The Bayonet Attack and 9...a5: The Safe Move Behind an Opening DVD Cover

A file sat in a folder labelled “Samarkand — Round 2, Open Board”. The headline said every strong team had won, though a few had come close to dropping points. I opened it expecting a team-event round report, the kind of document I still read to track the pulse of a tournament. The body was about an opening DVD, specifically a method of meeting the Bayonet Attack in the King’s Indian Defence with the move 9...a5.

Two unrelated texts, glued under one label. I sat still for a few minutes. Not because the content was hard, but because the label was wrong. In 2026, when I first worked as an analysis assistant for the SHB Da Nang coaching staff, I mis-recorded the position of a corner kick in the 63rd minute of the round-12 V-League match against Hanoi FC. The head coach read the report in front of the whole squad and stopped exactly on that line. I filed the wrong SHB Da Nang tape that year, and from then on I knew that football does not forgive carelessness. A wrong label does not ruin the match. It ruins the analyst, and after that it ruins the reader’s trust.

Chess is the same. A mislabelled file is contaminated data. In the rest of this piece I separate two things: the label, and the content. The content of that file turned out to be far more interesting than a round of team chess in Uzbekistan.

Context: a system that lives by conceding the centre

The King’s Indian Defence, KID for short, runs 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.Nf3 0-0 6.Be2 e5. Black lets White build a large pawn centre, then turns around and attacks that very centre. This is classical hypermodernism: you do not contest the centre by occupying it, you contest it by overloading it.

The sharpest branch of the system is the Mar del Plata Variation. The name comes from the tournament held in Mar del Plata, Argentina, in 2026, where Black’s ...Nc6 and ...e5 idea was pushed by leading players into a complete system. In classification terms, the branch sits in the ECO range E97 to E99.

In the Mar del Plata, after 7.0-0 Nc6, White pushes 8.d5 Ne7. The centre locks. White has space and a queenside pawn majority. Black has the bishop pair, a potential ...f5 break, and a kingside pawn majority. Both sides race down opposite wings. Whoever arrives first wins. It is the kind of game where every move has a price, and the price is usually paid in time.

To me, the Mar del Plata is a personality test rather than a memory test. Some young players I coached in Da Nang knew the first twenty moves of the branch by heart, then sat bewildered on move twenty-one because they did not understand why the f5 pawn had to go at that particular moment. They had learned the shape of the game without learning its rhythm. That is why I always tell students to read a variation by structure, never by sequence of moves.

The Bayonet: when White refuses to play the old script

For decades, White in the Mar del Plata played moves such as 9.Ne1, 9.Bd2 or 9.Nd2, preparing a two-wing race strictly by the book. Then, through the 1990s and 2000s, another wave appeared: White pushes 9.b4. The move has its own name, the Bayonet Attack.

The idea behind 9.b4 is concrete. White grabs queenside space before Black can organise ...f5. The b4 pawn prepares b5, drives the black knight off c6, and turns the queenside into a front that can open quickly. White no longer waits for Black to choose the moment of explosion on the kingside. White builds a second clock.

The Bayonet makes the Mar del Plata structure uncomfortable for Black because it strips away rhythmic comfort. In the older branches, Black knew roughly how many moves were available to complete ...f5 before the queenside collapsed. In the Bayonet, that window is compressed. Black must answer on the queenside immediately, while still keeping enough pieces for the kingside.

That is precisely why the Bayonet became one of White’s favoured KID variations in the computer era. It has both a clear positional basis and forcing short-term lines for White to memorise. For a professional, that is the ideal combination: a long-term plan payable in short-term precision.

Two answers: 9...Nh5 and 9...a5

Against 9.b4, Black has two broad directions.

The first is 9...Nh5. The knight jumps from f6 to h5, eyes f4, prepares ...f5 and puts direct pressure on White’s pawn structure. This is the main move, played most often, analysed most deeply, and it opens the game up. After 9...Nh5, play usually enters forcing lines: White pushes 10.Re1 or 10.c5, Black answers ...Nf4, pieces come off, and the result is often decided by details that only a database can fully map. This is the kind of opening where one bad move on move seventeen can end the game on move twenty-five.

The second is 9...a5. The a7 pawn advances one square and attacks the b4 pawn at once. This is a sideline, long in existence, not a new invention. But its character is completely different.

After 9...a5, White must choose. If White accepts 10.bxa5, Black recaptures with ...Rxa5, and the black rook lands on the a-file with an open file pointing into White’s rear. If White keeps the pawn and defends b4, the game closes into a slow two-wing race where both sides improve their pieces over many moves without a single concrete point of contact.

That choice is the entire commercial value of 9...a5. It converts the game from a sprint into a marathon. For Black, that is comfortable. For White, it is irritating, because White loses control of the tempo.

