Eight Columns, All 'Cannot Assess' — A Null-Result Audit of a Chess File

I opened the file on Tuesday morning, 9:45 Moscow time. Eight columns. The same sentence in every cell — "Cannot assess." Two badges at the top: "Insufficient source information," "Preliminary assessment." Below them a table where every row reads N/A — player, opening, rating, tournament, governance, risk, market. At the very bottom, one line I read three times: "Required input to proceed: please provide the original chess article or a completed Stage-1 deconstruction."

A chess analysis file with no chess inside it.

In May 2026, after I published the first public xG model for the Russian Premier League, two analysts asked the same question — where is the code? My answer was one sentence: here is the code, here is the sample. Today's file is missing the second half of that answer. No sample, no code, only an empty grid that nobody filled in carefully.

A model is a receipt, not an oracle. An empty receipt is still a receipt. This piece is about the date stamped on that empty one.

What a Stage-2 deconstruction actually is needs saying first, because half the work of this article finishes inside that definition. It is a verification layer. A chess story is first broken down at Stage-1 — headline, source, core claims, information points, entities involved, time sensitivity. Stage-2 then measures those fragments across eight dimensions: game and technical analysis, player and data, tournament system, competitive landscape, rules and governance, risk, public narrative, and industry transmission.

Each dimension has its own required inputs. The technical dimension needs PGN, engine match rate, ACPL, time control. The player dimension needs FIDE classical, rapid and blitz ratings, recent performance rating, head-to-head records. The tournament dimension needs tier, qualification path, prize fund, draw rate. The governance dimension needs anti-cheating clauses, eligibility registration, tiebreak regulations.

Today's document has raw material for none of the eight. So every cell says N/A, and under every conclusion sits a warning: "No Information Points supplied."

Two reactions are natural here. One is to fill the empty cells with guesses. The other is to close the file. Both are wrong. The first writes a lie; the second hides a truth. A third path exists, and I have held to it as a rule since 2026 — publish the null result. The trend that vanishes is also worth printing.

Three rules of my method apply directly. First, denominator before ratio — I will not print a count whose base I have not printed. Second, provenance before prose — every figure carries a footnote: sample size, competition, season, model version. Third, absence is also data — the tournament that was not held, the grandmaster who did not arrive, the budget line that was cut, all deserve filing as much as events that happened.

I did not invent those three rules. In June 2026 an email arrived from Dhaka — during the World Cup, while I was running a live shot-quality model across 64 matches and 169 goals and, between shifts, finishing a scrape of forty years of FIDE rating lists. That email did not offer a job. It offered a different way to count. The first number to come out of that road was this: Bangladesh has produced exactly five grandmasters. Niaz Murshed in 2026, Ziaur Rahman in 2026, Reefat Bin-Sattar in 2026, Abdullah Al Rakib in 2026, Enamul Hossain Razib in 2026.

Five. The base matters here: five dates, not five names. Dates arranged in order make a picture; names arranged in order make a story. The story is comfortable. The picture is not.

A pipeline spreadsheet has run ever since, refreshed every week. It belongs in this piece, because today's empty file is one of its columns — the column headed "source document," which is blank today.

Now the real work: what cannot be inferred from those empty cells, and why knowing that matters.

Without PGN, no technical claim survives. The technical dimension needs four things: opening classification, engine match rate, ACPL, and stability under the time control. If none of them exist, then "new system," "innovative preparation," "improved technique" are rumours set in type. Engine match rate requires move-by-move data from the whole game. ACPL requires an average of centipawn loss. These cannot be guessed; they must be measured. And before measuring, someone has to watch — a confidence I do not have. Coding the first sixty matches of the empty-stadium Bundesliga in 2026 taught me that what the eye sees and what the code counts are not the same object.

Without a rating ledger, player assessment is impossible. FIDE classical, rapid and blitz are three separate series. One number cannot carry all three. On top of that, performance rating and live rating are different things; one is a tournament result, the other a long-horizon position. Anyone writing "back in form" without a cross-check on 2700chess is writing feeling, and feeling carries no footnote.

