Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Chess Engines and the 50 moves rule
1 2 Previous Next  
- - By Linus (***) Date 2007-12-15 06:48 Edited 2007-12-15 06:51
On the Chessbase news I found an article about a RB vs NN endgame played between Sergey Karjakin (2694) and Alexei Shirov (2739) at the World Cup Khanty-Mansiysk which ended in a draw but is - according to six men tablebases - won for black in 208 moves!

http://www.chessbase.com/newsdetail.asp?newsid=4322
(scroll down to the second game in the article)

It is unlikely that humans can play this kind of endgame correctly, so I wouldn't worry too much about it in OTB chess. But it certainly does concern CC players who use engines and tablebases. FIDE rules are quite clear about the fifty moves rule. There have been extensions to 100 moves for certain endgames in the past. But as was stated in the article by a FIDE International Arbiter, all extensions to the 50-move rule were abolished at the Manila Congress in 1992.

So my point is: An engine (let's say Rybka) might "see" that a position will end in a won six piece endgame, simplyfy into this position deliberately and throw away half a point. Could (or should) the engine programmer avoid this?
Parent - - By Vempele (Silver) Date 2007-12-15 10:29

> On the Chessbase news I found an article about a RB vs NN endgame played between Sergey Karjakin (2694) and Alexei Shirov (2739) at the World Cup Khanty-Mansiysk which ended in a draw but is - according to six men tablebases - won for black in 208 moves!


That endgame's actually generally winning.

> So my point is: An engine (let's say Rybka) might "see" that a position will end in a won six piece endgame, simplyfy into this position deliberately and throw away half a point. Could (or should) the engine programmer avoid this?


Yes to both. Movei translates mate scores longer than 50 moves to high non-mates. This could be further refined (I don't know if Uri does this already) by considering how many pawns are left and what can be sacrificed to get rid of them - for example, KQPkq will (almost) never run into the 50-move problem.

Of course, you could also just generate DTZ (distance to zeroing move) tablebases.
Parent - By Permanent Brain (*****) Date 2007-12-15 14:36
The problem is, if Nalimovs indicate a mate in 50 or more, it does of course NOT mean that they are 50 move draws. Many of them will involve pawn moves or even captures, in time. In other words, the (big) mate distance in itself does not contain enough information.

I did take a look at the endgame mentioned, too: A useful online ressource for such positions is

http://chess.jaet.org/cgi-bin/dtx?fen=8%2F8%2F6k1%2F3NK3%2F4N3%2F5b2%2F3r4%2F8+b+-+-+0+49&.submit=+Lookup+FEN+
Parent - By Venator (Silver) Date 2007-12-15 10:49
Here is an interesting article about this ending:

http://www.xs4all.nl/~timkr/chess2/diary.htm
Parent - - By ebutaljib (****) Date 2007-12-15 13:01

> So my point is: An engine (let's say Rybka) might "see" that a position will end in a won six piece endgame, simplyfy into this position deliberately and throw away half a point.


No, I don't see your point.

First let's clarify what are we talking about here: OTB chess, correspondence chess or computer chess?

a) OTB chess
I agree, no human being can solve this, so the 50 move rule is just fine.

b) Correspondence chess
Correspondence chess is a competition between humans. Whether humans use engine assistance or not is irrelevant. It's always the human that decides about the move. If he trusts the engine and the engine is wrong, it's his own fault. So if the engine wants to simplify the position because it sees a forced mate in 208 moves which can't be achieved because of the 50 move limit, and human blindly trusts the engine and plays what it suggest than it's HIS OWN fault. It's no different than if the engine suggest some anti positional move that could be exploited many moves later in the endgame and the human plays it and loses the game because of this. It's his fault.

c) Computer chess
FIDE has no jurisdiction over computer chess, so the FIDE rules (like 50 move rule) don't apply.
Parent - - By Permanent Brain (*****) Date 2007-12-15 14:49 Edited 2007-12-15 14:55
No, the FIDE rules, articles 1 to 5 at least (= basic rules of play including the 50 moves draw rule) of course DO APPLY to computer chess. We don't want to play a different game, and that involves exactly the same rules.

(I also have official tournaments, like computer chess world championships, in mind.)

Nevertheless, for the problem in question, it is not required to omit the 50 moves rule. This is a rule which requires to claim the draw. So I can imagine that if opponents would 'informally' agree NOT to claim that rule, such a long tablebase mate could be winning and it would also be correctly FIDE compliant - because you can claim, but you don't have to. Of course, that would require that the opponents or all participants agree about this.

In automated engine matches and online engine games, things are a bit simplificated and such situations (also 3rd occurrance of same position) will always be draws unless the software has an option like "switch off 50 moves rule" or the like.
Parent - - By ebutaljib (****) Date 2007-12-15 16:22 Edited 2007-12-15 16:32

> No, the FIDE rules, articles 1 to 5 at least (= basic rules of play including the 50 moves draw rule) of course DO APPLY to computer chess.


