The Chess Mind

When I was a kid in the 1970s, there was a TV commercial featuring one person walking down a hallway enjoying his peanut butter (straight from the jar) and another person, approaching at a right angle, enjoying his chocolate. The men collide and their foodstuffs mix, but to their mutual delight, they find the combination of chocolate and peanut butter an improvement. Let's call this the Reese's Peanut Butter Cup Principle, or PBCP, for short: combine two things and you get a better thing.

Turning to chess: over the years I've worked as a chess teacher, many of my students have been fascinated by the complications of the Najdorf and Dragon Variations of the Sicilian Defense. Most have chosen one or the other, but every now and again, I find students who apply the PBCP to the Sicilian and play something like this:

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 g6 6.Be3 Bg7

So far, so Dragon.

7.f3 a6

Err...

8.Qd2 Nbd7 9.O-O-O b5



In the normal Dragon, Black doesn't waste time on a6, sometimes plays b5 but almost never this early, and finally, puts the knight on c6 rather than d7. Black's last three moves are Najdorf moves, not Dragon moves, while Black's 6th and 7th moves are Dragon moves virtually never to be seen in the Najdorf.

That's merely descriptive, however; it doesn't tell whether or why the combination of the two approaches is good, bad, or neutral. Those are normative questions, and my answer to them is, or at least was, that the combination is a poor one. In the Dragon, Black needs to generate queenside counterplay as rapidly as possible, and the problem is that the a6/b5/Bb7/Nbd7 setup will not put much pressure on White's king for a long time to come.

Imagine my surprise, then, when the strong French GM and openings specialist Igor-Alexandre Nataf played this very line against GM Aleksander Delchev just a few days ago! The game began with the exact moves above, and continued and concluded thus:

10.g4 Bb7 11.a3N Nb6 12.Bh6 Bxh6 13.Qxh6 Qc7 14.h4 d5 15.e5 Qxe5 16.Bxb5+! axb5 17.Rhe1



17...Ng8? 18.Qd2 Qg7 19.Ncxb5 Kf8 20.g5! Rc8 21.Qf4 Nf6 22.Rxe7! Nh5 23.Qd6 Nc4 24.Re8+! 1-0 (Delchev-Nataf, Herceg-Novi 2005)

Very pretty, and a nice object lesson for teachers wanting to scare their students away from this hybrid variation. Yet while Black's position was always precarious, White was not at all winning by force after 17.Rhe1. In short, I don't know if the "Najdorf Dragon" is a peanut butter cup or a combination of mustard and tuna fish. (Mmm...indigestion.)

Whatever its objective merits or demerits, it does have the advantage of being almost completely unexplored. So, for those of you who prefer the road less traveled, I suggest taking a look. Happy trails!

About a week ago, I received an email from "Daaim" suggesting I was perhaps unfair in my post on 1.e4 e5 2.Qh5 Nf6! He writes:

You show the game Hikaru lost on the ICC, but that shouldn't be your rationale for saying "I told you so." He's played many games with 2.Qh5.

Of course, the article was in good part tongue in cheek (re-read especially the first sentence of the second paragraph), but not entirely. First of all, that was the only ICC blitz game where his opponent played 2...Nf6; second, it's not just the result but how easily it came: by move 7 he's already clearly worse! Third, 2...Nf6 isn't just some "trappy" line - the computer in all its materialistic glory likes the move and thinks it gives Black an edge.

In an earlier comment to my post "Saidy-Fischer; not a Brilliancy?", MNb proclaimed the end of the "Fischer gambit". I think that even if he's right, it doesn't affect my argument in that post, which is that (a) the standard "refutations" of the sac are of at best dubious value, (b) even if the sac is bad, it's not obviously bad, and (c) its unsoundness, even if granted, doesn't suffice to eliminate the game's deserved recognition from the Informant voters back in 1969.

That said, it's still worth considering MNb's claims.

