Greengard's ChessNinja.com

Black Day in Poikovsky

| Permalink | 32 comments

Running out and no time to look, so please tell me that Riazantsev-Karjakin wasn't all theory. Really wild king chase. Three black wins today.


up to 12.Nf5 sacrifice it repeated Radjabov-Leko -Linares 2008,
Radjabov continued with 12.Qe5 and after 12..f6 13.Qe4 Qc8 15.Rd1 Nac7 played a knight sac 15.N:g6 h:g6 16.Q:g6+ Kd8 with complicated play

it is very sharp after 10.Qe4 thus it is not likely that anyone would enter such variation without home-prep but to find out when OTB started, and for who, a computer analysis would be needed.

( the same with that Q:b4 queen-sacrifice followed by recapturing the queen in a game which you pointed out few days ago, it was a sharp Vienna all theory up to black Nb4 [Ne5] and the sacrifice itself is thematic, the Nb4 and Q:b4 "sacrifice" followed by N:f6 taking back the queen could have been (and was of course in few games) 2 moves before, instead of Qc7..when exactly, and for who, the OTB game started it isn't easy to say)

There was also Svidler-Karjakin Sochi 2008 (later the same year as Radjabov-Leko):

[Event "Sochi FIDE GP"]
[Date "2008.08.11"]
[Round "10"]
[White "Svidler, Peter"]
[Black "Karjakin, Sergey"]
[Result "0-1"]
[ECO "E15"]

1. d4 Nf6 2. c4 e6 3. Nf3 b6 4. g3 Ba6 5. Qc2 c5 6. d5 exd5 7. cxd5 Bb7 8. Bg2 Nxd5 9. O-O Be7 10. Qe4 Na6 11. Nh4 g6 12. Qe5 f6 13. Qe4 Qc8 14. Rd1 Nac7 15. Nxg6 hxg6 16. Qxg6+ Kd8 17. a3 b5 18. e4 Nb6 19. Nc3 Ne6 20. b4 cxb4 21. Nd5 Nf8 22. Qg7 Rh7 23. Qg8 Nxd5 24. exd5 d6 25. axb4 Qf5 26. Be3 a6 27. Rac1 Rc8 28. Bb6+ Ke8 29. h4 Qg6 30. Qxg6+ Nxg6 31. Rxc8+ Bxc8 32. Rc1 Bf5 33. Ra1 Rg7 34. Kh2 Bc8 35. Rc1 Bb7 36. f4 Bd8 37. Bd4 Kf7 0-1

12. Nf5 smacks of Deep Rybka 4 to me. I got my nifty copy two days ago and it goes to 12. Nf5 within a few seconds and stays on it for at least the five minutes I kept it running on my quad-core. The players in fact stick with DR4's recommended best line until Riazantsev's 15. Nc3 (DR4 gives 15. Bd5 instead, but only by a tiny fraction of a point, and that eval may easily change if I'd let it run a little longer). The next major deviation by Rianzantsev from DR4 in the 15. Nc3 line comes with 16. Be4 instead of taking on d5 with either knight or bishop, which DR4 prefers (although again that's only with a few minutes search).

A quick and shameless plug for Deep Rybka 4; it is an absolute MONSTER. I'm only rated just over 2100 Correspondence, but the playing level seems extraordinarily higher. I've done about a dozen blitz matches on a quad-core vs. some of the highly touted "rogue" or free engines and so far they haven't even managed to DRAW a single game vs. DR4. SMOKIN!

@12. Nf5 smacks of Deep Rybka 4 to me

Well now, Fruit or Crafty would be enough if you have powerful hardware.

Exactly this kind of positions, complex and unbalanced (king insecurity/attacked but powerful center-pawns roller, in this game) are what one aims for when doing comp-prep because such positions are difficult to evaluate OTB, one can't play them well if that's the first time he sees them.

In the match Anand-Topalov one could the idea at work in the G7 involving the exchange sacrifice by Topalov : the position may have been equal or white (Anand) objectively slightly better but it was an unblanced, difficult to evaluate OTB,
position. White could have lost despite being a piece up because balck's passed pawns, if fact Anand went wrong with the "natural" 25.Nd2 and then had to find the only defence(26.Ra1) after Bb4. On the other hand Topalov was familiar with the position and its resouces.
And the same goes with this time Anand home-prepared Q-sacrifice for 2-rooks in G9. White has many powerful ideas 25.Be4 and a dangerous, long lasting, initiative.

