Champions Showdown: Improving the format

Advertising

Books, boards, sets: Chess Niggemann

Macauley Peterson Macauley served as the Editor in Chief of ChessBase News from July 2017 to March 2020. He is the producer of The Full English Breakfast chess podcast, and was an Associate Producer of the 2016 feature documentary, Magnus.

Discuss

Rules for reader comments

lajosarpad 11/3/2018 11:37

@celeje

Grandmasters were not stating concerns indeed. It is possible that they do not have concerns, or they have, but they did not express it, or they have, they expressed it but we are not aware of it. In any case, our lack of knowledge of any explicit concern is much weaker than the knowledge of explicit unkoncern, but even if Carlsen, for instance would say he is not concerned, this would not mean much, unless he would add that he thought a lot of it. And even in the case when he thought a lot of it, if he did not use a powerful methodology I would not be sure his explicit lack of concern would be strong enough to shift the null hypothesis. However, we do not know about a concerning statement pro or contra at this moment. I think the null hypothesis should be that chess960 as it is is playable, but since there are quite plausible suggestions for improvement, I think time allocated into thinking and debating them would be a time well spent.

"If there are lots of games played with enough variety in them, the horizon affect should be less of a problem, because the computers are actually playing moves, so the horizon keeps on advancing."

Let's say they see 20 plies ahead. If there is a move they discard in a given position which would be better much later, but seems to be less than satisfactory using the 20 plies, then the move will be discarded. And if something else will be played, then the discarded (better) move will not be analysed as the game goes on. The horizon effect is lowered only in the case when a discarded idea is still possible after the move is being made. If the opportunity is lost, then the move will not be found by the computer at all. So, the most serious cases of the horizon effect will not be solved by the engines continuing to analyse the game as it advances, nor by an increased number of analysis. Humans have a shorter horizon and due to this a lot of opening variations were discarded for decades, until computers with their better horizons were able to discard the human evaluation.

To reflect on your third answer to Petrarlsen: Let's have a thought experiment of a starting position where White has a mate in 1. If there were no other starting positions, then this would be a lot different and unique in comparison to the others. Yet, the position would be instantly discarded. Now, we know that there are no starting positions with the exact defect specified in the thought experiment, but this does not exclude the possibility of other serious defects. A position might be very unique and different from the others and having a very serious defect at the same time. My point is that the uniqueness of a position is not a decisive argument to keep it.

celeje 11/3/2018 10:18

@ Petrarlsen: (3)

Petrarlsen: "Because to forbid the Berlin, for example, would be to amputate the traditional chess' universe; with Chess960, you have 960 different universes; even if we were to keep only 300 of them, this would still be 300 times more than in traditional chess; I rather think this would be sufficient! And, furthermore, it must be noted that these 300 universes wouldn't be "amputated universes"; each of them would be available in its entirety (as opposed with what would be traditional chess without the Berlin defense)... "

Again, the problem is not the NUMBER of starting positions. It's that every starting position is unique, with subtle differences that are themselves of interest. No one thought: "We need 960 starting positions." That's just the number that it NATURALLY turned out to be with the rules proposed.

Just calling the different starting positions "different universes" does not change anything. If you want to call them "different universes", then you are still artificially amputating universes in the chess960 multiverse. But really, it's the same rules, so they are just different INITIAL CONDITIONS of the one theoretical universe.

celeje 11/3/2018 10:11

@ lajosarpad: (2)

celeje: "These computer-computer games won't suffer the horizon effect any more than human-human games."

lajosarpad: "It is premature to assume this. If both engines discard a better continuation, then better lines are not found. GM help is therefore useful. "

If there are lots of games played with enough variety in them, the horizon affect should be less of a problem, because the computers are actually playing moves, so the horizon keeps on advancing. We would need to make sure there was enough opening variety, though, as I said before.

But of course ideally we'll have more than just computer-computer games, which just supports my argument for more real competitive high-level human GM chess960 games.

celeje 11/2/2018 10:54

@lajosarpad:

lajosarpad: "I do not think tournament organizers are considering their tournaments to be experiments, so they will not suggest improvements and their tolerance of unique suggestions is probably limited, so collective."

I think tournament organizers consider the tournament conditions as experimental (e.g. whether paired, TC, how early positions are known, etc.), but not the game itself. That does not mean the games themselves cannot be used as experimental data by others. Of course that's only experimental data of chess960 & at the TC chosen.

celeje: "I've never seen or heard any GM claim that one of the 960 starting positions is too favorable to White or Black."

lajosarpad: "Because there was no such analysis done yet."

Yes, but none of them seem at all concerned that it might be the case. Not Carlsen or Kasparov or anyone that I've heard when they've been interviewed. That does not prove anything, but doesn't it suggest what should be the default hypothesis?

lajosarpad 11/2/2018 04:33

@celeje

The suggestions for change aim to improve the big picture, that is, the logical coherence, the fairness of the game, the playability of the game, these and more are all important aspects.

I do not think tournament organizers are considering their tournaments to be experiments, so they will not suggest improvements and their tolerance of unique suggestions is probably limited, so collective.

"I've never seen or heard any GM claim that one of the 960 starting positions is too favorable to White or Black."

Because there was no such analysis done yet.

"These computer-computer games won't suffer the horizon effect any more than human-human games."

It is premature to assume this. If both engines discard a better continuation, then better lines are not found. GM help is therefore useful.

"It's not a perfect analogy (no analogy is perfect), but not as bad as you suggest. Openings are players' choices only because that's what the current rules allow. If the rules didn't allow some openings, people would not think the opening choices were "limited", because those would be the rules they'd be familiar with. "

True, but still, the randomity is different if it is a choice in comparison when it is not a choice.

celeje 11/2/2018 10:56

@ lajosarpad: (3)

lajosarpad: "Opening options are players' choices, while the starting position in chess960 is not their choice. It is very different to limit the choice of the players from limiting the possible starting positions."

It's not a perfect analogy (no analogy is perfect), but not as bad as you suggest. Openings are players' choices only because that's what the current rules allow. If the rules didn't allow some openings, people would not think the opening choices were "limited", because those would be the rules they'd be familiar with.
Even with these rules, a player cannot play any opening he wants, because it depends on his opponent. It also depends on what color he has.

More importantly, historically there may have been something like opening rules. We'd have to go back to Murray and other books on chess history to check this. I think one historical source talked about a move the player "may" make on the opening move. It was not clear exactly what was meant, but it could have been a special move allowed only if you did it with your first move. This is a little different from what we're talking about, because it's a rule that allows something extra, not a rule that forbids something, in the opening, but it's along the same lines.

To us, this probably looks rather unnatural. But that's what the limiting of starting positions looks like to me. Very unnatural and ugly. You don't have such unnatural limitations on starting positions in chess960 or chess480.

celeje 11/1/2018 04:04

@ lajosarpad: (2)

lajosarpad: "And if the grandmasters say that a starting position is better for White, for example, but the engines do not confirm this, maybe due to the horizon effect, then why shouldn't we discuss the possibility to exclude the position? "

I've never seen or heard any GM claim that one of the 960 starting positions is too favorable to White or Black. I've never seen or heard any GM worry that might be the case either. Some people have tried to suggest a position that's like that & have not been convincing at all.

