitrag 137556)
Hoffentlich werden die Listen bald aktualisiert, um die neuesten Schachcomputer zu berücksichtigen. Dank der Tests von dir, Nick, mir und anderen liegen bereits genügend Partien vor, um eine gute Elo-Schätzung zu erhalten, und sobald weitere Tests durchgeführt werden, kann eine verlässliche Elo-Zahl ermittelt werden.

Hi Tracy,

Unfortunately, as Micha tried to explain, it is never going to really happen unless you get a few other computers around the King strength range. It needs ideally 4-6 computers of similar strength as King and enough games to be able to rate the next one. Vonset is possible as there are enough opponents for it. King when it first came out barely made the list. And to me it is still a surprise but a pleasant one that it made the tournament list:

Tournament List
1 Millennium The King Element 2591 70 (+ 34,= 24,- 12), 65.7 %
Millennium ChessGenius Exclusive 29 (+ 20,= 7,- 2), 81.0 %
Phoenix Revelation II Hiarcs 14.1 1% CPU 41 (+ 14,= 17,- 10), 54.9 %

As you can see it only ever played 2 opponents and if it wasn't for Hiarcs 14.1 1% beating it 10 times it would never have made the list. The gap against Chess Genius alone is too large to make a decent rating.

And so on it goes down the list.

Vonset could probably make the list, by playing King and Hiarcs and King 50 Mhz and ChessGenius Exclusive is doable.

But TCA2 scores against King and Vonset is a stretch because the gap is too great. You need something that gives you a Hiarcs kind of score +14=17-10 which means only a 54.9 percentage over 41 games. Instead of 80-90% performances because the risk is that the next 40 games could end up being 95%.

Ideally you need something (preferably more than one) that is established but slightly higher than King and so on and that does not exist at the moment.

Regards

Nick

tsmiller1 20.01.2026 02:27

Re: Aktivschachliste 2025
 
Zitieren:

Zitat von spacious_mind (Beitrag 137557)
Hi Tracy,

Unfortunately, as Micha tried to explain, it is never going to really happen unless you get a few other computers around the King strength range. It needs ideally 4-6 computers of similar strength as King and enough games to be able to rate the next one. Vonset is possible as there are enough opponents for it. King when it first came out barely made the list. And to me it is still a surprise but a pleasant one that it made the tournament list:

Tournament List
1 Millennium The King Element 2591 70 (+ 34,= 24,- 12), 65.7 %
Millennium ChessGenius Exclusive 29 (+ 20,= 7,- 2), 81.0 %
Phoenix Revelation II Hiarcs 14.1 1% CPU 41 (+ 14,= 17,- 10), 54.9 %

As you can see it only ever played 2 opponents and if it wasn't for Hiarcs 14.1 1% beating it 10 times it would never have made the list. The gap against Chess Genius alone is too large to make a decent rating.

And so on it goes down the list.

Vonset could probably make the list, by playing King and Hiarcs and King 50 Mhz and ChessGenius Exclusive is doable.

But TCA2 scores against King and Vonset is a stretch because the gap is too great. You need something that gives you a Hiarcs kind of score +14=17-10 which means only a 54.9 percentage over 41 games. Instead of 80-90% performances because the risk is that the next 40 games could end up being 95%.

Ideally you need something (preferably more than one) that is established but slightly higher than King and so on and that does not exist at the moment.

Regards

Nick

Actually, there are dozens of chess engines slightly below, right at, and slightly above the rating of the MKP: https://www.computerchess.org.uk/ccrl/4040/.

spacious_mind 20.01.2026 12:34

Re: Aktivschachliste 2025
 
Zitieren:

Zitat von tsmiller1 (Beitrag 137559)
Actually, there are dozens of chess engines slightly below, right at, and slightly above the rating of the MKP: https://www.computerchess.org.uk/ccrl/4040/.

Right, chess engines. There are probably several hundred suitable if you consider their versions and DOS that are suitable. That is what SSDF used since they became an engine rating list. And they played the Resurrection and Revelation against Desktops. They only kept a few dedicated computers for posterity.

Even SSDF could not really play them. If you look at their list they played a Revelation against the occasional Resurrection, never a Revelation against Revelation. Which means they never owned two Revelations to be able to play against each other. Computer played chess engines games don't make Schachcomputer.Info's list.

How many people do you think owned two Revelations or two Resurrections to be able to play them against each other? And of those how many do you think had the interest to actually play those matches for the benefit of a rating list?

I am sure there are some games collected but not enough to make a list. Mess did not exist. So even if you owned a Resurrection, you would need to also own an R30 to be able to play them (Millennium King did not exist). People tried online but you could not get enough online matches going because of the time it takes and interested players needed. A tournament game could take 6-8 hours, a 30 second game could take 3-4 hours, never mind time zone considerations and working for a living.

So in summary SSDF to Schachcomputer.Info is an apples to oranges comparison as SSDF's list is not pure as in dedicated chess computers only. :)

Regards
Nick

tsmiller1 21.01.2026 10:07

Re: Aktivschachliste 2025
 
Zitieren:

Zitat von spacious_mind (Beitrag 137567)
Right, chess engines. There are probably several hundred suitable if you consider their versions and DOS that are suitable. That is what SSDF used since they became an engine rating list. And they played the Resurrection and Revelation against Desktops. They only kept a few dedicated computers for posterity.

Even SSDF could not really play them. If you look at their list they played a Revelation against the occasional Resurrection, never a Revelation against Revelation. Which means they never owned two Revelations to be able to play against each other. Computer played chess engines games don't make Schachcomputer.Info's list.

How many people do you think owned two Revelations or two Resurrections to be able to play them against each other? And of those how many do you think had the interest to actually play those matches for the benefit of a rating list?

I am sure there are some games collected but not enough to make a list. Mess did not exist. So even if you owned a Resurrection, you would need to also own an R30 to be able to play them (Millennium King did not exist). People tried online but you could not get enough online matches going because of the time it takes and interested players needed. A tournament game could take 6-8 hours, a 30 second game could take 3-4 hours, never mind time zone considerations and working for a living.

So in summary SSDF to Schachcomputer.Info is an apples to oranges comparison as SSDF's list is not pure as in dedicated chess computers only. :)

Regards
Nick

That's OK, though. Chess engines are still chess AI, just in a different form than dedicated chess computers. They'd be able to provide appropriate strength opponents to finally get the TCA II, Vonset L6, DGT Centaur, and Sense Robot into the ratings list, since appropriately strong opponents are what is missing in getting these machines a rating. Simply a matter of calibrating the rating of any engines used so they'd be similar to the current Active Wiki Elo list.


Alle Zeitangaben in WEZ +1. Es ist jetzt 22:58 Uhr.

Powered by vBulletin (Deutsch)
Copyright ©2000 - 2026, Jelsoft Enterprises Ltd.
©Schachcomputer.info