A balanced approach to imbalances

[Lyudmil Tsvetkov]

Well, because knights with queens complement better. In the Q vs 3Ns you have 2 capacities in the queen - diagonal and linear, and only one in the 3 knights. In the 3Qs vs 2Qs + 3Ns you have 2 capacities for the Q side, and 3 capacities for the Q+Ns side. This brings an added value. Actually, while with a single queen or 3 queens you still have 2 capacities, in the 3Ns you have only one capacity, but in the 2Qs + 3Ns you have full 3 capacities! That makes a difference of 2 capacities, and that is very important. Do you agree now with my explanation?

Of course not. Because again, these are just words, offered without any proof. Obviously your explanation would fail if all the Queens were substituted by Chancellors, which move as Rook or Knight. Then 2C + 3N would still only have 2 capacities. Can you show evidence that in KCKNNN (+Pawns) that is equal, the chances do not swing strongly in favor of the Knights side when you add two more Chancellors on both sides??? This whole concept of 'capacities' is just sprouting from your imagination, without any basis or validity in real life.

You still did not answer why you claim that the knights would interdict access of queens to squares on the board, when I showed with counting mobile squares in the 3 diagrams I posted that mobility even with this strange imbalance is still a function of centralisation (quite explicable, as the piece values themselves are a function of a measurement of average mobility performance on an empty board), while at the same time in all 3 positions the knights enjoy extremely good defence. So that, actually, elephantiasis is due to unnaturally high mutual piece defence (with its corresponding bonus points), and not to square interdiction. Would you admit your theory was wrong?

Does this anser your question?
[d]8/6p1/p1knn3/2n3n1/2n3P1/2P1nn2/Q7/KQQ5 w - - 0 1

[d]1nn1k1n1/pppppppp/8/8/8/8/PPPPPPPP/3QK3 w - - 0 1
I do not know why, but in my analysis games I win all with white. It seems that 1.d4 is the best move for white, and it seems that black loses fairly quickly. Do you agree that white is winning here?

well, let's say I would not be surprised. The Pawns are spread out, with clearly disadvantages the Knights. Remove the 4 outermost Pawns on each side, and I would not be so sure.

More interesting seems whether you can also win this against Stockfish, when the latter plays black. Are you better at this than an engine, or worse? Is there any merit in listening to your analysis of this, or would it be better to just have the engines battle it out, and collect the statistics?

[d]1nnnknn1/pppppppp/8/8/8/8/PPPPPPPP/2QQK3 w - - 0 1
2Qs vs 5Ns. Natural initial position with advantage for neither side in terms of immaterial factors. I do not why, but in my analysis games, I am winning most of the games with white. And I definitely think this is won for white. Do you agree with that?

Not without proof. Even if with the same Pawn distribution it would be won with Q-3N, it is not obvious at all that you could force conversion to it against an opponent that consciously avoids such conversions. Perhaps you could beat Stockfish, but I don't believe for a minute you could beat QueeNy, (the latter with black). Show us the game where you do so.

[d]nnnnknnn/pppppppp/8/8/8/8/PPPPPPPP/Q2QK2Q w - - 0 1
3Qs vs 7Ns in a natural initial position. This is the Lucas position, but you might also use the position you posted with queens on b1 and g1. Do you agree that white is better here? I did not try analysing this extensively, as it is a bit complex, but I think it is still much better for white, although more difficult to play than the 2 other positions. Do you agree that white is better here? I think the fact that Disco managed to draw here points that the 2 other positions favour white even more, and maybe Disco did not play an optimal game for white (Lucas will excuse me, but I have my doubts). I think white is better even here.

Thousands of games with a variety of engines, and engines that were particularly modified to understand this position, have demonstrated beyond a shadow of a doubt that this position is utterly lost to white. No chance at all. so much for your intuition. QueeNy always wins this with black, even against tactically superior engines that know about the Q for 2N trick.

So, let us concentrate on factual assessment to see who is right. Any third party posting games/analysis with these 3 positions would of course be of great help.

OK, I am waiting for your wins against QueeNy.

The matter of fact is that I am able to win all games I am playing in analysis mode against myself in the first 2 positions, and most in the third.

Below a typical example of 3Qs vs 7Ns:
[d]1n1nkn1n/p1n2p1n/2p1p3/1p2Pn1P/2PP1QQ1/1P6/P7/4K2Q b - - 0 18
No chance for the knights, you see.