I don't agree with your worry about the horizon effect either, because both Petrarlsen & I wanted the computers not just to sit analyzing from the start position but to actually play out many games. These computer-computer games won't suffer the horizon effect any more than human-human games.

But I agree with an earlier comment that if all positions are theoretical draws some may be much harder to obtain than others. There human games will help determine practical difficulty for humans. In practice, if a theoretically drawn starting position requires play of quite high quality by Black & another starting position gives mediocre play the draw, that will show up when many human games have been played.

celeje 11/1/2018 12:57

@ lajosarpad:

I'll read & reply properly later, but first...

lajosarpad: "You perfectly know that finding the big truth in a position is next to impossible, so waiting to solve the positions in order to determine whether they are good positions is actually doing nothing."

I've been meaning to reply to a few of your comments that seem to be along those lines. There's a lot in between knowing nothing and just speculating with no evidence and complete mathematical proof. Just because we will not have complete mathematical proof doesn't mean we have to go to the opposite end of the scale.

The more "experimental" data you have, the more the error bars go down. The elo calculators you can download already conveniently calculate tournament performances in elo, including error bars. It is not true that we are just delaying for no reason when we collect more games.

But we also must keep in mind there are different questions we are discussing. The above will help for things like the "fairness of opening advantage" question. They will not help for arguments about aesthetics of a castling rule.

lajosarpad 11/1/2018 12:06

@Celeje

Petrarlsen already pointed out that if some positions can be better in comparison with others, the other positions can be worse compared to the better position. You perfectly know that finding the big truth in a position is next to impossible, so waiting to solve the positions in order to determine whether they are good positions is actually doing nothing. Let me give you an example: let's consider the case when we know that in an arbitrary chess position we know that White has a long term advantage due to some positional motif, which is outside the grasp of the engine, but the position is not solved. In that case it is safe to assume that White has the advantage, until we find the exact equalizing line for Black. Grandmasters tend to conclude in such situations that White has the advantage, yet, they did not play thousands of games in the position and the position cannot be found in the database as a frequently played position. It is not premature to conclude that the position is not balanced, because according to our current knowledge, White has the advantage. Can we accept as a quasi fact that the grandmasters in this case are right, or are all the grandmasters wrong? They do not know the final truth, the position is not solved. If we do not consider the grandmasters to be wrong in the case when they can give us very plausible reasons why White is better than Black in the given position and we can reach a compromise that according to our current knowledge they seem to be right, then why shouldn't we use this approach for chess960 starting positions? And if the grandmasters say that a starting position is better for White, for example, but the engines do not confirm this, maybe due to the horizon effect, then why shouldn't we discuss the possibility to exclude the position?

"I think our disagreement is maybe due to you being totally convinced that one castling rule is "bad" and there is some "better" castling rule. I think that is false."

I think it is better to formulate the original opinion in a different way: I think Petrarlsen thinks a possible rule change would improve the rule of castling because the logic behind the current rule cannot be known and it seems to be counter-intuitive.

Opening options are players' choices, while the starting position in chess960 is not their choice. It is very different to limit the choice of the players from limiting the possible starting positions.

Rule adjustments are comparable to revised editions of books, correcting some mistakes. In our case, the "book" was already "published" and you seem to advocate to not fix what seems to be wrong and do a second, revised publishing, but to wait and see factually that the item which seemed to be wrong is indeed wrong. In book publishing this might work, but in the case of chess960 this would, in practice mean that no changes will be done at all.

lajosarpad 11/1/2018 11:51

@Petrarlsen

"I must say that I don't really fully understand what you mean. Perhaps I am missing something very obvious! But could you elaborate a little, about this point?"

Indeed, I simply forgot to answer. Let's consider the case when all the pieces which blocks castling already moved, but in order to castle, according to your suggestion, one would need to make more than one moves, in order to reach a position where castling is possible, let's say one needs to make two moves to castle. I was thinking about making castling a possibility right away, by moving the king first and then castle. What if the rules were allowing the player to castle right away, but since two moves were played at once, we would give the other player the possibility to pass a move when he/she finds it feasible. Let's consider the example when White makes the castle right away, but it would take two moves as a minimum and play continues on, but the game reaches a position of mutual zugzwang where Black is to move. Since White made two moves, Black would be able to pass on the move to White and as a result White would be in zugzwang. The possibility to get out of zugzwang without a move seems to be a feasible compensation for me. This would speed up the process of castling in these cases, but the other side will have a compensation for the loss of tempo, namely, the possibility to lose a tempo when needed.

I also have the impression that we agree on the main point about the possibility that chess960's rules might be logical, but I was focusing on the scientific side and you were focusing on game play.

"I think that the first point is that, when we use the word "theory", we don't mean the same thing. When I use it about chess openings, I refer to the opening lines that have been analyzed in depth by grandmasters with (nowadays) the aid of powerful computer programs. "

I think you have meant opening theory here, as endgame theory, for example is very different. In terms of opening theory we have the very same concept for chess, that is, memorizing opening lines and understanding the idea behind the lines. However, my point was that in the case of chess960, opening theory will be different and will resemble more to endgame theory.

"So, when I say that Chess960 is a "non-theoretical game", I just mean that, in Chess960, there couldn't be long prepared opening lines, as in traditional chess. And I think that this difference gives globally a very different game."

Actually opening lines can be prepared and will be prepared, but due to the limitations of the human mind, full repertoires will be impossible for the human mind to remember.

"This is why I don't think that to include Position 518 is a good thing"

I can understand this point of view, however, on the long term if one will play chess960 and not play chess a lot, the person will not necessarily have an opening repertoire for the starting position of chess. Nowadays chess960 players are mainly chess players, but there is no guarantee this will be true in the future as well.

celeje 11/1/2018 10:44

@ Petrarlsen: (2)

Petrarlsen: "Another example that comes to my mind: If I were to write a book on a law theme (....), very probably, at one point, I would hesitate between several (2 or 3, probably) general plans, for the book. etc."

I see no analogy to writing a law book. When authors write a book, they go through many drafts, and they may even get feedback on chapters, but they complete the book and when they're happy they send the finished book to their publisher and that's it.

A less-bad analogy is writing an opening chapter of a fictional work. The opening chapter maybe gives the setting and introduces the main characters, but the author won't know what is going to happen to the characters, etc. But that's still a bad analogy, because there is nothing like playing and finding out more through that playing in the analogy.

If there are clear factual errors in the law book, typos, etc., then, yep, lots of people reading it may give feedback for its improvement, but that's about it. It just does not convince me as an analogy.

celeje 11/1/2018 01:13

@ Petrarlsen:

Petrarlsen: "So I think that it isn't logical to admit, on the one hand, that some positions can be much better than others, and, on the other hand, to consider that, for years and years, Chess960 should continue to be played as it is, without trying to do anything to determine if some positions would be much better than others..."