They only apply if you agreed for them.

> (I also have official tournaments, like computer chess world championships, in mind.)
>


LOL.
What OFFICIAL computer tournaments??? Governed by whom??? Which organization governs the computer chess??? There is none (at least not worldwide) so there are no rules and laws of computer chess. The 50 move rule and all the others only apply if you chose them to apply. Of course if you want to still play chess, the basic rules about piece movement, mate, stalemate etc. must apply, but 50 move rule was only "invented" to prevent players to deliberatly drag out the games just to annoy their opponents. Computers can't get annoyed, thats why this rule is absolete in computer chess. Does the FIDE rule about the conduct of players also apply to computers? Must they also go to the press conference? How come in some tournaments the computer doesn't have to be in the playing hall? Thats cheating and computer shall be penalized according to FIDE handbook :)

You can't blindly copy the FIDE rules and apply them to computer chess!!!
Parent - - By Permanent Brain (*****) Date 2007-12-15 16:41 Edited 2007-12-15 16:49
I am sorry to say it, but you don't know what you are talking about. You just have no clue, and you seem to live 'in your own world'. If you play chess, you don't 'agree for' rules and you cannot 'choose' which rules you like and which rules you don't like. The FIDE rules are the definition of chess. If you remove rules from that or invent different rules, you are playing a chess variant, not classical chess. (Which would be ok; I am not against chess variants and I have played shuffle chess or Chess960, sometimes.)

Also, there is a worldwide computer chess organization SINCE DECADES; if you didn't know it I must assume you discovered computer chess two days ago. :-D The ICGA (former ICCA) is holding computer chess world championships since 1974:

http://www.grappa.univ-lille3.fr/icga/

Chess programs have always been created to play just that very same game, classical chess. Your considerations about a different set of rules are pure fantasy. The fact that Nalimov (DTM=distance to mate) tablebases ignore the 50 moves rule, is just a design problem. There is no philosophy of computer chess behind it or whatever. There are endgame databases which include the 50 moves rule, that concept is called DTZ (distance to zeroing move).
Parent - - By ebutaljib (****) Date 2007-12-15 17:14 Edited 2007-12-15 17:17
Seems like you're the one who doesn't know what you are talking about.

Who is the president of this ICGA? Who are it's members? Where can I read their internal rules (statute)? What happens if one of their members doesn't comply to their rules? Which programs are eligible to compete in the world championships? How come this world computer chess organization doesn't have the official ratings list? Are tournaments in Leiden, Paderborn, Mainz etc also governed by this organization? If no, how come they are not? What tournaments (besides world championship) are governed by this ICGA?

You call that organization and compare it to FIDE or any other worldwide sporting organization???

> The FIDE rules are the definition of chess.


What was the definition of chess before FIDE? Don't tell me that it didn't exist before :)

Since you don't have a clue what you are talking about, I will tell you what FIDE rules are. They are the set of rules that was accepted by ALL it's members, and every game that wants to have any significance for FIDE must be played under those rules. Nothing more.  If FIDE rules said that the players must wear helmets, then they would have to, or the game wouldn't be held under FIDE rules and would be void. If the game isn't played under those rules than it wont be registered by FIDE. Thats it. FIDE handbook excludes computers thats why any games played between computers or human vs. computers is not in compliance with FIDE rules by default! No such game ever had any significance to FIDE, because it's against its rules. And now you want to tell me that computers must (and do) play under FIDE rules. Thats simply impossible!

If someday a real worldwide computer chess organization will exist, then it will have it's own rules. Of course some rules will be the same as FIDE rules, but other will be completely different. They will be written for computers, whereas the FIDE rules are written for humans!
Parent - - By Permanent Brain (*****) Date 2007-12-15 17:41
PLONK!! :-D

P.S. What is the maximum number of posters I can put on my ignore list? It constantly grows.

P.P.S. Maybe we should consider to only accept members who are 12+ years old :-D These stupid childs are just too boring and time consuming.
Parent - - By Fulcrum2000 (****) Date 2007-12-15 18:43
Why ignore him?. He doesn't curse, posts normal replies and as far as I can see doesn't try to start a troll.

You disagreeing with his opinion is a rather thin reason to ignore/ban someone...
Parent - By ebutaljib (****) Date 2007-12-15 18:47
Thank you Falcrum. Finnally a voice of reason.
Parent - By Permanent Brain (*****) Date 2007-12-15 18:50
He doesn't even know ICGA and he lives in his own fantasy world. My time is too precious for such bullshit.
Parent - By ebutaljib (****) Date 2007-12-15 18:46
Seems like you are uncapable of having an adult conversation based on arguments instead of attacking the other person.

