|
< Earlier Kibitzing · PAGE 237 OF 237 ·
Later Kibitzing> |
| Sep-19-07 | | MostlyAverageJoe: <Honza Cervenka: <15.exf6 e.p.> ... I am still not sure in this matter and I doubt that even Rybka the Magnificent can pierce the fog of this mess properly in just 60 sec per move analysis.> For what it's worth, at 18 plies, Hiarcs11.2MP still likes 15. Nc5, however, at 19 plies, exf6EP edges it out. I marked with <> the less-reliable portions of the analysis, as I did not do any sliding back or forth (the stuff below run for 5 CPU hours already). 1. (+0.60) 15. exf6EP Qe8 16. f7 Bxf7 17. O-O Bd5 18. Nf2 Bg7 19. e4 Be6 <20. Rfc1 Rxf3 21. Bxf3 Nxd4 22. e5 Nxf3 23. Qxf3 Rd8 24. Rc5 Rd2 25. b3 Qd8 26. Rd1> 2. (+0.53) 15. Nc5 Bd5 16. O-O <next move diverges from the game> e6 17. Rae1 Nb8 18. a4 Nd7 19. axb5 Nxc5 <20. dxc5 axb5 21. Bxb5 Qe7 22. b4 Rfb8 23. Bd3 Ra4 24. b5 Ra2 25. Be2> After 20 plies, both valuations drop a bit, but the diffrence increases: 1. (+0.59) 15. exf6EP Qe8 16. f7 Bxf7 17. O-O Bd5 18. Nfg5 Rf5 19. h4 Rd8 <20. Bf3 Bc4 21. Nd2 Nxd4 22. exd4 Bxf1 23. Nxf1 Bxg5 24. hxg5 Qd7 25. d5 Rxg5 26. Rc1 Qd6> 2. (+0.47) 15. Nc5 Bd5 16. O-O e6 <next move diverges from Rybka's 17-ply variation> 17. Rae1 Nb8 18. a4 Nd7 19. axb5 Nxc5 <20. dxc5 axb5 21. Bxb5 Qe7 22. Bd3 Ra4 23. b4 Rb8 24. b5> At lower depth, the best choice fluctuated a bit. It is still not quite clear which one is preferable, but the increasing difference seems to point to exf6EP as a better variation. |
|
| Sep-19-07 | | METOPEA: When will be the next? |
|
Sep-19-07
 | | Honza Cervenka: <Gypsy> I don't think that I have any special method though longterm training from CC can be quite useful in this. It is terribly time-consuming doing (especially in positions like was that after 15.exf6) and it consists of sitting with a chessboard before for hours (and sometimes for days) and trying and rechecking over and over again different possibilities in the position. Of course, without engine's assistance this is much more demanding as comp usually recognises quickly if the thing that just crossed your mind is complete trash or not sparing thus a lot of dear time and it can point out covert tactical holes as well as opportunities. But the difference is not that great in the end I would say. Btw, in this game I usually used ChessPad board on the screen instead of my good old wooden chessboard for analysing which spared time necessary for recording of analysed moves and lines and setting of the positions (and also it reduced probability of setting incorrect position which was my nightmare in the past when I played CC chess actively). But in all other ways I felt that I am back in early 90s...:-) |
|
Sep-20-07
| | Honza Cervenka: <OhioChessFan: Any good analysts think 33....Qa8 could have saved Black?> What is the plan of black after 34.Nf6+ Bxf6 (this looks pretty forced as after 34...Kf7 or 34...Kf8 white can play 35.Qf2 g5 36.Qf3 with penetration into black position or even 35.Rhxe3 fxe3 36.Qxe3 with next 37.Rf1 or 37.Qe4 and poweful attack) 35.exf6 Rf7 36.Qf2? 
click for larger view36...Rxf6 loses outright to 37.Qh4 but other moves don't seem to be satisfactory as well. For example, 36...Rd6 (stopping immedate 37.Qh4 due to Qxg2#) 37.Rh6 Rxf6 (37...Rxd4 38.Nxe6  ) 38.Rxe3 Rxd4 39.Rh8+ Kxh8 40. Rh3+ Kg7 41.Qxd4 ; 36...g5 37.Rh5 ; 36...Rh7 37.f7+ Rxf7 38.Qh4 . Probably the best but still quite hopeless is sac of exchange 36...Rxc5 37.dxc5 Qd5 which stops white's attack but produces an ending with quite significant material deficit. |
|
Sep-20-07
 | | Gypsy: Thank you <Honza>!
