How many points should you get for a draw?

8/20/2015 – Arno Nickel is chess fan and a GM of correspondence chess. But he is worried about the future of correspondence chess. In June he published an Open Letter, in which he proposed to change the scoring system to address the problem of the high number of draws in correspondence chess. With a survey he invited chessplayers to a discussion. Now he presents the answers.

ChessBase 14 Download ChessBase 14 Download

Everyone uses ChessBase, from the World Champion to the amateur next door. Start your personal success story with ChessBase 14 and enjoy your chess even more!


Along with the ChessBase 14 program you can access the Live Database of 8 million games, and receive three months of free ChesssBase Account Premium membership and all of our online apps! Have a look today!

More...

About the Problem of the Rising Rate of Draws in Correspondence Chess

Poll Evaluation (Survey from 15th June 2015)

By ICCF GM Arno Nickel

Arno Nickel

In an Open Letter published in June 2015 I put forward the idea to have an enhanced scoring system in which players get ¾ point for specific cases of a draw (for instance, stalemating the opponent or being a piece up against the bare king). This proposal was explained in detail in my open letter and is directed against the rising rates of drawn games. To increase acceptance for my proposal I also suggested a high-class invitational tournament to test the new drawing rules. Added to the proposal was a survey that asked ICCF members for their opinion. The deadline for this poll was the end of July.

ChessBase published the Open Letter on its English and German news sites where it triggered a lively discussion. The English version appeared on 20th June under the title "Correspondence Chess - the draw problem", the German version was published on 2nd July under the title “Kein halber Punkt mehr fürs Remis?”. In addition, members of the ICCF published the Open Letter on the websites of their national federations or spread it via other internet channels.

This led to a pretty good feedback which could have been even more impressive if all ICCF members had replied to the survey coming with the Open Letter. However, experience shows that no matter what they think about certain issues a lot of people refrain from getting active for a variety of reasons. Thus, the number of people who are interested in the subject of the “draw problem” and the future of correspondence chess is, of course, much higher than the number of people responding to the survey. And if one looks at a longer period of time one might even say that this topic never ceases to interest, especially if is connected to „chess engines in correspondence chess“. Whenever the subject of correspondence chess comes up this topic is discussed.

After these preliminary notes we could almost start to evaluate our survey. But first allow me one word about the ICCF. To initiate such a public debate does not mean to belittle the work the ICCF officials do or have done. On the contrary, I am convinced that the ICCF has done excellent work for many years, not least because they managed to develop an excellent chess server and modernized its rules and structures according to the requirements of the computer age. However, we are still waiting for long-term plans how to deal with the challenge posed by the increasing strength and domination of chess programs, and correspondence chess databases that are getting bigger and bigger. The answers given by ICCF appear to be evasive and defensive; to my mind they also fail to include the ICCF members in a discussion about the question how correspondence chess sees and presents itself. To some degree this is understandable as such discussions – as political parties know very well – tend to be awkward and risky. But they also offer the chance to define and establish a new position, not to mention the fact that such discussions can lead to a stimulating spirit of optimism. And this is the very intention of my survey.

Live game on the ICCF server

The answers to the survey show above all one thing: in view of the rising number of draws and the dominance of computer engines a huge number of ICCF members and a lot of chess fans feel the need for concrete action to keep correspondence chess attractive or, if there is no other choice, to reinvent its attraction. The many proposals and comments indicate the eagerness to experiment, and show the substantial wealth of ideas the ICCF members have and which certainly will not fall on deaf ears when delegates and commissions hear about them. I regret that I did not initiate this discussion many years before, however, as the saying goes: better late than never!

The complete documentation of the answers to the survey is summarized in an Excel table which you can download here in German and in English. All comments were left in English or in German but maybe will be translated some time.

However, not only ICCF members joined the discussion but other chess players also came up with interesting comments and proposals. These were added to the survey and defined as an extra group.

This is the distribution of entries:


Total number: 172
ICCF members: 123
Not known, whether ICCF members: 32
No ICCF members: 17

To highlight the numerous written comments, color marks were chosen:
Some relatively neutral, personal or general remarks were left unmarked.

