The Empty Column, the Filled-In Guess: Cricket Coverage's New Archival Problem

Page 14, register number 3, dated 12 July 2026. Sixty-eight boys at the Eden Gardens indoor nets over three days; fourteen selected. On that page I had drawn three columns for every boy — run-up length in yards, seam position, and third-spell accuracy. Forty-eight of the sixty-eight entries were completed in full. Twenty kept the third column blank, because those twenty had never bowled a third spell. In the empty space I wrote in ink: "The record is silent." That night a fellow observer asked me what the point of a blank column was. I said it meant we had not watched them bowl a third time, and what we have not watched, we do not write. An incomplete page is better than a false one. Twenty-nine years later, this month, another page arrived. It was not paper. It was a screen with twelve rows, and beside every row the same phrase — not available. Format: not available. Match nature: not available. Player: not available. Role: not available. Team: not available. Ranking: not available. Score: not available. Venue: not available. Date: not available. Only one thing survived, in brackets, a routing tag: cricket_asia. That was not a report. It was a file in which every field looked filled and nothing was inside. I thought of the night I first wrote "the record is silent" on a page. The difference was this: in 2026 a human being left the column blank, and in 2026 a machine returned the blank file. The truth is the same in both, and so is the danger — the next person in the chain will be in a hurry to fill the space. The true selection pressure in Indian cricket sits in the paperwork, not on the ground. I wrote that sentence first in 2026 and I repeat it on the first page of every file. The Vijay Merchant Trophy team sheet, the Cooch Behar Trophy register, the TW3 verification file, the school leaving certificate, the club transfer form, and now the board's NOC — the chain of these documents decides whether a boy moves from under-16 to under-19. Talent is an input into that chain. The chain is the competition. Part of the chain runs on a verbal economy. The name of the boy I watched bowl six overs at the Mohun Bagan nets in January 2026 came to me from a club secretary's mouth, not from a form. A fifteen-year-old seamer from Amroha, a twelve-yard run-up, the seam upright past the fortieth minute on a damp square. The club wanted him in the senior side that same season. I filed a one-page note recommending eighteen months of age-group cricket first, with the date of every sighting recorded. That note became my template; every report since carries a line reading "tracked since". A third layer has now joined this chain, and nobody has begun auditing it. Scoring feeds, auto-generated match summaries, live-score widgets, a bare stub behind a paywall, photo captions, and now the extraction pipelines that read all of it as content. In my 2026 register the first column of every entry was a name and the second was a date. In a 2026 pipeline the first column is a label and so is the second. That is the whole difference — a register verified names, a pipeline verifies labels. The file that reached me this month is a specimen of that layer. Twelve dimensions, eight analytical sections, thirty-three sub-tables, and zero information points. The schema was valid, the labels were populated, the content array was empty. The file did its job; the source did not. I will not read this as a report. I will read it as a document, and the most important part of a document is the field nobody could fill. The first gate is format. If Test, ODI, T20 or The Hundred cannot be established, everything after it is conditional, because the gravest error in cricket analytics is format contamination. A spinner's economy from the fourth innings of a Test and his economy from the twentieth over of a T20 do not belong in the same column — any more than a fifteen-year-old's six overs on a January square belong in the same column as a twenty-five-year-old's league figures. I refused that comparison in 2026 and the refusal cost me the club's approval. Fine. A register never asks to be liked. The second gate is entity resolution. No name means no role — opener, anchor, finisher, seamer, spinner, all-rounder, keeper. Without a name, whose average are you comparing, whose seasonal strike-rate swing are you laying against a career curve? The first question in every file I keep is: whose, since when, in what format. If all three are missing, the file closes. The third gate is the routing tag that survived — cricket_asia. That tag is not content; it is an address. It says something about which desk the coverage passed through and nothing about what happened. Reading an Asia tag as a India-Pakistan series, or an ACC tournament, is the same error as writing "a Bengal seamer" beside a boy before you have seen his school certificate. Treating a label as evidence is the same mistake, merely moved from paper to server. In 2026 I met the reverse face of that label error. Ishan Porel, from Hooghly, an eighteen-year-old Bengal seamer I had tracked for three seasons — twenty-four wickets in the 2026-17 Cooch Behar Trophy, most of them date-stamped in my notebook. In February 2026 he returned an Under-19 World Cup winner and editors began calling me for verification. That June, during the Russia World Cup, one outlet asked me to write on football academies. I declined and stayed with the Under-19 files. The reason was plain: I do not take up work whose verification grade I cannot establish. The same rule now applies to this file. Nothing in it is verifiable, so every sentence drawn from it, however sensible it sounds, is an inference. And an inference does not earn an analytical grade. In February 2026, at the first IPL auction, two franchise scouts came to Kolkata and put money on the table for my age-group register. I did not hand over the notebooks. I typed a three-page extract of eleven names instead, each carrying a minimum twenty-four-month tracking window. That year the Kolkata franchise academy took two of the eleven; of the other nine I wrote in print that they were not ready. Every auction has a margin, and the margin is where the truth gets filed. The two who were taken were the two at the top of the numbers. Numbers do not record who a boy sits beside at lunch, and that dressing-room column is absent from every transfer model. Franchise academies do not replace state associations — they select from their work, and that selection is now the fastest