Maurice Ashley: 12 Games Are Enough!

by Maurice Ashley
12/12/2016 – How to play World Championship matches? Yasser Seirawan did not like the 12 game format in Carlsen vs Karjakin and proposed a "Radical Solution". Now Maurice Ashley disagrees. After asking the top players for their opinion and looking back at previous matches he concludes that 12 games and a rapid tiebreak are enough!

ChessBase 18 - Mega package ChessBase 18 - Mega package

Winning starts with what you know
The new version 18 offers completely new possibilities for chess training and analysis: playing style analysis, search for strategic themes, access to 6 billion Lichess games, player preparation by matching Lichess games, download Chess.com games with built-in API, built-in cloud engine and much more.

More...

A Radical Solution: Unwarranted

Recently, my dear friend and esteemed colleague Yasser Seirawan wrote an article for this website decrying the format of the 2016 World Chess Championship. While many readers agreed with him and eagerly offered interesting proposals for change, I have to say that I think that his “radical” movement is way off track. Although I have sadly, and often disgustedly, watched FIDE shoot itself in the foot many times over the years, this is one area I feel that they and their organizational arm Agon have been absolutely trending on the right side of history.

There were two points that were essentially made in Yasser’s article. The more overt one is that the Classical Chess World Championship should be decided by classical means. No Rapid games (or, heaven forbid, Blitz) should mar the process of deciding the most important title in our game. After all, he argues, there are Rapid and Blitz tournaments expressly designed for the purpose of deciding who the best players are at those particular genres. Let the Classical Champion be decided by classical means, which usually coincides with giving draw odds to the champion. Yasser actually proposes that the match should go 13 games with draw odds given to the player who receives the extra Black, his point being that a winner should be decided at all cost but only by classical means.

Granted, in these matters, it’s impossible to prove one side wrong. Yes, classical chess does have a completely different time control from Rapid and Blitz. However, it should be noted that it was only in recent years that the title of Classical World Champion came into existence, but not initially for the reason one might suppose. The term is most often attributed to Vladimir Kramnik who, after winning the Braingames World Champion over Kasparov in 2000, ostensibly sought a way to differentiate his victory from the various title holders who had popped up as FIDE arranged their own knock-out world championships (a brave experiment that unfortunately was not implemented in a way that chess fans were prepared for). Kramnik was not looking to make a distinction between Classical and Rapid/Blitz, but to affirm that he had won the title by succession, i.e., that he had defeated the generally accepted legitimate champion (Kasparov) in a one-on-one battle. In his eyes, and many others, this is the one and only true (classical) way to consider oneself the new king of the block.

What the top players think

The objection of deciding the match by Rapid play has not, from what I know, received any vociferous objections from the many recent champions and their challengers who have played under the new rules (Kramnik, Topalov, Anand, Gelfand, Carlsen and Karjakin). When I asked Anand what he thought of Rapid tiebreaks, he said, “No problem. It’s the way of the world now.” A source close to Carlsen said that the World Champion hasn’t seen any reason to complain. Since I was in London getting ready to do commentary for the London Chess Classic, I decided to hold off submitting this article to personally ask the top players here if they objected to the current format. A full nine of the ten I questioned (Anand, Kramnik, Topalov, Adams, Aronian, Caruana, Giri, Nakamura, Vachier-Lagrave) felt that the present system of 12 games with Rapid tie breaks was absolutely fine, some even thought it was very near perfect. Only Wesley So hedged a bit, saying he saw the point of a purely classical solution of some sort. If the current titans of the game embrace 12 games plus Rapids as a good solution for deciding who amongst them is best, then one has to wonder why there is a need to gripe about it in the first place.

That said, at least one of the all-time greats has profoundly objected in the strongest terms. In an interview with SportBox, former champion Anatoly Karpov had these harsh words to say after viewing the Anand-Gelfand match:

"If evaluating the match from the qualitative and entertaining modes, I think that this was probably the worst World Championship Match played at least in the postwar period. One of the main reasons is the format of the encounter. 12 games is not that mockery on chess we observed during the knock-out system - but it is still not enough. At least 14-18 games are needed for full-fledged, creative fight: then the rivals have an ability to risk; whilst in a short match of the rivals whose strength is equal, the game is usually just hold, while the opponents are just trying to catch "a fail-safe chance." That's what we saw in Moscow and surely that made the match plain and boring... I'm firmly against of mixing different forms of chess. Determining the Classical World Chess Champion in rapid, and all the more, in blitz is just nonsense."

http://chess-news.ru/en/node/8423

Those are very strong words, but one which is clearly in disagreement with his colleagues of the present generation. Granted, the quote is four years old, but I bring it up because I find interesting is that this objection typically coincides with a nostalgic call to the days of 24 game matches whereas here Karpov posits that as few as 14(!) games are adequate to have a “full-fledged, creative fight”. It’s hard to understand the profound difference between 12 and 14 games because I have never come close to playing in a match at such a rarefied level, but his suggestion is a far cry from the call of some for the good old days of much longer matches.

Historical evidence

To consider the opposition to a shorter match a bit more thoroughly, let’s look at the history of 12 game contests. Of the seven that have been held under the current format, the reigning champion or higher rated player has won them all (see the table below):

Kramnik – Topalov 2006: Winner – Kramnik 2.5-1.5 in Rapid playoff after 3 wins, 3 losses (1 by forfeit).
Anand – Kramnik 2008: Winner – Anand in 11 games: 3 wins, 1 loss
Anand - Topalov 2010 – Anand in 12 games: 3 wins, 2 losses
Anand – Gelfand 2012: Winner – Anand 2.5-1.5 in Rapid Playoff after 1 win, 1 loss
Carlsen – Anand 2013: Winner – Carlsen in 10 games: 3 wins, no losses
Carlsen – Anand 2014: Winner – Carlsen in 11 games: 3 wins, 1 loss
Carlsen – Karjakin: Winner – Carlsen 3-1 in Rapid Playoff after 1 win, 1 loss.

These matches have been contentious affairs, mostly full of stress and anxiety (for the players at least) from the very beginning. Anand-Gelfand may not have been thrilling, but that may have been due to Anand not playing up to his highest level and Gelfand playing the match of his life (Garry Kasparov himself denigrated Anand’s play more so than anything: http://chess-news.ru/en/node/8423 ). The fact that the recent Carlsen-Karjakin match began with seven draws belies the fact that the players were badly trying to win, especially in games 3, 4, 5 and 9, which could all easily have been decisive if the player with the advantage had played more accurately. Over half of these matches had at least 4 decisive classical games, with 2 of them having 5 (not counting Kramnik’s forfeit loss against Topalov which would have made 6). Go back 50 years and you will find that these winning percentages compare favorably with many matches from the past. The fear of cautious play simply doesn’t hold water in the majority of cases nor, by the way, does the often stated worry that this kind of short match might somehow lead to more random results.

While there is a concern that short matches could be dull if the players choose to play things close to the vest, the data seem to support the idea that in many cases long matches can be quite anti-climactic. Kasparov-Short (1993 PCA World Championship; 24 games scheduled) was such a demolition by game 7 (Kasparov had racked up 4 wins) that some onlookers preferred that Short throw in the towel instead of having to play out the rest of the games (Final score was 6 wins to 1 loss in favor of Kasparov). Karpov vs Timman in the rival FIDE World Championship from the same year was effectively over by game 16, when Karpov held a 6-1 lead. Those were the last of the 24 game matches that more or less began the trend of shorter affairs. In the Kasparov-Anand match (1995 PCA World Ch.; 20 games scheduled), Kasparov had broken Anand’s spirit by game 14(!) - winning four out of 5 games at one point - that he was able to basically trot to the finish line. If one looks back at the history of long matches, this same pattern often repeats itself, leading to the natural assumption that a greater number of games simply isn’t necessary to prove who the superior player is at that moment in time. Even more interestingly, if we begin in the Botvinnik era (post-1951), it appears that any player who was leading a title match at the 12 game mark has gone on to win the title in the end (the lone exception I could find being Peter Leko against Kramnik in 2003).

To conclude...

The evidence strongly suggests that a 12 game match with Rapid tie breaks (no blitz) makes sense in the modern era. I realize I am biased, and here are some of my personal beliefs on which this bias rests:

  1. History has proven that the better player will usually win a match of any (decent) length with very few exceptions.
  2. Today’s general audience (and especially those age 34 and under) will not maintain much interest in a match that lasts over a month.
  3. Rapid chess is very often a reliable measure of strength (blitz may be up for question).
  4. Draw odds to the champion seems anti-sport. People want to see a clear winner.
  5. Even if the better player loses for some reason in Rapid, that in itself is a compelling storyline that suggests that the classical portion was hard-fought.
  6. Rapid chess is much more fun for and easy to explain to the general fan and media.
  7. Carlsen played one of the most beautiful moves in history to finish off a World Championship match…and he did it in a Rapid game!

