Some notes about openings!

[Sedat Canbaz]

Dear Chess Friends,

I decided to open a thread about short neutral engine openings (up to 8-10-12 moves)
Due to such thread is missing...!

To be honest,
I opened this thread it's not because to I wish to increase the Perfect book users...
I have thousands of users...it's ok...this is not the reason
The reason is that about why I opened this thread,
It makes me feel very sad...in case of testings with weaker openings...
Also I noticed that still many people have no clue about engine openings...
Plus I hope my current thread will make a little bit influence over the chess friends who prefer weak openings!

And now let's discuss about openings..because it's boring to discus only over origins, right ?)

So...I could not check all played openings of the available ratings...
Due to I have no much free time, plus some testings do not serve the games (e.g IPON)

And what I noticed (which I checked so far...),
Some top engines lost their games just due to using very weak opening lines...

Actually even in SCCT happens sometimes,
But not very often, because I make almost daily opening optimizations...
My main goal about Perfect 2014t book: its almost all lines to be approx. 55% Whites / 45% Blacks
I know very well that nobody work is perfect (including mine too)
But however,
The most important question: which neutral short book has more or less holes ?

So...I have some questions to those chess engine testers, who prefer oftenly weak opening lines:
-Why you prefer those weak openings, just because it is your favorite ?
-Or maybe because you have a lot of free cpu time ?
-Have you ever checked your played chess engines games?
-Did you look at the winning percentage of your used opening lines?
-Did you ever notice your opening lines to be less than 35 % winning percentage (played by your engines) ?
*Note:I mean about openings at equal engine strength based on many engine games!
*A little hint: e.g check please why Stockfish, Komodo, Houdini etc.. lost against weaker participants?!
-Did you ever make any opening move optimizations over the lines which suffer in your testings?

BTW, I can give you another example about the current issue (for those who use not so strong lines),
1)Imagine Michael Schumacher to drive a vehicle of 1990 years
or
2)Imagine Michael Schumacher to drive on damaged road?

Can we expect a good performance?)
Yes...a such picture will be fun...but only 'fun' no more no less...)!

Note also that we are in 2014...
And it's time for revolution for all engine testers too!

For more details:
http://www.talkchess.com/forum/viewtopi ... 9&start=30

And last (as I mentioned before)
I know very well too that working over openings is not so easy as it looks...
But difficulties are always way for success...
So we must overcome all difficulties and to bring the success!
My best wishes for less difficulties and to reach the great goal!!


Greetings,
Sedat

[vittyvirus]

Dear Chess Friends,

I decided to open a thread about short neutral engine openings (up to 8-10-12 moves)
Due to such thread is missing...!

To be honest,
I opened this thread it's not because to I wish to increase the Perfect book users...
I have thousands of users...it's ok...this is not the reason
The reason is that about why I opened this thread,
It makes me feel very sad...in case of testings with weaker openings...
Also I noticed that still many people have no clue about engine openings...
Plus I hope my current thread will make a little bit influence over the chess friends who prefer weak openings!

And now let's discuss about openings..because it's boring to discus only over origins, right ?)

So...I could not check all played openings of the available ratings...
Due to I have no much free time, plus some testings do not serve the games (e.g IPON)

And what I noticed (which I checked so far...),
Some top engines lost their games just due to using very weak opening lines...

Actually even in SCCT happens sometimes,
But not very often, because I make almost daily opening optimizations...
My main goal about Perfect 2014t book: its almost all lines to be approx. 55% Whites / 45% Blacks
I know very well that nobody work is perfect (including mine too)
But however,
The most important question: which neutral short book has more or less holes ?

So...I have some questions to those chess engine testers, who prefer oftenly weak opening lines:
-Why you prefer those weak openings, just because it is your favorite ?
-Or maybe because you have a lot of free cpu time ?
-Have you ever checked your played chess engines games?
-Did you look at the winning percentage of your used opening lines?
-Did you ever notice your opening lines to be less than 35 % winning percentage (played by your engines) ?
*Note:I mean about openings at equal engine strength based on many engine games!
*A little hint: e.g check please why Stockfish, Komodo, Houdini etc.. lost against weaker participants?!
-Did you ever make any opening move optimizations over the lines which suffer in your testings?

BTW, I can give you another example about the current issue (for those who use not so strong lines),
1)Imagine Michael Schumacher to drive a vehicle of 1990 years
or
2)Imagine Michael Schumacher to drive on damaged road?

Can we expect a good performance?)
Yes...a such picture will be fun...but only 'fun' no more no less...)!