No, you don't seem to be understanding me. Without solving completely every single position with every single castling rule, we cannot PROVE that one starting position is not any "better" than another. In that sense, "it is possible" that some starting positions, including the traditional one, are "worse" than others. We can HYPOTHESIZE that all starting positions are UNIQUELY VALUABLE and that no starting position is really "better" than another. That is IMO the sensible hypothesis to make. Each position is not really "better" or "worse", but different and special, so we should not want to lose it.

I have already stated some objective criteria we might want for being a "worse" starting position. Too much opening advantage for White or Black (but even "too much" may be hard for everyone to agree on). Or a certain dull draw. etc. I think those are very, very unlikely, but we don't have scientific evidence that they don't exist. We can only find out if such "bad" starting positions exist by "testing", etc. i.e. playing lots of games, etc.

Your "without trying to do anything" is odd. Having games played, having an "experiment" not terminated without much data, etc. IS "doing something".

I think our disagreement is maybe due to you being totally convinced that one castling rule is "bad" and there is some "better" castling rule. I think that is false. I should therefore return to explaining why.

Just cutting out opening positions is not a real solution, any more than banning the Berlin in tradtional chess is a solution to what some people claim is a problem.
Chess480, Chess480bg do not arbitrarily cut out any position. That is good.

Petrarlsen 10/31/2018 05:36

@ celeje (2/2):

- I find it interesting to take two passages out of your comments together:

* "(The proposers) do not know that the positions they want to cut out are not much more interesting, much more deep, etc. than those they want to keep, including the traditional starting position."(in your post from yesterday)

* "The bad thing about new proposals is they are so much about arbitrarily and clumsily forbidding certain starting positions." (in your last post)

In the first passage, you admit that it is possible that some starting positions can be much better than some others (if some positions can be "much more interesting, much more deep, etc." than some others, it is obviously that they can be much better), and, necessarily, if some positions can be much better, it also implies that other positions can clearly be not at all as good.

And, in the second passage, the main criticism you make about new proposals is that they intend to suppress some starting positions.

But, if it would transpire that some starting positions were clearly inferior, compared to others, it precisely stands to reason that these starting positions should be discarded; why would it be a good thing to keep subpar positions, when there are so many starting positions?

And, as for a lawyer who compares different plans for a book he is writing without having to test anything in practice, it would be possible to compare the different starting positions and to try to assess them, so as to define, if possible, which are better and which are inferior, and then to eliminate those last ones.

So I think that it isn't logical to admit, on the one hand, that some positions can be much better than others, and, on the other hand, to consider that, for years and years, Chess960 should continue to be played as it is, without trying to do anything to determine if some positions would be much better than others, and, if it would transpire that, indeed, some positions would appear to be much better than others, to discard the inferior positions. If it would appear, after trying to assess the starting positions, that it isn't possible to draw reliables conclusions about the starting positions, then, well and good; all the positions should be kept (at least for the moment), but if, on the opposite, some clear conclusions could be drawn, then, there would be no reason to continue to play this game for years and years with some inferior starting positions.

Petrarlsen 10/31/2018 05:27

@ celeje (1/2):

- "What matters is that each starting position is unique and worthy on its own of study and play."

As an example, if Carlsen was to play Caruana with, say, Position 325, and that it would appear that Position 325 would have been discarded (due to the implementation of some changes in Chess960), and that they will finally play with Position 856 (I haven't checked the positions; they are just random positions), will it really be an Irreparable Loss for Humankind? And will really Carlsen and Caruana be So (not Wesley!) Sad with this Tragic Situation that they wil inevitably play a completely dull and lifeless and uninteresting game?

...What will really happen will be that, in fact, there will be as much chances for this game to become a memorable and exceptional game with Position 856 (supposing that it is a good starting position as I didn't check...) than with Position 325. This because, one more time, a starting position is precisely not more than that: a Starting Position, i.e. the starting point for the game; from this point, it is up to the players to create masterpieces, and they can do it with every good starting position.

I think that your comparison with openings in traditional chess like the Berlin is inadequate for two reasons: 1) Because, precisely, an opening isn't a starting position; if you discard some openings, you intervene on the game itself, limiting it. If you exchange a starting position for another starting position, this will not prevent the game from being played completely normally, without any limitations. 2) Because to forbid the Berlin, for example, would be to amputate the traditional chess' universe; with Chess960, you have 960 different universes; even if we were to keep only 300 of them, this would still be 300 times more than in traditional chess; I rather think this would be sufficient! And, furthermore, it must be noted that these 300 universes wouldn't be "amputated universes"; each of them would be available in its entirety (as opposed with what would be traditional chess without the Berlin defense)...

And, one more time, the main idea is not to suppress haphazardly some starting positions, but to evaluate carefully all the situations, and to possibly discard some starting positions only after having meticulously evaluated the advantages and disadvantages of each type or category of starting positions. So, in this context, the discarded starting positions would be supposed to be less interesting or satisfying (for precise reasons) than the others.

Petrarlsen 10/31/2018 05:25

@ lajosarpad (2/2):

- About Position 518:

I think that the first point is that, when we use the word "theory", we don't mean the same thing. When I use it about chess openings, I refer to the opening lines that have been analyzed in depth by grandmasters with (nowadays) the aid of powerful computer programs.

For example, in traditional chess, when a top-player, playing against another top-player, plays 15 moves of well-known theory, and that, at this moment, he uncorks a powerful novelty that completely changes the evaluation of the position, it is something very specific that would not be found in this form in Chess960. And, for me, this isn't either good or bad; it is just simply (in my opinion) the most defining characteristic of Chess960: that such type of situations couldn't arise in this form for this game. This without having for a consequence that Chess960 would be better than traditional chess, nor the opposite.

So, when I say that Chess960 is a "non-theoretical game", I just mean that, in Chess960, there couldn't be long prepared opening lines, as in traditional chess. And I think that this difference gives globally a very different game.

This is why I don't think that to include Position 518 is a good thing; it tends to bring Chess960 closer to traditional chess, and, in my opinion, it stands to reason that, precisely, the interest of Chess960 is that it is different from traditional chess while being easy to play for a chess player. (If Chess960 was globally very similar to traditional chess, what would be the interest to have these two games?) In my opinion, this constitutes at least a possible drawback for the use of Position 518, and I don't see in which way it could be interesting to include Position 518 - so it seems to me that the balance between the advantages and disavantages is clearly in favor of discarding Position 518.

- "(...) careful planning is needed (note: for devising the rules of such a game as Chess960.). This was done by Fischer initially and he came up with some rules. But those rules are not carved into stone. We aim to achieve a state of a second phase of planning. Celeje aims to wait to see the results of the original plan. I do not agree with him, since adjustments could be done. I'm not saying I know what adjustments are needed, but determining the possibilities and debating them seems like a feasible approach."

I fully agree!!

- "If one wants to stick with the current rules for castling, for example in chess960, then it would be very good to actually know why the given person considers the current rules to be absolutely logical."

Yes; this is the heart of the problem. And, precisely, the Chess960's proponents NEVER explain why these rules would be logical; they don't even seem to be interested in this: "the rules are what they are and we must play the game as it is without changing anything". And yet, it would be interesting to compare the logic of possible new proposals (as, for example, my Chess480BG and Chess480M) with the logic of Chess960, but no-one never proposes any possible logic to explain the Chess960's castling.

