From Hong Kong 2026 to Nha Trang 2026: Fritz 20 and the Changing Seat of the Machine Coach

Four in the morning in Nha Trang, the surf reduced to a thin strip of sound outside the window. On the desk, a laptop holds a game that has reached move eighteen. The evaluation bar on the right jumps from 0.00 to +1.34, then drops to -0.62 within four blinks of the cursor. In front of the screen sits a fifteen-year-old boy, a middle-school student in the city who has just won the provincial youth championship. He points at move eighteen and says: The machine says this move is wrong. I ask him: wrong because of what. He is silent for a long time, then recites a line he copied into his notebook the night before. I have sat at boards with several generations of students, but I had never seen a fifteen-year-old speak about his own move in someone else's language like that. The move was his. The explanation belonged to a piece of software. He did not realise that he had just let somebody walk into his game and sit down in the empty chair beside him. The next morning, while I was rebuilding that exchange in my screenwriting notebook, the news arrived from Europe. ChessBase introduced Fritz 20 with a short positioning line: Your personal chess trainer. Your toughest opponent. Your strongest ally. Three roles in one sentence. I read it several times and thought of the boy in Nha Trang, who needs exactly those three roles, but in a very different order. I have been in this trade long enough to know that every time a chess program releases a new version, Vietnamese chess holds a quiet argument, usually inside a few closed groups on social media. One side says that a strong engine is everything. The other says strong engines are killing human chess. Both are half right, and neither addresses the hardest part: no engine, however strong, stays behind at two in the morning to ask a player why his hand shook on move forty. To write honestly about Fritz 20, I have to tell a longer story than a product announcement. That story begins in a different city, in a different year, when Vietnamese chess did not yet have the internet. In 2026 I sat in a studio for the first time, commentating a chess event for broadcast. There was no evaluation bar then, no numbers, nothing but a demonstration board on a stand and the moves of two players read out over the telephone by a man sitting in the stands. To learn whether a move was good or bad, the audience had to wait for the player to publish an analysis in a newspaper, usually three weeks later, usually in Russian and then translated back. In those years the teacher was a specific human being. It might be an old coach at a cultural centre who had lived through two wars and carried a few thousand games in his head. It might be a translated book with a worn cover, where every position came with a moral explanation: this move is greed, that move is patience. Vietnamese chess in that era was taught through character, not through scores. In 2026, a chess program called Fritz appeared in Germany, written by Frans Morsch and developed by Mathias Feist inside the ChessBase ecosystem. I read about it late, in a translated newspaper item, and could not remember the names of the two authors. The only thing I remember is thinking: another piece of software, probably just for playing around. I was wrong. In 2026, Fritz 3 won the World Computer Chess Championship in Hong Kong. Two years later, in New York, Deep Blue beat Garry Kasparov 3.5-2.5 in the rematch, and the whole world began talking about the end of human chess. At the time I was commentating a national final, and nobody in the studio mentioned Deep Blue. We assumed that what happened in America was a matter for people rich enough to rent a supercomputer. Wrong a second time. In 2026, in Bahrain, Deep Fritz drew 4-4 with Vladimir Kramnik in a match called Brains in Bahrain. In 2026, in New York, X3D Fritz drew 2-2 with Kasparov. In November and December 2026, in Bonn, Deep Fritz beat Kramnik 4-2, in a match that included game two, where Kramnik missed a mate in one right in front of him, a moment captured on camera and later turned into a textbook example of the difference between the human eye and the machine eye. In 2026, DeepMind published the AlphaZero results. Soon afterwards, an international community built Leela Chess Zero to reproduce that path in open source. In 2026, Stockfish integrated a neural network into version 12. From that point on, raw calculating strength stopped being the private property of any brand. Dozens of free engines can crush any world champion while running on a mid-range phone. In Vietnam, a parallel story was unfolding. Dao Thien Hai became an international grandmaster in 2026, the first Vietnamese player to reach that tier according to the International Chess Federation. Le Quang Liem won the World Under-14 Championship in France in 2026, the Aeroflot Open in Moscow in 2026, and the World Blitz Championship in 2026 in Khanty-Mansiysk, and now teaches and coaches at Webster University in St. Louis. Nguyen Ngoc Truong Son, Nguyen Thi Mai Hung, Le Kieu Thien Kim, Pham Le Thao Nguyen, Tran Tuan Minh and Nguyen