9...a5 Against the Bayonet: One DVD, Ten Claims, Zero Numbers

In an academy room in Hyderabad a fourteen-year-old had just played 9.b4. The coach opposite him said, without pausing, ‘9...a5, it is in the new DVD.’ My laptop was open with the same product description he was quoting. I read the whole text three times. Ten claims. Zero numbers.

No engine evaluation. No game count. No named model game. No presenter surname. No edition of the Olympiad whose games are supposedly annotated. Yet the coach was carrying that product into a classroom with total confidence, because in his frame ‘it is in the DVD’ was itself the evidence.

Ten claims, zero numbers — that sentence is not a discovery from one afternoon. It is the summary of six years of work. I built the dataset first and let the story arrive late, and that habit now bothers me even when the thing being sold is not a football match but a chess preparation.

Let me declare the stake first, because without it this piece is only an audit. In November 2026, in a Kolkata press box, a senior reporter told me chess tactics were ‘not a women’s beat’. I answered with a spreadsheet — centipawn loss and clock usage across 45 rapid games, Anand included. Since then every piece of mine carries a number, a source and a date. What I want to protect here is not a product. It is the basis on which a teenager chooses his opening preparation. If that basis becomes ‘it is in the DVD’, then the next time the same coach will tell the same boy that the engine is wrong and the book is right, and the boy will not have the language to ask.

Before the match there was a file nobody had opened. This is an attempt to open it.

Mar del Plata: a name, a date, a closed room

The King’s Indian Defence begins plainly: 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. Black has conceded central space, struck at the centre with ...e5, and is preparing ...f5. White has closed the centre with d5 and is looking for expansion on the queenside. In the classical Mar del Plata the two sides attack in opposite directions — White with c4-c5 and later b4-b5, Black with f5-f4 and ...g5-g4.

The variation’s name is tied to the Argentine city and the 2026 tournament where the structure appeared repeatedly. I hold that as naming convention, not as a photocopy of an original tournament bulletin, and I want that distinction visible. What I do hold as a document is the ECO classification: the Bayonet Attack 9.b4 and its sub-branches sit in the E97–E99 band, with E97 touching the Bayonet, E98 the 9.Ne1 lines and E99 the 9.Nd2 lines.

The Bayonet, 9.b4, became White’s main weapon over the last two decades for a simple reason: White wants to advance the queenside majority and build a b5 clamp so that when Black’s ...f5-f4 storm arrives, White has a second front open. Black who only plays on the kingside loses the queenside for free. That is why the Bayonet is the modern main road, and why every answer to it keeps being re-examined.

Black has two clear paths. The first is 9...Nh5 — the main line, with the knight heading to h5 to support ...f5 and ...Nf4; the position sharpens and both sides play concrete, forcing moves. The second is 9...a5, which the description calls the antidote. It claims queenside space, prevents White’s b5 clamp, and keeps ...Nd7, ...b6 or ...Nc5 in reserve. 9...a5 is not a novelty — it has appeared in practice before, and my unverified reading is that the product’s value lies in consolidation and explanation rather than discovery.

The six-link argument

Reading the description I found a clean chain of reasoning, and it has to be read as a chain, because every link sounds individually true while the whole thing hangs on one weak link.

Link one: the Mar del Plata produces mutual attacks and long-term strategic positions. True. Link two: computers do not understand these positions well. That was half true from 2026 to 2026 and is less true after 2026. Link three: therefore the human player benefits. This depends on link two and leans with it. Link four: but the fashionable Bayonet disrupts that comfort. True — the Bayonet became popular precisely because it slows Black’s kingside attack. Link five: and the main reply 9...Nh5 collapses back into forced, computer-friendly lines. Here the description concedes that after 9...Nh5 ‘computer analysis becomes important again’ — an admission that engine mastery is the norm and 9...a5 is the escape from it. Link six: therefore take 9...a5 and stay in the human domain.

The argument is internally coherent and it genuinely captures a real tension in King’s Indian practice — forcing, engine-checkable lines versus strategically opaque manoeuvring. The problem is not that the reasoning is wrong. The problem is that not one of the six links is tied to a number.

