When the King's Indian Refuses to Memorize: 9...a5 and the War Against the Bayonet

On the corner of my desk in Da Nang there is a yellowed slip of paper on which I have written four characters: 9...a5. I stuck it there last autumn, after an evening spent replaying a game between two players whose names I can no longer recall. What I remember is not the result. What I remember is the feeling that the board had suddenly gone quiet.

There were no rapid exchanges. No combination forced the opponent to respond within three moves. There was only a wing quietly stepping onto the a-file, as if the player of the black pieces had just declared that he would not play the script someone else had written. Months later, when an instructional opening DVD reached my hands promising an antidote to the Bayonet Attack, I realised that slip of paper was no longer about a single game. It was a small manifesto, carefully packaged and sold to players who do not wish to become memory machines.

The context needs rebuilding for anyone who has never sat before this position. The King's Indian Defence is one of the strangest systems in modern chess. Black voluntarily concedes the centre from the very first moves, only to reconquer it later with ...d6 and ...e5. It is a bet on time: enduring cramped space for twenty moves, in exchange for a stormy middlegame.

The Mar del Plata variation is the sharpest version of that system. Once both sides have castled, Black plays ...Nc6 and ...e5, White pushes d5 to lock the centre, and the board splits into two clear battlegrounds: White attacks the queenside, Black attacks the kingside. For decades this has been one of the most thoroughly analysed fronts in elite chess, with lines dissected out to move thirty.

Then came the Bayonet Attack. White's 9.b4, pushing the b-pawn forward before completing development, became the fashionable answer to the Mar del Plata problem. White's idea is pragmatic: instead of waiting, seize space at once and prepare a queenside offensive before Black can close the position. For the player of the black pieces, the Bayonet is a systematic headache.

The orthodox reply, treated as the main line in most theory books, is 9...Nh5. The knight leaps to the rim, preparing ...f5 and ...Nf4, opening the kingside counterattack. After 9...Nh5 the position becomes extremely sharp, with lines analysed to move twenty-five or thirty. That is the territory of memorisers.

And here the story becomes interesting. The product I am discussing does not sell 9...Nh5. It sells 9...a5.

The essential point is this: 9...a5 is not a new invention. It is a philosophical choice. The move does not try to prove that Black is better; it merely tries to drag the game out of the room White has already fitted with air conditioning.

Look at the technical substance. After 9.b4, White has just pushed the b-pawn forward on the attack. Black's 9...a5 asks the question in reverse: and you, what will you do with this pawn structure? If White takes with bxa5, Black answers ...Rxa5 and the a-file opens, unbalancing the queenside in Black's favour. If White maintains the tension, Black continues with ...Ne8 and ...f5, postponing the decision and keeping the position flexible. The move 9...a5 does not solve the underlying problem; it only changes the question.

The difference from 9...Nh5 lies in tempo. 9...Nh5 forces the game to happen: the knight leaps, White must calculate at once, the position opens, and the computer becomes the referee again. It demands precise memorisation move by move. Whereas 9...a5 slows the game down: Black steps back in time but reclaims the right to choose the script. It demands understanding of structures rather than recall of sequences.

If someone asks me which move is objectively better, I will answer honestly: the data gives no clear verdict. Modern theory still broadly places 9...Nh5 in the primary position, while 9...a5 is regarded as a respected sideline. But the question of which move is better is not the question this product wants to answer. It wants to answer a different one: which move is easier to play for someone who does not have three hours every evening to sit before a database?

Here I must acknowledge something as an observer. The real value of an opening does not lie in how strong it is when both sides play perfectly, but in how long it forces the opponent to think when both sides are human.

I have spent more than forty years watching chess, from the days standing behind the stage of a small tournament in India to the evenings commentating live in front of a screen in Vietnam. Over all that time I have learned something no table ever teaches. Most games at club level, and even at master level, are not decided by the best move, but by the move that disorients the opponent.

That is why 9...a5 has appeal. It is not a hammer. It is a small thorn. In chess the thorn is often more effective than the hammer, because the hammer puts the opponent on guard, while the thorn leaves them bewildered.

