Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / To many Draws in the engine room!
- - By mad max (***) [us] Date 2010-10-04 21:48
Could the cause be to many macros?..The floor is open:smile:
Parent - - By Nelson Hernandez (Silver) [us] Date 2010-10-04 22:11
Macros have nothing to do with it.  Level of play is up, way up in the last few years, and as Rybka dominance continues all games and all opening books tend to be Rybka-Rybka, which leads to more draws.  Things might change if another engine came along that was equal to the champ, but after so many years of supremacy it would take quite a bit of time to change the complexion of things.
Parent - - By Banned for Life (Gold) Date 2010-10-04 22:20
Disagree! The complexion would change overnight if the powers that be allowed all engines to play in the engine room. As far as the draw rate goes, playing a very small number of lines over and over again is a surefire way to end up with all draws, but this is hardly inevitable. If people put the effort into developing novelties and new lines, the draw rate would decrease dramatically. Of course no sane person will do this at this point in time because any played novelty will be in every book ten minutes later, due to the hyper kibitzing which the room is famous for. You really can't expect great things to come out of this sewer.
Parent - - By Nelson Hernandez (Silver) [us] Date 2010-10-04 22:35
You know, come to think of it, who benefits from restricting engines on Playchess?  Nobody is making a dime from engines being used or not used there.  Sure, the commercial guys don't want anarchy to ensue, but think about it.  This is a form of control, and I can't really understand what is the moral or institutional basis for it.  Why not let freedom ring?

As for the very small number of lines argument: if you actually examine the evidence, Playchess draws in B90 are not materially higher than all other lines.  This was a surprise to me, as I used to think like you.  What you propose is intuitive and logical, but the facts don't seem to support it.

Finally, I have a much better perspective from which to see how quickly lines are disseminated through the system.  While the kibitzers may put a move into book very quickly, they typically do not follow that line until it has been successful several times, and even then the copycats come one by one, not all in a rush.  From novelty move to wide propagation is more typically a couple of months and sometimes longer.  Ask me about any move in a particular position and I can tell you (privately) the history I have.  There are exceptions, of course.  Nonetheless if you wanted to start playing Nimzo-Indian I doubt anyone would follow you for many months.
Parent - - By Banned for Life (Gold) Date 2010-10-04 23:37
if you actually examine the evidence, Playchess draws in B90 are not materially higher than all other lines.  This was a surprise to me, as I used to think like you.  What you propose is intuitive and logical, but the facts don't seem to support it.

B90 has a reputation for being a very sharp opening, so the fact that it has only a slightly higher than average draw rate doesn't necessarily disprove my thesis. Can you give an example of the draw rate for a staid opening that's been played a gazillion times?

While the kibitzers may put a move into book very quickly, they typically do not follow that line until it has been successful several times, and even then the copycats come one by one, not all in a rush.

First, I am not concerned about people copying my lines. I see this as a good thing when it happens (very rarely). I am much more concerned with someone coming up with a drawing line and then having everyone copy it.

Second, there are actually two aspects to how kibizting interacts with .ctg type books and the chessbase gui to affect play. The first is the dissemination of lines, while the second is the tendency for everyone to play the same line (both because they are using the same database of games to generate their books and because of .ctg's preference for playing lines that have been played many times before). Since I play offbeat openings, the responding lines that people pick up and copy tend to be the less sharp and more drawish lines.

Nonetheless if you wanted to start playing Nimzo-Indian I doubt anyone would follow you for many months.

Probably I should work out some kind of arrangement with you to identify openings that are less well developed. Some of these will be more engine friendly than others (but not being engine friendly has its own advantages), but it's always fun to get people out of book early in the game, especially when there is a decent chance of getting into an unclear position.
Parent - - By Nelson Hernandez (Silver) [us] Date 2010-10-05 01:09

> Probably I should work out some kind of arrangement with you to identify openings that are less well developed.


If you picked up the last six months of games from the Chess Gladiators site, and mashed them with the last six months of CCRL and CEGT games, you could get a very accurate sample of the distribution and development of openings, which would come into even greater focus if you then excluded B90.  Then it would be a matter of grouping them thematically and figuring out what has the greatest promise and offers the player the most control.  You know how to do this; your whole career is based on canny judgments like these.

