9...a5 and the Escape from the Machine: Revisiting the Bayonet Attack in the King's Indian Defense

The chess clock clicked once and fell silent. On the board, White's pawn had just advanced to b4. I sat in the third row of a club tournament in Hai Phong, a half-filled notebook in my hand, watching the man playing Black in front of me die slowly inside a position he did not understand. He was not weak. He had simply memorized a road without anyone teaching him to read the map.

I have seen that scene many times. A club player, decently strong, opens the book correctly, plays the right moves, and then suddenly stands before a position that appears on no page. He loses because of what he does not know he does not know. And that death usually begins with a move that opening analysts still argue about today: 9...a5.

I trust footage over testimony, because footage cannot lie. That is why I never begin an analysis with a feeling. I begin with the position, with the number, with a frame I have rewound again and again. The move 9...a5 in the King's Indian Defense is worth dissecting because it sits exactly on the boundary between opening theory and practical psychology — between what the machine knows and what a human understands.

Before the specific move, the context must be rebuilt. The King's Indian Defense begins with 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5. It is a hypermodern defense: Black deliberately concedes central space, accepts that White will build a large pawn block, and then counterattacks that very block. This idea runs against the instinct of most beginners — those taught that the center must be seized at all costs. But it is precisely that contrarian logic that gives it its appeal.

This is not a system for the weak. Bobby Fischer used the King's Indian Defense as a weapon in some of the most important games of his career. Garry Kasparov turned it into one of his principal weapons throughout his peak years, especially in his rivalry with Karpov. When two world champions who disagree about everything both choose one system, people must stop calling it a sideline.

Within the King's Indian, the Classical Variation leads to the Mar del Plata system, where Black plays 7...Nc6. This is a crucial fork. After 7...Nc6, White usually continues 8.d5, pushing the pawn through the center and opening a two-wing race. White attacks the queenside, Black attacks the kingside. This is the kind of position where every concept becomes relative: the center is no longer something to occupy but something to lock, and the result lies in how fast each wing opens.

Then comes the ninth move. White plays 9.b4 — the Bayonet Attack. The name sounds aggressive, but its nature is not a killer blow. 9.b4 is a strategic declaration: White advances the queenside pawn early to gain space, open the bishop's diagonal to b2, and, more importantly, disrupt Black's natural plan of playing ...c5 or ...e5-e4 and then locking the wing. It is a tempo jab, not a punch.

What is interesting is that 9.b4 is not new. It has existed for a long time, but in recent years it has become one of the most popular choices at the highest level, because it forces Black off the paths that have been memorized in detail. It is a recurring pattern in modern chess: a move that is not stronger objectively but stronger practically, because it pushes the opponent out of his comfort zone.

Against the Bayonet, Black's orthodox reply is 9...Nh5. This is a difficult, sharp move that is willing to open the position. The knight leaps to the rim, eyeing f4 and g3, preparing kingside attacking schemes. 9...Nh5 opens the door to concrete, forced lines that demand precise recall from both sides. It is the road of those ready to fight with theory, with memory, with raw calculation.

And here the alternative appears: 9...a5. Black pushes the queenside pawn one square, attacking White's newly advanced b4 pawn while gaining space and preparing ideas such as ...Be6, ...c5, or ...Nh5 depending on the position. It is a legitimate move that has appeared in top-level practice, but it has never occupied the central position that 9...Nh5 holds. It is a road few travel.

What caught my attention was not the move itself but the way it is sold to the public. In the source material I read, there is an opening product — an instructional DVD set — promoting 9...a5 as the antidote to the Bayonet. The central argument sounds seductive: 9...a5 leads to predominantly strategic positions, with few concrete lines, where the calculating machine often does not know how to handle things, while the human player who understands plans feels comfortable.

This is where I want to pause longest. Because it touches one of the biggest questions in modern chess: when is a move good not because of how strong it is, but because of how much it confuses the opponent?

I began my career with chess, not with the running track. In 2026, I entered the profession as a player and tournament organizer, then gradually moved into media. Those years standing inside the organizing committee taught me something I later carried into athletics: inside the playing hall, what decides the outcome is not the move that is theoretically correct, but the move that the opponent across the board fails to recognize in time.

Based on my experience following games across more than three decades, I have noticed a rule: the most irritating moves are never the objectively strongest. They are the moves that are neutral in evaluation but alien in structure. A move the machine scores at 0.00 can devastate a human who is used to familiar positions.

Let us talk about the structure that 9...a5 creates. After White plays 9.b4, Black replies 9...a5. The black pawn advances to attack the b4 pawn, forcing White to decide: push to b5 to keep the space, trade on a5, or leave it and accept the tension. Each choice leads in a different direction. The key point is that none of them is a long forced line. Both sides retain many options. That is exactly what the product's promoter wants to emphasize.

Technically, 9...a5 has its own logic. It attacks the b4 pawn — the pawn that is the foundation of White's entire queenside plan. When White must defend or advance this pawn, the tempo of White's plan slows. At the same time, pushing the a-pawn frees the a7 square for the rook, opening ideas such as ...Ra7, ...Ba6, or ...c5 supported from the a-file. These are long-term strategic plans, not immediate tactical blows.

And this is why a club player is drawn to it. If you play 9...Nh5, you must remember. You must recall every move in a line both sides have analyzed to the end, where a move in the wrong order can cost you material. But if you play 9...a5, you can win by understanding the plan rather than by memorizing it. That is a seductive promise to anyone who lacks the time to sit for hours at the board.