Is the current system perfect? Nothing ever is. Personally, I think draws are the scourge of our game and something has to be done to deal with them (those who took a day off of work to watch the lifeless game 12 of the last match would most likely agree). I’ve written about this before, and I don’t think that Sofia rule solves the problem. My personal preference is the suggestion of Rustam Kasimdzhanov, - https://en.chessbase.com/post/kasimdzhanov-open-letter-to-fide-with-a-proposal - but that is truly a radical solution that would have to be tried multiple times in less important venues to test the validity of the concept. For the moment, I have to say that the current format (12 games with Rapid tie-breaks) holds water, but I am open to any suggestions that increase the tension from the word go and keeps the match within a solid time window that favors organizers, players and fans alike (less rest days probably make sense as well). Both Caruana and Kramnik told me they like the idea of playing a Rapid tie-breaker beforehand in order to give one player draw odds in the classical portion. That seems like splitting hairs to me, and I am certain it would not make the proponents of the classical-only model happy to know that Rapid games actually control the entire outcome of the match. Until some other fantastic proposal comes along, I have to say that the current model, sans Blitz, is not at all bad.

See what the Top players have to say about the format:

Maurice Ashley is an International Grandmaster well known for his dynamic brand of chess commentary and effective coaching style. He was a commentator for the Anand-Kasparov World Championship match as well as all of Kasparov‘s epic computer matches.

Discuss

Rules for reader comments

 
 

Not registered yet? Register

lajosarpad lajosarpad 1/1/2017 06:55
@Petrarlsen

you are right, we agree globally. Thank you for the interesting debate.
Dukenails Dukenails 1/1/2017 05:26
I don't understand what anybody is complaining about. The World championship match, despite some dull play possibly because of a bad stylistic match up, was exciting, tense, dramatic and riveting. I think as we have seen from the Rapid and Blitz World championships a playoff is necessary in case of a tie rather than any other method. Usually I would like to agree with Yasser but lets face it the rapidplay four game playoff was by far the most exciting day. I would personally prefer a slightly longer 16 game match and the rest days were simply bizarrely arranged with many Fridays and weekend days missed just when everyone is off work and wanting to watch! This would have been disastrous if the match had been played in the east. Make it 16 games, no rest days Friday to Sunday and I think it will be perfect.
chessdrummer chessdrummer 1/1/2017 01:22
"Kasparov-Short (1993 PCA World Championship; 24 games scheduled) was such a demolition by game 7 (Kasparov had racked up 4 wins) that some onlookers preferred that Short throw in the towel instead of having to play out the rest of the games (Final score was 6 wins to 1 loss in favor of Kasparov)."

Don't agree with this point. In that match, the games were very close with Short coming close to winning two of the game when Kasparov faced a hail of sacrifices. He was simply unable to close the games out. In some sports (like basketball) we can be blinded by the final score and not realize the competitiveness of the game. In many cases, the game is close before the win becomes academic and foul repeatedly to stop the clock. A one-point game becomes a ten-point game and we may surmise that the game wasn't close. Kasparov-Short was more competitive than the score indicated although Short was simply not strong enough. Those fans who say the "slaughter rule" be invoked misses this point.
Petrarlsen Petrarlsen 12/29/2016 04:40
@ lajosarpad : Globally, I think that we are now approximately in agreement on the essential points.

We don't completely agree on everything, but I think that if everyone had so close views on the Classical World Championship system than us, it would probably not be very difficult to conceive a new system !

Let's hope, now, that, in the future, a better system will be implemented by FIDE !...
lajosarpad lajosarpad 12/29/2016 03:32
@Petrarlsen 2/2

Let me show a paradox about the starting color. White may or may not have a practical advantage. If White does not have a practical advantage, then we can give White to any of the players at the first game, the Champion will still have the advantage. If White has a practical advantage, then we allow statistcial assumptions, therefore, let's see the percent of games of the Champion against strong opposition with Black. If that percentage is smaller than 50% then there is a higher probability the Champion will lose and the challenger will win. But I would be very surprised if this ever happened. So, if we would give White to the challenger, then the Champion would probably have an authomatic advantage and we could write the rules in such a way that they cope with the improbable exceptions as well. So if we accept color advantage assumption, then to measure White's supposed advantage, we need to see the ratio of White wins against strong opposition for the White player and Black losses against strong opposition for the Black player. They both will be below 50% with a high probability. If one is higher, we compare their relative distance to 50%. If their relative distance shows that Black was very unsuccessful with Black and White was very successful with White and Black happens to be the Champion and we believe that White has a practical advantage, then we reach your conclusion. But at that level we will always see players who are pretty successful with both colors.

"In fact, what I meant, for your one + one "long classical" games tiebreak, wasn't that the quality of the games would be inferior (it is marginally possible, but, if the players have sufficient time, for me, it is up to them to keep the quality level of the games, and if this isn't the case, the problem isn't with the structuration of the match, but with the players, who didn't manage to play at a "normal" quality level) ; what I meant was that, with this system, it is decided in advance that each of the two games of the tiebreak will be sufficient for the attribution of the Title. And, for me, compared with a mini-match of two "long classical" games, for example, this system puts a little less in the forefront the game quality, because, for example, a single mistake in one game can decide the Title. In a mini-match, for example, it can also happen that the last game could be decisive for the match's result, but it isn't decided in advance : it can only be the result of the first game of the mini-match that can bring this consequence. So, for me, your system is indeed a little less focussed on the game's quality. Personally, I don't consider this a real drawback at all, as long as the games are played with a "classical" time control, as this system is completely fair and logical, but this can be a matter of opinion... "

I see your point and we already have a few good approaches that we can choose from. The arguments between those are mostly based on subjective opinions from both parts. Taken the way we are arguing this means that the core is well-funded. You assume that White has a practical advantage, I think this assumption is premature. I think we will not convince each-other about this aspect, but I also think that both our thinking process from the starting point where we start from (and where we disagree) is mostly accurate.
lajosarpad lajosarpad 12/29/2016 03:09
@Petrarlsen 1/2

"In my opinion, this isn't a valid argument, because it is only an assumption, based on psychology and nothing else."
The quote above this thought contains the word "might", which excludes the assumption you mentioned here. I did not present assumptions, quite the contrary, I have shown that we can reason into the opposite direction, which was a clear attempt from my part to contrast your assumption that under the discussed tiebreaking conditions White has an advantage. Of course this contrast does not have the same value as your assumption, since your assumption has a foundation. But this does not validate your assumption. You used data gathered from past events to describe the future. Your reasoning relies on an unproved assumption based on observation, which is a very good basis to create a hypothesis about the future. Kasparov has always beaten Shirov. Does this mean that if we organize a match between Shirov and Kasparov, then Kasparov is the probable winner? Probably not, due to the fact that Shirov has invested great amount of time to study openings and train practical strength, while Kasparov has been working on something else. What was the logical mistake in such assumptions? We have missed facts which did not show in their past encounters. Let us imagine the case when we do not know Kasparov retired. In that case we could have the methodically valid hypothesis that Kasparov would probably emerge as the winner and we would be probably wrong, but from that information this hypothesis would be the best guess. From our current knowledge we would have the hypothesis that Shirov would be the probable winner. We are probably right, but we could be wrong. If, for example Kasparov maintained his training ways, he would have the advantage that he is more talented and has information about Shirov's more recent game. So, this statistical-based information can only found a hypothesis which is not factual knowledge, therefore White may have a practical edge, but we do not know whether this is the case.

Programmers do not need to be good players to program a good engine. They just need to know the heuristics to implement. And if the heuristics say that the player having an extra tempo has the advantage, then they will implement that and the engine will think in that way. Since their sheer calculating power is so enormous, computers see behind our horizon anyway and they would beat us even if they would be wrong somewhere where we are wrong as well. The computer evaluation only confirms that the human species thinks White is better at the start of the game and therefore cannot be used as proof. However, they can be used as an argument, as the computers may have shown the contrary if this assumption is mistaken. But this is, again, a hypothesis, a belief. It might be accurate though.
efan efan 12/29/2016 10:24
World Championship Tie-Breaker – New Idea Proposal

I agree with almost everything with Maurice Ashley; except for a small part that a tied Standard (classical) time matches are decided by rapid or blitz games. Most players are content, yet for long term principle we ought segregate Classical, Rapid and Blitz as different sub-events of Chess. This may not seem important for many players but for organizers who regard chess as a sport it matters.

If a long match such as the World Championship is drawn, I think that there should be a single game of standard time control tie-breaker. It is not a new idea; a single game decides and if drawn Black wins. But in this way the odd for Black is too great so Black gets less time. But how much less? My proposal is as follow;

1. Flip a coin, 2. The winner of the toss decides the amount of time for Black, 3. the other player decides the colour to play.

Perhaps White would get 90 min and Black will get anything between 10 and 80 min decided by the winner of the toss. (30sec ought be added for every move, and may be more time after 40 moves?)

This is a case of ‘one person cuts the cake into 2 and the other person decides which piece to have’

I believe it is a simple and effective tie-breaker that is worth trying out at least.


SONG Jinwoo
from Korea
Petrarlsen Petrarlsen 12/29/2016 12:25
@ lajosarpad (3/3) :

- "If White is given to the challenger, he might want to risk and press to avoid a solid draw and to have to fight for a draw with Black in the next game. The Champion would probably play for a draw. If Black is given to the challenger in the first game, he will probably try to play a sharp, but solid playing style to avoid problems, banking on the fact that if he does not lose, he will either be World Champion after the game, or White in the next game."

