Capablanca in the Headline, 9...a5 in the Body: The Empty Press Box of Chess Content

August 2026, Buenos Aires. Board one of the eighth Chess Olympiad. Cuba's José Raúl Capablanca scored 8.5 from 11 — six wins, five draws, no losses — and took the individual gold medal on first board. Midway through the tournament word arrived from Europe that war had been declared; several European players were scrambling to get home, and that Olympiad ended like a scattered ceremony. Capablanca died in New York in March 2026. The Buenos Aires gold was his last great title.

I knew all of that from the headline. None of it was in the article.

The headline arrived on my screen like this: "A farewell triumph: Capablanca wins gold in Buenos Aires 2026." I clicked. What opened had no Capablanca in it, no Buenos Aires, no 2026, and not a single named player. It was an advertisement for an opening DVD on the King's Indian Defence, Mar del Plata Variation, explaining why Black should play 9...a5 against White's Bayonet Attack.

What I found in the Jurong West press box was not silence but arithmetic. Here the arithmetic runs backwards. One player in the headline, zero in the body. One city in the headline, zero in the body. One year in the headline, zero in the body. In the body, not one cited game, not one database percentage, not one engine named, not one player. Two readings are possible. First, this is a fragment of a multi-headline newsletter page and the wrong header got attached to the right product text. Second, this is an evergreen product page that borrowed a great name to earn the click. Either way it is a content-integrity flag, and I log it and move on — because if you strip the header away, the body is not empty. It contains a real argument.

The move language of the King's Indian

The order of moves has to sit in front of us, because everything here hangs on the order: 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 King's Indian's basic bargain is this — White builds a big centre, Black encircles it, then breaks against it. The Mar del Plata branch, whose name is generally tied to the same Argentine city where the idea gathered force in 2026, replaces the Classical system's closed, slow strategy with a straight race: ...f5, ...f4, ...g5, ...g4, forward on the kingside.

White's Bayonet Attack, 9.b4, answers that race on the queenside. White abandons the Classical f4-centre plan and prepares c5 and b5 — shut the queenside, open the hands on the kingside. It entered the mainstream in the 1990s and has been a permanent headache for Black since. Black has two answers. 9...Nh5 is the main line: the knight heads for f4 or supports ...f5, the position opens, the forcing content is very high, and this is arguably the most deeply engine-mapped branch of the entire King's Indian. 9...a5 is the other road: an immediate queenside challenge, thinner concrete theory, fewer forcing moves, and therefore more judgement.

The economics of preparation

The product's claim sits here, and in its narrow form the claim is sound. The real price of 9...a5 is not equality; it is cutting the interest on White's preparation. A well-prepared White will have deep files on 9...Nh5 and almost nothing on 9...a5. The move buys time on the clock before the game, not an edge on the board. That is not a chess argument, it is a preparation-economics argument — and it is the honest part of the advertisement.

This is a memory-versus-judgement trade, and the trade is real. 9...a5 carries fewer concrete lines, so the memorisation load drops and the positional-understanding load rises. 9...Nh5 inverts that. For a player who does not want to rote-learn, the exchange is convenient. For a player who wants to rote-learn points, it is expensive. Which buyer the DVD is calling becomes clear early — the club player who ranks understanding above recall.

Where the engine loses the thread, and since when it does not

Now the sentence that is the weakest technical bone in the advertisement: the claim that computers often cannot handle the complex strategic positions that arise. That sentence was true around 2026. What changed between 2026 and 2026 was not merely speed but method: neural-network engines arrived — Leela Chess Zero in 2026, and NNUE evaluation with Stockfish 12 in 2026. The long-term positional-sacrifice evaluations in closed King's Indian structures that once confused engines are now routine work for them. What was a technical truth in 2026 is marketing in 2026.

That warning is aimed at the buyer, not at the product. When a repertoire's sales pitch rests on "the engine gets lost here," the first question should be: which engine, from which year, on what hardware. The answer is usually that the pitch is running on an old map. And where the pitch runs on an old map, the surviving argument underneath also needs checking.

