The Bayonet Attack and 9...a5: When Opening Theory Is Sold as an Antidote

In the fourth game of a Thursday evening at a small chess club on Nguyen Chi Thanh Street in Da Nang, a fifteen-year-old student moved a black pawn from a7 to a5 and then sat still. No crowd, no ticking clock worth mentioning. Only wood on wood and the hum of a ceiling fan. That boy had just played the ninth move of a King's Indian Defense, after his opponent pushed the pawn to b4 — what the chess world calls the Bayonet Attack. I sat across from him, notebook open, and realised I was watching something smaller than a game: a purchasing decision. The boy did not invent 9...a5. He had read it somewhere, in a lecture file or a video or a theory DVD someone sold him with the promise that this was the antidote to a fashionable opening.

That is why I am writing this. Not to teach you the King's Indian. But to describe the distance between a move and a marketing line, between a real game and a headline that does not match the text beneath it.

Context: an opening built to concede, then reclaim

The King's Indian Defense begins 1.d4 Nf6 2.c4 g6, with Black placing the bishop on g7. Its core idea is hypermodern: Black does not occupy the centre but lets White build a large pawn centre, then attacks that centre with the g7 bishop, the minor pieces and timely pawn strikes. It is one of the oldest and most stubborn openings in modern chess, surviving nearly a century as both a weapon for attackers and a nightmare for those who prefer certainty.

The branch in question is the Mar del Plata, named after the Argentine coastal city where it made its mark in a major mid-century tournament. Its skeleton: 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5 7.0-0 Nc6 8.d5 Ne7. Black pushes to e5, White locks the centre with d5, and the position splits into two fronts. Black's knight retreats to e7 to prepare f7-f5 on the kingside. White owns a broad queenside for pawn pushes and file openings. Two trains, two parallel tracks, and the winner is usually whoever arrives half a tempo earlier. In the ECO classification this region sits around E97 to E99.

I have followed this branch since I was a chess player in Da Nang, before I moved into chess media. In my early years I recorded games by hand, from videotapes or printed bulletins. And the earliest lesson has survived more than a decade: the Mar del Plata is not an opening to memorise. It is an opening to feel. You can remember twenty moves and still lose on move twenty-one because you never understood why your knight belongs on e7 rather than f6.

The Bayonet Attack: a blade aimed at the queenside

White's ninth move in the Mar del Plata has many options. One of the more modern is the Bayonet Attack, 9.b4. White gains queenside space, prepares b5 to evict the knight from c6, and — more importantly — opens a path for a queenside offensive that can arrive faster than Black's kingside attack. The name says everything. This is not a defensive move. It is a thrust.

In practice, the Bayonet forces Black to confront an uncomfortable question: do you enter a game your opponent has prepared better than you, or do you steer the game onto a plane where memory is no longer decisive? Most professionals choose the second. And for years, the second meant 9...Nh5.

9...Nh5 is the main move. It lifts the knight from f6 to h5, hitting the e2 bishop and the g3 pawn if White pushes, while clearing the way for f7-f5. It is highly forcing: it opens the position, generates concrete lines, and turns the game into a calculation race. For professionals that is a strength — they have analytical teams, databases, engines. For amateurs it is a weakness — they have only memory and time.

That gap is the market. And that market has a very specific product.

9...a5: a secondary branch sold as a main road

9...a5 has been known for a long time. Black pushes the a-pawn to a5, attacks White's b4 pawn, and deliberately declines to open the position. Instead of a calculation race, Black offers a strategic game: slower, more closed, where complex pawn structures must be handled, where knights and bishops must find squares in cramped space, where plans matter more than lines.

Here is the core point: 9...a5 is not a novelty. It is a known, secondary branch compared with 9...Nh5. Anyone selling it as an invention is overselling. Anyone selling it as a practical choice is entirely reasonable — its value lies not in novelty but in pulling the game out of the opponent's memory.

The sales argument, as far as I can reconstruct it, has four propositions. First, it targets the Bayonet Attack, a fashionable line. Second, it promises a strategic battle with very few concrete lines. Third, it claims the chances are balanced. Fourth, it rests on a human-versus-machine contrast: 9...Nh5 opens the position and forces lines, making computer analysis important again, whereas 9...a5 leads to sound positions with very few concrete lines, and computers often do not know how to handle such complex strategic positions.

The fourth proposition is the troubling one. Not because it is entirely false, but because it is dated.

Human versus machine: a story that ended before AlphaZero

I remember 2026, when AlphaZero appeared and defeated Stockfish in a match the whole chess world stopped to watch. I was in Da Nang, newly working as a commentator, and I remember the chill of watching that machine play moves that looked meaningless for twenty moves and then strangle its opponent by move sixty. That was the moment every claim of the form "computers do not understand strategic positions" became an artefact.

