The Invoice for an 'Antidote': A 9...a5 Promo Text With No Game, No Date, No Engine Number

The promotional text contains exactly three numerals: 7...Nc6, 9.b4 and 9...a5. Beyond those, not one figure — no Elo, no date, no engine evaluation, no centipawn loss, no database percentage, no price, no author's name. I read it three times, the third time running a finger along the lines to check whether a player, a cited game or a publisher had slipped past me. Nothing had slipped, because nothing was there.

A text that calls itself an "antidote" to the Bayonet Attack — recommending 9...a5 for Black in the King's Indian Mar del Plata — cites no game at all. The disease is named, the medicine is named, and the trial report is missing from the prescription. My trade has one name for this: a claim without a receipt.

The Classical Mar del Plata structure rests on a closed centre. Black plays ...Nc6, ...e5 and ...f5, White plants d4-e4, and two separate attacks are built on opposite wings — White on the queenside, Black on the kingside. For three or four moves nothing appears to happen, and then the whole game narrows to one question: whose attack arrives first.

White's 9.b4, the Bayonet, is a clamp. The pawn lands on b4, threatens b4-b5 and c4-c5, opens queenside files and presses on Black's ...c5 or ...b5 base. Black's standard reply has been 9...Nh5: the knight eyes the d2-pawn from the rim and steers the game onto clear, forcing, and therefore engine-checkable ground.

The promotional text offers an alternative at that exact point: 9...a5. A queenside pawn strike at the root of the clamp, pulling the game away from the forcing 9...Nh5 channels into a manoeuvring fight. Four claims travel with it — that computers often do not know how to handle the arising complex strategic positions; that there are very few concrete lines; that strategy beats memorisation; and that if Black knows what to do, things may quickly become dangerous for White.

What the text omits matters as much. No author, no publisher, no price, no publication date, no distribution channel, no game, no player, no engine match rate. The genre is unmistakable: a product introduction, not a tournament report. The archive does not remember the narrative; it remembers the timestamp — and there is no timestamp here, which means this text could have been published in any year of the last decade.

The word "antidote" is the settlement rule of the whole document. An antidote presumes a poison and promises a cure. Yet the text concedes in one sentence that "objectively chances are equal." That is unusual honesty for promotional copy, and it is also the largest technical tell. What is being sold is not a refutation but an equalising line with practical winning chances. For a Black player who needs +1 in the last round, the instrument is structurally wrong: an equalising tool is useless where a win has to be recorded.

I rebuilt the rules from the dispute backward, not from the handbook forward. The first question is therefore not whether 9...a5 is good or bad. The first question is under what claim it is being sold, and which document that claim rests on.

Audit the engine claim. The text says computers often do not know how to handle these complex strategic positions. That sentence is written in the vocabulary of the 2000s. In the NNUE and neural-engine era, closed King's Indian structures are evaluated far more reliably than they were then. The defensible version of the claim is much narrower: in closed positions engine evaluations are less stable and less humanly actionable — the number hovers near equality and the engine's first-choice move is hard for a human to convert into a plan. That is a statement about usability, not about engine ignorance. The text merges the two and lets the stronger, false version carry the marketing.

And a text making a claim about engines supplies not a single engine number. No match rate, no average centipawn loss, no evaluation range. This is the same error class I audit weekly in football data: xG does not explain a coach's in-game decision, it only describes the situation. An engine evaluation describes the position; it is not evidence of a player's understanding.

The economics of unfalsifiability begin here. "Very few concrete lines, strategy over tactics" is internally coherent and commercially self-undermining. A low-forcing line gives the buyer no forcing sequence to test before purchase, and no way to be refuted after it. The Centipawn Ledger was never a prediction machine; it was a receipt for every assumption. A receipt needs line items; there are none. A low-forcing line is a stress test for the ledger precisely because it contains no number to check.

A clean tactical story still stands behind 9...a5. White's b4 is a space clamp; Black's ...a5 cuts at the root of that clamp and pulls the game off the forcing 9...Nh5 channels. That is a legitimate design — avoid the machine, play the human. The price is equally clear: Black accepts a permanently smaller share of the position's dynamic assets in exchange for strategic clarity. The execution profile follows from that. Low forcing density means fewer immediate tactical blunders and a higher risk of cumulative strategic error; the burden shifts from memory to judgment.

Here I return to my own desk. Since 2026 I have kept a coded set of 1,842 archived classical and rapid games with centipawn loss and time usage both recorded. In lines like this one the error does not arrive as a blow; it arrives as drift — five small inaccuracies that never trip a tactical alarm. That is why such a line is hard to sell and equally hard to refute.

One terminological check remains. The text calls the Bayonet "fashionable." 9.b4 is not a fashion; it is a long-established main line — this is general theoretical knowledge on my side, uncited in the source. Calling it fashionable manufactures the sense of a new problem requiring a new solution. That is framing bias, not fact.

The text tells us who the buyer is. It assumes a player who already owns a Mar del Plata repertoire (7...Nc6) and is being hurt by the Bayonet. The rating band is roughly 1800 to 2300 club and tournament strength. The genuine sweet spot sits between 2026 and 2200: a lower-rated player gains most from the reduced memorisation load and gains least from the strategic reasoning the line demands, while a higher-rated buyer will engine-check on day one and ask for numbers. On time control, this is a classical product; blitz and bullet reward pattern recall, not planning. The psychological profile is written into the copy — players who prefer to think for themselves rather than memorise long variations. That is coherent segmentation, and it explains the missing data.

Which forces the inverted conclusion. The absence of data here is not carelessness; it is architecture. A strategy-first line selects for buyers who do not demand engine numbers. Marketing silence is the product's demographic, not its failure. The other edge of that argument cuts the author: a claim that cannot be falsified protects him from criticism and costs him the trust of the sophisticated buyer who could have paid the most.

The second inversion is the deliberate framing of 9...Nh5 as the bad, computer-friendly path, manufactured to create contrast for the recommendation. In reality both branches coexist in a Black repertoire; they are complements, not rivals. The product needed a villain, so one was written.

The third is the hedge that "if Black knows what to do, things may quickly become dangerous for White." That sentence moves the burden of proof off the product and onto the buyer's understanding. Standard marketing architecture, not evidence.

And who carries the cost? An unnamed buyer — say a 2026-rated club player who purchases the product, sits down in the position, and loses to a plan he was never taught, with no forcing line to fall back on. On the other side, the claim has no name attached: anonymous author, anonymous games, anonymous date. In my ledger an unsourced claim gets struck. Here there is no source line at all.

Three documents would make this auditable: the author's own games with 9...a5, with results and ratings; engine match rates and evaluation ranges across the model positions; and an explicit statement of which positions are treated as equal and which as playable for a win. Without them this is a promise, not a repertoire. Every odds column is a settlement rule waiting to be read out loud — and this column never wrote the rule.

The question that remains: if the line's real strength is unfamiliarity, what will it be worth on the day the opponent arrives prepared?