Fritz 20 and the Chess Training Revolution: When the Machine Sits Down at Your Board

Fritz 20 and the Chess Training Revolution: When the Machine Sits Down at Your Board

Three in the morning, and the board is still lit

It is three in the morning in Da Nang, and a small training room still has its light on. On the screen, the game stops at move twenty-seven, exactly where a nineteen-year-old girl let a bishop slip away. She rewinds it. Then rewinds it again. Then she opens the analysis pane, and a line of moves stretches out like a mountain path, leading to a position she had never imagined she could reach. There is no grandstand in that room, no arbiter, no applause. Only the ceiling fan, and the sound of her asking herself what she had been thinking on move twenty-seven.

When the cheering stops, I finally hear the real breathing of the game. I have sat in many rooms like that one, in Hanoi, in Ho Chi Minh City, at junior events where the spectators could be counted on one hand. What I have learned after more than thirty years in this trade is not who is stronger than whom. It is that most of a player's strength is built in the hours nobody watches.

That is why I read the news about ChessBase's Fritz 20 with unusual attention. The chess software house from Hamburg introduces its new product with three roles: your personal trainer, your toughest opponent, your strongest ally. One machine, three roles, and a question much larger than whether to buy it: can a piece of software change how a generation learns chess?

From Frans Morsch to a neural network

Fritz 1 appeared in 2026, the work of Frans Morsch and Mathias Feist, published by ChessBase, a company founded in Hamburg in the mid-1980s. Back then, computer chess was still a laboratory hobby. Four years later, Fritz 3 sat across from Garry Kasparov in Munich at the Intel Grand Prix, and the chess world began to look at machines differently.

In 2026, in Bahrain, Fritz drew 4-4 with Vladimir Kramnik. Four years later, in Bonn, Deep Fritz beat Kramnik 4-2, the first time a program defeated a reigning world champion in an official match. The second game of that match entered history in the bitterest way: Kramnik played a move that allowed immediate checkmate. He sat motionless for a long time before signing the scoresheet. I have replayed that game many times, and each time I realise I was not watching a machine calculate. I was watching a human collapse.

The next turning point came from elsewhere. In 2026, the efficiently updatable neural network, known as NNUE, entered the leading engines and changed how machines evaluate positions. The engine no longer merely calculates; it learns to see a position almost the way a human does, but with extraordinary precision, especially in closed positions where pawn structure and piece placement decide everything. Fritz 20 belongs to that current: a hybrid engine combining classical search with a neural evaluation.

Three roles of one machine

What matters is that Fritz 20 does not sell raw strength. Free engines are already strong enough to beat any active player, and everybody knows it. What Fritz 20 sells is the structure of learning.

Picture a machine with three modes. In opponent mode, it does not simply lower its Elo to your level; it imitates human error. It hesitates on a critical move. It picks a practical continuation over a beautiful one. It misses a tight defensive resource because it wants to win. Training against an opponent that makes your mistakes is far more useful than training against a perfect machine, because in a real tournament you will never meet anyone perfect.

In coach mode, the software reads your games and classifies your errors by severity, like a meticulous teacher taking notes move by move. It points at move twenty-seven and explains why you chose wrongly, rather than simply telling you that you were wrong.

In ally mode, it helps you prepare openings, store them, systematise them, and test your lines against a database of millions of games. This is the least glamorous part and the most decisive at professional level: a professional player earns a living by knowing what he will play on move fifteen.

What actually changes in the training room

The feature I care about most is not engine strength. It is the tools that teach the way humans absorb knowledge.

The guess-the-move mode, a long tradition in the Fritz line, gives the learner a real game by a great player and asks them to choose a move at every turn, then scores them. This forces a decision before the answer is known, exactly as in a real game. Being graded can sting, but it creates something reading annotations never can: a memory of your own failure.

The opening trainer built on spaced repetition, which prompts you to review just before you forget, is one of the most effective applications of learning psychology to chess. I once watched a young player memorise an opening system in three weeks, then forget all of it three months later. Spaced repetition addresses precisely that weakness.

Assisted analysis during play is a double-edged sword, and it deserves more attention, because it is where illusions are most easily born.

The built-in game database and cloud analysis let a player in a provincial town access the same material as a player in Europe. In Vietnam, that is a social change as much as a technical one. The Fritz line also includes packages for young children, where chess is taught as an adventure. For a market where school chess is growing every year, that first door matters far more than an advanced analysis feature.

Six weeks at Thong Nhat and the lesson of scarcity

I remember the spring of 2026. The entire national league was postponed indefinitely. I was working on a series about a women's club that had won the national title several times, and for six consecutive weeks I went to an empty Thong Nhat stadium to film training. No crowd, no match, only the sound of the ball and boots.

What I heard in those six weeks was a story about scarcity. Young players lacked worthy opponents, lacked tournaments, lacked even a teacher who could sit down and analyse situations with them. When there is no opponent, you train against yourself, and that is the kind of training that breeds the most illusion.

Chess suffers the same illness in a different form. A young player in Da Nang or Can Tho can train against the strongest engine in the world, but that does not mean they are being coached. Having an opponent without someone to point the way easily turns into copying. That is precisely the gap software like Fritz 20 aims at.