And a nice mating net:
[d]3Q4/8/1n4Q1/kn2Q3/1n6/8/Q2Q4/5QK1 b - - 0 1

Regarding complementarity rules, even Capablanca said that Q+N perform better than Q+B, because they have access to a wider range of squares on the board. Do you think Capablanca was stupid. I think Larry also conducted some experiments showing that Q+N perform better than Q+B. What do you think is the reason for this, if not complementarity? There are 3 basic piece types: rooks, knights and bishops, and you have to do something with that. The queen is just a combination of 2 piece types in many respects, at least in terms of the squares it has access to, but is more valuable than B and R as it moves quicker.

You did not answer if you acknowledge that elephantiasis, whenever relevant, is due to abundant piece defence, and not to interdiction of squares to opponent pieces. I showed 3 diagrams where this is evident, show me 3 where this is not so.

Regarding chancellors, what is the point of this? This is not a standard chess topic. 7 knights might still be, but chancellors. I once invented a chess set with a board of 14x14, 196 squares in all, 14 pawns and 14 pieces each side. Apart from a queen and a chancellor, there were pieces combining the moves of a bishop and a knight, and a very strong piece combining the moves of a bishop, rook and knight. I made myself a cardboard set and played some games on it, it was all very interesting, but complexity was beyond any grasp. So what, what s the point of this, how is this relevant to the topic we discuss on standard chess? I formulated some imbalances rules that are as practical as possible. If they work or not, only people that try to implement them might tell you. I think they make much sense indeed, but I am not a programmer to check this.

Regarding Queeny games, I am sure I will win most of the games with the queens, using more time to avoid some tactics, but what would be the point of this exercise? You need many games to come to a valid conclusion. OK, let us pick up the position with 2Qs vs 5Ns (a medium case) and try to analyse it further. I bet the queen side wins with a rate above 70%. If someone is able to play 20 games against different opponents of roughly equal strength at say, 5 minutes per game, that would be great, I just do not play too many engine-engine games nowadays.

[Rein Halbersma]

The matter of fact is that I am able to win all games I am playing in analysis mode against myself in the first 2 positions, and most in the third.

Please, stop right there. Do as H.G. asked you to: play against QueeNy, not against yourself. There is no objectivity otherwise.

[Lyudmil Tsvetkov]

The matter of fact is that I am able to win all games I am playing in analysis mode against myself in the first 2 positions, and most in the third.

Please, stop right there. Do as H.G. asked you to: play against QueeNy, not against yourself. There is no objectivity otherwise.

Queeny plays against itself, why should not I play against myself?
When I am analysing, I give myself enough time. When I have a moving opponent, especially with an unfamiliar position, the quality of moves will be lower. That is why I would prefer some engine-engine games, as engines would play tactically good chess even with little time. If I give myself 10 minutes for the entire game, and Queeny 5, and make some stupid tactical mistake, would that prove that my position is lost?

[d]1nnnknn1/pppppppp/8/8/8/8/PPPPPPPP/2QQK3 w - - 0 1

A reasonable suggestion would be for someone to play 10 5 minutes games from the above position, matching Queeny against 5 other opponents with reversed colours. The sparring engines might be Stockfish, Arasan, DiscoCheck, Crafty, Fruit 2.1, or other similar strength engines.

I would be happy if you could help with this, Rein.

[hgm]

The matter of fact is that I am able to win all games I am playing in analysis mode against myself in the first 2 positions, and most in the third.

OK, so you are a lousy player with Knights... Glad you wanted to share that with us. Now can we get back on topic?

Regarding complementarity rules, even Capablanca said that Q+N perform better than Q+B, because they have access to a wider range of squares on the board. Do you think Capablanca was stupid. I think Larry also conducted some experiments showing that Q+N perform better than Q+B. What do you think is the reason for this, if not complementarity? There are 3 basic piece types: rooks, knights and bishops, and you have to do something with that. The queen is just a combination of 2 piece types in many respects, at least in terms of the squares it has access to, but is more valuable than B and R as it moves quicker.

Capablanca might not have been stupid, but he did not have the advantage of the huge digitized databases of GM games later researches had. So I think it is more significant what Larry kaufman had to say about this. where did he say that Q+N was better than Q+B? And what were his statements about C+N vs C+B? (Because if there the Knight is better too, it has nothing to do with complementarity, but would just be a defect of the Bishop...)

