Post-Carlsen Chess: Denser Data, Emptier Throne

On 12 December 2026, inside a soundproofed playing hall in Sentosa, Singapore, Gukesh Dommaraju sat motionless at the board while Ding Liren buried his face in his hands. Game fourteen of the World Chess Championship had lasted nearly six hours. In the commentary booth, three leading grandmasters had used the word "draw" three times. They were wrong all three times.

Gukesh won. At eighteen, he became the youngest world champion in chess history, breaking a record Garry Kasparov had held for four decades. But the moment that kept me at the screen for two extra hours after the game ended was not the coronation. It was Ding rising, shaking hands, and leaving the room without saying a single word to the arbiter.

I replayed that game three times. The first time to learn who won. The second to watch the moves while both sides still had material. The third — the only genuinely useful pass — to watch what both players decided not to do.

On that third viewing, what stood out was not the final move. It was the long stretch before it, when both sides kept restructuring the empty squares around their own king, and neither wanted to be the first to break the balance. This is the kind of content that an engine evaluation calls "neutral", that a statistics sheet calls "nothing happened", and that an audience calls "boring".

That game taught me nothing new about Ding or about Gukesh. It taught me that most of the content of a chess game lives in what both sides avoid together — and that this is exactly where the industry's strongest analytical tools, from average centipawn loss to win rate, suddenly fall silent.

Context: an empty throne and a generation knocking

In July 2026, Magnus Carlsen announced he would not defend the world title. A few months earlier, in Madrid, he had just beaten Ian Nepomniachtchi to keep the belt. The decision was not a withdrawal from chess — Carlsen kept playing, kept winning, and remained the world number one on the FIDE rating list. It was a withdrawal from a format. He said the championship match format demanded too much opening preparation and too little actual chess.

That vacancy was filled in two steps. In April 2026, in Astana, Ding Liren beat Nepomniachtchi in a tiebreak to become China's first world champion. Fifteen months later, in Toronto, Gukesh Dommaraju won the Candidates at seventeen, earning the right to challenge. In December 2026, he defeated Ding in Singapore by a final score of 7.5-6.5.

Three years, three different champions, two different countries, two different generations. In the modern history of chess, no period has seen the crown change hands this quickly.

To see that speed more clearly, set it against an older marker. In May 2026, Carlsen reached a rating of 2882 — the highest any player has ever touched in the official rating system. From October 2026 to October 2026, he went 125 consecutive classical games without defeat, before Jan-Krzysztof Duda ended the streak at Altibox Norway Chess. Those two figures describe an era in which the ceiling of the sport was effectively occupied by one person.

Then the ceiling came down. On the FIDE rating list published in early 2026, the group of players above 2700 Elo was larger than at any previous point, and the average age of that group was roughly four years younger than a decade earlier. Most of that increase came from a single country.

Four Indian players sat in the world's leading group when Gukesh took the title: Gukesh himself, Rameshbabu Praggnanandhaa, Arjun Erigaisi and Vidit Gujrathi. Praggnanandhaa was once the youngest international master in history, qualifying in 2026 at ten, and became a grandmaster at twelve in 2026. Arjun Erigaisi crossed 2800 Elo in November 2026. Nihal Sarin, younger than both, has been a grandmaster since he was fourteen.

Outside India, the young wave also rose. Nodirbek Abdusattorov of Uzbekistan won the World Rapid Championship in 2026 in Warsaw at seventeen. This is not a phenomenon of one country. It is a phenomenon of one generation.

The question the chess industry is asking is simple: when the era of a single dominant individual ends, what replaces it? The answer most forums give is "data". I am not sure that answer is right. But I am sure it deserves more careful scrutiny than it usually gets.

The precision of machines and its price

Any elite chess analysis today starts from one fact: humans are no longer the strongest entities in the playing hall. Engines such as Stockfish or Leela Chess Zero evaluate positions at depths no player can match, and every professional player uses them to prepare.

The most common measure for assessing a player in that context is average centipawn loss, usually abbreviated ACPL — the average evaluation loss per move, measured in centipawns. The lower the ACPL, the more "accurate" the player. The problem lies in the word "accurate".

