Whatever is current - Amazing Leela

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
zullil
Posts: 5279
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Whatever is current - Amazing Leela

Post by zullil » Wed Jul 03, 2019 3:13 pm

Ovyron wrote:
Wed Jul 03, 2019 4:52 am
zullil wrote:Perhaps if you posted even one example of such an "opening-busting game" folks might pay more attention to your posts.
Engaging in these discussions has been very revealing to me about what the general public knows about the current status and advancements of opening theory. For this I'd like to coin the term of "corking" moves. Apparently, people are working really hard into discovering critical variations that bust what people are mostly playing, and then they save them to a private book and leave them "corked". Then, when an important game comes up and the line gets to be played, it's "uncorked", just to win them that one game, and if you analyze the line at corr time control, you realize it's the best a side could have played, and that the other side is best avoiding that line (to the extreme, it seems white should avoid 1.e4, suprisingly.)

Before these discussions, I wrongly believed that once that happened, everyone knew the line. It lost all its value, hard work should be put into corking some other line. But instead, what happens is that that one game fades into obscurity, nobody knows about the line, so it can pretend it's still corked, so that it's a lot more valuable to cork lines because, while the lines of public books are in the open and they become obsolete after a quick time, the lines I'm talking about will have long-lasting life, and can be uncorked and uncorked again against opponents that turn out to lack the knowledge, that turns out to belong to a few privileged individuals. The worst thing that can happen is the line being featured as the weapon of a public book, then it's over.

This is great to know and I hope this offsets shooting myself in the foot for this, I can only imagine people in the know reading my posts considering me a blabbermouth that would be better keeping all this secret. Because it's all worthless once it goes mainstream.
OK, so you can't provide even one example from these games "that have been busting entire openings" (your words). Because, even though they have completely overturned centuries of opening theory, only a few enlightened individuals are aware of them, so the secret lines can be used over and over to win ... very important engine-engine games.

Sounds rather ridiculous to me. :wink:

Uri Blass
Posts: 8455
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Whatever is current - Amazing Leela

Post by Uri Blass » Wed Jul 03, 2019 3:19 pm

Not everybody is interested in opening for winning games.

I would like to see a big opening project for analyzing opening at least as big as the stockfish framework when everything is public.

http://tests.stockfishchess.org/tests

I see 2034 cores are used to test patches for stockfish.
Why not use a similiar number of cores for analyzing the openings and share the analysis?

For analyzing the italian
White has many possible lines in the italian.

You need to refute not only the main lines but also lines like 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.a3 or if the refutation is 3...Nf6 that seems a more forcing move then you need to refute a line like 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Qe2

Even if you show black's advantage in everyone of the not popular lines there are many other lines when white deviate from the main line also after move 4 so I do not believe that people already analyzed all of them and showed that in all of them black get an advantage that is more than white's advantage.

Uri Blass
Posts: 8455
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Whatever is current - Amazing Leela

Post by Uri Blass » Wed Jul 03, 2019 3:22 pm

zullil wrote:
Wed Jul 03, 2019 3:13 pm
Ovyron wrote:
Wed Jul 03, 2019 4:52 am
zullil wrote:Perhaps if you posted even one example of such an "opening-busting game" folks might pay more attention to your posts.
Engaging in these discussions has been very revealing to me about what the general public knows about the current status and advancements of opening theory. For this I'd like to coin the term of "corking" moves. Apparently, people are working really hard into discovering critical variations that bust what people are mostly playing, and then they save them to a private book and leave them "corked". Then, when an important game comes up and the line gets to be played, it's "uncorked", just to win them that one game, and if you analyze the line at corr time control, you realize it's the best a side could have played, and that the other side is best avoiding that line (to the extreme, it seems white should avoid 1.e4, suprisingly.)

