Bayonet 9.b4 and 9...a5: An 'Antidote' or Merely a Change of Risk Profile?

In March 2026, in Mar del Plata, an Argentine seaside town, Svetozar Gligorić sat down at the board and deliberately handed the entire centre to his opponent. He fell back, built a wall of pieces on his last three ranks, placed a bishop on g7 like an eye open in the sky, and waited. The games played at that tournament, by Gligorić and Miguel Najdorf, produced one of the sharpest variations of the King's Indian Defence, and people named it after the town: Mar del Plata.

Seventy years later, in my inbox in Saint Petersburg, that name appeared in the promotional file for an opening video course: “Svitlana's Smart Moves – King's Indian Attack at the Olympiad”. The description promised something very specific: after White's 9.b4, Black may choose 9...a5 instead of the fashionable 9...Nh5, turning the game into a more comfortable strategic struggle that depends less on memory.

I read it three times. On the first pass I saw a sound pedagogical idea. On the second I saw a large evidential hole. On the third I saw a headline that does not match the body.

At 65, after more than four decades in tournament halls from Hanoi to Saint Petersburg, I have learned one thing about promises in chess: they tend to be right about the plan and vague about the data. When the board empties of pieces, memory begins to keep score.

The King's Indian Defence is the classic hypermodern system: Black lets White build a big centre and then attacks it from a distance with ...Nf6, ...g6, ...Bg7, ...d6 and ...e5. In the Mar del Plata Variation the two sides divide the board cleanly. White masses pieces on the kingside, opens the f-file, pushes the g- and h-pawns. Black opens the queenside with ...c5 and ...b5. It is the most naked footrace in top-level chess, where one slow tempo can turn into a catastrophe.

The standard move order, up to move nine, runs: 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. The move 9.b4 is called the Bayonet. White pushes the pawn to b4 to gain space, to prevent Black from playing ...c5 too early, and to prepare a queenside expansion of his own. From the 1990s onward the Bayonet became one of White's main weapons against the Mar del Plata, classified in the ECO range E97 to E99.

Black has two major answers. The move 9...Nh5 is the modern, forcing direction: the knight goes to the rim to clear the way for ...f5, to trade pieces, to dilute the position. It is strong, but it demands an enormous body of theory kept constantly up to date, because every move can be refuted by an engine line. The move 9...a5 is the more classical direction: Black blocks the queenside, keeps the structure intact, and accepts a slower game.

Over the past thirty years, analytical tools have changed the way people learn openings. A forcing line such as 9...Nh5 is nobody's secret anymore; it sits in databases, in courses, in opponents' memories. The maintenance cost of such a body of theory grows exponentially, while a club player's time does not. Demand for systems built on understanding is therefore real, and it does not require any particular product to exist.

The promotional file exploits exactly that demand. It says: choose the path with fewer forcing lines, less to memorise, more plans. Pedagogically, I agree with the underlying idea.

But a product description has to survive the buyer's scrutiny, and here three gaps appear very quickly. The first is the move order. The description stops at 7.0-0 and jumps straight to 9.b4, skipping the two obligatory moves 7...Nc6 and 8.d5 Ne7. An experienced buyer will fill in the blank, but it is a sign that the text was written for people who already know, not for people who need to know.

The second gap is evidence. Not a single game is named. No year, no round, no rating, no engine evaluation, no ACPL figure, no database statistics. The word “Olympiad” appears exactly once, in the headline, and never returns. For an instructional product, that is a very thin evidential base for a promise of that size.

The third gap is the headline. “King's Indian Attack” is an entirely different opening: White's 1.Nf3, g3, Bg2, d3 system. The body describes the “King's Indian Defence”, with 1.d4 Nf6 2.c4 g6. The two have nothing in common beyond the words “King's Indian”. For a magazine, that is an editing error. For a product, it is a signal about care.

The credibility-transfer mechanism normally used in this product category is: trust this line because it appeared at the Olympiad. But the Olympiad exists only in the headline. No game, no year, no board, no player. A prudent buyer should close that gap with his own database before spending money.

The old bird understands that landing is also a way of flying.

The move 9...a5 is not an invention. It is one of the two pillars of the Bayonet, coexisting with 9...Nh5 for decades and present in every open database. Calling it an “antidote” sets a higher expectation than the text itself concedes. An antidote, strictly speaking, means neutralising White's ambition. The text promises only two smaller things: equal chances, and practical danger for White if White is unprepared.

The real value of the product lies in changing the risk profile: from being forced to calculate to being forced to understand. With 9...Nh5, one wrong move inside a forcing line costs material or the game within the first fifteen moves. With 9...a5, the mistake does not surface immediately; it accumulates, and by move thirty Black realises he has drifted into an inferior position with no counterplay available. Two kinds of error, two kinds of pressure.

