Another endgame that the Rybka family fails at

[peter]

Have you or your engine seen all this too?

Well indeed they have, beeing shown your variants first and countig the moves given backward together with 6MOB.



Here is Deep Fritz' output being gone back your variant with 6MOB on harddisk at 3.Rf3?

2r5/5k2/8/5PP1/8/2p2R2/2K5/8 b - - 0 1

3...Tc5 4.Tf4 Te5 5.Tf1 Te2+ 6.Kd3 Tg2 7.g6+ Kf6 8.Kxc3
= (0.00) Tiefe: 21/43 00:00:42 16378kN, tb=50102
Of course, that's not as instructiv as your's, but the eval is o.k.

The initial endgame is not a study: white doesn't win by an unique series of moves (if it wins at all). Or at least it has not been proved yet to be a study. But it is a nice endgame.
A study doesn't need to be hard to solve to be a good one. Think of the famous Saavedra position. It's a four piece tablebase position.

Of course, very right said.
Thanks a lot for sharing your beautiful study with us, as a matter of fact I didn't realize that all before, but I have to admit, I didn't look as close, as I should have, thought, you'ld just wanted to show me the original endgame was white win, sorry for getting you wrong.
But as a real expert for endgame- puzzles, don't you think yourself, Robert's one is remis indeed?'
I for myself, I'm rather sure about it.

[Arpad Rusz]

All I can say now that the endgame is a draw... or a win for white.
I was trying to find the truth about Robert's endgame when my attention was distracted by the discovery of the positional draw with Kf7!-Kg7!.
I don't remember exactly how I get there, the line was something like this (but without that big blunders from the following help-play):

67...Rxc3+ 68.f3 Ra3 69.Rc7 Rc3!? 70.Rc5+ Kf6 71.Kf4 Ke6 72.g4 Kd6! 73.Rc8 Kd7!? the idea of black's play 74.Ra8 Rb3 75.Ra2? [75.Ra4! wins] 75...c3 76.Rc2 Ke6 77.Ke4 Rb4+ 78.Ke3 Rc4 79.f4 Rc8 80.Kd3 Rg8 81.Rg2 Rc8 82.f5+ Kf7?? 83.g5 Kg7! 84.Rf2 Kf7! 85.Rg2 Kg7 positional draw

[peter]

All I can say now that the endgame is a draw... or a win for white.

I was trying to find the truth about Robert's endgame when my attention was distracted by the discovery of the positional draw with Kf7!-Kg7!.
I don't remember exactly how I get there, the line was something like this (but without that big blunders from the following help-play):

67...Rxc3+ 68.f3 Ra3 69.Rc7 Rc3!? 70.Rc5+ Kf6 71.Kf4 Ke6 72.g4 Kd6! 73.Rc8 Kd7!? the idea of black's play 74.Ra8 Rb3 75.Ra2? [75.Ra4! wins] 75...c3 76.Rc2 Ke6 77.Ke4 Rb4+ 78.Ke3 Rc4 79.f4 Rc8 80.Kd3 Rg8 81.Rg2 Rc8 82.f5+ Kf7?? 83.g5 Kg7! 84.Rf2 Kf7! 85.Rg2 Kg7 positional draw

I see, that the thing is of somewhat other kind of interest for a composer.

Nice idea yet, but how the hell do you come to 68...Ra3?
I mean, I see it helps composing the end, but as for a game you simply wouldn't want to loose, wouldn't you prefer to at least try to advance the black pawn before the white ones both have crossed the middle of the board?

Anyhow, your variants help to understand, how close the original position is to a white win, if I don't find a refutation to your variant (really ending at 75. Ra4 of course), might well be, this one and only tempo (as a matter of fact it's the two of them, first one 68...Ra3? cause that makes 69...Rc3 as a second one necessary) you give away could make the difference.
Thanks again.

[peter]

...might well be, this one and only tempo (as a matter of fact it's the two of them, first one 68...Ra3? cause that makes 69...Rc3 as a second one necessary) you give away could make the difference.

However, going back your variant to 68...Ra3, this one is evaluated with +1.72 by Fritz 10 with 6MOB, after going into and backwards in my own variants after 68.(3.)...Rc1, here we have a clear difference in eval again:


8/5R2/8/6k1/2p5/5PK1/6P1/2r5 w - - 0 1

Analysis by Deep Fritz 10:


3.Tc7 Kf5 4.Tc5+ Ke6 5.Kf4 Kd6 6.Tc8 c3 7.g4 Kd5 8.Ke3 c2 9.Tc3 Ke5 10.f4+ Kd5
= (0.00) Tiefe: 24/76 00:05:40 980mN, tb=1612566

I'd really come to the conclusion, that simply giving away two tempos like you do, kills Black.
Maybe one could make a study out of the whole thing really.
Who says, a study must be beautiful and understandable allways.
This one was just one about the edge of predictability in chess endings with only 7 pieces on board.
What do you think about another one with Queens instead of Rooks?

8/5Q2/8/6k1/2p5/2q3K1/5PP1/8 w - - 0 1
Isn't it really crazy how much easier that makes the position?
Solution tomorrow won't be necessary, will it?


P.S. You'ld just have to find the starting position for it with a sinlge best move and one doubtless main variant...Must show the whole piece to Gerhard Josten, who was so kind to let me take part in his fine new book "A Study A Piece" , to Mighail Neghina and to Olaf Jenkner.
http://www.mynetcologne.de/~nc-jostenge/
Here's a german recension of it:
http://glareanverlag.wordpress.com/2010 ... n-magazin/

[Arpad Rusz]

I have checked the endgame by Deep Shredder 12 using 6 men tablebases, and indeed it seems to be a draw after Rc1.
Why I have tried Ra3? Because after Rc1, f4+ looked as winning at the first sight. Ra3 (or Rd3) keeps that pawn pinned.
It is very interesting that in many variations the best square for the black pawn seems to be c4!

Yes I know that book, Gerhard Josten invited me too to write something, but I didn't find time to do it. What study have you choose for the book?

[peter]

I have checked the endgame by Deep Shredder 12 using 6 men tablebases, and indeed it seems to be a draw after Rc1.
Why I have tried Ra3? Because after Rc1, f4+ looked as winning at the first sight. Ra3 (or Rd3) keeps that pawn pinned.
It is very interesting that in many variations the best square for the black pawn seems to be c4!

I'm more and more convinced, that it's just that one tempo, that makes the difference. I even had the illusion for a while, changing the right of first move and giving white one more half ply, it could be seen as a study itself:

[Event "Limit of 7 Men's Human Predictability"]
[Site "?"]
[Date "2010.05.21"]
[Round "?"]
[White "Peter, Martan"]
[Black "Robert, Flesher"]
[Result "1-0"]
[Annotator "Martan,Peter"]
[SetUp "1"]
[FEN "8/5R2/8/6k1/2p5/2r2PK1/6P1/8 w - - 0 1"]
[PlyCount "23"]

1. Rf8 (1. Rc7 $2 Rc2 2. Rc5+ Kf6 3. f4 Ke7 4. Re5+ Kd6 5. Kf3 Rc1 6. Re2 Kd5
7. Rd2+ Ke6 8. Rd8 c3 9. Rc8 Kd5 10. g4 c2) (1. Rg7+ $2 Kf6 2. Rg4 (2. Rg8 Ke5
3. Kf2 Kd4 4. Rd8+ Ke5 5. Ke2 Rc2+ 6. Rd2 Rc3 7. g4 Ra3 8. Rc2 Kf4) (2. Rc7 Ke5
3. Kg4 Kd6 4. Rc8 Kd7 5. Ra8 Rd3 6. Kf5 c3 7. Ra1 Kd6 8. Ke4 Rd2 9. g4 c2 10.
Rc1 Ke6 11. f4 Kf6 12. Ke3 Rg2) 2... Rc2 3. Re4 c3 4. Rc4 Ke6 5. Rc5 Kd6 6. Rc4
Ke6) (1. Re7 $2 Kf6 2. Re8 (2. Re2 Rd3 3. Kf4 c3 4. Ke4 Rd7 5. Ke3 (5. Ra2 Rd2
6. Ra6+ Kg5 7. g3 c2 8. Rc6 Re2+ 9. Kd3 Rg2 10. Rc5+ Kf6 11. g4 Rf2 12. Ke4
Re2+ 13. Kd4 Rd2+ 14. Ke4 Re2+ 15. Kf4 Ke6) 5... Kf5 6. g3 (6. Ra2 Rc7 7. Kd3
Kf4 8. Ra4+ Kg3 9. Rg4+ Kf2 10. Kc2 Ke3 11. Rg8 Kf4 12. Rg6 Rc5 13. Rg7 Rc8 14.
Rg4+ Ke3) 6... Rg7 7. Rg2 Re7+ 8. Kd3 Re1 9. Kxc3 Rf1 10. f4 Kg4 11. Kd4 Kf3
12. Rd2 Kxg3) 2... Rc2 3. f4 c3 4. Kf3 Kf7 5. Rc8 Ke6 6. g3 Rc1 7. Kg4 c2) (1.
Kh3 Rc1 2. g4 Kg6 3. Rc7 Kf6 4. Kh4 Rh1+ 5. Kg3 Rc1 6. f4 Ke6 7. f5+ Kd6 8. Rc8
Kd7 9. Rh8 c3 10. Rh2 Ke7 11. Kh4) 1... Rd3 (1... Rc1 $2 2. Kf2 Rc2+ (2... Rc3
3. g4) 3. Ke3 Kg6 4. g4) 2. Kf2 c3 3. g4 (3. Rc8 $2 Kf4 4. g3+ (4. Rc4+ Ke5 5.
g4 Kd5 6. Rc8 Kd4 7. f4 Rd2+ 8. Kf3 c2 9. Rd8+ Kc4 10. Rc8+ (10. Rxd2 c1=Q $19)
10... Kd4) 4... Ke5 5. Ke2 Rd2+ 6. Ke3 Rg2 7. g4 c2 8. f4+ Kd5 9. Kd3 (9. g5
Rg3+ 10. Kd2 Ke4) 9... Rxg4) 3... c2 (3... Rd1 4. Rc8 Rc1 5. Rc5+ Kg6 6. f4 Kf7
7. Rc6 Ke7 8. f5 Kd7 9. Rc4 Kd6 10. g5 Kd5 11. g6 Kxc4 12. g7) (3... Rd2+ 4.
Kg3 Rd1 (4... c2 5. f4+ Kg6 6. Rc8 Kf6 7. Rc6+ Ke7 8. Kh4 Rg2 9. Kg5 Kd7 10.
Rc4 Ke6 11. f5+ Ke7 12. Rc8 Kd7 13. Rc3) 5. Rc8 Rc1 (5... Rg1+ 6. Kf2 Rc1 7.
Rc5+) 6. f4+) 4. Rf5+ {#46} (4. Rc8 {#45}) 4... Kg6 5. Rc5 Rd2+ 6. Kg3 Kf7 7.
f4 Ke6 8. Kh4 Kd6 9. Rc8 Ke6 10. f5+ Kf6 11. Rc6+ Ke5 12. Kg5 {#36} (12. Rc5+
Kd6 13. Rc3 Ke5 14. Kg5 Rg2 15. Rc8 Rd2 {#34} (15... Kd6 {#33} 16. Kf6 Rd2 17.
Kf7 Rh2 (17... Ke5 {Zappa+6MOB #24} 18. f6 Kd6 (18... Rh2 19. g5 {#21}) (18...
Kf4 {#22} 19. Kg6 Rg2 (19... Kxg4 {#20} 20. f7)) 19. g5 Rd5 20. Rxc2 {#21} Rxg5
21. Rd2+) 18. Kg6 {#26})) 1-0



Finally I found 1.Kh3 winning too, so it's not single best move, probably there will be some more and of course it would'nt have fulfilles any other criteria of a study.
But as an example of predictability of 7MOB I still think it's fascinating.
If white wins with changed right of first move in less ways than there are remis variants, it's one more argument for me, Robert's endgame is proved remis.

Yes I know that book, Gerhard Josten invited me too to write something, but I didn't find time to do it. What study have you choose for the book?

I didn't have much choice. I helped Mighai Neghina in two of his studies and he choose the first one of those, first published in Glarean journal.

http://glareanverlag.wordpress.com/2009 ... n_neghina/

By the way, here's the other one:

http://glareanverlag.wordpress.com/?s=neghina