The Discipline of Not Guessing: Chess Commentary When the Data Is Empty

Before every broadcast I keep a ritual that borders on religion: two hours of preparation for each game I am assigned. My spreadsheet is always open, the left column listing opening systems, the right column holding notes on the player, and a blank line at the bottom where I ask myself what I will say if every prediction turns out wrong. That night, the data file I received was empty. No headline. No source. Not a single information point to hold on to. I sat staring at the screen, hands on the keyboard, and three excellent stories were already fully formed in my head — stories I could have written in forty minutes, fluent, compelling, and entirely invented. That was the moment I understood my profession is measured not by what I write, but by what I refuse to write.

The pitch never lies; we lie to ourselves with the sound of applause. In chess that sentence holds even harder, because the board is the one place where every falsehood leaves a trace in a database.

Chess is a data sport to a degree few others match. The International Chess Federation publishes official Elo lists every month, split into three systems: classical, rapid and blitz. Live-rating trackers update game by game while an event is running. Game databases store moves, thinking times, and opening variations that never appeared on a serious board. Engines produce average centipawn loss per move and an engine match rate — the share of moves identical to the engine's first choice. In other words, almost every statement about a player can be checked backwards within minutes.

That is why, when a source cannot be read, an analyst does not simply lack information. They face something more dangerous: the opportunity to invent specifics that are verifiable and therefore refutable on the spot. In a domain where every number can be looked up, invented detail is the most expensive kind of error.

The file I received that night ran on a two-layer process. The first layer extracts information from the source article. The second layer builds deep analysis on what the first layer extracted. When the first layer returned an empty list — no headline, no source, no entities, no viewpoints — the second layer had nothing left to analyse. Every cell in the table carried the label “insufficient information to assess”. Here is the point I want readers to hold on to: that label does not mean “no problem found”. It means “never checked at all”. The distance between those two readings is the distance between decent sports journalism and a machine that manufactures false confidence.

Had the document contained content, eight analytical layers would have opened in turn. I want to walk through them, because they show how far a commentator's responsibility actually reaches.

At the technical layer, the first question is always which opening system is on the board, and which family it belongs to. Average centipawn loss indicates move quality; engine match rate indicates preparation. But neither means anything unless it is tied to a time control. A classical game of more than 90 minutes per side allows an assessment of calculation depth entirely different from a blitz game of a few minutes. Armageddon — where White gets more time but must win — creates pressure unlike any ordinary game. Sofia rules, banning early draw offers, turn every move into a political decision. A novelty on move fifteen and a novelty on move thirty are two different stories. And time trouble is where errors are born, where nerve is measured in seconds. Without these facts, a commentator is telling stories, not analysing.

At the player layer, the frame of reference is even stricter. Three official rating lists, live rating, performance rating within a single event, position on the age curve, and the gap between over-the-board play and online play. Online results do not extrapolate to classical strength, and a twenty-year-old on the rise needs a different frame from a thirty-six-year-old holding the summit. The support team behind an elite player is also a variable, because it decides most of the pre-game preparation load. Without a player's name, that entire frame collapses in silence.

At the tournament layer, a misclassification drags every conclusion down with it. A world championship match is not a Candidates Tournament, nor a Grand Chess Tour leg, nor an open Swiss, nor a commercial online event. Qualification routes run through the World Cup, the Grand Swiss, a rating spot, Grand Chess Tour points and wild cards. Each route carries different weight, and insiders know that better than any ranking table. There are team events such as the Olympiad, where the pressure sits on a federation rather than an individual. There are commercial online events, where tempo and psychology are entirely different. Mistaking one event for another means mistaking the whole story.

At the competitive-landscape layer, I always draw four tiers: the throne tier from 2700 Elo upward, the chasing challenger tier, the rising-star tier, and the reserve pipeline behind them. Generational turnover does not happen in a single game but over years of accumulation, and it is usually misread because people only remember the final moment.

At the rules and governance layer, I am especially cautious. Anti-cheating enforcement, format and tiebreak rules, eligibility conditions, federation transfers — this is where one speculative sentence can follow a person for an entire career. The 2026 affair between Hans Niemann and Magnus Carlsen at the Sinquefield Cup is a traceable precedent. A FIDE fair-play panel later concluded there was no evidence Niemann cheated in the over-the-board games at that event, while platform reports concerned online games. Both facts coexist, and anyone who merges them has already broken the story. With this kind of content, silence is an ethical choice.

At the risk layer there are six familiar categories: competitive, career, financial, regulatory, psychological and systemic. But the largest risk sits outside that list. A decision taken on the basis of an article that was never actually read is the most dangerous kind of failure, because it makes no noise at all. It produces only a smooth report, no alerts, and a completely misplaced sense of safety.

At the public-narrative layer, every player tends to be handed a label: prodigy emerges, new king, dynasty ends, redemption arc, scandal, or breakthrough for women's chess. Every label has a life cycle — germination, acceleration, climax, backlash. The gap between public expectation and what actually happens on the board is exactly what a sober analyst must measure before it closes.

At the industry-transmission layer I look at three stages: upstream youth training and talent supply, midstream events and platforms, downstream content, commerce and derivative markets. A shock upstream takes years to reach downstream, while a shock downstream can travel back up very quickly. That is why shallow chess analysis tends to be right on the surface and wrong underneath.

All eight of those layers, in that night's file, lay still. Not one could run, because the first layer returned nothing at all.

The counterintuitive point sits here: we are used to reading the silence of data as a sign of safety. No injury announcement means no injury. No accusation means no cheating. No data means no problem. That is how collective memory deceives itself, and it is more dangerous than any rumour, because a rumour at least provokes argument, whereas emptiness presented as a conclusion does not.

In forty-one years in the studio, I have twice paid the price for misreading silence. In 2026 I left a World Cup feeling I had stepped onto the stage, only to realise I had merely walked the corridor. I rewatched seventeen tapes over a month, not to find other people's mistakes but to face my own. After July 2026 I understood that humility is also a tactic. Then came the empty-arena days, when I called dozens of games with not a single spectator present, and I learned that when the stands are empty you hear the game whispering in another language. That language does not speak of results; it speaks of the truth of each move.

Failure is not an own goal; it is seeing the goal clearly and still shooting wide. For a sports writer, that goal is verifiable truth. The shot wide is a detail invented because it reads smoothly.

What I want to leave for younger people in this trade is not advice about style but a technical habit. Label everything you have not checked. Record the source, the publication date, and the unit of every number. Say “cannot yet be assessed” when it cannot yet be assessed, and never let fluent prose substitute for the solidity of a fact. An analysis brave enough to leave a few cells empty is more honest than one stuffed with cells filled by guesswork.

That evening I closed the file, opened a blank page and wrote a single line: “Source unreadable, cannot yet be assessed.” The next morning I called the desk, postponed the analysis and asked for the original document again. It arrived three days later, complete. The eight layers ran from the beginning, and the finished piece was published later than planned. Nobody remembers it was three days late. But if it had been wrong, people would remember for a very long time.

So the final question I leave behind is not for the newsroom but for the writer: when the data file in your hands is empty, and three excellent stories are already waiting in your head, which story will you choose — or will you choose the truth that you do not yet know anything?