Without the tournament tier, qualification weight cannot be measured. A commercially glittering event and an official qualification event are not the same animal. Prize-fund scale, field strength, draw rate, calendar — with none of these, the question "how big was this win" has no answer. I know that distinction from the Russian side: in Moscow clubs and the state treat chess as infrastructure, so event tier and funding level stay visibly separate. In Dhaka the two often blur into one line.

Measuring the competitive landscape takes four tiers. World Championship tier, the 2700+ challenger tier, the rising-star tier, the national reserve pipeline. None of the four appears in today's document. So "a new generation is coming" and "the veterans are declining" are both guesses. And guesses cannot measure a national federation's capacity; a player's nationality does not establish the strength of that country's pipeline.

In the governance dimension, the biggest risk is a false accusation. Anti-cheating, eligibility registration, tiebreaks — no incident is recorded. So no allegation can be raised. FIDE, a national federation, an organiser, an online platform — which authority applies is itself undetermined. Accusing anyone inside a document that has no subject is a way of dishonouring your own method.

The industry transmission map is correct on paper and empty in numbers. Youth training → players and events → platforms → streaming and sponsorship → courses, equipment, media. Where the chain strains, and how hard, has to be measured. In today's file, direction, magnitude and time frame are N/A for all six segments. No effect on Chess.com or Lichess can be attached to it. No change in sponsorship or pipeline can be inferred either.

This is where the Bangladeshi ledger returns, because it is the most complete series I hold. Five grandmasters, the last in 2026. Then a gap. Home norm tournaments stopped after 2026 and did not return until 2026. The federation's annual budget sits around Tk 11 lakh. The school pipeline was switched off in 2026. Put those four facts side by side and the claim becomes clear: the decline is an accounting failure, not a talent failure. The question is not "who is the next grandmaster." The question is "who stopped paying, in which year, and what did the ledger look like before and after."

Answering it invites one trap — Niaz versus Anand. In 2026 Niaz Murshed became South Asia's first grandmaster, a year before Anand. Write that sentence and a "what if" forms in the reader's mind. I will not step into that "what if." I set the ledger beside it instead: one in 2026, five in total, no home norm tournament between 2026 and 2026. The counts do the mourning; I do not need the adjectives.

And the 2020s? This is where my old baseline needs updating. Rating-first thinking dismisses a 14-year-old national champion and a 59-year-old winner as noise. My rule is to hold the null result and the new result in the same frame. The gap is still there, and the new names are still there. Both truths have to be written together, or the audit stays incomplete.

An empty cell is not a zero. Miss that distinction and the whole audit becomes meaningless.

Zero means it was measured and came out at zero. Empty means it was never measured. What sits in all eight dimensions of today's document is not zero — it is absence. The practical difference shows up in field work. When the Bundesliga returned behind closed doors in May 2026, home win rate across the first sixty matches I coded fell to 32.7 percent, from 43.2 percent in equivalent fixtures a season earlier. That number is not zero; it is measured. In the same weeks, the FIDE Online Chess Olympiad final ended with India and Russia declared joint winners after a server disconnection — and I had no control group for the "effect" of that. Two events happened together without being the same event.

From there comes the largest trap of all: reading correlation as causation. A viral chess story does not mean sustained growth in participation. One tournament win does not mean a long-term form trend. A nationality does not mean the strength of a training system. Those three errors belong to one family, and all three are sitting in today's empty file waiting to be filled in.

Let me also file a charge against myself. Writing about Dhaka's chess from Moscow invites an omniscient tone I never earned in the print era. So I mark provenance separately: what I saw on the ground, and what I now read in the archive. The field experience in this piece is the Moscow model work and 64 World Cup matches; the Dhaka ledger comes from the archive — the federation budget line, the dates of discontinued tournaments, the final year of the pipeline.

What to watch in the next round is a short and verifiable list. The next FIDE rating list, the live slide on 2700chess, official event records, the federation budget line, FIDE announcements — movement in any of those five sources means rerunning the analysis.

One question stays open. The file that came back with "Cannot assess" in all eight columns — is that a failure, or is it the rare document that refused to lie?