Nine Analytical Dimensions, Not One Line of Data

The file sat on my desk in Busan for four days before I opened it for the third time. The cover page was explicit: Stage-2 Deep Professional Analysis. Domain: esports. Inside were nine chapters. Every chapter had tables, assessment grids, an "Evidence" section, a "Hidden Information" section, a "Risk Flags" section. The tournament-structure chapter carried four criteria rows. The club-finance chapter carried four columns: sponsorship revenue, league and publisher distributions, salary expenses, capital injection. The governance chapter carried five check boxes. The risk chapter carried a seven-row matrix. The final page was a four-item information-value table, each item rated one star out of five.

Not one cell was empty. Every cell contained text.

And the text in every cell was identical: "insufficient information."

The document ran to forty-one pages. It had a table of contents, a glossary of terms, a disclaimer at the end. It read smoothly, like professional work. It was missing exactly one thing: an event.

The source article's title was blank. The source was blank. The publication date was blank. Article type: unclassified. The list of information points: empty. Only the domain label survived, printed clearly: esports.

A label that outlived everything it was supposed to label. That is the detail I kept, and it is why I am writing this instead of filing the pages away.

Over eighteen months I collected eleven files of this kind. They came from four different content suppliers: two selling packages to esports organisations, one selling to sports news outlets, and one selling to a category of client I do not yet have enough evidence to name in print. The going rate in the East Asian market runs between three thousand and eight thousand won per thousand words, depending on how "personalised" the client wants it.

The production chain is nearly identical at all four. A collection layer scrapes source articles from news sites, forums, online leaderboards and transfer databases. An extraction layer pulls out information points, entities, viewpoints, time sensitivity and source quality. A deep-analysis layer builds a multi-dimensional assessment framework. Finally, a human editor approves the headline, adds an image, and hits publish.

The first three layers run on machines. The fourth runs on people. And the fourth is the only layer with no validation gate.

It is worth stating plainly why this chain exists rather than simply attacking it. Esports moves at a cadence no newsroom can match with human labour. Riot ships a patch every two weeks. Valve runs its majors at irregular intervals, sometimes with half a year of nothing meaningful. Tencent runs on a seasonal model. A single esports writer in Seoul today has to hold League of Legends, DOTA 2, CS2, Valorant and at least three other titles in their head simultaneously to keep the job. Nobody does that by hand.

The 2026 search algorithms add another layer. Content that wants distribution has to carry "information gain" — a finding the reader has not encountered elsewhere. That requirement is correct on the merits. It also creates a strange pressure: when there is nothing new, the chain has to manufacture novelty some other way. The cheapest way to manufacture novelty is to manufacture structure. Adding an analytical dimension costs far less than adding a verified source.

The nine-dimension framework was born from that. From the need to fill space, more than the need to understand.

Transfer windows are the harvest season for this kind of content. Across six weeks, the market pumps out thousands of articles a day about deals nobody has confirmed. Half of them end on the phrase "in negotiations." The phrase is not false. It is simply unverifiable. And across that entire content economy, readers have no instrument for distinguishing a story with three independent sources from a story with three repetitions of the same source.

Even the most thoroughly documented career in League of Legends — Faker's — becomes an untraceable line once it passes through a chain with no sourcing. The distinction between data and formatting dissolves at exactly that point.

The Vietnamese market receives this content later and through more layers. A Korean-language analysis is translated into English, then into Vietnamese, then condensed for social media. Across three translations, the sourcing section is usually the first thing cut. The final reader receives the conclusion without receiving the path back to where the conclusion was born. That is the ideal condition for an empty file to survive and reproduce.

In 2026, when I was an investigative reporter at the Kookje Shinmun in Busan, I read a kit-sponsorship press release from Busan IPark. It stated the contract with the domestic sportswear brand STX was worth 1.2 billion won per year. The wording was clean. There was a signature. There was a handshake photo. There was a quote from the chief executive.

The internal file I obtained from an anonymous source said 700 million won.