@ The players in fact stick with DR4's recommended best line until Riazantsev's 15. Nc3 (DR4 gives 15. Bd5 instead, but only by a tiny fraction of a point..

in this case both prepared but Karjakin home-work (and hardware ?) was of better quality

Maybe 15.Nc3 and 16.Be4 (eventually regaining the sacrificed piece in a different way) were deliberately played, despite or indeed because it's Rybka's second-best move? Riazantsev may have gambled that Karjakin's homework (and memorization) focussed on the engine's preferred line?

But with all the hype around Rybka, we shouldn't forget that humans also can play chess ,:) . I don't remember clock times in detail from watching live, but at some stage Karjakin was far behind on the clock - something like five minutes against an hour for Riazantsev. So it seems he was outprepared and had to think over the board.
"Karjakin home-work (and hardware ?) was of better quality" - in any case there is no evidence from moves 12-16, black's moves all seem forced and weren't that difficult or inhuman.

Good news that human beings can still survive against computer-assisted preparation. Anand-Topalov G7 is another example, a third (dated) one I remember vividly from watching on site is Ponomariov-Kramnik, Corus 2005
Kramnik far behind on the clock, live commentators were "preparing his funeral", draw in the end.

But the closest analogue to Riazantsev-Karjakin may be Gashimov-Grischuk from January this year:
Here the black king went even one step further (to b1) and took part in a mating counter-attack rather than "only" supporting passed pawns. Engines were laughing at both players, we still have to get their verdict on the entire game played today ... .

Daring to play this line against Karjakin, even with some computer backup, as it's the practical experience that counts. My "clone" of choice switches somewhat between Nf5 and Qc4, just as Rybka does. I don't know how you get your data, as multi cpu is not reproducible, but Rybka 4 on one core likes Qc4 at "depth 16" (about a minute). If you put the "clones" into multipv mode (don't try this with Rybka -- see below), then Qc4 looks better.

All the ratings lists, or at least those not controlled by the commercial interests, put Deep Rybka 4 at best a small margin (10 elo) above the so-called "clones". Also Rybka has some bugs, the most annoying being that it tends to "stall" in MultiPV mode. Simple draws are also missed. Try this one (w/o tablebases -- even though Rybka can give "tbhits" when you don't have them):

8/8/8/8/7p/7P/6P1/5kbK w - - 0 1

Rybka thinks the game is a draw, at whatever depth. I don't think there has an announcement as to whether such bugs will be fixed by Rajlich. For many, Rybka 4 has been a disappointment.

I tried popping the position you mentioned into my DR4, being skeptical of your claim. Guess what happened? The result was even more odd than what you suggested:

New game
8/8/8/8/7p/7P/6P1/5kbK w - - 0 1

Analysis by Deep Rybka 4 x64:

1. = (0.00): 1.g4 hxg3 2.h4[] g2#
2. = (0.00): 1.g3 hxg3 2.h4[] g2#

DR4 finds the mates, but evals the position(s) as "="! How odd...

Here is Rybkaforums discussion of the matter: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=17407

How about Rublevsky-Sokolov as well. The game went from a win to a loss in one move. White had 26.Rxc6! Qxc6 27.Rxc6 Kxc6 and now the black pawn looks unstoppable. But check out the great typo shot 28.Qxh7! and if the rook takes the queen the a-pawn queens with check and a winning skewer. Very cool.

What about 26.Rc6: Qc6: 27.Rc6: d2 28.Rd6!? Bd3 !?

a nice miniature "a la Morphy" today


1.d4 d5 2.c4 e6 3.Nf3 dxc4 4.e3 c5 5. Bxc4 a6
6.d5 b5 7.Bb3 exd5 8.Bxd5 Ra7 9.e4 Nf6
10.Nc3 Nxd5 ?! [10..Bg4] 11.Nxd5 Nc6 12.0-0 Be6 13.a4 Be7 ? [13..b4 ] 14.axb5 axb5 15.Ne5 ! +/-
15.. Na5 16.b4 Bxd5 17.exd5 cxb4 18.Rxa5 ! Qxa5 19.Nc6 Qa4 20.Qd4 Bf6 21.Qe3+ Kd7 22.Nxa7 b3 23. Qb6 1-0

Nice game.
Rublevsky reminds me of Sveshnikov; he has a very narrow opening repertoire and plays open games as White. This handicaps him a bit when he plays matches against heavy hitters, I think. They find it easy to prepare against him. I always follow his games with interest. Except his QGA games; with a few exceptions like the above, that opening bores the heck outta me for both colours.

