A test position of mine from last may 2010

[Robert Flesher]

I do not believe we ever discovered if white white or draws. I still believe white wins here! Engines may lead to false draws or wins due to weak play, prior to getting a tablebase position.

http://www.talkchess.com/forum/viewtopi ... highlight=


[d]8/5R2/8/6k1/2p5/2P3K1/2r2PP1/8 b - - 0 1

[Arpad Rusz]

1... Rxc3+ 2. f3 Rc1! 3. f4+ Kg6 4. Re7 Kf6 5. Re5 c3! (5... Rc2? 6. Kf3 Rc3+ 7. Kg4 Rc2 8. g3 Rc1 9. Rc5 Ke6 10. f5+ Kd6 11. Rc8 Kd7 12. Ra8 c3 13. Ra7+! Kd6 14. Ra2 Rd1 15. Kg5 Rd4 16. Rc2 Rc4 17. Kf6 Kd5 18. g4 Ke4 19. Rc1 +-) 6. Rc5 Rc2! 7. Kf3 Ke6 8. g3 Rc1 9. Kg4 c2 10. f5+ Kf7 (10... Kf6? 11. Rc7 zz +-) 11. Rc7+ Kf6 zz =

[Tom Barrister]

1 ... Rxb3+
2 f3 Rc1
3 f4+ Kg6
4 Re7 Kf6

Black's King will eventually need to go support the c-pawn; otherwise, White will simply advance his pawns. Black's goal is to force the trade of Rook for pawn before White's pawns are too far along to be stopped.

5 Re5 c3
6 Rc5

White needs to have the pawn under attack before it reaches c2.

6 ... Ke6

Continuing towards the c-pawn. If 6 ... c2, 7 Kg4! (the reason White's Rook needed to have the c-pawn under attack before it reached c2; now Black can't play 7...Rg1, because 8 Rxc2 protects the g-pawn. Black doesn't want to play ... c2 until the King is there to support that move, allowing the Rook to travel elsewhere.

7 Kg4 Kd6

Of course 7 ... Rg1, 8 g3 is a waste of time.

8 Rc8 Kd5

And not 8 ... Kd7, which defeats the purpose of supporting the c-pawn.

9 Kg5!

Actually 9 Kf5 might work, also. The King wants to be in front of the g-pawn and not the f-pawn, when the critical position is reached.

9 ... Kd4
10 g4 Kd3
11 Kg6!

The key move. White must be in front of the g-pawn, not the f-pawn.

11 ... Rf1
12 f5 c2
13 g5! c1=Q
14 Rxc1 Rxc1

White has reached a tablebase win, although it's not easy. Note that the win wouldn't exist if the King were on f6 instead.

15 f6 Rc6!

Black's best chance. Getting behind either pawn allows f7 with an easy win. Another try is 15 ... Ke4. The given move allows White more chances to slip up.

16 Kg7 Ke4
17 f7!

And not 17 g6? Kf5, winning one of the pawns.

17 ... Rc8
18 Kf6!

Of course, 18 f8=Q Rxf8, 19 Kxf8 Kf5 picks up the last pawn. With the given move, White keeps Black's King off of the pawns.

18 ... Rc6+
19 Ke7 Rc7+
20 Ke6!

White must play well here. 20 Ke8? Ke5!, 21 g6 (21 f8=Q Rd8+ trades pieces, after which Black gets the pawn with Kf5) Ke6, 22 f8=N (or 22 Kd8 Rxf7, or 22 Kf8 Kf6, 23 Kg8 Rc8+ trades pieces, followed by Kxg6) Kf6 gets the pawn. With the listed move, White makes sure Black's King can't get near the pawns.

20 ... Rc8

If 20 ... Rc6+, 21 Kd7, and Black can't stop the f-pawn.

21 g6

White's King has kept Black's King away, and now one of the pawns must promote.

21 ... Rc6+
22 Ke7 Ke5
23 g7 Rc7+
24 Ke8 Ke6

The last shot, hoping White will get careless and play 25 f8(g8)=Q?? Rc8#

25 f8=N+! Kd6
26 g8=Q

There are other defenses, but they seem to lead to the same general thing. Perhaps somebody else can find a way to hold the position.

[Arpad Rusz]

I think 6...Rc2! is better than 6...Ke6? (which I agree that it loses)
You made a small mistake: 10.g4? is wrong 10...Ra1 (and not 10...Kd3?) 11.f5 Ra5 12.Kg6 Rc5 = (but 10.f5! wins)

[Robert Flesher]

1... Rxc3+ 2. f3 Rc1! 3. f4+ Kg6 4. Re7 Kf6 5. Re5 c3! (5... Rc2? 6. Kf3 Rc3+ 7. Kg4 Rc2 8. g3 Rc1 9. Rc5 Ke6 10. f5+ Kd6 11. Rc8 Kd7 12. Ra8 c3 13. Ra7+! Kd6 14. Ra2 Rd1 15. Kg5 Rd4 16. Rc2 Rc4 17. Kf6 Kd5 18. g4 Ke4 19. Rc1 +-) 6. Rc5 Rc2! 7. Kf3 Ke6 8. g3 Rc1 9. Kg4 c2 10. f5+ Kf7 (10... Kf6? 11. Rc7 zz +-) 11. Rc7+ Kf6 zz =

That is the move I missed when doing my own analysis.! Time to check some other variations.

[Robert Flesher]

I do not believe we ever discovered if white white or draws. I still believe white wins here! Engines may lead to false draws or wins due to weak play, prior to getting a tablebase position.

http://www.talkchess.com/forum/viewtopi ... highlight=


[d]8/5R2/8/6k1/2p5/2P3K1/2r2PP1/8 b - - 0 1

I cannot find a win, it looks like my feeble human analysis is no match for the monster engines with tablebases. I stand corrected, yet again!

I find it amusing that one can study endgames or chess for years and then be made to look so inept by an engine, yet not due to the engine's superior knowledge, but their amazing calculation abilities, and tablebases. I stand also defeated, yet I enjoyed the lesson.

[Tom Barrister]

Picking it up after 6 ... Rc2, 7 Kf3 Ke6:

[d ]8/8/4k3/2R5/5P2/2p2K2/2r3P1/8 w - - 4 8

8 g4 might be better. Of note is that 8 ... Rc1, 9 g5 c2?, 10 Kg2! wins. Black's best looks like 8 ... Rc1, 9 g5 Kd6. I don't have enough time at the moment to look at the variations.

[PauloSoare]

Picking it up after 6 ... Rc2, 7 Kf3 Ke6:

[d ]8/8/4k3/2R5/5P2/2p2K2/2r3P1/8 w - - 4 8

8 g4 might be better. Of note is that 8 ... Rc1, 9 g5 c2?, 10 Kg2! wins. Black's best looks like 8 ... Rc1, 9 g5 Kd6. I don't have enough time at the moment to look at the variations.

After 8 g4 Rc1 draw.

[PauloSoare]

I can´t find a win too.
Arpad Rusz made good analyses.

[Robert Flesher]

Picking it up after 6 ... Rc2, 7 Kf3 Ke6:

[d ]8/8/4k3/2R5/5P2/2p2K2/2r3P1/8 w - - 4 8

8 g4 might be better. Of note is that 8 ... Rc1, 9 g5 c2?, 10 Kg2! wins. Black's best looks like 8 ... Rc1, 9 g5 Kd6. I don't have enough time at the moment to look at the variations.

This variation appears to draw. The rook cannot escape the kings harassment.