intervention in the age-group chain. The fourth gate is the one I built for myself in 2026: a minimum information threshold. No recommendation on a player until twenty-four months of observation and at least three dated sightings in the file. The architecture of that rule is simple — it fails closed, not open. It became the most useful habit of my early years, because every boy looks good on a first viewing, and enthusiasm is not an evidentiary category. This month's file handed that architecture back to me in another domain. Any system that withholds a conclusion until at least three information points and one resolvable name are present is doing exactly what I have done for twenty-eight years. Any system that does not is the club secretary who would have sent a fifteen-year-old into Ranji cricket on six overs — good intent, poor method, zero documentation. The fifth question is why an empty file arrives at all. Two explanations are honest. One: the original article existed and was lost in fetching or extraction — a page timed out, a paywall intervened, a widget stalled in JavaScript. Two: the source contained nothing to extract — a paywall stub's single sentence, a photo caption, or a live-score widget carrying digits and overs and no prose. The second cause is now far more common, because the fastest-growing "source" in cricket coverage is the scorecard, and a scorecard is not an article. The schema will look valid for months, the labels will look populated for months, the content array will stay empty. No pipeline should be sent to read a place where nothing was written, and that should be the first rule of any office that keeps metadata. The sixth point is the systemic risk, and the most expensive: the absent format. Without a format token, every conclusion the pipeline emits is silently contaminated, because it will compare across formats without knowing it has. This is the error that does the most damage in my profession and is never seen — not in a headline, not in a table, not in an editorial. It sits inside the record and later files a boy's career in the wrong column. When a scout judges a child on two frames he commits the same error; his version is paid not in a paragraph but in a career. The seventh and last point is the politics of the risk rating. This file cannot be graded low risk. Writing that would be the most misleading sentence I could produce this month. An empty availability array is a pipeline risk, not a cricket risk, and putting the two on one line is not technical carelessness but a serious breach of account-keeping. In my register, zero and unknown never shared a column. When a boy had not bowled a third spell I did not write zero; I wrote silence. When availability itself is zero, no register has the right to write "low" in the risk column. One thing I have actually done deserves mention. I keep a public unverified list — twenty-nine names of boys I have watched with my own eyes but whose format, date or transfer history I could not verify against a document. Not one of them has entered a recommendation of mine. That list is not a weakness; it is a tool. The same tool applies to this file: what is unverified goes on the list, not into the conclusion. A document has only three states — verified, disputed, undocumented. There is no fourth state, and the fourth state's name is "an inference we have temporarily agreed to call a fact". I have never signed a page under that name. And this is where I part company with what the profession currently treats as progress. The largest claim in cricket journalism right now is that machines will make coverage faster and richer — more matches, more numbers, more stories in less time. I read it the other way. The machine's most dangerous defect is not silence; it is confident noise. A clerk who cannot fill a field leaves it blank, because he knows a false entry will come back to his own office. A machine does not leave the field blank. It enters the most probable token, and that token settles in so neatly that the next reader takes it as verified. The result is an archive that looks complete and cannot be audited. That is the inverse of what an archive is. An archive's job is not to look complete; it is to be checkable. The register on which you can ask, beneath any entry, who wrote this and when and from where — that is an archive. The file where every field is full and no field can be questioned is not an archive. It is decoration. The second contrarian angle is the one least spoken in my trade: the answer to this problem is not more data. Indian cricket's records are not absent; they are unindexed. The district association's cupboard, the ledger kept in Bengali longhand, the school's physical-education book, the club secretary's register — all of it exists. I have been signing those pages since 2026. Nobody ever indexed them, and the indexing is the story, not another metric appended to the pile. A pipeline that never walks to the cupboard will produce a partial output no matter how fast it becomes. So the one thing I am taking from this month's file is not a cricket conclusion. It is a procedure: fail closed, not open. Fix a minimum threshold — at least three information points and at least one resolvable name. Make format a mandatory field, because without it the rest is void. Store the source address and publication date independently of extraction success, because when extraction fails you still need to know where the file came from. And treat taxonomy drift — cricket, cricket_asia, cricket_domestic — as a routing defect rather than a naming matter. A routing tag never becomes content, however clean it looks. I am leaving this file without a conclusion, deliberately. I have one condition that opens it: a second extraction from the raw source returning at least three information points, a format token and a name. On that day all eight dimensions run, and on that day I write. Until then the page stays empty — because the register does not lie; only the hands that forgot to write in it do. Or, more dangerously, the hands that wanted to write something in it anyway. Topsoil first. A career, like a trench, is read in layers, never in headlines. And the layer I removed today contained no boy, no score, no century. It contained one blank field, and beneath it one question: the next time a pipeline reaches that blank space, will it write silence, or will it enter the most probable name? The machine does not answer that. The office that runs it does — and that office has not signed a single register to date.