The Empty Space Between Capablanca's 2026 and the King's Indian's 9...a5

The headline on the screen read: “A farewell triumph — Capablanca wins gold in Buenos Aires 2026.” I opened it because I expected the room I had been looking for: the high ceiling of the Teatro Politeama, a fifty-year-old Cuban at the board, and across from him the coldest calculation Europe could produce. What opened was a product page. King's Indian Defence. Mar del Plata. Bayonet Attack, 9.b4, and in reply a recommendation of 9...a5. Capablanca's name did not appear once. Neither did Buenos Aires. Neither did 2026. There was a DVD, an opening system, and a sentence repeated three times — that computers cannot handle these complex strategic positions.

I closed the page twice that night and reopened it both times. The first time because I thought I had clicked the wrong tab. The second because I wanted to know whose error it was — mine, or the page's, which had stitched two entirely separate worlds under one heading.

The room the headline promised

Buenos Aires, August 2026. The eighth Chess Olympiad. Cuba was putting a man on first board who had shown the world two decades earlier that chess could be played like arithmetic. José Raúl Capablanca played eleven games there and scored eight and a half — 8.5/11. That was the board one gold medal. The last major international title of his career.

This was an era when no player's strength was measured by a number. The Elo system arrived in 2026; in 2026 a performance was recorded only in scores, opponents' names and small newspaper headlines. So that gold has no rating equivalent. What exists is this: eleven games, eight and a half points, and a man carrying his country's first board past the age of fifty.

And a date that chess books rarely give enough room — that Olympiad ended in September 2026, the same week Germany invaded Poland. Many players could not go home; they were stranded in Argentina and some settled there. Capablanca's gold is one of the last photographs of a world about to close. That is the room the headline promised me.

I like that room. I like it because I wrote the first long piece of my career out of a room like it — March 2026, a rented hall in Balewadi, a three-man webcast, one camera, one 4G dongle and a borrowed mixer. At least eleven hundred people were connected at once and none of them knew what they were part of. One camera, one dongle, and a nation that did not know it was watching. That year I stopped opening with the result. Every report began with a fifteen-hundred-word scene note — the smell of the hall, a coach's folded arms, a mother's tiffin box — and the analysis was built on top of it.

So when a page shows me Capablanca's room and serves me a King's Indian course inside, I feel cheated — not angry, cheated.

What was actually on the board

The King's Indian runs like this: 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. By now the board has split in two. White has the big centre and the space; Black has a heavy, rehearsed plan — ...f5, a strike on the kingside, and lightning out of the centre if White slips.

The whole structure takes its name from Mar del Plata because it surfaced repeatedly at a 2026 tournament in that Argentine coastal city. Players of that era worked out that letting Black abandon the centre for a wing attack pays, provided the calculation holds. For decades the system belonged to players who knew how to take risk.

Then came the Bayonet Attack — 9.b4. White says: fine, play ...f5, but first I will take space on your queenside. The move entered the mainstream in the nineties and 2000s and was played at the highest level. The moment 9.b4 lands, Black faces a clean question — what kind of chess do you want to play?

The fork at move nine

The page recommended 9...a5.

To understand it you have to understand 9...Nh5. That is the main reply. The knight goes to h5, the ...f5 threat is prepared, and the board opens quickly. These lines have concrete moves, forcing content, and they are among the most deeply mapped branches of the entire King's Indian. Any serious White player arrives with hours of preparation on them.

9...a5 is something else. It keeps the board closed, presses on the queenside, and says: I do not want a quick result today, I want to tire you out. Fewer concrete lines, more judgement.

The real trade here is clean and honest — a swap between memory and understanding. The player who chooses 9...Nh5 bets on his recall; the player who chooses 9...a5 bets on his comprehension. One carries a high memory load and low uncertainty; the other carries a low memory load and high uncertainty. For any club player that is the actual question, and in that sense the DVD's core idea is not wrong.

Memory versus understanding

In all the matches I have watched from beside the board, the biggest gap shows up in exactly that place — how many lines a player remembers, and how many positions he understands. November 2026, Kolkata, the first Tata Steel Chess India. Anand and Carlsen sat four feet apart and the hall fell silent before a move. I still remember that silence. What was happening inside it was not memory. It was understanding.