Note also that we are in 2014...
And it's time for revolution for all engine testers too!

For more details:
http://www.talkchess.com/forum/viewtopi ... 9&start=30

And last (as I mentioned before)
I know very well too that working over openings is not so easy as it looks...
But difficulties are always way for success...
So we must overcome all difficulties and to bring the success!
My best wishes for less difficulties and to reach the great goal!!


Greetings,
Sedat

Opening books should be limited to personal use. In tournament, there use should be banned

[Sedat Canbaz]

Opening books should be limited to personal use. In tournament, there use should be banned

Your suggestion sounds good,
But testing the engines without books in tournaments is too early for nowadays
The reasons (in case of without books):
1)Many engines include same/similar opening code and it will be boring to see same or similar openings
2)Many engines suffer in case of playing without books, I mean the engines play not so strong lines...
3)Especially the quality of the Blitz played games will be fall down a lot...

Btw, I am looking forward to see X engine which will start to play: B90 Sicilian Najdorf opening ))
I wonder also, who will be this hero? Which chess engine programmer will manage to do this ?
And as far as I noticed so far, no any chess engine can play B90, and we will wait a lot...?
Note: B90 Sicilian Najdorf is one of the strongest engine openings for both sides!

One thing more,
I think those testings without games should be banned!
And definitely the weak openings should be banned too!



Best,
Sedat

[Sedat Canbaz]

Hello there,

No any comment so far...exception Syed's view

Strange indeed...actually I expected to see more comments...
Still there is no any replay to my questions...why ??
Maybe your used openings are super strong, that's why you are in silence ??))

So far... I am only the person, who is trying to help about this issue!
And don't forget to say thanks next time to me)) because I spent some efforts ...

There is no doubt that,
This thread is really very important!

And I hope to see more postings about weak or super strong positions,
where other chess friends can benefit...

Note that I could not paste the below weak opening lines,
Because there are a lot of similar weak positions, no time for all sorry...

Note also that,
We (engine testers) play daily thousands engine games per day...
And more than million games per year!!!
And I think thousands of games suffer due to critical, weak opening lines!
It's true that mostly of the played games are belonging to the strong lines

But however,
Unfortunately I noticed many many weak opening lines (I am still shocked)!

And without to not mention this I can't,
SCCT includes less weak openings than all, which I've seen so far!

Just to be more clear,
This thread is not for online tournaments, correspondence games etc...
It's just for engine testers who prefer short neutral openings!

And see below what is going...please look at the winning percentage!
Do you wish to enable those weak openings in Perfect 2014t book?
Believe me, then my book's winning percentage will be very high as others ))

A book based on recent CCRL games:


.
A book based on recent CEGT games:


.
A book based on recent FCT games:


.
A book based on recent SCCT games:
Note: I already disabled this opening, what about you ))??




BTW,I have good news,
Soon I plan to release my latest book (Perfect 2015t.ctg)


Best Wishes,
Sedat

[Sedat Canbaz]

Just I'd like to add,
Engine testers should check those opening lines too, which are ended as draw
When I say as draw,
I mean for example: Stockfish could not win over X engine, which is more than 100 Elo weaker

There are a lot of similar weak positions...
Where the Top chess engines suffer and don't deserve this!
And using those weak openings...it's something like a Russian roulette))

Note also,
Even a single different move in the opening... plays usually a big role!!

One thing more (as I mentioned before),
Mostly of the neutral books/lines are based on GMs games, for example there are many openings,
Where GMs of 2500-2850 Elo perform good, but unfortunately exactly on the same opening positions,
This time the Top Engines fall in a big disadvantage and that affect a lot the Engines Elo strength
Note also that the current Top Chess Engines are at least 500 Elo stronger than the Top Grandmasters

For this reason,
It's time to changed our minds...computers are not humans!
We should use openings which suit for their playing styles!!

And instead of beauty,
We should concentrate on performance and strength!!



Hopes helps...

[jpqy]

Games : 9300 (finished)

White Wins : 2661 (28.6 %)
Black Wins : 2112 (22.7 %)
Draws : 4527 (48.7 %)
Unfinished : 0

White Score : 53.0 %
Black Score : 47.0 %

Hi Sedat,

I use your perfect books from the beginning..i find them just perfect to run eng-eng matches and to build my lists.
Above is a example where every engine plays 600games..when there comes a new version out ,is stronger ,it replace a older version..
With this little list i passed 45.000games.
The openings are very well balanced..a much stronger engine has more chance to win from both sites..so it doesn't matter so much if one from your openings is not that perfect..if both engines can easy win with white for example..well one day there comes a engine who can win from both sites again with this less perfect balanced opening.
When you play enough games it will be cleared out..just check these percentages from above and you get 53% for white & 47% for black!

