FCT1: New rules for FCT1 or better not?

[Frank Quisinsky]

Hi Ferdinand,

yes, better is to test without endgame bases. We can see more mistakes and programmers have a good chance to find and fix endgame problems.

Interesting what you wrote ... in combination to your words I can create the following rule:
If more as 2.000 games in the database, my test ended and I can added an other engine. Nice, will thinking about it.

To Point 4:
That's right but my book must be good enough that an engine started with playable position. I have to look in each of the games. Per 100 games 2-3 games I must replay, because the start position is bad. To many work here since I am working on my actual book. I think that the database is clear to 99% but to many work I make me. Main reason that I must create a better book for testing. I am sure you will be happy if I have the results. You can be find a documentation later ... I am starting here monday next week.

Possible with Shredder GUI I believe. I will try out this if my new book is ready.

To Point 1:
Yes, that is my opinion and since I am using more different engines I have a lot of more fun in testing. I am sure I don't lost my interest in testing again if I have the "RIGHT" group of engines in my tourneys.

Best
Frank

[Frank Quisinsky]

Hi Adam,

thanks for your comments (read a lot from yourself in TalkChess in the past).

2) Yes I agree but I have the opinion an engine should find easy endgame combinations without databases. I have nothing agains endgame databases, great for analyzes, but for engine testing?

Complicated topic!

All the time I am testing with 4-pieces. Main reason is to avoid boring endgame back and force. Clear I am using all the time resign = off and my move average is 15-18 higher if I compare with the others. Without endgame databases the move average will be again higher.

On the other hand:
During the games I will see the results calculated to 100% by the engine.
Must thinking on the French programmer of Capture. A lot of extra knowledge in combination with 5-pieces he added. After I delete endgame databases he release a big tirade by mail and Capture lost after testing around 30 Elo Points. I am sure today the progammer is not happy with me ... for that reason.

I say it ... complicated topic!

In my opinion endgame databases are more interesting for study interesting positions. For a rating list maybe more interesting to find out playing strenght without endgame databases.

4). Yes, I think 40 in 20 was a bad idea.

5). Yes!
You will like my new book and the ideas I have in my brain. Later you can read all in a open documentiton, step by step what I do. Others can used the material for creating books in an other format too.

Thanks Adam!

Best
Frank

[Frank Quisinsky]

Hi Sedat,

Ingo is for other reasons not happy with me.
But I like the work Ingo do because Ingo created own ideas for testing and only this is important. Very interesting in comparing results is, that I can see no bigger differents. Only to things the reason is clear (Junior ... Need more time ... other good know reasons for differents in ratings).

Yes, the idea to test engines not stronger as 3.000 Elo I like a lot.

Shredder engine is to old.
With Shredder GUI I am very happy.

The old problem ...
My heart and all my ideas are added in Arena. So many years I am working with Martin and that was a very nice time. If I start Arena, that is the same as to look in a glass. Not good for me ... but I am very happy with Shredder GUI. All works fine with all what I need for testing.

Best
Frank

[Frank Quisinsky]

Hi Krzysztof,

good news!

I switched for 3 years my last mails with Stefan. I had a very friendly connection to Stefan all the time, made a lot of interviews with him. I know that he is working on the GUI and I know different details he gave to me. I am sure that a very nice new version will be the result.

I hope that my new book is ready to this time.
So users of Shredder have an interesting new book concept in additional to a new version of Shredder. I hope Stefan don't changed the opening book options.

Best
Frank

[Frank Quisinsky]

Hm,

Longer time control ... OK
Tablebases ... hm, and now?

Really a complicated topic.
Have a look in my answer to Adam.

OK, I will thinking again about it.

Best
Frank

[tpetzke]

Hi Frank,

maybe a compromise

Allow the engine to access it own table bases, means Gaviota can access gaviota tbs as this is part of the extended engine design. The same for other engines with their own tbs (scorpio its scorpio bitbases).

An engine that has no own TBs system plays without. No worry, you wont be flooded with custom TB systems. To create one is at least as difficult as to create an engine.

Thomas...

[Ozymandias]

The way I see it, table bases are, in a way, the reverse of the opening book. They give you absolute certainty for 7 pieces and less, while opening books are a reliable guide for when you have 32. Is the middlegame, that matters.

In any case, I would have no problem with deactivating TBs, as long as the GUI can access 7-men, for adjudication.