Hosted by Darius, founder of Chessengeria.eu, April 2023.
Source: SedatChess
„(…) we have great talents in book making, where many people benefit from their work.”
Welcome to an interview with Sedat Canbaz a computer chess expert, who has been active in the field for over two decades.
In this interview, he shares his passion for computer chess, talks about his motivation for creating and improving chess opening books, and discusses the challenges he faces while creating a new versions. We will also learn about the history of Perfect Opening Books, the importance of using good quality databases, and the evolution of computer chess over the past two decades.
Let's dive in!
1. Darius:
When did you become interested in computer chess?
Sedat Canbaz:
My love over Computer Chess is started about 25 years ago, E.g during 1997-1998 and my 1st favorite programs were Arasan, Chessmaster, Fritz etc.
By the way, in these old good times, I often ran Duels (manually by hand): Chessmaster vs Fritz. Sure nowadays all my organized tournaments are done automatically)
But in the past, there were no alternatives to run all auto... Sure later I ran these kind of matches in Auto-232 mode too.
2. Darius:
What prompted you to create your first computer chess book ?
Sedat Canbaz:
I realized to create and work over opening books.. simply due to in the past, there were not so many well-tuned strong books. E.g I've noticed many BIG holes in others available books, where the engines were falling in a BIG disadvantage start.. So for this reason, the idea is born to work on better/stronger opening lines.
3. Darius:
Tell us about the history of Perfect Books.
Sedat Canbaz:
My 1st created book belongs to early of 2000s, and named it 'Draw'. Later, I followed a close friend's suggestion, renamed as 'Perfect'.
First Perfect book vers. were based on only GMs games, but later since 2006, I created deep books, which were based on engines games. E.g in 2007-2009, I tested Perfect books via online (on Playchess server).
That made me happy that in my each trying: I've got as best performance!
Even in those times, my used hardware was not as fastest, but I think that Perfect book's deep lines played serious role for getting better points. Plus in those times, a new idea is born: running book competitions too. I mean, Perfect books played as a 'key' of organizing these book tournaments. And in the past, there were no serious tournaments, e.g to be as book players (chess engines - Darius' addendum).
But later these kind of book tourneys became popular.. And now except SCCT, there are many other chess friends, who started running similar tournamentss!
And if I caused over this.. how I'm happy ! :)
4. Darius:
What motivates you to improve Perfect Books ?
Sedat Canbaz:
Well, usually the critical weak openings are forcing me to create New book version.
It's true that mostly my used opening lines are well-tuned/optimized... But like in every book, Perfect books include holes too and as other main motivation to improve comes from the name 'Perfect' in short, I'm doing my best the name to be deserved :)
5. Darius:
What are your goals in creating the next version of Perfect Books ?
Sedat Canbaz:
As usual, the target will be to work over new stronger lines, ideas... E.g in the past, some lines worked well with 2600-3000 Elo engines, but nowadays mostly latest Top engines (3500+) suffer with same openings.
In other words, my goal will be that: to create a book, which will have more chances to get a win... I mean not only as White plus as Black too. Because Chess is a game for both sides, but this is also true that with superior strong opening lines: it's not soft touch to get a win - sure i'm referring to especially if we run under optimal test conditions.
6. Darius:
Artificial intelligence has been increasingly making its presence felt in various fields
in recent months. In your opinion, can AI help improve chess opening books in the future ?
Sedat Canbaz:
Sure AI helps on improving..e.g Perfect 2023 book is based on only 3700+ NNUE games.
7. Darius:
What are the most important things you need to have a successful openings computer chess books ?
Sedat Canbaz:
A long story, in short, in my opinion, one of most important factors, which are required: the planning book have to be based on very good quality database, sure this is not all. Fast hardware is needed, testings via engines at least 3300+, sure higher better...
There are many other important things, which have to be used, but no way to explain all.
But there is one true, experience is required for sure...e.g just via importing games. No BIG chances to create a very successful book...at least my testings proves this...!
8. Darius:
What part of creating a new version of Perfect Opening Books do you find most challenging?
Sedat Canbaz:
I find them as most challenging if running matches on same PC, usually by Top engines (3500+).
9. Darius:
You are a well-known figure in the computer chess arena for more than two decades. Looking through the prism of these past years, has much changed in computer chess in the broadest sense compared to the present ?
Sedat Canbaz:
Yes, a lot of things are changed from the past to the present, except my love to Computer Chess :)
Actually another long story :) But in short, e.g in the past, the old machines were usually single-core (1 CPU) but nowadays dual machine may include hundreds of CPUs, now we can imagine the difference and in same time the chess speed too.
Plus in the past, the available Chess GUIs were quite limited with options...e.g without
Adapters were not possible to test all engines, but at present days as we know almost all engines can be tested in auto system on same platform and without using the adapters etc.