> I am much more concerned with someone coming up with a drawing line and then having everyone copy it.


The same holds true: it takes a while for drawing lines to become the main line because only some of the players are kibitzing, they are not necessarily incorporating your games every single day, and games are not released to the general public except for once a month. 

> Can you give an example of the draw rate for a staid opening that's been played a gazillion times?


OK, let's talk Petrov, ECO C42, considered a very drawish line.  I have it 57% drawn in Playchess contests year to date.  What does that prove?  I should mention that white players have an ELO 40 points higher than black in this line.
Parent - - By Banned for Life (Gold) Date 2010-10-05 17:55
What does that prove?

Without knowing the engine-engine draw percentage of B90, not too much. In human games, the draw percentage of C42 is 14 percentage points higher than B90. I would expect B90 to have a lower draw rate in engine-engine games too, if not for the huge number of games in this line.

The following OTB game statistics come from www.chessgames.com:

Number of B90 games in database: 6629
Years covered: 1926 to 2010
Overall record:
   White wins 37.9%
   Black wins 30.2%
   Draws 31.8%

Number of C42 games in database: 4752
Years covered: 1497 to 2010
Overall record:
   White wins 36.3%
   Black wins 17.9%
   Draws 45.8%

As far as picking out new lines, the criteria would be:

- Frequency of play - lower is better, never played is best! B27 was very good from that standpoint.
- High probability of leaving opening in an unclear position - An advantage for the side that's played many more games in the line. An even larger advantage for the side that has deep analysis in the line that can't be replicated in less than CC time.
- Engine suitability - Engines play some openings better than others. Picking openings that don't mesh with an engine is like swimming upstream.
- Positions that are badly evaluated by other engines - Engines are frequently unable to correctly evaluate opening positions. This combined with an uncommon line can lead to a significant exit advantage.
Parent - - By Ray (****) Date 2010-10-05 18:46
CCRL 40/40 database has:

B90
Games 7,980
Draws 31.2%
White score 52.8%

C42
Games 2,425
Draws 40.6%
White score 56.9%

http://www.computerchess.org.uk/ccrl/4040.live/eco_report_by_eco.html
Parent - - By Banned for Life (Gold) Date 2010-10-05 19:42
All things considered, those are pretty close, but I'll bet CCRL doesn't have any games from 1497! :lol:
Parent - By Ray (****) Date 2010-10-05 20:18

> All things considered, those are pretty close, but I'll bet CCRL doesn't have any games from 1497!


Quite correct !!
Parent - - By mikal (**) [us] Date 2010-10-04 22:55

>If people put the effort into developing novelties and new lines, the draw rate would decrease dramatically.


Examples of these lines? Many people already said this, but no one has the balls to prove it. If there are truly that much lines it wouldn't hurt to unleash a few of these then and prove you can decrease draw rate dramatically (that also depends on how you define "dramatic" draw decrease to a certain amount of time).


>Of course no sane person will do this at this point in time because any played novelty will be in every book ten minutes later, due to the hyper kibitzing which the room is famous for.


No, YOU can't do it. Just to defend your first statement and make your idea unprovable. Of course draw rate will decrease by letting your opponent win.

>You really can't expect great things to come out of this sewer.


Mind you, some top corr players certainly uses those as a reference. And I guess engine room also has some influence on commercial books like Rybka books or hiarcs books.
Parent - - By Banned for Life (Gold) Date 2010-10-04 23:21
Examples of these lines?

Most recently, I have been using B27 in my CC games with excellent results. This is somewhat surprising, given that people have figured out how to reduce similar lines that I play in A01 a tempo up to a dead draw.

No, YOU can't do it.

Actually, I've been playing off-the-beaten-track openings for years. I don't do database searches to see what's original and what's not, but there were bound to be a few novelties developed along the way.

Mind you, some top corr players certainly uses those as a reference.

As they would use any engine-engine game.