Before these discussions, I wrongly believed that once that happened, everyone knew the line. It lost all its value, hard work should be put into corking some other line. But instead, what happens is that that one game fades into obscurity, nobody knows about the line, so it can pretend it's still corked, so that it's a lot more valuable to cork lines because, while the lines of public books are in the open and they become obsolete after a quick time, the lines I'm talking about will have long-lasting life, and can be uncorked and uncorked again against opponents that turn out to lack the knowledge, that turns out to belong to a few privileged individuals. The worst thing that can happen is the line being featured as the weapon of a public book, then it's over.

This is great to know and I hope this offsets shooting myself in the foot for this, I can only imagine people in the know reading my posts considering me a blabbermouth that would be better keeping all this secret. Because it's all worthless once it goes mainstream.
OK, so you can't provide even one example from these games "that have been busting entire openings" (your words). Because, even though they have completely overturned centuries of opening theory, only a few enlightened individuals are aware of them, so the secret lines can be used over and over to win ... very important engine-engine games.

Sounds rather ridiculous to me. :wink:
I agree and if somebody really has an analysis that busted entire openings then he can earn money by selling this analysis to the top chess players(rating above 2700) who are going to use the analysis in important public games so we are going to see the relevant lines in public games.

todd
Posts: 24
Joined: Thu Apr 19, 2018 7:09 pm

Re: Whatever is current - Amazing Leela

Post by todd » Wed Jul 03, 2019 4:31 pm

Uri Blass wrote:
Wed Jul 03, 2019 3:22 pm

I agree and if somebody really has an analysis that busted entire openings then he can earn money by selling this analysis to the top chess players(rating above 2700) who are going to use the analysis in important public games so we are going to see the relevant lines in public games.
This is also correct. The amount of money I have been paid for opening analysis for use in human play exceeds the prize funds of these engine tournaments. It's not a big amount or anything - the hourly rate (especially after electricity and hardware costs) is quite low, but chess analysis is my main hobby anyway.

Dann Corbit
Posts: 9782
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Whatever is current - Amazing Leela

Post by Dann Corbit » Wed Jul 03, 2019 4:34 pm

Ovyron wrote:
Wed Jul 03, 2019 4:52 am
Before these discussions, I wrongly believed that once that happened, everyone knew the line. It lost all its value, hard work should be put into corking some other line. But instead, what happens is that that one game fades into obscurity, nobody knows about the line, so it can pretend it's still corked, so that it's a lot more valuable to cork lines because, while the lines of public books are in the open and they become obsolete after a quick time, the lines I'm talking about will have long-lasting life, and can be uncorked and uncorked again against opponents that turn out to lack the knowledge, that turns out to belong to a few privileged individuals. The worst thing that can happen is the line for being featured as the weapon of a public book, then it's over.

This is great to know and I hope this offsets shooting myself in the foot for this, I can only imagine people in the know reading my posts considering me a blabbermouth that would be better keeping all this secret. Because it's all worthless once it goes mainstream.
I analyze every new opening book.
I have routines to pull them apart and I collect all of the suggested moves that way.
Then I analyze them all (of course, I only analyze those positions I have not already analyzed.)
I also analyze private or commercial books that I have no access to.
For instance, when Sedat has a tournament for opening books, I collect all of the pgn when he is done,
Then I have a tool written by Les Fernandez that atomizes all the games and classifies all of the positions:
who was white, who was black, who won, how many seconds time, how deep, what was the score, etc.
For those positions without depth and scores, those are book positions, so I perform my own analysis on those.
I guess I am not the only person doing that sort of thing,

Not just Sedat's contests, of course. All of them. CCCRL, CEGT, TCEC, YCLET, whatever I can get my hands on.
The only exception is contests without PGN.
I even include human only games and mine them for statistics and openings (e,g, TWIC, OMCORR)
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Ovyron
Posts: 2143
Joined: Tue Jul 03, 2007 2:30 am

