King's Indian Defence: How 9...a5 Turns the Bayonet Attack into a Strategic Problem

On the ninth move of a King's Indian Defence game, when White pushes the b2-b4 pawn — the move analysts call the Bayonet Attack — Black stands at a familiar crossroads. One path is to retreat the knight to h5, opening the position and forcing both sides into sharp variations that every serious player must prepare up to move thirty. The other path is to push the a7-a5 pawn, a modest-looking move that creates no immediate threat but quietly steers the whole game into different waters. For decades, 9...Nh5 has been the main road, the thing every professional must memorise. But when a recent training product put 9...a5 on the scales as a solution to the Bayonet Attack, the question stopped being which move is theoretically stronger and became which move forces your opponent to think harder in real time. I started paying attention to the Mar del Plata branch back when I was still competing. I remember sitting for hours in front of an old computer, retyping every variation of this system to understand why a seemingly slow move like 9...a5 could coexist with the violent attacks of 9...Nh5. A mislabelled game file early in my career was the most expensive lesson I ever learned: the eye always needs verification. That is exactly why, whenever I read advertising claims about a new 'solution', I force myself to go back to the primary data. The King's Indian Defence is the system in which Black temporarily concedes the centre, builds a pawn wall on g6-g7-f7, places the bishop on the long g7 diagonal, and then counterattacks that very centre with moves like ...e5 and ...c5. The Mar del Plata branch — named after the Argentine coastal city where it was first dissected — is the sharpest version of the system, with Black developing the knight to c6 and pushing e5, opening a mutual wing-attack struggle. There, White typically storms the queenside pawns while Black masses forces on the kingside. The Bayonet Attack is precisely White's 9.b4, gaining space and opening plans like b4-b5 that threaten to shatter Black's pawn structure. In opening classification, this branch falls roughly into ECO E97 to E99, though the exact number depends on the specific variation and the next moves by both sides. The main reply for decades has been 9...Nh5. The knight retreats to h5 to support ...f5 or ...Nf4, opens the position, and forces both sides to calculate precise tactical variations. This is the branch engines analyse to terrifying depth, where any deviation is punished immediately. In other words, 9...Nh5 is the memoriser's road — the path of the player ready to spend thousands of hours committing variations to memory. By contrast, 9...a5 is a secondary branch that is nonetheless entirely real. The move puts a direct question to White's b4 pawn and guides the game into a slower strategic struggle. The core point is this: the chances remain balanced, but the character of the game changes completely. Instead of recalling a chain of sharp variations, both sides are pushed into a position demanding positional judgement, pawn-structure understanding and patience. White may choose b4-b5 to keep gaining space, but doing so fixes the pawn structure and gives Black time to organise a kingside counterattack on a rhythm very different from anything seen in the 9...Nh5 branch. This is what that training product aims at. It does not promise a new trick but a change in the game's character. Instead of 9...Nh5 opening the position, forcing lines and making computer analysis important again, the product chooses 9...a5 leading to sound positions with very few concrete lines. To a club player without thousands of hours to memorise the Mar del Plata, that promise sounds very attractive. It targets a specific market segment: players who believe they think independently better than they memorise mechanically. But years of analysis taught me something: attractive claims usually hide holes. First, 9...a5 is not a new invention. It is a long-known branch. Any seriously prepared Bayonet player knows how to face it. Second, the claim that computers often struggle with these positions is pre-AlphaZero language. Today, neural-network engines handle long-term KID positions far better than the advertising line suggests. Such a claim may be a way of differentiating from free tools rather than a technical truth. The breathing rhythm of a set of pieces on the board is no different from the breathing rhythm of a team on the pitch. Italy at Euro 2026 did not merely press — they taught all of Europe how a collective breathes. In a KID game, Black's pieces must breathe in the same rhythm: pawns on g6 and g7 shielding, the bishop on g7 as the spine, knights and rooks coordinating to a set tempo. If a single piece falls out of rhythm, the whole structure collapses. This holds true for both 9...Nh5 and 9...a5 — neither move frees the player from understanding the position. The difference lies in when that rhythm is felt and whether mistakes are punished instantly or accumulate silently. So where does the real value of 9...a5 lie? I would say in its ability to steer psychology. When your opponent has prepared thoroughly for 9...Nh5 and its sharp variations, choosing 9...a5 can throw them off balance early. That is practical value, not theoretical value. And that is exactly what the training product really sells: not theoretical superiority, but the ability to surprise during a short window of the game. I have seen this in amateur tournaments in Da Nang. A young player prepared deeply for the main branch suddenly faces a secondary variation and loses twenty minutes just to steady himself. The empty pitch turned out to be the most honest mirror of modern football. Chess is the same — when the crowd noise is gone, when all that remains is the board and the clock, differences in thinking rhythm become unmistakable. In the middle of the COVID season, sitting in an empty stadium, I heard the breathing of the tactical system. There I learned that a secondary variation chosen at the right moment is stronger than a main line played mechanically. But that is precisely why I must be wary of how the product is packaged. A move does not create value by itself; value comes from the person who understands its role. The transfer market is like a chess endgame: whoever reads the role first breathes a sigh of relief. That applies equally to buying a player and buying an opening course. Sellers promise things about the nature of the product, but what really decides is how the buyer uses it. The more serious problem I found while reading the original material was the mismatch between title and body. The title spoke of a team round report in Samarkand, while the body discussed an opening DVD. Two unrelated texts spliced together. That is a red flag about source quality — not because the opening content is wrong, but because the reader cannot trust the care of the messenger. In analysis work I always remind myself that a game file mislabelled from the start leads to a whole faulty report. The eye always needs verification, and this time it needs more than usual. The counter-intuitive angle here is this: if you buy 9...a5 believing it frees you from memorisation, you are fooling yourself. This branch is still part of the Mar del Plata — one of the most calculating systems in chess. It does not free you from calculation; it merely shifts the emphasis from memorising variations to judging the position. But positional judgement in the KID demands comparable expertise, only in a different form. The execution blind spot is that players routinely underestimate the difficulty of strategic positions and assume they are easier than tactical ones. Reality is the reverse — a tactical error is punished at once, while a strategic error accumulates silently and only surfaces twenty moves later. But I must also be fair. If my assessment stopped at rebutting the advertising, I would miss the real value. For a club-level player or an international master who cannot compete with grandmasters at memorising theory, a sound secondary branch has clear practical value. It offers another option, an escape from the memorisation race they cannot win. And in a world where every main line has been analysed to the end by engines, holding a lesser-known branch is itself a form of advantage. One more thing to note: this product is very likely a compilation of previously published theory rather than original research. Calling it a solution is a marketing frame, not a theoretical claim. For a serious buyer, what is needed before spending money is information about the author — who wrote it, their level, their results in the KID — together with sample variations and engine evaluations. All three are absent from the material I read. An opening product with no clear author identity, no sample games and no engine evaluation is hard to treat as a trustworthy reference. The question left for opening-theory watchers is not whether 9...a5 is stronger than 9...Nh5, but whether a secondary branch can survive as a practical weapon in the engine era. The answer will come from real high-level games — where the frequency of the Bayonet Attack, the score of 9...a5 and White's response will decide within a single theory cycle. The system takes a team far, the individual takes them home — and in chess, theory takes you into the game, while the ability to read the position takes you to victory. I will keep following whether this modest move produces a real wave, or just a small ripple in the long history of the King's Indian Defence.