When you are playing a game with your friend, do you play under FIDE rules (ALL FIDE rules)? You most likely don't! Wait, you said it yourself that we can't pick which rule to chose and which not. So what is it? Are you still playing chess eventhough not all the FIDE rules are met? FIDE rules are not equal to chess. They are just rules from one organization, nothing more. You can't just simply take the rules from one organization and apply it to the other. I'm sure that even you would agree that computer chess and human chess is not the same. Otherwise why do we have this conversation?

What computer chess world would need is a real organization which would make it's own rules. If they would write down the 50 move rule then it would apply, otherwise it wouldn't. It's as simple as that. Until such an organization and rules exists all computer chess tournaments are unofficial. Remember how the participants of the last Random chess world championship were simply handpicked and Hiarcs fans complained about this? Thats because THERE ARE NO RULES, no form of criteria, nothing! Just because the organization has the word 'International' in it's name and just because it organizes the so-called world championships, it's still all on the informal basis. It's just a group of enthusiast, nothing more.

Now instead of attacking me, try to defend your position with arguments for a change!
Parent - - By BB (****) Date 2007-12-15 18:37
The FIDE rules are the definition of chess.

The FIDE Laws of Chess are perhaps the definition of FIDE chess. If I play a game in the park against the local expert, we still call it "chess" even though (say) we might waive touch-move (Article 4). If you set up the board with h1 as a black square, is this still "chess" or is it a variant (Article 2.1)? If we label the rows a-h and columns 1-8 (shogi style), is this still chess (Appendix E to the Laws)? There are perhaps "official" answers to these queries, but I don't find them that interesting. Also, the FIDE Laws of Chess are preambled with The FIDE Laws of Chess cover over-the-board play, and there is no indication which of the Basic Rules of Play can be extended to other situations.

I would say that "chess" is defined by: the initial setup, the moves of the pieces, and checkmate/stalemate. The rules of play that concern draws, resignation (what happens if a player's mobile phone rings?), adjudication, outside assistance (kibitzers in the park), etc., are subject to local variation.
Parent - By ebutaljib (****) Date 2007-12-15 18:50

> The FIDE rules are the definition of chess.
>
> The FIDE Laws of Chess are perhaps the definition of FIDE chess.


Good. So there are people here who understand what I'm talking about.
Parent - By BB (****) Date 2007-12-16 18:14

> I would say that "chess" is defined by: the initial setup, the moves of the pieces, and checkmate/stalemate.


The 4th Kamsky-Shirov game brings up a (somewhat pedantic) question - is chess at draw odds a "variant"?

FIDE says: The objective of each player is to place the opponent`s king `under attack` in such a way that the opponent has no legal move.
Given that the objective of Kamsky was noticeably different (and, of course, this situation occurs quite often, at varying levels of complication), should the game be (say) FIDE-rated? [Since I personally think that ratings are over-rated, I really couldn't care less].
Parent - - By Banned for Life (Gold) Date 2007-12-15 17:46
The 50 move rule, or something similar, is necessary to keep games with increments from lasting virtually forever. This is just as true for computer chess as it is for OTB or CC.

Alan
Parent - By ebutaljib (****) Date 2007-12-15 18:56
I'm not argueing if its necessary or not, I'm arguing that it only aplies if you chose it to apply.
Parent - By BB (****) Date 2007-12-15 19:19
The 50 move rule, or something similar, is necessary to keep games with increments from lasting virtually forever.

Repetition will eventually kick in. :) But with some of the positions from Father's games, I'd put the over/under for that at about a million moves.
Parent - - By Sesse (****) Date 2007-12-16 00:23
Yes, but there's a big difference between the current 50-move rule and  "something similar". I recall seeing a rule set once (obviously not identical to the current FIDE rule set) saying that if "experts" had decided that a given position needed more than 50 moves to mate, the player would be allowed twice the number of moves the experts (whoever that are) deemed required. It still fixes the problem of infinitely long games, but doesn't convert long forced mates to draws.

/* Steinar */
Parent - By Banned for Life (Gold) Date 2007-12-16 01:50
Agreed there is nothing sacred about 50 moves. This allows mating in the great majority, but not all, cases. As tablebases get larger, we will find longer forced mates without captures or pawn moves. Whether it makes sense to change the rule to accommodate these rare cases will be up to higher powers than you or I.

Regards,
Alan
Parent - By BB (****) Date 2007-12-16 17:33
I recall seeing a rule set once (obviously not identical to the current FIDE rule set) saying that if "experts" had decided

Philidor's cafe operated under a 21-move rule, where if he thought no progress had been made, it was a draw. If I am not mistaken, there were times in Go (not sure which scoring rules, this was maybe in the 50s) where the accepted way to score territory in some disputed cases was to ask a panel of experts for their opinion.
Parent - - By BB (****) Date 2007-12-15 19:14

> You can't blindly copy the FIDE rules and apply them to computer chess!!!