<All> On a different topic: What is roughly the consultation strength of the cg Teams in sans-Fritzes games? Thus far, this game is the only sample point and it clearly was not played at the stratospheric levels of detailed accuracy that the GMAN and Yuri games were played at. But it does leave a good impression. I believe these teams could take head on rank-and-file GMs creditably. And, in a combined form, they probably wouldn't embarrass the <cg> mother-ship if Yuri asked for his rematch 'sans machines'. I am a partial observer, however. |
|
| Sep-20-07 | | whatthefat: <Gypsy>
I think certainly Team White played at a very high standard in this game, and that a full cg.com group could surely put up a good fight against Shulman. Of course though, the larger the group, and the higher profile the game, the more potential for cheating. |
|
Sep-20-07
| | Gypsy: <the larger the group, and the higher profile the game, the more potential for cheating> That is a disheartening thought. Well, maybe we should keep the game profile low then. All in all, playing high quality, high integrity games here among the friends is all that my heart desires. |
|
| Sep-21-07 | | MostlyAverageJoe: For what it's worth, I run analysis with Hiarcs 11.2, 40 CPU minutes per move, threshold 0.10, starting at move 10 or so. At this threshold, there are more mistakes indicated than what <nimh> found with Rybka (and some disgreement with Rybka's assessment of the magnitude of inaccuracies, too). Still, the accuracy is amazing. All inaccuracies happened in very complicated positions, judging by the fact that Hiarcs got only 18-19 plies deep on them. Here's the list.
19. ... b4 (+0.81) <better for BLACK would be: (+0.67) 19... e6 20. Bf3 Qh4 21. Qd3 Rfd8 22. b3 Bg5> 29. .. Ne3 (+1.17) <better for BLACK would be: (+0.95) 29... Kg7 30. Nbc5 Rc6 31. Nf6 Rxc5 32. dxc5> 30. ... Bg7 (+1.55) <better for BLACK would be: (+1.17) 30... Rc6 31. Nbc5 Kg7 32. Nf6 Qd8 33. b4 Rxf6 34. exf6 Qxf6> 35. ... Nd5 (+2.08) <better for BLACK would be: (+1.92) 35... Bxe5 36. Qb5 Bxd4 37. Qxc6 Rxh7 38. Qxe6 Kg7 39. Rxh7 Kxh7> 36. Nef6+ (+1.83) <better for WHITE would be: (+2.08) 36. Qf3 Rf5 37. Qg4 f3 38. Qxg6 f2 39. Nxf2 Nf4 40. Nf6 Rxf6> 40. ... Rd7 (+2.19) <better for BLACK would be: (+1.86) 40... Rxd4 41. Qxc7 e5 42. Qc8 Rd8 43. Qc4 Qf7 44. Rc3 Qxc4 45. Rxc4 Kf7> 43. ... e5 (+2.50) <better for BLACK would be: (+2.39) 43... Qh6 44. Qc4 Qf6 45. Rf2 Qf7 46. b3 Bf6 47. Rfe2 Rd7 48. Ra2 Qh5> Final position evaluates at (+4.18) after 18 plies. |
|
| Sep-21-07 | | Eyal: Of <MostlyAverageJoe>'s 7 alternative lines, I don't think the last 5 make any real difference - once Black loses the exchange after 35...Bxe5 36.Qb5 or as in the game itself, White's win is just a matter of technique; and <30...Rc6 31.Nbc5 Kg7 32.Nf6 Qd8 33.b4 Rxf6 34.exf6 Qxf6> also seems hopeless for Black after 35.b5 Rd6 36.Qc3! Qg5 (36...Qe7 37.g3; 36...g5 37.Rhxe3 fxe3 38.Qxe3) 37.Rhxe3 fxe3 38.Ne4 Qd5 39.Nxd6 Qxd6 40.d5+. But <29...Kg7 30.Nbc5 Rc6> is interesting - might give Black better defensive chances than 29...Ne3 etc. I suppose White can do better than 31.Nf6, though. |
|
Sep-21-07
 | | kwgurge: I'm not sure the first two lines make any difference either. In the best case presented in the first line the engine rating in favor of white is +0.67. Arguably that is enough to say that black may already be positionally lost with good play. It would be interesting to see what a very low threshhold evaluation of the first 19 moves of the game produced, e.g., how the advantage of +0.67 by move 19 was slowly accummulated (presumably, it was a slow accummulation and not one or two weak moves based on the results of <MAJ's> .10 threshhold analysis to that point). |
|
| Sep-22-07 | | MostlyAverageJoe: <kwgurge: It would be interesting to see what a very low threshhold evaluation of the first 19 moves of the game produced> Well, since the game gets out of the book after only 10 or so moves, I will re-run the initial phase with 2 hours/move until move 20. This will have to wait until next week, though... |