That argument, in its narrow form, holds. The preparation-economics part is solid, because it does not claim objective equality; it claims only that White's files will be thin here. The problem is that the product goes further and uses the word "antidote," and an antidote means a cure — a refutation of White's advantage. 9...a5 is not that. It is a move whose value depends on the opponent being under-prepared, which makes it a market claim rather than a chess claim.

The feature that cuts both ways

"Few concrete lines" is a feature that can be sold and can also be suffered. White has a cheap answer available: keep the queenside closed, play on space, fight with structure. The moment White does that, Black does not get the dynamic compensation without which the whole King's Indian bargain is meaningless. So the very feature that makes the move recommendable also limits Black's routes to winning against a strategically competent White.

The second edge is the time control. The DVD's entire premise is that you will think for yourself at the board. The place where amateur players lose the most points is blitz and rapid — where there is no time to think. The repertoire looks most comfortable exactly where it is weakest.

Third, and most important: the sample. How many games, from which database, by which players, from which years. Not one citation is offered. A repertoire without cited games is a promise whose account was never opened. I have learned to read every transfer window as a confession booth with a spreadsheet; the repertoire market is that same booth — a claim, a price, and a contract nobody shows you. Opening-content transfer windows run every season: every DVD calls itself an upgrade, and every upgrade carries the same fee, the risk of forgetting the old files.

The frame of absence

I began looking for the woman at the edge of the frame, because the frame is a choice. Here the edge of the frame holds Capablanca himself — name in the header, body nowhere. And the blind spot of collective memory is this: we call the mismatched header a small aggregator error and move on. The header is not an error; it is evidence. Chess's commercial layer sells the memory of great names because the product cannot show its games. The name arrives to fill the place where the citation should be.

Capablanca is ideal for that role, and it is no accident. In 2026 he was himself an anachronism — the Elo system arrived in 2026, which makes that Buenos Aires gold the last great victory of a world that had not yet been measured. The other end of his career reads the same way: the 2026 World Championship match ended 6–3 in Alekhine's favour with twenty-five draws, and that defeat is still kept in memory as a place where story does more work than number. That is why Capablanca suits clickbait so well: no database row can cut a claim attached to his name, because the claim was never a chess claim.

Who fills that silence is something I learned in Chennai, after turning down Doha. July 2026, the 44th Chess Olympiad: 188 teams in the open section, 162 in the women's; India-2 took open bronze, India-1 took women's bronze, and India won the Gaprindashvili Cup. The real work visible there was the Velammal-style school pipeline, the Tal Chess Club lineage, Tamil Nadu's cash awards, AICF patronage and PSU jobs — things you can count. At Budapest 2026 India won both the open and women's gold, and in December 2026 in Singapore, 18-year-old D. Gukesh beat Ding Liren 7.5–6.5. None of that came from the league promises typed in borrowed offices, promises I once believed; it came from arithmetic nobody was selling. The 2026 Chess Super League was a one-off online event; the Global Chess League launched abroad in 2026 and held its first India season in Mumbai in December 2026 — with Chennai, which produces all of it, still at the edge of the frame. What survives in chess is not the press release; it is the citation.

Four questions before you buy

So that page gives me not anger but a checklist. Put four questions to any opening repertoire: which games witness this line, which database, which engine and which version of it, and which time control it was built for. Content that answers all four may be expensive, but it is honest. Content that has only a great name is selling a headline, not chess.

And the image I keep is this: Buenos Aires, 2026, a fifty-year-old Cuban hunched over board one while war news came in, winning his last gold — and a screen in 2026 where that same name has been made the header of a page whose body does not contain him. A stadium is a room where absence learns to speak, and I listen. The question is not what the next DVD will teach. The question is whether, the next time a headline sells you a name, you will demand that name's game in the body.