On the other hand, FIDE might try to assert that its rules apply to other venues. I am reminded of the (two) bridge scoring changes imposed by the WBF in 1987 for duplicate play, which they then "imposed" on rubber bridge in 1993. As the Wikipedia entry notes as an aside:

(however, since there are no official competitions, rubber bridge players accept them as they see fit)

It seems (for instance) that rubber bridge at Yahoo! Games adopts one of the scoring changes, but not the other. Additionally, it is likely that a bridge pro would play by whatever reasonable rules a client desired, as they are interested in making a living, as opposed to being WBF-compliant. Similarly, if consumers want/demand no 50-move draws in certain endings, the makers of chess engines will likely accommodate such an option. Whether or not this should be called a "variant" or (alternatively) should be the default behaviour is a different question.
Parent - - By ebutaljib (****) Date 2007-12-15 19:57 Edited 2007-12-15 20:01
Yes, this is what I'm talking about. Before the game begins they can chose which rules apply and which don't. Before computer chess tournament they can chose whether the 50 move rule will apply or not. It's an informal agreement. When human players play in FIDE tournaments they can't chose anything. Thats the difference - computer chess world doesn't have general rules (because there is no organization that would make sure they are followed).

Perhaps someday there will be a world computer chess federation that will govern computer chess play, but until then everything is going to be on informal basis. Everything started on informal basis. Chess world championships were held on informal basis until 1948 when FIDE took over. I know that the 1886 match between Steinitz and Zukertort is considered as first 'official' world championship match, but everything was informal until after World War 2. There were no rules or criteria about who could challenge the World champion, there was no timeframe set in which the Champion has to defend the title, no prize found etc. - everything was on informal basis, thats why Lasker held the title a little bit longer than he deserved, Alekhine could postpone a rematch with Capablanca indefinatelly, etc. Thats why the world organization with it's rules is needed.

Until there is a world computer chess organization everything that happens in computer chess is informal and there are no strict rules.
Parent - - By Linus (***) Date 2007-12-16 07:50

> Yes, this is what I'm talking about. Before the game begins they can chose which rules apply and which don't.


I have been playing correspondence chess at a number of different organizations/clubs/servers. All of them had/have their rules or guidelines and ALL of them stated explicitly that the FIDE rules of chess apply. I am not talking about a Sunday afternoon game with a friend, where we could of course agree to any set of rules we like.

Now, this tread has deviated from my original question about if and how an engine programmer could or should take into consideration the fact that a tablebase mate in n moves (with n > 50, no pawn moves and no captures) is indeed a draw. Tablebases alone do not provide enough information, because they do not tell us if and when there are pawn moves or captures in the mate sequence.
Parent - - By ebutaljib (****) Date 2007-12-16 09:51
You are mixing apples and oranges. I was talking about computer chess and you are talking about correspondence chess.

Correspondence chess is part of FIDE, so of course FIDE rules apply. Computer chess is NOT part of FIDE thats why there are no strict official rules. The rules are agreed on an informal basis, thats why you can CHOSE whether 50 move rule (or any ohter rule) applies or not.

THAT is what I'm saying. Nothing else.
Parent - By BB (****) Date 2007-12-16 17:55
Correspondence chess is part of FIDE

This is not correct. At best, ICCF correspondence chess is affiliated with (and officially sanctioned by) FIDE. Other correspondence groups are not sanctioned by FIDE (and thus they can have their own rules about computer assistance, etc., without this worry).
Parent - - By BB (****) Date 2007-12-16 17:50
I have been playing correspondence chess at a number of different organizations/clubs/servers.. All of them had/have their rules or guidelines and ALL of them stated explicitly that the FIDE rules of chess apply.

Indeed, once you have a competition, you need to clarify the rules. However, given that the FIDE rules explicitly state that their applicability is to over-the-board games (as the digression about touch-move would evince), one can still left with various lacunae here - for instance, what type of outside assistance is allowed, and why is this a local decision rather than a FIDE decision? My main contention is that the Rules of Chess could (and should) be more explicitly separated from (FIDE) Tournament Rules, (ICCF) Correspondence Rules, (ICGA) Computer Rules, etc.

Tablebases alone do not provide enough information

"Tablebases" is generic term, but each instance of a tablebase is in a specific format and answers specific questions. Thompson, Nalimov, Haworth, Heinz, Edwards, etc., each have various indexing schemes and (more importantly here) various metrics are used.  There are also some tablebases (for problemists largely) that take castling privileges into account. It is true that some tablebases do not explicitly answer some questions.

if and how an engine programmer could or should take into consideration

This seems to assume that the programmer is using TBs that are unsuited to the desired purpose. If the desired behaviour is to consider the 50-move rule, it would be most proper for a programmer to use TBs that take this into account. [Admittedly, this is the "pure" solution, and most programmers would rather try an imperfect plug-and-play solution].
Parent - - By Linus (***) Date 2007-12-16 18:09
It seems that my thread has raised a much bigger question than initially intended, that is whether or not the 50-move rule is part of the chess rules (like castling, e.p. capture, underpromotion or the three-times repetition rule) or an extension to the chess rules applicable only if agreed to. Personally, I believe that it is (and should be) a standard rule applicable to all kinds of chess, including OTB, CC or computer chess.
Parent - By ebutaljib (****) Date 2007-12-16 21:11
50 move rule is (and in my opinion it should stay that way) a part of the chess rules for OTB and Correspondence chess.

