Move Nine in the Mar del Plata: 9...a5 and the Evidence-Free Claim of Opening Products

Pune, first floor of a coaching class off Fergusson College Road. A February evening, the ceiling fan slow, and on the board the position that has stood on the fastest road of the King's Indian for three decades — 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 Bayonet Attack. Across the board a fourteen-year-old with a pencil in his hand, and on the digital clock 1:29:31 remaining — ninety minutes for forty moves, thirty-second increment from move one.

He plays 9...a5.

The coach's eyebrow goes up, then down. The question is the one every real chess room asks: where did this come from?

The answer did not come from the room. The answer came from an advertisement.

Last month a promotional text for an opening product landed in front of me — a Black repertoire against the King's Indian Mar del Plata, in which 9...a5 is sold as the "antidote" to the Bayonet. Nine information points. Not one named game, not one named player, no author's name, no publisher, no price, no date. No engine match rate, no ACPL, no database statistic.

And yet it carried the sentence that drives the entire pitch: computers often do not know how to handle the arising complex strategic positions.

Whether that sentence is true is a separate question, and a braver one. First, look at what the sentence is selling.

Context: the Mar del Plata, the Bayonet, and the politics of move nine

The Mar del Plata is built as a race between two wings. White's centre is locked — a pawn on d5, a pawn on e4 — and everything White owns is aimed queenside: b4-b5, then c4-c5, then pieces down the b-file. Everything Black owns is aimed kingside: f5, then f4, then g5-g4, and if the door opens, a direct assault on the king. Attack on both flanks, speed on both flanks, and a closed centre nobody trades off.

In that structure 9.b4, the Bayonet, is White's bluntest instrument. The pawn lands on b4 and claims the queenside, threatening b5 and c5 immediately. This is no fashion. It is a long-established main line, played repeatedly at elite level, and every King's Indian player has to walk through it.

Black's established reply is 9...Nh5. The knight goes to h5, f4 follows, and the game becomes a direct tactical collision. Its defining quality is forcing density — each move nearly dictates the next, which is exactly why the branch is easy to verify with an engine. Open the machine, look four moves deep, and you know where you stand.

Now 9...a5. A quiet pawn move on the queenside. It does not attack anything directly. It simply says: the b4 clamp your entire plan rests on is where I am putting my finger. If b5 comes, a5 is already there. If c5 comes, there is one more point of tension. And most importantly, the game leaves the forcing tunnels where the engine's first choice and the human's first choice are the same move.

The product calls this an antidote, and attaches four claims: very few concrete lines; strategy over memorisation; aimed at players who like to think for themselves; and, objectively, chances are equal — but if Black knows what to do, things can quickly become dangerous for White.

I stopped at that fourth sentence. It is the sentence every opening advertisement contains, except this time it has been written down in plain sight.

Core: five things the pitch tells you, three it hides

First, and this is the most honest thing in it: the recommendation is not a refutation, it is an equalising tool. "Objectively chances are equal" is unusually candid for promotional copy. It means something precise. 9...a5 is not the weapon with which you obtain an advantage. It is the weapon with which you move the game to a place where you feel more comfortable. For the player who needs a win in the last round, the tool's utility is structurally limited.

Second, and this carries the entire marketing load: the "computers don't understand" claim is out of date. When the sentence was coined, engines genuinely struggled with closed positions; they ran on hand-tuned evaluation functions and miscounted depth. NNUE and the neural era changed that. Stockfish adopted NNUE in 2026, and closed King's Indian structures are now evaluated far more reliably than in the 2000s.

There is a narrow, defensible steelman, and I accept it. In closed positions engine evaluations are less stable and less humanly actionable. The number hovers near zero and arrives without a plan you can put on the board; the engine's first choice is frequently a move a human ruins three moves later. That is a statement about usability, not about engine ignorance. The advertisement fuses the two, and the fusion is the product.

Third, and here the product stands itself where it cannot be disproved: a low-forcing line can be neither tested before purchase nor refuted after it. If a product says there is a knight sacrifice on move ten and the engine shows 0.00, the buyer checks it in ten minutes and returns it if wrong. If a product says there are no concrete lines, only plans, the buyer has no instrument at all. And when something goes wrong later, the author has a reply waiting: you did not understand the plan.

That is clever architecture. It protects the author from criticism. It also damages the product's credibility most severely with the buyers who take chess seriously and keep an engine open.

Fourth, technically the 9...a5 idea is not absurd. White's 9.b4 is a space-grabbing clamp, and White's entire claim rests on that clamp. Black's queenside pawn move contests the claim at its root and pulls the game out of the forcing tunnels where the machine already knows everything. It is a legitimate "avoid the machine, play the human" design.

But the price has to be written down. Black is accepting a permanently smaller share of the position's dynamic assets in exchange for strategic clarity. Black's traditional stake in the Mar del Plata is the speed of the kingside attack — f5, f4, g5, g4, pieces in front of the king. Playing 9...a5 does not abandon that stake, but it slows it down. A player who wants to win by attacking will, in this line, attack his way to equality.

Fifth, a small piece of word theft: calling the Bayonet "fashionable". 9.b4 is a decades-old main line, not a trend. Calling it a fashion manufactures the sense of a new problem requiring a new solution — which happens to be this product. It is not a large lie. It is the small framing every market performs, which is why it needs to be named.

Who the buyer is: the 2026–2200 band, and that is the real target

No player is named here, so player-level analysis is impossible. What is possible is buyer-archetype analysis, and it is just as solid.

The text assumes the buyer already owns a Mar del Plata repertoire — that he plays 7...Nc6, plays the King's Indian, and is being hurt by the Bayonet. That assumption excludes the player still learning the opening.

The real contradiction is that the product's benefit runs inversely to the buyer's rating. The lower-rated player gains most from the reduced memorisation load, and gains least from the strategic reasoning the line demands. In a low-forcing line, errors do not show up immediately; they show up ten moves later as a position quietly shrinking. A player without planning experience will convert understanding into nothing and will not notice when it happened.

The genuine sweet spot is roughly 2026 to 2200: enough understanding to execute a plan, enough honesty to step away from asking the engine about every move.

Time control sits in the same place. A strategy-first line is built for classical, not blitz. Blitz is governed by pattern recall; you do not need a thirty-move plan, you need a shape you recognise in ten seconds. A line like 9...a5 will neither save you nor stop you in blitz — it will simply sit there while the clock runs out.

The truth about low-density lines: playable and unrefuted does not mean strong

Here I have to rein in my own reflex. My first instinct is to invert every claim, dethrone every crown, dig under every legend. But when the advertisement itself concedes that chances are equal, there is nothing to invert — it deserves credit. That is more honesty than most opening products offer.

So the objection is not to the move. The objection is to the evidence that was not sent.

If 9...a5 were objectively promising, it would sit much closer to 9...Nh5 in elite practice. Low theoretical density usually proves that a line is playable and unrefuted; it is not proof that the line is strong. The world holds thousands of lines nobody has refuted, which is precisely why nobody plays them.

And here is a distinction the pitch erases. 9...Nh5 and 9...a5 are not rivals; they are two parts of one repertoire. One forces and can be checked by a computer; the other is loose and demands human judgement. The same player can hold both — by mood, by opponent, by round. The advertisement needs a rivalry because rivalry manufactures product. The rivalry does not exist on the board.

I tested the line on a screen. My remote press box is the old one — screen recordings, timestamped notes, live-rating screenshots. The stadium was empty, but the server room was full of noise. Playing 9...a5 from the stem, the evaluation stayed around zero — sometimes +0.1, sometimes 0.0, sometimes −0.1. No explosion, no collapse. What was visible was that the engine's first choice kept shifting slightly, and turning those moves into a story a human could remember was hard. That is the true part of the claim. The engine is not blind; the engine's answer does not arrive in human hands.

The contrarian angle: the buyer's risk is one-sided

Here is what no opening advertisement ever prints.

When you play a forcing line and something goes wrong, you find out immediately. When you play a loose line and something goes wrong, you may never find out. In the first case the engine tells you that you were wrong on move eleven. In the second you lose on move thirty and conclude it was a bad day.

That is the asymmetry. A club player adopts 9...a5, works on it for three months, and if his planning skill never reaches the level the product assumes, he has no forcing line to fall back on either. He is stranded between an engine he does not consult and a plan he cannot execute.

And this is where the pipeline question arrives, the one I ask of every subject. Who made this product? Who tested it? How many bought it, how many finished the first chapter, how many played it in a tournament, and how many gained rating from it? Not one of those numbers is available. We count the pipeline that produces champions and never the pipeline that produces products — though both ask the same question: how many came in, and how many fell out.

I think of an afternoon at a district rapid in Balewadi. Forty digital clocks running at once, parents standing in the corridor, and an arbiter walking back to the same child again and again because he kept forgetting to check his own time after the increment. There was no 9...a5 in that hall. There was a thirteen-year-old trying to work out whether he had three moves pre-planned in a forty-five minute game. Opening products are sold at a distance from that reality — they address a player who already owns the thinking, and give him the confidence that his deficiency was knowledge.

One more thing, about media

I date chess's modern era to 2026 — not to a trophy, but to streaming and platforms like ChessBase India. That shift did more than decades of administration. It also created a new problem: reach grew, verification did not.

An opening product can now reach two hundred thousand people in four hours, and not one technical audit of it will be published anywhere. Because auditing is expensive, boring, and requires writing down moves nobody likes. What chess lacks most is not reach; it is long-form, verifiable writing. I wrote that sentence five years ago and it has become truer every year.

I also remember the night an Online Olympiad final was decided by a server collapse. I wrote three thousand words arguing that a shared world title had been settled by a server, not a bishop. Two outlets passed; a digital platform ran it whole. Since that night I have written move numbers and clock times with the same precision I once gave to 200m splits. I keep the stopwatch running long after the race has ended.

And in this advertisement the clock is nowhere. No game, no move number after nine, no time. Nine information points, zero verifiable facts.

Takeaway

I am not saying 9...a5 is wrong. I am saying that until a product shows who played it, when, what the engine said and what the database says, the move is not a recommendation — it is a promise.

The day an opening product prints twenty model games by name, publishes engine match rates and ACPL, and hands a White team a month to try to break it, I will buy it and write about it. Until then, the fourth lane remains the place where the official story starts to drift. And at the rewrite desk, the scoreline was never the whole sentence.