A player can achieve an extremely low ACPL by choosing safe, clear moves that engines rate highly but that generate no pressure at all. Another player may carry a higher ACPL because he dares to pick options that look dim on evaluation but are brutally hard to handle on a real board. The metric cannot tell these two apart. It only says the first one errs less.

The same happens with draw rate. Through the 2010s, the draw rate in elite events hovered at a very high level, at times above 60% in some round-robin events. People read that figure as proof that elite chess had been drained of ideas by machines. That reading ignores a detail: many of those draws were technically high-quality games that simply lacked a winner. A draw on move forty is entirely different in nature from a draw on move twenty-five.

This is why organisers tried to intervene in the rules. Sofia Rules — a set of regulations named after the tournament in Bulgaria that first applied them — prohibit agreeing to a draw before a set number of moves, usually thirty. Armageddon is used as a tiebreak format: White gets more time but must win, Black only needs a draw.

Both measures treat symptoms. They change visible behaviour without changing the underlying incentive. A player forced to play twenty more moves before being allowed to offer a draw will not become bolder because of it. He will simply play those twenty moves more cautiously.

The more interesting response came from the players themselves. From 2026, a group of leading players including Carlsen began entering Freestyle events — also called Chess960, where the starting positions of the pieces are randomised before each game. The purpose is clear: neutralise the entire opening-preparation industry and force players to think from move one.

This is a tactical statement, not an aesthetic one. It says that the value of an elite player in the engine era lies not in memory, but in handling situations that have never occurred before. Anyone who has followed chess long enough knew this before engines existed.

When the score does not tell the whole story

FIDE's Elo rating system is published periodically, usually monthly. Alongside it exists another system: live online ratings, updated after every game, tracked and published by services such as 2700chess. The existence of these two systems says a great deal about how the industry operates.

The official list lags. It does not reflect a player who has just won four games in a row this week. The live list reflects that, but is easily distorted by short result sequences. Both are correct within their scope, and both are misused in the same way: people take them out to draw conclusions about long-term form.

Based on my experience tracking matches across many seasons, I have grown used to separating three layers of numbers: the Elo rating describes baseline ability, recent results describe form, and the specific opponent describes context. Merging those three layers into one single metric is the fastest way to misread.

There is a phenomenon I call score-form divergence, and it appears regularly. A young player can gain several dozen Elo points in a season without a corresponding rise in the quality of his chess, simply because he plays more often and faces weaker opponents. Conversely, a veteran can hold his rating steady while playing at the highest level of his career, because he deliberately chooses fewer events and avoids risk.

For the Indian group, the third layer matters enormously. Praggnanandhaa and Gukesh grew up in the same training system, playing each other hundreds of games since they were juniors. Their head-to-head record does not reflect a gap in ability; it reflects who became accustomed to the other's style first. This is the kind of information a rating table never displays.

Another example of the limits of numbers is the age curve. In chess, most players peak between twenty-five and thirty-five. But the peak age has shifted. Players born after 2026 reach 2700 earlier by roughly three to four years than the previous generation. That does not mean they are better at the same age; it means they gained access to analytical tools and high-quality opponents earlier.

That difference matters for one specific calculation: career longevity. If a player peaks four years earlier, does the peak last four years longer, or end four years sooner? There is not enough data to answer, and anyone who answers with certainty is speculating.

The road to the throne

FIDE's qualification system for determining the world champion's challenger is one of the most complex structures in professional sport. It comprises several parallel paths: the World Cup, the Grand Swiss, the Grand Chess Tour series, a rating spot, and a wild card.

Each path produces a different kind of player. The World Cup is a knockout event where a single mistake ends the tournament. It rewards the ability to withstand short-term pressure. The Grand Swiss is a long Swiss-system event that rewards consistency. The rating spot rewards maintaining form across many different events throughout the year.

A player can be strong enough to be world champion yet never reach the Candidates, simply because the qualification structure does not match his playing style. This is the point data analysis usually skips: the qualification system is not neutral. It is a filter with its own bias.