But I must say plainly what people usually avoid. The claim that "the machine does not understand these positions" is steadily losing its value. It was a true statement in the 2000s, when chess software was still weak at evaluating long-term strategic positions. But in the era of Stockfish and Leela, the machines have learned to understand these locked-center positions very well. They are no longer confused by positions humans call "hard to understand."

This is the point the seller does not want you to consider. As the machine grows stronger, the advantage of a "strategic, few-concrete-lines" variation no longer lies in avoiding the machine — it lies only in avoiding the opponent's human memory. And that is a smaller, shorter-lived, more easily neutralized advantage, once your opponent has studied the same road.

What is more telling is that the source material provides no data at all to support its claim. No engine match rate. No average centipawn loss. No win rate, no draw rate, no sample size, no named player who has tested it at a high level. It is entirely a verbal argument.

I trust footage over testimony. And in chess, "footage" is the database. If 9...a5 were truly better than 9...Nh5, we would see it across thousands of games at the 2600+ level. We would see Black's win rate no lower, we would see it appearing in major events. The absence of data is not proof of weakness, but neither is it proof of superiority. It is simply a gap.

At this point I must mention something unusual in the very material I analyzed. Its title speaks of a European schools chess championship, held in Tsaghkadzor, Armenia, with young champions and a cultural festival. But the entire content inside speaks of an opening product — a DVD set teaching the anti-Bayonet 9...a5. The two have nothing to do with each other.

I have met this kind of mismatch many times in my profession. A flashy headline attached to dry content. A young, glittering event used as a shell for a commercial product. As a documentary filmmaker, I always ask: what is real, what is the shell? Because when the headline and the content tell two different stories, the reader is not merely misled — the reader is robbed of the right to judge the true nature of the matter.

If the headline is real, then this is a schools event in Armenia — a story about a young generation, about a talent pipeline, about a small country with a formidable chess tradition. That would be an entirely different story, and in my view a far more worth-telling one. Armenia is among the countries with the highest density of grandmasters per capita in the world. A schools event there is not merely a competition; it is a link in a talent-producing machine.

But if the content is real, then this is merely an opening product, and an opening product is not news. It is advertising. And advertising has no obligation to provide data, no obligation to prove, no obligation to admit its limits.

It is this ambiguity that compels me to write. Because I grew up in a field — sports media — where people constantly confuse two things: what attracts attention and what has value. And readers, especially club chess players, often pay for that confusion with their own money.

Now let us talk about the real audience for 9...a5. It is not for grandmasters. Grandmasters already have analytical teams, databases, powerful engines, and they will find the best path for themselves. 9...a5 is for club players, for strong amateurs, for those who want to understand rather than memorize. That is a real audience, with real needs, and with an undeniable truth: most chess players in the world do not have the time to turn themselves into computers.

For that group, the value of 9...a5 is real — but not because it is stronger than 9...Nh5, rather because it is different. It gives them a position they can understand through logic rather than memory. And in practical play, forcing your opponent to think instead of following the book is sometimes worth more than a small objective edge.

But I want to warn of this: that advantage is temporary. When 9...a5 becomes popular, it will be analyzed to the end, and it will lose its surprise. That is the fate of every road that flees theory. You escape theory today, and tomorrow your escape route becomes the new theory. Modern chess does not let anyone escape forever.

This is where I recall a line I have carried through my whole career. I have hurdled my whole life, I have simply never been timed. I write about people whose achievements do not appear in the official results sheet. And I think of club players — those who play 9...a5 not because it is strongest, but because it is the only road they can travel without an entire team behind them.

There is another aspect I cannot ignore: the format of the product. An instructional opening DVD is an aging distribution model. In the era of subscription platforms like Chessable, where content is continuously updated after each new engine discovery, a static disc is a paradox. It freezes opening knowledge at a moment in time, and cannot be corrected when the machine finds something new.

I once made films during the pandemic, when every shooting schedule collapsed and I had to draft a written crisis protocol to cope. That experience taught me: when the tools change, people must change with them. The chess industry is shifting from discs to interactive platforms, from static to updateable content. A DVD set cannot do what a living database can.

So why do I still write about 9...a5? Because behind the product is a real question. The question of where the value of understanding lies in an era when the machine has surpassed humans in every position. If the machine knows everything, why do humans still play chess? Why do we still sit at the board for hours, trading the time of our lives for moves a chip can compute in a few milliseconds?

I believe the answer lies where the machine cannot reach: the moment when the opponent across the board must stop, look up, and admit he does not know what to do. That is not a tactical moment. It is a human moment. And every move in the world, whether 9...a5 or anything else, is ultimately only a tool for creating that moment.

I still remember the image in the editing room during the pandemic years, when I discovered that the spiked shoes of the women players on the wooden floor formed a rhythm. The stadium stood empty, and I turned the sound of spikes into the heartbeat of the film. Chess is the same. In a silent playing hall, the ticking of the clock and the sound of pieces on the board are the heartbeat. And 9...a5, in the end, is only a small sound within that symphony.

What I want to leave behind is not advice on choosing a move. I do not have enough data to say 9...a5 is better than 9...Nh5, and I will not pretend otherwise. What I want to leave is a way of looking: when someone sells you a road out of theory, ask what evidence they offer. When someone says the machine does not understand a position, remember that the machine learns very fast. And when you choose a move because it makes you feel safe, make sure that safety comes from understanding, not from memorization.

Every documentary is a running track: the audience sees the finish, I live at every starting line. In chess, the finish is the game's result, while the starting line is every opening move — where everything is still possible, where a small decision can shape the whole game. 9...a5 lives at that starting line. It does not promise a victory. It promises only a road. And sometimes, in chess as in life, a road few travel is all you need to begin.