Chess 2026: Two Thrones and the Trap of Under-Evidenced Conclusions

In June 2026, I opened the spreadsheet our data-extraction system had just returned. Title field: empty. Source field: empty. Information-points field — the only basis our workflow permits us to analyse from — completely empty. Only one cell held text: the domain label "chess".

Everything else read N/A. Correctly formatted, correctly structured, tidy as a payroll sheet.

A perfect table that contains nothing. It only proved the system had run, classified, and emitted the right template — and had failed to read a single word of the source article. I sat looking at it for about twenty minutes, then realised I was facing the greatest temptation of the trade: filling the blank with what I already believed.

I knew exactly what I could write. Carlsen is still number one by rating. Gukesh is still the youngest world champion in history. Both sentences are true, easy to verify, and have been written a thousand times. People call that a shock; I call it unread data. That time, I did not write.

Why I did not write

My rule has three layers: a number must have a source; sources must be independent of one another; and if all three trace back to a single original report, then I have one source, not three. In the newsroom they call me the slowest writer on staff. They are right.

In July 2026, I was in Russia as one of seven female data journalists accredited for the World Cup. Japan versus Belgium in the round of sixteen: Japan led 2-0 and lost 2-3. The whole press room wrote about spirit. I stayed with the numbers.

Japan generated 1.2 xG in the first 45 minutes. After Fellaini came on in the 65th minute, Belgium won 14 of 18 aerial duels. The entire story sat in the data before it sat in the emotion. I wrote that piece, and a veteran editor waved it away: "Women looking at football data only pick the numbers that suit them." It ran on a regional sports daily and passed 200,000 reads.

Since then I have not written a single match without a data frame. xG does not replace emotion; it explains why our hearts race. That principle followed me to Evergrande in 2026, to Euro 2026, and now to chess.

Chess: the sport where everyone has an opinion and nobody has numbers

In April 2026, Ding Liren became world champion in Astana. In December 2026, in Singapore, Gukesh Dommaraju — born 29 May 2026 — beat Ding Liren 7½–6½, becoming the youngest world champion in history at 18 years and 6 months. Eight months earlier, he had won the Candidates Tournament in Toronto with 9/14.

Meanwhile Magnus Carlsen still sits atop the FIDE rating list, around 2830, with a career peak of 2882 set in May 2026. He no longer holds the world title.

Two thrones. One man holds the title, another holds the number.

This is the most important structural feature of today's chess world, and it is rarely stated plainly: since 2026, the world champion and the highest-rated player are two different people. The media call it a legitimacy crisis. The title answers "who won this match"; the rating answers "who is stronger over time". Mixing the two questions is a methodological error, not a difference of opinion.

When I moved to chess, what I brought with me was ACPL — average centipawn loss, the average loss per move measured in centipawns. Lower is better, and it is a close relative of xG: it measures the quality of decisions, not the result.

In the final game of the 2026 world championship, the match was decided by a single moment. The thirteen games before it were near level. One endgame mistake, and the throne changed hands. The match ACPL does not say Gukesh was comprehensively superior. It says he made fewer errors at exactly one moment. In chess, the result is one data point; the quality of the moves is an entire chain.

Based on my experience following elite matches, this is what most chess writing skips. People give the score, add an opening diagram, and call it analysis. But a score only tells you who won. It does not tell you who played better, who prepared deeper, who was exhausted at move forty.

Then comes the hardest part. In September 2026, at the Sinquefield Cup in St. Louis, Carlsen withdrew after losing to Hans Niemann, and the chess world plunged into its biggest cheating controversy in decades. FIDE set up a fair-play panel and tightened screening. Major events increased broadcast delays, restricted devices, and ran metal detection at playing-hall doors.

This is precisely where an empty data table becomes dangerous.

A silently failing extraction system does not produce an error. It produces an absence. And absence, after passing through enough processing layers, gets read as evidence. A dataset that later "records no cheating controversy in 2026" — not because there was none, but because nobody read the source.

The most dangerous error in this trade is not writing something false. It is writing something true about a thing you never read.

The counter-intuitive angle

The most repeated story of the past two years is "the human is back". An 18-year-old dethrones the champion, and chess escapes the shadow of the engine. It sounds lovely. It has no basis.

Gukesh grew up with engines. He trains with engines, prepares openings with engines, and what won him Singapore was the ability to stay calm across fourteen games — a psychological quality, not a technical one. Attributing that victory to "human intuition" is exactly the error I make whenever I rush: substituting correlation for causation.

Conversely, there is a stronger correlation that is rarely discussed. Recent champion generations are younger, and they became younger precisely as engines became free, ubiquitous, and phone-runnable. That is correlation. I do not have enough data to call it causation, and I will not.

What I will call causation, after years standing between two kinds of data, is something else: chess audiences are starved of verifiable information. They receive plenty of commentary. They receive very few numbers.

Numbers are asceticism: you must give up comfort before you can see the truth.

What to watch

Three signals I will track in the coming months. The monthly FIDE rating-list cadence, where the gap between the top-rated player and the world champion is a structural indicator rather than a private matter between two individuals. The qualifying path to the 2026 Candidates, where the depth of the post-2026 generation will reveal itself in points rather than praise. And fair-play protocol updates: each time FIDE tightens a step, organising costs rise, and that cost flows all the way down to junior events.

And if my data table comes back empty again, I will do exactly one thing: reread the source. Slowly, but without inventing.