The evaluation resulted in the following picture that reflects the opinions about the five main questions:

 

            Question 1

Question 2

Question 3

Question 4

Question 5

yes = 78,8 %

yes = 8,1 %

yes = 56,1 %

yes = 13,8 %

yes = 19,5 %

no = 11,4 %

no = 68,3 %

no = 22,8 %

no = 52,8 %

no = 40,6 %

abst.= 9,8 %

abst.= 23,6 %

abst.= 21,1 %

abst.= 33,3 %

abst.= 39,8 %

 

 

 

 

 

yes = 74,4 %

yes = 8,7 %

yes = 50,0 %

yes = 16,3 %

yes = 18,6 %

no = 11,0 %

no = 66,9 %

no = 23,8 %

no = 42,4 %

no = 32,0 %

abst.= 14,5 %

abst.= 24,4 %

abst.= 26,2 %

abst.= 41,3 %

abst.= 49,4 %

About one third of the ICCF members agreed to have their names published as supporters of the pilot project. Perhaps a list with the supporters of the pilot project will appear, but currently the names in the Excel table are encoded to protect the interests of all and to focus on the answers. (Of course it will be possible to identify some names because the statements quoted were made in a public forum but all in all this should be no problem.)

Correspondence card from 1984 - sent from the US to Germany

Before we look at each of the questions separately, one note to the involved ICCF members. I was positively surprised that players of all levels took part in the survey. One might have expected that mainly titleholders and players with an ICCF Elo of 2400+ would be concerned about the problem mentioned in the survey, but apparently this isn’t the case. Whether this had an influence on the statistics is hard to say. I have the impression that lower-rated players have a greater tendency to demand the banning of engines or to recommend engine-free play, but this does not stop them from recommending the pilot project. This approach is typical and can be seen in a large number of answers. Moreover, a number of officials, though very cautious in their judgments about individual issues, would like to see the results of a model tournament with different drawing rules. The main point is to gain insight and knowledge and to make progress in the discussion about the difficult question how correspondence chess sees and defines itself.

Question1:
Do you think that the rising percentage of drawn games is a major problem for correspondence chess?

There is a clear "Yes" to this question (78.8% →ICCF or 74.4% →All). This high value is no surprise but it is interesting to see how it correlates with the following answers. Moreover, there is a large variety of answers why the drawing rates increase. People do not just give the increasing dominance of the computer as a reason. The various explanations can be seen in the comments to questions 3-5.

Question 2:
Do you consider the rising percentage of drawn games to be an unavoidable side effect of correspondence chess which we should better accept instead of doing something against it?

Some people responding saw two questions in the question above and answered (1) yes, those draws are inevitable, but (2) no, we should not stop trying to change it; or they said (1) no, those draws can be avoided, and (2) yes, we should try do something against it. Both answers, however, agree to the main statement: “We should try do something against it”. But still, I have to admit that the question should and could have been clearer. There is also a strong approval of concrete steps (better sooner than later): 68.3% →(ICCF) or 66.9% →(All) advocate concrete steps. However, there is also uncertainty and helplessness as the significant number of abstentions with (23.6% →(ICCF) and 24.4% →(All)) reveals.

TThe survey offered no additional space for comments to the first three questions. Yet, a lot of people shared their opinions and experiences in regard to these questions in emails or forums. Thus, in the evaluation of the survey every question received an optional column with comments. In another column – at the very right – one could comment the survey as a whole. (In general there was a relatively high number of abstentions. Often people only commented some questions but not all. Such cases were seen as abstentions to these questions. In a few cases the answers were so unclear or contradictory that the entry was regarded as abstention. To mark these cases they received the symbol *.)

In the relatively small group of those who think the rising drawing rates (8.1% →ICCF or 8.7% →All) have to be accepted some still recommend the pilot project with an enhanced score system. This may appear contradictory, but they may have good reasons to make this distinction.

Times change - a biography of the first Correspondence Chess
World Champion (1953-1958), the Australian C.J.S. Purdy.

Question 3:
Do ou support the pilot project with an enhanced score system?

