Dann Corbit, where can I download your Opening Book ?

[Chessqueen]

It is the SuperBook created by Dann Corbit. The SuperBook has a book depth of 100 half moves to be used with Arena 3.5.1. I can NOT find the Link, I am also wondering if there are any other GREAT Opening books in ABK to be used with Arena that you recommend

[Dann Corbit]

I will see if I can find it.
If you want the best books, technology has come a very long way and the books from Sedat's contests are pretty well guaranteed to be stronger.
Any top ten book from one of Sedat's book contests will beat a book written twenty years ago that was derived from super GM and correspondence play.

[Chessqueen]

I will see if I can find it.
If you want the best books, technology has come a very long way and the books from Sedat's contests are pretty well guaranteed to be stronger.
Any top ten book from one of Sedat's book contests will beat a book written twenty years ago that was derived from super GM and correspondence play.

Thank Mr. Corbit, where can I find Sedat's book and is this the best opening book for Arena at the moment ? In your Opinion do you believe that it is better to use an opening book that is only about 6 moves instead of long opening book with 12 or 20 moves, since I want to give different Engines the chance to use their own opening knowledge instead of starting from the middle game. Another thing to considered is that some of these opening books are created by humans GM which in my personal opinion if you give any engines up to only 6 moves they can continue finding better lines then those created by GM specially engines rated above 3000

[Dann Corbit]

The new opening books were revolutionized starting with the Cerebellum book.
From that point forward, opening books became enormously strong, giving about 50 Elo to the top engines.

Here is Sedat's current contest:
https://sites.google.com/site/computers ... book-cs-53

[Dann Corbit]

This contest is probably a much stronger measure of book performance, because the programs are all the same, to the only variable that changes is the opening book:
https://sites.google.com/site/computers ... book-cs-52

[Chessqueen]

This contest is probably a much stronger measure of book performance, because the programs are all the same, to the only variable that changes is the opening book:
https://sites.google.com/site/computers ... book-cs-52

Okay, so knowing that 12 or 16 moves Opening books and also 7 piece EGTB are unfair to test Engines that are very close within 100 rating since all that they have to do is play about 40 to 60 moves in the middle game and if the program rated about 100 rating lower has an initial advantage playing with white it can keep that advantage up to the point where the 7 piece EGTB Kick in, therefore, testers are not really testing the best engines here, unless they only use opening boos that are up to 6 moves and EGTB of only 5 piece to give the better Engine a chance to show why there are better than the other engines with inferior chess knowledge. Don't you believe

[Dann Corbit]

I think we can measure several different things.

We can measure engine strength. The most important thing here is symmetry. In other words, if Engine 1 plays opening X as white, then Engine 2 should also play opening X as white. If the engines exit the book in different openings or at different points, then we are not measuring engine strength.

We can also measure book strength. To measure book strength we should use the same machines and the same engines and vary only the books. Then we know which book performs better.

We can measure Engine + Book strength. To do this we should keep the hardware the same and let the engines use their own books. This does not tell us how strong the engine is. It does not tell us how strong the book is. It tells us how strong engine + book is.

We can measure system strength. That means use any hardware, any engines, and any books on any system. What we measure here is total strength of system. But we do not know how strong the individual pieces are from this approach. Only that the winning system performed the best.

We can play engines without any books, but that gives us a zillion English and French opening games and everyone will get bored of it.

It is clear that a good book can make an engine play better. Same with better hardware or longer time control.
So I think it is important to try to understand exactly what your goals are. If you are running a chess engine contest, then what are you trying to measure? After you decide that, then you can pick the components that match that plan.

[Chessqueen]

I think we can measure several different things.

We can measure engine strength. The most important thing here is symmetry. In other words, if Engine 1 plays opening X as white, then Engine 2 should also play opening X as white. If the engines exit the book in different openings or at different points, then we are not measuring engine strength.

We can also measure book strength. To measure book strength we should use the same machines and the same engines and vary only the books. Then we know which book performs better.

We can measure Engine + Book strength. To do this we should keep the hardware the same and let the engines use their own books. This does not tell us how strong the engine is. It does not tell us how strong the book is. It tells us how strong engine + book is.

We can measure system strength. That means use any hardware, any engines, and any books on any system. What we measure here is total strength of system. But we do not know how strong the individual pieces are from this approach. Only that the winning system performed the best.

We can play engines without any books, but that gives us a zillion English and French opening games and everyone will get bored of it.

It is clear that a good book can make an engine play better. Same with better hardware or longer time control.
So I think it is important to try to understand exactly what your goals are. If you are running a chess engine contest, then what are you trying to measure? After you decide that, then you can pick the components that match that plan.

That is a very intelligent way to look at it, and we should be looking for better way to test Engines from now on, and like you stated " The most important thing here is symmetry. In other words, if Engine 1 plays opening X as white, then Engine 2 should also play opening X as white. If the engines exit the book in different openings or at different points, then we are not measuring engine strength." But also the EGTB should be no more than 5 piece to give a chance to the engine with better chess endgame knowledge.

[Dann Corbit]

I do not see any reason not to use 7 man EGTB files. In fact, if 8 man files were available, I would prefer if they use those.
There is no reason to fumble in the dark, when you have a lantern, and so does everyone else.