FRITZ 20 and the Silent War of Chess Software: When Free Engines Reprice Trust

In the winter of 2026, in a small internet cafe in Hanoi, I sat in front of a CRT monitor as thick as a refrigerator and saw Fritz think for the first time. The progress bar ran, the depth climbed: 12, then 14, then 16 ply. On one side was a fifteen-year-old boy who had just lost three games in a row at the city youth championship. On the other was a chess program published by ChessBase. Over three hours that afternoon, that program taught me more than the club had taught me in the previous three months.

Twenty years later, in January 2026, I read a press release introducing FRITZ 20.

The release called the product "a training revolution," a "personal chess trainer," aimed at both those taking their first steps on the board and players already competing at tournament level. By the last line, I realized I had learned nothing that could be verified. Not one technical specification. Not one comparison. No move-match percentage, no ACPL figure, no feature list. Only adjectives arranged carefully beside one another, like pieces placed on correct squares but taking part in no position at all.

I am not writing this to review FRITZ 20 — I have never run it. I am writing because that release inadvertently reveals the current state of the entire chess software market. A market where strength has become free, and value must be found somewhere else entirely.

From Deep Blue to AlphaZero: twenty years of strength sold as a product

In February 2026, in Philadelphia, Garry Kasparov beat IBM's Deep Blue 4-2. In May 2026, in New York, Deep Blue won 3.5-2.5. That moment entered history as the day machines overtook the world champion.

For the chess industry, the meaning lay elsewhere: it proved that computational strength could be packaged, labelled, and sold.

ChessBase was founded in Hamburg in 2026 by Frederic Friedel and Matthias Wüllenweber. Fritz was born in the early 1990s through the work of Frans Morsch, and from the middle of that decade the name Fritz became inseparable from ChessBase as a flagship brand. In November 2026, in Bonn, Deep Fritz defeated Vladimir Kramnik 4-2 in a six-game match. It was the last time a classical world chess champion entered an official match against a commercial machine and was beaten.

The chess software empire was built on a very reasonable assumption at the time: chess players need a program strong enough to trust, convenient enough to use, and they will pay for that trust.

Then in December 2026, DeepMind published its AlphaZero results. Across 100 games against Stockfish 8 under long time controls, AlphaZero won 28, drew 72, and lost none. It learned chess from the rules alone within hours, needing no opening database and no human-authored endgame tables. The technical report was later published widely, including in the journal Science in December 2026.

That event paved the way for Leela Chess Zero, an open-source project built by the community using the same method. And in another corner of the market, Stockfish — open-source, free, running on almost any device — still tops engine rating lists such as CCRL with an evaluation above 3600 Elo.

The point to emphasize is simple: the strength ChessBase once sold as a product has become a public good, free and stronger.

When an asset that was once a commodity becomes a public good, the business that owns it is forced to sell something else. FRITZ 20 is that transition, caught midway.

What the release says, and what it chooses not to say

Breaking the release into separate propositions, I count seven information points. FRITZ 20 is a chess engine. It is positioned as more than an engine — a training revolution. It targets ambitious players and professionals. It also targets those taking their first steps. It describes itself as a personal chess trainer. It promises efficient, intelligent, individualized training. And it offers no numbers at all.

That last proposition is the most important one.

In the chess software business, there are metrics any product developer knows and could publish if they wished: the engine's Elo on standard test suites, the percentage of moves matching top choices, the ACPL on a sample game set, the number of games in the bundled database, opening book size, the number of tactical exercises, system requirements and memory footprint. A release that names none of these tells the reader something very specific: this product wants to be judged by feeling, not by measurement.

That is not necessarily a bad sign about the product. It simply means the reader cannot evaluate it. And when the reader cannot evaluate, the power of evaluation shifts automatically to the seller. In any market, that is the moment when marketing quality begins to weigh more than technical quality.

I have followed many chess software launches over thirteen years. The most credible launches I have seen share one trait: they talk about what they do, not about how they make users feel. When you are given no number to argue with, you are left with two options — believe or not believe. And belief is a product with a very high margin.

The paradox of selling to everyone

One point in the release made me pause longer than the rest. The product is said to serve both beginners and professionals.

