Manually disabled problems (was 46848)

[drahacikfm]

Changed topic subject so we can find it.  Richard manually disabled a couple problems in that topic.  Now we can find the topic more easily.

52004 needs to be disabled.  The endgame is not winning, even in the computer variation given after white wins 2 pawns in that opposite-bishop ending.

[dougy]

I agree that this problem should be disabled, but because White scores 100% by playing a forced and obvious move, while to win the game he would have to play a long series of accurate endgame moves.

There are a couple of interesting plans for White after winning the g- and h-pawns:
1. Sacrifice the bishop for the pawn on f6 creating triple connected passed pawns.
2. Use the h-pawn as a decoy to win the f6-pawn creating connected passed f- and g-pawns.

Here's an example line for plan 1:
1. Bxc6+ Kxc6 2. Bc5 Bb3 3. Ke3 Kd7 4. Bf8 Ke8 5. Bxg7 Kf7 6. Bxh6 Bd1 7. Kf4 Bb3 8. Kg3 Ba4 9. Be3 Bc2 10. Bd4 Kg7 11. Kf4 Kf7 12. Bxf6 Kxf6 13. h6 Kf7 14. g5 Kg8 15. g6 Kf8 16. h7 Ke8 17. h8=Q+

[dougy]

Here's a good one.  Problem 12980.

After 3.Rxb1 is mate in 18 in the endgame tablebases.  But in the computer's analysis it plays 5.Re1? which only draws.

Rook pawns are surprisingly weak!  White can't allow Black's king in front of the pawn.

[revenant]

Problem 12980 shows 3. Rxb1 Kc6 4. Rb7 Bb6 5. Rg7 etc. as the final PV in the variations sidebar.  Is that in accord with the EGTB, dougy?  As Richard pointed out in a recent thread,

You can see this in play if you look carefully at the move list, often the later moves have different move sequences (especially further down the sequence) than earlier moves.

Thanks for the mini-lesson on rook & pawn endings!  I'll have to go dig out "Practical Rook Endings" by Edmar Mednis and study up.

[drahacikfm]

Hey revenant, that pointy-headed thing there is a bishop, not a rook!  So Mednis' book isn't going to help you    Good book though, I have it and it's worn out (but I still don't remember half the stuff in it).

Just so it's clear to Richard: this problem does not need to be disabled.  The 3 moves played by White are the best moves, and winning.  Only the analysis variation displayed after the first move has a mistake on move 5.  After the second and third moves the analysis variation corrected itself.

Yes, this endgame is one of those rare cases where White could not win with rook and pawn against bishop, if the Black King got to the corner a8.  In this endgame if Black gets to the corner, White wins if his pawn is farther back, a2 or a3 or a4, not sure exactly about a5, but I think that's a win too.  a6 is a draw.  "Passed pawns must be pushed" is bad advice for this endgame!

[revenant]

The good news is, I also have "Practical Bishop Endings" by Mednis which is equally commendable (students, take note!).  The bad news is, the genius instructor never got around to writing "Practical Rook-vs.-Bishop Endings" before he croaked in 2002 (although there may be some relevant advice in his "Practical Endgame Tips" or "Practical Endgame Lessons", neither of which I've seen).

Of course, what I *really* need is "Practical Queen+Rooks+Bishops+Knights+Pawns vs. Queen+Rooks+Bishops+Knights+Pawns Endings for the Impatient Player Who Doesn't Want to Have to Think".

Oh, wait... Is that what's called the "opening"?

[richard]

52004 disabled..

[dougy]

Problem 12980 shows 3. Rxb1 Kc6 4. Rb7 Bb6 5. Rg7 etc. as the final PV in the variations sidebar.  Is that in accord with the EGTB, dougy?  As Richard pointed out in a recent thread,

You can see this in play if you look carefully at the move list, often the later moves have different move sequences (especially further down the sequence) than earlier moves.

Ah, okay.  Yep that's fine - I didn't see that line (:.  Nonetheless, it's a rather interesting position that can save the half-point at times.

[oded ross]

Changed topic subject so we can find it.  Richard manually disabled a couple problems in that topic.  Now we can find the topic more easily.

52004 needs to be disabled.  The endgame is not winning, even in the computer variation given after white wins 2 pawns in that opposite-bishop ending.

I agree that this problem should be disabled, but because White scores 100% by playing a forced and obvious move, while to win the game he would have to play a long series of accurate endgame moves.