You did not answer if you acknowledge that elephantiasis, whenever relevant, is due to abundant piece defence, and not to interdiction of squares to opponent pieces. I showed 3 diagrams where this is evident, show me 3 where this is not so.

I did answer that: only partly. Abundant piece defense is a form of interdiction. So I don't know what you are trying to prove here. If it is due to abundant piece defense, that would confirm the elephantiasis theory. But there is more than that (and that still confirms the elephantiasis theory.) I already showed you 1 diagram, and I think it says it all! I don't see what analyzing the traits of positions that you first constructed to have that trait can accomplish. If you want to do something useful, calculate the average interdiction (split out by square occupant: mutual defense or board control) during an actual 7N-3Q game, in stead of positions that you dreamed up because your intuition decided they must be favorable.

Regarding chancellors, what is the point of this? This is not a standard chess topic. 7 knights might still be, but chancellors. I once invented a chess set with a board of 14x14, 196 squares in all, 14 pawns and 14 pieces each side. Apart from a queen and a chancellor, there were pieces combining the moves of a bishop and a knight, and a very strong piece combining the moves of a bishop, rook and knight. I made myself a cardboard set and played some games on it, it was all very interesting, but complexity was beyond any grasp. So what, what s the point of this, how is this relevant to the topic we discuss on standard chess? I formulated some imbalances rules that are as practical as possible. If they work or not, only people that try to implement them might tell you. I think they make much sense indeed, but I am not a programmer to check this.

If you have a theory that only works for orthodox Chess pieces, and totally fails for very similar pieces that by historic accident are not in orthodox Chess, that theory is suspect. It means you just have been rationalizing empirical facts of Chess by dreaming up fancy names for them, but haven't really learned anything on the mechanism that determines piece values. Eliphantiasis is a universal theory. It doesn't care at all whether the pieces happen to feature in orthodox Chess or not.

And what you say is nonsense. You don't have to be a programmer to test your rules, you can easily do that yourself. Just set up the imbalance, and let strong engines play out the games. If you to that often enough, you will know whether the imbalance is favorable (to engines), or not. And thus if it would be worthwhile for anyone to implement them.

Regarding Queeny games, I am sure I will win most of the games with the queens, using more time to avoid some tactics, but what would be the point of this exercise?

The point would be that your silly claims that you could win this would be squashed. Of course I understand that you don't like that. But then you should not make them. You would be crushed by QueeNy.

You need many games to come to a valid conclusion.

Don't worry. You would have millions of games before you win one!

OK, let us pick up the position with 2Qs vs 5Ns (a medium case) and try to analyse it further. I bet the queen side wins with a rate above 70%. If someone is able to play 20 games against different opponents of roughly equal strength at say, 5 minutes per game, that would be great, I just do not play too many engine-engine games nowadays.

It seems a good opportunity to change that, then. Any particular reason you want others to do all the work needed to satisfy your curiosity? I have never heard an excuse for shirking as feeble as this one! Is it that you cannot be bothered with uncovering facts because you are too busy fantasizing about evaluation of Chess positions?

[hgm]

Btw Larry Kaufman does claim, in his famous paper about imbalances, that a Queen would be equal to within a very small fraction of a Pawn to 2N+B. Considering that I found that all Knights in general cooperate better than mixed Knights and Bishops, even the Q vs 3N case should be slightly in favor of the Knights. Perhaps a 55% score in favor of the Knights if you average over a representative sample of Pawn constellations.

[Rein Halbersma]

A reasonable suggestion would be for someone to play 10 5 minutes games from the above position, matching Queeny against 5 other opponents with reversed colours. The sparring engines might be Stockfish, Arasan, DiscoCheck, Crafty, Fruit 2.1, or other similar strength engines.

I would be happy if you could help with this, Rein.

Now that round-robin sounds more like it. But the burden of proof is on you, since you are trying to making a point. I'm just a spectator.

[Lyudmil Tsvetkov]

Btw Larry Kaufman does claim, in his famous paper about imbalances, that a Queen would be equal to within a very small fraction of a Pawn to 2N+B. Considering that I found that all Knights in general cooperate better than mixed Knights and Bishops, even the Q vs 3N case should be slightly in favor of the Knights. Perhaps a 55% score in favor of the Knights if you average over a representative sample of Pawn constellations.

