John Watson Book Review (45)
Excelling at Chess
IM John Watson - Friday 9th August 2002
Excelling at Chess by Jacob Aagaard
Excelling at Chess
190 pages; Everyman Chess, 2001
I can't recall an example, although I imagine that some exist, of a chess book that directly attacks another at some length, so when Danish IM Jacob Aagard wrote me that he had written a book 'which goes into infight with your Strategy book', 'opposing yours more or less 100%', I was surprised, but tried to take it as a compliment. For any time that a titled player such as Aagard feels compelled to present his case in such a venue, it at least shows that the target (in this case my 'Secrets of Modern Chess Strategy') is attracting serious attention. When Aagard's 'Excelling at Chess' arrived, I found the relevant material in parts of Chapters 1 and 2, and just about all of Chapter 3, entitled 'No Rules?'. Here I'd like to invoke the self-defence principle and address his case as well as I can given such a complex topic. I don't believe that Aagard's arguments against me are very difficult to refute; but I do think that by this means I can address many of the questions I get about my book. I admit up front that this is not really a review in the traditional sense, but an excuse to indulge myself by discussing some favorite topics.
Aagard's introductory chapter quotes a New In Chess article from 10 years ago which said that every position has its own set of rules and linked that theory to Kasparov's play. Aagard says: 'I know that the former world champion would never say such nonsense'. Reti, who said precisely that, might be looking down with hurt feelings; but this gives us a feeling for Aagard's basic philosophic position. The next relevant point has to do with Kasparov's actual comment that there were only five or six 'real' chess players in the world. Aagard tries to interpret this for himself and comes up with: 'A Real Chessplayer is someone who knows where the pieces belong' and 'I believe that "Real Chess Players" would never put a piece on awkward squares, only lesser players do so.' The second claim (especially with the 'never') is a bit silly, but we can understand what he is saying. The first statement, in fact, is almost self-evidently true. But it is also too vague to be meaningful. A more interesting question addressed in my book is how strong players know where the pieces belong. Borrowing from what many others have written, I suggest that in most cases they employ: (a) pattern recognition (Rowson mentions 100,00 positions absorbed on the basis of experience - I suspect that these days the number is even higher); (b) calculation, e.g., however attractive an elegantly placed piece may be, calculation can and often does lead to the conclusion than an awkward placement is the superior one; and finally, players will use their (c) judgment/intuition (hard-to-define but sometimes unavoidable words), these last are also strongly informed by pattern recognition and by concrete examination of lines, of course, but in addition by creative balancing of many often subtle positional factors that would only be describable in words by a lengthy essay (i.e., not by abstract generalities). See page 103 of my book for an example of what I mean. This is one of the several senses in which I talk about the gradual divestment in modern chess thinking of the multitude of 'explicitly-stated generalities, abstract principles and rules' that still dominate our textbooks. I don't know whether those rules are useful in an instructional sense for beginning or low-level players, which is a different and complex issue. I do try to describe the nature of modern chess thinking employed by strong players. None of the above seems to me to be very controversial, much less brilliant - the more difficult and interesting issues arise when one discusses the specifics of pawn and piece play with these concepts in mind.