56.1% →(ICCF) or 50.0% →(All) approval with many abstentions (21.1% →ICCF or 26.2% →All) can be seen as a clear majority vote for the attempt. However, only for the attempt, not necessarily for a basic reform of the scoring system. The arguments for and against such a system are quite interesting. One could summarize them under the heading „hopes and fears“. Of course, most of the people who are against a different scoring system believe the problem can be solved by logical reasoning and do not consider practical tests to be necessary. They do not see their arguments as "fears" but as logical dismissals of the proposal based on their understanding of chess. Their standard argument is: Mate is the only criterion for gaining points; but this criterion would be diluted in favor of other objectives, if we introduce a ¾-point score; in short: “This is different chess“ or even “a different game“.

Some of those opposing the proposal would probably have no objection if the ¾-point score would be introduced as an additional offer – similar to Chess960. On the one hand this seems generous but on the other hand it means: „Do what you want, but stop bothering us!“ They seem to assume that the enhanced score system is nothing but a small playing field for a small number of people. Others do not even want to see such a "playing field" because they fear it might infect “true” chess sooner or later.

Another argument often put forward is: The ICCF should not do things on its own but stick to the FIDE rules! This kind of argument, which does not convince me at all, avoids any debate. I do not want to discuss it in depth here because it would take us too far from our main subject. But I think it is almost absurd to demand that the rules of correspondence chess follow FIDE rules blindly because the basic approach to chess is often different, for instance, in regard to help from others. In classical chess using computer programs during a game is “cheating”, in correspondence chess one is allowed to make use of whatever might help to find the best moves. My position is: Follow the rules of FIDE as far as they make sense for correspondence chess. (For the same reason I am also against doping tests for correspondence chess players. Just kidding.)

Sometimes supporters of the pilot project propose modifications:

  • Extending the trial to all classes of players (instead of having an invitational tournament);
  • Restriction of the enhanced score system to tie-break situations, similar to the Baumbach rule (which decides in favor of the player with the higher number of wins before resorting to different tiebreak-methods);
  • Giving 0.6 – 0.7 – 0.8 points for half a win – not ¾ point;
  •  Extending the ¾-point score to all draw agreements when one side remains with a material advantage of at least one pawn unit.

One proposal deserves special attention though it is currently not yet topical, namely, the impact of such draws on the rating system (adjusting of rules etc.).

Cover of one of the rare books about current correspondence chess

Question 4:
Should the ICCF consider other steps instead of the pilot project?

The answers to this question (similar to question three) show a high approval of the pilot project. 52.8% →(ICCF) or 42.4% →(All) do not want to look for solutions different to the one proposed in the pilot project, and with 33.3% →(ICCF) and 41.3% →(All) the number of abstentions is higher than it is in the first three questions. A number of answers did not distinguish carefully enough between question 4 and question 5 (which asked for measures that go beyond the pilot project) and the allocation of answers had to be corrected. Some participants also made proposals to both questions and left it to the reader to prioritize. Important alternatives (even if not yet numerically significant) to the pilot project are:

  • A 3:1 score as in football (also possible as 1.5 : 0.5 score); this incentive to take more risks  is one of the most popular proposals; especially as it seems to have proven itself in football. However, similar to other arguments that lead us away from our main subject – the ¾-point score – here is not the place to look at this in great detail. It might make sense to organize test tournaments similar to our pilot project to see whether the score system could stimulate players to take more risks and whether this shows in their choice of opening, and last, but not least in the results. However, I doubt this because chess engines will still reduce the willingness to risk too much – on both sides of the board. But the basic question whether it makes sense to value one win and one loss higher than two draws is a different question. To some degree the Baumbach rule already adopts this approach though it only applies to tiebreaks. At top level the Baumbach rule is not often applied because often players who did not lose a single game finish with the same score and Sonneborn-Berger is used for tie-break. With an increased ratio of draws we will soon have more and more cases in which none of the existing tie-break systems will be able to decide which of the players with a score of +1 or +2 finishes better.
  • A radical measure is put forward again and again is the ban of engines. This is different from "engine-free play" which is supported, for example, by the BdF and by many free chess servers as a kind of competition based on mutual agreement. A strict "ban" requires more control and sanctions to guarantee fair competition and to avoid grey areas. Some players believe they can copy methods from real time online-play as control mechanism. If, for example, the moves of a certain player correlate to more than 70% to the first choices of the engines this player might be considered as convicted of cheating. However, dreaming of such control mechanisms is no more than wishful thinking as it ignores the differences between real time online play (mostly blitz- and rapid chess) and correspondence chess. Even using a database in correspondence chess would run counter to such control mechanism and thus the questions arises where such controls should start. Moreover, in real time online play the time players need for their moves is an important indicator of possible cheaters (players who use the same amount of time for easy and for difficult moves) but in correspondence chess this plays no role. Moreover, it is unclear how to evaluate several moves of roughly equal value that correspond to engine proposals. This problem could easily arise in positions with a lot of positional maneuvering. The correspondence to the engine proposal could be accidental – while one would also have to find a way to account for corresponding moves that are forced, e.g. if they are the only moves that stops one side from losing a piece (or more). There are many arguments why an enforced engine ban is bound to fail no matter what we think about using engines in correspondence chess.

