We should reconsider the value of the Queen.................

[pichy]

Each and every position you showed so far is either a 5 or 6men endgame.
I guess you have recently discovered them, this is fine but bad for a discussion about Q values.

Guenther

Then show me a better position to support my argument about the diminish value of the Queen as it enter the endgame versus Two Rooks or Rook versus Bishop

[pichy]

Each and every position you showed so far is either a 5 or 6men endgame.
I guess you have recently discovered them, this is fine but bad for a discussion about Q values.

Guenther

Okay here is another fine example where the Queen is not that powerful, Black to play and tell me who has the better chance

[d]6r1/6rk/8/8/8/8/8/Q6K b - - 14 1

And Black to play and draw
[d]8/5Q2/8/8/1k6/b1n5/8/K7 b - - 14 1

[Sven]

Each and every position you showed so far is either a 5 or 6men endgame.
I guess you have recently discovered them, this is fine but bad for a discussion about Q values.

Okay here is another fine example where the Queen is not that powerful [...]

Again, as I already wrote: all your examples are 100% "tactical" (see below). I'd propose to continue with examples where the "static evaluation" aspect is more important than search, tactics, or even TB lookup.

These were your examples so far:

1) 8/8/8/3kb3/3r4/3KQ3/8/8 w - - 0 1
Tablebase draw

2) 8/8/8/3kn3/3r4/2QK4/8/8 w - - 0 1
Unreachable from legal games (check by rook + knight at the same time).
If it were reachable: a tablebase draw

3) 3k4/2rr4/8/8/8/8/8/3KQ3 w - - 0 1
Forced mate in 8 for black, white immediately loses its queen in the next move

4) 4k3/5r2/8/8/8/8/3P2Q1/1r2K3 w - - 0 1
Tablebase draw

5) 3k4/2rr4/8/8/5Q2/8/8/3K4 w - - 0 1
Tablebase draw

6) 6r1/6rk/8/8/8/8/8/Q6K b - - 14 1
Forced mate in 6 for black, easy to see e.g. after 1...Rh8

7) 8/5Q2/8/8/1k6/b1n5/8/K7 b - - 14 1
Tablebase draw

Maybe you find a good example with Q + some pawns vs. RB + same number of pawns where you can explain your idea?

Please note also, there have been thorough investigations based on many real games on that topic, e.g. by Larry Kaufman, which have been followed and used by quite some people since then. Furthermore, you stated that the value of a queen were "fixed over the whole game", this is not true, many engines use the well-known opening-endgame interpolation technique where the actual piece values depend on the game stage.

Sven

[S.Taylor]

You made a good point Muller, if it is White to move in this position where the King is not in check the best that White can do is get a draw:

If it is the wrong person to move, then EVERYTHING is always different.
Next move could be Pawn takes king. Or if you don't want to include a position of check, it could be a position of pawn takes queen, when it is not the colour of the queen to move!

[pichy]

Each and every position you showed so far is either a 5 or 6men endgame.
I guess you have recently discovered them, this is fine but bad for a discussion about Q values.

Okay here is another fine example where the Queen is not that powerful [...]

Again, as I already wrote: all your examples are 100% "tactical" (see below). I'd propose to continue with examples where the "static evaluation" aspect is more important than search, tactics, or even TB lookup.

These were your examples so far:

1) 8/8/8/3kb3/3r4/3KQ3/8/8 w - - 0 1
Tablebase draw

2) 8/8/8/3kn3/3r4/2QK4/8/8 w - - 0 1
Unreachable from legal games (check by rook + knight at the same time).
If it were reachable: a tablebase draw

3) 4k3/5r2/8/8/8/8/3P2Q1/1r2K3 w - - 0 1
Tablebase draw

4) 3k4/2rr4/8/8/5Q2/8/8/3K4 w - - 0 1
Tablebase draw

5) 6r1/6rk/8/8/8/8/8/Q6K b - - 14 1
Forced mate in 6 for black, easy to see e.g. after 1...Rh8

6) 8/5Q2/8/8/1k6/b1n5/8/K7 b - - 14 1
Tablebase draw

Maybe you find a good example with Q + some pawns vs. RB + same number of pawns where you can explain your idea?