What the description discloses, and what it does not

I made two columns on one page. On the left, everything present: the variation’s name, the identity of the Bayonet, the recommendation of 9...a5, the ‘strategy, not tactics’ promise, the pairing of ‘objectively equal but dangerous for White’, and the Olympiad in the title. On the right, everything absent: the presenter’s name and title, the publisher, video runtime, the number and list of model games, database size, engine-corpus statistics, any single evaluation, any frequency percentage, any results split, and the specific Olympiad edition the games come from.

The right column came out longer than the left. That is the most honest picture of this entire analysis.

One caveat matters: every technical claim here comes from marketing copy, not from independent review or database output. That distinction matters because in opening literature the function of marketing copy is to sell, and its grammar is different. I am not blaming that grammar. I am recording that when the same grammar reaches a coaching desk, it turns into evidence.

The audit: six metrics that could have been measured

Six metrics exist for any opening product, all of them publicly derivable, and not one of them is here. Engine match rate — the share of moves identical to the engine’s first choice. ACPL — average centipawn loss per move. Database frequency of 9...a5 within Bayonet games. The results split for that frequency, meaning Black’s score with 9...a5 versus 9...Nh5. The list of model games — which games, which event, which year. The presenter’s credentials — title, peak rating, and whether they have played this line in serious classical chess.

Zero of six.

That does not mean the product is bad. It means every technical judgement about it has to be carried at reduced confidence. The most solid thing I have is not the product but the structure: 9...a5 is a known alternative, 9...Nh5 is the main line, and the description admits it. That admission limits any claim to novelty — and, honestly, it is the most trustworthy sentence in the entire text.

The NNUE date stamp

The weakest technical link is the claim that computers cannot handle these complex strategic positions. That sentence was roughly true every day between 2026 and 2026. Since 2026 it is less true. Modern engines run on NNUE evaluation — neural-network assessment, materially stronger than older engines in closed, strategic positions. The old engine stalled near zero in blocked positions; the new one finds planning-level differences in the same position. A dated heuristic is not fresh evidence.

This does not destroy the product; human judgement still matters in strategic positions. But it ages the central selling sentence. And when a product’s main argument is aged, the buyer has at least the right to ask on which engine version, at what depth, on what date the line was assessed.

Two statements in one paragraph

Here is what stings most. The description collides with its own logic. On one side, computers cannot understand these positions. On the other, after 9...Nh5 ‘computer analysis becomes important again’.

Put together, the meaning is: the real sale is not that engines are weak, but that engine grip loosens in one specific branch while staying firm elsewhere. That may well be true. But it gets converted into a universal claim, and then it outruns the evidence. Analysis says: at this node, to this depth, the evaluation is unstable. Marketing says: computers just do not understand. The first sentence is useful. The second sells. They can share a page; they are not the same truth.

One branch of a large tree

The most practical point, and the one the description never touches: the product addresses exactly one White branch, 9.b4. In the Mar del Plata, White retains fully independent options on move nine — 9.Ne1 and 9.Nd2 remain popular and create different structures — plus earlier deviations such as 8.Be3 and 8.Re1. A Black player adopting the Mar del Plata still needs answers to all of them.

That is not a flaw in the product. It is a flaw in the perception the sale creates: ‘the antidote to the Bayonet’ and ‘a Mar del Plata repertoire’ are different objects, and conflating them is the mistake repeated every opening season. Buy it as a supplement, not as a substitute. If the buyer already owns a complete King’s Indian repertoire, it sits exactly where it should — as the consolidation of one node.

Olympiad model games: good source, biased selection

The Olympiad is a good source: a biennial team championship with most of the world’s top hundred players and roughly 180–190 national teams, fully public games. I hold that figure as general knowledge, not as an official entry list, so treat it as pending verification. But team events carry a selection bias. In a team match a draw is sometimes sufficient, so play becomes result-oriented and less critical than at a Candidates or a pure round-robin. A repertoire built mainly on team-event games should be cross-checked against individual elite events.

And there is a hard gap: no game, round, player or year is named anywhere. The Olympiad here is not an analytical instrument. It is a prestige label.