And I guess engine room also has some influence on commercial books like Rybka books or hiarcs books.

I assume the Rybka and Hiarcs books are over-represented in the openings commonly played in the engine room. I consider this a weakness, not a strength.
Parent - By mad max (***) [us] Date 2010-10-05 01:21
Thnx guys for the input..
Parent - - By mikal (**) [us] Date 2010-10-04 22:37
No, you're not just creative enough to introduce something new. (I know this will hurt your ego :D )

And if you're friendly enough you can get games too from other people, no need to be paranoid about those macro guys, in fact I welcome them. Of course you can't do this with your Gortt9 and Ghost Soldier nicks for having a rude reputation or until you solve your narcissistic ego problem.
Parent - By mad max (***) [us] Date 2010-10-05 01:19
not that this is part of the subject here, but since u brought it up, you must be one of those pathetic multi macro users yourself.a cheap unethical scum bag who's only reason for being  is to rip lines off to further your own selfish need for elo...haveing said that, you have a nice day.
Parent - By SR (****) [gb] Date 2010-10-05 06:41
Some people have reported your post as a blatant abusive attack that is not warranted .
I agree with this assessment. We want  to keep moderation to the minimal so please avoid this type of attacks (this also applies to some other posts in this tread).

So to all of you! Try to keep a more friendly tone. It should be possible to checkmate someone without also spitting and kicking them ;-)
Parent - - By Loboestepario (****) [us] Date 2010-10-05 00:28
What percentage are we talking about?
Parent - - By dragon49 (****) [us] Date 2010-10-05 02:59
In the context of the first post, what are "macros?"
Parent - - By ppipper (*****) [es] Date 2010-10-05 08:52 Edited 2010-10-05 11:40
macros are pieces of software used to automatically kibitz a big number of games without human presence. That way, you can store a big number of games quite quickly and if you know how to, you can use them next day in your books. More games, more information.

Theoretically, they are not allowed in the engine room. However, IMO playchess agree with macros, since they don't do nothing to avoid them. Thanks to macro users, most of playchess games are published for free every month.
Parent - - By Harvey Williamson (*****) Date 2010-10-05 08:59
Where does it say they are not allowed? There is an upper limit to how many games can be kibitzed in 1 session.
Parent - By ppipper (*****) [es] Date 2010-10-05 11:40
just a guess, search in the forum
Parent - By Fulcrum2000 (****) [nl] Date 2010-10-05 11:44

> There is an upper limit to how many games can be kibitzed in 1 session.


There is an upper limit to how many games can be kibitzed in 1 day.
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-05 13:28
pipper this is a never ending story ....
To see if you understand that we use macros to copy the games is just a matter of convenience and time savings. The same could be done manually.
The few who do, never copy a portion of the book. We only use the databases to query purely statistical. Or really believe that Eros Riccio, Alberto Guecci, or other strong players run are based on games played in Playchess??
On the other hand, is the only way to preserve the rounds, but would be lost forever.
I have a group of players that we exchange the games, they sent me everything they could copy and / or played, and then collect all the games, I will forward a large database of about 40,000 sets monthly.
It is not just a matter of accumulating games, you have to know what to do with them!

pipper esto es una historia de nunca acabar....
A ver si se comprende que los que utilizamos macros para copiar las partidas es sólo un tema de comodidad y ahorro de tiempo. Lo mismo se podría hacer manualmente.
Los pocos que lo hacemos, jamás copiamos una partida al libro. Sólo utilizamos las bases de datos para consulta meramente estadística. O realmente piensan que Eros Riccio, Alberto Guecci, u otros fuertes jugadores corr se basan en las partidas jugadas en Playchess???
Por otra parte, es la única forma de preservar las partidas jugadas, sino, se perderían para siempre.
Yo tengo un grupo de jugadores que nos intercambiamos los juegos, ellos me envian todo lo que pudieron copiar y/o jugaron, y luego de recopilar todos los juegos, yo les reenvío una gran base de datos de alrededor de 40,000 juegos mensuales.
No es sólo una cuestión de acumular juegos, hay que saber qué hacer con ellos!!
Parent - - By ppipper (*****) [es] Date 2010-10-05 14:00 Edited 2010-10-06 08:53
To be honest, I don't really care very much about people using macros and kibitzing. 80% of the games are based on the same lines, 90% of the not_draw_games are lost by engine's blunders which suddenly they analyze until very low depths, or they are won by 12+ cores machines, needless to mention the use of clones... Besides the mere interest of preserving games forever, I don't understand very well how serious opening books war can be based on 3 / 16 minutes games, when the statistical info may often be confused by deep 13 moves, or super (at the moment) 12 cores machines.