As for me, my impression is that the only logic of the present Chess960's castling is to keep at all costs the traditional chess' castling arrival positions for all the 960 starting positions. But this seems to me clearly much inferior to what could give a well-thought-out solution, finely tuned to be as well-suited as possible to the specific characteristics of Chess960.

Petrarlsen 10/31/2018 05:23

@ lajosarpad (1/2):

- You forget to answer my question about an idea you had about Chess480M (or you missed my question...).

This was my question: "When you say: "(...) we might consider adding that the whole process could be a single move, but the opponent could pass as many times as many additional moves were needed for castling by the incriminated rook and king.", I must say that I don't really fully understand what you mean. Perhaps I am missing something very obvious! But could you elaborate a little, about this point?"

I repost this, as I would quite like to understand better what was your idea!

- About the question whether there is or not a logic in the Chess960's present castling, taking into account your last posts, I think that we approach this differently, but that, in fact, we agree. In any case, I agree with what you wrote about this in your posts of today!

celeje 10/31/2018 11:58

@ Petrarlsen:

Petrarlsen: "To have a little more starting positions or a little less is only a matter of detail; it doesn't really change anything, provided that there still remains an important number of starting positions."

It's not the exact number that matters. It does not have to be 960 or 480 or anything. What matters is that each starting position is unique and worthy on its own of study and play. A small change in the starting position changes everything re. the possible games, geometry, etc. This is what Kasparov said after he and Svidler accidentally looked at the wrong starting position.

The bad thing about new proposals is they are so much about arbitrarily and clumsily forbidding certain starting positions. That is a bit like trying to forbid certain openings in traditional chess. If we just make up a rule that no one is allowed to play the Berlin, that is not an improvement to traditional chess, even if people don't like the Berlin.

The Berlin example is not a realistic risk of happening, but centuries ago there probably was a time when it was thought dishonorable not to play the King's Gambit Accepted. Playing only the King's Gambit Accepted is not an improvement to traditional chess.

The good thing about chess480, chess480bg is that they are not fixated on cutting out starting positions. They don't cut out any starting positions.

lajosarpad 10/31/2018 11:58

Experimenting is possible through thought experiments as well. If we do not have many games, we are making thought experiments for all the relevant and representative cases. If one wants to stick with the current rules for castling, for example in chess960, then it would be very good to actually know why the given person considers the current rules to be absolutely logical.

lajosarpad 10/31/2018 11:56

"So, globally, I maintain that the best possible approach would be to try to select the best rules, and if the best rules are slightly different from the current Chess960 rules, to try to make organizers and players adopt these rules rather than the current Chess960 rules."

Agreed. I think this is the most important starting point we should agree in order to have the very possibility of improvement.

lajosarpad 10/31/2018 11:55

@Petrarlsen

"So, if I understand well, you mean that, when someone devises something, he ALWAYS has ONE and ONLY ONE possibility in his mind??"

I think he/she meant that creating is the first step. Choosing from candidates is not excluded from the creation. However, I quite agree with you that Celeje's approach is problematic. Imagine the case when someone has to create a space ship. Creating it first, before thorough planning would be less than satisfactory. It might work, but not necessarily in the ideal way. In the case of chess960 there is no danger of death, but nevertheless, careful planning is needed. This was done by Fischer initially and he came up with some rules. But those rules are not carved into stone. We aim to achieve a state of a second phase of planning. Celeje aims to wait to see the results of the original plan. I do not agree with him, since adjustments could be done. I'm not saying I know what adjustments are needed, but determining the possibilities and debating them seems like a feasible approach. We do not need thousands of years to implement the changes. We would be deprived of some of the excitement we could have in the meantime.

People will not memorize the first fifteen moves, as you accurately pointed out, but they will memorize concepts. If, by memorizing a concept and knowing which are the criteria of its applicability, then, if the criteria are met I will be able to find most of the correct moves, if not all. If several such concepts will be remembered, then one will have an opening repertoire, so opening preparation will not be gone at all. It will change and resemble more to ending preparation.

"To elaborate a little more, what I meant was that, yes, it is absolutely necessary to have a scientific approach to be able to affirm that a given castling system is logical and coherent or not, but that, in my opinion, it isn't at all absolutely necessary to have a scientific approach to discard one or the other type of castling; for example, if it is too complicated to determine if one type of castling is good or bad, the simplest solution is quite simply to try to devise a system for which it is simpler to determine if it is good or bad..."

I did not react to this in my previous comment. I am reading the comments chronologically and write my reactions in the meantime, then, when the character limit is filled, send a comment. I understand your point, but would like to improve it by saying that if we factually know that we do not know something, then it is a scientific fact. And since we do not know what is the hidden logic if any in all the positions in the context of castling, we need to work with what we have. We do not have knowledge of a hidden logic, but if we provide some suggestions where the logic is not hidden at all, then the suggestions seem to improve the rules. And the fact that this "seems to be an improvement" means that according to our current knowledge we have a lot of reasons to implement the improvement.

"You affirm that it isn't possible to choose between 2 or 3 possibilities without testing them: if this is true, there are absolutely NO reasons for this to be different for the inventor than for another person. And thus, like anyone else, the inventor MUST test everything before choosing one given possibility, following your theories. Or else, if it is intellectually possible for the inventor to make a choice without any tests, it it also necessarily possible to do so for another person. There is no conceivable reason for this to be per se different between the inventor and another person. "

Excellent point!

lajosarpad 10/31/2018 11:33

@celeje

"But that argument assumes we agree that Y is an improvement over X. We do not agree on that. "
We cannot agree on that if we postpone the discussion to an arbitrary point of time when we have "enough" information. My point is that it is good to debate possible improvements with the aim of reaching a consensus. I do not have a concrete set of improvements which must be implemented. But we already have enough information to debate what the suggestions for improvements would yield.

"No, you create first. Then you test."

The very first step is to think what should be created. Then the second step is to plan what you create. Before a thorough conceptual analysis and thought experiments it is unwise to create.

@Petrarlsen

"Yes, I agree, but I don't find very interesting to make complicated researches to know if something haphazard shouldn't be by pure chance something logical!"

I think I did not formulate my point clearly. I do not think we should allocate time to research whether there is such a logic by accident before we make any changes. It would be unfeasible. But, as a thought experiment, if we found such a logic, then we would not necessarily need changes. If we implement changes and after that we find such a logic in the original rules, then we need to compare the original rules with the changed rules in view of the knowledge we found. Until we find such a knowledge, the null hypothesis is that there is no such logic and changes are needed. So, according to our current knowledge, changes seem to be needed, but in view of the truth, which we do not know, the changes might not be improving the game, even though it's better to change something which seems to be illogical to something which seems to be logical than to keep the seemingly illogical state and wait to see the logic.

