How are we supposed to make this intereactive? Or are we just suppose to make it instructional?
Endgame Positions
Sort:
Nov 13, 2008
0
#3
immortalgamer wrote:
How are we supposed to make this intereactive? Or are we just suppose to make it instructional?
you just need to help me come up with the FEN string and objective. then we will build the application that makes this amazing :)
Nov 13, 2008
0
#4
I do not understand the point of that King and Pawn ending, since any move by White wins.
More interesting would be removing the h4 pawn, which means White has only one winning move.
Nov 13, 2008
0
#5
Another interesting position is K + Q vs. K + R. I know Walter Browne got so good at that ending (from playing against a tablebase), that he was virtually unbeatable with the Rook side in 5 minute chess, even against grandmasters.
These endgames you (erik) mention are all "book" endgames which the user just has to know, to be able to win them. So either he mates the computer, or he struggles and tries various times and doesn't succeed, because he doesn't know. So imo you could join them in one category with, although ranging in difficulty, endgames the player should just know. No "missing a winning move" or "overlooking a defense", just either the player knows what he's doing, or he's just trying and guessing.
Nov 13, 2008
0
#7
In between "knows what he's doing" and "trying and guessing" is "thinking ahead a little bit and applying some knowledge". The latter is better for incrementally learning, than memorizing some technique, I believe.
It is also good to distinguish different kinds of endings, because some have radically different principles or techniques.
Checkmating
- K + Q vs. K
- K + Q vs. K + B
- K + Q vs. K + N
- K + Q vs. K + R
- K + R + R vs. K
- K + R vs. K
- K + R vs. K + N (is this a win or a draw?)
- K + R vs. K + B (only certain positions)
- K + B + B vs. K
- K + B + B vs. K + N
- K + B + N vs. K
- K + N + N vs. K + P (only certain positions)
- ...
Preventing checkmate (easy to hard)
- K + B vs. K + R
- K + R vs. K + R + N
- K + R vs. K + R + B
- ...
Pawn promotion
- K + P vs. K
- K + a/hP + B vs. K
- K + a/hP + N vs. K
- K + R + P vs. K + R (Lucena)
- K + R + aP + bP vs. K + R
- K + Q + P + P vs. K + Q
- K + Q + P vs. K + Q
- K + a/c/f/hP vs. K + Q (drawing)
- K + B/N + P vs. K + B/N
- Breakthrough positions
- Zugzwang positions
- ...
Preventing pawn promotion
- K vs. K + P
- K + R vs. K + R + P (Philidor)
- K + Q vs. K + P (winning)
- K + B/N vs. K + B/N + P
- ...
Others
- Kamikaze rook (desperado): K + R vs. K + Q, K + R vs. K + R + P, K + R vs. K + R + B
- ...
DavidForthoffer wrote:
I do not understand the point of that King and Pawn ending, since any move by White wins.
More interesting would be removing the h4 pawn, which means White has only one winning move.
I think Erik's idea was more like what I've posted here.
Certain endgame fortresses are also important to know, I think; things like bishop and wrong rook pawn versus king or rook and pawn on the second rank versus queen. I suppose the question with these sorts of positions is: should the user be reqiured to play the necessary moves to create the fortress, or should they just be given the fortress ready-made, with the task being simply to hold it? For example, we could give the user a position like this one (just a rough idea, it could probably be improved):
Or we could give them something like the one at the bottom (sorry for my rubbish formatting), where everything is already done, and the task is just to not mess it up. Of course, this position is far easier, but the point is much clearer, and you must understand why this position is a draw before you can hope to solve problems where the fortress has to be created.
DavidForthoffer: Certain endgame fortresses are also important to know, I think...
A very Interesting topic that I for one would love to know more about -- and I have't seen it taught before (not that I've looked) --
It also touches on a concern that Camembert has brought up in another thread (the Problems with Engine thread) ... which is to keep in mind that these postions must be played against an engine, and engines try to find the best move, which is not necessarily the most instructive move.
Do engines play against fortresses properly? I know currently engines don't understand/evaluate some fortresses properly. This is on my mind because it just happened in fact in the Anand Kramnik world championship match where Kramnik built a fortress and all the Patzers (me, etc.) were looking at our engines going, "ohhh, Kramnik's lost another one...." and of course the result was 1/2 -1/2
Come to think of it, that Anand Kramnik position might make a good Technique Test position (though I think it would be a very difficult test). Now, which game was it again?
Nov 14, 2008
0
#12
1. Be5 Qxe5 {The only reply} (1... Rxe5 2. Rf8+ Ka7 3. Nc8+ Ka8 4. Ne7+ {
with a devastating fork on c6.}) 2. Rf8+ Ka7 3. Nc8+ Ka8 4. Nb6+ Ka7 5. Ra8+ $1
Kxb6 6. Rxa6+ Kxa6 7. axb5+ Kb6 8. Ra6+ Kxb5 9. Ra5+ Kc4 10. Rxe5
with a devastating fork on c6.}) 2. Rf8+ Ka7 3. Nc8+ Ka8 4. Nb6+ Ka7 5. Ra8+ $1
Kxb6 6. Rxa6+ Kxa6 7. axb5+ Kb6 8. Ra6+ Kxb5 9. Ra5+ Kc4 10. Rxe5
Then You're on your own!! 
