9...a5 in the King's Indian: When a Rare Move Is Sold as a Truth

Chengdu, a Saturday morning late last month. Board seven of the open rapid. The digital clock ticks, the tea on the organisers' table went cold long ago. Black puts a pawn on a5 on move nine. White — a young player, 2,150 Elo, who spent a full week memorising the main line of the Bayonet — leans down, hands on his thighs, and goes quiet for fourteen minutes.

Nothing loud happens in those fourteen minutes. No combination, no sacrifice, no hand slapping the table. Just a man rifling through what he has learned and failing to find it.

When White finally plays b5, the position shifts into a different kind of silence: no more forcing sequences, only hesitant moves. The game runs to move 47. White loses. But what I wrote in my notebook was not the result. I wrote: fourteen minutes.

A month later, a DVD advertising exactly that move appeared, promising to be the antidote to the Bayonet. I read its sales copy with the same eyes that had watched those fourteen minutes.

The King's Indian Defence, Mar del Plata Variation — a name taken from a 2026 tournament in Argentina — is one of the densest theoretical complexes chess has produced. Black concedes queenside space, locks the centre with a pawn chain, then throws everything at the kingside. The system's fingerprint sits in 7...Nc6 and 8...Ne7, after White has pushed d5.

From the early 1990s, White found a way to break that rhythm: the Bayonet, 9.b4. Instead of waiting to be counter-attacked, White opens the queenside first, forcing Black to answer structural questions before the kingside pieces are placed. Since then the theory has grown a new floor every year.

Black's main line is 9...Nh5. It is thick, it is correct, and it is cruel to the half-prepared: hundreds of branches, long forcing sequences, and the dry fact that at a high level memory is part of skill. Forget one move at move 18 and you lose at move 25 without understanding why.

9...a5 is different. It exists in practice, but at low frequency; it has never climbed to main-line status. One producer chose it as the backbone of an anti-Bayonet repertoire and called it the antidote. That sales copy is the subject of this article.

Consider what the document does not contain: no author name, no publication date, no product ID, no sample games, not a single line of engine output, no ACPL, no numerical evaluation. The move sequence even jumps from 7.0-0 straight to 9.b4, skipping 7...Nc6 and 8.d5 Ne7 — skipping precisely the two moves that make the Bayonet exist.

That omission is not a chess error. It is the trace of a text written to persuade rather than for someone who intends to sit at a board and execute the line. An executor needs to know where the road starts.

So what does 9...a5 do on the board? It attacks the b4 pawn — the thorn the Bayonet has just planted in the queenside. White must choose: push b5, fix the structure and create a target; or trade on b4, handing Black an open a-file and a rook with a road. Both choices are commitments. Structurally, the document's claim holds.

What 9...a5 actually buys is a change of character: from forcing sequences to manoeuvring, from reciting to playing. Selling that as an objective superiority inverts the frame. The real value of 9...a5 lies in reducing preparation cost, not in being stronger than 9...Nh5. That is a practical benefit, worth money to the right buyer, mislabelled.

The weakest sentence in the whole text is “computers don't understand these positions”. That is a legacy claim. Before 2026 it was partly true. Since Stockfish 12 brought NNUE in 2026 and Leela Chess Zero matured, closed positions are no longer an engine blind spot at the level the text implies. A closed position may not be scored higher by an engine, but it is still scored, and scored fast.

The real argument is far narrower: 9...a5 contains fewer forcing lines to memorise, so the human-versus-engine gap is a memory gap, not an understanding gap. White's team can still engine-check the position in fifteen minutes. “The engine doesn't understand” and “the engine hasn't checked” are different statements, and the second is false.

Then comes “objectively, the chances are equal”. No evaluation, no reference games, no database citation. In practice the Mar del Plata structure gives White durable space and a queenside majority plan; Black trades that for the initiative, central strikes and the choice of when to open the kingside. That is a high-variance balance, not a symmetric equality. The claim may be roughly true at engine depth, but it is stated as a fact while being an unverified belief.