"I don't remember exactly what I said, but I don't think I said that this logic is missing;"

You've pointed out a part of my reasoning which seemed to be a bad approach for you, while my reasoning was just not assuming that there is no such logic. Probably a misunderstanding. To save time I did not search the exact quotes, but if you would like to debate this part, I would search and find them. In this moment this seems to be unnecessary, since we agree on the most important point, that is, while we do not see any hidden logic, changes are not premature. When I said that "It would be unscientific.", I meant the exclusion of the possibility of the existence of a hidden logic would be unscientific. If we do not exclude the possibility, we can still implement changes. So we need to separate the concepts: the possible existence of a hidden logic is a scientific question, the system to be used is a legal system which needs to be the result of a scientific analysis. This is what you do with the suggestions about castling. It is true that you do not attempt to find the final truth, as you are clearly not capable of achieving it (no offense), but nonetheless, the philosophical and logical coherence you aim to achieve requires a scientific analysis. I am sure you agree, the very fact that you are giving scientific arguments about king safety proves it. However, again, I do think that while we are not aware of the final truth, we need to approach it with the means we already have.

We approach chess960 differently. For me it is a game which will have its own theory and players will adapt their preparations to the game the best way they can.

"I would rather feel as if, eating in an Indian restaurant, I would finally find Japanese cuisine in my plate;"

I understand your point and find it plausible, but it is subjective. Just like my approach to chess960 :)

Petrarlsen 10/30/2018 04:45

@ celeje:

About your last post:

In my opinion, the only thing which is really important is to try to have the best possible game.

To have a little more starting positions or a little less is only a matter of detail; it doesn't really change anything, provided that there still remains an important number of starting positions.

And if a potentially interesting starting position would be wrongly discarded, there would still remain several hundreds of interesting starting positions, so this wouldn't at all prevent the game from being an interesting game, and this is the only really important thing, in the end.

On the basis of a good starting position, what really makes the interest of a top-level game is the imagination and creativity of the players, and they would use there imagination and there creativity as well with 300 starting positions as with 480 or 960 (or 1500, etc.) starting positions. And Chess960's general idea would still apply, because, even with more or less 300 starting positions, the limits of human memory being what they are, no-one could memorize 15 (and even less 30) moves of theory.

I think that this is an important point: what makes mainly the quality and the interest of a chess game resides in the players and NOT in the starting positions; a starting position is, precisely, only a starting point. If the starting point is bad, it will mar the game, but any good starting position will have the potential to be the starting point for an immortal game, if the players are up to the task. (And no starting position, whichever it is, will give an interesting game if the players are bad players.)

(And, furthermore, if you take my two proposals, the modified Chess480 (Chess480M - for modified - to have a name for it) and Chess480BG, one of them, Chess480M, doesn't discard any starting position.)

So, globally, I maintain that the best possible approach would be to try to select the best rules, and if the best rules are slightly different from the current Chess960 rules, to try to make organizers and players adopt these rules rather than the current Chess960 rules.

celeje 10/30/2018 03:01

@ Petrarlsen:

Petrarlsen: "You affirm that it isn't possible to choose between 2 or 3 possibilities without testing them."

I'm not really thinking of the "testing" (playing) as "choosing between" different rules. I'm thinking of it as validating the hypothesized fairness between White and Black of every starting position and the hypothesized interest of every starting position.
My guess is that's true for chess960, chess480, etc.
A starting position would be "bad" if it gives too much advantage to one side (but how much is "too much" is debatable too). The starting position can be bad in other ways too. e.g. it can be bad if it almost always leads to dull draws. (I think that's even more unlikely than too much advantage, but it's possible).

The issue is that different proposals seem intent on cutting out starting positions. The proposers do not know that the starting positions they want to cut out are not very interesting, very deep, etc. They do not know that the positions they want to cut out are not much more interesting, much more deep, etc. than those they want to keep, including the traditional starting position.

If anything, we can ask: why stop at 960? But I accept that e.g. having 4 white-squared bishops on the board is too "strange" and not "pleasing".

Petrarlsen 10/29/2018 01:29

@ celeje:

- "Maybe though, you don't mean Fischer so much as any person formulating a game..."

Yes, more or less; what I mean is that it isn't possible to exclude that Fischer envisaged several possibilites, for the Chess960's castling. And if it was possible for him to choose, we can also choose.

- "As for the "ONE and ONLY ONE" part, what MIGHT be a LITTLE relevant is whether AT THE END (when he's ready to promote the game publicly) he was satisfied with ONE and ONLY ONE, or whether AT THE END he was still thinking of SEVERAL, perhaps none of which was totally satisfying."

But this doesn't change anything at all!!

You affirm that it isn't possible to choose between 2 or 3 possibilities without testing them: if this is true, there are absolutely NO reasons for this to be different for the inventor than for another person. And thus, like anyone else, the inventor MUST test everything before choosing one given possibility, following your theories. Or else, if it is intelectually possible for the inventor to make a choice without any tests, it it also necessarily possible to do so for another person. There is no conceivable reason for this to be per se different between the inventor and another person.

celeje 10/28/2018 09:27

@Petrarlsen:

Petrarlsen: "So, if I understand well, you mean that, when someone devises something, he ALWAYS has ONE and ONLY ONE possibility in his mind?? "

That's not what I wrote at all. I wrote that it's odd that you repeatedly blamed people liking chess960 on them revering Fischer and then you turned around and came up with some argument that starts with speculating about Fischer's thinking.

It might be very interesting to hear from historical figures who knew Fischer if he told them anything about his thought processes. But speculating about what he was thinking is not.

Maybe though, you don't mean Fischer so much as any person formulating a game...

As for the "ONE and ONLY ONE" part, what MIGHT be a LITTLE relevant is whether AT THE END (when he's ready to promote the game publicly) he was satisfied with ONE and ONLY ONE, or whether AT THE END he was still thinking of SEVERAL, perhaps none of which was totally satisfying.

Petrarlsen 10/28/2018 07:57

@ celeje:

Another example that comes to my mind: If I were to write a book on a law theme (...I didn't for the moment, but, as a lawyer, I know to what it amounts to write a law book...), very probably, at one point, I would hesitate between several (2 or 3, probably) general plans, for the book.

As a law book (as many other books...) is supposed to teach something, to enable the reader to have access to a certain number of informations, the only means to "test" these plans would be to write entirely 3 (for example...) versions of this book, with 3 different plans, and to select a certain number of readers (for example students or other lawyers who wouldn't know very well this given theme, for the test to be significant), and to compare what they would think of the 3 different plans.

Following your theories, it wouldn't be possible to do anything else; if I would have 3 different possible plans, I wouldn't have ANY means to know if one can be better than the other without testing them all.

But with such an approach, it would take a whole life (...or perhaps even "several" lives!...) to write even a small book!!! (..as, obviously, for each little detail, it would be necessary to use the same approach...)

If this approach suits you, it is all very well, but I will leave it to you for your exclusive use!!

Petrarlsen 10/28/2018 07:10

@ lajosarpad:

In my first post of today, I wrote: "In my opinion, for this precisely, it isn't necessary to have a scientific approach (...)", about the comparisons between different possible sorts of castlings.

