FRITZ 20 and the Trust Problem When Chess Engines Have Become Commodities

During a commentary session at a Central Vietnam youth tournament in Da Nang, I sat next to a parent. She opened her phone and showed me everything she had bought for her son: two analysis apps, a tactics library, an endgame reference tool, and a boxed chess software package. She asked me a question that left me silent for a few seconds: “Which one actually teaches my child?”

I looked at the board beside us. A fifteen-year-old boy was bent over move twenty-seven, his hand resting lightly on the knight. He did not need more software. He needed someone who knew what they were talking about. The mother was not asking me which program was the strongest. She was asking which one was the most honest.

That question haunted me through the whole tournament. It returned once more when I read the ChessBase press release about FRITZ 20, the latest product in a Fritz line that has been part of chess culture for more than thirty years.

ChessBase is a German chess software developer, the company behind Fritz since the late 1980s. The Fritz line entered history when Deep Fritz versions faced Garry Kasparov in 2026 and Vladimir Kramnik in 2026 in man-versus-machine matches. Back then, a commercial engine beating a world champion forced the entire chess world to follow every move.

The FRITZ 20 release I read described the product in beautiful phrases: “a training revolution,” “a personal chess trainer,” “efficient, intelligent and individualized training.” The stated audience ranged so widely that it was suspicious: from “players taking their first steps” to “ambitious players and professionals.” A product aimed at everyone, by market logic, tends to be a product aimed at no one.

What I searched for and could not find matters more. No engine Elo rating. No calculation speed per second. No match rate against leading engines. No release date. No feature comparison against competitors. No price. No independent test cited.

In the commentary trade, I am used to how a match is measured: time, move count, accuracy rates, error indices. When a sports product is introduced with no reference metric to check against, I place it in the advertising folder, not the technical folder. And this applies beyond chess. In recent years, sports equipment brands, performance-analysis apps and personal training platforms all use the same language: they promise transformation, they avoid explaining mechanism.

ChessBase was once the name that shaped how people stored and searched games. Their database was the standard tool for generations of players. But the old business model — boxed software, yearly-updated databases — is under pressure from every direction. When free analysis is as strong as paid analysis, a vendor is forced to shift from selling tools to selling a learning experience.

To read FRITZ 20 properly, it must be placed on today’s chess-software map. That map has four main groups.

The first is open-source engines, led by Stockfish. This is raw strength, free of charge, continuously updated by the community. Anyone with a smartphone already holds an engine stronger than any commercial software of the previous decade.

The second is neural-network engines, notably Leela Chess Zero, which learn from millions of games instead of relying purely on search.

The third is integrated platforms, where Chess.com and Lichess offer free analysis engines, exercise libraries, communities and rating systems.

The fourth is structured training platforms such as Chessable, where value lies in teaching method rather than engine strength.

FRITZ 20, by its own positioning, wants to sit in the fourth group while carrying the legacy of the first. That is not an easy stance, and its chosen approach is to stretch the product definition to its maximum.

The key point the release accidentally reveals: in modern chess, engine strength has become a commodity. The real value of training software no longer lies in how well it calculates, but in how it teaches people to think.

Looking back, Fritz once held an advantage no one else had. In 2026, Deep Fritz defeated Kramnik in a man-versus-machine match in Bonn. Back then, software beating a world champion was a thunderbolt. Twenty years later, it no longer surprises anyone. A mid-range laptop running Stockfish can beat any living player. The event became the default, and when a miracle becomes the default, it no longer sells tickets.

The reason FRITZ 20 must call itself a “training revolution” lies not in changed technology but in the need to change the story.

There is, however, a pedagogical problem. In my years of watching matches and training sessions of young players, I noticed something counter-intuitive: the stronger the tool, the more passive the learner becomes. When software says “this move is wrong,” the learner memorizes the right move. When nobody explains why that move is right within the structure of their own position, that knowledge evaporates after three weeks.

This is the point every piece of training software must answer, and the answer does not lie in technology. It lies in whether the software can teach the learner to ask the right questions.

A product aimed at both beginners and experts must choose one. Designed for beginners, it must simplify, provide a path, show patience. Designed for experts, it must go deep, with endgame tools, a vast opening library, multi-layer analysis. One interface cannot be both a kindergarten class and an advanced laboratory without sacrificing one end.

This does not mean FRITZ 20 is a poor product. Fritz has a long tradition, a friendly interface and a loyal European user base. The problem is not quality. The problem is how it is sold to the public. And here I see a familiar gap: the chess industry has learned very quickly how to make products, but very slowly how to tell the truth about them.

Reading only the release, one can easily draw the wrong conclusion. The first mistake is thinking the problem lies in the product. The second is thinking FRITZ 20 is weak against Stockfish. Both miss the point.

The real problem lies in chess’s information infrastructure. ChessBase is simultaneously manufacturer, publisher and news source. When one entity plays all three roles, the line between journalism and advertising disappears, and no one in the chain has an incentive to push back.

In other sports, a new product goes through multiple layers of independent testing. In chess, we have no such mechanism. The global chess community relies on a few forums, a few personal YouTube channels and a few anonymous reviews. That is a fragile evaluation system for a market that globalized long ago.

The irony is that buyers often do not know they lack information. A parent sees a nice interface, attractive advertising, an excited child, and assumes the product has been validated. Chess’s silence toward unproven claims quietly becomes a form of endorsement.

The real concern is not FRITZ 20. The real concern is a parent in Da Nang, willing to spend two million dong on their child, with no way to tell a genuine tool from a leaflet written in the language of tools.

That is also why I do not trust single evaluation metrics when it comes to training. A metric such as the match rate against an engine says nothing about whether a young player understands their own position. It measures obedience, not understanding.

The question I carried home after that tournament is no longer whether FRITZ 20 is good, but: who will tell the buyer the truth?

The chess world has taught the world how a machine can beat a champion. The next step, far harder, is to teach the world to read an advertisement the way one reads a position — slowly, carefully, and without trusting a beautiful move before checking all the lines.

As for the fifteen-year-old in Da Nang and the mother beside him, I think what they need is not another piece of software. What they need is a chess culture mature enough to say: prove it before you sell it.