In any software market, a product serving both ends of the skill spectrum usually pays for it with a loss of clarity. Beginners need guidance: a simple interface, lessons with a path, understandable feedback, few choices, sensible defaults. Experts need almost the opposite: deep access to raw data, the ability to tune engine parameters, batch analysis tools, control over import and export formats. Those two requirement sets pull a product in different directions, and every feature added for one group tends to make the experience heavier for the other.

The platforms that succeeded over the past decade each chose one direction and went deep. Stockfish chose raw strength and open source. Lichess chose a free, ad-free, community-run experience. Chess.com chose an integrated ecosystem: play, study, analysis, tournaments, content. Structured training platforms chose learning paths and human-authored material.

A product claiming to be good for everyone puts itself in the hardest position of all: having to prove it is not shallow anywhere. And when a product dares not narrow its audience, it is usually because it has not yet identified who actually pays.

Strength has become free — and that is good news

The story usually told about free engines surpassing commercial ones is a sad story. I do not think so.

When Stockfish and Leela Chess Zero are free and stronger than most commercial engines, what happens is not the death of analytical chess. What happens is that top-level analysis becomes universally accessible. A student in Nghe An, an undergraduate in Can Tho, a fourteen-year-old girl in a town with no chess club — all of them hold a tool that twenty years ago only national federations could afford.

I have watched the consequences of this over years of following youth events. The gap between a young player in a big city and one in a province used to be decided largely by access to materials and good coaches. Today much of that gap has been flattened at the tool level. What remains — and this matters far more — is the ability to ask the tool the right question.

This is where the phrase "training revolution" actually lives, and it never lived in the engine. It lives in the interface that asks the question.

An engine twice as strong teaches nothing more to someone who already has an engine strong enough. An interface that knows how to ask the right question — that points out you went wrong on move twelve for structural reasons, not calculational ones — does teach. That is the domain of pedagogy, not hardware. And it is precisely the domain a purely promotional release cannot prove anything about.

What ChessBase actually sells

Looking only at Fritz would lead to the wrong conclusion about ChessBase. The company does not live on engines. It lives on databases and interfaces.

ChessBase sells ChessBase — software for managing and analysing game databases, now through many major versions on an annual update cycle. Alongside it comes Mega Database, a historical collection of millions of games, updated periodically, annotated, indexed by opening, searchable by player, event, and variation. For a professional coach, a chess journalist, an opening book author, or a national federation, it is a tool with no free equivalent at the same level of workflow refinement.

ChessBase's value rests on three things. First, data cleaned and standardized over decades. Second, search tools optimized for how chess professionals actually work — speed of positional search, report generation, database merging. Third, a format ecosystem that makes switching platforms costly in time.

The Fritz engine, within that structure, is a component of the package — a brand component. It is strong. But it is no longer the reason to buy.

When a company markets a component that is no longer the reason to buy as though it still were the reason to buy, we are watching an unfinished repositioning. The FRITZ 20 release is a snapshot of that moment: an old product being retold in the language of a new one.

The forgotten queens

In the story of chess engines, there is one group of players who benefited most from analysis becoming free, and who are almost never mentioned in promotional releases.

Women chess players.

Before analysis became free, the path of a young female player was blocked at several levels. No coach within thirty kilometres. No training partner at the same level. No material on games by top-level women players to study from. Judit Polgar reached the world top ten and the highest rating in women's chess history, but she was an exception built by a family with its own training programme — a condition almost nobody has.

Hou Yifan won the women's world championship multiple times and entered the world's top sixty on the open list. Vietnam's Le Quang Liem has entered the world's top twenty. Those milestones only appear when a player has access to the highest quality of analysis, continuously, over many years.

What free engines did was hand that access to anyone with a computer. A girl in a provincial town can now analyse her own games at a depth that twenty years ago only national team members could reach. She does not need permission from a federation, does not need to wait for a scholarship slot, does not need to be selected onto a list.

Imagine an afternoon where your little sister sits down at the board without asking anyone's permission.

That is the real revolution, and it issued no press release.

In the world of chess, the only boundary is the boundary of skill.

And the value of a player is not on a price list; it lies in the fate of girls who dare to dream.

The economics of chess software in the subscription era

There is a structural reason releases like this are becoming more common: the one-time software sale model is narrowing across the industry.