To elaborate a little more, what I meant was that, yes, it is absolutely necessary to have a scientific approach to be able to affirm that a given castling system is logical and coherent or not, but that, in my opinion, it isn't at all absolutely necessary to have a scientific approach to discard one or the other type of castling; for example, if it is too complicated to determine if one type of castling is good or bad, the simplest solution is quite simply to try to devise a system for which it is simpler to determine if it is good or bad...

Petrarlsen 10/28/2018 06:03

@ lajosarpad: When I use the phrase, about Chess960, "non-theoretical chess", I quite agree that things are more complicated than this, and that, if Chess960 would really become a successful game with many top-level tournaments played with classical time controls, some sort of a theory would necessarily appear.

But what seems to me certain (because of the limits of human memory!) is that there couldn't be 15 (or more) moves of pure theory in a Chess960 game (apart, obviously, from Position 518!), so, in my opinion, the general idea of a "theoretical game" (traditional chess) opposed to a "non-theoretical game" (Chess960) is nonetheless globally true...

Petrarlsen 10/28/2018 05:50

@ celeje:

- "(...) Now you speculate on Fischer's thinking to justify not "testing everything extensively"."

Yes! And why not???!!!???

- "No, you create first. Then you test."

So, if I understand well, you mean that, when someone devises something, he ALWAYS has ONE and ONLY ONE possibility in his mind??

I don't know how your mind works, but it musn't work in exactly the same way than most people!

Most of the time, when someone devises something complex, he envisages several possibilities, and chooses the best one, after careful considerations. Or, for example, he can choose 2 or 3 possibilities out of 6 or 7, and proceed to test them (without testing all the 6 or 7 original possibilities, as, very frequently, several of them will quickly appear to be inferior).

If you insist, I will search for some examples, but this seems SO obvious, that I don't see the point to illustrate my meaning with examples, as, in my opinion, more or less every person who devises complex things uses, at least part of the time, this approach... For example, as a lawyer, when I devise a complex contractual system, very frequently, several possibilities appear to me for some details (...and, in this sort of things, even the smallest details can in fact be extremely important...); I really wonder how I would be supposed to test them all before making a choice!!!! This would probably be possible in "Alice in Wonderland", but not in real life!

Petrarlsen 10/28/2018 05:50

@ lajosarpad (2/2):

- "However, I never accepted the original reasoning, which included that the game gets too theoretical. So, if I never accepted that reasoning, why should I accept your conclusion taken from this premise?"

I don't accept at all either the idea that traditional chess would be too theoretical. But, for me, the main interest of Chess960 is nonetheless to have a "non-theoretical chess" (which isn't the case for Position 518); not because traditional chess would be too theoretical, but because, with Chess960, we would have precisely a "theoretical chess" (traditional chess) and a "non-theoretical chess" (Chess960), this while Chess960 can be played quickly by anyone who already plays chess, this allowing to have two quite different forms of the same game.

(To illustrate this with an example, for me, if I would follow a top-level Chess960 tournament - which I would do without any doubts if the rules were improved -, and find Carlsen playing against Caruana in a Ruy Lopez because the starting position would have been Position 518, I would rather feel as if, eating in an Indian restaurant, I would finally find Japanese cuisine in my plate; even if the Japanese cuisine would be REALLY top-notch, I wouldn't be very pleased!; I wouldn't have choosen an Indian restaurant if I wanted Japanese cuisine!! "There is a time and a place for everything", so to say!)

For me, this is for these reasons that I am interested in Chess960. And I don't see quite well what would be the interest of Chess960, apart from this theoretical aspect? What do you think about this?

- "So far I was the only one who pointed out that this is a logical fallacy in this discussion. Celeje unsurprisingly did not consider his/her argument to be a Nirvana fallacy, but you did not react to this point, which was a key argument in my opinion."

As I generally agree with most of your points, when I don't react to one of your points, most of the time, it means that I agree with it! (possibly with some slight nuances, but not more than that...)

And I agree indeed that celeje's reasoning corresponds to the Nirvana fallacy; in my opinion, the present state is that we don't know if the Chess960's castling is really good or not (X state); we could try for example to devise other solutions and to compare them with the present castling, which would be obviously better, as it could possibly permit to discover a better system (Y state), but the idea that heaps of games would permit us to obtain possibly even better results (Z state) justify to stay at the X state without changing anything. So, indeed, I agree with you about this...

- "I agree with you that pure reasoning is the first step. Without pure reasoning chess960 would not have been created in the first place."

Exactly!

Petrarlsen 10/28/2018 05:49

@ lajosarpad (1/2):

- "I think the proponents of chess960 do not consider chess tournaments to be part of the chess960 tournament circuit (...)"

This wasn't what I meant; my meaning was that I think that a certain number of Chess960's proponents consider more or less traditional chess in the context of a traditional chess' competition as being some sort of an inferior form of Chess960. Idea with which I quite disagree!...

- "There is a huge difference between the players knowing that they will play chess, or accidentally playing chess."

Quite true!

- "(...) the outright rejection of all the reasoning and arguments due to lack of "enough" experimenting (...) is an attempt to finish the debate without any result."

I quite agree...

- "(...) there might be some logic by accident (...)"

Yes, I agree, but I don't find very interesting to make complicated researches to know if something haphazard shouldn't be by pure chance something logical! In my opinion, it is much better to simply devise a new solution, as there are no particular reasons to think that the present system is particularly good... And, furthermore, even if there was a "logic by accident" in the present system, I find much simpler and convincing to use an easy-to-understand system, corresponding to a logic which would be possible to understand intuitively, without entering complicated reflections... In my opinion, if it is necessary to enter complicated reasonings to demonstrate that the Chess960's castling is coherent and logical in every situation, it means that this system isn't the best one.

- "(...) some logic, which you (...) assumed to be missing (...)"

I don't remember exactly what I said, but I don't think I said that this logic is missing; I just meant that I don't see the interest to make complicated researches to find a logic where there doesn't seems to be any reason to find one, when it is simply possible to devise a new system, and to try to make it as logical and coherent as possible.

- "It would be unscientific."

In my opinion, for this precisely, it isn't necessary to have a scientific approach, because it isn't necessary to find the "Absolute Best System" (...and this would be rather a tall order, because, certainly, many castling systems could be devised...), but simply to find a coherent and logical system, corresponding to the basic conditions to have something well-balanced, for the castling. If the choice was made of another castling system, for Chess960, and that it would transpire one day that, finally, the original Chess960 castling was perfectly logical, if the new castling was itself quite logical and coherent, it wouldn't be a problem, in my opinion. And, as I said previously, I think that a castling whose logic is simple to understand is anyway better, so, even in this (unlikely) case, I would consider better to use another castling.

celeje 10/27/2018 07:11

@ Petrarlsen:

Petrarlsen: "If, for example, Fischer hesitated between several possible sets of rules, for Chess960's castling, he certainly didn't consider it necessary to have heaps of games played to select what he considered as being the best set of rules! When someone devices something new (be it a game or anything else), very frequently, he hesitated between several possibilities - if it would be necessary, in every case, to test everything extensively, nothing would never be created! "

Huh? Before, you were speculating people defer too much to Fischer. Now you speculate on Fischer's thinking to justify not "testing everything extensively".