In computer chess there are no official rules so everything is optional - a matter of agreement. Of course if you still want to call it chess the rules about piece movements, castling, en passant, mate, stalemate must be folowed, but all FIDE rules simply can not be followed and applied to computers. The 50 move rule was invented from practical point of view, so that one opponent doesn't deliberatlly drag out the game just to annoy his opponent. They shortly reconsidered this rule when developing of tablebases begun and they found out that some endings need more than 50 moves without pawn movement or captures to checkmate. But then they realised that such endings can never be 'solved' by a human so they re-established the 50 move rule. The right decision, in my opinion.

In computer chess however this rule is unnecessary, because of two reasons:
- computers don't get annoyed (sooner or later the game will end as a 3 fold repetition)
- there is always a human operator who can adjudicate the game.
Parent - - By Linus (***) Date 2007-12-16 08:06 Edited 2007-12-16 08:12

> First let's clarify what are we talking about here: OTB chess, correspondence chess or computer chess?


I specifically mentioned in my post, that it was about computer assisted correspondence chess and not OTB chess.

> Correspondence chess is a competition between humans. Whether humans use engine assistance or not is irrelevant.


I disagree. Correspondence chess is very much a game of computer assisted humans today. This applies to servers/organisations where engine use is allowed as well as to those where it is not allowed. On all organisations where engine use is banned, you have players complaining about opponents using engines and the arbiters have their problems proving/disproving the accusation. I know this problem first hand, because I used to be Senior Arbiter of an email correspondence chess club.

Now, I do not want to start yet another tread about wheather it is good or bad that engine use is allowed. I just made my decision for myself. Since on those servers/organisations I played which officially ban engines I ended up losing against opponents who - in my opinion - did use engines anyway and I could not prove it, I decided only to play correspondence chess on servers which explicitly allow engine use. For me this is a clear and easy solution to a complex problem. I know other people will disagree. That is everyones choice.

> FIDE has no jurisdiction over computer chess, so the FIDE rules (like 50 move rule) don't apply.


I am not talking about computer chess in general. I am talking about correspondence chess servers/organisations who to my knowledge play according to FIDE rules. I have not seen an organization which does not.
Parent - - By ebutaljib (****) Date 2007-12-16 09:56 Edited 2007-12-16 10:05

>> Correspondence chess is a competition between humans. Whether humans use engine assistance or not is irrelevant.


> I disagree. Correspondence chess is very much a game of computer assisted humans today.


Are you saying it is not a competition between humans??? If that is what you are saying, then racing is not a competition between humans either.
Also if it's not you that is competing, but your computer program, then you don't feel anything when you win/draw/lose the game. Afterall what is it to you, you weren't competing, so you didn't achieve/lose anything.

You know this is not the case. YOU were competing. It was YOU who was trying to outsmart your opponent. Whether you used computer to help deciding about the moves or not is irrelevant.

>> FIDE has no jurisdiction over computer chess, so the FIDE rules (like 50 move rule) don't apply.


> I am not talking about computer chess in general. I am talking about correspondence chess servers/organisations who to my knowledge play according to FIDE rules. I have not seen an organization which does not.


Agreed. I never said correspondence chess isn't or shouldn't be played under FIDE rules. Read my posts more carefully.
Parent - - By Linus (***) Date 2007-12-16 10:33 Edited 2007-12-16 10:35

> Are you saying it is not a competition between humans???


No. What I am saying is that most correspondence chess games today are played by computer assisted humans, even at servers, where this is not officially allowed.

I expect that an engine knows the rules by which chess is played, as much as I expect a pocket calculator to give me correct results. It would do me no good, if I had an engine that plays incredibly strong, but does not know the e.p. capture rule. Such an engine would give me bad advice. The same goes for underpromotion, castling, three times repetition and the fifty moves rule.

> FIDE has no jurisdiction over computer chess, so the FIDE rules (like 50 move rule) don't apply.


The 50 rule move is part of the chess rules as much as e.p. capturing, castling, underpromotion or the three time repetition rule. It is not a choice of the operator of a chess site to allow/disallow the 50 move rule. The chess rules are generally accepted all over the world as is. To my knowledge there are no (serious) CC servers that deviate from these known rules.

> Read my posts more carefully.