|
Sep-22-07
| | kwgurge: Thanks <MAJ>. It may show us if the whole line is really unsound for black. |
|
| Sep-27-07 | | Hesam7: <kwgurge: Thanks <MAJ>. It may show us if the whole line is really unsound for black.> I agree, looking back I think our first mistake was 9...Nc6. Kasparov recommends 9...c5 and gives the following line: 10 dxc5 Bb7 11 O-O Nxe4 12 Nxe4 Bxe4. <It is here that the analytical duels are continuing to this day, and Black succeeds in maintaining the balance in them> -- Kasparov |
|
| Sep-28-07 | | MostlyAverageJoe: <Hesam7: I think our first mistake was 9...Nc6> This is still a "book" move. Hiarcs started real analysis at move 10, using 100 minutes/move (this is running on Intel duo, so it is really 200 CPU minutes per move). Once it is done, I can tell it to disregard the book and do perhaps moves 8 and 9. |
|
| Oct-01-07 | | Hesam7: <MostlyAverageJoe: <Hesam7: I think our first mistake was 9...Nc6> This is still a "book" move. Hiarcs started real analysis at move 10, using 100 minutes/move (this is running on Intel duo, so it is really 200 CPU minutes per move). Once it is done, I can tell it to disregard the book and do perhaps moves 8 and 9.> The fact that it is a book move does not mean that we can equalize after it, Kasparov's recommendation on the other hand seems to work. I am not so sure that using engines this early on in the opening is that productive. |
|
| Oct-03-07 | | MostlyAverageJoe: Well, I run a 40 CPU-hour analysis for the black's move number 9, and 9...Nc6 shows up on top (23-plies deep). More details later. I agree, however, that this is a bit early for computer analysis to be of much use. |
|
| Nov-01-07 | | The Chess Express: <hidude: i am devising a system for evaluating these positions without a computer... stay tuned> What did you come up with? |
|
Nov-15-07
 | | OhioChessFan: < Still awaiting MAJ's analysis > |
|
| Feb-17-08 | | D.Observer: Black is really hopeless since the ♕ exchange. |
|
| Mar-07-08 | | jovack: I don't see how you can enforce the "no computers" rule here... but interesting game nonetheless. |
|
Mar-18-08
 | | Sneaky: <MostlyAverageJoe: Well, I run a 40 CPU-hour analysis for the black's move number 9, and 9...Nc6 shows up on top (23-plies deep).> I remember when that move was played, its proponents argued that there is some theory behind it (it's not a TN) but I felt that it was positively horrible and that we should be trying to play ...c5, not making moves that make ...c5 impossible. It would be too strong to say that it was the losing move, but I really hated it myself. My interest in the game dwindled from 9...Nc6 onward. |
|
| Mar-24-08 | | The Chess Express: <Sneaky> What problems have you found with the <10. Be3 Ng4 11. e5 Nxe3 12. fxe3 Bh6 13. Ne4 Be6 14. Qc3 f5 15. Nc5 Bd5 16. O-O Nb8 17. Nd2 Nd7 18. Ndb3 Nb6 19. Na5 c6 20. Qd2 Nd7 21. Nab7 Qc7 22. Nxd7 Qxd7 23. Nc5 Qd8> line? |
|
| Jan-24-09 | | WhiteRook48: no need to gloat over the win, the dancing ♖ is annoying |
|
| Feb-08-09 | | WhiteRook48: the exchange rules! |
|
| Feb-09-10 | | Eyal: A recent tournament game where White finally tries the idea of Be3-e5 instead of 10.d5 or e5; however, he seems to go wrong almost immediately with 13.a4 (instead of Rd1 or 0-0), allowing Black to play c5 and shatter White's center. Geirnaert S. (2364) - Kovchan A. (2554) 2009 Ghent
1.d4 Nf6 2.c4 g6 3. Nc3 d5 4.Nf3 Bg7 5.Qb3 dxc4 6.Qxc4 O-O 7.e4 a6 8.Qb3 b5 9.Be2 Nc6 10.Be3 Ng4 11.e5 Nxe3 12.fxe3 Bb7 13.a4 Na5 14.Qb4 c5 15.dxc5 Nc6 16.Qe4 b4 17.Rd1 Qa5 18.Nd5 Qxc5 19.h4 Rad8 20.h5 Nxe5 21.hxg6 Nxg6 22.Rh5 e6 23.Qxb4 Qxb4+ 24.Nxb4 Rxd1+ 25.Kxd1 Rd8+ 26.Kc2 Be4+ 27.Bd3 Rb8 28.Bxe4 Rxb4 29.Bd3 Rxa4 30.Kb3 Ra1 31.Ng5 a5 32.Rxh7 Ne5 33.Bc2 Rc1 34.Be4 Nd7 35.Rh4 Nc5+ 36.Ka2 Nxe4 37.Rxe4 Rc2 38.Rf4 Rxb2+ 39.Ka3 Rb7 40.Rc4 Bf8+ 41.Ka2 Bb4 42.Ne4 Kg7 43.g4 Rd7 44.Ng5 Rd2+ 45.Kb1 Rd8 46.Nf3 Rd5 47.Rc7 Rd1+ 48.Kc2 Rf1 49.Nd4 Rf2+ 50.Kd1 Kf6 51.Nc2 Rd2+ 52.Kc1 Rd5 53.Kb2 Bd6 54.Rc4 Be5+ 55.Kb1 0-1 |
|
 |
|
|
< Earlier Kibitzing · PAGE 237 OF 237 ·
Later Kibitzing> |
Kibitzing
