CCRL Discussion Board

I propose a new topic about chess endgame studies composed with tablebase help.

Árpád Rusz - Sakkvilág, 2010
8/3b2KP/8/2R5/8/5R2/pk6/7r

White to move and win

After 1.Ra5 a1Q 2.Rf2! Kb1 3.Ra1 Ka1 4.Rf1! Rf1 5.h8Q Be6 we get the following position:

7Q/6K1/4b3/8/8/8/8/k4r2


Try to guess which is the only move by the white king that wins!

How do I get the FEN out of the diagram?
(I can see it when I put the mouse in the diagram, but I cannot copy it...)

If you click the Quote button on the post with diagrams you can copy the FEN.

I now added automatic display of the FEN string above the diagram. Enjoy.

Kirill Kryukov wrote:I now added automatic display of the FEN string above the diagram. Enjoy.

Thanks !!!!!!!!!!!
Good move !!!!!!!!!!!!

Arpad Rusz wrote:I propose a new topic about chess endgame studies composed with tablebase help.

Nice, I always enjoy these irrational positions. Please post more if you'll find any! Perhaps some day I'll try to use them for the EGTB test suite (when I'll get time to resurrect the project).

ernest wrote:Thanks !!!!!!!!!!!
Good move !!!!!!!!!!!!

I should have done it years ago. Welcome!

7Q/6K1/4b3/8/8/8/8/k4r2


The question was: "Which is the only move by the white king that wins?"

The answer:
Only 7. Kh7!! is winning! 7...Ka2 8. Qa8+ Kb3 (8... Kb2 9. Qg2+ +-) 9. Qb8+! (9. Qb7+? Kc3 10. Qc6+ Bc4 =) 9... Kc4 (9... Kc3 10. Qe5+ +-) 10. Qc7+ Kd5 (10... Kd4 11. Qb6+ +-) 11. Qa5+ Ke4 12.Qb4+ (12. Qa4+? Kd3 13. Qa6+ Bc4 =) 12... Kf3 (12... Kd3 13. Qd6+ +-) 13.Qf8+ (13. Qb7+? Kg3 =) 13... Kg2 (13... Ke2 14. Qe8 +-) 14. Qg7+ Kh1 (14... Kh2 15. Qe5+ +-) 15. Qh6+! +- Only now we see the reason for not playing 7.Kh6?

(As there are no winning queen moves, playing 7.Kh7!! is the only way to win.)

More compact format:
8/3b2KP/8/2R5/8/5R2/pk6/7r w - - acd 29/81; acn 62258191501; acs 432000; bm Ra5; c0 "Árpád Rusz - Sakkvilág, 2010"; ce 1355; pv Ra5 a1=N Rf2+ Kb1 Rf1+ Rxf1 Rxa1+ Kxa1 h8=Q Be6 Kh7+ Ka2 Qa8+ Kb3 Qb8+ Kc4 Qc7+ Kd5 Qa5+ Ke4 Qb4+ Kf3 Qf8+ Ke2 Qe8 Kd3 Qxe6 Rf4 Kg6 Kd4 Kg5 Re4;

dann corbit wrote:More compact format:
8/3b2KP/8/2R5/8/5R2/pk6/7r w - - acd 29/81; acn 62258191501; acs 432000; bm Ra5; c0 "Árpád Rusz - Sakkvilág, 2010"; ce 1355; pv Ra5 a1=N Rf2+ Kb1 Rf1+ Rxf1 Rxa1+ Kxa1 h8=Q Be6 Kh7+ Ka2 Qa8+ Kb3 Qb8+ Kc4 Qc7+ Kd5 Qa5+ Ke4 Qb4+ Kf3 Qf8+ Ke2 Qe8 Kd3 Qxe6 Rf4 Kg6 Kd4 Kg5 Re4;

1...a1N is a weak move, 2.h8Q Rxh8 3.Kxh8 wins...
Engines beat us but they don't really understand chess.

I have for a long time deprecated the epithet 'EGTB Study', assigned originally by John Roycroft to any study which entered the sub-7-man 'EGT'd' zone of chess.

AJR seemed to be implying that such studies should not be allowed - or should be confined to their own ghetto, and that their composers were somehow cheating.

This was something of a volte-face for AJR who had always championed the creation and use of EGTs.

Unfortunately, his objections to EGT-assisted creation are no more than luddite. A study is much more than a position, and the fact that the value/depth of a position can be looked up in a file does not mean that a study involving it is 'anticipated'. This has now been made clear by what was the FIDE PCCC.

New methods and tools change old practices, and this is true also of study composition.

So, as you can tell, I'm really not in favour of dividing up studies into those that are independent of EGT'd-chess and those that are not.

g

Since 'studies' and sub-7-man chess are being discussed in the same breath here ...

... I should say that I've just written a short piece for the ICGA Journal reporting the data-mining of Harold van der Heijden's excellent ENDGAME STUDY DATABASE IV.

Eiko Bleicher and I evaluated all sub-7-man positions in Harold's corpus to see which studies contained:

- mainline positions with values incompatible with the stipulation of the study
- mainline positions where there was more than one DTM-optimal move
- moves which were not DTM-optimal, indicating at least a dual of some significance

The 3-pager in the December issue of the ICGA Journal basically reports on studies cooked by this EGT-based analysis.

g

The problem was that there were some composers who added a few moves (usually forced and with captures) to a position from the reciprocal zugzwang list and called that a "study". But AJR was wrong: there is nothing bad with a good EGTB study.

I let you to discover the solution of the following study from Heijden's database:

S. Osintsev (1st Prize Kaljagin JT 2009)
8/4p3/8/2K1P3/N7/4p3/4k3/8 w - - 0 1

White to move and win

After 1.Nb2! there are two nice variations starting with 1...Kd2 and 1...Kf3.

The zugzwang is a big theme in studies, and AJR used to publish lists of zugs - often in EG. Unfortunately, these lists were often found to have errors.

See http://ilk.uvt.nl/icga/games/chess/endgames.php for the complete list of value-critical ('Type A') zugs.

See http://centaur.reading.ac.uk/view/creat ... 00763.html and specifically http://centaur.reading.ac.uk/4518/ for definitions and examples of Type B and Type C zugzwangs - and an explanation of why the term 'reciprocal zugzwang' is not 100.000% apposite!


I would like to have examples of studies which anyone thinks have been over-assisted by EGTs, and examples of judges who thought so.

A fundamental problem in the studies community is that judges appear to have very different criteria, all of which are 'secret'. A better way forward might be that each judge declares a one-page manifesto on their judging stance.

John Nunn has written most clearly on the issues around studies and EGTs, and I agree with his perspective. The means of composition is irrelevant: it is the output that is being judged. Perhaps the confusions between 'process' and 'output' is confused by the fact that the word 'composition' describes both.

There is another instrinsic problem. There are certainly studies which are proved by EGTs to have an absolutely unique solution - not even a single 'ignorable dual' - but which have too long a solution for human consumption. It may be because of this that they are deemed not to have artistic merit, or maybe they did not have artistic merit anyway.

'Length' is not the only technical parameter testing human understanding. 'Width' is another. If a thicket of side-variations have to be explained away, the explanation of the solution becomes indigestible. The composer needs to explain away not only alternative moves for White which meet the goal but moves which appear to meet the goal but in fact do not.


On the other side of the study/EGT problem scenario, I found AJR's arguments, effectively against studies now going into 'EGT territory', to be too aggressive: at times they degenerated into goal-based rhetoric. I was reminded of the Stalinistic logic that sent more than one Russian study composer to the gulag because their compositions were not accessible enough to 'the people'.

I challenged AJR to produce an 'EGT study' based on casually picking up a position from an EGT and turning it into a study: he hasn't done so and I don't believe he would find it easy.

AJR's stance caused a most unfortunate hiatus in the judging of the Bent Memorial Tournament, when the other two judges did not agree with his stance on studies going into 6-man territory. All of this subtracts from AJR's reputation which is a pity as he has given much to the world of the chess study.

g

... #75,270 in HvdH's Endgame Study Database IV.

Thank you for highlighting that: it's very special, and dbIV/EGTs help to understand why it is so.

A combination of the Trebuchet and N-converting-P themes with a neat preamble. Excellent

g

I have modified this position (Z07) found here: http://centaur.reading.ac.uk/4518

8/8/8/2p5/1pP1p3/kP2P3/Pp1P4/1K6 b - c3 0 1


Now the following position is a chess study with only one winning move (1.c4!):

8/8/2p5/2p1p3/1p2p3/kP2P3/PpPP4/1K6 w - - 0 1


1.c4! (1.c3? c4! 2.cxb4 cxb3 3.axb3 Kxb3 -+) 1...bxc3 e.p. 2.dxc3 c4 3.b4 (3.bxc4? c5 -+) 3...Ka4 4.Kxb2 Kb5 5.Ka3 Kb6 6.Ka4 Ka6 7.b5+! (7.a3? Kb6 8.b5 c5 stalemate) 7...cxb5+ 8.Kb4 Kb6 9.a4 bxa4 10.Kxa4 Kc5 11.Ka5 Kc6 12.Kb4 Kd5 13.Kb5 1-0

Do you know any other examples of this kind?

Position #405206 in the Sub-7-man DTM metric zugs pgn is wrongly listed as type A8.

Yes: you have a study based on Noam Elkies' zugzwang position 'Z07'.

Noam was focused on creating zugs of types a4-a6 rather than creating studies.

Not sure what the authorship attribution should be: 'Ruiz after Elkies' I think. I'm certainly not in the back-story of your study's creation. As to its quality, I'm not sure there's enough of a 'surprise element' but technically, it's fine.

I think 9.a3 works as well but may be a 'minor dual'.
12.Ka4 just puts a time-wasting loop of 4 ply into the line.
14.Kb6 (beyond the 'unique line') is better than the obvious 14.Kxc4 but the latter also wins.


I think all Noam's zug-compositions were in http://centaur.reading.ac.uk/4518


I assume your comment about position #405206 is correct: there isn't a type A8 but it sounds like it's type A4, A5 or A6. I'll check later: thanks.

g

There is also possible to have two e.p. captures in a mutual zugzwang position.

8/8/1p1p1p1p/1P1k1P1P/2p1p3/2P1P2P/2PPK3/8 w - - 0 1


Only 1.d4! (reciprocal zugzwang!) wins. White needs the extra tempo h3-h4 in some variations.

The author of the following study presumed that the RRN/RR 7 piece ending is a general draw...

E.Vlasák (Hlinka JT, 2004)
8/8/kr6/8/1P3r2/4R3/3N4/3KR3 w - - 0 1

White wins

1. Ra3+ Kb7 2. Re7+ Kc6 3. Rc3+ Kd6 4. Rc4 Rxb4 5.Ree4!

8/8/3k4/8/1rR1Rr2/8/3N4/3K4 b - - 0 5


5... Rxc4 (5... Rxe4 6.Nxe4+ Kd5 7. Rxb4) 6. Nxc4+ Kd5 7. Rxf4 +-

It would be very nice but Bourzutschky and Konoval proved that RRN/RR is a general win, so white wins even if it loses the pawn. The study is demolished...

I was trying to save it: here's my position (first publication):

Á.Rusz (after E.Vlasák)
5R2/2r5/1N2P1r1/2P1kbP1/8/4b3/8/3RK3 w - - 0 1

White wins

1. Nc4+ Kxe6 (1... Ke4 2. Nd6+ Ke5 3. Rxf5+ Kxe6 4. Rdd5 +-) 2. Rd6+ Ke7 3. Rxf5 Rxg5 (3... Rxc5 4. Rxc5 Rxd6 5. Nxe3 +-) 4. Nxe3 Rxc5 5. Rdd5!

8/4k3/8/2rR1Rr1/8/4N3/8/4K3 b - - 0 5


5...Rc1+ (5... Rxd5 6. Nxd5+ Ke6 7. Rxg5 +-; 5... Rxf5 6. Nxf5+ Ke6 7. Rxc5 +-; 5... Rg1+ 6. Kf2 Rxd5 7. Nxd5+ Ke6 8.Rf6+ +-) 6. Kd2 Rxf5 7. Nxf5+ Ke6 8. Rd6+ +-

While in the first study 5.Rcc7 is a dual, here 5.Rff6? is only a draw!

Thematic try: 1. Rd5+? Ke4! (1... Kxe6? 2. Rfxf5 +-) 2. Rfxf5 Rxe6 3. Nc4 Rxc5 (3... Bf4!?) 4. Nxe3

8/8/4r3/2rR1RP1/4k3/4N3/8/4K3 b - - 0 4


4...Rxd5 5. Rxd5 Kxe3 =