Since you have not addressed the question in my original post (should/could engine programmers take the 50 move rule into consideration when it comes to tablebases information) it is you who has not read my post carefully. You just keep going on about technicalities in the generally accepted chess rules.
Parent - - By ebutaljib (****) Date 2007-12-16 11:44 Edited 2007-12-16 11:52
ebutaljib: Corrrespondence chess is a competition between humans.
Linus: I disagree.
ebutaljib: Are you saying it is not a competition between humans?
Linus: No. What I'm saying is that most correspondence chess games today are played by computer assisted humans, even at servers, where this is not officially allowed.

Make up your mind. Is it or is it not? I said 2 times that is totally irrelevant if the human uses a computer or not! It's still humans that are competing.

You are the one who is steering the conversation away and when I'm going to start to give my opinion whether engine assistance should or shouldn't be allowed, then you will say that I'm deviating from the post.

If we return to your original post:

> On the Chessbase news I found an article about a RB vs NN endgame played between Sergey Karjakin (2694) and Alexei Shirov (2739) at the World Cup Khanty-Mansiysk which ended in a draw but is - according to six men tablebases - won for black in 208 moves!
>
> http://www.chessbase.com/newsdetail.asp?newsid=4322
> (scroll down to the second game in the article)
>
> It is unlikely that humans can play this kind of endgame correctly, so I wouldn't worry too much about it in OTB chess. But it certainly does concern CC players who use engines and tablebases. FIDE rules are quite clear about the fifty moves rule. There have been extensions to 100 moves for certain endgames in the past. But as was stated in the article by a FIDE International Arbiter, all extensions to the 50-move rule were abolished at the Manila Congress in 1992.
>
> So my point is: An engine (let's say Rybka) might "see" that a position will end in a won six piece endgame, simplyfy into this position deliberately and throw away half a point. Could (or should) the engine programmer avoid this?


I asked you about which form of chess are we talking about (because it is not so obvius).
1. We agree on OTB chess, that this long mates are irrelevant, because no human can solve this.
2. If we are talking about correspondence chess (and it seems like we are) then I say this is a competition between humans. It is the human who decides about the moves, not the engine! If the engine is wrong and human makes that move, and the move loses the game or lets opponent to get away with a draw, than it is his own fault. Do you at least agree on this?

But your last paragraph then talks about Rybka (or any other engine) that draws when it could win. Again, if we are talking about correspondence chess, then the engine doesn't lose/gain anything. It was YOU who was wrong because you trusted Rybka. Get it? YOU should have analysed the position to see if you can win within the 50 move rule or not!!! It was your choice to make that move that draws, not Rybka's.

3. If your last paragraph was talking about engine vs. engine game (computer chess) then the 50 move is not obligatory, because there are no strict official rules about computer chess (no governing body, etc.). If you agree with other operator that it applies, then it does, but you could also agree that it doesn't. You have a choice. See my discussion with Permanet brain who tried to convince me that FIDE rules is equal to chess. IT IS NOT!!!

I just wanted to clear things up and to make you aware, that OTB chess, correspondence chess and computer chess are 3 different forms of chess. OTB and correspondence chess have worldwide organizations with strict official rules which must be met - and computer chess doesn't have this!

But I see people on this forum are nort aware of this. You keep mixing correspondence chess with computer chess, and keep insisting that FIDE rules are chess. Read my posts carefully again and you'll see what I am talking about.

Now to answer your original question if programmers should avoid that their engine trades down when there is a mate which can't be acomplished within the 50 move rule:

My answer is: I don't know and don't really care :)

Like I said I wanted to see if you can distinguish between different forms of chess and it seems like you're not.
Parent - - By Linus (***) Date 2007-12-16 12:44 Edited 2007-12-16 12:46

> I just wanted to clear things up and to make you aware, that OTB chess, correspondence chess and computer chess are 3 different forms of chess.
> Like I said I wanted to see if you can distinguish between different forms of chess and it seems like you're not.


I made it very clear from my first post that I was talking about computer assisted correspondence chess, not computer vs. computer chess. This sentence is from my very first post:

> ... I wouldn't worry too much about it in OTB chess. But it certainly does concern CC players who use engines and tablebases.


If you had read my original post, this would have been obvious to you.

>Now to answer your original question if programmers should avoid that their engine trades down when there is a mate which can't be acomplished within the 50 move rule:
>My answer is: I don't know and don't really care :-)


Then why the hell are you posting here? I started this thread raisng a simple question. If you don't have to say anything about the subject at hand, then don't respond. With this comment you just confirmed that you are nothing but a troll, trying to stir up trouble. And I was stupid enough to go along.
Parent - - By ebutaljib (****) Date 2007-12-16 13:00 Edited 2007-12-16 13:07

> Then why the hell are you posting here?


Because I'm entitled to give my opinion on the matter just like you or anyone else.

My first sentence in this thread was that I don't see your point. Everything else that I wrote was in response to Permanent brain (who tried to convince me that FIDE is chess) and you (who couldn't decide if correspondence chess is a competition between humans or not).
I explained why I don't see your point concearning the 50 move and the engine trading down in CORRESPONDENCE CHESS. Because in correspondence game the human has to decide if he trustes the engine evaluation or not. And since you deliberatlly chose to ignore that part of my previous post, I will take it as if you silently agree with me on this.

