The Tag Was Wrong, the Year Was Missing: The Valle de Santiago File and the Chain of Evidence

Friday, 2 October.

The date is on the first page of the file. The year is nowhere. On a transfer desk, an entry like that stops me cold. A date without a year is a fee without an instalment schedule: the number looks large, but nobody can say when the money moved, to whom, or against which reference. The report of an armed attack in a public area of Valle de Santiago, Guanajuato, Mexico — with a preliminary toll of two dead — carries exactly that defect. The incident is serious, violent, and captured on camera. The document itself is incomplete.

I work in sports data. My job in the football transfer market is to reconcile paperwork. Who receives what, in which instalment, which add-on triggers when, whether the player was registered inside the window — until those four questions are answered, I do not call a deal complete. That habit makes me slow. It has also saved me from conclusions I later had to break.

This piece is not about football. It is about a wrong tag, a missing year, and a video — and why those three together put the credibility of a news pipeline in question.

Context: what the report holds and what it does not

The incident occurred in Valle de Santiago, in the state of Guanajuato, on Friday, 2 October. An armed attack was carried out in a public urban area. According to preliminary information, two people were killed. The attackers used long guns, and after firing once they returned and fired again. A security camera recorded the scene. The footage was later circulated on social media. Authorities have not disclosed the identity of the victims. The Guanajuato Prosecutor's Office keeps the investigation open.

Beyond those seven sentences, the report holds little else. The source is not specified.

Now consider what a proper information file should contain: the exact date including the year, the location, the identity and tier of the source, the number and identity of the dead, an official classification of the attack type, a list of evidence, and a timeline for the next official update. Three of those seven are missing here — the year, the source's identity, and any official statement on the nature of the attack.

Guanajuato context matters, because the state has long appeared in reporting as one of Mexico's most violent regions. Organised crime, extortion and armed group conflict are not new there. But knowing that context brings its own danger: the tendency to confuse a pattern with an incident. "Things like this happen in Guanajuato" and "this happened for this reason" are separated by a large evidentiary gap. In my trade, that gap is the most abused space of all.

One more element belongs in the context. The spread of security-camera footage of a violent incident across social media is now a routine process. The shorter and faster the clip, the more it is shared. Platform algorithms reward sensation, not evidence. The document that receives the most attention today may be the least verified document in the pile.

Core: how a crime report acquired the "football" tag

Here is the central matter. The document sent for analysis carried the domain label "football". Not one of its twenty information points references any club, player, coach, competition, match, transfer or governance matter. The tag is wrong. This is not merely an error; it is a systems failure.

On the transfer desk I work to tiers of rumour. Tier one: official statements, registration documents, published league lists. Tier two: club briefings, confirmed reporting by reliable outlets. Tier three: agent statements, intermediary signals. Tier four: aggregators, headlines of the "agreement nearly done" kind. Tier five: clips, screenshots and unsupported claims circulating on social media.

Security-camera footage is the strongest example of that fifth tier — and strong is not the same as reliable. Footage proves that something happened. It does not prove who did it, why, or who fired first. Yet in the publicity machine, that footage speaks loudest.

There is a journalistic truth I learned first-hand. At the 2026 World Cup in Russia, in the press area at Nizhny Novgorod, I was one of two women among eighty journalists. Every day brought a flood of transfer rumour. The man with the most dramatic claim drew the crowd. Editors, though, called the number where a date, a release-clause figure and a wage band were written down. The timeline I had already logged — Antoine Griezmann's Atlético Madrid release clause dropping from €200m to €120m on 1 July 2026 — did not go viral, but it held.

In this document, there is not one football-related fact across the whole file. A single word — perhaps "video", perhaps "attack", perhaps a partial match on a place name — misled an automated tagger. A wrong job was done on the conveyor belt. Downstream, that wrong job does real damage.

Consider what a wrong domain tag can do. First, transfer-rumour models. If a crime report enters a football corpus, any language or statistical model built later learns its vocabulary, its place names and its numbers. Second, location-based analysis. If the name Guanajuato enters a layer of football geography, someone later may assume the document contains club-related sourcing from that region. Third, and most dangerous, the devaluation of the subject. If the investigation of a double homicide is filed under sport, attention to that investigation falls.

That is my strongest objection. As a data auditor I would say: a wrong tag is not just wrong, it is a silent error. It produces no error message and no crash. It quietly degrades the quality of analysis. Football finance knows this problem well. A wrong age entry, a wrong loan clause, a wrong sell-on percentage — these carry forward for years and eventually cost a club millions.

I kept the list of sixty-seven names, because expiry is a quiet form of power. In 2026, when football stopped, the only database tracking the 67 Premier League players whose contracts expired on 30 June was mine. I checked each club's wage deferrals, short-term extensions and loan conversions against the league rulebook. That Willian and Pedro would leave Chelsea on free transfers, and that Bournemouth's Ryan Fraser would not sign a short-term extension, required no "source". The rulebook was enough. Agents then began calling me, because I quoted clauses, not gossip.

The same discipline applies to this crime report. What the file lacks is the source's name and tier, an official classification, and any time verification.

A transfer is never a single event; it is a chain of signatures, each with a date. A criminal investigation follows the same rule. The shooting is one event. The chain of evidence is much longer: who owns the camera, between which dates the footage was retained, who first copied it, who circulated it on social media, who edited it, who stamped a date on it. Every joint in that chain needs a date. A missing year means the first link is already broken.

