|
< Earlier Kibitzing · PAGE 4 OF 4 ·
Later Kibitzing> |
Jan-20-10
 | | HeMateMe: I think White has a winning position, here. I'll go out on a limb, and say that Smeets was wise to resign. |
|
| Jan-20-10 | | MaxxLange: <zanshin> I hear you.. there is a concrete line of defense that refutes the idea, so: not sound. I remember in college, the logic teacher told us that "valid" is not a term of praise, it just means that there is no bust. A "sound" sacrifice is like that valid philosopher's logic: it is sound if it CAN'T, with best play, go from a win to a draw, or a draw to a loss |
|
| Jan-20-10 | | zanshin: And in case anyone thinks I am trying to minimize Shirov's play, I will reiterate that he was the one who described his sac as unsound. |
|
| Jan-20-10 | | Hesam7: <zanshin: I think we disagree on the definition of a sound sacrifice. I disagree that a complication that results in a win defines a sound sacrifice.> Fair enough.
<zanshin: I think they are consistent. Rybka lowered the evals because <31.Qg5> was not the preferred move.> They are not, the second eval contradicts the first one. In other words the machine does not realize 31. Qg5 is better until you feed it to it. That is an indication that it's evaluations are not correct. |
|
| Jan-20-10 | | patzer2: After 33...Rcd8?? (instead of 33...Rce8=), the obstruction sacrifice 34. Rxd4!! Qxd4 35. Ne5  is even stronger than the winning game continuation. AFter 34. Rxd4!! Qxd4 35. Ne5, Black might try 35...Rd7 but then 36. Nxd7 or 36. Qh6 wins easily. |
|
Jan-21-10
| | HeMateMe: <I remember in college, the logic teacher told us that "valid" is not a term of praise,> If you scream "fire" in a crowded forest, will people in a movie theater see you fall? |
|
| Jan-21-10 | | RandomVisitor: A final look at the proposed improvement 28...Kh8:
click for larger view Rybka 3:
<[-0.33] d=24 29.Qf4> Rg8 30.Rce1 Qd5 31.g4 Bd8 32.Nd6 Rc7 33.Re8 Qg5 34.Qf3 Bg6 35.Rxg8+ Kxg8 36.Qb3+ Qd5 37.Qxd5+ cxd5 38.h4 Rd7 39.Kg2 b6 40.h5 Bf7 41.f4 |
|
| Jan-21-10 | | GlennOliver: "atragon: Another line at 32 may be
Bxf2+ and after 33. Kxf2 there is Rf8+, 34. Ke1 Rce8+ which leads to an interesting endgame where white edge is very small."Agreed, a sound counter-attacking gambit with possibilities for Black to take at least a draw. |
|
| Jan-21-10 | | zanshin: <They are not, the second eval contradicts the first one. In other words the machine does not realize 31. Qg5 is better until you feed it to it. That is an indication that it's evaluations are not correct.> <Hesam7> Here's a quick head-to-head: 
click for larger view [-0.53] d=17 31.Rxg7 Qxg7 32.Qxg7 Kxg7 33.Nd6 Rcd8 34.Nxe8 Rxe8 35.Rc3 Re7 36.Rb3 Kf6 37.Rf3 Kg6 38.Rf8 Rd7 39.Rf4 Kg5 40.g3 (0:03.55) 27134kN [-0.54] d=16 31.Qg5 Bxd4 32.Rd1 Re2 33.Kh2 Rf8 34.Ne5 Rxe5 35.Rxd4 Rxg5 36.Rxd7 Rxg3 37.Kxg3 Rg8 38.Kf3 Rg7 39.Rxg7 Kxg7 40.g4 b5 41.cxb6 axb6 42.Kf4 c5 43.Ke5 c4 (0:03.27) 24018kN Sure, the evals may change later and the moves may flip-flop at deeper iterations. But are we really going to start quibbling about a centipawn or two? You know that it's very common for evals and preferred moves to change as you slide down the line. This does not mean all previous analysis is suspect. |
|
| Jan-21-10 | | whatthefat: <I think we disagree on the definition of a sound sacrifice. I disagree that a complication that results in a win defines a sound sacrifice.> To me a <sound> sacrifice is one that does not objectively change the evaluation of the position. This one seems to be in the gray area. Before the sac, the position was objectively drawn, with maybe a slightly better position for White. After the sac it was still objectively drawn but now Black needed to make a difficult defense in the short term that should have yielded a slightly better (but still drawn) endgame for Black. |
|