(1) MNb refers to an alleged refutation of the following line of the Grand Prix Attack (GPA): 1.e4 c5 2.Nc3 Nc6 3.f4 g6 4.Nf3 Bg7 5.Bc4?! e6 6.f5 Nge7 7.fxe6 fxe6 8.d3 d5 9.Bb3 with 9...b5. (His point is that this is equivalent to 1.c4 e5 2.Nc3 Nc6 3.g3 f5 4.Bg2 Nf6 5.d3 [because the pawn gets to d4 in two moves, the resulting position is exactly equivalent [albeit with colors reversed] to that reached in the GPA line] Bc5 6.e3 f4 and now 7.Nge2 [instead of the game's 7.exf4] fxe3 8.fxe3 d6 9.d4 Bb6 10.b4)

Unfortunately, while Cor van Wijgerden may have refuted the variation, I can't assess the claim here: I don't have access to the magazine in question, and as it's a position that has occurred in dozens of games going back to 1969, a look at the databases isn't sufficient. As a discussion opener, here's how GM Dorian Rogozenko continues the GPA line in his 2003 Gambit book Anti-Sicilians: A Guide for Black, pp. 27-28: 10.O-O c4 11.dxc4 dxc4 12.Qxd8+ Kxd8 13.Nxb5 cxb3 14.axb3 Bd7 15.Ng5 Kc8 16.c3 e5, when the position is somewhere between White's having adequate compensation and Black's having an edge (my computer thinks Black has about a +.3 edge if White continues with 17.Be3).

Let's suppose the worst, that Black is slightly better. In some sense, that's a "refutation" of the line, as it's silly for someone to choose a variation with White that leads to a slight disadvantage. But recall that we're really considering the line with colors reversed, so it would be Black with the slight disadvantage. Perhaps that's not ideal, but it's not really a big problem, either. So more needs to be said before this line should be considered refuted.

(2) Returning to the game, after 1.c4 e5 2.Nc3 Nc6 3.g3 f5 4.Bg2 Nf6 5.d3 Bc5 6.e3 f4 7.exf4 O-O 8.Nge2 Qe8 and now, instead of the game's 9.O-O, MNb suggests 9.h3, continuing 9...d6 10.Be3 exf4 ("[10...]Bxe3 is less strong now") 11.Bxf4 Nh5 12.O-O, which transposes (with colors reversed) to the game Glek-Gavrikov, Minsk 1983.

Two comments here: the first concerns the proposed analysis, the second its applicability to what I wrote and discussed on the program.

First, on his analysis: while the Glek-Gavrikov game wound up a very sharp draw, but I think he's objectively right about the position after 12.O-O - I think White is in very good shape there.

On the other hand, he might be underestimating 10...Bxe3. After 11.fxe3 exf4 12.exf4, the computer finds a very interesting move: 12...Nb4!? If White lets the steed remain, then Black may get good play with ...Qg6 or ...Bf5 - at best, White may get a very small edge with 13.O-O Qe3+ 14.Kh2 Qxd3 15.a3 Qxd1 16.Raxd1 Nc2 17.Nd5 Nxd5 18.Bxd5+ Kh8 19.Rd2 Ne3 20.Rc1.

The obvious question is this: why in the world White shouldn't just kick it away with 13.a3? The answer is that 13.a3 Nc6, White achieves nothing with the otherwise attractive 14.Qd2 (taking e3 away from the Black queen and preparing queenside castling), because of 14...Nd4, when 15.O-O-O?? Nb3+ reveals the point of inducing a3!

Unfortunately, White does get a good position with 14.Kd2!, when I was unable to find a foolproof way for Black to achieve adequate counterplay against White's plan of kingside expansion.

In sum, he's probably right about his analysis, but the next question is whether it's relevant to Fischer's gambit (as opposed to the way Fischer followed it up).

First, as I argued here, White can respond to 7...O-O(?) with 8.fxe5, when as far as I could determine, he achieves a large advantage. Second and more importantly, however, I think Black can improve on 7...O-O with 7...d6, and after 8.Nge2 O-O 9.h3, Black need not transpose to MNb's line with 9...Qe8 10.Be3, but can deviate with either 9...Bf5 or 9...Nd4 (for example). I think White has an edge here with accurate play, but I'm not at all persuaded we're in 1-0 territory.