There are a couple of interesting plans for White after winning the g- and h-pawns:
1. Sacrifice the bishop for the pawn on f6 creating triple connected passed pawns.
2. Use the h-pawn as a decoy to win the f6-pawn creating connected passed f- and g-pawns.

Here's an example line for plan 1:
1. Bxc6+ Kxc6 2. Bc5 Bb3 3. Ke3 Kd7 4. Bf8 Ke8 5. Bxg7 Kf7 6. Bxh6 Bd1 7. Kf4 Bb3 8. Kg3 Ba4 9. Be3 Bc2 10. Bd4 Kg7 11. Kf4 Kf7 12. Bxf6 Kxf6 13. h6 Kf7 14. g5 Kg8 15. g6 Kf8 16. h7 Ke8 17. h8=Q+

I think it's a mistake to disable this problem.
First of all, the endgame seems winning to me, after White put hit B on d4 and the K on f4, unlike the (once again) lame computer line.
The problem need to go on at least up to 3.Ke3 , then it will become an okay problem - not a 2000 standard one but not 100% solved either.
Ideally the solution should stretch to 10-11 moves but I guess that would require too many alternatives.

[drahacikfm]

Ok, I agree now that this problem (52004) is winning.  But the question remains whether it should be enabled again.

The first move is absolutely forced, because if you don't take the knight, you lose a piece.  Then the problem stops.  And it's a bishops-of-opposite-color endgame with equal material at that point.

The win is found by deeply analyzing a position that occurs many moves after the problem already stopped.  Rybka needed to get to depth 27 before it saw that it was winning.  The question is: is such a problem appropriate for a tactics site, or should it be in some advanced opposite-bishop endgame textbook!  It certainly should not be enabled as a one-move problem with a forced first move and a rating under 1400.

[uri blass]

Ok, I agree now that this problem (52004) is winning.  But the question remains whether it should be enabled again.

The first move is absolutely forced, because if you don't take the knight, you lose a piece.  Then the problem stops.  And it's a bishops-of-opposite-color endgame with equal material at that point.

The win is found by deeply analyzing a position that occurs many moves after the problem already stopped.  Rybka needed to get to depth 27 before it saw that it was winning.  The question is: is such a problem appropriate for a tactics site, or should it be in some advanced opposite-bishop endgame textbook!  It certainly should not be enabled as a one-move problem with a forced first move and a rating under 1400.

I think that every problem that is about winning material for no compensation is
appropriate for a tactics site.

The only point is that it may be not appropriate for the chess tempo site because the chess tempo site is only about winning tactics and not about tactics in general.

I consider winning a pawn  even if it does not win the game as tactics and I am going to consider using chess tempo for more time as soon as these problems appear in chess tempo(it can be also winning a pawn to get equal position when alternatives simply cause the opponent to be pawn up with no compensation).

Uri

[drahacikfm]

53180 should be disabled.  1...Rxc3 is completely winning but is not an allowed alternate.  Rybka +3.23 advantage.  Strange that Toga missed this.  1...Rxc3 is the only "mistake" that people have made so far.

[richard]

re:53180  I think Rybka probably has much better inbuilt endgame evaluation, Toga wouldn't consider that line even after a fair bit of thinking. Forcing toga down that line a few steps it does start to see around +2.5.  I assume that endgame is a theoretical win? If so I'll disable this one also.

Richard.

[drahacikfm]

53180:  Not sure which line you were forcing Toga down.  But 2.Rxc3 bxc3 leads eventually to Q+2P vs Q which is winning.  Any other move by White, and Black plays 2...Ra3 and wins the a4 pawn, with two connected passed pawns extra in the rook ending.  Even after 2.Ke2 Black plays 2...Ra3 because 3.Rxa3 loses immediately.

What's interesting is that when I set Rybka to show the top 4 moves, it does not display 1...Rxc3 until depth 21.  On depth 20 it's still not in the top 4.  Your Toga analysis on the problem shows depth 19, so even if you had been using Rybka you wouldn't have had 1...Rxc3 as an alternate.  I think depth 21 is needed for the engine to see that Black goes two pawns up in the queen ending.

[richard]

Forcing toga down the 1..Rxc3 line lets toga see the 2 pawn advantage (and the promotions for both sides), I'll disable it.

Richard.