FRITZ 20: ChessBase's Revolution Without Numbers

I once misread a player's name, then read the position correctly the moment the replay crawled across the screen. But it took me nearly twenty years to recognise a different kind of error: trusting the blurb printed on the lid of a software box. This week ChessBase announced FRITZ 20. The release calls it a training revolution, a personal chess trainer, built for players taking their first steps and, at the same time, for tournament-level and professional players. I read the whole thing. Not one ACPL figure. Not one price. Not one release date. Not one comparison against Stockfish. Seven paragraphs. Not a single number. To understand why that silence matters, you need to remember what Fritz once was. ChessBase was founded in 2026 in Hamburg, Germany, by Frederic Friedel and Matthias Wüllenweber. Fritz arrived in the same decade and quickly became one of the strongest commercial engines. Fritz won the World Computer Chess Championship multiple times. In 2026, Deep Fritz drew 3-3 with Kramnik in Bahrain. In 2026, again in Bonn, Kramnik lost 2-4 to Deep Fritz, with the second game entering history as the first time a reigning world champion lost a classical game to a machine. I remember that night in Bonn in 2026. Kramnik missed a move, and Fritz delivered mate in one. The commentary room went silent. That was the moment a commercial engine crossed a line nobody thought it would cross that decade. Eighteen years later, every phone in every pocket does the same thing, for free. The landscape has changed completely. Stockfish is open source, free, and has topped most engine rating lists since around 2026. Leela Chess Zero runs on neural networks, also free. Lichess offers unlimited engine analysis at zero cost. Chess.com bundles analysis, tactics, a database and a community into one subscription. Engine strength has become a commodity, and the only thing left to sell is how humans use it. That is the market FRITZ 20 is walking into. Let me be precise about ACPL, because it is the only metric that means anything for a training tool. Average Centipawn Loss is the average error against the engine's best move, measured in hundredths of a pawn. Strong club players typically sit between 30 and 40. International Masters hover around 20. Super grandmasters stay under 15. If you want to know whether a piece of software can actually teach, look at whether it dares publish this number. The FRITZ 20 release has no ACPL. No move-match percentage. No description of mechanism. Only four words recycled through different phrasings: efficient, intelligent, individual, revolutionary. Then comes the contradiction. A product that claims to serve players taking their first steps while also serving tournament-level and professional players. In chess software, no product has ever done both ends well. Stockfish serves deep analysis and cannot teach a beginner how to trade pieces. Chessable sells opening courses to people who already read notation fluently. ChessKid is built for children. Nobody covers the whole span. That vagueness does not come from clumsiness. It comes from calculation. Chess is midway through a noise cycle. Football's transfer window has rumours; the chess market has product releases. Both teach the same lesson: when the source is the seller, information becomes noise. I have organised tournaments and I have sat down evaluating software for students. A good training tool has to answer three questions: what does it measure, what does it compare against, and where does it show me progress. FRITZ 20 answers none of them. The empty stadium that year, I found a tactical diagram lying still under a seat instead of the roar of the crowd. Now I find the same thing: what is valuable is never in the headline, it is in the part nobody bothers to write. ChessBase is not a weak player. Its database is one of the largest assets in the game: millions of annotated games, hand-edited opening collections, an interface three generations of players know by muscle memory. Fritz in my memory is a desk to work at more than a scoring machine. And yet the release does not sell that desk. It sells a slogan. From a competitive standpoint, the strategy gives something away: ChessBase knows engine strength is no longer an advantage. When free rivals have passed the threshold where the human eye cannot tell the difference, selling a stronger engine is selling an unverifiable promise. So they pivot to selling experience. But experience cannot be verified by data. And that is precisely the problem. ChessBase has sold boxed software since the 1990s. The buyer pays once, uses it for years, upgrades when they feel like it. That model is dying. Chess.com charges monthly. Chessable sells courses in packages. Lichess charges nothing. A box of FRITZ 20, whatever it costs, has to compete with an ecosystem users can walk away from at any moment. If I were the buyer, I would ask the three questions the release does not answer: how strong is the engine against the free version of Stockfish? What is new in the edited content compared with FRITZ 19? And what do I pay, once or monthly? One more detail worth noticing: the release carries no publication date. For a software product, a release date is the minimum. Without it, the reader cannot tell whether this is news or old advertising pushed back into circulation. Where could I be wrong? There is a more generous reading. Fritz's training revolution was never about the engine. Fritz became famous for an easy interface and the edited content that came with it: a game database, opening courses, a tool that lets many players analyse a position together remotely. That is human work first, algorithm second. If FRITZ 20 really is an edited-content platform, then publishing no ACPL is normal. Nobody measures a syllabus in centipawns. Mistakes are not what frightens me; kneeling down to pick the tape back up at two in the morning is what makes a forecaster. I read the release, I have not held the product. Judging a tool from a leaflet is exactly what I have scolded others for doing. But even on the most generous reading, one foothold does not hold. A company with forty years of history, with a data archive no free rival can copy, chooses to talk about a new product exactly like an app out raising venture capital. No price. No date. No specs. That says nothing about the product. It says a great deal about the pressure ChessBase is under. Here is a verifiable prediction: within twelve months, FRITZ 20 will appear as a subscription, or it will be accompanied by a technical disclosure far fuller than this one. If neither happens, I will correct myself, as I always do when I misread a player's name. And if you are weighing a purchase, the cheapest test is to load Lichess, run Stockfish, and measure your own ACPL across thirty games. It is free, and it is more honest than any slogan.