I just can say..continue what you are doing..and i can't get your perfect 2015t fast enough

You do a process for years..so your books can only grow in quality and give joy for engine testers!

Kind regards,
JP.

[Adam Hair]

Opening books should be limited to personal use. In tournament, there use should be banned

Your suggestion sounds good,
But testing the engines without books in tournaments is too early for nowadays
The reasons (in case of without books):
1)Many engines include same/similar opening code and it will be boring to see same or similar openings
2)Many engines suffer in case of playing without books, I mean the engines play not so strong lines...
3)Especially the quality of the Blitz played games will be fall down a lot...

Btw, I am looking forward to see X engine which will start to play: B90 Sicilian Najdorf opening ))
I wonder also, who will be this hero? Which chess engine programmer will manage to do this ?
And as far as I noticed so far, no any chess engine can play B90, and we will wait a lot...?
Note: B90 Sicilian Najdorf is one of the strongest engine openings for both sides!

One thing more,
I think those testings without games should be banned!
And definitely the weak openings should be banned too!



Best,
Sedat

Unfortunately, using a book (depending on how it is weighted) can lead to many games played with the same opening.

I was experimenting openings and noticed that certain openings were used many times when using Perfect 2014t (both the Arena book and the Polyglot book). I have seen this before with other books (which is one reason why I use a pgn of opening positions rather than a book), but it seems to be accentuated with Perfect 2014t because it is a smaller book (less possible number of openings). For example, in one 580 game match between Komodo and Stockfish using the Polyglot version of Perfect 2014t, the following position occurred 34 times:

[d]r2qkb1r/1p1b1ppp/p1nppn2/6B1/3NP3/2N5/PPPQ1PPP/2KR1B1R w kq - 0 8

With Arena, the situation was a little better. In a 380 game match, this was the position that occurred the most (16 times):

[d]r1b1kb1r/pp1n1pp1/2p1pq1p/8/2pP4/2NBPN2/PP3PPP/R2QK2R w KQkq - 0 8

All in all, only about 25% of the openings in the match using the Polyglot book were unique, and only about 35% of the openings in the match using the Arena book were unique.

Then I checked the games from your 15m 3s tournament that involved Komodo, Houdini, and 2 versions of Stockfish:
http://www.sedatcanbaz.com/chess/?page_id=135

This position occurred 44 times in that tournament:
[d]r2qkb1r/pb1n1ppp/2p1pn2/1p6/3P4/2NBPN2/PP3PPP/R1BQK2R w KQkq - 2 9

Since there is only 12 possible pairings in that tournament (treating reversed colors as separate pairings), some pairings used this opening more than once in your tournament.

I bring this up not because I think that Perfect 2014t is a bad book. Except for it being too drawish IMO, I think that it is a great book. However, I do not think that it is a suitable book for tournaments with a small number of participants. At some point the openings will be reused. And as I said above, this is something that can occur with other books too.

I recommend extracting all of the good openings to a pgn. You can ensure that the participants do not reuse openings if you use a pgn of openings.

[Ferdy]

There is no problem with probably weak openings as long as both sides will handle it. I have seen Graham did this in his series of amateur of tournaments.

[shrapnel]

There is no problem with probably weak openings as long as both sides will handle it. I have seen Graham did this in his series of amateur of tournaments.

In this case I support ol' Sadat Khanbaaz !
This is EXACTLY what he was criticizing . " Michael Schumacher driving a 1990 car OR driving on a broken, pot-holed road ! "..... the PERFECT analogy of what Mr Graham Banks is doing !

[Ferdy]

There is no problem with probably weak openings as long as both sides will handle it. I have seen Graham did this in his series of amateur tournaments.

In this case I support ol' Sadat Khanbaaz !
This is EXACTLY what he was criticizing . " Michael Schumacher driving a 1990 car OR driving on a broken, pot-holed road ! "..... the PERFECT analogy of what Mr Graham Banks is doing !

These best engines that we have have to be tested on unfamiliar grounds too. It can also be considered as chess opening exploration, we want to see how the best engine handles a little bit of a disadvantage. Probably we can see novelties here and there. We could be happy if a strong engine find a nice idea on how certain openings have to be played. This will contribute to human knowledge.