(3) MNb proposes a third way for White: 7.gxf4 (instead of 7.exf4 as in the game, or 7.Nge2, heading for the GPA with colors reversed). He writes, "Black's best try seems exf4 8.d4 Bb4 as 7.gxf4 d6 8.Nge2 o-o 9.h3 Qe8 10.a3 is a GPA again; White has the useful extra move a3."

By way of reply: first, in the latter line, I think White has an edge but I'm not sure it's anything special. It's easy to overestimate the value of an extra tempo in a reversed opening - a line that's equal for Black rarely becomes clearly better by the addition of a single tempo. And second, his main line (7.gxf4 exf4 8.d4 Bb4) looks pretty healthy to me for Black - White is at best slightly better.

In sum, I think MNb has offered some good suggestions for White - not enough to win, as far as I can tell, but enough to make the gambit a dubious idea for postal chess or against a well-prepared opponent who can calculate well and play good defense.

Finally, three quick points about the variation. First, if Black plays the line with 3...f5, White should play 5.e3 instead of 5.d3. The point is to get in d4 in a single move - White might as well have an extra tempo if Black is going to head for a reversed GPA.

Second, after 5.d3, Black has good chances for equality with 5...Bb4. And third, after 5.d3(?!) Bc5(?!), it probably behooves White to flick in 6.a3, as the kneejerk response with 6...a5 weakens the b5 square. After White plays an eventual Nb5, d4 will be a well-supported positional threat to the Black position.

One of the best things a chess player can do if she's serious about improving is to find a complicated position and attempt to analyze it into the ground. So where does one find these rich, complicated positions for analysis? A great place to start is with games in the Poisoned Pawn Variation of the Najdorf Sicilian.

Earlier we looked (very briefly) at some historic games in the variation; here's a contemporary game featuring the current record-holder for youngest-ever GM, Ukraine's Sergei Karjakin (he achieved the title at the illness-inducing age of 12 years, 6 months):

Korneev,Oleg (2602) - Karjakin,Sergey (2634) [B97]
Dos Hermanas Dos Hermanas, Spain (7), 07.04.2005

1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Qb6 8.Qd2 Qxb2 9.Rb1 Qa3 10.f5 Nc6 11.fxe6 fxe6 12.Nxc6 bxc6 13.e5 dxe5 14.Bxf6 gxf6 15.Ne4 Qxa2 16.Rd1 Be7 17.Be2 0-0 18.0-0 Ra7 19.Rf3 Kh8



I was able to watch the game for a few minutes at this point and tried to make 20.Nxf6 work. Black's position looks disjointed, but his defensive resources are considerable. For instance: 20.Nxf6 Rxf6 21.Rxf6 Bxf6 (21...Rd7? 22.Qh6 Rxd1+ 23.Bxd1 Kg8 (23...Bxf6 24.Qf8#) 24.Rf3+-) 22.Qh6 Rf7 (22...Kg8? 23.Qxf6 Rd7 24.Rxd7 Bxd7 25.Qd8++-) 23.Bh5



at first glance looks pretty good for White, but Black survives with some tactical tricks. 23...Qxc2! 24.Rc1 Bg5!! (24...Qf5 25.Bxf7 Bg7 26.Qh5 Bf8 27.Qxf5 exf5 28.Rxc6 Bb7 is a little better for Black, but nothing too serious.) 25.Rxc2 Rf5!!



