The Sämisch KID Fog: Wei Yi's 'Masterpiece' and the Arithmetic of Opening Labels

I watched from the tribune, then went back to measure what the eyes missed. That sentence is the rule of my trade, and last week, when I opened the preview of a weekly chess teaching video — the title said 'Sämisch King's Indian Defense,' the description said 'Keymer System' — I realised that this time the measuring would have to happen on the label before the moves.

The frame that formed on screen

Wei Yi, Olympiad eighth round, playing Black. The video claimed Wei Yi kept the queenside closed, absorbed White's attack, then counter-attacked. Commentator Robert Ris called it 'very precise play.' A second model game — Mendonca vs Praggnanandhaa — was also presented from Black's perspective. This is one episode of a weekly series called 'Fast and Furious,' meaning the product is subscription-based, not news.

But when I placed the title and the description side by side, two different opening frameworks stood before me. The title says: 1.d4 Nf6 2.c4 g6 3.Nc3 Bg7 4.e4 d6 5.f3 — the Sämisch King's Indian. The description says: 1.Nf3 d5 2.e3 — the Keymer System. These are not the same opening. This contradiction is the single biggest technical flag on the entire product, and any review that ignores it is selling the buyer the wrong product.

The ledger of an opening label

Chess games carry no copyright — which is good, minimal legal friction. But a teaching product keeps its own ledger on the label. When you write 'Sämisch KID,' the buyer comes looking for the f3 structure: White's big pawn centre, the kingside storm, and Black's question — close the queenside, absorb the first wave, then open the side where White is weak. That is a classical motif, mainstream elite understanding, not a new discovery.

Now look at the other side. The 'Keymer System' — 1.Nf3 d5 2.e3 — is a flexible, non-committal White scheme. It can transpose into many structures, and sometimes does. So only two paths remain. First: the video's title is selling the broader 'Sämisch theme,' while the Wei Yi game is a separate Keymer-System example used to illustrate the same 'attack versus defence' balance. Second: the game began with a 1.Nf3 move order and transposed into a Sämisch-type structure after ...d5/...g6/...Bg7. Both are inferences, not facts. And in chess journalism, selling an inference as a fact is the same sin administrators commit with budget figures.

What can be measured, and what cannot

Here I recall an old habit. In 2026 in Chennai I watched Anand–Carlsen from the tribune, then came home and counted heart rates and hours of sleep. That exercise taught me that every match analysis needs a 'cost line' attached. Here that line is blank. The Stage-1 output has no move list, no ACPL, no engine match-rate, no result margin, no move count. 'Very precise' is a commentator's opinion, not an engine-verified measurement. Teaching products routinely inflate precision for pedagogical effect — that is standard industry practice, and I won't call it a crime, but I won't call it a measurement either.

Wei Yi's actual opponent in the eighth round is not stated. 'Against the Keymer System' does not mean Keymer was sitting across the board. Many people misread this distinction — the name of a system is not the name of a player. Without the opponent, preparation-targeting analysis is impossible. Without ratings, a form signal is impossible. And most importantly: teaching products select games precisely because they are exceptional. This 'selection bias' is structural, not incidental. Two hand-picked Black wins cannot be proof of a player's general level.

The contrarian angle: the ledger of attack versus defence

Now the part where I am in the minority. The video's theme is 'the balance of attack and defence' — that is, Black's question: when to close the queenside, when to open the other wing. It is a good question. But two 'model games' are not enough to teach it. I remember in 2026 I sent Sagar Shah a nine-page proposal — 'the space ledger': control percentages for all 64 squares, sampled at moves 15, 25 and 35, cross-checked against pawn-lever timing. Eleven pieces ran, including a National Premier study in which the champion's average controlled squares climbed from 31 to 38 between moves 20 and 40. That is my method: table first, prose second. Here there is no table. Only a story of how Black absorbed and struck back.

The problem is that 'model games' in closed positions teach what worked — not when it will fail. Decision rules are needed: under what conditions should the queenside be closed, and under what conditions not. Two games are not a sample of that rule, only illustrations. And teaching a lesson on a closed, absorption-based position under the 'Fast and Furious' branding is a stylistic mismatch — a marketing matter, not a chess-analytical one.

Forward verification

I watched from the tribune, then went back to measure what the eyes missed. This time what was missed was not in the 64 squares but between the title and the description. Before the next review, three ledgers should be demanded: the game's move list (did a transposition occur?), ACPL or engine match-rate (can 'precise' be measured?), and the opponent's name (is preparation-targeting analysis possible?). Until those three arrive, the word 'masterpiece' remains a curator's opinion — and I enter opinions in the ledger, not in the shrine.

Next time a teaching video sells an 'Olympiad masterpiece,' ask the first question: is this Sämisch KID or the Keymer System? Because a product that does not know its own opening cannot teach you to close the queenside.