Before moving to specific examples, I should note as well (and this may be a language problem) that rule independence in my book refers to not being dependent upon certain rules, abstractions and generalities in practice; the phrase doesn't imply that there is no truth in any general consideration whatsoever. For example, it would be dogmatic to never consider sacrificing the exchange due to an abstract notion of material. And modern players, it is generally acknowledged, are increasingly aware how often compensation (sometimes very subtle) exists for the exchange, i.e. they are increasingly independent of the older conceptions of material. I make precisely this point in my book. But to deny that material imbalances in general have significance, or to have no awareness that rooks are generally more valuable than minor pieces, would be wholly detrimental to one's play. Similarly, Aagard tries to strengthen his case against me by saying that 'winning pawns is a good idea' is a rule. Well, I wouldn't call this a rule, but the comment is a red herring anyway, because for obvious reasons my book doesn't take issue with the validity of factoring in material gain. It concentrates instead upon specific generalities that are losing relevance in today's game. Thus I repeatedly make the point that the movement in modern chess is away from general rules and towards a more open, concrete, and realistic view of the board. Aagard argues as if the debate centers around whether a rule (or generality, or even a positional consideration) has an abstract validity in the sense of being true in a majority of cases, rather than discussing, for example, whether it is useful to think in terms of that rule when confronted with a specific position. The burgeoning number of exceptions and the willingness of strong players to consistently ignore classical rules and conceptions have characterized modern chess, and players have expressly indicated their growing preference for concrete discussions (verbal as well as analytical). Note too that 'concrete calculation' doesn't mean just lining up moves in your head. It can involve seeing further into the position and understanding that at one point the opponent won't be able to stop you from getting passed pawns or some such. That is a positional insight, and not a rule. Neither, for example, would be the desire to coordinate your pieces. Finally, it's not that logic has become less relevant, as Aagard seems to think the phrase 'rule independence' implies. Indeed, one could argue that logic is more strictly applied in modern chess. The now-standard moves 1.d4 Nf6 2.Bg5 Ne4 3.Bf4 c5 4.f3 Qa5+ 5.c3 Nf6 6.d5 Qb6 7.Bc1 are logical, but someone who learned chess from classical principles might look askance at moving his bishop three times to arrive at its starting position; or at having no pieces developed, even after 7...g6 8.e4 d6 9.a4 or 9.c4, for example. My book examines how strong players have adopted a more realistic, creative, attitude towards the game, sometimes instantiated in ways that would have shocked their predecessors.
Turning to examples that Aagard argues with, I have a four-page section (starting on page 104) about moving pawns in front of one's king, referring to Steinitz' old admonition not to do so. I argue that in the modern era, old inhibitions about such moves have gradually broken down. Even in theoretical positions that have been analysed for decades, players have discovered helpful pawn thrusts in front of their king, ones that at first sight seemed unthinkable. But the notion behind them soon became accepted and spread to other positions with more or less resemblance to the original. I find this a characteristic of modern play and adduce reasons for it. Does Aagard try to show that this is not the case? No, he says only that 'Those arguing against any form of rules in chess will find it difficult to counter this: Pawn moves in front of the castled king create structural weaknesses and make life easier for the attackers.'
My claim, however, is that players are becoming less and less inhibited about making such moves (which by the way often stop the opponent's pawn attack cold). As Aagard knows, I never suggest that one should move the pawns in front of one's king in most games. Rather, I contend that adhering to some principle about not moving such pawns can limit one's play. In today's chess, I propose, strong players are not thinking in terms of this rule. Instead, they are more open-minded, looking concretely at the actual position and seeing for themselves whether such a pawn move 'makes life easier for the attacker', or whether it has various advantages that make it worth playing. That's pretty simple to understand. One might compare my discussion of the now-routine attack upon the front of a pawn chain rather than at its base.
In discussing the old saw about developing knights before bishops, Aagard says that we should judge rules with the proviso 'all things being equal' before we 'cynically judge them as false.' But as Aagard well knows, I make no such judgment, cynical or otherwise. Rather, with a detailed introductory explanation of how this was probably no more than a general guideline in Lasker's day, and that 'it turned out to be a usable rule in the classical openings', I proceed to mention eight examples of its appropriateness in that context. Furthermore, I explain the reasoning that supported the principle then. I then turn to many examples of modern openings in which the bishops are developed before the knights, including some exact positions in which players of earlier times developed their knights first. I give a philosophic basis for preferring the bishop development in certain positions, and suggest that the rule is of doubtful value as the basis for playing the opening today. Of course this idea (Ns before Bs) is more of a textbook notion and probably the least important aspect of my minor-piece discussions; still, the ubiquitous counterexamples are an indication of how modern opening play has evolved. Aagard then resorts to aspersions: 'When Watson uses a lot of new opening ideas to invalidate a wise old observation, he is not doing it in the service of research, but purely to promote his own argument.' He compares me to a politician simplifying to promote false arguments. And as if that were not enough, he gives me the charming backhanded compliment that he loves my book (which he calls trendy), but only because he considers it 'a kind of "greatest hits" of the last ten years' best books.' Apart from being insulting, this isn't even remotely accurate, as one can easily verify.