A gap of 500 million won a year, and nobody on the club's board could explain where that line sat in the ledger. I spent six weeks cross-referencing tax settlement figures against audit reports, reconstructing entry by entry, before publishing a 4,200-word investigation. The board had to account for itself before the ownership council. The chief executive resigned.

What I learned in Busan was not that someone lied. What I learned was this: a fully populated data cell can perform exactly the same function as an empty one. Both travel through the system without anyone asking what stands behind them. The 1.2 billion won figure in that press release behaved precisely like the phrase "insufficient information" in the file I read this week: it took up the space, it looked correctly formatted, and it stopped the reader from asking the next question.

Three years later, when the pandemic emptied the stands, I met the same mechanism again at Seongnam FC. The club announced a thirty per cent pay cut for players. Colleagues reported the statement as given. I pulled the second and third quarter financial reports from the Korean professional football portal and found 2.8 billion won in unpaid wages and transfer fees dating back to 2026. I cross-referenced the date those liabilities arose against a 5 billion won preferential loan from the Gyeonggi provincial government. The rescue money entered the club. It did not reach the players. The piece ran on 15 July 2026 and forced the provincial council to order a special audit.

By 2026 I met a more sophisticated version of the same mechanism. A Spanish broker sent me a forty-seven page dataset on Lee Kang-in's contract with RCD Mallorca. Inside was a 17 million euro release clause and a twelve per cent agent fee routed to a shell company registered in Malta. I did not publish for three weeks. I verified the digital signatures on the documents, benchmarked them against the publicly filed contracts of five other Mallorca players sourced from Transfermarkt and La Liga records, and only then wrote. The piece ran on 12 August 2026. Three days later Mallorca issued a denial. In November 2026 the Spanish anti-corruption authority opened an investigation, and the article became part of the basis for the proceedings.

The common denominator across all three cases is this: money has no name, but a contract always does. And the name on a contract is the one thing formatting cannot manufacture.

Back to the nine-dimension file. I read it the way I read a contract: slowly, twice, starting from the smallest lines.

Chapter one, patch and meta. No game title, no version, no balance-change description, no win rate, no pick-ban data. The "meta direction" cell read insufficient information.

Chapter two, tournament system and format. No name, no tier, no qualification path, no schedule density. Nothing against which to assess upset rates or the stability of strong teams.

Chapter three, teams and players. Not one name. Not one role. Not one form curve. Not one injury.

Chapter four, regional landscape. No region placed, with a methodologically correct footnote: regional standing depends on the game title, and when the title itself is unresolved, placement is impossible.

Chapter five, club finance and business. Four columns — revenue, cost, trend, warning — all empty. With one sentence I wanted framed: the absence of a signal here reflects an empty input, and it does not confirm financial health.

Chapter six, rules and governance compliance. Five check items blank. With a second warning: a null input must not be read as "no violations found."

Chapter seven, risk profile. The only chapter with a genuinely scored entry. That entry sits in none of the competitive, financial or personnel risk categories. It sits under analytical-integrity risk — taking decisions on a null input and producing fabricated findings beneath a format that sounds professional. Level: high. Probability: high. Impact: high.

Chapter eight, public narrative and expectation. Empty. Chapter nine, industry transmission. Empty.

The most notable point, and the reason I had to write, lies in the structure of the failure. The "esports" label survived. The article type was flagged "unclassified." Meaning even the classifier would not commit to which domain the source belonged to. The labelling layer and the extraction layer disagreed with each other, and that disagreement was swallowed silently before it reached the final human reviewer.

Engineers call this a silent failure. The system completes, returns a structurally valid payload, and the payload is semantically empty. No warning. No error code. Nobody woken at two in the morning.

Three hypotheses are the most plausible explanations. One: the source body was empty, paywalled, or image-and-video only, leaving no text to extract. Two: the extraction layer threw an error, the error was swallowed, and the system returned a default empty schema. Three: the source genuinely was not esports content, and the label is the residue of a classifier that never committed.