In my opinion, this isn't a valid argument, because it is only an assumption, based on psychology and nothing else. It might be true, but you seem to have absolutely no element to prove that the possibility that you imagine could indeed have serious chances to realize itself. And without any proof, anyone can always imagine an apparently coherent hypothesis that would constitute an argument for anything...

- "Under no means should we give the same importance of a game played by Kramnik when he was a little boy to his most recent game, also, we should avoid attributing the same importance to a game played by Kramnik 20 years ago when he was already a super GM as to a game played by him recently."

I couldn't find the criteria that are used for the selection of games used by FIDE for the winning / drawing / losing statistics. But, obviously, these games are NOT all the games played by the players : a very obvious sign of this is that Anand and Topalov, who are one and the other above 40, have much less games in these statistics as Rapport and Wei Yi, who are both juniors ! So, certainly, the statistics for Kramnik (for example), who is also above 40, must refer to games from a period when Kramnik was approximately at the same level than today, which makes this statistic quite significant.

But, even if I think that these elements are sufficient to draw reliable conclusions, I quite agree that it would be better to make a "fine-tuning" of these statistics, as you suggested. And, as a mathematician, all your ideas would very obviously be particularly interesting in that respect !

- "About the changes in quality of games in different scenarios"

Perhaps I didn't make my point quite clear on this subject.

In fact, what I meant, for your one + one "long classical" games tiebreak, wasn't that the quality of the games would be inferior (it is marginally possible, but, if the players have sufficient time, for me, it is up to them to keep the quality level of the games, and if this isn't the case, the problem isn't with the structuration of the match, but with the players, who didn't manage to play at a "normal" quality level) ; what I meant was that, with this system, it is decided in advance that each of the two games of the tiebreak will be sufficient for the attribution of the Title. And, for me, compared with a mini-match of two "long classical" games, for example, this system puts a little less in the forefront the game quality, because, for example, a single mistake in one game can decide the Title. In a mini-match, for example, it can also happen that the last game could be decisive for the match's result, but it isn't decided in advance : it can only be the result of the first game of the mini-match that can bring this consequence. So, for me, your system is indeed a little less focussed on the game's quality. Personally, I don't consider this a real drawback at all, as long as the games are played with a "classical" time control, as this system is completely fair and logical, but this can be a matter of opinion...
Petrarlsen Petrarlsen 12/29/2016 12:24
@ lajosarpad (2/3) :

- "The tiebreak odds is a provably stronger advantage than White in the first game, since the Champion does not need to win those games to secure the title, he only has to avoid a loss." and "However, giving Black to the World Champion in the first tiebreak game does not constitute, at least according to our current knowledge such a disadvantage that it would make it more probable to the challenger to win the match than the Champion"

If we take the 1st tiebreak game independently, if White wins, he is the Champion, if Black wins, he is the Champion, and if there is a draw, the tiebreaks continue. So, for me, at this stage, White has a slight advantage.

I quite admit that, extremely probably, the application of the "draw odds to the Champion" system after the 2nd tiebreaking game will tip the scale in favor of the Champion, even if the Challenger has White in the first game, but, for me, even this very strong probability isn't sufficient, because those are different things : on the one hand, the Challenger having White in the first tiebreaking game, on the other hand, as a third tiebreak (counting separately the two tiebreaking games), the application of the "draw odds to the Champion" system.

As I said before, if there are two contradictory advantages in a tiebreaking system, I consider that, for it to be acceptable, it is necessary to have the possibility to compare these two contradictory advantages with objective and numerical arguments. So, if it is possible to prove that, for example, the "White advantage" in the first tiebreaking game has a value of "1", and the application of the "draw odds to the Champion" system after the second tiebreaking game a value of "3", for me, that is completely satisfying, because it is then mathematically proven that, globally, the advantage goes to the Champion.

And, for me, if such a demonstration is not possible, the other choice must be made : to give the first White game to the Champion (as I said before, it has no significant drawback, so it can be used "by default", so to say...).

- You say that computer analysis isn't an argument in favor of an advantage to White. I agree the "computer argument" isn't a strong one for the reasons you state, but I nonetheless think it is all the same an argument, even if not a strong one at all.

Why ? Because all is proven by results, in chess. And the level ("Elo level" equivalent) of chess programs is determined by games played by chess programs against other chess programs. If the fact that some programs give an advantage to White was an error, there would inevitably be someone who would conceive a program that does NOT give an advantage to White (or else, that would mean that ALL the programmers who conceive chess programs think that White has an advantage, and, as these programmers are clearly chess experts in their own right, this would also be quite significant), and, as the analysis of this program would be better, it would have a better "Elo level" equivalent that the other programs (for a similar "base level"). It seems obvious that, if an erroneous bias is programmed, it will have negative consequences on the program's results, so the fact that the best programs give an advantage to White seems indeed to be, to a certain extent, an argument in favor of an advantage to White.
Petrarlsen Petrarlsen 12/29/2016 12:23
@ lajosarpad (1/3) :

I thought I could find enough time to answer to your lasts posts reasonably quickly, but I finally couldn't ; I'm sorry for this delay... and I hope that you will still read this at one moment or another !

Before discussing the content of your lasts posts, I would like to put a short preliminary word about our respective approaches : as you are a mathematician, you resort naturally quite frequently to mathematical arguments ; for me, as I am a lawyer, I logically use the type of reasonings that I master best. And, although reasonings are also quite fundamental in legal work, they are nonetheless quite different from a mathematician's reasonings. I think that the two approaches are both quite valid, but I will not enter myself into the realm of mathematics, because mathematics aren't my specialty, and it is more logical to reason in the way you master best...

- "Past events are by no means proving future events unless there is a provable implication relation."

For me, this was in fact the most obvious element, so obvious in my opinion that I didn't even think it useful to demonstrate it. But perhaps this point could be indeed discussed, too...

For me, generally, in all human activity fields where there is a progressive development, there is always a significant inertia : things will never change completely from one day to the other. This, quite simply, because the state of knowledge at a given moment can't permit such a sudden change ; the state of knowledge will not change from one day to the other, and, as a consequence, what is produced in the given field of human activity will not change either from one day to another. Three examples : 1) In architecture : the architect of the Palace of Versailles (or any architect of the same period) couldn't possibly have designed the Empire State Building. 2) In music, Bach (or any other Baroque composer) couldn't have composed music in Debussy's style (for example), because, notably, harmony developed progressively, and the necessary knowledge for that wasn't available at Bach's time. 3) Even the most gifted computer engineer of the 1950s couldn't have designed a present-day computer.

The same is true in chess : the chess approach of the 2700+ and 2800+ grandmasters is submitted to a progressive development, and nothing will ever change completely from one day to another.

So, for example, for their respective approach of chess play with the White or Black pieces, this theme is linked (in my opinion) to three factors : 1) Opening theory. 2) Strategical approach. 3) Psychological approach. It is obvious that, for all the 42 2700+ or 2800+ grandmasters, these three elements can only evolve quite progressively ; none of these 42 players could possibly change completely their approach on any of these three factors from one day to the other ; such an evolution would take some time.

So, yes, for me, there is quite a "provable implication relation" between the games played by the 42 top-players and the games they will play tomorrow. And, if White has clearly better results than Black today, it will still be the case tomorrow.

And thus, I still maintain that, for the moment, White has a practical advantage in Chess !
Petrarlsen Petrarlsen 12/22/2016 12:46
@ lajosarpad : I'm sorry ; I didn't have time to answer your last posts until now...

I think it will be possible for me tomorrow.
lajosarpad lajosarpad 12/20/2016 11:44
@Petrarlsen

IV. About the changes in quality of games in different scenarios

In short classical games, the quality drops compared to longer games either because of time trouble or because of quicker decision making to avoid time trouble. You can come up with the decent argument that the quality will still be acceptable and I will agree with you. My worry is due to the fact that time troubles are more probable in this case, but that is a very subjective worry and cannot be used as a challenge of the system's logic.

In a minimatch where Champion retains unless the challenger has a win before the Champion the very fact that the challenger has to play for a win adds additional tension to the game and he will be more willing to play in a risky style. On the other hand, if they have plenty of time to think about the game, then mistakes could happen due to

- blackout due to tiredness or nerves
- time trouble
- not understanding a complicated position where the challenger takes risks
- purposeful playing weaker moves to complicate the game

The first case can occur in any situation, the second will have an increased chance in this case, as at least a player will play for a win. The third and four possibilities have high chances and they will drop the quality of the games on the one hand, but would more than compensate in entertainment on the other hand. I am not worried of putting the challenger into a must-win situation in the last tiebreak game, because must-win situations have occurred before in such matches and those have more often than not led to interesting play and in many cases the must-win situation was the ingredient of an evergreen game, think about the very last game in the Kramnik-Lékó match where Kramnik was in a must-win situation and produced the game of the year.
lajosarpad lajosarpad 12/20/2016 11:30
@Petrarlsen

III. About using previous relative results to determine the color distribution in the tiebreak

While you are right that past results are true regardless of their actual reason and while I have shown why at least according to my opinion past events cannot be used to determine future events, including, but not limited to color distribution, I consider this idea to be sound. It is evidently superior compared to coin flip to determine the color distribution. A hypothesis is undoubtedly a stronger base than complete randomity.