Using a bizarre line of reasoning, Aagard tries to dismiss my discussion of the rule about knights being weak on the edge of the board: 'Watson writes that this rule has its main validity in the endgame, when there are fewer pieces left on the board ... He is right but this does only underline the validity of the rule. The fewer pieces there are on the board, the more importance individual pieces are accorded. Therefore if knights cannot be on the edge in the endgame, when only a few pieces remain, then this illustrates how little they contribute when out there.'
Let's get this straight: if a piece placement has a certain strength in a simple ending, then that somehow shows its fundamental contribution, so situated, during any part of the game? So as a rule, kings should advance and roam the board in the opening, and a rook should get behind a passed pawn in the middlegame? Essentially he's saying that endgame positions are somehow more significant than middlegames with respect to assessing pieces' fundamental strength - why on earth would that be? Here we have abstraction that limits understanding, although I'm sure that Aagard himself wouldn't dream of playing chess with such an artificial philosophy. This is not just a statistical issue; it's about how decisions are actually made. Sure, as I myself point out, knights are less effective on the edge of the board in a majority of cases. But are strong players less dependent than they used to be (or not dependent at all) upon a knights-on-the-rim rule? Or does the rule, if applied as it has been in the past, merely limit our ability to make the correct, creative decision? The increasing and very large number of cases where a knight is placed on the rim to good effect indicates to me that strong players are probably not too concerned about putting the knight on the rim on principle, but just look to see how good the move is in reality. That is, they use their judgment, calculations, and experience with similar positions to come to a decision. This is also confirmed by players' verbal annotations to their games.
Aagard even insists that knights on the rim are always bad, using a unique theoretical approach. He says for example: 'I have found that many people find it difficult to accept that a knight can be objectively poorly placed on the rim while the position remains acceptable, that the knight can contribute positively and still be a positional weakness.' Okay, that's one of several ways to model what's happening, but an unnecessarily burdensome one. As for applicability, isn't it a lot more practical (and better in terms of being able to play well) to simply deal with the actual position at hand, where the knight might be of decisive influence, rather than worry about (or take comfort in) the obscure abstraction that the knight is in some irrelevant sense 'bad'? In which way does Aagard believe that the best players are thinking about such positions? Would they find Aagard's rule useful to follow?
While on the subject of knights on the rim, I can't decide if Aagard is really missing the point or just feigning ignorance when, referring to my stance on this issue, he says 'I understand the argument', and plays devil's advocate: 'if a knight on the rim is dim, what about a position such ashe following.' The position he then provides is one in which a knight on h4 participates in a direct and winning kingside attack. He adds: 'I could have used other examples where there is no defence...' Since this kind of position doesn't support my argument, he concludes that counterexamples are irrelevant, i.e., 'no example will change [the fact that knights on the rim are dim]'. But Aagard knows very well that this isn't the argument at all. I have no example in my book that is remotely related to a mating attack, for obvious reasons. Likewise, one wouldn't argue that rooks are strong on open files by having them deliver a one-move back rank mate. This is knowingly misleading writing and does not reflect well upon the author's intent.
Similarly, Aagard tries to undercut my argument about the rules regarding bishops and knights in open and closed positions. He uses five lone-bishop-versus-knight endgames as illustrations, and concludes that 'bishops are better than knights in open positions' (presumably with 'all things being equal'). He appends the strawman argument that 'The rule is not "Bishops against knights win the game" - only mate has that kind of strength!', as if anyone would dispute that. But let's see in what sense I have, as he says, 'come to the wrong conclusion' in this case. First, my book is about opening and middlegame ideas, not about four-piece endings with only two kings, a knight and a bishop. Since my book doesn't deny that an open board is generally favorable for bishops in endgame positions, what point is Aagard making? Apparently, as before, that principles for all positions derive from endgames. We've discussed that. But I don't even say, and would never say, that knights are as good as bishops in most open positions even in the middlegame - quite the opposite! What I do assert, as before, is that modern players are much less inclined to use such a rule to direct their play. There’s another way to look at this: if a rule has too many exceptions, following it will severely limit your play (and creativity). Why not take concrete positions into account with as little prejudice as possible, use your experience (perhaps in similar positions), test some actual lines, and form your own judgment about the reality of whether a piece is good or bad in a particular situation? 'All things being equal' sounds nice, but to have any usefulness it must mean that the player is balancing an extremely large number of factors on the board, at least in the opening and middlegame. How would he do so? Does a strong (or any) player actually think to himself: 'Well that knight is pretty bad on the rim, but it attacks a square that if vacated - but I'm not sure how likely that is and when it would occur - would allow it to put pressure on a pawn; whereas I have a bad bishop, and other things being equal, that's bad, but it does protect a weakness. Now, does central control count as another "thing being equal" or does my queen being developed at an early stage cancel it out more or less, and what about my doubled pawns, his initiative...' and so forth? You could multiply this interior monologue by 10 or more for certain decisions in the middlegame. The whole idea is absurd. With some exceptions, the strong player knows (or makes his best judgment that) a piece or pawn is correctly placed via a less abstract process, as described before. He may consider increasing the harmony of his position, or concentrating his pieces in one area of the board for an attack, but those are hardly 'rules', and such considerations are outside of my critique anyway. The question in practice is whether anyone can decide by explicit means to what extent other factors in a typical middlegame are equal. Since the diverse considerations are interdependent and also time-dependent, one normally requires experience and analysis to get around the problem.
