FRITZ 20 Knocks on the Training Room Door: The Chess Engine and a Revolution Without Data

In November 2026, in Bonn, Vladimir Kramnik sat for a very long time over a single move. The reigning world chess champion was losing to Deep Fritz — an engine running on ChessBase software — and in the sixth game he overlooked a mate in two. Kramnik lost the match 2-4. For the first time in history, a world champion had fallen to a computer under classical time controls. I have replayed that game more times than I can count, each time in a different mood. In 2026, I read it as a full stop. Years later, I understood it as an opening parenthesis. Last month, ChessBase announced FRITZ 20. The press release calls the product "a training revolution," a "personal chess trainer," aimed at both ambitious players and those taking their first steps on the board. After reading the whole document, I went back to look for the technical specifications. There were none. This article begins with a small detail: a company that once beat the world champion now speaks about itself only in adjectives. To anyone who has followed chess for more than two decades, Fritz is a familiar name. The first version appeared in 2026, developed by Frans Morsch and Mathias Feist, published by ChessBase — the German company founded by Frederic Friedel in 2026. In 2026, in Bahrain, Deep Fritz drew 4-4 with Garry Kasparov. In 2026, X3D Fritz drew 2-2 with Kasparov in New York. By 2026, Kramnik lost. Within fifteen years, the Fritz line moved from "formidable opponent" to "the machine that beat a world champion." Over the same stretch, something else was growing in silence: open source. In 2026, Stockfish — a descendant of the Glaurung line — was released for free. In 2026, Leela Chess Zero arrived, learning chess through the method AlphaZero had introduced. Both are free. Both are stronger than any commercial engine of their moment. Both run on an ordinary laptop. Then came the platforms. In December 2026, Chess.com completed its acquisition of the Play Magnus Group, in a deal reported at around 80 million dollars. Lichess remains free, open source, and serves millions of users every month. Chessable built its own market on spaced repetition for opening study. That landscape puts ChessBase in a difficult position. It no longer holds a monopoly on computing power. It no longer holds a monopoly on game data. And most importantly, it no longer holds a monopoly on reaching people who study chess. That is why I read the FRITZ 20 release with unusual attention. When an engine that was once a symbol of raw strength begins to define itself with the word "training," that is a market signal more than a product announcement. One notable detail is the contradiction in audience targeting. The release describes FRITZ 20 as being for "ambitious players and professionals," while also serving "players taking their first steps." These two groups have fundamentally different training needs. A player rated around 2,100 needs deep opening analysis tools, a strong grasp of structures, and the ability to cross-reference a large game database. A beginner needs tactical puzzles of graded difficulty, explanations of principle, and feedback delivered stage by stage. A product that embraces both usually sacrifices depth on one side. I have followed the chess software market since the 1990s, when a ChessBase CD was a valuable asset in Vietnamese chess clubs. Back then, software was the only route to top-level game data. Today, a twelve-year-old in Da Nang can open Lichess, launch Stockfish, and within thirty seconds hold a tool stronger than anything Kramnik faced in 2026. That shift changes the nature of the question. The issue is no longer how strong the engine is. The issue is what the engine helps a human understand. Computing strength has become a commodity. Stockfish is free. Leela Chess Zero is free. Positional metrics — centipawn, win probability, ACPL — have become a shared language displayed on any platform. When the foundation of strength is flattened, competitive advantage must come from somewhere else: pedagogical skill, lesson structure, and the learner's experience. ACPL — Average Centipawn Loss, the average loss measured in centipawns per move — is a common measure of move quality against the engine's suggestion. The metric is useful for review. It also carries a side effect rarely discussed: it teaches young players that the best move is the move that matches the machine. In Vietnam, I have sat behind young chess players during many training sessions. They open the engine after each game, glance at the error rate, nod, and move to the next game. Few pause to ask themselves: without the engine, what would I think at move twenty? When I watch these players train, I do not look at the screen; I look at how they keep their composure when the machine tells them they have just erred. Reacting to a machine's rejection is a skill, and it appears in no feature menu. Vietnamese women's chess has a notable generation: Pham Le Thao Nguyen, Vo Thi Kim Phung, Nguyen Thi Mai Hung, alongside a rising cohort of younger faces. Looking at the broader picture, the share of women among rated players worldwide remains low — a fact the global chess community has known for a long time and has not solved. Judit Polgar once entered the world's top 10 with a peak rating of 2,735, but she was the exception rather than the norm. Hou Yifan reached 2,686 and was the world's top-ranked woman for years. This matters to the