There are still some questions to be answered though. Which color should be given to the Champion, the one with the better or worse result? You believe that the color with the stronger result should be given to the Champion. I am not sure but I could accept such a decision without any problem.

Also, if we apply this, I would like to suggest a fine-tuning of it: newer results are more relevant and we either weight their results with numbers associated to fuzzy categories (https://arxiv.org/abs/1410.1478) or we use only the recent results. Under no means should we give the same importance of a game played by Kramnik when he was a little boy to his most recent game, also, we should avoid attributing the same importance to a game played by Kramnik 20 years ago when he was already a super GM as to a game played by him recently.
lajosarpad lajosarpad 12/20/2016 11:17
@Petrarlsen

II. If White has a practical advantage, should it be given to the Champion?

In my previous post I have shown that it is premature to think that White has a practical advantage, except for the psychological state of the players playing given colors and shown that the arguments behind this idea are strong-enough to form an educated guess. This post assumes that White has a practical advantage and contemplates about the validity of giving White to the Champion under these circumstances.

The reason one would give White to the champion is that White given to the challenger in the first tiebreak game in my proposed system could form as a base of a counter-argument to the advantage of the Champion. The tiebreak odds is a provably stronger advantage than White in the first game, since the Champion does not need to win those games to secure the title, he only has to avoid a loss. However, a win for the Champion can not be excluded. The challenger has to play all in. As you elegantly pointed out a tiebreak of games weakens the advantage of Champion retains even if the supposedly advantegeous White is given to the Champion in the first tiebreak game, since that constitutes a chance to the challenger to still become World Champion in a situation the other system would already make that impossible and there are no compensating cases. However, giving Black to the World Champion in the first tiebreak game does not constitute, at least according to our current knowledge such a disadvantage that it would make it more probable to the challenger to win the match than the Champion because it is easier to at least hold two games than to become the first winner in two games.

As a result, even if we give Black to the challenger in the first game, if we can contribute any advantage to that, it is not more than a compensation.

Interestingly, we can use the same assumptions and facts as base of counter-arguments to your arguments in the following way:

If White is given to the challenger, he might want to risk and press to avoid a solid draw and to have to fight for a draw with Black in the next game. The Champion would probably play for a draw. If Black is given to the challenger in the first game, he will probably try to play a sharp, but solid playing style to avoid problems, banking on the fact that if he does not lose, he will either be World Champion after the game, or White in the next game.

So, basically I have to correct my previous thought, where I mistakenly believed that arguments can be used against Black given to the Champion in the first game. In fact, the argument is not strong-enough and has a symmetrical counter-argument.
lajosarpad lajosarpad 12/20/2016 10:36
@Petrarlsen

I. The validity of the argument that White has a practical advantage

Let us define what practical advantage is. Proposed definition:

Practical advantage in chess is the phenomenon when a player (the one having the advantage) has better chances than his opponent to achieve a win due to either provable advantage on the board or environmental properties not related to the moves already played by the two players.

Such a practical advantage can constitute to several possible things, including, but not limited to relative strength, health of the players, relative ability to play the already reached position, clock situation, psychological state.

The hypothesis this post discusses is that White has such a practical advantage. The fundament of the hypothesis is that in past games of strong players, their relative result with White were better than their relative result with Black. An additional argument is that computers favor the White position in many cases.

I challenge the fundaments of this hypothesis for scientific reasons. We can accept that White seems to have an advantage because of past results and computer evaluation, but it is premature to factually state that White has such and advantage over Black because while the fundament and the further argument are strong enough together to form an educated guess, but can not be used to prove the hypothesis.

Past events are by no means proving future events unless there is a provable implication relation. Without that provable implication such a conclusion is a misuse of the mathematical induction (https://www.mathsisfun.com/algebra/mathematical-induction.html). In the URL shared we have the example of the dominos. In our case the mistake is that based on the fact that a lot of dominos have already fallen we assume that the next dominos will fall as well, without proving a connection of implication between the past and future events. Without that connection, the past events can not be used as a proof for future events.

Computer evaluation uses heuristics implemented by human beings, who quite frequently tend to believe in the myth that White has an advantage due to the extra tempo in the initial position and due to the limits of analysis, since an extra tempo is considered to contribute to the position unless a zugzwang is found, which is a mistake in zugzwangs beyond the limits, a phenomenon which can be used by strong GMs to beat computers. This argument, addmittedly with an exageration can therefore be translated to "White is practically better, because we practically believe that White is better."

The only reason I can think of we could attribute a practical advantage to White with our current level of knowledge is that players who believe that White has a practical advantage will tend to be more confident with White and more unconfident with Black (see psychological state among the reasons of a possible practical advantage).

I call White's practical advantage a myth, since it is unproven and widely spread, but it could be actually accurate. However, without being able to determine its level of accuracy, we should not accept it as a fact.
Petrarlsen Petrarlsen 12/19/2016 09:21
@ fgkdjlkag : I think that the problem isn't with chess itself, but rather with chess as you conceive it.

For you, the World Championship has for purpose "to determine the stronger player". For me, the World Championship is a competition where the only thing that counts is that, following a given set of rules, one player can be determined as the winner. And I consider that, if a system is fair and logical, it is sufficient (for the "classical" World Championship, I find also much preferable to avoid anything that isn't classical chess). And, for me, this can be compared with sports like 100 metres, long jump, or pole vault, where the only goal is to win, even if, for example, two participants have nearly the same results : if the system used for the World Chess Championship is a good system, I don't see the necessity for more than a minimal difference in results between the players.

And clearly, the large majority of persons interested in chess doesn't think either that a particular difference in result is needed : the posts under the two Seirawan, the Sutovsky, and this Ashley article show it clearly - there are many very different propositions, but nearly none of them are oriented in this direction.

I think, that, indeed, in chess, it would be difficult to create a realistic system for the World Championship that would require an important difference in results between the two players.

And, furthermore, I think that, perhaps, you don't take fully into account in your reasoning the fact that there is significantly more than 50 % of draws in chess at top-level : in chess (contrary to Go, for example), when a 2700+ or 2800+ grandmaster wins a game, it means that he managed to come out of this "central draw zone". And we see again and again (and the Carlsen - Karjakin match was a particularly good illustration of that topic), that, even in difficult positions, when a top grandmaster defends well, it isn't easy to win the game for his opponent. So a win really demonstrates something, for the winner.

So, in fact, for me, even a minimal difference in result between the two participants of a World Championship match is already quite significant.

As for the mistakes, in my view, they are not a problem ; mistakes are an integral part of chess (no players are completely immune to them, even at the top level and in classical games), and they don't bother me as long as they aren't due to an inadequate (for a classical games title) time-control. But, for example, I don't find that 30 mn. + 30 s. games (used as tiebreaking games) would be too short in that respect (notably due to the 30 s. increment that prevents the players from getting into a too severe time-trouble).
Petrarlsen Petrarlsen 12/19/2016 09:19
@ lajosarpad (4/4) :

As for your last post :

- The three points that you propose : Besides the question of giving White to the Champion or the Challenger, for which I developed my own opinion before, I agree completely with these three points.

- As to the 30 mn. + 30 s. games in my system, I think that the following idea could be used :

After my first tiebreak (the "virtual mini-matches" system), that, in my system, I would maintain in all cases (because, in my opinion, this system would not affect the quality of play, would not have any consequence on the match if this match isn't tied after all the "normal" games have been played, and would not increase the number of draws), I would propose four possibilities :

1) Applying directly the "draw odds to the Champion" system (as my "virtual mini-matches" system would have already been used, it would not occur very frequently, so this would seems to me an acceptable solution).

2) Using your two games tiebreaking idea, with the Champion having White for the first game.

3) Having a playoff match of 4 games with a 60 mn. + 30 s. time control, on two days (applying again my "virtual mini-matches" system in case of a tie, at the end of this playoff match).

4) Having the playoff that I originally intended to use : 6 games with a 30 mn. + 30 s. time control, on two days also, and still applying my "virtual mini-matches" system in case of a tie at the end of the "playoff match".

The criterion of choice, in my opinion, would be very simple : either the choice is made of favoring completely the quality of play, and it would point to the first solution, the other (extreme) possibility being to think that the "draw odds to the Champion" must only be used as very last resort (to avoid mixing the results of two different World Championship cycles), and this would have as a result the choice of the fourth solution.

And the second and third solutions are intermediate solutions, which favor the quality of play, but not in a completely absolute manner, for your solution (I consider that, with this solution, the quality of play is a little less in the forefront because the two tiebreaking games are taken individually, and that this has for a consequence that the Title is finally played on a single game, whereas, with the first solution, one isolated game isn't sufficent to decide the Title), or which rather favor avoiding to use a result of the previous World Championship cycle as a tiebreaker (the "draw odds to the Champion" system), for the third solution.