On the other hand, I think one should have the right to avoid his games can be macro-kibitzed, in the same way one may choose to send evaluations or not. This should be the next playchess update of great value for players, IMO.

In any case, it this is not possible at all, I would prefer that all the engine's games were made public directly by playchess, so that everyone could play with the same cards.
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-05 14:11
We can all play by the same letters, depends on the work and time invested.
It's like if you said you should donate a computer Playchess twelve cores to each player to be even ....
I reiterate, it is only for reference. Sometimes they appear interesting movements in a blitz game in a single machine. Ultracorr as looking, full of trash items, or Megadatabases, full of body parts wrong. Only consultation, final decisions, as you know, making each player on each move, unless the items are copied to the book, which will loose a book, or at best, a book drawish.

Todos podemos jugar con las mismas cartas, depende del trabajo y el tiempo invertido.
Es como si dijeras que Playchess debiera donar una computadora  de doce nucleos a cada jugador para que sea parejo....
Reitero, es solamente para consulta. A veces aparecen movimientos interesantes en un juego blitz, en una maquina single. Es como consultar Ultracorr, llena de partidas basura, o las Megadatabases, llenas de partidas humanas malas. Sólo se consulta, las decisiones finales, como tu sabes, las toma cada jugador en cada movida, salvo que se copien las partidas al libro,  con lo que obtendrá un libro flojo, o en el mejor de los casos, un libro drawish.
Parent - - By mad max (***) [us] Date 2010-10-05 20:21
Crawley i've been watching you defend your habitual kibitzing in these posts for months. I can see if you want to run "A" macro 24/7 which you do( sorry harvey but he does and we all know he does) But my question to you Crawley is why do u have 3 maybe 4 macros all going at the same time?I've seen the crawley macro, the ruben combs macro, the argentum macro,and your other one that i cant remember at this time..really bro why so many?:smile:
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-05 20:27
My nick name is not Crawley, is Crawly.
I have scary lost data, I make backup....   :cool:
Parent - By mad max (***) [us] Date 2010-10-05 21:55
?:yell:
Parent - - By ppipper (*****) [es] Date 2010-10-06 08:54

> Sometimes they appear interesting movements in a blitz game in a single machine


Please, could you post an example? Thank you!
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-06 09:09
Sure! 

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5.
Nc3 a6 6. Bg5 e6 7. f4 Qb6 8. Qd2 Qxb2 9. Rb1 Qa3 10. e5 h6 11. Bh4 dxe5 12.
fxe5 Nfd7 13. Ne4 Qxa2 14. Rd1 Qd5 15. Qe3 Qxe5 16. Be2 Bc5 17. Bg3 Qd5 18. c4
Bxd4 19. Rxd4 Qa5+ 20. Rd2 O-O 21. Bd6 Rd8 22. g4 Nc6 23. O-O Nde5 24. h4!!

By Auryn:

"As I suggested in another post, it wasn't me who found that h4 in that poisoned pawn game. It was in the book of the player "Baxus" (that's his nick on Playchess); all the merits for that move go to him, no idea how he found it. All I did was to have some luck to have found that game he played (Baxus - Cryptonite) with that strong novelty, and check if it was really a strong new move, and I saw it was, so I tried it in this corr game against Lowrance, hoping he didn't have that Borstnik - Quaresma corr. game, as if he had followed it and had continued with Rxd6 (instead of resigning) the game would have probably ended in a draw... but I was lucky again:cool:
As for finding moves in general (like that 23...Kg8 in the other game with Dronov I posted) which are not suggested as engines... you should consider that still exist some players who have human ideas in corr. games and don't rely only on engines' suggestions. Anyway that 23...Kg8 was not a human idea I had, it was just the "investigation" of the third or fourth move suggested by engines. "