Financially, elite chess has improved markedly. The 2026 World Championship in Singapore carried a total prize fund of 2.5 million US dollars, split 60-40 in favour of the winner. Grand Chess Tour events carry total prize funds in the millions of dollars each season. Those figures remain small next to popular team sports, but they are enough for a few dozen players to live entirely from the profession.

Further down, the picture is very different. A grandmaster rated below 2600 often has to supplement income through coaching, commentary or content creation. This is why many young players switch to fast-prize online events rather than the classical circuit. That economic structure creates a pull that anyone analysing the growth of the sport must account for.

A power map being redrawn

If elite chess is divided into tiers, the most sensible division today has four layers.

The top layer holds one person, and that person does not hold the title. Carlsen remains the highest-rated player and still wins most of the events he enters. His presence in this layer creates a paradox: the official world champion is not the strongest player in the world, and everyone knows it.

The second layer is the group rated above 2700 Elo. It is the largest and the most brutally competitive. The gap between the third-ranked player and the fifteenth is far smaller than a decade ago. Within this layer, one good tournament can lift a player from outside the top twenty into the top ten.

The third layer is the group of unstable young talents. It has the largest volatility and also attracts the most media attention, because a story about a prodigy always sells better than a story about a thirty-year-old playing steadily.

The fourth layer is the development system behind them. It is the most important and least discussed. India has dozens of academies and thousands of junior events each year. Uzbekistan has a centralised training programme. China has a national squad-based coaching system. The United States relies on school systems and private centres. Russia maintains a long coaching tradition but is affected by restrictions on international competition.

A development structure determines national results with a lag of roughly eight to ten years. That means the current Indian wave was decided between about 2026 and 2026, and the next wave is already decided — it simply has not appeared yet.

On women's chess, the gap remains clear. Judit Polgar once broke into the world's top ten at her peak and is the only woman to have done so. Hou Yifan of China once reached a rating close to 2700. Koneru Humpy of India held a leading position in women's chess for many years. But the number of women in the 2700 group remains countable on one hand, and the cause is not ability. It is structure: participant numbers, investment levels, and the number of elite playing hours available.

Rules of the game and the integrity question

In September 2026, at the Sinquefield Cup in Saint Louis, Carlsen lost to Hans Niemann and then withdrew from the event. He made no direct accusation at that moment, but withdrawing from an event he was already playing carries meaning. Weeks later, in a public statement, he was clearer that he believed Niemann had cheated more than Niemann had admitted.

In January 2026, Niemann filed suit in Missouri seeking 100 million US dollars in damages, alleging a coordinated boycott. The case was dismissed in June 2026.

What stands out about this affair is not the legal outcome. It is that professional chess's anti-cheating apparatus was built mainly for the online environment, where every move is logged, thinking time is measured to the fraction of a second, and anomaly-detection software has enough data to compare against. In over-the-board play, the data is far sparser and the standard of proof far higher.

This is a structural problem, not a personal one. A strong anti-cheating system in one environment does not automatically become a strong system in the other. The chess industry spent years improving the online side, where audiences are largest and commercial risk is greatest. The over-the-board side still relies heavily on manual process.

Another regulatory line draws little media attention but directly shapes the power map: federation transfers. A player can change the national federation he represents, subject to waiting periods and fees. That rule determines who plays team events, who receives national quotas, and who benefits from which federation's financial support. In a sport where state sponsorship remains an important revenue source, this is a major lever.

Risks that do not sit on the board

The biggest risk to an elite player is not losing a game. It is burnout. The current schedule for the world's leading group is dense enough that many players must choose between protecting their rating and protecting their health. Withdrawing from events is no longer an exception.

The second risk is income structure. Most of a mid-ranked player's income comes from tournament prizes, which swing sharply. When there is no event, income is zero. This creates an incentive to over-play, and that incentive directly conflicts with the need to recover.

The third risk belongs to the analysis industry. When the data is complete, people read the data. When the data is empty, people fill it with story. And story always sounds better than data.

When the data is empty

This is the part I want to give the most room, because it concerns my own trade directly.

For years I have built my own recording routine for every major event: watch live once, rewatch focusing on structure when the side in control of the ball — in chess, the side holding the initiative — has it, and rewatch a third time when the position turns. I assign codes to each pattern for quick retrieval. I draw the diagram before writing the words, because a diagram cannot be bent by language.