Engine-free play can only be an additional offer and requires that all players just look for fun and honor – prizes would invariably lead to cheating. But experience has shown that some players even cheat in fun tournaments in which engines should not be used. Black sheep can be found everywhere.

A lot of people see the main problem of an increased number of draws in short draws and in a lack of competitive spirit. The suggested countermeasures range from a general ban of draw offers to a ban of draws before move 30. We could of course try out such traditional methods, but I fear they will not have the same success as in over-the-board chess. I recommend to look for new ideas and to stimulate the competitive spirit.

That is what the ¾-point score is all about. The additional option of a half victory by stalemate or material advantage against the bare king could offer a strong incentive to play on – even so-called drawish positions. Another interesting proposal is to reward draws with black before move 40 with a small bonus. This might stimulate and inspire White players to follow more combative lines. And maybe some of these games would lead to a win instead of a draw or a ¾-point win.

  • Shorter time limits are also suggested sometimes but this proposal came up only a few times in this survey. I already explained that shorter time limits contradict the spirit of correspondence chess (high level chess!) and neglect the working and living conditions of many players. There is nothing new to add.
  • A relatively modern approach was some years ago suggested by Vladimir Kramnik: to draw lots which opening moves a player has to make. This method, which is limited to the first moves, has its followers, although at the moment there are not many. But in view of the databases we have to consider whether this could work in correspondence chess. I would think that this approach is more suited to OTB-chess, in which you play more games in a shorter period of time and thus have better chances to see your favorite opening on board. This would not be the case in correspondence chess. Apart from that, I very much appreciate the freedom to choose the openings I want to play.
  • Much to my surprise virtually no one playing correspondence chess recommended to play Chess960. It seems as if only a few correspondence chess players prefer Chess960 to starting from the classical position. Even those who play it (like me) see Chess960 more as a supplement than as a substitute to classical chess.
  • Another rather new proposal, as far as I can see, is to ignore draws in calculating the ICCF Elo. This way players with a relatively high rating might be more motivated to play in tournaments with lower rated players because they would not have to fear losing rating points when drawing with Black against much lower rated players, who just defend with White and use databases and chess engines efficiently. And players would no longer have the incentive of gaining rating points with bloodless draws with White against higher rated players. – About two years ago I mentioned this proposal in a much smaller circle and there it met with approval. However, though I assume that the ICCF rating commission would not agree to such a radical step (which would endanger the unity with FIDE and while dramatically reducing the number of evaluable games) I believe the commission is working to adapt the Elo-formula to the changed situation in correspondence chess. This means that it wants to adjust the rating expectations with a view to high number of draws. Hopefully, such an adjustment will come soon and show the same positive effects we hope to achieve with the more radical suggestion.

A draw - but worth more than half a point?

Question 5:
Should the ICCF consider additional steps apart from the pilot project?

As already said, the differences between the answers to question 4 and 5 were not always clear. The high number of abstentions, 39.8% →(ICCF) and 49.4% → (All), points to the fact that many participants did not know what to make of this question. Some comments and suggestions made in reply to questions 3 and 4 are also valid as answers to question 5, and there is no need to repeat them.

  Finally, it should be pointed out again that in the context of this survey a broad approval has been achieved for the suggested pilot project. In an uncertain situation which is, however, not completely desperate, and in which people have ideas, many correspondence chess players do expect concrete initiatives and a continued exchange of ideas to the benefit of the future of correspondence chess.