One technical problem the document never touches is the central one: asymmetric payoff. A rare line pays only the side that has studied it, and only until the opponent catches up. The half-life of surprise is inversely proportional to how often the move appears in public. Every public game, every video, every analysis piece deducts from that premium. The document neither quantifies it nor tries.

The refutation front is absent too. White's most natural replies to 9...a5 — b5, Ba3, Nd2 — are never discussed. An honest opening document spends most of its space exactly there: on what the opponent will do to you. Here the space goes to selling the idea.

There is a further, very practical risk: time. A strategy-first line is hungrier on the clock than a memorised forcing line. In rapid and tiebreaks, where every minute is paid for, “think for yourself instead of memorising” is a more expensive product than it looks.

And there is an unmentioned price. 9...a5 spends a move on the queenside while Black's strategic clock on the kingside keeps running. If White holds the structure with b5, White has space, a clear target, and time to redeploy without immediate punishment. If White trades, Black gets the a-file, but the structure simplifies — and simplification cools the kingside attack that is Black's main asset in the Mar del Plata. No line in the document mentions that possibility.

So who should buy it? The buyer profile is legible even without a single named player. Below 2026 Elo, utility is medium-to-high: opponents rarely reach a prepared Bayonet, so strategic guidance beats memorised theory. Between 2026 and 2400, utility peaks: opponents do prepare main lines, and a rare move at this band produces the largest surprise dividend. Above 2400, utility drops to medium-low: opponents and their seconds already have the move in their files, the rarity advantage decays, and 9...Nh5 reasserts itself.

What the document does not say but reveals is that the product targets a very specific pain: the player who loses because the opponent prepared better, not the player being outplayed positionally. “Think for yourself instead of memorising” lands exactly on that pain. And 9...a5 was very likely chosen because it is underexplored in public databases — an economic decision about what sells as a surprise weapon, not a conclusion that it is objectively best.

A credible opening document looks different. It carries an author name and title, annotated sample games, database statistics with game counts, a dedicated section for the opponent's strongest replies, and guidance on time controls. All of that is missing here. For an instructional product that absence is not neutral: it makes style fit impossible to assess — the single most important variable in whether a repertoire works for a given buyer.

Based on my experience following games — thirty-two years commentating chess for VTC, with more than a few Mar del Plata games passing through my headphones — I have learned that people usually buy an opening line to solve a problem that does not live in the opening. In 2026, when I began digital commentary, I thought I was stepping onto a stage. It turned out I had only entered the corridor. In 2026, when every playing hall stood empty, I heard something I had never heard before: pieces set down on wood, breathing, and the sound of people hesitating. A rare line creates exactly that silence on the board — and silence, in chess, is always information.

Chess's collective memory carries two durable beliefs. First: a rare move is a hidden truth. Second: if the engine does not understand it, the engine cannot check it. Both are relics of an era that closed around 2026.

A chessboard never lies; we only lie to ourselves with applause. A rare line is not stronger for being rare. It differs only in having been questioned less. Its real value is that it shifts the game from memory to judgement — and judgement only pays dividends to those who already have enough of it. For everyone else it swaps one trap for another: from forgetting the line to misreading the position.

The phrase “very few concrete lines” in the sales copy is a confession disguised as a selling point. It says the product does not hand you answers; it hands you the obligation to answer yourself. For some buyers that is the gift. For most, it is a second invoice they have not seen yet.

From keyboard to board, the speed of words never matches the speed of a move. And failure in chess is rarely a wrong move; it is the moment we see the square clearly and still place the piece outside it.

The measure of an opening product, after forty-one years watching this industry, is how much thinking it hands back to the buyer. A DVD that teaches three lines expires in three months. A DVD that teaches how to think inside a structure can last ten years. 9...a5 is, in the end, a pretext for learning how to think — and the pretext is worth more than the move.

White's fourteen minutes at board seven that morning remain the clearest answer. The young player did not lose because he did not know the move. He lost because nobody had taught him what to do when memory runs out. Can a repertoire ever be sold as such a promise — or can it only be bought as a curriculum?