For the basic mates, I'd suggest the following positions. Obviously, almost any position would do, but these are the positions where the route to mate is the longest (according to tablebases). I suppose the ideal would be to give the user a different position each time, but maybe that's a different sort of application to the one we're dealing with here:
Three other pretty fundamental positions (I don't know if the computer would play a sensible defence on the last one):
Nov 14, 2008
0
#15
Lesson: So what does it take to win?
In most cases, you need to be able to queen a pawn.
However, as with most rules, this one is most helpful when making transistions, that is, when changing the nature of the struggle.
(Trading pieces for instance)
In the diagram you see a strong GM throw away his last chance to advance in the candidates' matches by forcing an endgame that wasn't nearly as favorable as the middlegame he was in.
Why?
Because he went with the instinct based on the rule, that told him that if you capture the other guy's queen, there is no endgame. Instincts are dangerous to have in the endgame.
In this game white played f8=Q+
Much stronger is 1. h5!, which probably would have caused immediate resignation.
1..... Qxf8
2. Rxf8 Kxf8
3. Qxe6 ( of course Qxg6?? Rg7 loses) ....Nxh4
White can only try to win now by grabbing pawns and trying to promote his own a or d pawn. Black set up an impregnable fortress and forced the draw.
Nov 16, 2008
0
#16
Answer key
1. Be5+ Ka7 (1... Kc8 2. Bc7 b4 3. axb4 a3 4. b5 a2 5. b6 a1=Q 6. b7#)
2. Bc7 Ka8 (2... Ka6 3. Bb6 b4 4. axb4 a3 5. b5#)
3. Kb6 $1 b4 4. axb4 a3 5. b5 a2 6. Be5 1-0
Nov 16, 2008
0
#17
JG27Pyth wrote:
DavidForthoffer: Certain endgame fortresses are also important to know, I think...
A very Interesting topic that I for one would love to know more about -- and I have't seen it taught before (not that I've looked)
Andre Cheron had a great 5-volume set on endgames with lots of these kinds of positions. It is early 1900's, and very likely in the public domain by now.
immortalgamer wrote:
Lesson: So what does it take to win?
In most cases, you need to be able to queen a pawn.
However, as with most rules, this one is most helpful when making transistions, that is, when changing the nature of the struggle.
(Trading pieces for instance)
In the diagram you see a strong GM throw away his last chance to advance in the candidates' matches by forcing an endgame that wasn't nearly as favorable as the middlegame he was in.
Why?
Because he went with the instinct based on the rule, that told him that if you capture the other guy's queen, there is no endgame. Instincts are dangerous to have in the endgame.
In this game white played f8=Q+
Much stronger is 1. h5!, which probably would have caused immediate resignation.
1..... Qxf8
2. Rxf8 Kxf8
3. Qxe6 ( of course Qxg6?? Rg7 loses) ....Nxh4
White can only try to win now by grabbing pawns and trying to promote his own a or d pawn. Black set up an impregnable fortress and forced the draw.
Although I agree h5 would have been better, I don't agree 1. f8=Q is horrible, throws away the win, relies too much on bad instincts etc... I'd like to see anyone defend the draw as black after 3... Nxh4 4. Qc8+ Kg7 5. Qxb7+ followed by 6. Qxd5, when white has two extra pawns and a queen for a rook and knight. Unless you can prove to me that it is not winnable for white, you can't include this in the database (since if it is winnable, 1. f8=Q+ is much easier to calculate than 1. h5)
Nov 16, 2008
0
#20
Phobetor wrote:
Although I agree h5 would have been better, I don't agree 1. f8=Q is horrible, throws away the win, relies too much on bad instincts etc... I'd like to see anyone defend the draw as black after 3... Nxh4 4. Qc8+ Kg7 5. Qxb7+ followed by 6. Qxd5, when white has two extra pawns and a queen for a rook and knight. Unless you can prove to me that it is not winnable for white, you can't include this in the database (since if it is winnable, 1. f8=Q+ is much easier to calculate than 1. h5)
Since you would like to see anyone defend the draw. How about Korchnoi? Does that work for you?
Here is the game: http://www.chessgames.com/perl/chessgame?gid=1083569
I'm just brainstorming here about the important endings in chess:
Checkmating
K + R vs. K
K + R + R vs. K
K + Q vs. K
K + B + B vs. K
K + B + N vs. K
Pawn promotion
King with pawn in FRONT (try to draw being a pawn down)
King in front of pawn (to win)
3 pawns vs. 3 pawns (capablanca breakthru i believe?)
lucena position
philidor's position
opposition
breakthru pawn sacrifice
... just a start! :)