9...a5 in the King's Indian Defence: a repertoire that sells surprise, not advantage

Hanoi's night falls faster than I expect. On the wooden board by the window I rebuild a position I have looked at hundreds of times: 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. Black stands before two doors. One is 9...Nh5, the knight jump every opening book prints in bold. The other is 9...a5, the move a promotional guide is now calling an antidote to the Bayonet Attack — and the move I have never seen sold through a finished product. That guide carries a literary title: "A night unlike any other." It sounds as if the author is writing poetry rather than selling an opening system. But titles are not small matters. This one evokes a military image: a bayonet raid in the dark. That is exactly how 9.b4 got named — the Bayonet Attack. At sixteen I opened a tactical blog with a piece on Vietnam's U20 shape and collected more than forty comments of the "what does a girl know" variety. I cried, then told myself I had to answer with data. Years later the habit survives: when a sharply marketed opening product appears, my first move is to mute the marketing and set the position up on the board. People say girls know nothing about tactics. So I write for them to read. WHERE THE BOARD ACTUALLY SITS The King's Indian Defence is one of the strangest systems still alive at the top level. Black surrenders the centre entirely, accepts being squeezed for the first fifteen moves, and pays it back with a storm on the kingside. The basic pawn structure — Black pawns on d6, e5, g6 — produces a closed centre and wide-open flanks. That is why it has survived every trend: from Garry Kasparov through Judit Polgar to Teimour Radjabov and Vladimir Kramnik, the system has never disappeared. The Mar del Plata Variation is its heavyweight branch. The name comes from the tournament held in Mar del Plata, Argentina, in 2026, where the world's leading players first popularised a defensive structure with ...Nc6 followed by ...Ne7 and a kingside pawn attack with ...f5. Since then thousands of games at every level have built a theoretical corpus so thick that students now use software to memorise it rather than reading books. In the main line, after 9.b4, Black usually plays 9...Nh5, aiming for ...f5, ...Nf6, opening the kingside with two pawns and driving a wedge through the centre. The price is absolute precision: one inaccuracy and White breaks through first. This is not a branch for the inspired improviser. The product in front of me takes another road. It proposes 9...a5 — instead of striking the kingside, Black strikes the very prop of the Bayonet Attack: the b4 pawn. Three selling points are emphasised: the position is strategic rather than memorised, there are very few concrete lines, and it suits players who think for themselves. WHAT 9...a5 ACTUALLY DOES Look at the geometry. After 9...a5, Black opens the a-file and hits b4 head-on. White must decide: push bxa5, or hold the tension by defending b4 and accepting continued pressure. Each choice leads to a different structure, and here is the key point — none of them is a theoretical win. If White trades, Black gets an open a-file, a White pawn leaves the board, and b4 stops being a spearhead. If White holds, Black prepares ...c5 or ...c6 to hit the White centre from the other side. The beauty of this structure is that it demands no forced blow. The whole plan is to slow the game down, pulling the fight away from lines both players know by heart. For a club player that is a real advantage, not an imagined one. The biggest edge on a chessboard is not the strongest move but the move that makes the opponent think more than you do. 9...a5 strikes precisely there. But honesty about technical standing is required. 9...a5 is not an invention. It is a legitimate minority branch that has existed in theory for a long time, neglected because the main line 9...Nh5 is more attractive to calculate. In other words, this is repositioning, not discovery. That distinction matters because it changes the buyer's expectation: you are not buying an advantage, you are buying preparation asymmetry. The author, in a fairly candid passage, admits that objectively the chances remain equal. That is the sentence I underlined hardest. It shows the product does not promise to win from the opening — it promises you will enter the middlegame in a position your opponent has prepared less. FOUR THINGS I CHECKED First, sophistication. Against 9...Nh5, the move 9...a5 drops one tier in forcing power. That is classification, not criticism. The less forcing a branch is, the more it requires the player to understand the position — which is exactly what the guide concedes when it says it suits independent thinkers. Second, practical stability. For a well-prepared Black, the "very few concrete lines" argument sounds plausible. But I think that claim is overstated. A repertoire is never one branch. If you commit to 9...a5, you still must prepare for White leaving the Bayonet with 9.Ne1, 9.Nd2 or 9.Bg5. Those are three doors, three different pawn systems, and no guide can fold them into one chapter and stay honest. Third, product scope. The document never states whether it covers only the 9.b4 branch or the whole King's Indian after 7...Nc6. That is no small detail. If