The King module is announced on Schach Niggemann

[Yarc]

I've been busy attending to affairs since losing my father recently and I want to thank the people on this forum for their kind thoughts. Anyway, I have just noticed that Schach Niggemann have announced the new King module. There are no product pictures yet but it's possible to pre-order this new module. In an effort to cheer myself up I have indeed pre-ordered and know I will enjoy it when it arrives. Here is the link:

https://www.schachversand.de/en/compute ... -link.html

I wonder what the King is doing tonight Regards
Ray

[Steve Maughan]

Interestingly the piece says Johann has spent one year improving The King engine!! It should be an interesting product.

Tempted regards, Steve

[Yarc]

Interestingly the piece says Johann has spent one year improving The King engine!! It should be an interesting product.

Tempted regards, Steve

From what I have been reading over on the German forum it is indeed an interesting opponent and the improvements Johann has made will surely make this a must for dedicated fans. There were rumours abound about it's playing strength being 2500+, but Niggemann have quoted it as being 2450. Either way, it's very good and more importantly has a very exciting style of play which should make playing against it and computer matches very entertaining.

I don't think it's going to be out until sometime in November but

That's not far way Regards
Ray

[Steve B]

Interestingly the piece says Johann has spent one year improving The King engine!! It should be an interesting product.

Tempted regards, Steve

From what I have been reading over on the German forum it is indeed an interesting opponent and the improvements Johann has made will surely make this a must for dedicated fans. There were rumours abound about it's playing strength being 2500+, but Niggemann have quoted it as being 2450. Either way, it's very good and more importantly has a very exciting style of play which should make playing against it and computer matches very entertaining.

I don't think it's going to be out until sometime in November but

That's not far way Regards
Ray

The King module just turned in a 2600+ performance rating(30 sec.) in Michael's tournament
2450-2500 would be a disappointment actually


Under Achieving Regards
Steve

[Schachspieler]

Found it here some days ago https://www.topschach.de/chessgenius-ex ... -4119.html

There will be a match between International Master Roman Vidonyak and the new King Module. Maybe this time Roman Vidonyak will lose.

Last time he won the Match against the Lang Module:

http://forum.topschach.de/chess-genius- ... t1125.html

Should be available very soon

[Yarc]

The King module just turned in a 2600+ performance rating(30 sec.) in Michael's tournament
2450-2500 would be a disappointment actually


Under Achieving Regards
Steve

I understand where you are coming from and at this stage it's probably too soon to get an accurate rating. I have been following the German forum where the official testers have been posting and there has been much debate over it's strength. In BT-2630 tests it has scored in excess of 2500, but some have said that its moves in actual games are not of that standard, hmmm. What they have all been saying is that it's extremely good fun to play and provides very active and sometimes risky moves which can either devastate the opponent or cause it to lose the game. We are in with a chance then!


The jury is out on ELO Regards
Ray

[Steve B]

I have been following the German forum where the official testers have been posting and there has been much debate over it's strength

a lot of "official" testers it seems
I guess these folks get their Tasc modules for free

think its time for me to take a trip to Germany

European Vacation Regards
Steve

[spacious_mind]

Wonder if Millennium will sell the Exclusive board separately. Don't need two Lang's.

Millennium 2000 is down for a few days. Have they changed the link or gone bust?

Best regards

[Frank Quisinsky]

Hi,

I updated my Risc 2500 with the Montreux 1.0 eprom and 2048 RAM for hash. Maybe the same that Risc 2500 will be stronger as the new Millennium computer with the newer program by Johan de Koning with 50MHz. Mephisto Berlin 68020 is with 24Mhz clearly stronger as Millennium Genius with 48Mhz.

The quality of the good and old Mephisto Exclusive or Muinch boards are higher as the new Millennium Exclusive chess computer in my humble opinion. Also I am thinking that other programs can be more interesting as a program by Johan de Koning.

The programs by Johan de Koning in Tasc R30 or Risc 2500 are very aggressive, good for mate problems or in endgames with pawns, queens or rocks but bad in positional positions or often in complicated openings. Opening stratety isn't good!

SSDF made start of the year 2018 a correction with around 100 Elo down. A Risc 2500 here with 2.100 Elo ... I am thinking more right is a correction with 120-150 Elo, because chess computers like Risc 2500 haven't more as 2.050 - 2.075 Elo. A Berlin 68020 not more as 2.075 - 2.100 Elo.

For 20 years we have that opinion after rating games vs. humans. All the time Elo for the older chess computers are clearly to high.

Today the current Wasp with humen like playing style is 400 Elo stronger as TheKing 3.23. Wasp is one of the most aggressive engines with many pieces on board and perfect for opening understandings. Other engines with a great style of play are much more interesting as the old TheKing program. But chess computer lovers like that to have a newer TheKing program in a newer chess computer.

I wish me a newer chess computer with Wasp or Rodent. Aggressive engines with many pieces on board, good for learning openings.

And the new Genius Pro with 150 Mhz is only 50 Elo stronger as Mephisto Berlin 68020 with 24Mhz. I hope not the same with the current TheKing version for Millennium Exclusive chess computer in comparing with Risc 2500.

Best
Frank

[spacious_mind]

SSDF made start of the year 2018 a correction with around 100 Elo down. A Risc 2500 here with 2.100 Elo ... I am thinking more right is a correction with 120-150 Elo, because chess computers like Risc 2500 haven't more as 2.050 - 2.075 Elo. A Berlin 68020 not more as 2.075 - 2.100 Elo.

For 20 years we have that opinion after rating games vs. humans. All the time Elo for the older chess computers are clearly to high.

Today the current Wasp with humen like playing style is 400 Elo stronger as TheKing 3.23. Wasp is one of the most aggressive engines with many pieces on board and perfect for opening understandings. Other engines with a great style of play are much more interesting as the old TheKing program. But chess computer lovers like that to have a newer TheKing program in a newer chess computer.

Hi Frank,

Thanks for your posts it is good to see some new posters at Hiarcs. You are right that King at 50 Mhz should be competitive with Risc or Montreux.

However, sorry but I don't buy into your assumption that a Risc should be around 2000 ELO. You can only do that by placing it into an ELO chart with engines. Comparing to humans it is absolutely not true based on human games.

Nowadays with Lichess it is very easy to compare Chessbase human Tournament games and their evaluations against chess computer games and compare the errors and centipawns.

I did that for around 200+ London 33 Mhz TM games 200+ London 33 MHz 30 sec/move games. Mephisto Nigel Short and for of course Novag Super Constellation. I also did the same with Novag Constellation 3.6 games.

For the human games I used Chessbase matches between 2009 and 2014 from the Mega bases using at least 250 matches from players for example rated between 1975 to 2025 published Chessbase ELO to obtain the average of 2000 ELO for the human player. I ran every single game through Lichess. For all the computers and for all the humans. The human players who played against each other where from the German Bundesliga, Russia, France, Italy, South America, USA, Serbia, Croatia and so on and on all officially rated.

I did the exact same for the same years 2009-2014 matches for 1775-2025 ELO 2175-2225 ELO, 2375-2425 ELO etc all the way up to SuperGM matches above 2800 ELO to obtain an accurate comparison.






In the above first chart in yellow you can see the rating that London, Nigel Short and Super Constellation achieved with exact same lichess evaluation as the chess base humans.


Interestingly in these evaluations Super Constellation was evaluated at exactly 2018 USCF ELO when you adjust the score to USCF. This is the exact same rating that the USCF Chess rated the computer in tournaments against humans in the 80's. So therefore you can conlude that a 2018 ELO player in the mid late 80's is no different to a 2018 ELO player in the years 2009-2014


Therefore it is a complete myth that Risc 2500 should be at 2000 ELO. Sure we all know who have these computers that we can beat them, because of how we have adjusted ourselves to play them. But dress up a Risc 2500 computer into a human form and put him into a tournament against humans playing 2000 ELO and Risc 2500 will smash them.


SSDF does a good job trying to match where a dedicated computer lies playing engines and lists them against engines that play up to what 3.4 GHz? it does not match humans against dedicated computers.

Tibono is currently playing CP2150 PC version at Lichess live against humans and it is performing at 2400 ELO. For gawd sake that needs a step back and thinking about. CP2150???? rated at what 1675 ELO or something similar playing at 2400 ELO against humans after a few hundred games???? It throws engine rating lists out of the window!!

Best regards

[Frank Quisinsky]

Hi,

at first:
Have many thanks for your interesting website to chess computers. I am looking often and compare different information with my own and with schachcomputer.info. Your information and site are really great for me.

To Elo of older chess computers.
Normely not measurable for older chess computers.

Example:

MM V, SSDF = 1.878 ... exactly 52 Elo to high after ratings games vs. humans I have.

Not measurable, easy to explain.
Elo is bad for chess computers generally.

MM V: After openings 2.100 Elo (because very aggressive with many pieces on board). Transposition into endgame with 1.500 Elo, endgame with 1.300 - 1.400 Elo.

And now what will you give for an Elo for such a constellation. The right answere can be only: Elo not possible, not measurable!

So we know that for 25 years and we can play against the weaknesses of chess computers. Passive opening system and try to go to the endgame ... players in our club have no problems with 1.700 Elo vs. MMV 5Mhz Standard very often.

And all this you can see for all the chess computers in the past. Julio Kaplan (tactic problems), Johan de Koning (bad positional moves after openings), Ed Schröder (much problems in endgames), Richard Lang (passive playing style but in all playing phases with good chess, same for Kittinger programs). But all of the chess computers are really very nice at this times and of course today. I like all of it.

U.S. Elo is around 350 Elo higher!

After my own calculation Mach IV is playing with 1.950 Elo. In the new SSDF 25 Elo stronger. 1.975 SSDF + 350 US Elo = 2.325 Elo (should be right). But I am thinking again 25 Elo to high.

Around 1990 - 1992 the chess club from St. Petersburg visiting our chess club to Neuss / Germany for tournaments. We had a lot of fun with the strong players of St. Petersburg. Like IM Donchenko (2.420 Elo) for an example and other strong players from 2.250 - 2.400 Elo. Also Germans like Rene Borngässer (2.450 Elo) plays the tourneys we organiced. All the persons plays a lot of games vs. chess computers at this time.

Example:
Rene Borngässer on 486 / 33Mhz Genius 2 / 3 made around 70-80% of points without any problems. Vs. Risc 2500 players with 2.250 / 2.350 Elo made 70% of points. I am playing at these times with 2.100 Elo and made 55% vs. Risc 2500, 60% vs. Sparc 20Mhz, 45% vs. Lyon 68020. And I collected much more games from many chess players vs. chess computers.

So, if Elo is measurable for chess computers ... I think in reality not measurable ... max. level for Risc 2500 with 2.075- 2.100, Berlin 68020 a bit higher with 25 points, Mach IV are playing with 1.950 Elo. Polgar / MM V with 10Mhz ... max. 1925 Elo.

So Schachcomputer.info ELO is 150-160 to high for active and tournament rating systems, SSDF for the moment only 20-30 Elo to high.

But it's right that chess computers can used hashtables with faster processor are the first really good machines for strong club players. All other computers with 8bit and without hash (hash is around 100-125 Elo, in endgames often much more) have in most of cases big problems in endgames. Players with 1.800 Elo in our chess club in the past made more as 50% vs. Polgar, Milano, MMV with 5Mhz.

Also Nigel Short update for Milano is max. 25 Elo better as MMV / Polgar / Milano, not more.

Biggest problems we have with chess computers can be:
First time we can see a high level (with many pieces on board). Second time we see the bad engames. Humans are merciless.

First 20 games Human with 1.800 Elo vs. MM V 5Mhz = 12:8 for MM V.
Second 20 games from the same player vs. the same MM V 15:5 for the human player. That is fact and again the biggest problems we have with Elo calculation for older chess computer.

Same with the current situation in computer chess.
Stockfish with 1 Core on 4Ghz = 3.350 Elo.

GMs are surprised about the knowlege in openings for Elo monsters, like Stockfish, Komodo, Houdini.

Stockfish = 2.700 Elo after openings, 3.100 in the early middlegames, 3.400 Elo in the transposition into endgame and 3.600 Elo in endgames. So long we have not a better rating system, so long all the computer chess ratings are bad. We can't give Stockfish generally 3.350 Elo if the situation in the different playing phases are quit different.

And I know what I wrote, I have over 500.000 of such games on my site and over 2 millions of such games computer engines plays here all the years on my systems. I have also 9.200 chess computer games I never create an official download from my first times from the years 1979 - 1994 with older chess computers.

So I build a very hard opinion about it and know that the most of readers and fans of chess computers will not read it. But that is the reality, ratings for chess computers are all the years in SSDF wrong and in reality is ELO for older chess computers not measurable.

Best
Frank

PS: But I will looking in the afternoon in your interesting statistics. I like such stats becasue I know ... not measurable, but I try with my own Elo rating systems again and again to give chess programs an Elo.

[Yarc]

There will be a match between International Master Roman Vidonyak and the new King Module. Maybe this time Roman Vidonyak will lose.

Last time he won the Match against the Lang Module:

http://forum.topschach.de/chess-genius- ... t1125.html

Should be available very soon

Yes, I remember reading that article. After the match Roman Vidonyak said he enjoyed playing the Lang module so much he was going to buy one.

I shall be very interested to see the result of the new match with the King module to see what he makes of that.

Ray

[Frank Quisinsky]

Hi,

what I like in the past ...
The dealer "Niggemann" in Germany. I purchase many chess computers from him. Often I search a disussion with him because his opinion about playing strenght to chess computers are great. All the time he has right with Elo for chess computers.

Example:
I order a chess computers, asked for Elo strenght in young years. Niggemann gave information about Elo strength = 200 Elo lesser as SSDF. I like that.

Niggemann wrote a review to Saitek Blitz. He explain that the Saitek Blitz with tournament level will be much stronger. And from 3 to 6 minutes Elo goes high with 125 Elo or maby 150 points. People like Niggemann saw all that in the past and gave users like me realistic opinions.

"Schach der Großmeister" in German TV. I have the chance to asked in the past one of the grandmaster ... why not analyzes from Lyon 68030. The answere ... not strong enough with 2.250 Elo max.. And the grandmaster are right. Yes, if I looking in Lyon 68020 results I have ... max. 2175 Elo.

So computer chess are for people like GM Pfleger not really a topic at this times. But all of the grandmaster know ... with many pieces on board chess computers can find interesting things. And the computer chess area started ...



Best
Frank

An other example:
Millennium 2000 publicity for Chess Genius 48Mhz ... 2.000 Elo (I am very happy to read that, absolutely OK).
Millennium 2000 publicity for Chess Genius 120Mhz ... 2.200 Elo (I am not happy, 100 Elo to high).

[Yarc]

Wonder if Millennium will sell the Exclusive board separately. Don't need two Lang's.

I think they should really especially for people wanting two separate boards one for each module. It's crazy otherwise!

Millennium 2000 is down for a few days. Have they changed the link or gone bust?

I've also noticed this but guessing they are updating their site with information on the new King module. However, their web site has not been working properly for a long time. Sometime ago they had information for both the King and Blutooth modules but this disappeared.

I don't think they have gone bust, or at least hope they haven't. The new King module is set to be released this November so they are still around. Just hope their web site appears again!

Watching and hoping Regards
Ray

[Frank Quisinsky]

Hello Ray,

TheKing is max. 100 Elo stronger as Genius. So I think with 300Mhz 2.300 - is realistic. But maybe Johan de Koning made big steps in the last years, I don't know.

More interesting are programs with human style like Wasp, Hiarcs or programs very aggressively like Wasp or Rodent, Spark, Junior!

Wasp is around 400 Elo stronger as the last TheKing versions tested in CEGT. Such a program with 300 Mhz can make clearly over 2.600 points with a fantastic style of chess and a lot of opening understandings.

At the moment a test-run from Wasp is still running on my site. Load the games I have and looking ... how many fast games Wasp lost vs. clearly stronger opponents. More sense ... to search all the games Wasp won vs. cleary stronger engines very fast. Wasp is the optimal engine with many pieces on boad ... what humans like to see.

But again ... let us wait of the first TheKing results. In the past TheKing is playing very nice computer chess (older chess computers, Montreux, Risc 2500 and R30/R40). I can understand all the fans from Johan de Koning. But all in all I am thinking, today much more interesting engines are available as the old TheKing.

Rodent is really a very nice chess engine!

Best
Frank