But there is one kind of situation that routine cannot handle: when the data does not exist.

I have run into this many times. An event with no stream. A score table with gaps. A game whose record is incomplete. In those cases the professional pressure is enormous: you must write, must publish, must produce content. And the easiest way to produce content is to use your own background knowledge to fill the hole.

That is when the error happens. Not a factual error, but an error of nature. An empty data field has two entirely different explanations: nothing happened, or something happened that was not recorded. The writer cannot tell these apart by looking at the blank. But the reader does not know the writer cannot tell them apart.

The result is a text that sounds extremely confident, full of numbers, full of terminology, and entirely unsupported. Worse, it is silently dangerous: it does not say anything false, it merely says things no one can verify.

In chess, this error appears in familiar places. A quick draw is labelled "a lack of ideas". A losing streak is labelled "a form crisis". A young player winning three games is labelled "the new generation has arrived". All three conclusions may be right, all three may be wrong, and none of them is supported by actual data.

I believe in the diagram, but I believe more in the space between diagrams. That space is where information has not been recorded, and the analyst's job is to say clearly that it has not been recorded — not to fill it with a plausible-sounding hypothesis.

The press conference has no seat for me. Tactical history always does. But that history is only worth something if the person writing it is willing to say "I do not know" in the right place.

The person standing at the edge of the pitch sees the whole match, not only what is inside the goal. In chess, the person standing at the edge of the board sees the whole position, not only what is on the scoresheet.

Public narrative and the expectation gap

Every phase of a sport is told through a handful of narrative labels. Chess today is being told through four.

The first label is "the new king". It appeared when Gukesh won at eighteen. It has a factual basis: the age record is real, and it will stand for a long time.

The second is "the dynasty is over". It appeared when Carlsen declined to defend. It has a much weaker basis, since Carlsen still won most of the events he entered afterwards.

The third is "the Indian wave". It has the strongest basis of the four, resting on the number of young players reaching grandmaster level in one country within a short window.

The fourth is "chess is booming". This is the most questionable. The surge in online players is real, but most of that growth comes from casual players, not from people who follow competitive chess. The label is often used to infer that sponsorship will rise accordingly. There is no clear evidence for that inference.

The largest expectation gap concerns Carlsen's position. The market still prices him as world champion, while the official system already has a different champion. That gap cannot last. It will resolve one of two ways: either Carlsen returns to the championship format, or the market adjusts its expectation downward. Both take time.

The flow of the chess industry

The entire chess industry can be described through three connected layers.

The upstream layer is the junior development system. It determines the supply quality of players over the next decade. Countries that invest here harvest late but surely.

The middle layer is the tournament system and the playing platforms. Here, one structural event has already occurred: the world's largest online platform acquired the group owning the traditional chess brands, centralising a significant share of the rights to distribute competitive content. Alongside it, an open-source platform maintains a counterweight with a non-commercial model.

This concentration has two faces. On one side, it generates the resources to run online events with large prize funds, something that did not previously exist. On the other, it leaves a single entity holding decision power over scheduling, format and audience access at the online level.

The downstream layer is content, commerce and derivative markets. Here the clearest shift is from expert commentary to personalised content. Many professional players now earn most of their income from streaming and content production rather than from tournament prizes. This changes incentives: a player may choose events based on audience appeal rather than sporting value.

The effect on the sport's public image runs both ways. It widens access, but it also pushes value toward entertainment capability. A solid, steady, reserved player will struggle to compete for reach against an impulsive player with a strong personality, even when the first is stronger on the board.

Takeaway

The past three years taught the chess industry something it never had to learn in the four preceding decades: how to operate when no single central figure monopolises attention.

The question worth verifying next season is not who will win. The question worth asking is this: as the group above 2700 Elo keeps thickening and the gaps at the top keep narrowing, will the quality of elite chess rise — or will it merely become harder to predict?

Those two possibilities differ in nature, and there is only one way to tell them apart: record accurately what actually happened, and say clearly when there is nothing to record.