No, you create first. Then you test. (Otherwise you have nothing to test, so obviously you must create first.) What you don't do is insist on knowing the answer before you do the testing. The testing will help you find the answer.

celeje 10/27/2018 07:03

@lajosarpad:

lajosarpad: "It is the Nirvana fallacy, which rejects a solution due to the lack of perfection and rather keeps the inferior state of affairs. The fallacy is essentially the approach that from a non-ideal X state we should not switch to Y state with some improvements, because there might be a Z state which is the idealistic best and which does not necessarily match Y. "

But that argument assumes we agree that Y is an improvement over X. We do not agree on that.

We also don't agree on what X and Y are. You approach it like X and Y are versions of a computer program. But if X and Y are "experiments" and switching from X to Y means both X and Y are incomplete, then you effectively have neither X nor Y.

lajosarpad 10/27/2018 06:00

@Petrarlsen

"If a set of rules appears to be much better than another, why should we wait for years and years to implement a change that would appear useful?"

Good question! The approach of many proponents (including Celeje) to postpone all the improvements until we get enough information is fallacious. It is the Nirvana fallacy, which rejects a solution due to the lack of perfection and rather keeps the inferior state of affairs. The fallacy is essentially the approach that from a non-ideal X state we should not switch to Y state with some improvements, because there might be a Z state which is the idealistic best and which does not necessarily match Y.

"Why would it be positive that, 1 time out of 960, the game would be played in a completely different way, compared to the 959 other starting positions?)"

You in your thought process take the original reasoning for the existence of the game (avoiding theory) for granted. However, if I play chess960, I do not aim to avoid theory. As a result, your reasoning is completely implausible for me in this case, even though I know that the existence of the original position of chess among the original positions is a defect from the point of view of the original reasoning. However, I never accepted the original reasoning, which included that the game gets too theoretical. So, if I never accepted that reasoning, why should I accept your conclusion taken from this premise?

My point is that we need to differentiate the original reasoning from the objective value. In our case the original reasoning was very subjective and as such it cannot serve the purpose of being the premise of any objective evaluation, it is certainly not meeting my subjective criteria and I would be surprised if I was the only person whose subjective criteria does not 100% match the original reasoning.

"if it would be necessary, in every case, to test everything extensively, nothing would never be created! "

Exactly. This is why we need to identify the argument to postpone changes until we know "enough" as an instance of the Nirvana fallacy. So far I was the only one who pointed out that this is a logical fallacy in this discussion. Celeje unsurprisingly did not consider his/her argument to be a Nirvana fallacy, but you did not react to this point, which was a key argument in my opinion.

I agree with you that pure reasoning is the first step. Without pure reasoning chess960 would not have been created in the first place.

"In my opinion, to eliminate reasoning and to eliminate experimenting are the two opposite errors; to eliminate reasoning is an impasse; it isn't possible to verify everything in practice; to eliminate experimenting will give many bad results, because pure reasoning is frequently insufficient to draw reliable conclusions. The correct solution is to use pure reasoning as much as possible, and to use experimenting as a complement, when pure reasoning isn't sufficient. "

Agreed.

In my previous comment instead of "much an exact pattern" I meant "match an exact pattern".

lajosarpad 10/27/2018 05:24

@Petrarlsen

I think the proponents of chess960 do not consider chess tournaments to be part of the chess960 tournament circuit, they consider chess960 having the starting position of chess to be equivalent to chess and as such, a part of chess960. I consider this to be disrespectful, because a game which might start in the starting position of chess is different from chess. There is a huge difference between the players knowing that they will play chess, or accidentally playing chess. While the game has similar rules after the starting position was randomized and chess's starting position chose as a chess game, I consider preparation to be part of the game and the preparation phase is different even if the starting position and the rules are identical.

"In fact, when Fischer devised Chess960, he very probably took into account several possibilities, and he choose some of these possibilities and discarded some others, without having heaps of games to compare the diverse solutions."

Excellent point! Proponents tend to accept Fischer's solution, yet, discard the very methodology he used to avoid changes. Actually experimenting never hurts, but the outright rejection of all the reasoning and arguments due to lack of "enough" experimenting is a logical no go, since it does not justify the reason why shouldn't we experiment with different rules, it does not attempts to convince us about the quality of the current rules, instead, it is an attempt to finish the debate without any result.

"The answer is "no"; it just follows from the pieces' positioning in the different Chess960's starting positions. For example, if the reason for these differences was that, in some positions, the King would be safer in the center, and in some other, safer in a corner, it would be quite correct, but such isn't the case: sometime the King goes toward the center, sometime toward a corner, but there isn't any particular reason for this, apart from the random positioning of the pieces in the starting position of the game. "

We do not know the actual value of castling in all the positions. There might be a plausible knowledge even if it is there by accident and not by design. I think it is worth analyzing. I'm not saying no changes are needed, but I certainly say that there might be some logic by accident. You are assuming we cannot find a very plausible logic and you might be right, but, frankly none of us knows. So, to make sure my point is not misunderstood: I welcome any analysis, the knowledge which comes from those, but I am not against changes before we get all the needed information. I do not consider my approach to be faulty, I was just taking into account the possibility that there might be some logic, which you - prematurely - assumed to be missing, due to the reason that there was no any guarantee in the planning phase that there will be some logic of the kind you are looking for.

"when the King goes toward the center, it isn't because the center, for this position, is supposed to be safer; the only justification for these discrepancies is pure chance; nothing else."

This is true, but your mistake is that you exclude the possibility that a plausible explanation could be given. In this moment it looks like your proposal is right, so I tend to think it is worth implementing, but I am not sure at all that your assumptions are correct. Believing them requires faith, which is not my strong point. All I can say now is that your suggestion seems to be an improvement and it looks like they are worth implementing, but maybe some studies will convince me about the opposite if they show that the king in every position ends up in a safer point. I know there is not much chance of this happening, firstly, because a lot of uncharted territory should much an exact pattern, which usually does not happen, secondly, because the question is difficult to answer, but I am not open to exclude this possibility. It would be unscientific.

celeje 10/27/2018 12:39

@ Petrarlsen: I'll read and comment soon, not necessarily in that order.

Petrarlsen 10/27/2018 12:33

...And this article just became The Most Commented Article Ever on ChessBase!...

Chess960 REALLY seems to be an inexhaustible theme!...

Petrarlsen 10/27/2018 12:18

I have tried to devise a new system to replace the current Chess960's castling, on the basis of the "b/g files castling" (placing the King on the b1/b8 square or the g1/g8 square). For the moment, I will call it Chess480BG.

(Reminder: I also consider that Chess480 is an interesting idea, and I proposed an improvement for it under this same article.)

1) The starting positions:

a) Suppression of the positions in which the King is on the b or g files (to have a castling clearly oriented toward a King's move from the center towards the flanks, as in traditional chess and as in 60 %, if I am not mistaken, of the Chess960's castlings, the "toward the center castlings" representing only 11 % of the Chess960's castlings: as castling is, in most cases, a "toward the flanks" move, I think that it makes sense to chose this solution rather than another one).

b) Suppression of the positions in which the King is on the c-file, a Rook being on the b-file, and, symmetrically, of the positions in which the King is on the f-file, a Rook being on the g-file. (The reason being to ensure that both castlings can be used in practice; if, for example, the King is on c1 and a Rook on b1, this castling can be played immediately, while, for the other castling, it would be necessary to move all the pieces between the c1 and the g1 square; I consider that this makes too great a difference between the two castlings, and that it would favor too much the easier castling).

