Two Faces of Move Nine: Is 9...a5 Really an Antidote to the Bayonet Attack?

Late last week, past eleven at night in my Mymensingh room, the position that kept returning to my laptop screen was nothing new. King's Indian Defence, Classical branch, Bayonet Attack. White has played 9.b4—pushing the queenside pawn to tighten the grip on the centre. Black sits and thinks. Two moves ahead: 9...Nh5, or 9...a5?

That single moment hides the whole question. A DVD blurb has landed in my hands claiming 9...a5 is the Bayonet's "antidote." The language is familiar: "very few concrete lines," strategy as your friend, and the most eye-catching sentence of all—"computers often do not know how to handle these complex strategic positions." My hand stopped at that line. In the engine era, someone is writing that computers cannot understand?

Context: what the Bayonet is, and why it matters

Let me clear the ground. In the Classical King's Indian, White's 9.b4 has become fashionable over two decades. The philosophy is hypermodern—Black hands over the centre first, then attacks it. In the Bayonet, White squeezes space on the queenside and tries to crack Black's c5–b7 structure.

Black historically has two roads. One, 9...Nh5—the Mar del Plata tradition, a forcing move where Black takes f4 and breaks the structure, dragging White into a response. Two, 9...a5—less forcing but more patient; it shuts the queenside and puts a question mark behind White's b4 pawn.

The DVD says 9...Nh5 is "computer-analysis-important" but heavy for humans. So it recommends 9...a5, where there are "very few concrete lines"—less memorisation, more understanding. On theory alone, the argument sounds decent. But there is no engine evaluation, no database figures, no win-draw-loss ratio. Only language. And language can sell an opening; it cannot prove one.

A market note matters here. ChessBase-style instructional products arrive in the same format year after year: a strong player picks a line, then it is marketed as a "secret weapon" or an "antidote." The product need not be bad—but marketing language and evidential language are not the same. When readers blur the two, the confusion begins.

Core: a wager on structure, or on memory?

Now let us step inside the position.

After 9.b4, White's plan is clear—gain queenside space, threaten c4–c5 or b4–b5, and add kingside pressure with Ne1–Nd3–f4. The Bayonet's core idea is exactly this: press on both wings and force Black into piecemeal responses.

9...a5 is the first thorn. The move puts a finger on the base of White's b4 pawn. If White plays bxa5, Black recaptures Rxa5, seizes the a-file and erases White's extra-pawn structure on the queenside. If White plays Ba3, Black keeps that same file with the axb4–Bxb4 exchange. In other words, 9...a5 forces White to decide—and the longer the decision, the better for Black.

This is the DVD's selling point. It argues that 9...Nh5 leads into forced lines where computer memory wins, while 9...a5 leads into a battle of structure where human understanding pays off.

The theory is attractive on its own. But where does the claim come from? Which engine, which version, which evaluation? How many games in which database? What percentage of wins at which time control?

Zero. The narrative stands on zero data.

This is where an old habit returns. I once tagged 640 league games, sorting them only by pawn structure and endgame type. The sample grew until the room itself disappeared—but the lesson stayed: a claim is only true when a sample sits behind it. With 9...a5 my first job is the same: pull the games, run the engine, test whether "objectively equal" holds.

"Objectively chances are equal" is another DVD line, thrown out with no evaluation. Yet modern engines—Stockfish or Leela—evaluate closed King's Indian structures with remarkable accuracy. Leela's neural network can grasp the strategic subtleties of closed positions that were largely impossible a decade ago. So "computers cannot handle it" was true in 2026, not in 2026.

That does not make 9...a5 a bad move. The opposite. It is honest, tested, a weapon of many strong players. But it is not new. It is a known strategic sideline sitting in the database. The word "antidote" is commercial positioning, not technical discovery.

There is a structural truth too. If Black chooses 9...a5, the game's character changes—from an open knife-fight to a closed war of patience. That shift suits Black in Classical time controls, where there is time to think. But in Rapid or Blitz, where seconds are counted, "few concrete lines" means "less memorised support." If you misremember the strategic idea, the structure will not save you.

There is also a reciprocal risk the DVD never mentions. If 9...a5 becomes your antidote, opponents will know it. Bayonet players will prepare against this branch in advance. The more popular a weapon, the faster an antidote to the antidote appears.

This game is even harder for club players with day jobs. Someone who works by day and studies two hours at night finds "few lines, more understanding" genuinely tempting. But if the understanding is wrong, the cost is large. That is the frame that matters: an instructional product can give a foundation, not the last word.

Contrarian: the gap between headline and body

Now to the biggest flaw of all.

The headline speaks of the European Schools Chess Championship—from Tsaghkadzor to the Grand Final, a cultural celebration. But the entire body is an advertisement for a King's Indian DVD. No player, no result, no rating from the championship—nothing.

That gap between headline and body is the real story. It is a small breach of trust with the reader. You came for schools-championship news; you got opening-theory marketing.

At the Batumi Olympiad I built federation dossiers, filing each opposing team's board-one habits. There I learned that a dossier must be read like a weather report—first separate the true page from the mere preamble. This article never managed that, because two pages of unrelated material were stitched together.

Still, there is a human side that data-lovers skip. Who is this "computers don't understand" marketing really for? For those who did not grow up in the engine era, who want to believe human understanding still wins. It is a nostalgia market—largely for adult improvers who want understanding over memorisation.

That is not wrong. But marketing and information are not the same thing. And when headline and body pull in opposite directions, the reader learns the piece did not come from a news desk—it came from a product feed. That is the real blind spot: editorial, not technical.

Takeaway: the next move's question

So next time?

With 9...a5 my plan is simple. Database first, belief second. How many rated games exist in this line, what percentage the engine calls equal, and how much Classical differs from Rapid. The "objectively equal" claim will then answer itself.

As for the European Schools Chess Championship—if its real results exist anywhere, they must be found separately on the European Chess Union calendar. Because in a piece where headline and body walk in opposite directions, there is more marketing than news.

And finally the question that must be asked of every opening product. Will 9...a5 save your game—or only sell a DVD? The answer sits on the board, not in the advert.