Before the neural-network engine generation, the claim held. Search-based engines with crude evaluation struggled with positions whose value lies in long-term structure — exactly the positions the King's Indian Mar del Plata produces. After AlphaZero and its successors, that gap nearly vanished. Modern engines handle closed, structure-heavy, low-forcing positions better than most humans ever will.

In other words, "computers cannot handle it" is marketing language from a bygone era. It is emotionally compelling — who would not want a corner of the board where humans still rule — but it no longer stands technically.

This is where I open my second notebook, the one for feeling. I understand why people want to believe it. When you sit opposite an opponent who has memorised thirty moves by machine, when the game feels like a contest between two hard drives rather than two minds, you crave shelter. The DVD seller did not create that craving. They simply stood where the craving passes.

The missing data is the loudest thing in the room

A serious opening product should offer four kinds of data. First, engine match rate — whether the recommended move coincides with the engine's top choice at reasonable depth. Second, move-quality metrics, such as average centipawn loss for the side using the system. Third, actual win, draw and loss rates in a database, broken down by rating band. Fourth, a list of reference games, ideally at grandmaster level.

This product supplies none of them. No engine evaluation. No sample games. No database statistics. No time-control or practical-game data. All it offers is an assertion: the chances are balanced, and it works.

In my trade, an assertion without data can still be true. But it cannot be verified, and something that cannot be verified should not be bought at the price of something that can.

I say this as someone who once paid for a mistake in front of a microphone. In 2026, as an intern in Da Nang, I was assigned to commentate an U21 match between Vietnam and Timor-Leste at Hoa Xuan stadium. On minute twenty-three I mispronounced the opposing number nine's name three times in a row. Listeners in a central Vietnam group said I talked like I was reading a newspaper. I was ashamed but I did not quit. I spent a month rewatching the tournament tapes, transcribing every player name, and by the final I knew nearly two hundred names by heart.

Mistakes in front of a microphone taught me that perfection is only the draft of sincerity. They also taught me something drier: a mispronounced name can be corrected; a number that does not exist has nothing to correct. When an opening product presents a secondary branch as an antidote without a single figure, the buyer is paying for belief, not for a conclusion.

Samarkand: a headline pointing one way, a body pointing another

This is the part I want to dwell on, because it concerns chess only indirectly. It concerns how we receive information.

The document I analysed carries a headline about the second round of a team event in Samarkand, suggesting that despite close calls all the top teams won. That is a tournament report. Yet the entire body discusses a commercial DVD promoting 9...a5 against the Bayonet Attack. No tournament in the body. No teams. No rounds. No results.

Two unrelated texts, spliced into one package.

For a writer, this is the most frightening kind of error, because it is not technical but systemic. A headline is what readers see first; it decides whether they click. If a headline promises a tournament report and the body delivers an opening advert, the reader has been misled at the doorway. And if someone like me cites the document without checking, I become the next link in the same chain.

A microphone does not make a storyteller. It is the silence of the stands that signs every line.

I think about Samarkand differently. Samarkand is a city in Uzbekistan, a country with a strong chess tradition, where team events are held regularly and carry prestige. If a team event really took place there with a second round in which every top team won, it deserves to be told with its own data: pairings, board results, dates, format. And if an opening DVD really launched, it deserves to be assessed with its own data: variations, engine evaluations, sample games.

What does not deserve to exist is a Samarkand headline sitting on top of a text about 9...a5.

For years I have kept one habit before writing anything: opening three tabs in parallel — one for the source, one for cross-checking data, one for opposing views. That habit grew out of this exact fear: that I might tell a beautiful story out of pieces that never belonged together.

Two notebooks and a pawn on a5

Back to the club in Da Nang. The boy is still there, hand on the a5 pawn. I ask where he read the move. He names a lecture file, cannot recall the author, remembers only the promise: learn one move, escape one opening.

I do not blame him. I have been in that position, with photocopied books instead of files. The problem is not seeking a shortcut. The problem is a shortcut wearing the wrong label.

The truth about 9...a5, stripped bare, is this. It is a legitimate branch. It leads to sound positions. It reduces the memory load compared with 9...Nh5. It suits players who prefer thinking to memorising. But it does not remove calculation. The King's Indian Mar del Plata remains objectively sharp, and no line removes the demand for calculation. A well-prepared Bayonet player can still find play. Its real value most likely lies at club and international-master level, not at the very top.

That goes in my data notebook. In my feeling notebook goes this: a fifteen-year-old in Da Nang believes he has found an escape, and that belief, even resting on an inflated promise, is a beautiful thing. It should not be crushed. It should only be placed correctly.

When the stands are empty, I first hear the pieces speak for thousands of people.

I learned that in 2026, when the pandemic forced V.League matches behind closed doors. I was a new employee, assigned to commentate Viettel against Hanoi FC, a 0-0 draw. With no roar, my voice flattened and the commentary died emotionally. My editor said something I still quote verbatim: you said everything correctly, but no one could hear your heart. I doubted my ability and nearly left the profession. After weeks of brooding, I interviewed fifty supporters over video calls, asking which stadium sound they remembered most.

