A Capablanca 2026 Headline Over a King's Indian Body: Auditing the 9...a5 Preparation Trade

Hook

Buenos Aires, 2026. The headline says so — Capablanca takes gold, a farewell triumph. Now open the body. Inside is a commercial blurb for an instructional DVD on the King's Indian Defence, Mar del Plata branch, recommending 9...a5 against the Bayonet Attack. I counted every information point in that body by hand, line by line, the way the old desk reconciled ledgers. Mentions of Capablanca: zero. Buenos Aires: zero. 2026: zero. Names of any player, living or dead: zero. The headline has arrived from one book and the body from another. Two lines, one page, unacquainted.

This is not a minor editorial slip to be waved through. A headline is a promise; a body is a transaction. When promise and transaction disagree, what is left to the reader is trust, and trust is the cheapest currency in this trade. So this piece does two jobs. First, it reconciles the mismatch, because that mismatch contaminates everything downstream. Then it audits the technical claim actually being sold beneath the advert: is 9...a5 genuinely an antidote, or merely a sensible piece of preparation?

Context

The move order first. 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 — that is the Bayonet, and over two decades it has become White's sharpest, most concrete, most heavily filed attempt. White takes space on the queenside, presses the black centre, and — this is said too rarely — forces Black to decide where his dynamic compensation will come from, because the space is going into White's pocket.

Black has two roads. 9...Nh5 is the main line: violently forcing, deeply analysed, engine-mapped. 9...a5 — the recommended move — is a structural gesture on the queenside, thin on concrete lines and heavy on judgement.

Now the part the advert sells as "computers get lost here." That sentence was true once, in a specific window. Before the NNUE era — broadly, before 2026 — engines genuinely struggled to evaluate closed King's Indian structures where Black plays for long-term structural sacrifice. I have watched those evaluations wobble, watched them flip two moves later. But Stockfish-NNUE-class and Leela-style networks have largely retired the trope. Engines now read these positions with far more confidence. Read the sentence as marketing framing, not current technical fact.

There is another dating error folded into the headline itself: the Elo system arrived in 2026. In 2026 there was no rating coordinate in existence. The headline invokes an era its own body's vocabulary — rating, ACPL, engine match rate — cannot grip. Those who watched Capablanca lose the 2026 World Championship to Alekhine counted the score: Alekhine 6–3, twenty-five draws. That count remains the standard reference, and it too is absent from this text.

Core

So the real question: what is this recommendation good for?

One honesty first. I have not watched the DVD. I have only the advert text, which reached my desk and not a board. I cannot say how many games the author showed, because the text carries not one game citation, not one engine match rate, not one ACPL figure, not one database win or draw rate. A recommendation published without a sample must be defended from first principles, not from data. What I am measuring is therefore the product's claim, not the product. That distinction has to be held.

Technically, the choice between 9...Nh5 and 9...a5 is a real trade — and it is a trade between memory and judgement, not between right and wrong. In 9...Nh5 the moves force, the variations run deep, the memory load is high; a prepared White will catch you on a move whose name you have heard and whose analysis he has done. In 9...a5 the moves force less, the memory load drops, and the positional-judgement load rises. You remember less and decide more.

That is the defensible version of the advert's argument, and it is a different claim from the antidote. The real value of 9...a5 sits in preparation economics: a well-prepared White has thick files on 9...Nh5 and a thin one on 9...a5. You are reducing his preparation return. That economic argument holds. It does not establish that the move is objectively equal, and the advert's phrase "equal but dangerous" slips exactly where it leans hardest.

Open the structural mechanics and the picture clarifies. 9...a5 contests the b4 square, prepares ...axb4 and a later ...Rb8, and thereby opens a queenside file to press on. It keeps the e5/d6 pawn chain intact and it delays ...f5. That delay is both the point and the cost. In the Mar del Plata Black's traditional weapon is the kingside explosion — ...f5, ...f4, ...g5, ...Ng6, ...Rf7, then the break on the file. 9...a5 pushes that explosion a move or two later, and in exchange hands White time to convert space into a permanent structure.