Anh Khoi followed one another into a generation that no longer fears anyone technically. That is precisely why Fritz 20 raises a difficult question. When everyone owns the same strongest engine, the engine is no longer an advantage. What remains to separate winners from losers? Three roles The Fritz 20 positioning line splits neatly into three roles: trainer, opponent, ally. To me that is a useful structure for talking about Vietnamese chess, because the three roles demand three different things, and we possess all three at a far lower level than our potential. The first role: trainer. An engine does not become a teacher by itself. It becomes a teacher when it leaves behind a trace that can be read again. In sport we call that a training log. The track athletes I have filmed never run 400 metres without timing the splits. They know exactly whether their first 200 metres was one tenth faster or slower than the previous session, what their heart rate was at 300 metres, where they usually collapse. Chess has exactly one equivalent tool: an error log. Not a beautiful game, but a list of the moments where you threw away an advantage, together with the specific circumstances: which move, how much time was left on the clock, was the position tense or open, was the opponent an adult or younger than you. An engine such as Fritz 20, used properly, allows a player to gather hundreds of games and find the pattern beneath hundreds of individual mistakes. It took me six weeks to understand something a statistics table could have said in six seconds: my student in Nha Trang is not weak in the middlegame. He is weak from move fifteen to move twenty, always the same way, and always after thinking more than ten minutes about a simple move. The problem lies in his biological clock, not in his opening knowledge. Software with time annotations would have seen it by his twentieth game. In the age of the 30-second video, I spent three months listening to one man talk about a 400-metre race in 2026. The price of listening slowly is time. The price of not listening is that we coach the wrong thing for years. In 2026 I followed Nguyen Van Hai, an 18-year-old boy in Nha Trang who ran 400 metres in 47.80 seconds at a grassroots meet. He was not called up to the provincial team because there was no money for training camp. I filmed him on a phone for two months, from the beach to a red-earth track. The finished video reached 200,000 views in a month, and a local company decided to sponsor him. What made the difference in that story was not Hai's talent, but the fact that someone recorded his numbers for the first time and showed them to other people. A chess engine does exactly what my phone did for Hai: it turns the invisible into the measurable. But measurable is not yet progress. Without a training plan, a data sheet is only a piece of paper that knows how to accuse. The second role: opponent. Before engines, Vietnamese chess players trained by sparring with one another. In the clubs, people queued for a board, and the beauty of that system was that you always met opponents of roughly your level, people who could beat you three games and then lose four. Nobody improves fast when they are crushed every day, and nobody improves when they win every day either. First-generation engines broke that balance. They were absolutely strong. There was no game for you. Many young Vietnamese players I have spoken with went through months of playing only against the machine at maximum strength, losing repeatedly, and then losing confidence in real tournaments. This is a form of professional psychological injury that nobody writes into a medical file. Engine developers recognised this long ago. The recent trend is to make machines play more like humans, to think more slowly, to blunder where humans blunder. Fritz 20 sits inside that trend by putting the emphasis on training modes, opponent simulation and strength adjustment. For Vietnamese players, this is the part worth discussing, not the extra few dozen Elo points of strength. In athletics, nobody trains sprinting by racing Usain Bolt every day. You hire a pacemaker two per cent faster than you, and the pacemaker is the real coach. A chess engine should be treated like a pacemaker. An opponent two hundred points above you is medicine. An opponent eight hundred points above you, every single day, is poison. The third role: ally. In professional chess, every player has a research assistant before a tournament. The assistant prepares a dossier on the opponent: what this opponent opens with, what he avoids, which option he chooses when pushed into a cramped position. In the past this took months and depended on whether the assistant remembered a particular game. Today, most of that work lives inside a database. This is where I believe Vietnamese chess is leaving the most value on the table. Our national teams have players of international standard, dedicated coaches, and a regular international calendar at Olympiads, Asian championships and SEA Games. But we do not have a repository of our own games, fully recorded, classified and published. To understand how a young Vietnamese player