And by the way, CC can mean Correspondence Chess or Computerr Chess.
Parent - - By pedrox (**) Date 2007-12-16 14:40 Edited 2007-12-16 14:45
All organizers (all) of tournament for chess programs are based on the normal laws of chess (FIDE), a program of chess must know what is a castling, which is eating to the passing (ep), the rule of repeating a position 3 times or the rule of the 50 moves. Sometimes, these four things are difficult to handle well for a person who is doing for the first time a chess program and we can find with engines that do not. The engine will be able to play in tournaments without these four rules? Yes. Why? Because simply on its list of moves is not the castling or ep and other engines, the GUI or the tournament director who will decide based on the FIDE rules when a repetition of moves or rule 50 moves occurring and may establish an result for the game.  But these tournaments director ask developers to resolve these problems, perhaps they we will not continue playing tournaments.

All programmers have assumed that our engines must play under the same rules as FIDE, at least if they want to compete with other programs.

On the other hand I do not agree to extend at least for the moment the rule 50 moves in computer chess. To access the tablebases of Namilov requires a code source, there are programmers who are permitted access to the code and others do not, so I not think is just extend the rule of the 50 moves, while the code to access the tablebases of Namilov is not free.
Parent - - By ebutaljib (****) Date 2007-12-16 14:59
I'm tired of this. I don't understand you people. I talk about one thing and you keep on replying with something else. I think my English is good enough that you can understand what I'm talking about.
Parent - - By pedrox (**) Date 2007-12-16 15:19 Edited 2007-12-16 16:09
I am not the person who has been talking, my comment above is the only one who did.

I am sure that Mike is a big fan of the engines of chess and of the chess tournament between programs, as he says in his comments he had no knowledge of programming and I often differ from the comments he makes, but I agree with him in his comments with you.

You ignore that there are world championships of chess programs and they are governed by the same rules as FIDE. The World Championship is organized by the ICGA and respected by all programmers. Yes, I can make a different organization, as Kasparov did at the time with the ACP, but in this case the basic rules with which play tournaments would have the same rules as FIDE, I would not change as Kasparov not changed in its day.

You have all the information on the ICGA at:

http://www.icga.org/

If you do not recognize the ICGA you do not acknowledge that the current world champion (the 15th) is Rybka, you asked Vas whether he agrees with it, Shredder is the  World Computer Speed-Chess.
Parent - - By ebutaljib (****) Date 2007-12-16 20:51
I don't ignore anything, all I'm saying is that this is all unofficial. Thats perfectly OK. Rybka won the last computer chess world championship, so it is the World Champion. But it is all on informal basis. Steinitz, Lasker, Capablanca, Alekhine and Euwe were all (and still are) considered World champions eventhough everything was informal. There is absolutelly nothing wrong with that.

Since many of you don't understand what I'm trying to say, I will give you an example.

a) World championship match between Anand and Kramnik.
There are strict game rules, rules about playing condiditons, press conferences, conduct of players, etc. (even dresses of the players). Players must accept this. The only thing they can chose is the time controls: they can chose between classical or FIDE time controls (which are a little faster).

b) Match between Rybka and Zappa
They agreed on everything. There was nothing that restricted them. They decided if the computers will be physically in the playing hall or not, they could chose any time controls they liked, they chose the number of games, they agreed on what to do if the computer crashes, what kind of hardware will be used etc.

See the difference? In the first case the match is governed by FIDE and FIDE rules must be met, and in the second case EVERYTHING is the matter of agreement. They could agreed that the 50 move rule will apply strictly, they could chose that the game is simply adjucated if the position is won/drawn by tablebases, etc. Thats only possible because in computer chess there are NO OFFICIAL RULES and worldwide organizatuion that would govern them.

P.S.: If you want to argue that Rybka vs. Zappa was just a show (and thus inofficial), you can apply all this to the next computer world championship. It's all a matter of agreement.
Parent - - By pedrox (**) Date 2007-12-20 00:32
We agree on something, the match between Rybka and Zappa has been unofficial and to develop this match there has been a series of agreements between the two sides. Yes, Anthony and Vas could have come to an agreement to change the rule of the 50 moves.

But when you are an author of a program and presents your program to the World Championship of the ICGA, everything is official, you must accept the rules of the ICGA and if you do not agree to stay away. Indifferent you reach an agreement with the other programmers who are in the world championship, is the ICGA who organized this tournament and you have to play with their conditions.

On page of the ICGA you can find the following word:

1. Implement the rules of chess correctly. Incredibly some programs still cannot recognize draw by repetition or apply the 50-move rule.

