A Samarkand Headline, a 9...a5 Body: Three Data Gaps in a Chess Theory Product

A fourteen-point document landed on my desk one December morning with a headline attached: “Samarkand R2 (open): Despite close calls, all top teams grab wins.” I read the headline, underlined the word Samarkand, and turned to the body. There is no Samarkand in the body. No teams. No pairings. No round. Those fourteen information points describe something entirely different: a promotional blurb for a chess opening-theory product, a tutorial disc selling the ninth move 9...a5 in the King's Indian Defense, Mar del Plata variation, as an antidote to the Bayonet Attack. I trust footage over testimony, because footage does not lie. Here, the gap between the headline and the body is itself the first data point, and it says that something in the sourcing chain has failed. Everything drawn from the body must therefore be re-verified from scratch, inheriting no credibility from the headline above it. Thirty-two years of sports documentary work taught me one plain thing: the first sign of damage is always on the cover of the file. When the cover says one thing and the contents say another, the reader is entitled to either stop checking there, or to continue reading at a discount. The King's Indian Defense opens with 1.d4 Nf6 2.c4 g6. Black develops the bishop to g7, deliberately concedes the centre for some twenty moves, then strikes back with ...e5 and later ...f5. This is the hypermodern school: do not occupy the centre with pawns, let the opponent occupy it, then drill into it with pieces and flank counterstrikes. The Mar del Plata variation is the sharpest branch of that system. The canonical move order runs 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. White seals the centre with d5, squeezing Black's pawns, and forces both sides to pick a wing. From there the game becomes a footrace at opposite ends of the board: White pushes c4-c5 and b4-b5, opening the b- and c-files; Black pushes ...f5-f4, ...g5-g6, ...h5-h4, opening the g- and h-files. Whoever arrives first wins. The variation is named after the Argentine city of Mar del Plata and sits in the ECO range E97 to E99. From my own experience watching top-level games, this is one of the few opening branches where a spectator can actually feel the tempo quickening minute by minute, the way spikes on a track announce the final lap. The Bayonet Attack is White's answer. After eight standard moves, White plays 9.b4. The idea is concrete: seize queenside space before Black can stabilise, prepare b5 and c5, and slow Black's ...f5 counterplan by forcing an immediate answer on the opposite wing. From the 1990s onward, 9.b4 became the most uncomfortable test a King's Indian player had to face. Black has two routes. The first is 9...Nh5, the main move. The knight jumps to h5, eyes f4, supports ...f5, opens the position, forces concrete lines, and drags the game back into the territory engines analyse best. The second is 9...a5, the move this product is selling. Black's a-pawn hits b4 directly, forcing White to decide the queenside pawn structure at once before the race on the two wings begins. The game slows down, with fewer forced lines — and more fog. That is the entire technical content the document offers, and it stops exactly there. The rest is the reader's job. Three gaps appear the moment I go back through each information point and mark what is missing. There is no ACPL, no average centipawn loss for either side across the lines being taught. There is no engine match rate. There are no database statistics: how many games have reached this position, win rate, draw rate, results broken down by rating band. And there is not a single sample game with a player's name, an event, or a year attached. For a product selling opening theory, those three gaps are not a footnote. They are the floor the buyer stands on. Someone buying an opening disc is not buying inspiration; they are buying a promise that their positions will improve afterwards. That promise can only be measured with data. When the data does not exist, the promise converts into faith. What is interesting is that the product advertises having very few concrete lines. That is a market position, not a figure. Few concrete lines means what — three, ten, a trimmed tree that requires no memorisation? The document does not say. What it does say is a promise that runs against the grain of the Mar del Plata: the buyer will not have to memorise. I have spent years in edit suites, rewinding footage of races decided by a tenth of a second. Chess is the same. What decides is not whether you remember, but whether you understand the structure. Understanding a structure also takes time, and time is the most expensive commodity in a game played with a clock. Here, the product promises to save study time by trading forced lines for strategic positions. That is a defensible pedagogical argument. The problem is that it is presented as a conclusion rather than a hypothesis awaiting verification. The second gap concerns the true standing of 9...a5 in the opening tree. The move is a long-known sideline, not an invention. The document does not claim a novelty, and that is honest. But the packaging implies something else: that this is an antidote that solves the Bayonet problem. The main move remains 9...Nh5. Anyone who opens an opening tree in a database sees that within thirty seconds. So what is the product actually selling? It sells a shift in the character of the game — from a concrete, forcing fight where engines and memory dominate, to a strategic struggle where the player must improvise. That is a real value, but a psychological and pedagogical one, not a claim of theoretical superiority. The third gap is human. There is no author's name. No title. No competitive record in this very variation. No game of the author's is cited. In an opening product, the author's credibility is part of the goods, not decoration. A buyer needs to know that the person teaching them has sat in that position in a real game, with a clock, an opponent, and a rating behind them. Without