The 9...a5 Answer to the Bayonet: A Solution Built on Understanding, Not Data

In a small coffee shop in Saigon, the player with the black pieces rests his finger on the b4 pawn, then pulls it back.

The board in front of him has already passed nine moves – 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 9.b4 – and White's pawn has just stepped forward like a declaration. He does not open a book. He looks toward the queenside, where an empty square waits for the move few people mention: 9...a5.

I sat two tables away, recording that moment – not for the move itself, but for the slowing down. A man spent nearly four minutes on a move that an engine returns in a thousandth of a second. Those four minutes are the subject of this piece.

The King's Indian Defence is the system in which Black deliberately concedes the centre, places the bishop on g7, and strikes back with ...e5. The names attached to it – David Bronstein, Garry Kasparov – all used it as a double-edged weapon: defensive and provocative at once.

The Mar del Plata variation takes its name from the 2026 tournament in Argentina, adding 7...Nc6 and turning the queenside into a battlefield before anyone has thought about the king. White answers with 8.d5 Ne7. Then 9.b4 appears, carrying the name Bayonet Attack: the rifle spike.

The name is no mere metaphor. White does not charge at the black king. White opens another front, pushing the pawn to b4 to break up Black's queenside pawn structure, and only turns toward the king once the defensive line has been torn. In the ECO opening codes, the Bayonet sits in the E97–E99 group – the group anyone who has played Black knows as the most hours-hungry territory on the board.

What makes the Bayonet uncomfortable is that it gives Black no clear target. White does not attack; White expands. And in chess, slow expansion is often harder to meet than a loud assault.

The orthodox answer is 9...Nh5. It has been analysed so densely that variations multiply, forced moves pile up, and memorisation becomes a second job. The author of the product I am examining admits as much himself: after 9...Nh5, computer analysis becomes important again.

That product – an instructional opening course on disc, the familiar distribution format of the European chess market – proposes a different route: 9...a5.

That is the entire news portion of the story. Everything else lies behind it.

Black does not contest the centre. Black strikes directly at the b4 pawn – a piece that has only just stepped forward and is already a target. The move is positional, not forcing: no combinations, no traps, only a proposal that the queenside will open in a different way, and that Black will seek counterplay there rather than in the centre.

The seller describes it with three phrases: sound positions, very few concrete lines, and a focus on strategy rather than tactics. He also concedes plainly: objectively, the chances are equal.

I read that sentence three times. In the language of opening authors, equal is a modest word. It does not mean Black is better. It means Black is not being squeezed.

The real value of 9...a5 is not that it is stronger than 9...Nh5. It is that it changes the kind of difficulty: from hard to calculate to hard to understand.

And this is where I have to be clear, having watched far too many opening products over five years: not a single metric comes with it. No ACPL. No win-draw rate. No game sample. Not one line of independent data proving that 9...a5 is genuinely equal. Every favourable adjective comes from the seller.

That does not make the move wrong. It only makes it unverified.

The target audience is stated plainly: players who prefer to think for themselves rather than memorise long variations. In practice, that is the 1800-to-2400 band – players with enough understanding to handle a positional position, but not enough hours to maintain a thick engine file.

The disc format says something too. It is a traditional distribution channel, shrinking in the face of subscription variation libraries, where learners pay monthly rather than buying once.

There is one practical advantage the seller does not need to exaggerate: White prepares less against 9...a5. The first fifteen minutes at the board belong to Black, simply because the opponent has never opened a file on it.

But one line in the introduction made me stop: computers do not know how to handle these positions.

I do not believe it. And notably, the author does not fully believe it either – a paragraph later he says that after 9...Nh5 the computer matters again. If engines evaluate the position after 9...Nh5 well, they evaluate the position after 9...a5 well. Both are King's Indian structures.

This is an old advertising line, and it is unnecessary. The move is good enough to stand without it.

The second risk is subtler. A low density of forced moves sounds like a gift: less calculation, fewer prepared traps. It also means a thin margin of safety. With no variation to follow, the player must decide every move himself. The memoriser breaks at exactly move eleven.

The third risk is cyclical. The Bayonet rises and falls with elite fashion. When it rises, 9...a5 has value. When it falls, the move sits quietly in a file. The real death of an opening variation is when we stop mentioning it.

And one more thing: this entire argument lives on calm. Fast chess does not reward calm.

I would still tell a Black player to learn 9...a5. Not because it is the answer, but because it teaches something 9...Nh5 does not: how to read a position when there is no variation left to read along.

An empty stand still has an echo – if we are quiet enough to hear it.

And I think of the player in the coffee shop. His four minutes were not wasted. He was doing exactly what a chess player must do: choose understanding before choosing memory.

A move can fade, but the way we understand it never grows old.