Re: Whatever is current - Amazing Leela

Post by Ovyron » Sat Jul 06, 2019 4:19 am

Dann Corbit wrote:
Wed Jul 03, 2019 4:34 pm
I analyze every new opening book.
I have routines to pull them apart and I collect all of the suggested moves that way.
Then I analyze them all (of course, I only analyze those positions I have not already analyzed.)
I also analyze private or commercial books that I have no access to.
For instance, when Sedat has a tournament for opening books, I collect all of the pgn when he is done,
Then I have a tool written by Les Fernandez that atomizes all the games and classifies all of the positions:
who was white, who was black, who won, how many seconds time, how deep, what was the score, etc.
For those positions without depth and scores, those are book positions, so I perform my own analysis on those.
I guess I am not the only person doing that sort of thing,

Not just Sedat's contests, of course. All of them. CCCRL, CEGT, TCEC, YCLET, whatever I can get my hands on.
The only exception is contests without PGN.
I even include human only games and mine them for statistics and openings (e,g, TWIC, OMCORR)
Have you contended with what you have, against someone that wants to beat you really badly? What I had seen with similar approaches was Nelson Hernandez's book, which I have no idea of his size (but at some point he was an authority on novelties), and in practice, the best performing variations with the best engine scores could turn out to be losing. But he didn't know, until he actually played them out in a tourney and the opponent showed him what's what.

I wonder what is your main line for the Italian (both sides play the best moves that they can), though. Since we disagree on the outcome (black can force an out of book position where black has a "true score" higher than white), I wonder what would happen, if your mainline has black going wrong somewhere, or it goes into the critical line where it appears white has the edge but it's actually the opposite.

But at this point it'd be just selfish from my part to ask, since I can't discuss these variations, I guess I have become a leech.
Make someone happy today.

Dann Corbit
Posts: 9782
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Whatever is current - Amazing Leela

Post by Dann Corbit » Sat Jul 06, 2019 5:43 am

Chess analysis is a kind of quest for truth.
For some position, what's the best plan?
But it's not simple truth like "What's the integral of one over x from 1 to x?"
Well, sometimes it is, like a mate in nine where all other moves lose?"
More often it is more like, "For an arbitrary value of y, find the number x, which (when raised to it's own power) is equal to y."
Once in a while, like if you are given y=27, the answer is easy, namely 3.
But in the general case it cannot be solved in terms of elementary functions.
In the general case, we will perform some cyclic approximation which improves with each algorithmic cycle.
Similarly for chess, we can search from the root or use statistical processes or forward and backward analysis or some similar method to approximate the solution.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

corres
Posts: 1315
Joined: Wed Nov 18, 2015 10:41 am
Location: hungary

Re: Whatever is current - Amazing Leela

Post by corres » Sat Jul 06, 2019 6:48 am

As a position is closer to endgame as it has an exact solution and as a position is closer to start position as it is more possible the end result is draw.
Obviously with the spread of powerful PC-s and chess engines some of (human) chess variations became
obsolete and there are variations what became playable albeit they were considered being disappointed.
This is the natural course of the development.
But it is very improbable that the full of a very old opening will be considered being totally bad.
In the field of chess playing there are actual modes, that is all.

Uri Blass
Posts: 8455
Joined: Wed Mar 08, 2006 11:37 pm
Location: Tel-Aviv Israel

Re: Whatever is current - Amazing Leela

Post by Uri Blass » Sat Jul 06, 2019 9:03 am

Ovyron wrote:
Sat Jul 06, 2019 4:19 am
Dann Corbit wrote:
Wed Jul 03, 2019 4:34 pm
I analyze every new opening book.
I have routines to pull them apart and I collect all of the suggested moves that way.
Then I analyze them all (of course, I only analyze those positions I have not already analyzed.)
I also analyze private or commercial books that I have no access to.
For instance, when Sedat has a tournament for opening books, I collect all of the pgn when he is done,
Then I have a tool written by Les Fernandez that atomizes all the games and classifies all of the positions:
who was white, who was black, who won, how many seconds time, how deep, what was the score, etc.
For those positions without depth and scores, those are book positions, so I perform my own analysis on those.
I guess I am not the only person doing that sort of thing,