26.g4! Bxh6 27.gxf5 exf5 28.Rxc6 Be3+ 29.Kf1 Bd7-/+]

20.Rg3 Rd7

I thought this move was a novelty (but see below), one which unfortunately (but typically) seems to resolve the position into a forced draw, believe it or not - a result that has repeatedly occurred in the 10.f5 lines over the past 15-20 years. 20...Rg8? was played in an earlier game won quickly by Black, but don't judge by the result - Black is completely lost now! 21.Qh6 Rxg3 (21...Qxc2 22.Nxf6+-) 22.hxg3 Rd7 (22...Qxc2 23.Rd8+ Bxd8 24.Qf8#) 23.Rxd7 (23.Nxf6 isn't as good, but it's still sufficient. 23...Qxc2 24.Bh5?? But this is a terrible blunder. Instead, the nice interference move (24.Rd3! wins after 24...Qc5+ 25.Kh2 Bxf6 26.Qxf6+ Kg8 27.Qxe6+ Kh8 28.Qe8+ Kg7 29.Rxd7+ Bxd7 30.Qxd7++-) 24...Rxd1+ 25.Bxd1 Qxd1+ 26.Kh2 Qd3 0-1, Hakki(2310)-Iskov(2345), Eksjo-B 10th 1982.) 23...Bxd7 24.Bh5! Kg8 (24...Qxc2 25.Bg6 forces Black to pitch the queen with 25...Qc1+ in order to avoid mate.) 25.Nxf6+ Bxf6 26.Qxf6 wins.

21.Qh6 Rxd1+

[21...Rf7 22.Qh5 (22.Rxd7? Bxd7 23.Qh5 Rg7-+ and e8 is covered.) 22...Bf8? (22...Rxd1+ 23.Bxd1 transposes to the game.) 23.Rxd7 Rxd7 24.h3 and the joint threats of 25.Nxf6 and 25.Qe8 give White a decisive advantage. 24...Qb1+ 25.Kh2 Qb4 26.Nxf6 Qf4 27.Nxd7 Bxd7 28.Bd3 Bh6 (28...e4 29.Bxe4 Qxe4? 30.Qf7 forces mate.) 29.Bxh7 is completely won for White (29...Kxh7? 30.Qg6+ Kh8 31.Qg8#)]

22.Bxd1 Rf7 [22...Rg8?? 23.Nxf6 with mate in two.] 23.Qh5



It looks like White is winning: the rook on f7 is under attack, and there doesn't seem to be any good way to defend it. If 23...Rg7?? 24.Qe8+ mates in at most two, while 23...Rf8?? 24.Qg4 allows Black only the privilege of choosing death on g7 or g8. Therefore, the rook can't move, and counterattacking doesn't seem to work either. For instance, if 23...Qa4 or 23...Qc4, 24.Qxf7 Qd4+ (24...Qxe4 25.Qg7#) 25.Nf2 wins. Or if 23...Qa1, 24.h3 wins - the Rf7 still can't move away, but if it doesn't 25.Qxf7 Qxd1+ 26.Kh2 leaves Black unable to prevent 27.Qg8#. But there are still more possibilities...

23...Qa5

[23...Qd5 also seems to draw after 24.Qxf7 (24.Qg4 Qd8-+) 24...Qxd1+ 25.Kf2 Qxc2+ 26.Kf3 Qd1+ 27.Ke3 Qd4+=, while 23...Qb1 leads to insanity: 24.Kf1 (24.h3 Qb7-+) 24...Qb5+ 25.Be2 Qb1+ 26.Kf2 Qb7 (26...Qb6+ 27.Kf3 f5 28.Qxf7 fxe4+ 29.Kg4 Qd8 30.Kh5 Qf8 31.Qxf8+ Bxf8 32.Rb3 Bd6 33.Rb6 Bd7 34.Kg5 Be7+ 35.Kg4 Bd6 36.Rxa6+-) 27.Bxa6 Qa7+ 28.Kf3 Bd8 29.Nd6 e4+ 30.Kxe4 Rf8 31.Qh6 Rg8 32.Rxg8+ Kxg8 33.Bxc8 Qa4+ 34.Kf3 Qa3+ 35.Qe3 Qxd6 36.Qxe6+ Qxe6 37.Bxe6+=]

24.Kf1

[24.Qxf7?? Qe1# is terrible, of course, but 24.h3 again seems to leave Black on death's door. For instance: 24...Rf8 (24...Qe1+ 25.Kh2 Qxe4 26.Qxf7+-) 25.Qg4 Qe1+ 26.Kh2 Qxg3+ 27.Nxg3 and while Black isn't doing too badly on a superficial material count, his pieces don't work well and the pawns are more likely to be vulnerable than a strength. White is winning. But it turns out that Black has another nifty tactical resource here: 24.h3? Qc7!! and Black wins: 25.Qg4 (25.Qxf7?? Bc5+ 26.Nxc5 Qxf7-+) 25...Qd8 and although White's attack is over, Black's three pawn advantage remains.]