The name problem: KIA versus KID

The title says King’s Indian Attack; the moves given are King’s Indian Defence. The Attack is a distinct system built on 1.Nf3, 2.g3, 3.Bg2 and d3. The Mar del Plata begins 1.d4 Nf6 2.c4 g6. The names look alike and the products are not the same. In a search-driven market that mislabelling pulls the wrong buyer — a King’s Indian Attack player — and confuses the right one, the Defence player hunting an answer to the Bayonet. It is correctable in cataloguing and tagging, and until it is corrected it charges interest on the product’s credibility.

Equal and dangerous: two promises, no evidence

The description says the position is objectively equal but dangerous for White. That pairing is a known device in opening literature: it calms the buyer’s fear that the line is refuted while promising practical upside. Both can be true. Here neither is evidenced — no evaluation, no results split, no named model game. When a product promises in two opposite directions and supplies a number for neither, it is reassurance-plus-upside packaging, not analysis. I will not call that deception. I will call it unsubstantiated, and that is enough.

Fewer forcing lines does not mean less work — it means more

Here the problem stops being technical and becomes psychological. The promise is fewer concrete lines, understanding instead of memorisation, aimed at the player who prefers to think. That is a real and underserved segment, and I accept it fully. But the same description also states that Mar del Plata positions demand long-term planning and tactical sharpness, and that second sentence cuts the first. Where moves are not forced, the demand on judgement rises. In a forcing line the player only has to recall the next move. In a non-forcing position he must choose, every move, which plan is right, when to switch fronts, and where to accept a simplification under clock pressure.

For a club player that difference is decisive. The player who wants less memorisation is often the player who bleeds advantage fastest in a complex, opaque position. The promise and the audience walk away from each other. ‘Easy’ and ‘non-forcing’ are not the same thing, and treating them as one manufactures a disappointment that begins within weeks of purchase. That is the most under-priced risk here, and it is commercial, not technical.

The limit of my own audit

Marketing copy is not a database, and reading it as one is a mistake. The comparison class for an opening product is not a scientific paper; it is a product description. Almost none of the descriptions I have read in recent years — complete King’s Indian repertoires or single-branch modules — print an engine evaluation, a results split or an ACPL. This product is not unusual within its category; the category is the limit.

So my finding is not ‘this DVD is bad’. My finding is that the whole genre of opening material has no evidentiary standard, and that by the time a marketing sentence reaches a coaching desk it has become proof.

And because balance is one of my own rules: what checks out deserves equal prominence. The chess foundation here is sound — 9...a5 is a real, playable alternative, 9...Nh5 is the main line, the Mar del Plata genuinely carries strategic depth, and the ‘strategy versus memorisation’ segmentation is real. In the PGN-and-clock archive I have kept since 2026, the recurring pattern is that lower-rated players lose the most time in complex positions — and that is precisely this product’s likely buyer. The problem is not the line. It is the claims made about the line.

Every contested server leaves a timestamp, if you wait long enough. Here the timestamp is missing, and that is the story.

What to watch next

I will not predict whether this line reaches elite practice, and I will not name the next world champion. My job is to keep the ledger, not to prophesy. Six signals I will be recording.

First, current NNUE evaluation of the critical 9...a5 positions and whether it holds as depth increases; a practically meaningful drift would undermine the core claim. Second, elite uptake — a 2650+ classical appearance with a good Black result would refresh the line’s standing. Third, White’s move-nine preference: if the database share shifts from 9.b4 toward 9.Ne1 and 9.Nd2, the addressable audience shrinks. Fourth, publisher disclosure — presenter name and title, runtime, model-game list, database inclusion. Fifth, the next Olympiad cycle and its game corpus, the raw material for this category. Sixth, competing repertoire releases, which would compress this product’s market window directly.

One question remains, and it belongs to that fourteen-year-old. The next time a coach says ‘it is in the DVD’, will he have the words to ask who assessed that line, when, and to what depth? The day that question becomes ordinary in a classroom, opening products will have to ship with numbers.**

---

Methodology note — Sample: ten information points, all from a single product description. Source: marketing copy, not independent review or database output. Limitation: no engine evaluation, game count, results percentage or named model game was available; presenter name, title, publisher and Olympiad edition are unverified. All technical conclusions are therefore preliminary and should be checked against a current database and engine before use in serious play. This article is not a purchase recommendation and not betting advice of any kind.