A Tiebreak Crown, an Engine Advertisement: The Ledger Missing from Fritz 20's Training Claim

Last week I read a piece twice. The headline carried the Bavarian Championship; the headline carried Sasikiran, champion on tiebreak, one man out of four players on the same score. Forty-seven years of reading chess results means my eye goes automatically to the position that decided the tiebreak arithmetic. The piece ended. There was no move, no diagram, not even the names of the four players who had tied. The crown in the headline is real; the square beneath it is empty, and into that square has been placed an advertisement for a piece of software called Fritz 20.

Selling software is fine; the game loses nothing by it. My objection is to the accounting. Every training claim made in chess over the past two decades has rested on at least one measurable number: engine match rate, average centipawn loss, results in a specific opening, a user's rating gain. This piece contains no number other than the tiebreak in its headline; the rest is adjectives. Borrowing the result of a regional event and slipping a product inside it is not a new manoeuvre, and it is the most familiar trap in chess journalism today.

Fritz's own history explains where it came from. In 2026, in game two of Deep Fritz's match against Vladimir Kramnik, a mate in one appeared on the board; the human lost the match, and engine strength was news. Once Stockfish and Leela Chess Zero were released as free open source, that market for strength collapsed. Anyone can now run an engine at maximum depth on a phone for nothing. The only survival route for a commercial engine is to step away from strength and move toward interface, curriculum and care. Fritz 20 is making exactly that move, and it is not weakness but market reality. The trouble begins when the relocation itself is offered as the proof.

The Bangladeshi context is unavoidable here, because we do not lack engines. Five grandmasters, the last one in 2026; roughly eleven lakh taka a year in government support for the federation; the school chess pipeline shut down in 2026. When the Premier Division was suspended in 2026 my coaching stipend went to zero, and I filled the emptiness with three hundred hours of engine work and Leonine Chess Club's online events. Twelve viewers watched the first live broadcast. Those twelve taught me what no advertisement teaches: the shortage is not of tools, it is of audience. In Dhaka's Premier Division an Indian guest grandmaster is brought in for board one, and the club itself builds his preparation file. So the question is direct: will a new training program produce our sixth grandmaster?

Fritz 20 makes four claims. First, it is a particularly difficult opponent. Second, it is a strong ally. Third, it is more than a chess engine. Fourth, it is a training revolution. Each requires a different kind of evidence, and none of the four appears in the article.

A difficult opponent is proven by engine-versus-engine match rate. Whether an engine genuinely becomes humanly complex when its playing level is lowered can be measured by its score against rival engines, by average game length, by results in fixed opening sets. A strong ally is proven by user outcomes: how far average centipawn loss fell in three months, how much the hit rate on repertoire answers rose. A coach needs a week to build those two tables. The article has not one cell.

More than an engine is the third claim, and the only verifiable one, which is where my interest lies. Being a trainer does not mean calculating faster; it means deciding which forty positions get studied. That is editorial work, not computational work. In 2026 I accumulated twelve hundred positions, sifting roughly fourteen hundred games of Bangladesh's five grandmasters, and began dividing the board into sixteen zones and counting pawn influence in each. Arguments about space end under that method, because 'Black controls nine of sixteen zones' is a sentence no editor can cut. But the engine did not build that zone map; I did. An engine never decides which question to ask, it only answers.

Revolution means comparison with a previous state. What the previous state was, who measured it, what percentage of users gained how many rating points — not one sentence. In 2026, just before the Chennai Olympiad, I built a sixty-page preparation file on a 2540-rated Indian grandmaster for a club: fourteen opening trees, every game since 2026. The file was flawless. I missed the registration deadline. He arrived two days late, played four games, scored 2.5. A preparation tool and a preparation calendar are two different objects, and software owns only the first.

That is the hidden gap in the promotional structure. It always sells one assumption: the tool is the absent coach. Bangladesh does not lack a coach; it lacks a calendar. A schoolboy with a phone already holds a deep engine; he lacks a person who says these forty positions and not those forty. Fritz 20 wants to occupy that person's place, yet what that person actually does is not described anywhere in the advertisement.

The second gap is more uncomfortable. Personalised engine training naturally pushes a player toward the engine's taste — precise, safe, and often wrong. Annotating forty-one of Ziaur Rahman's games last year I kept seeing the same thing: moves the engine calls second-best that were right on that board, at that clock, against that opponent. His fifteen national titles came from decisions a modern evaluation would call inaccurate. If a training program teaches only the engine's taste, it is not a trainer; it is a machine for producing sameness.

So what I want to see in the coming months is specific. One, Fritz 20 should publish engine match data and user progress figures; a release that never gives a number is a slogan. Two, somebody in Dhaka should build a three-hundred-position curriculum for the generation after Manon Reja Neer, because no company will build it. Three, at the next national championship I want to see whether a sixth name reaches our list of five grandmasters. The Bavarian tiebreak is already history; the question is which square of our chessboard was actually filled by an advertisement that borrowed its headline.