In the perpetual-license model, a business sells one product to one buyer, collects once, and must then persuade existing customers to buy the next version by creating enough difference. In the subscription model, a business does not need a shocking new version. It only needs to keep users inside its ecosystem, month after month.

Chess.com proved the power of the second model: the platform announced surpassing 150 million registered accounts during 2026, with revenue coming mainly from recurring paid tiers. Lichess proved that a fully free model can still operate sustainably through donations and a volunteer community. Two different models, but both point the same way: chess users today are used to paying for a continuous stream, not for a box.

A product called FRITZ 20 — with an incrementing version number, in a packaged-product presentation format — sits on the other side of that stream. That does not mean it will not sell. It has a loyal base: coaches, federations, people who have used ChessBase since the 1990s and have no reason to switch. But it does mean every release must work harder to persuade a generation of players accustomed to everything important being free.

The contrarian angle: the revolution does not come from a stronger engine

The assumption we all carry — that chess improves because engines get stronger — has probably been wrong for a long time.

Over ten years of watching training rooms and youth events, what I have observed is that the fastest-improving players are not the ones with the strongest engines. They are the ones with a teacher, a training group, or a regular rival willing to tell them: that move was wrong for this reason, not that one.

An engine cannot do that. An engine gives the right answer to the question it asked itself. It does not know where the learner is stuck, because it does not know who the learner is.

This is where the chess software industry has hit a limit, and FRITZ 20 may be trying to cross it by calling itself a "personal trainer." But to be a personal trainer, a program must do something no release can describe: it must understand the mistake of a specific person, in a specific position, at a specific point in their learning process.

The paradox of selling to everyone and the paradox of calling yourself individualized share the same root: a mass-market product cannot individualize without data about each individual. And that data lives on the platform, not in the engine. That is why integrated platforms — where users play, err, get analysed, and learn — hold a structural advantage over a boxed software package.

The real training revolution will come from collecting and interpreting the errors of each learner, not from adding ten more ply of calculation depth that nobody can feel.

I have seen a generation ready to press, and I believe they will never retreat.

Signals to watch over the next six months

With a release carrying no publication date, no author, and no specifications, the only way to assess FRITZ 20 is to wait for facts from outside the manufacturer.

First, pricing and distribution model. If FRITZ 20 appears as a recurring subscription rather than a perpetual license, that is the clearest sign ChessBase is moving with the industry. If it remains a one-time purchase with an incrementing version number, competitive pressure stays exactly where it is.

Second, the arrival of independent testing. An Elo comparison on standard test suites, a YouTube testing video, a thread on chess forums — any third-party data will be worth more than the entire original release. When a product is genuinely strong, the manufacturer does not need to praise itself.

Third, a concrete feature list. What distinguishes FRITZ 20 from FRITZ 19, and what distinguishes it from free tools. If the answer lies in the interface and the training workflow rather than engine strength, then the release got the direction right but the emphasis wrong.

Fourth, the moves of open-source engines. Stockfish and Leela Chess Zero continue to be updated steadily by their communities. Every step forward makes the "strength" component of any commercial package a little less valuable.

If none of those four signals appears within six months, the answer to whether FRITZ 20 is a revolution will be delivered automatically.

What I believe after twenty years in front of an analysis screen

In that internet cafe in 2026, what stunned me was not that Fritz was strong. It was that there was a companion who never got tired, never lost patience, and was always ready to play one more game at eleven at night.

Twenty years later, what I believe has changed is not chess. Chess being more beautiful now is only as beautiful as it ever was. What changed is who is allowed to reach that beauty.

A promotional release tells me there is a revolution. But the real revolution already happened, more quietly, and it is not inside any software box. It is in the fact that a girl in a town with no chess club can open a computer and find the right move on the thirtieth move, simply because she wanted to know. She needs no one's permission, no one's selection, no one to open the door.

Whatever product helps her do that faster will have a place. Whatever product merely says it helps her do that will have to wait until someone verifies it.

That is the whole story of FRITZ 20, and perhaps the whole story of the chess software market this decade: strength has been distributed for free, and what remains to be sold is the ability to teach. Whoever proves that with data wins. Whoever only talks will keep issuing releases.