That is total ...
Anyone who has a reasonable amount of chess knowledge knows that Q is stronger than 2Ns + B, even stronger than the mostly imaginable case of 3Ns, but weaker than 2Bs + N. That is an empirical fact. OK, play 20 1 minute games of Queeny from that position:

[d]1nn1k1n1/pppppppp/8/8/8/8/PPPPPPPP/3QK3 w - - 0 1

against a small sample of engines with reversed colours and report the results. As the position is actually won in just a couple of moves for the queen side, I would be surprised if the knights get more than 1 or 2 points, probably none, as the position is very simple. Would you please do that?

You should understand me why I would not like to play the games myself: I am not a programmer, nor a statistician, I have even some problems with software applications, so I would really appreciate some help with the games.

[Lyudmil Tsvetkov]

The matter of fact is that I am able to win all games I am playing in analysis mode against myself in the first 2 positions, and most in the third.

OK, so you are a lousy player with Knights... Glad you wanted to share that with us. Now can we get back on topic?

Regarding complementarity rules, even Capablanca said that Q+N perform better than Q+B, because they have access to a wider range of squares on the board. Do you think Capablanca was stupid. I think Larry also conducted some experiments showing that Q+N perform better than Q+B. What do you think is the reason for this, if not complementarity? There are 3 basic piece types: rooks, knights and bishops, and you have to do something with that. The queen is just a combination of 2 piece types in many respects, at least in terms of the squares it has access to, but is more valuable than B and R as it moves quicker.

Capablanca might not have been stupid, but he did not have the advantage of the huge digitized databases of GM games later researches had. So I think it is more significant what Larry kaufman had to say about this. where did he say that Q+N was better than Q+B? And what were his statements about C+N vs C+B? (Because if there the Knight is better too, it has nothing to do with complementarity, but would just be a defect of the Bishop...)

You did not answer if you acknowledge that elephantiasis, whenever relevant, is due to abundant piece defence, and not to interdiction of squares to opponent pieces. I showed 3 diagrams where this is evident, show me 3 where this is not so.

I did answer that: only partly. Abundant piece defense is a form of interdiction. So I don't know what you are trying to prove here. If it is due to abundant piece defense, that would confirm the elephantiasis theory. But there is more than that (and that still confirms the elephantiasis theory.) I already showed you 1 diagram, and I think it says it all! I don't see what analyzing the traits of positions that you first constructed to have that trait can accomplish. If you want to do something useful, calculate the average interdiction (split out by square occupant: mutual defense or board control) during an actual 7N-3Q game, in stead of positions that you dreamed up because your intuition decided they must be favorable.

Regarding chancellors, what is the point of this? This is not a standard chess topic. 7 knights might still be, but chancellors. I once invented a chess set with a board of 14x14, 196 squares in all, 14 pawns and 14 pieces each side. Apart from a queen and a chancellor, there were pieces combining the moves of a bishop and a knight, and a very strong piece combining the moves of a bishop, rook and knight. I made myself a cardboard set and played some games on it, it was all very interesting, but complexity was beyond any grasp. So what, what s the point of this, how is this relevant to the topic we discuss on standard chess? I formulated some imbalances rules that are as practical as possible. If they work or not, only people that try to implement them might tell you. I think they make much sense indeed, but I am not a programmer to check this.

If you have a theory that only works for orthodox Chess pieces, and totally fails for very similar pieces that by historic accident are not in orthodox Chess, that theory is suspect. It means you just have been rationalizing empirical facts of Chess by dreaming up fancy names for them, but haven't really learned anything on the mechanism that determines piece values. Eliphantiasis is a universal theory. It doesn't care at all whether the pieces happen to feature in orthodox Chess or not.

And what you say is nonsense. You don't have to be a programmer to test your rules, you can easily do that yourself. Just set up the imbalance, and let strong engines play out the games. If you to that often enough, you will know whether the imbalance is favorable (to engines), or not. And thus if it would be worthwhile for anyone to implement them.

Regarding Queeny games, I am sure I will win most of the games with the queens, using more time to avoid some tactics, but what would be the point of this exercise?

The point would be that your silly claims that you could win this would be squashed. Of course I understand that you don't like that. But then you should not make them. You would be crushed by QueeNy.

You need many games to come to a valid conclusion.

Don't worry. You would have millions of games before you win one!

OK, let us pick up the position with 2Qs vs 5Ns (a medium case) and try to analyse it further. I bet the queen side wins with a rate above 70%. If someone is able to play 20 games against different opponents of roughly equal strength at say, 5 minutes per game, that would be great, I just do not play too many engine-engine games nowadays.