progressed over four years, you have to ask people one by one. I once did something similar for another sport. In 2026, when every competition was suspended by the pandemic, I cleaned out a storage room and found an old VHS tape about Nguyen Thi Minh Phuong, an 800-metre record holder with a time of 1 minute 58 seconds, who had been removed from the 2026 Olympic list because of an administrative error. I made video calls to twelve former athletes of her era, pieced the memories together, and built the documentary Empty Stadium - Dreams Frozen. The stadium was empty, but the applause of 2026 still echoes inside every old record. The lesson I drew was not about one forgotten athlete. It was that a VHS tape preserves more than an administrative file, and in chess we do not even have the tape. Every stadium I walk through has a secret door, and the entrance is always a person. In chess, that door is a game nobody wrote down. If Fritz 20 wants to prove itself an ally of Vietnamese chess, its greatest contribution will not be calculating strength. It will be the ability to keep automatic logs, annotate games, classify errors, and turn what happens on the board into reusable data across many years. An engine without that function is still a chess machine, not a training system. The gap lies elsewhere In many years of commentating chess, I have always had the feeling that Vietnamese people play chess better than Vietnamese chess's position on the world map suggests. There are thousands of serious players in Hanoi, Ho Chi Minh City, Da Nang, Nha Trang and Can Tho. There are clubs that meet every weekend. There are families who put their children into chess because they believe it teaches patience. But when I look at the structure, I see three gaps. The first gap is continuity of coaching. A young player improves fast for two years and then stalls, usually not because the talent ran out but because the money ran out, the coach ran out, or the tournament calendar was no longer dense enough. In countries with strong chess cultures, a young player can play thirty events a year, mostly low-cost open tournaments within the country. In Vietnam, the number of suitable events for a promising player is still small, and most are concentrated in a few cities. The second gap is data. We do not have a habit of archiving. A beautiful game at a provincial youth championship, if nobody records it, vanishes from history within a week. Years later, when someone wants to write about that generation, they will have to rely on memory, and memory is always prettier than reality. The third gap is an analytical staff. A professional player needs someone to help him see his blind spots. That cannot be bought with software, because software does not know what this player fears, what he likes, where he loses his composure. I have been tempted to display a row of statistics to be safe, but a table of metrics not tied to human memory is just dead text. These three gaps cannot be filled by a software update. And that is why I do not want to write about Fritz 20 as a product. I want to write about it as an excuse to restate old work. What actually changes One thing I am certain of after more than thirty years of commentating chess: when a tool becomes commonplace, value migrates from the tool to the person. When good cameras became cheap, what distinguished a photojournalist stopped being the camera. When strong engines became free, what distinguishes a player stopped being the engine. Vietnamese players today enjoy unprecedented access. An eighth-grader in Nha Trang can analyse his own game with a tool stronger than any supercomputer of the 1990s. But that access does not automatically convert into rating points. It converts only when somebody turns it into a training process with rhythm, measurement and supervision. I have seen two kinds of students. The first opens the machine, looks at the suggestion, nods, and goes to bed. The second opens the machine, writes down the three moves he got wrong, investigates why he thought the way he did at that moment, and sets himself an exercise for the following week. After two years, the distance between these two people is not two hundred rating points. It is one rung of a career. There is one detail I always remember when discussing this difference. In game two of Deep Fritz against Vladimir Kramnik in Bonn in 2026, Kramnik missed a mate. It was a painful moment for a world champion, and it was captured on camera for posterity. The lesson there is not that machines are stronger than people. The lesson is that even the best have a specific blind spot, and if that blind spot is not recorded, it repeats. For Vietnamese chess I want to stress something that could be called information gain. When every player uses the same global opening database, shared knowledge rises and technical gaps narrow. In such an environment, the person who stands out must be the person who produces private knowledge nobody else has. For example: a ten-year dossier on how young Vietnamese players handle rook-and-pawn endgames. No software sells that data. It can only be produced here, by people here. Why strength is no longer