Not just Sedat's contests, of course. All of them. CCCRL, CEGT, TCEC, YCLET, whatever I can get my hands on.
The only exception is contests without PGN.
I even include human only games and mine them for statistics and openings (e,g, TWIC, OMCORR)
Have you contended with what you have, against someone that wants to beat you really badly? What I had seen with similar approaches was Nelson Hernandez's book, which I have no idea of his size (but at some point he was an authority on novelties), and in practice, the best performing variations with the best engine scores could turn out to be losing. But he didn't know, until he actually played them out in a tourney and the opponent showed him what's what.

I wonder what is your main line for the Italian (both sides play the best moves that they can), though. Since we disagree on the outcome (black can force an out of book position where black has a "true score" higher than white), I wonder what would happen, if your mainline has black going wrong somewhere, or it goes into the critical line where it appears white has the edge but it's actually the opposite.

But at this point it'd be just selfish from my part to ask, since I can't discuss these variations, I guess I have become a leech.
I wonder what you mean by "true score".

I understand that it is a score based on analysis of some tree but
scores of stockfish and lc0 may be different and score based on analysis may be different dependent on the time you give to analyze the leaves of the tree.

I think that it is not something that is well defined and as long as you did not prove win for black or draw then deeper analysis may give a different conclusion.

Dann Corbit
Posts: 9782
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Whatever is current - Amazing Leela

Post by Dann Corbit » Sat Jul 06, 2019 9:08 am

Uri Blass wrote:
Sat Jul 06, 2019 9:03 am
Ovyron wrote:
Sat Jul 06, 2019 4:19 am
Dann Corbit wrote:
Wed Jul 03, 2019 4:34 pm
I analyze every new opening book.
I have routines to pull them apart and I collect all of the suggested moves that way.
Then I analyze them all (of course, I only analyze those positions I have not already analyzed.)
I also analyze private or commercial books that I have no access to.
For instance, when Sedat has a tournament for opening books, I collect all of the pgn when he is done,
Then I have a tool written by Les Fernandez that atomizes all the games and classifies all of the positions:
who was white, who was black, who won, how many seconds time, how deep, what was the score, etc.
For those positions without depth and scores, those are book positions, so I perform my own analysis on those.
I guess I am not the only person doing that sort of thing,

Not just Sedat's contests, of course. All of them. CCCRL, CEGT, TCEC, YCLET, whatever I can get my hands on.
The only exception is contests without PGN.
I even include human only games and mine them for statistics and openings (e,g, TWIC, OMCORR)
Have you contended with what you have, against someone that wants to beat you really badly? What I had seen with similar approaches was Nelson Hernandez's book, which I have no idea of his size (but at some point he was an authority on novelties), and in practice, the best performing variations with the best engine scores could turn out to be losing. But he didn't know, until he actually played them out in a tourney and the opponent showed him what's what.

I wonder what is your main line for the Italian (both sides play the best moves that they can), though. Since we disagree on the outcome (black can force an out of book position where black has a "true score" higher than white), I wonder what would happen, if your mainline has black going wrong somewhere, or it goes into the critical line where it appears white has the edge but it's actually the opposite.

But at this point it'd be just selfish from my part to ask, since I can't discuss these variations, I guess I have become a leech.
I wonder what you mean by "true score".

I understand that it is a score based on analysis of some tree but
scores of stockfish and lc0 may be different and score based on analysis may be different dependent on the time you give to analyze the leaves of the tree.

I think that it is not something that is well defined and as long as you did not prove win for black or draw then deeper analysis may give a different conclusion.
I guess the only sensible meaning for true score is won, lost, or drawn. Anything else is clearly an approximation.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

Post Reply