What the text does not mention is the structural price. By choosing 9...a5, Black concedes queenside space, must decide the fate of the a-file, must handle the a5 pawn in the endgame, and must pick the right moment for ...c5 or ...c6. Those are strategic decisions with long-term consequences, and an honest course should state them rather than only promising “fewer forcing lines”.

The claim that “computers often do not know how to handle these positions” is the most dated part of the whole description. The era of engines that assessed closed positions within a few centipawns is over. Engines built on neural networks, from Stockfish to Leela Chess Zero, read closed Indian structures with high accuracy, enough to point out the best plan for both sides. The honest version of that claim is: your human opponent is unlikely to have memorised an engine-checked route to an advantage. That is a statement about the opponent, not about the computer.

The pair of claims “objectively, chances are equal” and “things might quickly become dangerous for White” is the sharpest and most misreadable point. They belong to two different categories: one is an objective assessment, the other a practical scoring probability. They are compatible only if the reader understands that the product sells practical advantage against an unprepared opponent, not objective advantage against a prepared one.

One selling point is wasted: 9...a5 is far easier to play under short time controls. With a few minutes on the clock, nobody can calculate a ten-move forcing line accurately, but anyone can follow a familiar queenside plan. The text never says this, although it is the clearest advantage of the system.

Based on my experience watching games at club events and city championships in Saint Petersburg, I notice a recurring pattern. White players rated 1800 to 2200 often push 9.b4 because they know it is “strong”, but stop analysing at move fifteen. Black players in the same band often choose 9...Nh5 because they know it is “modern”, then get dragged into a forcing line they have no time to maintain. Both sides buy openings on faith rather than on plans. A course that teaches plans has a genuine place in that gap.

The presenter's name appears in the headline, but her title, rating and competitive record are not stated. In this product category the presenter's authority is part of the quality, because anyone can look up the line, while the explanation cannot be looked up. What is scarce and being sold here is pedagogy, not theory. And a buyer cannot assess pedagogy from a promotional file with no sample game and no preview clip.

In content economics, video and DVD opening courses face double pressure: free databases supply the theory, free streaming supplies the explanation. Their survival value lies in curation and structure. My prediction is that within a few years this category will migrate toward a model of a repertoire system plus an update service, because opening theory decays faster than the life cycle of a disc.

The opening market is where people sell memory to buy time.

The biggest blind spot in the collective memory of the King's Indian Defence is the romantic memory. We remember it as the attacking weapon of players willing to concede the centre and then counterattack, from Svetozar Gligorić and Miguel Najdorf at Mar del Plata 2026, through a young Garry Kasparov, to Teimour Radjabov as its last defender at the top. That memory is beautiful, and it makes club players buy a system they believe to be an attacking tool.

But its real standing has changed. At the elite level the King's Indian's frequency has narrowed under pressure from engine-refined anti-Indian set-ups. At club level it holds firm, because those set-ups do not travel downward quickly. The elite decline is precisely what sustains the lower-tier market: amateurs buy the memory of a weapon, not a weapon that is currently winning.

The inverted risk sits right there. If you choose a low-theory system in order to avoid study, you will meet exactly the most dangerous kind of opponent: a Bayonet specialist who already keeps a file on 9...a5 and understands the queenside structure better than you do. In a position where both sides follow a plan, the side that understands the plan more deeply wins. The advantage of “less to remember” becomes the disadvantage of “less to understand” the moment the opponent stops being surprised.

The anti-computer argument in the product description is a tribal signal, not a technical argument. It resonates with the amateur's sense of being suffocated by engine theory, and its persuasive power is entirely independent of its accuracy. The worrying part is the practical consequence: a player who accepts “computers don't understand” as truth will gradually drop the habit of engine verification, which is the fastest route to losing analytical discipline.

What would change my mind? Three things. A named sample game, with a year and a round, in which 9...a5 is played to move thirty with good annotation. A database table showing the actual scoring rate of 9...a5 above 2400 over the past few years. And a preview clip showing that the presenter genuinely explains plans rather than reciting variations. With those three, the “antidote” promise becomes a testable claim. Without them, it remains a promise.

The day the clock's hand stops moving, I learned to move slower but deeper.

Choose openings by maintenance cost, not by nominal evaluation. That is the lesson I have drawn from watching hundreds of club games ruined not because the player chose the wrong system, but because he chose a system he had no time to feed. The move 9...a5 is a reasonable choice for a specific group of players, and it deserves to be presented as what it is: a mature strategic option, not a dose of medicine.

If a line is only worth something when the opponent is unprepared, is it a weapon or a lottery ticket? The buyer should answer that himself before opening his wallet, and should verify the first fifteen to twenty moves with his own engine and database. The rest of the course, if the presenter is good, will teach how to play those positions to move forty. That is the part worth paying for. When the board is at its quietest, the plan is the only thing still speaking.