the story Most writing about chess engines stops at comparing strength. Based on my experience following tournaments, that is the least valuable comparison for amateur players, because the threshold needed for training was crossed years ago. What remains to say about an engine is what it teaches, how it teaches, and whether the learner can hear it. There is a paradox worth stating here. The stronger the engine, the lower its pedagogical value, if the user does not know how to turn it down. A perfectly playing engine can never explain why a human played badly. To teach a person, the machine has to know how to pretend to be a person. Fritz 20's direction suggests the developers understand that. Positioning a product around the word training, rather than around the word strongest, is a statement about the market: the buyer of chess software today is no longer someone who wants to beat the machine, but someone who wants to beat his own older version. That is a far more credible claim than adding a few dozen Elo. But there is a downside I want to state plainly, even though it is not pleasant for those selling software. Engine-based training is making games at every level more similar in the first fifteen moves. A boy in Nha Trang, a student in Hanoi and a player in Sao Paulo can all read the same opening file. Stylistic diversity is compressed, and the interesting part of human chess is pushed to the back end of the game. This leads to a consequence few discuss: in a technically flattened environment, the decisive factors shift to intellectual stamina and psychology. Whoever keeps a clear head on move thirty-five of a five-hour game, after a week of continuous play, wins. That is the territory of athletics, not of software. Vietnamese supporters in Russia do not watch football, they watch themselves in every play. I used to think that sentence applied only to football, until I sat through a youth chess tournament and realised the parents in the stands were doing the same thing. They were not watching chess. They were watching their own youth, and the chances they never had. That tells us something about the value of this sport. Chess audiences do not come to watch perfection. They come to watch someone suffer and get up. If we fill in every error with an engine, we will have better players and a duller sport. Another storm from the same wind While developers were discussing training, another story shook the chess world. In September 2026, at the Sinquefield Cup, Magnus Carlsen withdrew after losing to Hans Niemann, and a cheating controversy dragged on for months, complete with investigation reports and litigation. To me that story has a very concrete side related to what is written here. When a strong engine becomes free and ubiquitous, it simultaneously becomes the best training tool and the cheapest cheating weapon. Both roles live in the same installation file. In online chess, this leads to an expensive arms race between platforms and those who want to cheat. That cost is ultimately paid out of prize funds, which means out of the players' own income. It is a paradox the administrators of this sport have not solved: the tool that grows the game is also the tool that threatens its integrity. In Vietnam the problem is still latent. Online events run by clubs or schools often lack the resources to verify. The worry is not that someone cheats, but that there is no detection mechanism, so clean players get tainted by suspicion. A young chess community that loses faith in fairness will recover far more slowly than it would from losing a few sponsors. Compared with the refereeing field I have followed for years, the story is very similar. When referees cannot explain their decisions to the crowd in the stadium, the crowd writes its own version, and that version is usually worse than the truth. In chess, the equivalent mechanism is publishing the anti-cheating process. At the very least, players need to know by what standard they are being judged. The software is not at fault in this story. It simply makes clear something people already knew: any powerful tool will be used in both directions. What remains after the installation finishes I do not think Fritz 20 is big news in the sense of changing world chess. The strength of leading engines plateaued years ago, and the gap between versions has no practical meaning for players below 2500. What is notable lies elsewhere, and it is smaller but more durable. It lies in the engine shifting from referee to scribe. Previously, the machine told us who was right and who was wrong. Now it tells us where we usually go wrong. Previously, we opened the engine to check an opponent. Now we open it to check ourselves. For Vietnamese players, that shift means more than an update. It opens a possibility we never had: building a long-term record of ourselves. A fifteen-year-old who starts a log today will have ten years of data at twenty-five. That is an asset no engine, no coach and no opponent can buy back. In exchange, record-keeping demands a character trait not everyone has: patience with oneself. It requires a player to spend time on games that are already over, while