24...Qd8

Black threatens to play 25...Rg7, when White will have no attack to compensate his material deficit, so it's time to cash in and force the draw.

25.Qxf7 Qxd1+ 26.Kf2 Qxc2+ 27.Kf3

[As I had Powerbook 2004 loaded when I first analyzed the game (rather than the 2005 model), it was only later that I discovered that almost the entire game was a repetition of an earlier effort. (Sadly, that too is typical of Poisoned Pawn games - see my post on fake draws). It's only here that we have a new move, as the different but equivalent 27.Ke3 led to the inexorable draw after 27...Bc5+ 28.Nxc5 Qxc5+ 29.Kd2 Qf2+ 30.Kc3 Qd4+ 31.Kc2 Qf2+ 32.Kc3 1/2-1/2 Vallejo Pons,F-Kasparov,G/Moscow (Armenia vs. ROW) 2004.]

27...Qd1+ 28.Kf2 Qc2+ 1/2-1/2

An exciting and interesting game in its own right, though sadly a sham (see the previous post for more thoughts on this topic!), as it's inconceivable that they were unaware of the Kasparov game and the "novelty" offered nothing new. (I hope that at least the live spectators were entertained, but I expect the internet audience was rightly disappointed.)

Apropos the relatively recent posts (see here and here) on 1.e4 e5 2.Qh5, I'd like to discuss one way in which most - if not all - of us have been miseducated. Many players go through the following early chess history:

Stage 1. Get mated in 4 moves by 1.e4 2.Bc4/Qh5 3.Qh5/Bc4 4.Qxf7#

Stage 2. Endeavor to mate everyone else that same way.

Stage 3. Use the queen as the chess equivalent of a wrecking ball even when not attempting (or succeeding) in employing the 4-move mate.

Stage 4. Learn via hard experience, beginners' books and/or a mentor that bringing one's queen out early (especially towards the center) is a bad idea, leading to such unpleasant outcomes as her getting trapped or the other side's gaining numerous tempi by attacking the queen with developing moves.

Stage 5. Eschew early queen moves, enforcing an almost Boo Radleyish existence on the lady through at least move 15 or so.

Most of us, I think, gradually overcome the bad advice of stage 4, even though we still do our best to propagate it with those juniors and other new players with whom we have influence. Why do we do this? Two reasons come to mind, and I'll endeavor to address both: first, we still, deep down, believe that early queen moves are bad; second, we think the rule is a useful fiction.

Let me start the attempt to break the rule's grip on our chess psyches by providing some examples of perfectly good early queen moves:

For White:
1.e4 e6 2.Qe2!?
1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.Qg4!?
1.e4 e5 2.Qh5 (on any but 2...Nf6!)
1.e4 e5 2.d4 exd4 3.Qxd4 Nc6 4.Qe3/Qa4
1.e4 e5 2.Bc4 Nf6 3.d4 exd4 4.Nf3 Nxe4 5.Qxd4
1.e4 e5 2.Bc4 Nf6 3.Nc3 Nxe4 4.Qh5 Nd6 5.Bb3 Nc6 6.Nb5 g6 7.Qf3 f5 8.Qd5
1.e4 e5 2.Nf3 Nc6 3.d4, when 3...exd4 4.Nxd4 Nxd4? 5.Qxd4 is great for White.
1.e4 e5 2.Nf3 Nc6 3.Bc4 Be7 4.d4 exd4 5.c3 dxc3 6.Qd5!
1.e4 e5 2.Nc3 Nc6 3.Bc4 Bc5 4.Qg4!
1.e4 e5 2.Nf3 d6 3.d4 exd4 4.Qxd4
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Qxd4
1.d4 d5 2.c4 c6 3.Nc3 Bf5?! 4.cxd5 cxd5 5.Qb3
1.d4 f5 2.Qd3
1.c4 Nf6 2.Nc3 g6 3.Nf3 d5 4.Qa4+ (Bd7 5.Qb3)
1.c4 Nf6 2.Nc3 g6 3.Nf3 d5 4.cxd5 Nxd5 5.Qa4+ (Bd7 6.Qh4)