I came up through football, so let me translate once, because one lesson carried across. The night at Luzhniki I watched Spain hold 74 per cent of the ball to Russia's 26 and out-shoot them 25 to 7, and go out on penalties; my model had Spain at 78 per cent to advance. I stayed until four in the morning re-charting all twenty-five shots and landed on an aggregate xG near 1.1 — volume without penetration. After that night I banned possession as an opening paragraph and made shot quality the first thing on the page. In chess the translation is direct: for Black, 9...a5 is not possession; the question is how much space he surrendered and what he is charging for it. Giving up space is not a result.

An old habit surfaces here. I charted 1,140 MLS sequences by hand before I trusted a single number, because at my price point nobody was selling pressing data. I chart the silence between events; that is where the model usually hides. In this DVD's text the silence is enormous — no games, no sample. So I have to treat the recommendation's entire strength as theoretical, and my audit stays theoretical too.

And what theory produces runs against the move, at least against the marketing claim. The moment you choose "few concrete lines," you have also thinned your own routes to a win. Against a strategically competent White, 9...a5 does not raise Black's winning chances; it relocates them — long, less fertile work instead of a fast attack. Draw rates rise, and that matters to anyone who sits down with Black expecting to win.

A third technical fixture worth counting: the time control. "Think for yourself" is heard in classical. In blitz and rapid the same position turns against its owner, because the time control where amateurs shed the most points is exactly the fast one, and under time pressure the quality of judgement in a low-forcing position decays fastest. The DVD's whole philosophy is weakest precisely where its buyer loses most.

One more signal the advert omits and should not. The practical value of 9...a5 decays the day it enters opponents' files. Every opening sideline has a shelf life, and the word "antidote" shortens it, because whoever sells an antidote has announced where the antidote is.

Another football lesson applies, and I will translate it once as well. In any transaction two things are separate — the claim and the settlement. On the football desk I called it settlement: the market closing and money actually arriving, which is a different object from a model's projection. The night the FIDE Online Olympiad final between India and Russia broke mid-game on server failure, the market voided and I spent it reconciling refunds, not results. The chess advert carries the same gap. An antidote is a claim, not a settlement. The advert sells you the claim; the settlement happens on the board, and there you are alone.

Contrarian

Now the part where I stand against my own preferred argument.

The trap in all of this is mistaking correlation for causation. The fact that strong players employ opening sidelines does not establish that the sideline is strong. The reverse holds too: thin database counts for 9...a5 do not establish that it is weak. Thin counts say only that few people play it, or that those who do never post.

Here I have to write down one of my own errors, because a receipt has two sides. At the 2026 Chennai Olympiad I went to watch Gukesh on board one — 9/11, individual gold. On the flight back I rebuilt my rating model around age-adjusted improvement curves, because static Elo held him near 2690 and my eyes held him elsewhere. On that call I was early and right. That same year I misread something else: I assumed that a sideline thin in the database was a sideline not worth playing. My archive from that year records that I treated thinness as a quality signal. It was wrong. The database was thin because players at my level were not playing it. That is the gap between claim and evidence, and I do not forget it.

So my position on 9...a5 is not that it is theoretically weak. It is that the buyer should know which currency he is paying in. He is reducing memory load and increasing judgement load. If his opponent is strategically strong and blunder-averse — and every club tournament has three of those — the memory saved will be repaid in judgement. The cost sits exactly where the advert is selling the saving.

And a final lock on the headline. This could be an aggregation bug on a newsletter page, two headlines printed together with the wrong header stuck to the right product. It could be a deliberate clickbait frame. I am not alleging bad faith, because I do not have evidence of it. But error and bad faith produce the same outcome for the reader: information that did not answer his question. If a piece of content must carry the weight of a headline, that weight belongs on the technical claims, not on a dead grandmaster's farewell.

Takeaway

Three things to watch. First, whether 9...a5 frequency in club-level databases rises over the next two seasons — if it does, the preparation-economics argument is working; if it does not, buyers made their own decision. Second, whether the next edition corrects the "computers cannot understand" line for the NNUE era or merely repeats it — that small sentence will reveal whether the product is being updated or repackaged. Third, whether a single game citation ever appears. Where a text carries no games, I can only say this: I counted, and that is what the count showed.