It is not too odd that Rybka gets the PV. After all, the PV comes from the hash, and White's moves are forced, so they should all be there. Actually, in text UCI mode, it doesn't give me the PV, so maybe the GUI is doing something with forced moves?

The oddity is that 0.00 appears, and moreso, that it did for the previous 10+ moves in that game. If Rybka shows the mate with TBs, just like mating with B+N, it will probably go into the growing list of things (like bishop underpromotion) that are "Rybka quirks" as it were. Even the blind bishop problem with Rybka 3 was thought not too important because usually TBs would suffice there. Fortunately that is solved now.

I'm looking for Zappa to make a comeback with Zach Wegner at the helm now. He also said he could have big hardware for the WCCC game. Hopefully it won't be a boring draw.

The famous f4?? blunder of Rybka against Zappa (into a B+3P vs B draw) back in the 2007 challenge match (Game 9) shows why you can't just rely on search but do need knowledge too. I think Kaufman's work on trading off pieces helped solve that specific problem, but the Game 4 disaster (RRB vs Q+Ps, and Rybka gives away pawns to avoid the 50-move rule, going on to lose) might still exist.

I don't see a thread for Capablanca 2010, so I'm hijacking this one...I really enjoyed the Short-Chucky game. Fun-fun-fun from the second move. Chucky convincingly clobbered Alekseev today as well.

Off topic at the moment but I was reading the bios for the players in the King's Tounament which starts in a couple off days.

"Wang Yue born at 31 March 1987 in Taiyuan (China).
Best Chinese chessplayer all-times, the first to reach the rating of 2700 (and the 3rd in Asia to make it)."

Who was the 2nd?

Maybe they overlooked that Wang Yue was already at 2703 in the October 2007 list, but then fell below that magic number again. In April 2008, actually both Bu Xiangzhi (2708) and Ni Hua (2704) were ahead of Wang Yue (2689), but later Wang Yue was the only one who not only stayed above 2700, but made further progress.

I believe that Sasikiran touched 2700 in Jan 2007.

Rustam Kasimjanov

I stand corrected .

Both chessplayer and Umesh are right. This shows that touching or crossing 2700 isn't enough to "make history" unless you maintain and further improve on such a rating - it isn't even enough to become FIDE knockout world champion once ... .

Five or ten years from now, how many people will remember that players like Vallejo Pons or Naiditsch are or were part of the 2700+ club?

Kasimdzhanov crossed 2700 a long time ago, back in 2001, when 2700 is not common, and perhaps still relatively significant. It brought him to 11th spot. By the time Sasikiran, Bu, Ni, and Wang Yue reached it, 2700 was already more common.

That's why I always thought rank are more significant than rating (for example, being top 5, whatever the rating). In the last 40 years, since the beginning of Elo ratings in 1971, only 42 players have ever reached top-5.

More people than will remember Thomas.

Both Vallejo Pons and Kazimdzhanov are great players. They just lack the consistency to be at the top of their game all the time.

"To want fame is to prefer dying scorned than forgotten."

"In the last 40 years, since the beginning of Elo ratings in 1971, only 42 players have ever reached top-5."
Interesting, where do you find such information? No doubt Vallejo and Kasimdzhanov are great players on a good day or week, even greater ones (Topalov and Anand) picked them as their seconds ... .

BTW, I had ignored or forgotten about Kasim for two reasons:
- somehow I didn't think about Uzbekistan, because he is by far the strongest player from that country
- and I may have subconciously lumped it with the Soviet Union, not including it within Asia.

But I _was_ thinking about Eugenio Torre from the Philippines. His peak rating was 2580 in January 1983, good for #20. Back then, 2600 (Korchnoi and Seirawan) meant shared 12th place, and Karpov (2710) was the lone member of the 2700+ club. So much about rating inflation ... .