Based on my experience following matches and training rooms, good training software does not create strong players. It shortens the time it takes for a player to realise where they went wrong. The distance between those two things is an entire human career.

Vietnamese chess and a narrowing gap

In 2026, Le Quang Liem won the World Blitz Championship in Beijing, one of Vietnamese chess's greatest milestones on the international stage. Nguyen Ngoc Truong Son once crossed 2,600 Elo and stayed among Asia's leading players. Tran Tuan Minh and other young faces continue that lineage.

Look at the structure behind it, however, and most Vietnamese chess results are built on individual effort and a handful of centralised training centres. A player from a provincial town rarely has regular access to an international coach. Software fills that gap at a minimum level: it gives a learner an opponent strong enough for daily practice, an error checker that never judges, and a library that fifteen years ago existed only in big centres.

On the other side, strong engines also bring an unsolved problem: cheating in online competition. Anyone who has organised an online event knows the feeling of suspicion when an unknown account plays with unusual precision on move thirty. Platforms have invested heavily in detection, but the race has no finish line. Good tools always have two faces, and chess will live with that longer than any other sport.

The paradox of a machine beside the board

Strategy is like people: you must know who you are before you try to become someone else. A machine does not know who you are. It only knows where the position stands and gives the best move for that position. This is the inherent limit of all training software, and it is the thing advertising rarely mentions.

The first paradox is the homogenisation of openings. When thousands of players prepare with the same database and the same engine, they find the same systems. Professional games today are often uncomfortably similar in the first twenty moves, because everyone follows a map drawn by machine. Fritz 20 is, after all, part of that trend, even as it tries to personalise the learning experience.

The second paradox is speed. An engine evaluates a position in milliseconds. A human needs minutes, sometimes dozens of minutes, for a complex decision. Training at machine speed erodes the ability to tolerate one's own slowness. Players can calculate thirty variations but cannot sit through a six-hour game, which is why so many talents look beautiful on a screen and fade in long games.

The third paradox, and the most dangerous, is memorising engine moves without understanding why. A player memorises an opening to move twenty, the opponent deviates by one move, and the whole building collapses. Chess is a discipline of decision-making under incomplete information. No software decides for you.

When the machine becomes a fairer teacher

There is another side to the story, and this side makes me optimistic.

Women's chess was explained for decades by vague theories about aptitude. Judit Polgar shattered that by entering the world's top ten, and she did it through a training programme designed by her family, not through magic. Hou Yifan won the women's world title several times and stood among the leading players at a very young age. In Vietnam, Le Kieu Thien Kim, Pham Le Thao Nguyen and other generations of women grandmasters proved the same thing on a smaller scale.

The gender gap in chess comes mostly from access. A girl in a provincial town rarely gets an international coach every week. Software like Fritz 20 cannot replace a teacher, but it raises the minimum floor enormously. It gives a nineteen-year-old female player in Da Nang an opponent strong enough to train with, a fault-finder that never judges, and a library that fifteen years ago existed only in major centres.

In chess, I do not look at the board; I look at how a player stays calm while losing. And the best tool is not the one that hands you the strongest move. The best tool is the one that shows you your weaknesses without costing you your belief.

Three standards for judging training software

From my experience following tournaments and training rooms, three criteria matter more than the rest.

First, does the software help the learner understand rather than merely know? A line showing the best move is meaningless if the player cannot grasp the idea behind it. The best tools explain causes: structure, weak squares, bad pieces, time.

Second, does the software teach patience? Chess is a sport for those who can wait. Any feature that encourages players to solve everything in three seconds works against the nature of the game.

Third, does it turn practice into a durable habit or just a few weeks of enthusiasm? Signing a contract is a matter of black and white, but staying is a matter of the heart. That is true of a club, and it is true of a piece of software you open every evening.

What Fritz 20 cannot do

It cannot sit across from you and notice that you have lost confidence after three straight defeats. It cannot know that the nineteen-year-old is considering quitting chess to take a job, and needs encouragement more than a red analysis line. It cannot tell you to go to sleep.

Those things still belong to people: a teacher, a training partner, a family, a community. And in Vietnam that community grows every year, from school clubs to provincial tournaments.

What Fritz 20 can do is make sure that when you sit down at the board at three in the morning, you are not entirely alone. There is a patient machine there, never tired, never irritated, ready to replay game number twenty-seven if you want.

What is happening now

Thirty years ago, a young player who wanted to improve had to find a good teacher, and luck was a decisive factor in a career. Today the decisive factor has shifted elsewhere: who knows how to ask the machine the right question, and who can keep a healthy suspicion of the most beautiful analysis lines.

That is the greatest change Fritz 20, and the software of its generation, brings. It does not make chess easier. It makes learning chess fairer, and fairness is always more uncomfortable than privilege.

The next morning, the nineteen-year-old replays the old game against herself. She reaches move twenty-seven, pauses, and chooses a different move. Nobody applauds. There is no medal. Only a correct move at a moment where she once went wrong.

In chess, that is the entire meaning of progress. And that is why I still believe the board is, after all, where human beings learn to grow up.