Here is a great example and I decided to played it versus Komodo, I was playing White:

[Event "Queen vs Bishop+Rook"]
[Site "?"]
[Date "2012.04.08"]
[Round "?"]
[White "Pichy"]
[Black "Komodo3-32"]
[Result "1-0"]
[BlackElo "3080"]
[Time "18:17:34"]
[WhiteElo "2200"]
[TimeControl "0+20"]
[SetUp "1"]
[FEN "6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "26"]
[WhiteType "Pichy"]
[BlackType "program"]

[d]6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1

1. Bf8+ Kh7 2. Rg7+ Kh8 3. Be7 Qe1 4. Bf6 Qxe2+ 5. Kh1 Qc4 6. Ba1 h4 7. Bf6 h3 8. Ba1 h2 9. Bf6 Qe6 10. Bd4 Qh6 11. Rg2+ Kh7 12. Rxh2 Kg6 13. Rxh6+ Kxh6 *

And after this position there is no need to continue since White has an easy win.
[d]8/8/2p4k/2P5/3B1p2/5P2/8/7K w - - 0 14

Please note also, there have been thorough investigations based on many real games on that topic, e.g. by Larry Kaufman, which have been followed and used by quite some people since then. Furthermore, you stated that the value of a queen were "fixed over the whole game", this is not true, many engines use the well-known opening-endgame interpolation technique where the actual piece values depend on the game stage.

Sven

[Sven]

Here is a great example and I decided to played it versus Komodo, I was playing White:

[Event "Queen vs Bishop+Rook"]
[Site "?"]
[Date "2012.04.08"]
[Round "?"]
[White "Pichy"]
[Black "Komodo3-32"]
[Result "1-0"]
[BlackElo "3080"]
[Time "18:17:34"]
[WhiteElo "2200"]
[TimeControl "0+20"]
[SetUp "1"]
[FEN "6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "26"]
[WhiteType "Pichy"]
[BlackType "program"]

[d]6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1

1. Bf8+ Kh7 2. Rg7+ Kh8 3. Be7 Qe1 4. Bf6 Qxe2+ 5. Kh1 Qc4 6. Ba1 h4 7. Bf6 h3 8. Ba1 h2 9. Bf6 Qe6 10. Bd4 Qh6 11. Rg2+ Kh7 12. Rxh2 Kg6 13. Rxh6+ Kxh6 *

O.k., this example might look better but it is still one of those 0.0x% where the exact value of the queen does not matter much due to "short-term" tactical issues (though not so short-term here considering the number of plies until the black queen is finally lost). Here these issues are caused by the extremely weak position of the black king into which black is being forced tactically with the first four white moves.

You are presenting a nice endgame study where the solution typically deviates from what you expect as a normal outcome for the vast majority of cases. With the given material of QPPP vs RBPPP and given typical piece values of Q=9.75, R=5.0, B=3.25 you get an advantage of +1.5 pawns for black. Without any further positional scoring the overall evaluation would be right within the draw margin for positions with equal number of pawns, so we would expect a draw, albeit with some black advantage. But in the given position the king safety is extremely bad for black, as stated above, so even a static evaluation would show at least a small advantage for black.

Queens can be stronger than rook + bishop, though, in positions with this kind of material, for instance if
- their own king is safe,
- rook and bishop are not cooperating well, and/or
- the rook/bishop side has several weak points that can be attacked simultaneously be the queen, showing the main strength of a queen which is its ability to act at many places on the board "at the same time".