I compiled the information from checking all Elo rating lists since 1971. That's a lot of lists :-), but it's not that difficult since I only needed to check the top 5 (and the 6th, just in case his rating is equal to the 5th, I will count him as well). Besides, most of the time top-5 names repeat, so I only needed to browse quickly through the lists.

Here I copy the list of players who have been top-5 in the Elo list since 1971, in alphabetical order:

Adams, Anand, Andersson, Aronian, Bareev, Beliavsky, Carlsen, Ehlvest, Fischer, Gelfand, M. Gurevich, Huebner, Ivanchuk, Jakovenko, Kamsky, Karpov, Kasparov, Korchnoi, Kramnik, Larsen, Leko, Ljubojevic, Mamedyarov, Macking, Morozevich, Petrosian, Polugaevsky, Ponomariov, Portisch, Radjabov, Salov, Shirov, Short, A. Sokolov, Spassky, Speelman, Svidler, Tal, Timman, Topalov, Vaganian, Yusupov

Sorry, "Macking" should be Mecking

Torre was by far the strongest asian player in the 1970s and 1980s, and he went to the candidates. However, as far as rankings are concerned, he didn't go higher than the margins of top 20, and only for a short period. Wang Yue and Kasimdzhanov (and Anand of course) seem to have done better in that respect.

As a comparison, I also checked Sonas' list for approx. the same period, starting 1971, but ending in January 2005.

There are 40 players in the top-5 in Sonas' list for that period. Most (33) names appear in both Sonas and FIDE top-5, as expected.

Sonas' top-5 included some that are not on the FIDE top-5 (for that period): Geller, Grischuk, Keres, Polgar, Romanishin, Stein, Yudasin

On the other hand, Sonas does not include: Ehlvest, M. Gurevich, Huebner, Speelman

Also, since Sonas' list ended in 2005, its top-5 doesn't include several recent top players: Aronian, Carlsen, Jakovenko, Mamedyarov, Radjabov

If we count everybody who has ever been in top-5 since 1971, either in the FIDE or Sonas list, we get these 49 names:

Adams, Anand, Andersson, Aronian, Bareev, Beliavsky, Carlsen, Ehlvest, Fischer, Gelfand, Geller, Grischuk, M. Gurevich, Huebner, Ivanchuk, Jakovenko, Kamsky, Karpov, Kasparov, Keres, Korchnoi, Kramnik, Larsen, Leko, Ljubojevic, Mamedyarov, Mecking, Morozevich, Petrosian, J. Polgar, Polugaevsky, Ponomariov, Portisch, Radjabov, Romanishin, Salov, Shirov, Short, A. Sokolov, Spassky, Speelman, Stein, Svidler, Tal, Timman, Topalov, Vaganian, Yudasin, Yusupov

[White "Carlsen, Magnus"]
[Black "Wang, Yue"]

1. e4 e5 2. f4 d5 3. exd5 exf4 4. Nf3 Nf6 5. Bc4 Nxd5 6. O-O Be7 *


Since the tournament started Carlsen has been trying hard to win a game but all his attacks (like those against Ponomariov and Gelfand) went to nowehere and Magnus had to backpedal fast to save himself.
Let's see how he does today, but the 5.Bc4/6.Bxd5 line is not dangerous for Black, 5.Bb5+ is the main line

King's Gambit score!

In the game mentioned, while my opponent was playing d4, c3, e3, a3 and h3 plus putting a couple pieces on passive squares I had managed to complete development except for castling and connecting R’s. In the process he had also abandoned any semblance of influence on the center by playing d4xc4. The result was crushing defeat for him.Players are often encouraged to not study openings at all or go to the opposite extreme and study an opening (usually an inferior one) until they know it like the back of their hand. In the latter case they are told that’s a way to win more games. Once out of their book knowledge everybody plays to their rating though.

Twitter Updates

    Follow me on Twitter



    About this Entry

    This page contains a single entry by Mig published on June 10, 2010 12:38 PM.

    Draw Odds in the WCh was the previous entry in this blog.

    KiB is the next entry in this blog.

    Find recent content on the main index or look in the archives to find all content.