the sweetest reward always lies in the next game. At sixty-five, I still call friends who sat with me in the studio thirty years ago to check a small detail: was that final in the afternoon or the evening, was the crowd large, what colour shirt was the player wearing. I do that because I know memory is a form of archive that can decay, and the only way to preserve it is to cross-check multiple sources. Vietnamese chess needs the same, only at a larger scale and with machines. A record never grows old; it is only waiting for someone who knows how to listen. In chess, those records are scattered in students' notebooks, in phone photos of boards, and in the memories of people who have sat at the board for decades. The work to be done If I could set one task for Vietnamese chess this year, I would not ask for a software licence. I would propose a very small rule: every game at national youth championships must be recorded electronically and published publicly, with both players' names and the date. That costs almost nothing. One recorder, one standard file template, one archive folder on a server. But after ten years, Vietnamese chess will have something no money can buy: a stream of memory thick enough to reveal trends, long enough to spot late-blooming talent, and detailed enough to answer questions we currently answer only by intuition. I once watched a VHS tape rescue one person's career and pull twelve other witnesses out of silence. In chess, the equivalent of that tape is a four-kilobyte text file, recording a game whose two players had no idea it mattered. I believe that in twenty years, when Vietnamese chess has a generation crossing 2700 consistently, people will trace back to today and find the cause was not an expensive new engine. The cause was that somebody decided children's games needed to be written down. About the children in Nha Trang My fifteen-year-old student returned to his unfinished game. After that evening I set one requirement: every game he plays must be recorded with three handwritten lines of self-commentary, no machine. Those three lines must answer one question: what was I thinking when I played that move. For the first two weeks he wrote very little, along the lines of I thought this move was good. By the fourth week he wrote more, and by the sixth week he discovered for himself what had taken me six weeks to guess: he is afraid of positions with few pieces on the board. The fewer the pieces, the more he loses his composure, and he tends to push a pawn early to create a feeling of safety. No engine told him that. The engine only said move thirty-two was wrong. It was his own writing that made move thirty-two meaningful. I do not tell this story to criticise software. Without an engine we would never have seen the problem and gone looking for its cause. But with the engine alone, that boy would keep reciting other people's lines and calling it understanding. What I want to say to parents buying chess software for their children is this. A powerful tool in the hands of a child with no habit of writing produces a good copyist. A powerful tool in the hands of a child who asks himself questions produces a player. The difference between those two children is not the price of the software. I have talked with many coaches in the provinces, people who teach chess out of their own pockets and out of love. They do not lack passion. They lack time, management tools, and a place to compare their students with the rest of the country. If a program like Fritz 20 can do one thing for them, it is to help them work more systematically, not to help their students play more strongly. Looking forward I cannot predict how many grandmasters Vietnamese chess will have in twenty years. Such predictions are usually wrong, and usually written so the writer looks knowledgeable. What I dare say is something else. In the next ten years, the gap between Vietnamese players and the top players of Asia will no longer be decided by access to tools, because tools are everywhere. It will be decided by three human things: a decent record-keeping system, a stable coaching staff, and a calendar dense enough that young players get to collide with each other. Fritz 20 can help with all three, but only if someone decides to do the second and third. No software organises a tournament by itself. No software pays a provincial coach's salary for another five years. There is a sentence I often repeat to younger colleagues when they ask for the secret of the trade: I do not go looking for stories, I go looking for people. The story will come. The same is true for Vietnamese chess. Stop looking for a stronger engine. Go looking for someone willing to sit down after the game and write. Four in the morning in Nha Trang, the laptop is still lit. The boy has gone home, leaving on the desk a sheet of paper with three handwritten lines and a question mark at the end. I read it a third time, then fold it and put it in the drawer with other sheets from ten years ago. A record never grows old; it is only waiting for someone who knows how to listen. Perhaps this boy's game is the same, and my job is simply not to let it disappear.