Closed King's Indian structures demand precisely that. You do not lose them to a wrong move; you lose them to a wrong plan. You can play thirty moves perfectly and then discover your king has been opened and nothing is left.

And this is where preparation economics does its work, which is the strongest part of the recommendation. A well-prepared White player has deep files on 9...Nh5. His files on 9...a5 are thin. So the move's real strength is not that it is better — it is that it makes the opponent's preparation worthless. This is a preparation-economics argument, and it is sound even if the claim of equality cannot be verified.

“Computers don't understand” — an expiry date

Now the sentence that appeared three times on that page and stopped me hardest — that computers cannot handle these complex strategic positions.

That sentence was once true. Engines before roughly 2026 did wobble in closed King's Indian positions, especially where long-term positional sacrifices were involved. They calculated deep and did not see direction. That era is over. Neural-network engines — NNUE-class and Leela-style networks — now evaluate these structures reliably. “The engine gets lost here” is folklore now, not technical fact.

This does not mean 9...a5 is a bad move. It means part of its sales pitch rests on a claim whose date has passed. And in chess, dates matter. What was true in 2026 becomes marketing in 2026.

The numbers that are not there

There was not a single figure on that page. No engine match rate, no ACPL, no database win-draw ratio, no game citation. Saying “equal but dangerous” is easy; proving it takes games.

This is my habit and I do not hide it — before every long piece I run a separate verification pass. Facts only, prose untouched. On this page that pass ended immediately, because there was nothing to verify.

What can be inferred is that 9...a5 is almost certainly not the engine's first choice, and that its benefit begins to decay the moment the line enters an opponent's files. Preparation economics cuts both ways.

The antidote illusion

Here is the part where I have to be most careful, because this is where language sells, not chess.

“Antidote” means a cure — something that removes White's advantage. 9...a5 is not that. It is a legitimate, coherent, practical alternative that reduces the payoff of the opponent's preparation. Buyers who take it expecting a genuine cure will be disappointed, and that disappointment will have been manufactured in their expectation, not on their board.

Second, time control. This DVD talks about thinking. But where do club players lose most of their points? In fast time controls. And precisely there, where pressure is highest, low-forcing positions are most expensive. The positions that need thinking time are the ones with no thinking time. That contradiction was nowhere on the page.

Third, and least said — in these closed structures White has low-risk answers. Keep the queenside shut, play on space, do not break the structure. Then Black never gets the dynamic compensation without which the King's Indian is incomplete. The “few concrete lines” feature cuts both ways: it saves Black's memory, but against a strategically competent opponent it also narrows his routes to a win.

The empty space

I come back to that first night, and to the page that joined Capablanca to the King's Indian.

Two readings are possible. The first is that this is part of a multi-headline newsletter page where several headers live together, and the wrong one got attached to the product text during assembly. That is a technical error, and technical errors get fixed. The second is that it is a deliberate wrapper — borrowing the warmth of a historical triumph to earn a click on an evergreen product page.

Either way the result is the same: a content-integrity flag. The reader did not get what he came for, and what he got was not sold under its own name.

And there is a larger thing visible here, one that is becoming normal in chess publishing. We have a finite set of great historical moments — Capablanca's 2026, his 2026 match against Alekhine that ended 18.5–15.5 with twenty-five draws in thirty-four games — and these moments get borrowed repeatedly because they generate feeling for free. But borrowing has a cost. When a real triumph is used as a wrapper often enough, one day the real triumph itself starts to look like a wrapper.

That is the most expensive loss here. 9...a5 did not need Capablanca to sell. The move could stand on its own — as a recommendation that says: your memory is weak, so bet on your understanding. That is an honest sentence. It does not require 2026.

Looking forward

What is left is a question: the next time a chess page opens, what will the reader trust — the headline, or the board?

My guess is that trust will drift slowly toward the board. Because the board does not lie. Play 9...a5, and if White keeps the queenside shut, you will learn exactly what you have and what you do not. No headline can change that.

Buenos Aires 2026 deserves its own page — separate, complete, with no DVD attached. And 9...a5 deserves its own page — with no Capablanca attached. The day those two separate, chess publishing will be a little more honest. And perhaps that day some club player will understand that the DVD he bought did not come from the room he was promised — it came from a different room, and in its own way, a better one.