FRITZ 20 story. If training tools determine progress, and if the best tools are free, then the greatest barrier is no longer cost. The barrier lies elsewhere: time, a mentor, and a competitive environment. A genuine training revolution would have to answer some hard questions. How do you diagnose a player's weaknesses without relying solely on game results? How do you build a training path that adapts over time? How do you measure progress in a meaningful way, rather than stopping at a rating number? The FRITZ 20 release touches none of them. ChessBase holds an advantage few rivals can copy: more than three decades of data. Its Mega Database contains millions of games, an asset younger platforms need years to accumulate. But raw data trains no one by itself. The value lies in how data is turned into lessons. That is the fragile boundary. A vast database can be an excellent research tool for someone who knows what to look for. For someone who does not, it is a room without a door. Good training software must do the work of a guide: it must show the learner which room to enter first. Tactics resemble people: you must know who you are before you try to become someone else. I read that line inside the FRITZ 20 release in a different way. ChessBase is trying to become something else — a training platform — while its old identity was the strongest machine in the analysis room. The transition is not wrong. But if the new positioning comes without new data, readers are left with promises alone. For someone learning chess in Vietnam, the more practical question is: what does it take to improve in the first two years? My answer, after years of observation, revolves around a few very old things: playing many controlled games, having someone point out repeated mistakes, and having a peer group of similar strength to measure against. A computer serves all three, but none of them is something a computer does on a human's behalf. A machine can tell you that your move 17 was inaccurate. It cannot tell you why you chose that move, or what you should ask yourself next time. This brings me back to a memory from 2026. When over-the-board chess tournaments were postponed en masse, I spent many weeks sitting in silent training rooms, documenting sessions without spectators, without clocks, without scoreboards. The empty corridor of 2026 taught me that the board never dies, it only knows how to wait. When the cheering stopped, I could finally hear the true breathing of the game. And that breathing did not come from the engine. It came from the person sitting in front of it. The counter-intuitive angle sits here: FRITZ 20 calling itself a "training revolution" reflects a business model being inverted more than a technical leap forward. When computing strength becomes free, value migrates to people — the coach, the club environment, and a competitive system dense enough for young players to clash in. For women's chess in Vietnam, this carries concrete meaning. The gateway into the discipline is no longer a piece of software costing a few dozen dollars. The real gateway is a teacher who knows how to ask the right question, a regular tournament schedule, and a space where girls do not feel like a minority in the playing hall. A deeper paradox lies in data. The more machine analysis there is, the less stylistic diversity there is. When every young player learns from the same engine, the same database, the same metric, playing styles tend to converge on a single point. Distinction — the thing that makes games people still remember — must come from elsewhere: from how a player accepts risk, from how they read an opponent, from choices that are not optimal but are uncomfortable to face. That is why I always tell the young players I get to talk with: use the machine to check, not to decide. Checking expands understanding. Deciding narrows it. There is another dimension of the story that matters for the Vietnamese market. Cost. A training software package priced between a few dozen and a few hundred dollars is a serious investment for many families. When that investment is not accompanied by verifiable data, the buyer is paying for belief. Belief can be justified. But it needs to be fed with evidence. Over many years in this profession, I have learned that the way an organisation speaks about its product often reveals more than the product itself. When the specifications section is missing, when the target audience stretches from beginners to professionals, when the word "revolution" appears without a definition attached — that is the moment to ask questions, not the moment to open a wallet. None of this is meant to dismiss FRITZ 20. The product may be good. ChessBase has three decades of experience and a team that understands chess deeply. The problem lies in the gap between claim and evidence, and that gap is where consumers are most vulnerable. What I carry away after reading the FRITZ 20 release has little to do with FRITZ 20. It has to do with the fact that a fifteen-year-old girl in Da Nang today holds — for free — analytical power that Vladimir Kramnik had to respect in 2026. That revolution has already happened. It simply came without a press release. If there is a question worth carrying into the coming season, it is this: when the tools are equal, what creates the difference between those who go far and those who stop? The answer, from what I have observed, lies in the number of hours someone is willing to sit down again after the engine has been switched off.