Series on my old Dragon games

As you may know if you are a reader of Dennis M's blog, I am an aficianado of the Sicilian Dragon. I have always found this opening very exciting and fun, and the obsessive-compulsiveness of my high school and early college years has funded me with a pretty good understanding of this opening. Actually through much of high school I played the Accelerated Dragon, attempting to avoid the Yugoslav attack and spring traps on people who tried to play the Yugoslav against the Acclerated move order. Then I lost confidence in my pet line against the Maroczy, so I decided to switch to the normal Dragon. I've won some nice games with the Dragon, but also have ended up with my head on a plate. Some of these games will be wins, but some will be losses. This one is a loss.

This is the game that started it all. Joe Polchinski was the number two player on what would prove to be the first of a series of superstar teams to come out of Tucson. Before there was Will Wharton, Spencer Lower, Robby Adamson, Ken Larsen, or Tal Shaked, there was the Canyon Del Oro high school team. Its top player, Keith Nelson, had rocketed his way from class D to expert in about a year, and eventually became a master. None of these players are active now to my knowledge; Joe gave up the game after going to Cal Tech. These guys were so good they sponsored the Tucson Invitational and then retired the trophy if the same team won three years in a row. They retired the trophy.

Polchinski (A) - Reppert (B) Arizona Championship Semifinals 1970
1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 g6 6. Be3 Bg7 7. f3
O-O 8. Qd2 Nc6 9. Bc4 Bd7 10. h4 Qa5

This was the way the Dragon was played in those days. It fell into disrepute, but has been revived of late largely thanks to GM Chris Ward. I'm just guessing the move order here; but if you want to play the Moles line discussed on Dennis's site, you have to delay 10. h4 and play 10 O-O-O or 10 Bb3. For example, 10. Bb3 Qa5 11. O-O-O Rfc8 12. Kb1 Ne5 13. Bg5.

11. O-O-O Rfc8 12. Bb3 Ne5 13. h5

The line with 13. Kb1 actually put the Qa5 system our of commission for a long time, but Ward has had some success rehabilitating both ...Nc4 and ...b5. I once played ...Rc5 in this position and won. I hope to annotate a fascinating game I played against Ilan Brand with 13...b5.

13....Nxh5

Ward now recommends 13...Rxc3, which transposes to some of the established lines in the variation, but does not permit 14 Nd5 Qxd2 15 Rxd2 Kf8 16 g4. Ward forgot this idea in a game against nemesis John Nunn, who beat him badly in this queen-trade variation. However, I looked at 16...Ng3 and decided it wasn't bad for Black.

14. Bh6

14 g4 Nf6 15. Bh6 Bxh6 16. Rxh6 Rxc3 is considered at least OK for Black.

14. Nd3+!?

Two years earlier Karpov had defeated Gik after Bxh6 15.Qxh6 Rxc3 16.bxc3 Qxc3 17.Ne2 Qc5 18.g4 Nf6 19.g5 Nh5 20.Rxh5 gxh5 21.Rh1 Qe3+ 22.Kb1 Qxf3 23.Rxh5 e6 24.g6 Nxg6 25.Qxh7+ Kf8 26.Rf5 Qxb3+ 27.axb3 exf5 28.Nf4 Rd8 29.Qh6+ Ke8 30.Nxg6 fxg6 31.Qxg6+ Ke7 32.Qg5+ Ke8 33.exf5 Rc8 34.Qg8+ Ke7 35.Qg7+ 1-0. However, as Gufeld and Stetsko point out, Black has 18...Bxg4, which looks OK for Black. But while I was trying to set this up on my computer, I ended up finding 18 Rxh5!. Now if 18...gh 19 Rh1 Ng4 looks forced, but after 20 fg Bxg4 21 Nf4! and now if 21...Qe3+ 22 Kb1 Qxe4 23 Qg5+ Kf8 24 Nxh5 Qg6 25 Qxg6 hg 26 Nf6! Bc8 27 Nd5 b5 28 Nc7 Rb8 29 Rh8+ Kg7 30 Rd8 and White is winning. I think Dennis found a move to put up somewhat better resistance, but it's not good enough.