Aagard tries to avoid the issue: 'It is true that the rules of the masters from around the year 1900 are limited and need to be revisited, but discharged for a notion of rule independence is not the way to do it' [sic]. First, for the record, I also question the usefulness of some of the rules and generalizations that come from players of the 1910s and 1920s, and even from Nimzowitsch himself (e.g., 'attack the pawn chain at the base', and his view that the isolated queen's pawn was just bad). But more importantly, how does he propose to revisit those rules? The implication is that they can be qualified and/or limited. Okay, but if Aagard would give us an example of doing this, I think that one would see precisely why many rules are impractical to apply over the board by comparison with a direct approach. [I myself describe a few hypothetical 'micro-rules', but they don't seem to help much].
In addition, Aagard persistently oversimplifies my comments. When I talk about Kasparov's decisions in one particular, extremely dynamic, game (a King's Indian vs. Shirov, page 126), I describe them as being based upon concrete calculation and opening preparation, and intuition 'to some extent'. In the game, Kasparov's dark-square control and a large host of potential ideas (both attacking and positional) are matched against Shirov's extra pawn and the breakdown of Kasparov's centre. There are of course other considerations such as pawn structure and piece effectiveness. Aagard complains that 'Shirov is a great calculating player, far greater than Kasparov...' [one could certainly argue this characterization], so that it must be 'Kasparov's superior understanding' that is decisive. It is also the case, according to Aagard, that Kasparov calculates not 'better' than Shirov, but 'more precisely'. First of all, this is pretty murky: I'm not sure that being more precise might not also be better. But in any case, none of this contradicts my description at all!
Aagard also uses a strawman argument here, saying that Kasparov's wordless notes from Informant (to which I referred) '[don't] mean that he uses only calculation when he is sitting at the board.' As if I had said that. The issue, remember, is whether explicit rules, in particular classical ones, are being used in such a case, even in combination with ach other. Let's say, hypothetically, that I'm wrong about the degree of intuition required in this particular game, which Aagard believes shouldn't be qualified by my 'to some extent'. It still seems obvious that Kasparov is drawing heavily upon pattern recognition, especially when you look at his related King's Indian games. In fact, this exact opening variation was a Kasparov speciality, and the key position to assess and analyse was one directly stemming from theory that he created. So preparation was probably a huge factor in deciding whether his bishop and other factors would outweigh White's advantages. In any case, I can't imagine a rule-based decision here. In fact, even without recourse to rules, an assertion like 'the dark-squared bishop, threats by the knight, and queenside line-opening ideas will beat the extra pawn and queen infiltration' would be insufficient to assess this precise position, and risky to depend upon. The are simply too many other factors at work.
I should end by showing Aagard in a better light, so let's look at a fascinating example in which he makes a good point, but about which there is much more to say. The fragment, which I give on page 136, comes from Kasparov-Kamsky, Linares 1993:
1.e4 c5 2.Nf3 e6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 d6 6.Be3 a6 7.f3 Nbd7 8.g4 h6 9.Rg1! Qb6 10.a3 Ne5 11.Bf2! [jw: 'Moving this piece twice achieves the goal of driving away Black's queen; this takes priority over development.'] 11...Qc7 12.f4 Nc4?! 13.Bxc4 Qxc4 14.Qf3 e5? 15.Nf5 Bxf5 16.gxf5 d5 17.fxe5 Nxe4 18.Rg4!. [jw: 'With a clear advantage. In this example, it almost seemed as though White forgot that he was supposed to get his pieces out; and yet in the end, his development was superior.']