| Jan-22-10 | | Hesam7: <zanshin: Sure, the evals may change later and the moves may flip-flop at deeper iterations. But are we really going to start quibbling about a centipawn or two? > You did not get my point, I am comparing the following (both posted by yourself): <After <30. ... Bg7>[-0.54] d=16 31.Qg5 Bxd4 32.Rd1 Re2 33.Kh2 Rf8 34.Ne5 Rxe5 35.Rxd4 Rxg5 36.Rxd7 Rxg3 37.Kxg3 Rg8 38.Kf3 Rg7 39.Rxg7 Kxg7 40.g4 b5 41.cxb6 axb6 42.Kf4 c5 43.Ke5 c4 (0:03.27) 24018kN After <31. Qg5>
[-0.24] d=19 31...Re6 32.Ne5 Bxe5 33.dxe5 Rg8 34.Qe3 Rxg3 35.fxg3 Qd3 36.Qf4 Qd8 37.Kh2 Bg6 38.Rc4 Kg7 39.Rb4 b5 40.cxb6 (0:09.46) 69821kN> It is clear that Rybka should not be trusted in these positions. |
|
| Jan-22-10 | | zanshin: <Hesam7> I knew that's what you were referring to. I simply made another run to compare the moves directly. In my previous post that you cited, Rybka may have changed the evals, but as I said, this is quite common. It doesn't mean everything is now suspect. |
|
| Jan-22-10 | | Hesam7: <zanshin> my point is simple: "the machine realizes the strength of a move only after you play it." If despite knowing this you still trust its evaluations then I have nothing more to say. |
|
| Jan-22-10 | | zanshin: <Hesam7> I sense we are close to agreement on one thing: it's time to wrap this up. I got two different evals for <31. Qg5> ([-0.24] d=19 31...Re6; and [-0.54] d=16 31.Qg5). I could get different evals depending on the state of my hash tables, length of analysis, lines I had examined, etc. It's not going to be exact all the time - hence my point about quibbling over centipawns. The evals are still negative - hardly indications for a strong move And definitely not <practically almost lost> as you stated earlier. If you want to pursue this, let's move to another forum, yours or mine. I suspect the others are growing tired of us. |
|
| Jan-22-10 | | Hesam7: <zanshin: I got two different evals for <31. Qg5> ([-0.24] d=19 31...Re6; and [-0.54] d=16 31.Qg5). I could get different evals depending on the state of my hash tables, length of analysis, lines I had examined, etc. It's not going to be exact all the time - hence my point about quibbling over centipawns.> The difference of evals before and after White plays 31. Qg5 is not explained by the factors you have described. Rybka simply does not understand the potentials of White's position. <zanshin: The evals are still negative - hardly indications for a strong move And definitely not <practically almost lost> as you stated earlier.> Again I am disputing the fact that evals being slightly negative means anything here. Also by <practically almost lost> I meant that if you have to defend this position OTB against Shirov in time trouble chances of you surviving are very slim. |
|
| Jan-22-10 | | zanshin: <Rybka simply does not understand the potentials of White's position.> <Hesam7> I guess we'll keep going ;-) Would you be convinced if I got the same negative evaluations with a different engine? My guess is, no. <I meant that if you have to defend this position OTB against Shirov in time trouble chances of you surviving are very slim.> Well, that's the root of our misunderstanding. I thought we were analyzing this as objectively as possible. Again, the objective soundness or strength of a move has nothing to do with the complications and errors induced by time pressure, personal dislike for the position, etc. which includes all other off-the-board issues. |
|
| Jan-22-10 | | euripides: <the objective soundness or strength of a move has nothing to do with the complications and errors induced by time pressure, personal dislike for the position, etc. which includes all other off-the-board issues.> Butting in here: I think this raises an interesting semi-philosophical point about what these evaluations mean. There's one sense of 'objective' in which there are only three accurate evaluations for any move: winning, drawing or losing assuming both sides play perfectly. Personally,I would call a sacifice that produced a change in this evaluation 'unsound' and one that didn't 'sound'. For two moves that share an evaluation in that sense the question is which move is more likely to win, or harder to play against, given imperfect players. I don't see how 'off-the-board' considerations can be excluded from the comparison of two moves in this case - given that human imperfection is itself 'off-the-board' and with perfect players the two moves would be equivalent. The analogous philosophical point is that the value of an approximation depends on the context in which it is used. |