But of course 16... Qxc3 is wrong, but 16...Nf6 is perfectly good for Black, leading to wins for Black unless White plays perfectly, in which case it's a draw. This has pretty much put the 14. Bh6 line out of commission at the master level, and so it is no longer regarded as a critical line for the future of the ...Qa5 Dragon.

15. Kb1

An invention of German player Juergen Dueball. The alternative is 15. Qxd3.

Nxb2

When Dueball first played this against Mista in 1968, Mista played 15...Bxd4 and was rung up with 16. Nd5!

16. Kxb2 Bxh6 17. Qxh6 Rxc3

16...Qxc3 has been analyzed out to a loss for Black. But looking at it with the help of the silicon monsters, I am not so sure. It would be a good exercise to figure out whether White is really winning, or not.

18. g4 Nf6 19. g5

An important alternative is 18. e5, which Bob Rowley played against me the following year. That game requires separate treatment.

19...Nh5 20. Rxh5

The sub-text of this game is that Joe had been working out of his copy of the Dragon book of the time, which was a small Chessman Quarterly pamphlet written by Ray Keene, which was called Yugoslav Attack 1969. Mine was to arrive the next day. I was following analysis by, I think Honfi in Informant 7. According to the Informant,

20....gxh5?

is satisfactory for Black after 21 g6 Rxb3, so I played it. Keene had quoted analysis by Dueball, according to which 20...Rxb3+ 21. axb3 gxh5 22. f4 Rc8 23. f5 Qe5 24. f6 exf6 25 gxf6 Qg3 leads to a win for White. Looking at it with a couple of computer programs, I haven't been able to find the win (and Bob Rowley can't find it either), so I suspect they just had that wrong too. But the first thing to notice is that if the queen is headed for g3 in can save a tempo with 22...Qc3+. But after 23. Kb1 Qe3 is crushing, as I pointed out in Larry Evans' column a few months later. Unbeknownst to me, there was a game Segi-Velimirovic in 1970 which went 20. f4 Rc8 21. f5 Rxb3+ 22. axb3 Qc3+ 23. Kb1 Qe3 24. Rxh5 gh which reaches the same position. A game on the database from 1994 ended in a draw after 23. c4 in the Keene-Dueball line, but it looks as if Rxc4! secures an advantage for Black.

21. Bxf7+!

So much for Chess Informant. Does it perhaps occur to you that analysis coming from the books at this time maybe wasn't trustworthy? That we should have been looking for improvements? Of course, you can trust what's written in books today. And, by the way, did you see that ad I put in the paper? Some oceanfront property is available, right here in Arizona. (Dennis has some for sale in Indiana also).

Actually, analysis that reflects actual top-level practice probably won't have holes this big in it. That's doesn't mean it won't have holes, that just means the holes won't be quite so large.

21...Kxf7 22. Qxh7+ Kf8 23. g6?

Missing 23. Qh8+ Kf7 24. Qxh5+ and now either Kf8 25. Qh8+ Kf7 26. g6+ leads to mate, or 24...Kg8 25. Qg6+ Kf8 26. Qh6+ Kg8 27. g6 and the party's over, as Joe pointed out when he annotated the game for the Arizona Woodpusher. What he didn't know at the time was that Boleslavsky, in a book yet to be written would overlook this (just as he did during the game) and declare the game satisfactory for Black! So, let's take a toll here. Polchinski busts the Informant analysis over-the-board during the game, I bust the Dueball-Keene analysis, and Polchinski busts Boleslavsky analysis that hadn't been published yet!

23...Rxc2+ 24. Nxc2

Polchinski's notes to the game, as I recall, said that he could have won here with 24. Kxc2, but my programs and I don't see it after 24...Qxa2 25. Kc1 Qg8.

24....Qe5+ 25. Nd4 Qg7 26. Qxh5

Strangely enough, this analysis was given in a book by Boleslavsky that came out in 71 or 72 that gave all these moves as correct. What did I say about trusting published analysis? Black is OK here.

26...Rc8 27. Kb1 Rc5??

Missing a simple combination. Rc4 was OK for Black, in fact he looks a bit better here.

28. Qxc5! dxc5 29. Ne6+

And White won.