Personally, and purely personally, the solution that I would prefer would be the 4th (with the 30 mn. + 30 s. games), because I would subjectively prefer, as much as possible, to see the title being decided between the two players on the basis of the results of the current match, and because I would think that the game quality, in the 30 mn. + 30 s. games, would still be sufficient to have a match result that would really results from a better level of play by one of the players. But the three other solutions would suits me quite well too...
Petrarlsen Petrarlsen 12/19/2016 09:18
@ lajosarpad (3/4) :

And I maintain that, as a complement, I think that using the opening evaluations of the present day chess programs can be useful (even if, as you said quite rightly, these programs are programmed by humans...), because they are globally very efficient tools for evaluating a position (...this is confirmed by their extremely high "equivalent Elo rating"...), and that, taking this into account, it seems obvious to me that it is more probable that their evaluation is right in a given position than the contrary. So, in addition to a study of the results of the 2700+ and 2800+ grandmasters, this, in my opinion, is useful to confirm the result of the "White vs. Black contest", by adding a complementary approach to the first one. If the two approaches give the same result, it seems logical to me that the drawn conclusions must be stronger (even if only marginally so) than if only one of the two methods was used.

So, for me, in view of this conclusion (this being obvious that if this conclusion is not accurate, the following must also be reconsidered...), I consider that it would be more logical that the Champion would have White in the first game, because, in my opinion, the problems linked to the two contradicting advantages that would exist if the Challenger plays White in the first game constitute a real counter-argument for giving White to the Challenger in the first game, as, on the contrary, I don't think there are significant arguments that work in favor of giving White to the Challenger in the first game.

(It could be argued that the advantage of White for the first game to the Champion + the implementation of the "draw odds to the Champion" system would be too much of an advantage to the Champion, but, in fact, yes, there are two advantages that are given to the Champion, but, taken together, they give less of an advantage to the Champion than the pure and simple application of the "draw odds to the Champion" system - if the "draw odds to the Champion" system is directly applied after the last "normal" game of the match, the Champion is directly awarded the Title ; if two games are added before applying this system, even if the Champion has White in the first game, the Challenger has potentially two supplementary possibilities of winning the Title, so this tiebreaking system is in fact advantageous for the Challenger, in comparison of the direct application of the "draw odds to the Champion" system. So, as I consider the direct application of the "draw odds to the Champion" system acceptable, I can only conclude that the "two tiebreaking games with first game with White for the Champion" system is all the more acceptable.)

I have a further argument in favor of giving White to the Champion in the first tiebreaking game, and, thus, White to the Challenger in the last game : for me, it would be also more logical for the "general structure" of the match : if a tie subsists until the very last tiebreaking game, if the Champion is given White in the first tiebreaking game, then the Challenger will have White in this game. And, as the goals are completely clear for this game, the Challenger being under the obligation to win this game and the Champion having only to draw it, it would, in my view, seems more logical that, for this very last game of the match, the Challenger would have the advantage of having White for his last attempt to win a decisive game. Otherwise, this last game will really constitute for the Challenger a "desperate attempt", since he would have the double handicap to be under the obligation to win this game, and to play with the black pieces...
Petrarlsen Petrarlsen 12/19/2016 09:17
@ lajosarpad (2/4) :

4) The method I would propose would be to take into account the winning AND losing (this is a new improvement of this system...) FIDE statistics (using for the present moment the official list for December 2017) of every 2700+ and 2800+ grandmasters (the players who objectively master best the game of Chess) with both colors. For the moment, I don't see other methods that could take into account a comparable number of games, so I will not compare this method with another one, but I will only try to evaluate it independently.

- Is this method using a sufficient number of games ? : For each player, in the FIDE statistical data, there are several hundreds of games for each color. For a comparison, a Candidates Tournament, all players included, represents 56 games. Therefore, it seems to me that several hundreds of games for only one player clearly represent a large number of games, sufficient to draw reliable conclusions from it.

- I don't see any reason to think that these data could be unreliable as to their content (for example, if the players would generally play against stronger opponents when they have Black than with White, which would introduce a bias in the results ; it could maybe happen for one or the other player, but I don't see how it could be the case for a significant part of the 42 2700+ and 2800+ grandmasters).

- Two elements can, in my opinion, be used to say that the global result of this analysis is quite conclusive :

a) All the 42 studied players have both a better winning percentage and a better "non-losing" percentage (i.e. adding wins and draws, the interest of this last statistic being to show with which color it is most difficult not to lose a game) with White than with Black.

b) For certain players, the difference between their results with White and with Black can be very significant. For example, if we take only the 4 2800+ grandmasters (the players that are presumed to have the nearly best possible level for a human player of today, so that, logically, they shouldn't have big gaps in their mastery of chess playing, with either color), the player who has the biggest difference between his results with White and with Black is Kramnik : he wins more than two times more with White than with Black and he loses two times less with White than with Black - an enormous difference in results. So it isn't only that White has slightly better results than Black (which is in fact only the minimum result for every 2700+ and 2800+ grandmasters) ; even for certain absolute top players, there can be a very great difference in results between their White and Black results.

As for me, in view of these elements, I conclude that, for the present moment, White has a (slight) advantage in classical games, between 2700+ and 2800+ grandmasters.
Petrarlsen Petrarlsen 12/19/2016 09:17
@ lajosarpad (1/4) :

As I had already began to write a "project" for a post about the White or Black advantage question before reading your last post, I will also begin by this question...

So, notwithstanding your argumentation, I must say that, for the moment, I continue to consider that, currently (and only currently - I completely agree that this evaluation can change in the future), White has a (slight) practical advantage ! (You will probably not agree with me, but the reason for which you will not agree, if this is the case, will be very interesting for me !...)

This is my reasoning, step by step, from the very beginning :

1) One of the most fundamental ideas, in Chess, is to rank players, using for that purpose the result of games played by these players (to rank them in a tournament, in a match, in the Elo list, etc.).

2) On the same bases, it seems logical to me that, if we can rank and compare players so using the result of chess games, it must also be possible to "rank" and compare the results obtained by the two existing "colors", in Chess, White and Black, in the same manner (considering for that purpose that the colors are "represented" by the players : as, necessarily, a player in a given game tries to obtain for himself the best possible result, given the circumstances, I think we can safely consider, too, that, in a way, he "champions" the color for which he plays, as he tries to play the best game possible with this color : it gives the same result - in details, some nuances could be introduced, but I think that, when taking into account a great number of games, this general idea must nonetheless be true).

And it would also seem logical to me that, as for a match or tournament result or for the Elo ratings, that the only elements that count are the "raw results", taken purely concretely, without taking into account the reasons that caused these results (for example, for the White and Black question, the results obtained can be caused by many factors : obviously, an objective chess advantage at having White if there is one, but not only, and by far ; it can also notably be due to the opening theory of the considered period of time or to a different strategical or psychological approach of the game by the players according to whether they have White or Black, but, in my opinion, these factors must not be taken into consideration for the global result - whether White or Black has an advantage, at the considered period of time). The reasons for which a player, for example, wins a World Championship are not taken in account for the awarding of the World Title, and I don't think either, that, for evaluating the respective chances of White and Black, the "reasons of the result" must be taken into account.

3) It seems to me logical that, to have a really reliable result, the followed method must take into account a great number of games. For example, the results of a given competition (a tournament, for example - a double round robin tournament could be probably one the best possibilities in this direction) would not, in my view, take into account a sufficient number of games to give a really reliable result.
fgkdjlkag fgkdjlkag 12/18/2016 09:17
The more I think about this issue, the more I realize that the problem is not with the format or the players, but with chess itself. The previous article I cited (http://math.bu.edu/people/aki/j.pdf), demonstrates that in standard time control classical games, in previous world championships, even 20-24 games and more is sometimes not enough to establish who is the stronger player. The purpose of having any additional games in shorter time controls is only to induce mistakes, which were not happening as much, or as severely, in the longer time controls. Using draw odds to the winner is an admission that the stronger player is not known. Using the first to win a set of games as a tiebreak during the match, does not prevent the overall match from ending in an equal score, and also introduces an asymmetry, in that the player who receives white first will have a small advantage. In Go, there are very few draws, likely because Go is complex enough to distinguish between players of similar ability. This being the case, and the purpose of the world championship being to determine the stronger player, no world championship format/tiebreak system will be agreeable to all, nor have all the ideal features, save for very long classical matches, which are also not ideal. The best solution seems to be to switch to a more complex form of chess (eg, seirawan chess, finesse chess, capablanca chess, fischer random, etc).
lajosarpad lajosarpad 12/18/2016 02:06
@Petrarlsen

You are absolutely right that short classical games would represent a major improvement compared to the current system. If your system would be applied then I would agree that justice has been done and we have a decent system. My subjective counter-argument, which, under no means challenges the logic of the system is that short classical games have a much higher chance of leading to the scenario you described than longer games. This makes me slightly worried about the system and my subjectively based opposition is based on my opinion that other decent systems might not have this disadvantage. You can correctly argue that other systems have their own disadvantage too. For instance my proposed system has the danger of decision in the odd game due to a bad day for the loser, having an uneven color distribution. Champion retains has the disadvantage of having a high chance of undecided result. I consider these to be lesser dangers.