I do not yet have three independent sources to settle on any one of them. I present all three, because that is the most honest way to handle an open question. It is also what the file itself did correctly in chapter seven: it scored its own failure instead of dressing it up.

Based on my experience following matches — I have watched LCK since the 2026 season, and I still believe the most reliable signal in a BO5 sits at the timing of the pick-ban phase rather than in the scoreline — I think esports analysis is entering the phase European football passed through fifteen years ago. When data becomes a commodity, data provenance becomes an asset. Whoever controls provenance controls the interpretation market.

There is another layer the file never touches, but anyone working in the trade has to touch it themselves. Automated content operations serve readers, and some portion of them also serve odds listings. A technically-sounding paragraph about a team's form can exist purely to hold a reader for another twenty seconds before they click a different link. Nobody calls that manipulation. It is content in the wrong place. But when not a single cell in the document can name a player, that paragraph is no longer information. It is a piece of interface.

Here I have to separate out the legitimate part of the story that critics usually skip.

The nine-dimension framework does not cause this problem. In one respect it is the thing preventing a worse one. A system willing to write "insufficient information" into every cell has refused to fabricate. Most automated content in the market does the opposite: it always has something to say, and that something is usually speculation presented as observation. Against that baseline, an empty file is an act of integrity.

I also have to record that I found chains doing it correctly. Three of the suppliers I surveyed run hard validation gates: if the information-point list is empty and no entity is resolvable, the system returns an error and halts the entire pipeline, even when the client has already paid for that day's content package. That is correct engineering, and it exists. Its existence is why I am not writing this as an indictment of technology.

The problem sits at the business layer. A validation gate generates no traffic. An empty article sells no advertising. An error message does not reach the front page. When the newsroom's performance metric is page views, the decision to remove the gate is not made by an engineer. It is made by whoever carries revenue responsibility, in a meeting nobody minutes. And when the empty article escapes, it still carries enough form to survive.

There is a fairer reading still, one that is kinder to working journalists. Most editors at the end of these chains cannot read the source code. They receive a finished package with a table of contents, tables, and a disclaimer. To them, that file looks exactly like a vetted product. The responsibility does not lie with the person who presses publish. It lies with whoever decided the publish button did not need a door in front of it.

This framing matters, and I learned it at a literal price. In 2026, covering the Asian Games in Jakarta, a Korean sports medicine official told me that three weightlifters in the 62kg and 69kg categories had shown abnormal blood results before the competition, but the inquiry had been suspended for lack of a B sample. I used my press credentials to access the Asian federation's doping control office and logged seven storage-procedure failures from the sample-handling log. A three-part series on the urine-sampling loophole ran in October 2026, and the Asian weightlifting federation was forced to reform its monitoring procedure before the Tokyo 2026 Olympics.

Had I written that story hunting a culprit, it would have ended with a name and changed nothing. I wrote it asking which stage of the system failed, and it became a preventive document. The same principle applies to the nine-dimension file.

The failing stage has a name: the input validation gate. The fix has a name too: force a hard error when the information-point list is empty and no entity is resolvable, instead of returning a passing-but-empty payload. The minimum conditions for a chain to be allowed to run analysis also have names: game title, at least one substantive information point about a team, player, patch or transaction, the patch version, tournament name and tier, team and player names, region, publication date, and source-quality metadata. Without the game title, every downstream dimension is meaningless, because each title operates under a different governance regime.

Every season ends, but a file does not. The forty-one pages on my desk will stay in the drawer, and in a few months they will have siblings: a newer version, tighter formatting, a cleaner table of contents, still no defendant.

For readers, the only instrument is a small question, askable of any esports analysis this transfer window: which game does this name, which player, which tournament, and which source does it cite. If all four answers are blank, what remains is formatting.

The truth sits in the smallest lines nobody bothers to enlarge. In that file, the smallest line was the one for the publication date. It was blank. Across all forty-one pages, it was the most honest line there.