|
| Jan-22-10 | | euripides: ...I think people sometimes have a third concept in mind, but it's hard to make this precise. Consider an 'objectively' drawn position (drawn with perfect play). Say a sacrifice produces very difficult tactics for ten ply, with lots of chances for the opponent to lose, but with perfect play an endgame is reached that is still objectively drawn, but is difficult in practice for the player who made the sacrifice. The computer will then give a negative evaluation and some people will call the sacrifice 'unsound'; but really it's a question of how you weigh the initial complications, which make life difficult for one side, against the tricky ending, which is difficult for the other side. I don't see a clear rationale for calling such a sacrifice 'objectively unsound' without referring to the two kinds of difficulty - which will vary across human players and even more so between humans and computers. |
|
| Jan-22-10 | | Hesam7: <zanshin: <Rybka simply does not understand the potentials of White's position.> <Hesam7> I guess we'll keep going ;-) Would you be convinced if I got the same negative evaluations with a different engine? My guess is, no.> So far what you have been saying is something along these lines: "oh look the eval changed from +0.1 to -0.3." But this is not an argument, nobody has demonstrated a way for Black to get an advantage, most lines lead to unclear endgames. The fact that some engine's evaluation went down about half a pawn is completely irrelevant. <zanshin: Well, that's the root of our misunderstanding. I thought we were analyzing this as objectively as possible. Again, the objective soundness or strength of a move has nothing to do with the complications and errors induced by time pressure, personal dislike for the position, etc. which includes all other off-the-board issues.> You obviously did not read my post carefully, so let me refresh your memory: <Hesam7: My claim is that after 31. Qg5 Black does not have any advantage and at best the position is objectively equal but practically almost lost.> and I still stand by what I said. |
|
| Jan-22-10 | | zanshin: <euripides> You and <whatthefat> have raised some good and interesting points, but I'm afraid we are getting off-topic here. I have some thoughts on separation of off-the-board factors and objective evaluation. We can pursue them further in my forum if you want. |
|
| Jan-22-10 | | zanshin: OK <Hesam7> I think we have exhausted all arguments here. I have a better understanding of your point of view and hope you have one of mine. I don't think we are making any progress. By your definitions, the sac is sound and that's fine with me. |
|
| Feb-22-10 | | Doug4Chess: When Shirov played 36. Ng6+, Smeets should have answered with Bxg6. After that, I see nothing that Shirov could have done to mount a mating threat. In fact, the advantage would have been with black. |
|
| Feb-22-10 | | Shams: <Doug4Chess> 36...Bxg6? 37.Qh6+ and now: i. 37...Qh7 38.Qxf8+ , or
ii. 37...Kg8 38.Rxg6+ Kf7 39.Rg7+ wins the queen |
|
| Feb-23-10 | | Doug4Chess: Shams: Yes, you're pretty much correct. Several hours after making the 02/22/10 posting, I took a second look at the board after 36. Ng6+, and realized how mistaken my analysis had been. |
|
| Feb-23-10 | | Shams: <Doug4Chess> Murphy's law of posting, you notice your blunder right after the <delete> button disappears from next to your offending post. Happens to me all the time. :0 |
|
 |
|
|
< Earlier Kibitzing · PAGE 4 OF 4 ·
Later Kibitzing> |
Kibitzing
sac: 26.Bxh6