Our disagreement about the preferable first color distribution in the tiebreaks in my proposed system comes from our different judgement about the initial position. I, however do not say that White is preferable for the challenger, therefore our disagreement is in essence not due to opposing arguments and I believe that the goal is to make the arguments clash and learn from that and hopefully reach either an agreement, or opinions close-enough to make sure that both ideas are acceptable for both. I think we already reached that point and thus the goal of our debate, however, I am still interested to see your arguments due to the hope that I could learn from them. Therefore I will check this discussion tomorrow.

But before that, acknowledging that you do not consider my idea to be completely satisfactory unless Champion has White in the first game, I am asking you whether we agree on the following points:

1. Champion retains is a completely decent system, acceptable if chosen.
2. Short classical games tiebreak is a completely decent system, acceptable if chosen.
3. A finite set of games (originally 2 were proposed, but arguable) as a tiebreak with the first winner winning the match is a completely decent system, regardless of the choice of color for the first game but with the nature of distribution known in advance.

I do not ask whether all combinations are satisfactory for you, I already know the answer. I am asking whether you have objective, provable arguments against any of the scenarios above. If not, then all these can be presented as a proposal, possibly with other, newer ideas included. If so, a referendum in the chess world is to be advocated. If not, then we need to reach a conclusion about the debated scenarios before we do such a proposal.
Petrarlsen Petrarlsen 12/18/2016 03:11
@ lajosarpad :

There are still some nuances for which I don't agree with you about the White - Black advantage problem...

Unfortunately, for now, I haven't have enough time to sort this completely out ; I will write something here tomorrow or the day after ; I hope that it will be possible for you to come back to this page at this moment, because I would be very interested to know what you will think about it ! (even if you probably will still not be in agreement with my arguments !!...)

About the 30 mn. + 30 s. games, when you say : "I am afraid it could lead to time troubles when two tired persons play them", for me, the advantage, with the 30 s. increment, is that the worst possible situation would be that the players would play exclusively on the 30 s. increment. But that would not in fact be a worse time trouble that what can be possible in the "normal" World Championship games, with a 100 mn. + 50 mn. + 15 mn. + 30 s. time control, because, even with this last time control, it occurs sometime that the players end up by playing solely on the 30 s increment.

And this would make, in my view, an enormous difference with the present rapid 25 mn. + 10 s. games of the World Championship playoff, where, when the players play on the 10 s. increment, it isn't really anymore anything comparable with classical chess, in my opinion !...
lajosarpad lajosarpad 12/17/2016 03:35
@Petrarlsen

"As for the "White advantage" question, what would you think of the idea of checking periodically this question by using the two elements that I used this time ? (winning percentage for 2700+ and 2800+ GM and opening's computer evaluations) I would notably say that if all the top-GMs have better results with White at a given moment, we can safely say that for a given game between two players at the same period, White has a (practical) advantage. "

I almost agree. The point I would change here is that I would not say White has a practical advantage, but I would rather say that White seems to have an advantage. And this difference between points of view leads us to different conclusions. I have nothing against the Champion having the Black color in the first tiebreaking game, as before the advantage is proven I cannot rely on that. I would not give a proven advantage to the challenger, but a psychological advantage is feasible to motivate the Champion to play aggressively. (Forgive me, I am a mathematician and programmer). If the advantage is proven, your reasoning about contradicting advantages will have a strong base, but as it stands, the draw odds is a provable advantage, while the color advantage is speculative. And if these statistics of recent games are used to determine the initial color of the players in the tiebreaks, giving the seemingly better color to the Champion, then I have no objection against it, but I would also not object of giving it to the challenger as compensation until the color is proven to be better. But even if White is better, due to the tie break odds of the Champion, the challenger will (over?)press with White.

In this moment, before a proof is given I think all three variants have strong arguments and Champion having White in the first game has no strong counter-arguments, so it might be slightly preferable to the other two variants. Subjectively we can debate any systems. A good example to that is the person who attacked you, because he did not know that your proposal is really about short classical games. However, your system is perfectly logical. I am not for it because I am afraid it could lead to time troubles when two tired persons play them, but that is subjective. If we look at the starting point, that the match was already drawn when they reach the tiebreak, we have a pattern of equality between the players which could be reached by drawing many games or many decisive games. If the number of decisive games is more than the number of draws and among the decisive games White was the winner, then it seems that White has an advantage. But if we see a match where many games were decisive and most of them with Black, then giving White to one of the players could be a disadvantage to the player. So I am still not sure what color should be given to whom, I am only sure that in this system the two games should be played with alternate colors.
Petrarlsen Petrarlsen 12/16/2016 11:52
@ lajosarpad :

Yes, I also find it quite pleasant to discuss with persons who always stay polite, even in disagreement ! There aren't really many of them on the Internet, but, nonetheless, under the four "World Championship Format" articles, there where several of them ; I must say (and quite obviously, you agree with me !) that it is MUCH appreciated !! To see the positive side of things, when a thing is rare, you appreciate it more !...

To come back to the "chess things"... I indeed think that we are in agreement on the most important points.

I've thought a little more about your "two classical games" tiebreaking system, and I think I now see better why I feel a little uncomfortable about giving White to the Challenger in the first game (...and I must say that I am quite curious to see what you will think of this argument, because it is an argument which, in my opinion, could also be used in other cases...).

About your system, I quite agree that there are two "contradicting" advantages, one for the Challenger (having White in the first game), and one for the Champion (who is benefitting from the "draw odds to the Champion" system), and that the Champion's adantage SEEMS to be quite stronger than the Challenger's advantage.

The problem is that, for me, that it seems only possible to make a SUBJECTIVE comparison between the two : I don't think that there is any means to make an objective assessment of these two advantages that could be used for a comparison. For example, if we say (I would very approximately evaluate it this way) that we can attribute 3 for the "draw odds to the Champion system" advantage, and 1 for the "Challenger having White in first tiebreaking game" advantage, I don't think that we could really demonstrate (or refute) it objectively in any way.

So, for me, I would think that it isn't wholly satisfying to have contradicting advantages that cannot be compared on an objective basis. This because there isn't a PROVEN certitude that it is one or the other player who is given an advantage by the given system (we can "feel" it, be personnally certain that one or the other player has the advantage, but without any objective and numerical argument to demonstrate it).

This is why, in my opinion, it would be better to use a system with contradicting advantages only when it is possible to attribute, on an objective basis, numerical data that permits to know without any possible discussion which player is advantaged by the given system.

As for the "White advantage" question, what would you think of the idea of checking periodically this question by using the two elements that I used this time ? (winning percentage for 2700+ and 2800+ GM and opening's computer evaluations) I would notably say that if all the top-GMs have better results with White at a given moment, we can safely say that for a given game between two players at the same period, White has a (practical) advantage.
lajosarpad lajosarpad 12/16/2016 02:56
@Petrarlsen

Your thoughts were interesting again, it is good to debate with someone who can graciously disagree, which is unfortunately not common on internet forums. We agree in several important points:

1. The match should feature classical games only. Your idea of short virtual matches complies to this idea of quality over quantity as well.

2. It is unfair to give the challenger an advantage over the Champion who already earned his privileges.

3. Randomity should not be a decisive factor.

4. Champion retains either as the sole tiebreak or as a last resort is reasonable, although, agnostic approach is recommendable, we should listen to other ideas as well, as we do not know where are the hidden gems in advance.

The idea of maximum two tiebreak games with alternating colors seems reasonable to me, like other plausible systems. If I understood your point, then you can only accept this system if the Champion has the White pieces in the first tiebreak game. The reasoning is that White seems to have an advantage at least from a practical point of view, since the very best players have better success with White than with Black and engines evaluate many positions as slightly better for White. This is a good basis for a hypothesis, but it is not adequate for more. The better success rate with White for the very best can be at least partly attributed to the widely spread belief that White has a better starting position due to the extra tempo. Engines are also programmed by humane workforce and they use something very similar to alpha-beta pruning for a limited depth. The evaluated positions are evaluated according to a heuristic, which, again, is planned and implemented by humans. This means that people biased in White's favor do the implementation and therefore if you give a completely symmetric position to an engine, the engine will favor the player who moves first unless a zugzwang is in the horizon of the engine, in which case the side not getting into zugzwang will be favored. For the sake of simplicity I ignored those zugzwangs which do not change the evaluation. This means that the widespread belief that White is better reappears in the evaluating algorithm, therefore the limited-depth algorithm which uses that belief cannot be used as a proof of the belief due to the inadequancy of argument self-proving.

So, if we accept that White is better in the starting position, the seeming validity of the hypothesis must be periodically checked and trigger system adjustment if needed.

But let us assume for a moment that White is better and the challenger having White in the first tiebreak game would have an advantage. Since the Champion is favored in case of 1-1, the advantage of the Champion would be much bigger than the (arguable) advantage of the Challenger, making the advantage of the Challenger only a remedy, or compensation if you will.

Randomity used as a factor would not be decisive in this case, as we cannot foresee the challenger winning unconditionally in the first tiebreak game against the Champion with White and it is very possible that the Champion will win with Black if the challenger (who does not want to "waste" the last White) overpresses, then the Black pieces could become decisive psychological factor for the Champion.