By Crawly

It is only an example.
If you look blitz games analysing your games, and found a new move, and maybe is strong, investig, analyse, try, test, and can to be or no to be a strong move, but have a way for work...
I fix my book and play corr with my database, it have all games (blitz and slow).   :wink:
Parent - - By Banned for Life (Gold) Date 2010-10-06 12:54
24. h4!! is almost certainly NOT the result of engine play in a blitz game, but rather long term (probably interactive) analysis. After spending a great deal of time on this move, the discoverer made the mistake of playing the move on the Playchess server where it was quickly scooped up by Eros, who to his credit, actually looked at the game and recognized the high quality of someone else's analysis. Eros was then fortunate enough to get a chance to use this move in a high level CC game. Now the move is well known and no longer has any real value, as anyone with any brains will avoid this line unless they can be assured of a draw (if this is an acceptable result).

So let's recap:

1) Someone did some serious analysis to come up with 24. h4!!
2) Eros, showing his skills as the ultimate kibitzer, scooped it up off PCC.
3) The discoverer probably never got to use this innovation in a serious game.

This pretty much summarizes the pluses and minuses of allowing kibitzing on the Chessbase server.
Parent - - By Rubén Cómes (****) [ar] Date 2010-10-06 14:24
Theirs are only hypothetical conjecture.
The only truth is the reality.
Search Baxus games and you will see the progress line. Baxus improved the line game after game..
This is the only game Corr I've found with the line in question:  Borstnik, Ales 2466 - Quaresma, Eng. Luis Manuel T 2335 (1/2-1/2) European TC VII Final (Chequia-Eslovenia) 31.12.2008
Again, this is only an example. There are many cases. But, mine is not teaching, is the kibitz ...

Good luck.  :cool:
Parent - By Moz (****) Date 2010-10-06 17:27
It speaks volumes that 24.h4 was largely ignored for two years despite appearing in publicly available high level corr game...
Parent - By Kreuzfahrtschiff (***) [de] Date 2010-10-05 12:05
its the opening books, they are to good now, if u havent them it would be more decisive games
Parent - By FWCC (***) [us] Date 2010-10-05 13:59
The level of play is up of course and with Rybka people find refutations at an alarming rate.If you want to win at PCC you had better at least purchase a 6 core system.The winners over at PCC seem to be the BIG hardware owners with x980 or better.A good book will put you ahead for a couple of days until another comes out and your lines are crunched.I would say simply the level of play has increased hence the increase in draws if your stats say so.
Parent - By M ANSARI (*****) [kw] Date 2010-10-05 19:45
There have always been a lot of draws in the Playchess room.  I haven't been on for some time, but I would have thought that with Rybka 4 that things would change dramatically.  Rybka 4 has a lot of ELO points to improve upon from the default depending on the time control you want to "specialize" in.  There usually is a dramatic change in productive lines with a new release of Rybka, but then again some of the lines that are in most books on Playchess are extremely solid and have been analysed to death to what is the equivalent of more than cluster strength hardware.  In some of these critical lines some people have had their computers analysing that one critical line for days or even weeks.  Add to that the fact that tools such as Monte Carlo and Idea can be used and you can be pretty sure those lines are leak proof.
- - By Moz (****) Date 2010-10-05 23:21 Edited 2010-10-05 23:26
From my perspective as a corr player, there are two big problems infecting the engine-room culture:
 
a) Severe myopia
b) Epistemic closure (i.e. far too much deference for ancestors, predecessors and engines in general) 

Everyone's so busy trying to break new ground, they never consider the possibility that the original site survey might be faulty.
Parent - By Nelson Hernandez (Silver) [us] Date 2010-10-07 17:56
An enigmatic post.  I wish you would elaborate.
Up Topic The Rybka Lounge / Computer Chess / To many Draws in the engine room!

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill