Another Tarrasch beauty

[Rebel]

[d]3r4/7p/Rp4k1/5p2/4p3/2P5/PP3P1P/5K2 b - -

This is Tarrasch-Rubinstein, San Sebastian 1911. Black is a pawn down and the b6 pawn is under attack. Passive defence with 1...Rd6 or 1...Rb8 is doomed to failure. Rubinstein grasps his only chance: To mobilize his rook......... Rd2! is a draw.

Which of the top-engines?

Not an unusual position in chess.

[Jouni]

This is familiar from some testsets and unsolved so far. My guess: INCORRECT.

[Rebel]

I think a GM will play 1..Rd2 instantly.

[Terry McCracken]

I think a GM will play 1..Rd2 instantly.

Rb8 should work too...no?

[Jouni]

10s/move test games after -Rd2:

[Event "Asema"]
[Site "?"]
[Date "2012.03.26"]
[Round "3"]
[White "Critter 1.4 64-bit"]
[Black "Houdini 1.5a x64"]
[Result "1-0"]
[Annotator "0.98;0.90"]
[SetUp "1"]
[FEN "3r4/7p/Rp4k1/5p2/4p3/2P5/PP3P1P/5K2 b - - 0 1"]
[PlyCount "46"]
[TimeControl "60/600:0/0:0/0"]

{Pentium(R) Dual-Core CPU E5300 @ 2.60GHz 2599 MHz W=24.6 plies; 3
272kN/s; 869 268 TBAs B=23.5 plies; 3 131kN/s; 730 551 TBAs} 1... Rd2 {Both
last book move} 2. Rxb6+ {0.98/19 15} Kg5 {0.90/20 22} 3. a4 {1.29/22 17 (Ke1)}
Kg4 {0.74/21 12} 4. Ke1 {1.29/21 0} Rc2 {0.74/19 0} 5. a5 {1.29/22 18} Kf3 {0.
74/24 13} 6. a6 {1.29/21 0} Re2+ {0.74/23 0} 7. Kd1 {1.32/23 16} Kxf2 {0.75/22
3} 8. Rf6 {1.32/21 1} Rxb2 {0.80/25 10} 9. Rxf5+ {1.32/25 17} Ke3 {0.80/24 0}
10. Ra5 {1.32/24 0} Kd3 {0.84/25 11} 11. Kc1 {1.33/26 16} Rb8 {0.84/23 0} 12.
a7 {1.33/26 16} Ra8 {0.84/22 0} 13. Rd5+ {1.33/27 15 (Ra3)} Kxc3 {1.35/26 78}
14. Rd7 {1.33/26 0} h6 {0.79/24 0} 15. Kd1 {1.33/31 15} Kb4 {1.50/25 8 (Kc4)}
16. Ke2 {1.33/30 14} Kc5 {1.50/26 9 (Kb5)} 17. Ke3 {1.33/28 14} Kc6 {1.50/25 1}
18. Rf7 {1.50/26 13 (Rh7)} Kd5 {1.50/27 8} 19. Rf5+ {1.76/25 12 (h4)} Kc6 {2.
86/27 61} 20. Rf6+ {1.76/23 0} Kb5 {2.94/22 11 (Kb7)} 21. Rf7 {1.98/25 13} Kb6
{2.95/24 7} 22. Kxe4 {2.49/23 12 (h4)} Rg8 {4.49/22 54 (Rd8)} 23. h4 {2.50/21 0
} h5 {5.92/20 47 (Ka6)} 24. Kf5 {3.50/23 12 adjud. (Rh7)} 1-0

[Event "Asema"]
[Site "?"]
[Date "2012.03.26"]
[Round "4"]
[White "Houdini 1.5a x64"]
[Black "Critter 1.4 64-bit"]
[Result "1-0"]
[Annotator "0.93;0.89"]
[SetUp "1"]
[FEN "3r4/7p/Rp4k1/5p2/4p3/2P5/PP3P1P/5K2 b - - 0 1"]
[PlyCount "140"]
[TimeControl "60/600:0/0:0/0"]

{Pentium(R) Dual-Core CPU E5300 @ 2.60GHz 2599 MHz W=21.7 plies; 3
427kN/s; 302 547 TBAs B=24.4 plies; 3 907kN/s; 3 387 177 TBAs} 1... Rd2 {
Both last book move} 2. Rxb6+ {0.93/18 13} Kg5 {0.89/19 16} 3. a4 {0.89/20 9}
Rc2 {1.27/20 18 (Rd1+)} 4. a5 {1.00/22 11 (Rb4)} Kg4 {1.31/22 18 (f4)} 5. a6 {
0.87/20 8} Kf3 {1.32/22 0} 6. Ke1 {0.87/19 0} Re2+ {1.32/22 17} 7. Kd1 {0.87/
24 13} Kxf2 {1.33/22 1} 8. Rf6 {0.87/22 0} Rxb2 {1.33/24 14} 9. Rxf5+ {0.88/25
9} Ke3 {1.33/23 0} 10. Ra5 {0.88/24 0} Kd3 {1.33/24 16} 11. Kc1 {0.88/25 12}
Rb8 {1.36/23 0} 12. a7 {0.88/27 10} Ra8 {1.36/26 6} 13. Rd5+ {0.88/29 12} Kxc3
{1.36/27 32} 14. Rd7 {0.88/28 0} h6 {1.37/28 14} 15. h4 {1.47/30 72 (Kd1)} Kb4
{1.37/29 13 (Kc4)} 16. Kd2 {2.44/22 3} Kc5 {1.67/29 68 (Kb5)} 17. Ke3 {2.58/21
3} Rg8 {1.97/25 22 (Kc6)} 18. Kxe4 {4.07/19 9 (Rg7)} Rg4+ {2.17/24 11 (Re8+)}
19. Kf5 {8.92/18 8 (Ke5)} Ra4 {2.31/25 9 (Rg8)} 20. Rh7 {16.21/15 7 (h5)} Ra5 {
2.65/23 21} 21. a8=Q {#46/14 0 (h5)} Rxa8 {1.80/22 6} 22. Rxh6 {#45/13 0} Ra1 {
#44/0 0 (Rf8+)} 23. h5 {#44/0 0} Rf1+ {#43/0 0} 24. Ke6 {#43/0 0} Rh1 {#42/0 0}
25. Kf7 {#42/0 0} Kd4 {#41/0 0} 26. Kg7 {#41/0 0} Kc5 {#40/0 0} 27. Rh8 {#40/0
0} Rg1+ {#39/0 0} 28. Kf6 {#39/0 0} Rf1+ {#38/0 0} 29. Ke6 {#38/0 0} Re1+ {#37/
0 0} 30. Kf5 {#37/0 0} Rf1+ {#36/0 0} 31. Kg4 {#36/0 0} Rg1+ {#35/0 0} 32. Kf3
{#35/0 0} Rh1 {#34/0 0} 33. h6 {#34/0 0} Kb6 {#33/0 0} 34. Kf4 {#33/0 0} Kc6 {
#32/0 0} 35. Kf5 {#32/0 0} Kc7 {#31/0 0} 36. Kg6 {#31/0 0} Rg1+ {#30/0 0} 37.
Kh7 {#30/0 0} Kd7 {#29/0 0} 1-0

[MM]

I think a GM will play 1..Rd2 instantly.

+1

(It looks draw to me, active rook, white king has limited movements, black king can invade white field, central black pawns can be pushed to create a strong counterplay).

Regards

[Don]

[d]3r4/7p/Rp4k1/5p2/4p3/2P5/PP3P1P/5K2 b - -

This is Tarrasch-Rubinstein, San Sebastian 1911. Black is a pawn down and the b6 pawn is under attack. Passive defence with 1...Rd6 or 1...Rb8 is doomed to failure. Rubinstein grasps his only chance: To mobilize his rook......... Rd2! is a draw.

Which of the top-engines?

Not an unusual position in chess.

Is it well established that Rd2 draws or is it an opinion? It could be that Rb8 also draws, or that neither draws. I agree that Rd2 is the best try but nothing else is obvious to me, although I'm not that strong a player.

[BubbaTough]

I suspect most masters would reject ...Rb8 because it seems likely white can get 2 connected passed pawns and advance them quickly for a win. Thus, they would play something active (Rd2) even in bullet chess. It would take a lot of work for a human to confirm or deny whether this is really the best move. The computer is going to be much better than humans at working out the details.

Rook endgames are probably the area human intuition most often is superior to computer calculation (other than perhaps locked or fortress positions) because human heuristics about what is likely to win and what is likely to draw are pretty good, and most top programs lack good approximations of these heuristics. Nevertheless, the ability to work out the details is so important and so useful, I suspect even in rook endgames computers are better than humans at most non-postal time controls. Sure, you can cherry pick certain endgames where humans are better, but generally speaking, its darn likely computers are still better.

On a week a move though, I would pick Kramnik in any rook endgame over a cluster of computers. Somehow, I doubt that hypothesis will be tested.

-Sam

[Don]

I suspect most masters would reject ...Rb8 because it seems likely white can get 2 connected passed pawns and advance them quickly for a win. Thus, they would play something active (Rd2) even in bullet chess. It would take a lot of work for a human to confirm or deny whether this is really the best move. The computer is going to be much better than humans at working out the details.

Rook endgames are probably the area human intuition most often is superior to computer calculation (other than perhaps locked or fortress positions) because human heuristics about what is likely to win and what is likely to draw are pretty good, and most top programs lack good approximations of these heuristics. Nevertheless, the ability to work out the details is so important and so useful, I suspect even in rook endgames computers are better than humans at most non-postal time controls. Sure, you can cherry pick certain endgames where humans are better, but generally speaking, its darn likely computers are still better.

On a week a move though, I would pick Kramnik in any rook endgame over a cluster of computers. Somehow, I doubt that hypothesis will be tested.

-Sam

I noticed that computers often play differently and this is often criticized when it almost always turns out that the computer is correct or else it doesn't really make a difference. In many cases the human's move choice is simply more comfortable or easier to play for a human or it's judged to be more difficult or unpleasant for the opponent. Almost all human players would rather attack than defend even if that is not objectively best but computers see things more clearly most of the time. What is scary and complicated to a human is often boring to a computer which can often see past all the complexity to see things for what they really are.

In this position I don't care what humans prefer because if black sacrifices the pawn he is now down 2 pawns and there is an interesting defense to blacks gambit based or Rb3 to keep out the black king. So now white has 3 connected passed pawns and black does not seem to have anything. Or maybe he does, that is what I'm trying to figure out now.

But what I am saying is that this smells like one of those positions where black plays coffeehouse chess as a bluff. It probably is the best way to play against another human but I have my doubts that it's any sounder in an objective sense than Rb8.

[Alibaba]

Hi,

seems interesting to take a notebook with a modern Chess Engine (but without TB, and 1 core should be enough) use a timemachine and visit the chess legends like Steinitz,Tarrasch, Lasker, Capablance etc at there best time and play a match of lets say 30 tournament time control games.

I would pay a lot of money to see such a match!

I think expect Lasker to be able to win 1 or 2 games, but not the match, and would expect Capablanca to play a lot of draws.....

Regards
Andi