Looking at the pawn structure after 9...a5, I see a point few people mention: Black does not rush to close the kingside. While 9...Nh5 accelerates the tempo, 9...a5 keeps the kingside undefined. That means White does not know where to attack. They have prepared for a race, but Black refuses to run. In a race where one side refuses to start, the one best prepared for the race is precisely the one left most confused.

I remember an evening at Da Nang Radio, commentating a game in which Black had chosen 9...a5. The young people in the studio, accustomed to opening a database for reference, seemed lost. They could not find the familiar sequence. They asked me: where is this position heading? I answered that it was heading nowhere at all, and that this was exactly the problem. Modern chess has become a sport of pre-drawn roads. When a player decides to fold up the map, he forces the other to travel without a road.

At this point I must say what I believe matters most.

The marketing of this product rests on an old argument: that computers often do not know how to handle complex strategic positions. I have read this sentence many times, and each time it feels dated. Before the age of neural-network engines, the argument had some basis. Today the leading engines handle the long-term strategic positions of the King's Indian far better than the sentence implies. Selling an anti-computer idea with an argument the computer long ago overtook is selling expired goods.

But here is the counter-intuitive part. The obsolescence of the argument does not make the product useless. It merely returns the marketing to its proper place: marketing.

The real problem with 9...a5 is not whether computers understand it. The real problem is that databases already contain it, and therefore a well-prepared opponent may have the answer ready. Any antidote printed and sold to the public ceases to be a secret on the very day it is published. A respectable sideline such as 9...a5 does not defeat the player who has studied it. It only defeats the player who has never met it.

So what is this product really selling? It is not selling a win. It is selling time. The time your opponent needs to realise what you are doing. And that time, at club level, is often worth more than a few dozen Elo points.

There is another paradox I want to name. The product is pitched at players who like to think for themselves rather than memorise. Yet to play 9...a5 decently, you still have to learn a sequence, only a shorter and more flexible one. This is an identity-based sales pitch, not a technical one. It targets the image a chess player wants to have of himself. I understand that, because I too once wanted to be that person.

In the opening-publishing world this is no new story. Hundreds of opening products appear each year, most of them consisting of sequences already available in free databases, repackaged with a layer of storytelling. Buyers do not pay for the moves. They pay for structure, for guidance, for the feeling that someone has gone ahead and cleared the path. Because chess, at some deep level, is not a game of information. It is a game of the fear of being left alone in the position.

In Vietnam, where I have chosen to live and work, this story has its own shade. Our young players grow up with computers. They can look up every line of the King's Indian in seconds. But that does not make them invincible. It only makes them dependent on the opponent playing exactly what has been recorded. When someone produces 9...a5, a move the computer rates as sound but that rarely appears in elite games, their familiar map is suddenly missing a few pieces.

That is the real value of this product, and also its real limit. It is useful against the unprepared. It is harmless against the prepared. And it tells no one that it is only useful for a certain window, because once it becomes popular it ceases to be useful.

There is one thing I always tell the young people who come to ask me about buying opening products. Before you pay, demand three things: the author's name and their record in this system, a few sample lines with engine evaluations, and one real game in which the author himself played that line. If a product cannot supply all three, you are buying a belief, not a weapon. Belief has value, but you should know what you are paying for.

I wonder what will happen to 9...a5 in the coming years. If it appears in classical games at the 2700 level, it will have a chance to become a permanent part of theory, and the time it buys will shorten. If it only circulates in club events, it will live forever as a small weapon, a thorn people learn and then forget. Neither scenario is a tragedy. Both are chess adjusting itself.

What I want to leave behind is not a verdict on a particular DVD. What I want to leave behind is a remark on how we read opening advertising: the product does not lie when it promises to make the game easier. It simply does not say that the ease has an expiry date.

One morning I peeled the 9...a5 slip of paper off my desk and pasted it into a notebook. I did not do so because the move had lost its value. I did so because I realised it was no longer mine alone. Once a move is packaged and sold, it belongs to everyone who bought it, and soon it will belong to everyone who must play against it.

I thought I had been pushed away from the board the day I left the newsroom. It turned out the board took me deeper into myself. Silence on the board is not emptiness. It is a move not yet written down, and sometimes it is the move that makes an opponent think longest of all.