Discussion and Feedback Join the public discussion or submit your feedback to the editors


Discuss

Rules for reader comments

 
 

Not registered yet? Register

genem genem 8/21/2015 11:08
Arno Nickel showed cover of a rare chess book by Fritz Baumbach. On Amazon.com today I see another chess by him, also apparently very rare, and supposedly extremely valuable :-)

Fernschach: Tips Und Tricks Vom Weltmeister1991
by Fritz Baumbach
$2,089.31 used
genem genem 8/21/2015 11:03
Arno Nickel mentioned V.Kramnik's idea of draw lots to assign the opening moves of a correspondence chess game. Decades ago the world of championship Checkers found it necessary to switch to such a system due to their high draw rate problem.
Nickel said he would probably dislike Kramnik's suggestion because it would take away Nickel's - "freedom to choose the openings that I want to play."

Nickel also mentioned his surprised that almost no corr. chess respondent wrote in favor of trying chess960 as a possible solution. Nickel says he does play chess960 but he sees it as merely a "supplement" and not a "substitute".

Nickel's overall assessment of his respondents textual comments is an interest in trying different things on a one-time or trial basis, apparently just to learn more that might help devise another solution.
One way might be to COMBINE the 'draw lots' and 'chess960' ideas into one, by having all the games of one corr. chess tournament begin with the following legal chess moves:

1. e3 e6 2. Qf3 Qf6 3. Be2 Be7 4. Bd1 Bd8 5. Ne2 Ne7 6. Ng3 Ng6 7. Qe2 Qe7 8. Qf1 Qf8 9. Qg1 Qg8 10. Nf1 Nf8
(Now discretionary play begins.)

Corr. chess960 would be an exceptionally fast way for humanity to learn more about the abstract phase of chess we call the opening phase, without the constant confound or contamination of all the esoteric tactics inherent in the only start position humanity has ever used.

- - - - - - -

For a draw, better than .75-.25, or .6-.4, might be:
.5-.4 where White earns only .4 points.

But I remember the Slugfest tournament near Seattle several years ago where the points system was even more skewed to be anti-draw. The end result was - No different effect of any kind on (A) what the players tried to accomplish on the chess board, no on (B) the final standings.

When you lift up the covers, the underlying purpose of all these proposals of point imbalances is to compel players to make unsound moves in hopes of draw-busting complications. Not a good idea; unless...

Best idea is to make a Draw Offer last longer.
Make the draw offer expire only after the receiver of the offer has complete 3 subsequent moves, instead of just 1 subsequent move. Then the offer receiver can risk an unsound move and hope the complications lead to a weak reply by the opponent.
Wijnand Wijnand 8/21/2015 08:22
You can vary the amount of points you get for a draw if you have more /less material.
Seems a good idea.

But when every draw is the same value, like in the present system, the value has to be 0.5 points.

In a tournament where wins were 3 points and a draw only 1, You can, in a drawn position, agree with your opponent to flip a coin to decide the outcome of the game (who has to give up).
Flipping a coin gives you a theoretical value of 1.5 points, a draw only 1.

JBoyd333 JBoyd333 8/21/2015 08:12
I have been interested in this problem for many years, and have done a lot of analysis of computer engines and human games in relation to draws. After analyzing millions of games, and noting that the top engines and the top human players rarely win with black, I am convinced that white has around a 10% advantage, amounting to about 60-70 ELO points. Taking this into consideration, the following point chart resolves the inequality by giving black more points for a win or a draw. (It does not take into consideration HOW a draw was reached, which I think deserves more consideration.)

The percentages for White-Black-Draw are around 36-26-38 for humans, and 35-29-26 for computers (note LESS draws).
Giving 1 pt to the winner regardless of color, and .5 pts for a draw, regardless of color, ignores the built-in advantage for White. (Although, this is equalized if the same number of white/black games are played.)

As for the value of draws, and specifically if a win and a loss should score more, or less, than two draws, the new scoring system would give the edge to a WL (1.0) or a LW (1.3) over two draws (.7). It would take 3-4 draws to beat a win by white, and 4-5 draws to beat a win by black. This gives MUCH more incentive to play for a win!