If I am not mistaken, this would keep more or less 564 starting positions (but, as this version is symmetrical, only half of these positions would be different one from the others, so this would make more or less 282 really different starting positions).

2) The rules:

The general idea would be to center castling more than in Chess960 or traditional chess on the idea of a King's move. This because, in many cases, I think that the Rook's displacement isn't useful.

So, in this version of the game, the castling would be a move which would bring the King from its starting position (keeping the rule according to which the King musn't have moved previously) towards either the b1/b8 or the g1/g8 squares. As for the other pieces, it would be necessary that all the squares between the King and its arrival square would be free, with special dispositions for the Rooks: on the side chosen for castling, the Rook musn't have moved from the beginning of the game; if the Rook is on a1/a8 or b1/b8, it will be placed on c1/c8 after castling and if the Rook is on g1/g8 or h1/h8, it will be placed on f1/f8 after castling (to avoid blocking a Rook in a corner - for the a1/a8 and h1/h8 squares - or to free the King's arrival square - for the b1/b8 and g1/g8 squares); if the Rook is on any other square, it will simply stay on its original square when castling is played.

The other current conditions, for castling (that the King musn't be in check, etc.) would be kept.

In comparison with the current Chess960's castling rules, I rather think that both this system and my improved Chess480 system (with two-squares castling, as in the original Chess480) would be more coherent with the "King's safety" idea.

In my opinion, it would be interesting to compare more globally these two systems with the current Chess960's system, to evaluate which is the best system.

(Note: To calculate the repartition between "towards the flanks castling" and "towards the center castling" in Chess960, I have used the data of an article by Mark Weeks, whose title is "Introduction to Chess960 Geometry"; as we are not normally supposed to put links in the comments section, I will not put a link to it, but it can be very easily found by searching "Introduction to Chess960 Geometry Weeks" - without quotation marks - on Google.)

Petrarlsen 10/26/2018 09:25

@ celeje and lajosarpad (5/5)

- @ celeje:

"Playing the game or games IMO should be the first step, not the last step."

If, for example, Fischer hesitated between several possible sets of rules, for Chess960's castling, he certainly didn't consider it necessary to have heaps of games played to select what he considered as being the best set of rules! When someone devices something new (be it a game or anything else), very frequently, he hesitated between several possibilities - if it would be necessary, in every case, to test everything extensively, nothing would never be created!

And I don't see the difference with the present situation; if we are considering other, new, possibilities, we can compare them to the current rules to see if one of them is clearly better than the others. And, yes, Chess960's competitions are played, but what does it change? If a set of rules appears to be much better than another, why should we wait for years and years to implement a change that would appear useful? Just because some Chess960's competitions are played?? The - very paradoxical - result would be that the fact that some Chess960's competitions are played would have for a consequence that even the most useful changes couldn't be implemented; this wouldn't be at all logical, in my opinion!

To go back to your statement: "Playing the game or games IMO should be the first step, not the last step.", in my opinion, this is a wrong reasoning; the first "filter" is necessarily made by pure reasoning - it is best to go as far as possible by pure reasoning, because it is much simpler, and much less time-consuming. And, when pure reasoning isn't sufficient, THEN it is necessary to experiment.

In my opinion, to eliminate reasoning and to eliminate experimenting are the two opposite errors; to eliminate reasoning is an impasse; it isn't possible to verify everything in practice; to eliminate experimenting will give many bad results, because pure reasoning is frequently insufficient to draw reliable conclusions. The correct solution is to use pure reasoning as much as possible, and to use experimenting as a complement, when pure reasoning isn't sufficient.

So, to apply this to Chess960's castling rules, the best would be to compare the present rules with other possible rules, and to see whether it is possible to draw a sufficiently reliable conclusion by pure reasoning. And if it isn't possible to have a clear result for this, then it would be useful to experiment.

An imaginary example: If Fischer would have made public 3 versions of his game: Chess960, Chess480, and Chess480BG, what could have been done about it?

Following you, as there couldn't have been any means to determine which one would be better, it would have been absolutely necessary to experiment these 3 versions.

But, in my opinion, it is rather obvious that it wouldn't be possible to multiply these sorts of experiments; the only result would have been that Chess960 et alii would have been completely forgotten.

So, the correct approach would have been to try to determine which one is the better version of this game, and then, to organize competitions with this version.

This is exactly, in my opinion, what must be done today: to compare the different versions, and, if the current version wouldn't be considered as been optimal, to try to implement the necessary changes.

Petrarlsen 10/26/2018 09:25

@ celeje and lajosarpad (4/5)

- @ lajosarpad:

* "I do not see why the starting position of chess being possible is a defect."

I don't quite understand your viewpoint; in my opinion, for this, it is rather simple: To use Position 518 has, in my opinion, a serious disadvantage (the point that I developed earlier about opening theory), and I don't find any concrete advantage for using this starting position. (What can it bring to a player? Or to the game itself? Why would it be positive that, 1 time out of 960, the game would be played in a completely different way, compared to the 959 other starting positions?) So, for me, the fact that Chess960 includes Position 518 is indeed a defect of the game.

* When you say: "(...) we might consider adding that the whole process could be a single move, but the opponent could pass as many times as many additional moves were needed for castling by the incriminated rook and king.", I must say that I don't really fully understand what you mean. Perhaps I am missing something very obvious! But could you elaborate a little, about this point?

- @ celeje:

"Focusing so much on a simple castling "explanation", which to me is like a folktale (...)" (About the "King's safety" idea, for castling.)

My reasoning is quite simple, for this:

Step 1: Castling is a special move.

Step 2: If a special move is implemented in a game, there must be a reason for it.

Step 3: Lajosarpad and I agree that the "King's safety" idea is quite satisfying, to explain castling. And I don't think I have read anyone proposing another possibility. So, in my opinion, the "King's safety" idea is, for the moment, the best - and even the sole existing - explanation, for castling.

And it seems simply logical to verify if, concretely, the implementation of this move corresponds to the move's general idea. Just a question of coherence, in fact...

So there isn't any link with any "folktale", as for this...

News

SHOP

SHOP

Champions Showdown: Improving the format

ONLINE SHOP

ChessBase Magazine 226

Chess960 is here to stay

Sinquefield is sold

Room for improvement

Picking the position

Kasparov's final thoughts on the Champions Showdown

Links

Discuss

ChessBase '26 - Mega Package

Understanding Middlegame Strategy Vol.14 Reversed Colour Systems – Benoni, Blumenfeld and Benko Gambit

The Ultimate Antidote to the London System

London System Powerbase 2026

London System Powerbook 2026

Mastering the London System

ChessBase Magazine Extra 230

Opening Encyclopaedia 2026

Pop-up for detailed settings