it is one branch, the buyer must cover the rest — and the rest is far bigger than one branch. If it is the whole system, the value jumps. For me, this ambiguity is the guide's largest gap. Fourth, verifiable data. Nowhere in what I read is there a named player, a reference game, a database figure, an engine evaluation, or a publication date. For an opening product, that is a serious shortfall. An opening branch earns trust when it arrives with evidence: how many games, at what level, with what results. Without that, the buyer is purchasing an argument, not a proof. I am not saying the author is wrong. I am saying the guide's credibility sits at medium by professional standards and cannot rise without data. THE BIGGEST BLIND SPOT IS THE SENTENCE ABOUT COMPUTERS The guide contains one notable claim: computers often do not know how to handle complex strategic positions of this kind. That was once true. It is no longer true in the way it is written. The turning point came in December 2026, when AlphaZero showed that a reinforcement-learning system could assess closed, manoeuvring, non-forcing positions far beyond classical engines. Then came the neural-network wave, and Stockfish's move to an NNUE architecture from 2026, which changed the picture entirely. Positions that older engines judged superficially for lack of a clear tactical blow are now scored with stable, verifiable numbers. This does not refute 9...a5. It means the selling point "computers don't understand" has lost most of its weight. If you choose this branch hoping a machine-prepared opponent will miss something, you may be betting on an outdated belief. Tactics never lie; only those who read them do. What survives for 9...a5 is something else: surprise in live play. At club level and in many open events, opponents prepare the main line and rarely look deep into sidelines. There, the value of 9...a5 lies not on the board but on the clock. You save thinking time; your opponent does not. A second, professional blind spot: the title "A night unlike any other" clearly puns on the night bayonet raid. That is a marketing headline, not a technical description. It is not false, but readers should know whether they are reading advertising or analysis. Most opening buyers cannot separate the two, which is why products like this sell. There is one more thing I always check: who has played the branch. No names in the guide means no empirical evidence. In Vietnam, many young players approach the King's Indian through foreign material with nobody to cross-check against. Le Quang Liem, Nguyen Ngoc Truong Son and the younger generation such as Nguyen Anh Khoi have all had to build their own systems for lack of Vietnamese-language resources. For them, a guide like this is useful — but only if it states clearly what it covers. HOW TO READ AN OPENING PRODUCT I have a habit of muting the sound when reviewing a game, so that I see only movement. With openings I do the same: mute the advertising, keep the moves, and ask what problem each move solves. Applied to 9...a5, I find a coherent plan: slow down, open a file, shift the fight from the arena of calculation to the arena of structure. Seen that way, the product holds up logically. Where it does not hold up is the commitment. It promises a position requiring less memorisation but offers no evidence about how many moves must be remembered. It promises difference but provides no reference game for the reader to judge. It promises strategy but never says how much background the buyer needs before starting. And here is what I want to say to women readers who have been told they do not understand tactics. Coping with a glossy product has nothing to do with how many variations you have memorised. It has to do with asking the right question. You do not need the whole Mar del Plata to ask: how far does this branch reach, who has played it, are there games. Those are the questions of a serious reader, not a beginner. In a transfer window — chess's equivalent being the season of buying guides and personal coaches — noise always exceeds signal. Sellers speak in adjectives. Buyers should listen for nouns: game counts, player names, publication dates, scope. Without those nouns, every promise of "strategy" is just a pretty title. Sometimes, to understand a position, you must stand in silence longer than others can bear. I sat still with this 9.b4 position for several nights. Every night looked the same, until I realised the most interesting thing was not 9...a5 itself, but the way it is being sold. WHAT MUST BE VERIFIED IN THE NEXT GAME I will try 9...a5 in a real game at a long time control and record three things: how long I take to decide after White's reply, whether I face any forced calculation, and whether I genuinely save time compared with 9...Nh5. That is the only way to turn a marketing claim into personal data. If the results match the logic of the branch — a slower opponent, a more comfortable me — then 9...a5 deserves a place in the repertoire of anyone who wants to play the King's Indian without memorising an entire library. If the results show I still must remember just as much, then this product sold me a better name rather than a better method. The only way to know is to set the pieces on the board. No advertisement can replace a real game.