From that I built a library of remembered sound. Applause. Sighs. Chants sung with hands and feet. In chess I learned to hear something else: the silence before a difficult move, a silence thicker than any noise.

What is actually being sold

Seen as a market, this product belongs to the most stable and least glamorous revenue stream in chess: opening-theory publishing. It is the market of answers to popular openings. It exists because of a very human need — the need not to be surprised.

Its real competitor is not another DVD. It is the free database and the free engine. Anyone with a computer can filter a database by structure and find thousands of games. So why buy? Because raw data does not tell stories. A database gives you a thousand games and no plan. An authored product gives you a narrative, an order of priority, a way to remember. That is its genuine value, and the reason it can coexist with free tools.

But precisely because the value lies in narrative, the pressure to exaggerate also lies in narrative. When you cannot win with variations, you win with framing. And the most effective frame in this industry is "antidote".

I have seen the same mechanism elsewhere, in sports business, where representation contracts stop athletes from speaking their minds and where performative correctness replaces personality. The mechanism is identical. When you cannot sell truth, you sell posture.

Risk for the buyer, risk for the writer

If I had to rank the risks of buying a product like this, financial risk would not be first. The sums are usually small. The biggest risk is expectation: the buyer believes they will escape memorisation and discovers they are still playing a sharp opening that demands calculation.

The second is obsolescence. A secondary branch like 9...a5 can be neutralised once White analyses it. In a heavily mined branch like the Bayonet, this happens faster than you expect. A move that surprises for three months can be refuted in three weeks.

The third is psychological: overconfidence. When a product tells you that if you know what to do, things quickly become dangerous for your opponent, you enter the game with a false sense of safety. In chess, false safety is the most expensive feeling there is.

Sport does not promise victory; it promises only a heartbeat willing to continue.

For a writer, the risk sits elsewhere. The risk is that I tell this story so well that readers forget the original problem was never whether 9...a5 is good or bad, but that a Samarkand headline is sitting on an opening advert. If I let the story overwhelm verification, I betray the very principle I live by.

Sports culture is a graveyard of names, existing so their stories can live one more innings. That graveyard does not need another name buried in the wrong plot.

What I know, and what I leave open

I know the King's Indian Defense is real, old and played at every level. I know the Mar del Plata is among its sharpest branches, its structure running through 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5 7.0-0 Nc6 8.d5 Ne7, around ECO E97 to E99. I know the Bayonet Attack is 9.b4, that 9...Nh5 is Black's main reply, and that 9...a5 is a known secondary branch. I know a product promoting a secondary branch as an antidote, without engine evaluations, sample games or database statistics, is selling belief more than conclusions. And I know a document whose headline concerns a team event in Samarkand while its body concerns an opening DVD has an information-integrity problem, however good its technical content.

What I do not know: who the author is, what their results in the King's Indian are, whether 9...a5 has appeared in recent elite classical games, and whether Bayonet frequency is rising or falling. These questions need data I do not have.

I leave them open. An honest writer does not close doors for which he holds no key.

Da Nang, that evening, and a heartbeat willing to continue

After the club game ended — the boy won, with a crude but correct plan — I asked how 9...a5 felt. He said: I felt I could breathe.

It was the best answer I heard that week. And it reminded me why products like this exist, however much they exaggerate. They do not sell moves. They sell room to breathe.

Our job as writers is not to deny that room. It is to teach players which room is real and which is a door opening onto another corridor, no less cramped.

The boy will meet better-prepared opponents. He will lose again. He will seek other antidotes in other openings, and someone will sell them to him with new promises. What I can leave him, as a storyteller, is a habit: whenever someone says this is how you escape memorisation, ask for the numbers. If there are no numbers, ask for the author's name. If there is no name, ask what you are reading, and whether its headline describes what it actually contains.

This is not advice about chess. It is advice about living in a world where everything is sold with a story, even the driest thing imaginable — a branch of an opening on move nine.

That night I went home, reopened my data notebook and wrote three lines. First: the King's Indian Mar del Plata is alive, but as a surprise weapon rather than a mainstay, because it concedes space and invites sharp, well-theorised play. Second: the Bayonet Attack is the reason this market exists, because it is the fashionable, irritating line, and every irritating line breeds an antidote market. Third: the most suspicious element of the whole document is not 9...a5 but a Samarkand headline standing above a DVD blurb.

Then I opened the second notebook and wrote one line: the boy won, and the boy breathed.

I keep those lines far apart on the page. I do not merge them. One is verified fact, the other is personal feeling, and my job is to let them stand side by side without letting one blur the other.

That is all I can promise a reader. Not that I will always be right. Only that I will always show you which part I verified, which part I inferred, and which part I am still asking about.

As for 9...a5, let it stay there on the board, waiting for the next person to sit down and ask it a different question.