ChessBase's FRITZ 20: A Launch Pitch With Not a Single Benchmark

I replayed the SEA Games 29 tapes to find what people had left behind. Kuala Lumpur, August 2026. Nguyen Thi Oanh entered the final lap of the 3000m steeplechase, and I sat there counting her stride rate frame by frame: 47 per minute, while most of the field had already dropped below 42. Someone in that meeting said women do not understand running, so just write it with feeling. I did not argue. I counted, and then I put the counting sheet on the table. That method followed me to another arena: the chessboard. This week I opened ChessBase's product page for FRITZ 20 and read the entire description. Seven lines. Seven lines saying this is a personal chess trainer, a revolution in training, built both for players taking their first steps and for professionals. I looked for one measurement. There was none. A tool sold as a training solution, with no benchmark a buyer could check. The position of the source matters before anything else. ChessBase is a German chess software company, founded in 2026 in Hamburg, built first on game databases and then on the Fritz engine line developed by Frans Morsch and Mathias Feist. Fritz produced milestones that can be verified: in 2026 in Bahrain, Deep Fritz drew world champion Vladimir Kramnik 4-4; four years later in Bonn, Deep Fritz beat Kramnik 4-2. Every milestone like that has a date, a place, a score, a transcript. The FRITZ 20 pitch is not that kind of document. It is issued by the manufacturer itself, meaning the source holds a direct commercial stake in the outcome, and it carries no publication date. In documentary work, such a source is still usable, but only in the right way: read it as a leaflet, then go find the data the leaflet left out. The chess software market restructured long ago. Stockfish has topped most engine rating lists since the mid-2010s and ships free as open source. Leela Chess Zero runs on a neural network, following the AlphaZero path. Houdini was once a strong commercial engine, then ran into controversy in 2026 when its author acknowledged using Stockfish code. At user level, Lichess offers free analysis, Chess.com folds everything into one platform, and Chessable sells structured opening courses. A paid software package today has to answer a very specific question: what does it do that the free tools cannot? The FRITZ 20 description does not answer that question. It says the product is for players already competing at tournament level, and also for players taking their first steps. Those two groups have almost opposite needs: beginners need a curriculum, graded exercises and readable feedback; tournament players need opening-repertoire matching, position evaluation, game database management and opponent scouting. A product that embraces both usually sacrifices depth at both ends. What a training tool must be measured by If a track coach told me he makes his athletes faster, I would ask three things: the 100m time before and after, the number of sessions per week, and whether any of it was filmed. Nobody would be asked to prove anything beyond those three. Chess software has an equivalent trio. A training tool claiming to improve the learner must show ACPL, the average centipawn loss per move, falling across a defined set of practice games, measured on a fixed position set that does not change between runs. Alongside that comes the engine match rate, the share of moves matching the engine's top suggestion, measured over the same game set. And then the conditions: which machine, which time control, which platform, what price per year. Those four facts together are enough for a buyer in Hai Phong to decide whether to spend money. The FRITZ 20 pitch contains none of them. If FRITZ 20 really is a training machine, the description should list its components: a themed exercise library, a scoring system after each session, multi-week plan generation, an opening-repertoire matcher, play modes against the engine at various levels, and desktop-to-phone synchronisation. The phrase training revolution describes none of those components. In documentary work I call such lines narration without pictures, and they always get cut in the final edit. I trust footage over testimony, because footage does not lie. A single frame tells me whether the athlete led with the wrong leg over the hurdle; an analysis file tells me which move number the player got wrong. A sales pitch tells me nothing except what the seller wants me to believe. Engine strength stopped being a selling point For more than twenty years, the Fritz line lived on computing power. That era is over. When Stockfish is free and stronger than any commercial engine at most time controls, selling a pure engine is like selling a stopwatch to someone who already owns a phone. Nobody pays for what they can download in three minutes. So ChessBase moving Fritz toward a training-tool position is a logical, even mandatory step. The hard part lies elsewhere: training is not a feature, it is a process. A process has inputs, an order of exercises, a check after each cycle, and data showing where the learner stands. The bottleneck in modern chess software is turning free computing power into a repeatable practice habit, not calculating a few percent deeper. The FRITZ 20 pitch uses exactly the vocabulary of that bottleneck, but it does not describe a process. In Vietnam, the gap is not the engine Based on my experience tracking matches and youth events I have filmed, the Vietnamese picture differs considerably from what a European advertising campaign imagines. At chess clubs in Hai Phong, Hanoi or Ho Chi Minh City, twelve-year-olds already run Stockfish on their phones, use free Lichess analysis, and know how to review their own games. What they usually lack is not an engine. It is a coach sitting beside them, able to read each child's games and assign the right work for the following week. Chess lessons are already a meaningful expense for many families, so every paid piece of software has to prove its value through visible results. Looking at the generation that came through, Le Quang Liem won the World Blitz Championship in 2026 in Khanty-Mansiysk and has crossed 2700 Elo; Nguyen Ngoc Truong Son has crossed that mark too. Both grew up when chess software reached Vietnam mainly through unlicensed copies and translated books. In other words, their results came from training discipline and specific coaches, not from owning an expensive software package. On the women's side, Pham Le Thao Nguyen and Vo Thi Kim Phung are among the players who have delivered international results for Vietnamese chess, and notably they are still asked questions their male colleagues never face. In 2026, while making a film about Quach Thi Lan, I heard the line girls should not run hurdles again, and realised that in chess the same prejudice simply wears different clothes. The only answer I know is to measure performance, measure technique, measure the training process, and write not a single word about appearance. I have kept that rule ever since, including when writing about a software package. The counter-argument Here I have to challenge myself. My demand for ACPL and engine match rate is the reflex of a documentary maker, not necessarily the buying standard of the mainstream. A parent in Hai Phong does not buy ACPL for their child. They buy a clear curriculum, a readable interface, and the feeling that the child will sit at the board more consistently. Seen that way, ChessBase's strategy may be right: sell the learning experience, not engine strength. The mistake lies in the language. The same product cannot promise beginners a revolution and promise professionals a personal trainer, while offering that second group no specification at all. Professionals are the hardest customers in this market: they have Stockfish, they have their own databases, they can measure everything themselves. To sell to them, you need data. To sell to beginners, you need a curriculum. A leaflet claiming both usually sells to neither. I also set a test for myself: if an independent party, unconnected to the manufacturer, ran FRITZ 20 on an ordinary laptop in Vietnam, analysed the same 200 positions, compared it against Stockfish and published the results with its method, I would rewrite this piece. I wait for data, not for inspiration. We need a shared yardstick I have run hurdles all my life, I have just never been timed. I say that about myself, but it applies to most young Vietnamese players training every evening with nobody measuring them. Every documentary is a running track: the audience sees the finish line, I live at every starting block. With training software, I want to stand at the starting block a little longer, long enough to put a yardstick on the table that anyone can check again. What I am waiting for is not an update. It is a table of numbers.