As other main difference (in the past), the number of the engines were small..but in 2023 I'm not going to count all, because I may miss some..but probably over than 1000 engines.
And before 2000s, the strength of mostly chess engines were below than Super GMs, but nowadays, there are a lot of chess engines with playing strength over than 3000+ Elo...
Meanwhile, especially since 2006 (after Kramnik vs Deep Fritz), the official Man-Machine duels are stopped, reason no way to beatable these kind of Top engines, except in Handicapped condition).
10. Darius:
There are many opening books on the market, both commercial and free ones created by enthusiasts.
What should a chess player and a fan of chess engines pay attention to when choosing an opening book ?
Sedat Canbaz:
Sorry, here I'm not going to suggest which books exactly...due to I may miss to mention some names :)
But no doubt that we have great talents in book making, where many people benefit from their work.
However, anyone is free to visit and check about the latest played book competitions results. Note also that, the strongest deep opening theory is changing dramatically very fast... Today we may have a book with superior performance, but tomorrow the same book may suffer. Sure to avoid this (for better performance): it's required quite frequent updates for sure.
On other hand, if you are online player...go for deep opening books. E.g just in that case, the latest Perfect opening book will suffer, because the online opponents will use to play
likely a book via deeper opening lines... Something like NNUE vs NON-NNUE eng matches). But on the same PC (if our goal is determine the real engine strength) Perfect 2023 seems to be not bad.
11. Darius:
In the struggle of the strongest chess engines, which present a generally similar level,
is the chess opening book is or can be a decisive factor in determining the victory of one of the sides ?
Sedat Canbaz:
Well, I think that (to see many wins) it all depends on our used conditions, but I agree that tournaments with 3700+ NNUE engines (under serious conditions): fast hardware, strong openings, MP, Ponder ON etc. even at Blitz, no way to appear many wins, close to 99% draws. The reason due to NNUE engines are gaining from previous played data (Evalfiles). But in case as NON-NNUE (engines), sure then the situation is changing, e.g the overall stats show minus plus 10%-15% more wins.
Of course, there are ways to reduce the Draw-Ratio, but if using very weak openings, sure then we see more FUN, but then thousands of games (per player) are required, otherwise the Error Margin can be too high... I mean in the case of determining the real engine strength.
12. Darius:
Does a chess opening book intended for inter-engine play differ from an opening book dedicated to human chess players ? If so, with what ?
Sedat Canbaz:
The strong deep engine books may help especially to Correspondence players, actually they gain...
Frankly quite frequent, many people contact me, asking for a advice about which books to use. And as usual, my replay over this issue is to check my latest organized book competitions.
Btw, there are books, which are quite strong in TREND openings, some are better in OVERALL. That means, for better conclusion: End user have to check about which are valuable for his needs.
On other hand, the weak human players may try to gain as well.. but how much will be useful ? :)
This is other question of course.. for this, I suggest to be used a book based on humans games, because the strongest engine opening theory vs human theory are completely different to each other!
Just one example, many humans are doing well with Alekhine, Dutch, Scandinavian etc. openings, where mostly of the Top chess engines (as Blacks) e.g with the mentioned openings suffer...
At least I can say, Top Engines winning percentage of these openings are below than 40%-35%.
13. Darius:
Do you set any goals to achieve in the field of computer chess for the future?
Sedat Canbaz:
Usually for the future, my goals are to run and test new engines plus new books too.
And as you may see too, I'm trying each new tournament to be different than previous one.
Otherwise, it's boring for my side, e.g if each time running same kind of testings.
14. Darius:
What tools / software / sources do you use when working on Perfect Opening Books
and testing chess engines ?
Sedat Canbaz:
Well, for opening book creations: I work mainly with ChessBase, SCID..
For testings, I use various GUIs...but nowadays mainly CuteChess GUI. It's a good choice for engine testings, especially for Concurrent matches..
15. Darius:
Are there any chess openings that are particularly favored by you?
Sedat Canbaz:
One of my favorite openings are Sicilian, Giuoco-Piano, Semi-Slav, English etc.
Darius:
Sedat, I would like to express my gratitude on behalf of the readers of Chessengeria.eu and myself for taking part in the interview and sharing your expertise in developing one of the top-rated chess opening books worldwide.
I hope that you will continue to thrive in your work with Perfect Opening Books, and that your enthusiasm for the computer chess will stay with you for many more years to come.
Sedat Canbaz:
As final words,
I wish to say many thanks for your great efforts too.
Chessengeria seems to be one of the best chess sites!
Keep up the good work.
Darius:
Thank you once again for your time and contribution, and I wish you all the best in your future endeavors.
NOTE:
Further publication and reproduction of this interview is permitted only with the prior permission of Darius, founder of chessengeria.eu
The condition is that the content is preserved in its entirety and that a link to chessengeria.eu is included.
In the Files area you will find free chess engines for download.