It seems a good opportunity to change that, then. Any particular reason you want others to do all the work needed to satisfy your curiosity? I have never heard an excuse for shirking as feeble as this one! Is it that you cannot be bothered with uncovering facts because you are too busy fantasizing about evaluation of Chess positions?

Larry did say somewhere that he extensively tested Q+N vs Q+B, or checked some database, coming to the conclusion that Q+N performed somewhat better. I read that somewhere, but I do not remember now where. Probably Larry could help with this.

C+N, or R+N+N, would be of course worse in terms of complementarity than C+B, or R+N+B, 2 against 3 piece types, I think it is obvious.

Abundant piece defence is not a form of interdiction, I think the diagrams I posted showed that.

I do not know if Queeny will beat me or not, but I do know that I am able to score well against considerably stonger engines, so that Queeny should not really be a problem.

OK, if we want to be at least a bit helpful, let us concentrate on the crux of the matter: who is better in the below position?

[d]1nnnknn1/pppppppp/8/8/8/8/PPPPPPPP/2QQK3 w - - 0 1

I say white is much better, you would claim otherwise.

Would you please play (and I already explained why it is not very convenient for me to do that) 20 1 minute games of Queeny against 5 different opponents of roughly equal strength with reversed colours and report the results?

All I can do is post my analysis, but as no one is interested, it is better for you or someone else to quickly play those games.

[hgm]

Queeny plays against itself, why should not I play against myself?

Because queeNy is the strongest player of this sort of game on the planet, and your play utterly sucks?

That is total ...
Anyone who has a reasonable amount of chess knowledge knows that Q is stronger than 2Ns + B,

So GM Larry Kaufman does not have the slightest Chess knowledge, and when he says that masters in general value 3 minors even half a Pawn higher, he is just lying through his teeth, right? And that the statistics of GM games favors the minors of course is totally irrelevant. It only means they win more, despite the fact that they are far, far weaker.

This discussion starts to be a bit surreal...

even stronger than the mostly imaginable case of 3Ns, but weaker than 2Bs + N. That is an empirical fact. OK, play 20 1 minute games of Queeny from that position:

[d]1nn1k1n1/pppppppp/8/8/8/8/PPPPPPPP/3QK3 w - - 0 1

against a small sample of engines with reversed colours and report the results.

To quote from a famous western movie: "You must be soft between the ears".

I am not a programmer, nor a statistician, I have even some problems with software applications, so I would really appreciate some help with the games.

You don't seem to have any problems posting here. The greatest idiots on this planet can download and run engines. Even you should be able to learn how to do it.

Abundant piece defence is not a form of interdiction, I think the diagrams I posted showed that.

It is, because I say it is. I coined the term, so I define what it means. Capice?

I do not know if Queeny will beat me or not, but I do know that I am able to score well against considerably stonger engines, so that Queeny should not really be a problem.

Except for the tiny detail that you will be starting from a lost position. Can you also beat Houdini in KQK, when you are playing the bare King?

[enhorning]

While I find the material imbalances interesting, I think this discussion has too much emotionally based arguments, and not enough games.

So, as the discussion started with 7 Knights versus 3 Queens, I ran QueeNy against a gauntlet of Komodo 6, Critter 1.6a, Stockfish 4, Houdini 1.5a and Gull 2.3, at a time-control of 1 minute + 1 second increment, starting from this position:
[d]nn1nknn1/2nppp1n/8/8/8/8/3PPP2/2QQKQ2 w - - 0 1
(The most pawns one can have and still have a Fide legal position. Two knights have been moved forward a single step from the home line to avoid having material en prise in the initial position.)

As white, the side with the 3 queens, QueeNy got 5 draws.

As black, the side with the 7 knights, QueeNy got 4 wins and 1 loss.

I would note that QueeNy has multiple handicaps. The other engines are running with my default settings of 4 threads and 256 MB Hash. QueeNy only has 1 thread, and whatever its default hash is - considerably smaller, looking at the memory footprint.

Games can be downloaded from: https://www.dropbox.com/s/pm5fgtniucixwt5/3Q-7N.pgn

I'll leave drawing conclusions up to others - I am not a chess theoretician - I just like playing oddball variants, and watch them get played, and this extreme inbalance of 7N vs 3Q was intriguing me.