I present two charts, the first adjusted to be closer to the traditional score, and the second more statistically accurate (=fair), but less traditional looking:

1. Win Draw Loss
White 1.0 .3 0
Black 1.3 .4 0

2. Win Draw Loss
White 100 35 0
Black 135 41 0

I have compared this new scoring system to the current one in several international tournaments, and found that in a few cases the standings are affected. In a few cases greatly so.

I am calling this scoring system the "Performance Based Scoring" system, or PBS for short.








Side Note:

Another interesting option is:

Black W 5 Black W 3
White W 4 or White W 2
Black D 2 Black D 1
White D 1 White D 0

but these, although easier to use, are less statistically accurate. In the 5 point system black has about a 5% advantage, in the second black has a huge 40% advantage!

John
Keesje Keesje 8/21/2015 04:32
Engine-free correspondence chess is practically impossible. You can have a gentlemen's agreement not to use engines for analyzing the game, but what about game data bases and books? Most of them also contain moves that were at some point in time evaluated by one or more engines. So to be consistent, you shouldn't use data bases nor books. You might argue to use books that were written before the computer era. However, these books also contain moves that were created by somebody else, not by the player himself. So, what's the difference between a move that was invented by a different person or by a different machine? Essentially, there is no difference, in my opinion.

However, there is also no problem for correspondence chess. It's true that a lot of pre-computer players have stopped, but a whole new generation of new people has started with correspondence chess. For them, a chess engine is just a tool, like a book or a data base. The average quality of the games has improved enormously, which is a good thing. Of course, for the same reason, there are a lot of draws now, but I don't see what the problem is. You can play a nice and interesting game that ends in a draw. So what? As long as you're having fun and like what you're doing, everybody is happy.
Radagast23 Radagast23 8/21/2015 03:38
I think engines will kill correspondence chess. Right now they have a heavy impact on correspondence games. I use correspondence chess to improve and deepen my opening repertoire. The only solution is an "engine-free" chess game, reserved to gentlemen players.
Offramp Offramp 8/21/2015 10:58
As Herman Goering said, "When I hear of alternative scoring systems, I reach for my gun." The best alternative to the traditional method is 3 for a win, 1 for a draw - [i]tout simple![/i]
Aighearach Aighearach 8/21/2015 09:31
An extra major tournament, of course most people support that. What is shocking is that it was less than 90%. Heck, I don't even like the proposal but I strongly support holding an extra tournament if you have funding.
luishon luishon 8/21/2015 06:51
How many points should you get for a draw?
for a draw a player should get less than half
of a point
same as his opponent
that will out balance the make a draw and catch up with a winner
a game of chess is for blood not for peace
Rama Rama 8/21/2015 02:14
I had long ago proposed that the simple way to get less draws (when color allocation is equal) was not to drastically change the scoring system or the rules of the game but to simply (slightly) take the color into account.
If a draw with Black was scored as .6 and a draw with White as .4 then it would cause the least disruption since a game is still a battle for a single point and a player who wins or draws all of his games would end up with the same score as they do now. However it would be impossible for two players entering the last round with the same score to finish the tournament with the same score. Further, an imbalance would be created from the moment the first round ends; ergo White will be motivated to fight for the full point.

Rama Gitananda
tigerprowl tigerprowl 8/20/2015 10:57
I don't think there should be a distinction between correspondence chess and over the board chess. I like the idea of revamping the point system for draws overall. However, I don't like the idea of using fractions. This will greatly confuse people. Why not make it 3 points for a win, 1 point for a "normal" draw, and then award 1.5 instead of 3/4 when necessary? This does two things, it allows for different draws to be awarded more appropriately and it also encourages people to play for a win to get the extra 2 points. I think the extra 3/4 (or my suggestion .5) point should be given on basis of material after the 30 move rule has been applied (after 30 moves).

Stalemating is a result of play from the start, and it is hard to determine if it was intentional or not. A draw on the other hand is usually agreed to. This is an important distinction I feel Arno Nickel rejected or overlooked. You shouldn't be allowed to force a 3/4 or .5 point on your opponent. Same goes with perpetual check.
1