The Bayonet Attack and the 9...a5 Promise: An Antidote Without Data

On Tuesday night I opened the database and typed four characters into the search field: 9...a5. The result was not empty. That is the detail the product description never mentions. The move has sat in archived games for a long time, surfacing in scattered events across many levels, never quite disappearing and never quite becoming main line. Analysts call branches like this durable sidelines — old enough that nobody is forced to learn them, effective enough that somebody still pulls them out. The interesting part is not the move. It is how the move is being sold. Theoretical background The King's Indian Defence, Mar del Plata system, is one of the densest theoretical complexes still alive at the top level. Black lets White take the centre, then counterattacks with ...Nc6, ...e5 and a kingside push. Both sides race on opposite wings: White advances queenside pawns, Black advances kingside pawns. Typical games can be decided by a single tempo around move thirty. Inside that system, the Bayonet Attack, 9.b4, is the most uncomfortable answer White can produce. The branch is associated with Vladimir Kramnik in the late 1990s and early 2000s, and it has since become a main choice for White players who want queenside space before Black organises the counterattack. Black's orthodox reply after 9.b4 is 9...Nh5. It is a forcing move: it opens the position, drives both sides into concrete variations, and pushes the game toward what an engine can evaluate clearly. Learning 9...Nh5 means memorising, and memorising means constant updating. The product on offer proposes a different turn: 9...a5. Not a novelty, not an invention, but a known branch repackaged as a strategic answer. The central argument is tidy: instead of entering the forcing lines after 9...Nh5, Black shifts the game into a less theoretical fight, where both sides play on plans rather than recall. It is an appealing argument. That is exactly when I started checking. The analysis One thing should be stated plainly: this product does not claim to invent a move. It claims to supply a way of playing. The distance between those two things matters more than it appears. If it were a novelty, the standard of proof would be engines and databases. If it is a way of playing, the standard of proof is practical results at the level the buyer actually plays. For the second case, a buyer needs three kinds of data: Black's win rate in the 9...a5 branch across rating bands, sample games with analysis, and how closely the recommended moves match top engine choices. In the description I read, none of those three existed. Not one sample game. Not one rate. Not one engine evaluation. No named author with a record in this specific branch. That detail is not small. With an opening-theory product, the author's credibility is the product. Buyers are not paying for dozens of variations; they are paying for someone's filtering — for the fact that a specific person read hundreds of games and worked out which branch is worth learning. When the author is anonymous, that mechanism disappears. An engine evaluation can lie, but five consecutive failed pawn breaks cannot. The product's second argument sets two kinds of player against each other: the memoriser and the independent thinker. The 9...Nh5 branch is assigned to the first group, the 9...a5 branch to the second. This is a familiar identity-marketing technique: it does not sell a move, it sells a self-image. I have no ethical objection to that. I only note that the Mar del Plata is, by nature, a sharp system. It allows White a large central and queenside space advantage, and in return Black must generate concrete kingside pressure before White converts space into material. No branch inside that system removes the requirement to calculate. Some branches reduce the memory load. None turns it into a purely strategic game. The promise of "very few concrete lines" therefore has to be read as a relative promise, not an absolute one. For a club player, halving the theoretical load is a real gain. Against a well-prepared White opponent, the 9...a5 branch is still inside the crosshairs. On practical value, I would place this product's usefulness between club level and the international master boundary. That is the group with enough study time but not enough time to refresh every branch monthly. Above that level, where every branch has already been dissected, a known sideline creates no surprise. The contrarian angle One line in the description made me stop longer than the rest: the suggestion that computers often do not know how to handle these complex strategic positions. That line is dated. It belongs to the era before neural-network engines transformed positional evaluation. Today, King's Indian positions with closed structures, separated wings and long-term plans are precisely the kind of position engines handle better than humans, and by some margin. Using that argument to sell a product means using an expired premise. No tactic is ever old; only the way we read a game goes stale. There is also a problem at the metadata layer. The document's original headline refers to a team round in Samarkand, while the entire body discusses an opening product. Two unrelated texts were spliced together. For a reader, that is a source-quality signal: if the shell and the contents do not match, everything inside has to be re-verified from scratch. I once learned a similar lesson at a different stadium. That V-League summer taught me one thing: a formation only looks beautiful when the opponent agrees to stand still. Chess works the same way. A plan only looks beautiful when the other side agrees to play into it. Closing If you are considering buying an opening product, ask for what every serious opening product can supply: an author's name with a record in the branch being sold, a few sample games with engine evaluations, and a chapter explaining when the branch fails. Without those three, a buyer is paying for belief. 9...a5 may well be a good choice. But between "a good choice" and "an antidote" there is a gap, and that gap is normally filled with data rather than promises. If this branch is that effective, why has it never appeared consistently at the top events?