9...a5 and the Data Vacuum Inside a King's Indian Repertoire DVD

Three in the morning in Shanghai. I open a file in my opening folder and find an unfamiliar name on the cover: a chess opening DVD titled "A farewell triumph: Capablanca wins gold in Buenos Aires 2026". I click in. There is no Capablanca inside. No Buenos Aires. No 2026. Inside is a purely technical story about the King's Indian Defence, about Black's ninth move, about the Bayonet Attack and a move sold as an antidote: 9...a5.

A cover speaking of the dead. Content speaking of a living position. The two do not touch at any point, and in the gap between them sits the thing my trade is paid to see: a data vacuum. Not one reference game cited. No engine match rate. No practical scoring. A product selling a positional idea while carrying a blank passport.

That is why I sat down to write this while it was still dark. The real story lies in how the move is packaged, more than in the move itself.

The King's Indian and the name Mar del Plata

The King's Indian Defence has an odd history at the top of the game. Structurally it is a declaration that the centre need not be occupied immediately. Black plays 1...Nf6, 2...g6, 3...Bg7, then 4...d6 and 5...e5. White is allowed to build a large centre with 1.d4, 2.c4, 3.Nc3 and 4.e4. Black concedes space in order to acquire a target. For decades the system was treated as eccentric, until Bobby Fischer and later Garry Kasparov turned it into a world championship weapon.

Its sharpest branch is the Mar del Plata Variation, named after the 2026 tournament in the Argentine city where the structure was pushed to its competitive peak. The main line runs 7.0-0 Nc6 8.d5 Ne7. White freezes the centre with the pawn thrust to d5 and pours his pieces onto both wings. Black retreats the knight to e7 and prepares ...f7-f5 to open the kingside. The result is a two-wing race, each side with its own plan and its own clock. Whoever arrives first wins; whoever is one tempo late is buried.

For decades White answered with slow queenside plans. Then a completely different ninth move appeared: 9.b4.

The name describes the thing precisely. Bayonet Attack. Not a developing move, not a consolidating move. A thrust into the queenside that seizes space, threatens b5, and above all forces Black to answer the queenside question before being allowed to think about ...f5. White does not try to play the familiar position better. White changes the position entirely.

And when an opening line becomes fashionable, the opening market responds by a law that closely resembles the transfer market: demand is generated not on the attacking side but on the receiving end. Nobody sells White a way to play 9.b4. People sell Black a way to survive 9.b4.

A move ignored for thirty years

The traditional answer is 9...Nh5. The knight goes to the rim, frees the f4 square for the pawn, and prepares ...f5 at all costs. It is an aggressive move and an honest one in the spirit of Mar del Plata: Black accepts a poor structure in exchange for a free hand on the kingside.

The price of aggression is compulsion. After 9...Nh5 the position opens, the sequences harden, and once the sequences harden the player stops playing chess and starts playing memory. At elite level memory is a commodity with buyers and sellers; at club level it is a debt nobody can service.

The remaining move is 9...a5. A pawn on the rim. No piece developed. No immediate threat. To the naked eye it looks like a concession, which is exactly why it was underrated for decades.

Read structurally, though, 9...a5 does three things at once.

It forces White to decide about the queenside immediately. White's b4 pawn faces a choice: advance to b5, exchange on a5, or stand still and absorb pressure. Each choice leads to a different position, and none of them is free. White loses the right to postpone.

It removes the timing of White's lever. In the Bayonet Attack the strength of b4 lies not in the pawn itself but in the moment b5 is pushed. Push early and it is weak; push late and it is strong. 9...a5 narrows that window, and once the window is narrow, White's plan loses flexibility.

And it preserves the ...f5 plan while wrapping it in strategy. Black will still push the kingside pawn. Only now the queenside question has been settled first. The order of operations is reversed, and order is the entire content of this move.

Some would call that a slowdown. I call it buying time with one small square at the edge of the board.

The most valuable sentence in the whole package

Amid a forest of promotional language, one sentence made me read it three times: the objective evaluation of the position is equal.

An opening product admitting that the move it sells confers no advantage. That is rare enough that I made a note. Most products in this industry are packaged in the language of overthrow: a novelty, a refutation, a forced win. Here it is the reverse. The seller says plainly that the position is balanced, and that the value lies in the opponent suffering more than you do.

That is an argument about behaviour, not about advantage. And because it is an argument about behaviour, it is far more honest than its packaging suggests.

I once trusted feeling, until a number knocked on my door at three in the morning. This time what knocked was the absence of numbers. And that absence taught me something else: when a seller voluntarily lowers his own promise, the rest of the package deserves a more serious hearing.

The position is equal but difficult to play. For Black it demands a plan rather than a memory. For White it demands that he find a plan instead of following a signposted road. One side is handed a map; the other has his map taken away. That is the entire commercial value of 9...a5.

The data vacuum and what it costs

Now the part I am obliged to state plainly, because stating this part plainly is what my trade is for.

The package supplies no measurable indicator. No engine match rate. No average centipawn loss. No practical score for the line in serious games. No sample size. No named author, no named second.

For a product selling safety, that is a systemic omission. The buyer can verify nothing except the seller's tone of voice.

And I know that feeling exactly, because I have been on the other side of it.

Based on my experience following games at open tournaments and team events over many years, I have noticed a recurring pattern: lines advertised as having "few forced variations" tend to have shorter lifespans than lines advertised as sharp. The reason is not the quality of the move. It is the structure of the market. A sharp line defends itself with complexity; to break it, an opponent must invest heavily. A strategic line defends itself with ambiguity; to break it, an opponent needs one evening with an engine and one correct idea.

In 2026, at forty-eight, I sat down after a game in Shanghai and asked myself a question I still ask every week: which move in this position is the move my opponent does not want me to play? I answered it by reading positional data rather than commentary. Since then I have held one rule: every opening judgement must answer the question of who suffers more over the next twenty moves.

Apply that rule to 9...a5 and the answer is fairly clear. The sufferer is White. White has more space, but space only has value when there is a plan to use it. Black has less space but a clear plan and a clear target. In chess, a clear plan usually beats wide space, at least for the first forty moves.

But here I must lower my voice. Everything above is my structural reasoning, not the seller's data. I can argue that 9...a5 is sound. I cannot claim it works, because nobody handed me a single figure to check.

The business model of repertoire content

One thing is worth noting about the timing of this product. It does not arrive as an invention. It arrives as an answer to a fashion.

9...a5 has long existed in the arsenal of serious King's Indian players and has appeared in the practice of some leading grandmasters. This product did not invent it. It systematised it, placed it inside a learnable structure, and above all attached it to an emotional promise: you will not have to memorise any more.

That is the core business model of the entire repertoire-content industry. It monetises the gap between opening fashion and preparation effort. When a line becomes popular the gap widens and a product appears to fill it. When the line goes out of fashion the gap closes and the product becomes an archive item.

I do not consider that fraud. I consider it a structural feature of the trade, much like the way the transfer market sells young wingers from smaller leagues at the price of a dream. The transfer market does not buy the past. It buys what the data has forgiven. The opening market is the same, except the currency is study time rather than euros.

Capablanca, Buenos Aires and a cover that lies

The 2026 Chess Olympiad in Buenos Aires was the first held in South America. José Raúl Capablanca, by then past his peak, played board one for Cuba and took the individual gold medal on first board. He died in March 2026, three years later. Across his whole career Capablanca was never a King's Indian player. He belonged to the classical school, to clean positions and well-preserved structures. Attaching his name to a King's Indian DVD has no technical basis whatsoever.

There are players the world forgets, but data never forgets them. Here it is the reverse: data forgets and the name is remembered. The cover remembers Capablanca. The content never mentions him once. A small paradox, but it says a great deal about how this industry operates: the name sells, the content does not.

I have no evidence that the pairing was deliberate. It may be an editing error. It may be a template headline attached to the wrong file. Whatever the cause, the consequence is identical: the reader receives an information package whose metadata cannot be trusted. And when the metadata cannot be trusted, the content must be read with matching scepticism.

In my working vocabulary I call this a misidentified resurrection. A name is dug up to decorate an unrelated product. The name is revived but not respected.

What computers actually do

One claim in the package stopped me longest: that computers often do not know how to handle these complex strategic positions.

I understand what the writer means. There is a truth about humans in it. The strategic mazes of Mar del Plata demand a long-range plan, and long-range planning is precisely what amateur players are not trained to build. A club player entering that position without a guide will wander for forty moves and lose to a blow on the opposite wing.

Read literally, though, the claim went out of date long ago. Modern engines handle these structures very well, usually better than grandmasters do. They are not confused by a closed position. They simply evaluate it differently.

This is where I want to draw a line, because it is a fine example of a very common reasoning error: conflating "difficult for humans" with "difficult for computers". The two propositions are not equivalent, and in many positions they point in opposite directions. What exhausts humans is often what engines find easiest: clear structure, clear target, few choices. What confuses engines is often what humans enjoy: chaos, many equivalent choices, no clear plan.

A pressing sequence is not noise. It is a question that data is whispering. The question whispered here is: if the position is objectively equal, where does the product's value actually reside? The honest answer is that it resides in an asymmetry of effort, not an asymmetry of evaluation. That is a good answer. It is simply not the answer the cover is promising.

The paradox of learning less

Here is an objection the seller may not have considered.

The entire appeal of the product rests on a promise to reduce memorisation. Fewer forced variations, fewer sequences, more room to think for yourself. It sounds sensible, and emotionally it is exactly what a club player needs.

But look at the clock.

Strategic positions, where no sequence is forced, are the positions that demand the most thinking time. There is no single correct move to find in three seconds. There is a plan to build over twenty minutes. In rapid and blitz that is a serious handicap. Conversely, the forced sequences of 9...Nh5, however long and hard to learn, are things that can be played from memory under time pressure.

In other words, this product promises to reduce the study load while increasing the thinking load. For a player with plenty of clock and little study time, that is a good trade. For someone playing online blitz every evening, it is a reverse trade.

This is the kind of detail a repertoire product with data would have to state clearly, because it determines who the product is for. Without data the buyer must guess which group he belongs to. Most buyers will guess wrong, because most buyers do not know how they play until they are beaten in exactly that way.

Four signals I will track

I will not close with a summary table, because a summary table belongs to stories that are finished. This one is not.

The frequency of 9.b4 in top-level games. If it keeps rising, the product retains value. If it stalls, the product has expired without notification.

The practical score of 9...a5 in serious games. This is the only figure that can confirm or refute the product's whole argument, and it is entirely retrievable. Its absence from the promotional package is a choice, not a technical oversight.

The appearance of a clear counter-preparation for White. Once someone finds a systematic way to meet 9...a5 with 9.b4, the advantage of "few concrete lines" becomes the disadvantage of "few lines to choose from".

The identity and credentials of the author and second. This is the decisive variable the package leaves blank. A respected grandmaster saying one thing and a player who has never competed at the top saying another are not the same product, and the buyer cannot tell them apart unless he is told.

I light candles for data. But I always let the flame of feeling light the way to the question. And the question I kept after closing the file near dawn was not whether 9...a5 is strong. It was this: if a move is good enough to sell, why will nobody produce a single number to prove it is good enough to buy?

When that answer arrives, I will open the file again at three in the morning. Until then, I keep it in the folder of things worth watching and not yet worth trusting.