those three things, the buyer is left with the tone of the promotional copy. And tone cannot be cross-checked. One sentence in the document stopped me longest: computers often do not know how to handle these complex strategic positions. This proposition carries the whole framing of the product, and it has a date. Claims of this kind belong to the era before neural-network engines. From late 2026, when AlphaZero published its self-play results, the picture changed completely. Modern engines handle the long-term, closed, manoeuvring positions of the King's Indian Defense far better than that sentence implies. If the claim is technically wrong, the question becomes what it is doing there. The answer is fairly clear: it is a differentiation tool. If engines do not understand these positions, the buyer cannot replace the product with free lookups. It turns the product into the only guide through terrain where free tools are said to be blind. That is effective sales copy — and dated sales copy. I do not blame the author. I check myself first: if I write a script built on a false premise about technology, how long before my audience loses faith? For me, the answer is one screening. For an opening product, the answer is one theory cycle — a few months — before a well-prepared White player finds a way to neutralise the line. Let me be precise about my own argument. 9...a5 is not useless. It is a legitimate branch. It can produce balanced positions. But balance is a claim about the position, not a claim about advantage. And at elite level, balance does not sell. People buy advantage, or they buy safety. This product is selling safety to one specific group: players who do not want to memorise. That group is real. I meet them every week at chess clubs around Hai Phong and on online servers. They have day jobs, they play in the evening, their ratings sit somewhere around the strong amateur band, and they hate being dragged into a memorisation race they have no time to run. For them, a slower position with fewer forced lines is a gift. And this is precisely where I want to turn to the contrarian angle. The most suspicious thing in this entire document is the part I most want to defend: the Samarkand headline. When a document splices a mismatched headline onto a body, there are two possibilities. The first is that the product is at fault. The second is that the aggregation is at fault and the product is innocent. I do not have enough data to adjudicate. In either case, I can assert one thing: this document cannot be used as a source, regardless of whether the product inside it is good or bad. This is where I must apply the same standard to myself. If I demand ACPL, engine match rate, database statistics and an author's name from a disc seller, I must accept that my own verdict on this product sits at medium confidence, not certainty. I have no game proving 9...a5 is weak, and none proving it is strong. What I have is a file with holes, and the holes are themselves the finding. Zooming out, this is a market story. Opening theory is one of the most stable and least glamorous revenue streams in chess. People sell answers to fashionable lines. When the Bayonet becomes common in open tournaments, demand appears for a counter. Products appear to fill it. That is the ordinary law of a small market, not a conspiracy. But this market has a peculiar competitor: free databases and free engines. Any player with a laptop and a connection can rebuild the entire 9...a5 tree in one evening. A product survives only if it sells what the database does not: a structure, a hierarchy, a story. That is why the “engines do not understand this” argument matters so much to the seller. It does not describe engines. It describes the product's competitive position. For Vietnamese players, that distance is shorter than people assume. A high-school student in a provincial town can reach the same database as a grandmaster in Europe. That means domestic buyers already have the tools to verify before spending. It is an advantage, and a responsibility. If I had to compress this entire analysis into one action, it would be this: before buying, open a database, count the games played from the position after 9.b4 a5, look at results by rating band, then ask yourself whether you can live in that position for three hours on a clock. The whole process costs one evening. It is far cheaper than one bad purchase. And if I had to compress it into one sentence about the industry: opening-theory products will not die because engines get better. They will die because buyers learn to verify. I have taken my share of criticism for writing drily. People say I cover sport like an inventory clerk. I keep writing that way, because I have seen what happens when a script is built on emotion without a data annex. It collapses at the first rebuttal. Chess is the same. A move can be sold with prose, but it only stands with data. And data, unlike prose, does not depend on which side the reader is on. One detail in this file keeps returning to me. Fourteen information points. Not one mentions a specific game. A product that teaches people to play chess, with not a single game inside it. From a documentary maker's perspective, that is the clearest sign the product was assembled from marketing copy rather than field material. Footage does not lie — but here there is no footage to trust. So what would change my mind? An author's name. A verifiable title. Three sample games with opponents, events and years. An ACPL table for each line taught. Database statistics with an explicit retrieval date. That is a short list, and it asks for nothing extraordinary. Until those appear, 9...a5 remains a legitimate sideline worth studying, and entirely unworthy of belief as an antidote. The distance between those two statements is the distance between a chess game and a flyer. And if this line ever does appear in top-level games, I will be the first to sit down and rewind the footage to count what changed. That is the only way I know to read a move: not by reading the copy about it, but by reading the move itself.