For Black:
1.e4 e5 2.f4 Qh4+ 3.g3 Qf6/Qe7
1.e4 e5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Qh4!?
1.e4 e5 2.Nf3 Nc6 3.d4 cxd4 4.Nxd4 Bc5 5.Be3 Qf6
1.e4 c5 2.c3 Qa5
1.e4 c5 2.c3 d5 3.exd5 Qxd5
1.e4 c5 2.Nf3 e6/Nc6 3.d4 cxd4 4.Nxd4 Qb6
1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6 6.Bg5 e6 7.f4 Qb6
1.e4 d5 2.exd5 Qxd5
1.d4 Nf6 2.c4 e6 3.Nc3 Bb4 4.Qc2 d5 5.cxd5 Qxd5
1.d4 Nf6 2.Bg5 Ne4 3.Bf4 c5 4.d5 Qb6
1.d4 e6 2.c4 b6 3.e4 Qh4

and so on.

Not all of these lines are equally good, but they're all at least playable, despite violating the so-called rule against developing the queen early - especially towards the center. In fact, the rule is baloney, because there are so many exceptions and borderline cases that such a coarse-grained principle is of almost completely valueless. Worse, because the advice is offered at a formative stage in the player's career, they might have a relatively tough time unlearning it.

As a parallel case, to show just how powerful the effects of powerful but misleading lessons can be, consider the long-lasting trauma suffered by the Advance Variation of the Caro-Kann, brought about by the famous old game Nimzovich-Capablanca, New York 1927.

That game, which began 1.e4 c6 2.d4 d5 3.e5 Bf5 4.Bd3 Bxd3 5.Qxd3, put the variation out of business for 50 years, with a brief interlude in the 1960s, thanks to Tal's not-particularly-successful employment of the variation in his world championship matches with Botvinnik.

I suspect the reasoning went something like this: after 3...Bf5, either we play 4.Bd3 or we don't. If we do, then we have a French Defense in which Black has painlessly exchanged his bad bishop; if we don't, then the Bf5 will be beautifully placed! Only in the late 80s and early 90s did the strongest and most creative players on the planet turn to the position with fresh eyes, and then the line returned to the scene with a vengeance. (For those of you don't know how this happened, I leave it to you as a matter of research and, better still, an exercise in creative thinking.) One game - one single bit of "textbook orthodoxy" - dimmed grandmaster eyes for 60 years!!

Thus, if even the elites of the game can be so powerfully influenced - practically snowed - by a single game, then how much more are we harming young players with our pseudo-Siniatic commandments.

So what should we do? I think two approaches are in order, and both will achieve the same positive goods without any accompanying harms.

First, one can tell juniors that it can be dangerous to send the queen on early fishing trips, and explain why, but then go further and try to show what some of the common perils are (so they know what to look out for) but also when the queen is perfectly healthy in the center. A more fine-grained rule, worked out with the student him- or herself, will be both more accurate and more likely to foster their development as a thinker and a chess player.

Second, present all the positive things one can do with the pieces, so that they look for good moves (both for themselves and their opponents) rather than going through some artificial process of excluding bad ones. Let them keep the active mindset that motivates the queen-as-godzilla approach, but help them see that every piece - both individually and especially when working in concert - can have tremendous power as well.

In sum, it's better to assume something can be done and learn one's limitations than to assume limits and not even try. That's not a good blanket rule for life (kids: don't assume you can fly from the rooftop until gravity teaches you otherwise), but at least at first glance, it does seem the right way to approach the game.