If a developer does not like these conditions is free to make another official association and hold another World Championship alternative, as Kasparov did in its day. Then you can change the rules, such as when Kasparov did not want to face against his challenger officer who was Shirov by not considered commercial.

You have forgotten that we had simultaneously 2 world champions? Or your only acknowledge the world champion of FIDE?

You have all the information from the World Computer Chess Championship in wikipedia.

http://en.wikipedia.org/wiki/World_Computer_Chess_Championship

I do not understand because you do not believe that this is official.
Parent - - By Permanent Brain (*****) Date 2007-12-20 02:53
These are very theoretical considerations anyway. I am following computer chess news and events fairly closely since 20+ years, and NEVER has a participant of one of these big tournaments 'not accepted' a FIDE rule. This simply isn't a topic...
Parent - - By BB (****) Date 2007-12-20 03:23
This simply isn't a topic...

Even at the time of the Thompson CDs, the question of (say) adjudicating 5-piece endings rather than playing them out was topical (not sure if the ACM or ICCA tournaments ever had any formal discussion along these lines, but there is Roycroft's 1984 article A Proposed Revision of the `50-Move Rule'. ICCA Journal, Vol. 7, No. 3, pp. 164-170.). Using an Internet search engine showed up a 2004 discussion of this question.

NEVER has a participant ...'not accepted' a FIDE rule.

The question is not for one participant, but for a mutual agreement by all participants.
Parent - - By BB (****) Date 2007-12-20 23:53
there is Roycroft's 1984 article...showed up a 2004 discussion  of this question.

Actually, that was a 2004 archive of a discussion from 2000. In the ICCA Journal, there is also H. J. van der Herik and I. S. Herschberg, The 50-move Rule Revisited,
ICCA Journal 12, 3 (1989), 192.

Concerning my opinion, I can't believe that engine tournaments still have "operators" - at a time when we have a multitude of usable interfaces, this is so anachronistic as to be laughable. It seems to be a leftover from the early ACM days to desire that engine tournaments emulate human conditions.
Parent - - By pedrox (**) Date 2007-12-21 01:13
Concerning my opinion, I can't believe that engine tournaments still have "operators" - at a time when we have a multitude of usable interfaces, this is so anachronistic as to be laughable. It seems to be a leftover from the early ACM days to desire that engine tournaments emulate human conditions.

At the World Championship can go any person who has made his own chess program, it does not have to be compatible with the protocols that you know, UCI or Winboard, nor does it bear Auto232, so you need an operator. Besides is not the same as playing every program in a different computer or two programs that do the same computer. In the world championship anybody can choose the hardware with which to play.

And Hydra?, You will tell me as Hydra playing with a chess program for the PC.

These meetings also serve to allow programmers fun and change ideas.
Parent - By BB (****) Date 2007-12-21 04:33
so you need an operator

Or, compliance with some protocol could be demanded (and public interfacing code for this made available). As a side benefit, there should then be a lack of operator errors.

These meetings also serve to allow programmers fun and change ideas.

I agree. There is no reason to soil such fun with the yoke of operating, a task most easily performed by the machines themselves.

Besides is not the same as playing every program in a different computer or two programs that do the same computer.

This is easily solved by a local network. Each engine runs on whatever computer (possibly remote) the team has. All engines link up to a main tournament server in the playing hall. The programmers sit and watch the moves magically appear on the screens rather than use a clumsy board. VoilĂ .
Parent - - By BB (****) Date 2007-12-20 03:58
On page of the ICGA you can find the following word:
1. Implement the rules of chess correctly. Incredibly some programs still cannot recognize draw by repetition or apply the 50-move rule.


When Marsland originally wrote this after the 14th WMCC in Jakarta (1996), it was thought so important that he labelled it #0 [however, it seems largely to be aimed at ameliorating the settling of disputes (particularly operator errors) rather than (say) making such a recognisance capability a formal requirement for a programme to enter the event].
Parent - By BB (****) Date 2007-12-20 23:49
When Marsland originally wrote this after the 14th WMCC in Jakarta (1996)

It was pointed out to me that I should recall that there was a non-trivial (though minor) boycott of this event after the Israelis of Junior were denied entry to Indonesia.
Parent - By BB (****) Date 2007-12-20 04:25
you must accept the rules of the ICGA

Upon reading the latest WCCC rules more closely, it is depressingly vague about this:

> 5. Unless otherwise specified, rules of play are identical to those of human tournament play.


Maybe they just don't want to mention "FIDE" as such, but "human tournament play" seems a bit too inexact for my taste. For instance, the rules for the 6th Open Polish Computer Chess Championship largely copy the WCCC rules, but also add here "accordingly to the updated Chess Code of the Polish Chess Association" so as to clarify. [In fact, I think the WCCC rules are simply copying the old ACM phrasing, see rule #3 on the last page of this pdf].
Up Topic Rybka Support & Discussion / Rybka Discussion / Chess Engines and the 50 moves rule
1 2 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill