New release Pedone 1.1

[Graham Banks]

......CCRL are testing to many clones.........

Define a clone.
Who decides what is a clone and what isn't, and what criteria do they use?
Are the criteria widely accepted by the computer chess community.

[Frank Quisinsky]

Hi Graham,

I am looking in the playing style and stats I created. I can see a lot and in most of cases I can see derivative. I have furthermore a database of test positions, this helps me.

You are right, not in all cases I can see it ... in the most directly.

Each clone in a Rating List produced bogey opponents.
The result are influences.

It's complicated what we do in calculating Ratings today. A decision what we do, which engines we can added will never right or wrong. But it's the wrong way to added all and to have such an opinion you give.

I think you are right and you are not right here. Say it before, all isn't easy to handle.

But I try to prevent each influences with the experience I have after so many years in testing engines. And which all my experience I produced often mistakes.

Important is more the fun factor we have.
Ratings will be never to 100% good.

Best
Frank

[Engin]

"Column bonus for backward, isolated and doubled pawns" ?

hmm... why bonus ? as i know mostly backward, doubled or isolated pawns are bad and they must give a penalty instead of bonus.

Ok, if you are right then i am understanding now why the texel method want give me everytime a bonus for doubled, isolated or backward pawns instead of penalty ... the world are going to crazy

[Fabio Gobbato]

I used bonus as a generic word, of course they are malus.

[supersharp77]

Thank you....Fabio Gobbato......loaded and it works.......now v0.2.1.. v0.3 and v1.1 are working.......Dunno if engine is close to 2700 though.....Thx
AR

[Modern Times]

Hi Graham,

1.300 games for 4 versions isn't not enough. OK, better as to have nothing. I prefer CEGT with more games and better to compare (Shredder = 2.800 Elo). CCRL Elo goes an own way and is clearly to high. Not easy to compare CEGT results with the results from others.
We discuss about it TalkChess.
IPON do it, CEGT do it ... CCRL goes an other way.
Frank

Ever looked at the SSDF ratings list Frank ?

http://talkchess.com/forum/viewtopic.php?t=55793

[Frank Quisinsky]

Hi Ray,

yes, 200 Elo to high.
In chess Computer times 350-400 Elo to high.
In the 90er years SSDF made some corrections and goes down.
But today and all the latest years clearly to high.

For around two weeks I discuss by mail with a gradmaster and I asking again. This guy analyze with Houdini his games. After his opinion not more as 3.050 on his actual i5 Notebook with 1 core (I don't now the MHz from his Notebook).

Allways the same opinions I got from strong chess players. In the beginning of the game Houdini maybe not more as 2.950, in the very late middlegame maybe 100 Elo or more better ... his opinion!

Good to compare with Shredder 2.800 Elo or Hiarcs 2.850 Elo.

In American the Elo calculation is around 150 Elo higher (I know).

Best
Frank

PS: I gave the grandmaster the information that for the beginning of the games he should try out Spark. He will do that, waiting of his answer I sent him around 10.000 Spark games, sorted to move number. The only comment I have so far ... a lot of fast won games to interesting actual opening systems, not many lost games ... I will have a look. Best tipp we can give such a strong player for his opening analyzes.

PS2:
Today I repair much things in my Book (C57, C58, C59, E96). To E96 (critical opening for engines) I have a lot of lines cancelt with "f" in my book. Engines need deeper lines. I analyse with Spark only and Spark gave perfect analyzes to the games I added from Sahovski Informator.

I know more and more why such Systems aren't very populare in GM-Theory. Very rarely because complicated. I will asked the GM why C59 is so rarely these days. Can't not see a reason because good for both sites if the right moves are played!

[Frank Quisinsky]

Hi Ray,

E96 is a good example.
Shredder give out to E96 often E95. ChessBase have also different problems, example to C16. More exactly is Shredder in give out the correct ECO code. Both programs have problems with transpositions comes very late.

Or in my chess books are mistakes, can't believe it.
Sorry, my actual Topic, working a lot on my Shredder Classic "Live Book".

Best
Frank

[Frank Quisinsky]

Hi Fabio,

now around 900 games are played.
Your engine isn't an attacker ...
14 games lost up to move number 50, 0 games won.
Some tactical holes, a lot of stuff for analyzes you have if you download the games.

Different of the "weaker" engines in my test group are very strong in endgames. Here the performance is not more as 50%. But the strength seems to be clear too. In late middlegame very interesting combinations, engine hold often draw in critical positions. A bit the same style Vajolet2 have (Vajolet2 is stronger in endgame). But maybe I am wrong, working more on my book so far, but made some anlyzes with the games I have.

No technical problems, time management is good.
Games can be download after round robin, I think in the night or tommorow in the afternoon.

Very good work after the time you are working on it.
I am happy to have your engine in my rating list.

Best
Frank

[Fabio Gobbato]

Thank you Frank!!! Thank you for your analysis, I will work to improve the engine.