So we do not agree in some of the points, but I believe that we are in agreement about the most important points.
Petrarlsen Petrarlsen 12/16/2016 11:54
@ fgkdjlkag (3/3) :

- As for the "draw-odds to the Champion" system, you misunderstood what I said : what I really meant was that, globally, under the two articles by Seirawan, the article by Sutovsky, and this article by Ashley, there was approximately 500 commentaries (a little more, now, because commentaries continue to appear !), which is rather enormous (I don't think I've ever seen such a number of commentaries on one isolated theme on ChessBase).

And an astounding proportion of these commentaries are in favor of the "draw odds to the Champion" system (in one form or another - it can be also the idea of giving another sort of advantage to the Champion).

I must say that I was myself really extremely surprised to see that so many persons stood up for this system.

And I think there also is a great difference between deciding that, after 12 games, if there is (globally) a draw, the Champion will stay the Champion, and to use this idea as the very last resort, after several other tiebreaking ideas. For example, in my system, realistically, I don't think it would happen much more frequently than between one and three times by century. I could be mistaken, but, in any case, this is the way that I see it...

As for me, I don't find shocking that, if it occurs in rare occurrences, if the Challenger didn't manage to demonstrate being capable of beating the Champion, the Champion keeps his title... It is the simplest of tiebreaks (using, in fact, the result of the previous World Championship match as a tiebreaker), and it solves the "tiebreak problem" easily in cases where, either "strange ideas" would be used (as the Armageddon game, for example...), or multiple classical games would be played as a tiebreak, with potentially very serious financial consequences...
Petrarlsen Petrarlsen 12/16/2016 11:54
@ fgkdjlkag (2/3) :

3) "The Sonneborn-Berger System" : As we know, this system (schematically) indexes the results obtained by a player on the results of his opponents (for a given result, a player gains more points if his opponent scores well overall in the tournament, and less points if his opponent scores poorly).

This system seems to me to be rather artificial (for example, the "value" of a game played the first day of a Candidates Tournament will be known only after the end of the tournament - more than two weeks after ; also, the "tiebreak winner" can very well be decided, on the last day of the tournament, by the fact - for example - that the tail-ender of the tournament will win or lose his game). And the results of this tiebreaking system are very difficult to anticipate by the players - the results of each player depends on the results of all the other players, so it can't really be taken into account in the player's general strategy.

Thus, I think that this system is less satisfying than my "virtual mini-matches" system, because it is much more artificial and unforeseeable.

So, globally, it seems to me that my "virtual mini-matches" system doesn't at all compare unfavorably to the tiebreaking systems that are used in the Candidates Tournament.

And, as the "virtual mini-matches" system seems to me to be potentially very useful to reduce expenses (in effect, for a 12 games match, it permits the same result as 6 "classical 2 games mini-matches" in a playoff, without requiring the players to play a single additional game), I think it could be interesting to implement in a match...


- As for the question of the "short classical games", I think that, probably (this would obviously have to be checked in practice !) the "chess ideas" would be less deep and interesting than in "long classical games", but that the global quality of play would nontheless remain correct.

My idea is that, for a "classical match", tiebreaking games must NOT operate a selection on speed, but on quality of play. And I think (and hope !) that with 1 hour of thinking time (in equivalent - main time + increment) and with a 30 s. increment (...that would make impossible too big a time-trouble...), what would differenciate a player from another would be the quality of his play and not essentially his speed.

This while keeping the costs inside a reasonable limit, the 30 mn. + 30 s. time control allowing a total of 6 games in 2 days.
Petrarlsen Petrarlsen 12/16/2016 11:53
@ fgkdjlkag (1/3) :

- As for tiebreaking systems in tournaments and matches, my opinion is that it is only because it is not usual to use tiebreaking systems in matches that a certain number of persons think that such tiebreaking systems can work in tournaments an not in matches.

It is nonetheless quite true that a difference between tournaments and matches, in relation to tiebreaking systems, is that there is only two players in a match, contrary to a tournament : it makes it much more difficult to create tiebreaking systems for a match (besides, I think that none of the tiebreaking systems that are commonly used in tournaments could be used in a match), and I don't think it could be possible to use a whole series of successive tiebreaking systems, as it is frequently the case in tournaments.

But I don't think that, because of that, the tiebreaking systems that could be used in a match would bear worse results than those of the tiebreaking systems that are used in tournaments.

I think it is interesting, for example, to compare my "virtual mini-matches" system with the three tiebreaking systems used in the Candidates Tournaments :

1) "The results of the games between the players involved in the tie." Most of the time, there are only two players involved, so it ends up also into a sort of "virtual mini-match" between the two players. In my view, this system doesn't seem to be very different from my "virtual mini-matches" system. As in my "virtual mini-matches" system, it gives a slightly greater importance to a single pair of games (in the tournament system, they are the two games between the two tied players ; in my match system, they are the two first games - by pairs of games - that are won, as an ensemble, by one of the players), so these two systems seem rather similar to me.

2) "The total number of wins in the tournament of every player involved in the tie." This system seems to me quite logical (it gives an incentive to win, for the players, which, objectively, can only be a good thing, because it isn't possible to rank players if there aren't any wins ! so to encourage the players to win games is necessarily a good thing), BUT it also seems to be quite artificial, because this tiebreaking system also favors the player who has lost the most games ; in effect, if a player has one more win than a second player with the same global score, the result will be that, for 1 win and 1 loss of one player, there will be 2 draws for the other (for example, in the 2013 Candidates, Carlsen had 5 wins, 2 losses, with 7 draws ; Kramnik had 4 wins, 1 loss, with 9 draws, and Carlsen won on tiebreaks ; Kramnik's 2 supplementary draws were in a way "exchanged" for 1 win and 1 loss for Carlsen). And I don't think that so "exchanging" 1 win and 1 loss against 2 draws has any significant meaning in terms of quality of play.

Thus, I think that this system is rather less satisfying that my "virtual mini-matches" system, because it is more artificial. (In the "virtual mini-matches", to win on tiebreaks, a given player must really demonstrate something in terms of quality of play : he must be the winner of a "virtual mini-match". And, as this rule would be more than obviously annouced in advance to the players, the player who would have won the match because of this tiebreaking system would really have demonstrated something chesswise, because he would have managed to do what his opponent wouldn't have done : to be the first to win a "virtual mini-match". While the tournament tiebreaking system based on the total number of wins doesn't seem to really demonstrate anything chesswise.)
fgkdjlkag fgkdjlkag 12/16/2016 03:27
@Petrarlsen

I believe all the tiebreaking systems only work in tournaments, not in matches, as they are dependent on having at least 3 players (sonneberg-berger and the like, I'm not referring to draw odds or things like that).

I think the question is, are we going to see as many deep brilliancies in short classical games as we are in long classical games. I don't know the answer to this.

I missed the 500 commentaries in favor of draw odds idea, are you talking about the comments sections of the chessbase threads? I personally do not like the idea of one player getting the advantage. In historical tied matches with draw odds, we do not know who was the better player. Thanks for the post about boxing @mdamien.

@Resistance, I don't know if you saw the article I linked to (http://math.bu.edu/people/aki/j.pdf), but it has the same idea as you, a long enough match to determine the better player. I certainly wouldn't mind a long match.

Anyone organizing the next world championship just has to read all the comments here and in a few other threads for a plethora of ideas and opinions.
Petrarlsen Petrarlsen 12/15/2016 08:35
@ lajosarpad :

I agree that there is no definite proof that White has an advantage in Chess.

To conclude that, practically, White seems to have an slight advantage, I took into account two elements :

1) The fact that the winning percentage is globally higher for White than for Black for top-level players (I have checked : there isn't a single 2800+ or 2700+ GM that hasn't a better winning percentage with White than with Black).

2) The fact that, when analyzing with modern engines the supposed "best" openings of today (at least the openings used in present day World Championship matches - thus obviously excluding slightly "outlandish" openings like the King's Gambit, that are not really conceived to give the best possible position to the player who choses it, but rather a certain type of position that this player is aiming at), from the starting position until the end of the opening, White is given as having permanently a slight advantage.

It is a completely "pragmatic" approach, and it certainly isn't a definitive proof, by I would rather think that, for practical purposes, it is sufficient to consider that White has a slight advantage.

As for the question of the choice of color in your system, personally, what I would absolutely exclude, would be to use a drawing of lots for that ! I wouldn't find at all satisfying the idea that the World Title could be decided, even for a rather slight part, by a drawing of lots !

I agree that, contrary to what I thought originally, there can be arguments in favor of giving White to the Challenger in the first tiebreak game, in you system.

Personally, though, I would nonetheless prefer to give White to the Champion.