In the example you gave above all of this is not the case. But it is not a property of the queen that leads to a white win, instead mainly the position of the black king together with the special case of white being able to enforce the Bf6/Rg7/Kh8 position is responsible. By the way, even evaluating the queen as, say, only 7.0 would not help to find the white win early based on static evaluation, it is necessary to see the forced tactical line where black loses its queen due to discovered check.

I still don't see why evaluating the queen as only 8.5 in the endgame should have any value for more general cases.

Sven

[pichy]

Here is a great example and I decided to played it versus Komodo, I was playing White:

[Event "Queen vs Bishop+Rook"]
[Site "?"]
[Date "2012.04.08"]
[Round "?"]
[White "Pichy"]
[Black "Komodo3-32"]
[Result "1-0"]
[BlackElo "3080"]
[Time "18:17:34"]
[WhiteElo "2200"]
[TimeControl "0+20"]
[SetUp "1"]
[FEN "6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "26"]
[WhiteType "Pichy"]
[BlackType "program"]

[d]6R1/8/2pB3k/2P4p/5p1q/5P2/4P1K1/8 w - - 0 1

1. Bf8+ Kh7 2. Rg7+ Kh8 3. Be7 Qe1 4. Bf6 Qxe2+ 5. Kh1 Qc4 6. Ba1 h4 7. Bf6 h3 8. Ba1 h2 9. Bf6 Qe6 10. Bd4 Qh6 11. Rg2+ Kh7 12. Rxh2 Kg6 13. Rxh6+ Kxh6 *

O.k., this example might look better but it is still one of those 0.0x% where the exact value of the queen does not matter much due to "short-term" tactical issues (though not so short-term here considering the number of plies until the black queen is finally lost). Here these issues are caused by the extremely weak position of the black king into which black is being forced tactically with the first four white moves.

You are presenting a nice endgame study where the solution typically deviates from what you expect as a normal outcome for the vast majority of cases. With the given material of QPPP vs RBPPP and given typical piece values of Q=9.75, R=5.0, B=3.25 you get an advantage of +1.5 pawns for black. Without any further positional scoring the overall evaluation would be right within the draw margin for positions with equal number of pawns, so we would expect a draw, albeit with some black advantage. But in the given position the king safety is extremely bad for black, as stated above, so even a static evaluation would show at least a small advantage for black.

Queens can be stronger than rook + bishop, though, in positions with this kind of material, for instance if
- their own king is safe,
- rook and bishop are not cooperating well, and/or
- the rook/bishop side has several weak points that can be attacked simultaneously be the queen, showing the main strength of a queen which is its ability to act at many places on the board "at the same time".

In the example you gave above all of this is not the case. But it is not a property of the queen that leads to a white win, instead mainly the position of the black king together with the special case of white being able to enforce the Bf6/Rg7/Kh8 position is responsible. By the way, even evaluating the queen as, say, only 7.0 would not help to find the white win early based on static evaluation, it is necessary to see the forced tactical line where black loses its queen due to discovered check.

I still don't see why evaluating the queen as only 8.5 in the endgame should have any value for more general cases.

Sven

I give up, I am going to sleep

[Adam Hair]

Yes, you have supplied several positions where the side with the queen places too much value on her. However, if you analyze the results of many games, the value of the queen is close to the value of two rooks (no minors on board).

If the value of the queen was set to 8.5, there would be many endgame positions where the queen would be undervalued.

[pichy]

Yes, you have supplied several positions where the side with the queen places too much value on her. However, if you analyze the results of many games, the value of the queen is close to the value of two rooks (no minors on board).

If the value of the queen was set to 8.5, there would be many endgame positions where the queen would be undervalued.

I have more than 60 endgames so far where the value of the Queen is overvalued, but I have not seen too many instances where the value of the Queen is undervalued after I made change to the value of the Queen by setting it to equal 8.5. What I do believe is that the value of the Rook should be set to 5.25 and the bishop to 3.35 in the endgame if the Queen value remains equal to 9.0.