What 9...a5 actually achieves

I want to strip this part of promotional language, because it is where deception is easiest.

9...a5 does not solve the Bayonet. It postpones the Bayonet. More precisely: it forces White to define the pawn structure before White can complete the b5 and bxc6 plan. Black trades a move for a decision. In elite chess, a decision forced in the opening phase is worth about half a pawn in practical terms, sometimes more, if the opponent is unprepared.

Positionally, 9...a5 keeps the bishop pair for Black longer than the main lines do. Black’s kingside pawn structure stays intact: f7, g6 and h7 untouched, so ...f5 remains a real break rather than a weakened one. Black can play ...Nc7 or ...Nd7, returning the knight to the centre once the queenside is settled. It is the kind of game where Black endures the first twenty moves before starting to ask questions.

The price is equally clear. 9...a5 creates no immediate threat. White can use that time for 10.Ba3, holding b4 and clearing the a1 or b1 square for a rook. White can play 10.Nd2 followed by 11.b5, pushing the queenside another way. White can also accept the a-file line and trust the central space advantage. None of those branches loses for White automatically.

To be exact: 9...a5 is a legitimate, well-founded move that leads to playable equality. It is not an antidote. Anyone selling it as an antidote is selling something else: reassurance.

The engine premise is a decade out of date

This is where I want to linger longest, because it is the biggest crack in the product’s entire commercial argument.

The argument says that 9...Nh5 opens the position, forces concrete lines, and makes computer analysis important again, while 9...a5 leads to sound positions with very few concrete lines, positions computers often do not know how to handle.

That last sentence belongs to the pre-AlphaZero era.

In December 2026, DeepMind published the results of AlphaZero’s match against Stockfish. The way AlphaZero played positions with a locked centre, two-wing structures and long-term plans that were not immediately visible in the evaluation — that was a turning point in how the whole field understood machine analysis. From that point on, neural-network engines such as Leela and modern NNUE builds of Stockfish handle long-term positional positions far better than the older, pure-search generation of engines.

In other words: if an opening product today still sells the idea that “computers cannot understand this position”, that product is telling you it was designed in a world that no longer exists. That does not automatically make it useless. But it puts the foundational reasoning out of step with reality.

I remember the COVID season in Da Nang. The city was locked down and matches were played in empty stadiums. I had to analyse from a screen with limited camera angles, and I hated losing the wide view. Sitting in an empty ground in the middle of COVID, I heard the breathing of a tactical system — and the only way to hear it was to match GPS data from eleven players with three fixed cameras. I did that for the match against Ho Chi Minh City FC, found that young forward Nguyen Phi Hoang covered an average of 46 metres per minute inside the box but touched the ball only 21 times, and wrote a twelve-page report on how the team lost distance discipline without crowd noise. The coaching staff applied it immediately to reshape the pressing block.

The lesson from that is identical to the chess lesson: when the familiar tool disappears, what remains is structure. Not feeling. Not assertion. Structure.

The opening-theory market: an old, cold trading floor

The product in that file is almost certainly from the Fritztrainer line — the instructional DVD format developed and sold for decades by ChessBase, the chess software company founded in 2026 in Hamburg. The format follows a stable formula: a titled player presents a system, chapter by chapter, with database files so the buyer can replay it.

Such a product does not exist in a vacuum. It sits in a market where free opening data has become unbelievably good: open databases, community studies online, and strong engines running on an ordinary laptop.

The irony sits here. The product tells the buyer it is for players who like to think for themselves and dislike memorisation, while the industry’s most successful business model of the past decade was built on exactly that memorisation. Chessable, an opening-learning platform built on spaced repetition, turned memorising move sequences into a measured daily habit, was later acquired by a chess group, and ultimately ended up inside the largest chess platform in the world.

So the market paid for memorisation, not for independent thinking. That is data, not opinion. And it makes the promise of “very few concrete lines” a narrow sales advantage aimed at a very specific buyer: someone over thirty who has lost games to forgotten theory and is convincing themselves the problem is method rather than volume.

I was in that group. I bought similar products hoping to find a system with less to remember. After a few years I realised something simple: no system has less to remember. There are only systems where what must be remembered lives on a different floor — the floor of plans rather than the floor of move sequences. And the floor of plans is harder to learn, not easier.

Who actually buys an opening DVD

Here I have to be blunt, because it concerns one of my own habits.

An empty pitch turns out to be the most honest mirror of modern football. Without a crowd, without the roar, you hear the boots, the instructions, the breathing. Chess in a training room is the same. Strip away the spectacle of a game with spectators and what remains is one person sitting at a board, asking themselves what they are doing.

Someone buying an opening DVD usually is not buying theory. They are buying three other things.