The contract clock was already running before the window opened. The clock on this incident was running before the report was published — from the moment the camera switched on. We are seeing only the final scene, not the beginning.

Another layer: how the numbers should reconcile

I saw what a proper ledger looks like at the Qatar World Cup. In 2026 I applied the same method to all 32 squads and flagged 128 players entering the final six months of their contracts. Lionel Messi's PSG expiry and Cristiano Ronaldo's free-agent status after Manchester United terminated his deal became my reference cases. Beside each name sat three things: the contract end date, the wage band, and registration eligibility for the next window.

This document has none of that structure. No source, no year, no procedural timeline. It is not a ledger. It is a story.

Newcastle did not buy a squad in January 2026; they bought a sequence of deadlines. Trippier £12m, Wood £25m, Bruno Guimarães £40m, Burn £13m, Targett on loan — and I was first to report Guimarães's fee as three instalments plus €4m in add-ons and a sell-on for Lyon. Until the settlement schedule reconciles, I do not write a deal as done. That rule earned me agents' trust, because I asked for deal sheets, not headlines.

I do not trust the rumour; I trust the registration window and the amortization schedule. The same principle belongs in a news pipeline.

Before a document enters a database, four questions should be asked.

First: does the tag match the content? If not, the file goes back.

Second: is the date complete, including the year? If not, it is a note, not a document.

Third: what is the source tier — official, media, unnamed, or unknown?

Fourth: which evidence is direct and which is inference? Are the two layers written separately?

If the answer to any of these is no, the document is not fit for analysis. Today's document fails all four in part or in full.

Contrarian view: where the crowd looks, the truth is not

The security-camera footage circulating on social media is the weakest document in this file. It is also the most discussed. The reason is simple: footage shows, text must be read. Fear and anger travel directly. But when the video of a violent incident circulates on its own, it is not evidence — it is a sensation. A sensation does not answer questions; it multiplies them.

There is another gap few will notice. The report states that the attackers fired once and left, then returned and fired again. In a random robbery, that return is unusual. Returning means confirming the target, which suggests the incident was probably not random but targeted. That is an inference, and I label it as one, at medium confidence. Long guns, a return, and two dead sitting together leave the imprint of organised planning. But an imprint and a proof are different things.

There is a further danger, and it belongs to my own trade. Guanajuato has football. It has clubs, supporters, and matchday security budgets. Some will want to attach this incident to that football context — let us see what matchday security looks like now. The document offers not one point of support for that link. Drawing a connection between the name of a city and the geography of a sport is easy, not responsible. I will not do it. A geographic resemblance is not a document.

And the deepest failure, deeper than the tag, is the missing year. 2 October fell on a Friday in 2026, in 2026, in 2026. The document cannot be pinned to a specific day. A date that cannot be pinned does not enter an archive. And if it does not enter an archive, it does not join the trend count. We cannot propose policy from this incident, cannot measure investigative progress, cannot hold any authority to account. Only the video remains.

The archive remembers what the deadline-day broadcast forgets. I see this principle every transfer window. Half of what makes the biggest headline on deadline day is proven false six months later. Meanwhile the small deal, the small clause, the small sell-on — those are still on the books five years on. The same applies here. Today everyone is watching the video. Without the year, in three years nobody will be able to find the incident at all.

In a male-dominated profession I learned that in a male-dominated press box, the quietest number often told the loudest story. This file holds two numbers: two dead, and one missing year. The first creates sensation. The second renders the document immobile. An auditor's job is to look at the second.

Why the quiet number speaks loudly

I have seen many times that in coverage of violence, the least-spoken fact is procedure. Who is investigating, for how long, how many witnesses have been questioned, who is retaining the camera footage, when it will be erased. Without that procedural information, a homicide report becomes a story rather than an investigation. And if the file moves into the wrong domain, nobody notices the procedure at all.

When reconciling a club's accounts, I follow one rule: beside every number I write a date, a source, and a margin of error. Without those three, the number does not count. A double-homicide investigation deserves no less. "Two dead" is a preliminary count, not a final one. That distinction must be written, because a preliminary count can change the next day. A report that omits the distinction hands the reader a permanent truth that is actually a temporary estimate.

One more procedural question matters: who is retaining the security-camera footage, and for how long? In many countries footage is automatically erased after a fixed period. If the investigation runs long and the footage is wiped, the chain of evidence breaks permanently. That is not speculation. It is a procedural risk anyone can verify.

The next domino: what to watch

Three things to watch from here, all date-dependent.

First, formal updates from the Guanajuato Prosecutor's Office. Identification or arrest of suspects becomes a document at the official tier — not rumour.

Second, date disambiguation. The year must be recovered from the archive. Otherwise the incident remains ineligible for the trend count.

Third, an audit of the tagging system. If the pipeline called this document "football", how many other errors does it hold? That question cannot be answered without sample verification. And the longer the audit is delayed, the more wrong documents accumulate inside the database — quietly, without a single error message.

The real work on a transfer window begins after it closes. Then you see who truly arrived and who arrived only on paper. News works the same way. Today the video is spreading. But six months from now the file that survives will be the one where the date was complete, the source was named, and evidence and inference were written apart.

Friday, 2 October — but which year? Until that question is answered, I will not call this a document. I will only say: there is a video, there is an open investigation, and there is a tag sitting in the wrong place.