Aagard disagrees strenuously with my comments. Regarding move 11, he says that, 'It is important to note that White is still leading in development while Black has done nothing at all in this regard'. The second part is not true, obviously, but this is still a little better than what I said. His main gripe is with my second sentence (after the assessment). He says, accurately: 'In the game, White was always leading in development, and Black's failure to develop contributed fully to his defeat.' Of course, I never said that White was behind in development, just that he wasn't developing. And I wasn't just talking about breaking rules, as he asserts in his next paragraph, but discussing characteristics of modern play. Nevertheless, I agree that this sentence was badly written and misleading; indeed, I already had marked it for rewriting.
But with an open mind, Aagard might have noticed the main thing: this game is still a wonderful example of modern chess and how it is changing. This is true with respect to both development and prophylaxis. Let's take a closer look. Aagard says 'Out of the first eight moves, Black made six pawn moves. He followed this by moving his queen, moving a knight for a second time, then the knight yet again and then his queen! Of course some of these moves were forced by White.' [jw: ...Nc4, ...Qxc4, and to some extent ...h6]. This is an important thing to point out. In the second chapter of my own book [page 16], I say almost the same thing about another Sicilian Defence: '...and yet, Black can make a considerable number of pawn moves with only one or two pieces out after 1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6, for example, ...e5, ...h6 (to prevent Bg5), ...b5 might all follow shortly, making seven pawn moves before the development of other pieces...and often, one of the first pieces moved is the queen...' I also point out (page 17) a line in the Poisoned Pawn Sicilian in which Black's development sequence goes pawn, pawn, pawn, knight, pawn, pawn, queen, queen, queen, pawn, the same knight, queen! As we both say, an emphasis on structure over development is characteristic of the Sicilian Defence.
In view of this, it is very interesting to see how Kasparov deviates from the White's traditional development in the Sicilian (and this is the point I was trying to make, however badly, in my brief notes to the game). Given Black's many pawn moves described above, we expect White to have a lead in development in the Sicilian. By move 5 in the openings above, he has one more piece out and has the move. At such a point, we are used to seeing White bring his other pieces out quickly in order to maintain or increase that lead, with perhaps one more pawn move (often the f-pawn) or at most two. Here are some examples of what typically happens after Nc3 and Nf3xd4 are in: (a) in the Dragon, we see Be3, f3, Qd2, 0-0-0, and Bc4; (b) in the Scheveningen, Be2, Be3, 0-0, f4, often followed by things like Bf3, Qe1-g3, and Rd1; (c) in the Rauzer, Bg5, Qd2, 0-0-0, f4, Be2; (d) in the Sozin, Bc4, Be3, Qe2, 0-0-0, and maybe even Rg1 before even one more pawn move, which in this case would be g4.
But look at our game: Kasparov is more interested in prophylactic and space-gaining moves. He plays Be3, f3, g4, Rg1 (This is not even on an open file! But as his notes show, it discourages Black's counterplay), a3, Bf2, f4 (we're up to 4 pawn moves, and both the bishop and pawn just moved for the second time). Then, after a combination of exchanges and the normal Qf3, White has cleared the way for his rook's second (and very unconventional) move to g4! Can we really imagine players from 20 years ago placing the pieces and proceeding as Kasparov did, much less those from 40, 60, or 80 years ago? Even a simple g4-g5 idea was considered quite radical when Keres played it, and this is obviously something else again. And yet such play by White (in a somewhat less extreme form, to be sure) is not so unusual today.
In conclusion ('At last!' cries the crowd in relief), I hope that by examining Aagard's criticisms, I have shed some light on what my book does and doesn't mean. I'm sorry that my discussion has been so theoretical and possibly difficult to follow; I don't believe that my book is. If you want to read it and like most of us are short of time, I would suggest that your main focus of attention be upon specific examples. Page 267 has a summary of 29 themes of the book, most of them relating to modern piece and pawn play. Rather than invest a lot of time dealing with abstract arguments, one might want to start there and then browse through the book to examine positions with those themes in mind.
Next time, more books!