The first is time. Rather than build a variation tree from scratch, they buy a pre-built tree with someone accountable for it. For a club player with a full-time job, that is a reasonable trade.

The second is permission. A system signed by a titled player is easier to trust than a system you invented. That is a psychological function, not a technical one. It releases the player from the burden of judgement in the opening phase.

The third is identity. And this is the most skilfully sold part. The line “for players who think for themselves, not memorisers” is an invitation to join a group of people. The buyer does not buy the move 9...a5. They buy an image of themselves. Opening products sell identity more effectively than they sell variations, which is why the line survives even though free data has long surpassed it in accuracy.

I do not consider that fraud. I consider it a market feature that needs to be recognised correctly, much as I read transfer reports. The transfer market is like a rook endgame: whoever reads the role first breathes easy. The opening market works the same way. Whoever reads a product’s role before buying does not waste money.

The contrarian angle: whose safety does this safe move protect

This is the part I consider most important, and it is not in the original file.

People usually read 9...a5 as a technical choice: Black wants a slow, low-theory, easy game. That is true, but only at the level of the board. At another level, 9...a5 is a career choice made by the teacher, not by the player.

I have watched this in football for eighteen years. The return of the back three is not football progress. It is how a coach protects his reputation when a back four gets torn apart: switch to three centre-backs and the goal conceded looks more like a system error than an individual one, and system errors are easier to argue before a board of directors. The shape changes to protect the person who chose the shape.

In chess, that mechanism works identically but more subtly. A coach who recommends 9...Nh5 to a student takes responsibility for a vast theoretical territory where every move can be checked by an engine and proven wrong within fifteen moves. A coach who recommends 9...a5 does not. If the student loses, the cause is “not understanding the plan”, a verdict that cannot be verified with data at the board. If the student wins, credit goes to the system.

Put differently: 9...a5 is a safe move for Black at the board, and an even safer move for the person selling it. Those two safeties overlap, and together they hide the real question: whether it is the best choice for the player.

I have no data to answer that question. The original file offers no games, no win-rate figures, no engine evaluations, no author name, no record for the author in this very branch. That is too large a gap to fill with argument.

I always check myself at this point: if my counter-intuitive conclusion is not backed by data, it is only an opinion wearing makeup. On 9...a5, I stop at saying it is a legitimate move, oversold, and serving two parties at once.

The breath of a collective, and the breath of an opening

Italy at Euro 2026 taught Europe the breathing rhythm of a collective, and that lesson applies to the chessboard in ways few consider. Roberto Mancini’s team held under sixty percent possession yet still produced twenty-five shots in the opening match against Turkey. I doubted that left-back Spinazzola’s constant advances would leave a hole, until the data showed seven successful crosses and five recoveries in the opponent’s half. The advance was not an individual risk; it was a function of the system: when one man goes forward, three others shift to compensate.

That structure is identical to the structure of a good opening. A move like White’s 9.b4 is only strong when 10.b5, 11.Ba3 and central control arrive on the right beat. A move like Black’s 9...a5 is only safe when ...f5, ...Nc7 and the retention of the bishop pair arrive on the right beat. Italian pressing is music that must be played together; one beat off and the whole symphony collapses. An opening is music too, except the audience never sees the conductor.

And here I return to my own tracking experience. Based on my experience following matches, on grass and on the board alike, I keep finding one repeating rule: the system that wins is not the one with the most brilliant moves, but the one with the most average moves. 9...a5 survives at club and international-master level because it keeps most moves at a safe average. It dies at the highest level, because there nobody pays for average safety.

What to verify in your next game

If you are Black and considering 9...a5, here is what I would check before putting it in my arsenal.

First check: does your White opponent play 10.bxa5, and if so, are you genuinely comfortable with a rook on an open a-file while the centre stays locked. That comfort must come from games you have played, not from a chapter in a DVD.

Second check: what is your plan on move twenty. Every opening ends, and the ending of 9...a5 is a position with very few points of contact. If you have no plan for improving pieces there, you have traded a sharp game for a blurry one.

Third check, and the hardest: find a real game by a strong player in this branch, not an illustrative game inside a teaching product. Illustrative games are chosen because they illustrate. Real games are chosen because they were played. Those two kinds of data differ in nature.

I spent a month after the SHB Da Nang mistake in 2026 reviewing footage from five rounds and building a notation system that divided the pitch into eight zones. In the end, my report helped the team neutralise seventy percent of the dangerous set-piece situations in the return leg. A tape wrong at the very start is the most expensive lesson: the eye always needs verification.

The same principle applies to every opening product. The label on the cover says nothing about the quality of the contents. Verification belongs to the reader, and a reader who accepts that is never led by a wrong label.

Your next game will answer the question no product can answer: are you playing a system, or are you playing someone else’s memory of a system.