Why ? Because, for me : 1) I think that in such a competition as the World Championship, all must be completely fair ; giving an advantage to the Champion is fair (I know that you agree with me on this point), but to give an advantage to the Challenger would not be fair. 2) For a solution to be fair, there musn't be a realistically plausible analysis of this given solution that could reach a conclusion showing that this solution isn't fair. 3) In your system, if you give White to the Challenger in the first tiebreak game, the Challenger is given a slight advantage, having White in this game. Yes, this advantage is probably more than compensated by the final use of the "draw odds to the Champion" system, but still, in my opinion, a slight doubt remains as to the fact that this version of you system would favor the Champion or the Challenger, and thus, I wouldn't find this completely satisfying.
ulyssesganesh ulyssesganesh 12/15/2016 01:22
remember karpov--kasparov had 24 games .... so let us have at least 16 games instead of the present 12....
ulyssesganesh ulyssesganesh 12/15/2016 01:20
sorry, morris. yasser is right!
lajosarpad lajosarpad 12/15/2016 01:00
@Petrarlsen, your question is very interesting and thought-provoking indeed.

Does White have an advantage against Black in the starting position of the game? Even the question is ambiguous, since it can be answered based on several possible and conflicting philosophical directions. If we approach it in a purist way, then "advantage" would mean that there is a line in which White wins no matter what Black plays. This might be true, but it has not been proven. Perfect play might result in a win for White due to the extra tempo he has. Or in a draw due to Black's defensive resources. Or to a Black win as the extra tempo might turn into a Zugzwang. Or a draw due to White's resources to avoid that Zugzwang. A result-statistical approach would tell us that White has the advantage since the relative results statistically favor White. Or it can come from an analitical statistic approach, pointing out the ratio of good positions for each colors following semi-decent play. Or we can approach it socio-culturally as most of us favor White as a color. Or we can approach it from a sporting point of view. I am not sure at all that White has an advantage after the first move and I do not see real chances that this will be decided in the near future.

To conclude, we have no proof of evident advantages for White, but White still has a subjective advantage, the openings in many cases seek equality for Black. A revolution of more challenging Black openings might come into play due to strong analitical resources, we have to see that possibility in advance. So your reasoning is valid from a sporting, humane point of view, but a purist validation would render the answer uncertain. I do not have anything against the Champion having White at the first game precisely due to your points, but I see rationale in allowing the challenger having White in the first game of the playoff as well, as it will allow him to really challenge the Champion. It would remedy the Champion's favor and would create a psychological, but mathematically not necessarily existent advantage to the challenger for the playoff. This would give incentive for the champion to try to win the match before the playoff as well if the Champion is afraid to play an all-deciding game with Black. In that case, the conflict of interest would be total, as a draw would not quite be a perfect result for either player, but would be worse for the Challenger.

In short, I am not sure which color should be given to which player. I can certainly accept your approach or the opposite. Or randomity. The idea is to make sure we have a great match which is fun to watch, just and creates a conflict of interests, which will not prevent uneventful draws, but will reduce their chance and their frequency.
A7fecd1676b88 A7fecd1676b88 12/15/2016 06:36
@Jacob woge - yes, it is 42 indeed.
This then leads to the statement: "This isn't climate science, I have no grant money at risk, and can easily admit when I have made an error in a calculation."
Thank you for the correction.
Your correction does not, unfortunately for Mr. Ashley, change the original conclusion.
yesenadam yesenadam 12/14/2016 10:38
Very nice to hear from the great Maurice Ashley on this site. :-D
Nice to hear from the players, although to hear the question and their exact words would be nice. Maybe they would prefer playing 1 game and getting $1 million. hehe. In general, you make a lot of sense, thank you.
Having the rapid tie-breaks beforehand seems like "splitting hairs"?! Gee. The idea is (as I'm sure you know) that it would avoid 'playing for draws' in the classical section, as we saw in the match. And notoriously that candidates a few years ago, when people were just playing for a quick draw with white, putting their faith in rapid/blitz. I'm not sure why you can't see the huge difference this would make. At least, it would prevent the spectre of "drawing games to reach the rapid/blitz deciders". Wouldn't it?
One other point - from a couple of your numbered points, it seems that the true deciders of how things should be - are people who know nothing about chess. (i.e. the general public). That's my main problem with these 'improvements'. How many other sports are massively changed by considering what people who know nothing about them want? That would be silly. Oh..but it's all about money. Ensuring the last dollar is milked out of the game. And given to who? Sigh. Anyway. I think there's already more than enough 'monetization' of everything in the world without everything about chess being deformed/'transformed' in whatever way will maximize profits for investors, and maximize interest for people who don't play chess. Unless you are arguing that somehow these deformations coincide with what the chess world wants, from GMs to patzers and everyone in between. Also, it seems like you would really rather have all the games rapid, because that's what the public and everyone wants. All your points argue that way. Don't they? A 'solid time window' gee. Well, for me - a bit or a fair bit longer would be good. (Why must the match be so short? Why do people want to make it short? Don't these people making money off it, make more with a longer match?) Although with the blitz tie-breaks first, the debacle of 'playing for draws to get to the rapid' won't exist. Hmm but it may destroy most of the excitement of the blitz play-offs, as no-one will know at the time whether they're significant or not. But it will make for a better classical match.
Lastly, if a player really hated blitz playoffs, don't you think maybe they would be better off hiding that from their peers? Assuming that they are free to speak without wanting to show weak points to peers, or rudeness to organizers, seems a bit..unwarranted. But I don't know.
Thanks again! I dig your approach, of 'here's some evidence, here are my beliefs' rather than 'here's my crazy idea' (not to name names hehe), and you've affected my opinion on these matters. p.s. I have to disagree - some things are perfect. Just not voting systems, systems of government, WCC formats etc.
mdamien mdamien 12/14/2016 04:39
@fgkdjlkag

You ask if there is any other sport where the champion gets an automatic advantage. The wording itself belies a view of a match on equal terms, rather than a match between champion and challenger, since to say the champion "gets an automatic advantage" by virtue of holding the title suggests a view whereby he automatically gives up his title, in order to play for it anew on equal terms.

But to answer the question, yes boxing has a long history with the same format: you must defeat the current champion to become champion, and the champion retains his crown in case of a tie. John Sullivan was the last champion of bare-knuckle boxing, and the first of gloved boxing. He was defeated in 1892 by "Gentleman Jim" Corbett. That unbroken line continued with names you recognize, like Jack Johnson, Jack Dempsey, Gene Tunney ... right down through the likes of Joe Louis, Rocky Marciano, Floyd Patterson ... to Muhammad Ali, Spinks, Norton, Holmes ... to Mike Tyson, etc. Boxing has had its own "FIDE" problems with disputed crowns at times, including the present, but it is almost the perfect analogy to chess and its own history.

Let's drop the boxing world championship down to eight rounds. If there's no knock out and the fight is tie on the judges scorecards, we'll go to a "haymaker" tiebreak system: they just stand there and take turns receiving punches to the chin, until one is knocked out.
Resistance Resistance 12/14/2016 08:16
My ideal, my dreamed Classical World Chess Championship Match would consist of 30 Classical Chess games, and in case of the match ending in a tie, the Champion retains the title (i.e., no tiebreaks).

W h y ?
30 Classical Chess match games seem (to me) to be enough to prove which one of the players is the better player; the better player in that specific moment in time, at least.

You give enough room for even the most conservative of players to risk and not be afraid of losing one or two (miserable) games. (--Because you enter the match in a much more relaxed frame of mind; 30 games will allow the players to actually focus on the pure chess aspect of things: it will encourage them to be creative and combative, even when down in the score --it will encourage them to overcome a 1,2, 3 or even higher number of points deficit--, and it will not promote avoiding defeat at all cost from the very beginning, something a 12-game match is bound to encourage. A 30-game match will allow the players to play chess in the full sense of the word, and it will disincentive the player with an early lead in the match to to play for a draw the rest of the match, since it is difficult to maintain or guarantee such thing when there are so many games still ahead--). This, together with the absence of tiebreaks, would really stimulate the players to fight for that title of Classical World Chess Champion till the very end; it would really set the mood for bigger, grander fights! Epic fights. Epic World Chess Championship Matches...

Isn't that what we all want?

.
calvinamari calvinamari 12/14/2016 04:40
Here is the problem with this "survey" of top players. In countless examples of broadcast interviews with Maurice, the superGM being subject to his hyperventilating style invariable looks pained by the ordeal. The inevitable result of these interviews is that you always and unmistakably come to know full well what Maurice's views on the subject he brings up are, but either (a) the player barely gets a word in edgewise or (b) the player is vaguely complacent with whatever Maurice is spouting just to get the interview over with as promptly as possible. We have all seen this again and again. So I would not be the least bit surprised if this "survey" were more of the same. "Sure, Maurice. Right. Whatever you say. Sorry gotta go." The bloviating sportscaster style is a well established media type that Maurice tries to apply to chess. Maybe it has a place in some contexts where the goal is it introduce chess to a new audience, but I rarely have seen players warm to it.
Petrarlsen Petrarlsen 12/14/2016 03:36
@ fgkdjlkag :

"Why does the defending champion get an advantage?"

I don't know what are the arguments of PawnMates in favor of a certain degree of use of this system, but, as for me, I explained my own point of view in a previous post under this article ("12/13/2016 07:22"). (And I also gave a number of arguments on this same theme under the Sutovsky article, in the last pages of commentaries.)
123Next