About the Open Files. These are live. Not a case study written after the fact with the answer already in the room — a reading published while the outcome is still open, built from public information only: reporting, official statements, observable actions, published corrections. No insider access, no internal documents, no privileged briefings.
That means this is not a diagnosis. A diagnosis takes structured observation inside an organization, and no outsider can supply one. What follows is a structural reading — the architecture that best explains the public record on the date below, what should appear next if that reading is right, and what would force me to weaken it.
The forecasts are made in advance, not in arrears. Every file is cryptographically timestamped before it is published, and updates are appended rather than substituted. When a reading gets out-explained, the original stays where it is and the correction goes underneath it.
Written and cryptographically timestamped 25 August 2026. The file and its OpenTimestamps receipt are here — drop both at opentimestamps.org and check it yourself. It does not prove I was right.
Something still on the drafting table
A while back I started a watchlist. Not a reading list — a test bench.
The framework I work from has been built, argued, and run inside client engagements for years, and every one of those is private and stays private. Which leaves me in an awkward spot when somebody asks whether it works. I can describe it. I can walk you through a case with the identifying details filed off. What I cannot do is hand you something you can check.
So I started watching organizations everybody already knows, with public records, where the work could be done somewhere anyone could look over my shoulder.
Several are on the list. FIFA moved fastest.
And here is the thing that decided it for me.
Business forensics are autopsies. An organization fails, we open it up afterward, we explain what killed it, and everybody nods — because by then the answer is already in the room. The analysis is careful. It is often correct. And as proof of anything it is worth almost nothing, because nobody made a call while the outcome was still in doubt.
Nobody follows the patient in from the ambulance.
Nobody says, before the tests come back, here is what I think is wrong, here is what should show up next if I am right, and here is what would tell me I am not.
That is what this is. The reading was written and sealed before the outcome existed, because a reading that arrives after the fact cannot be tested and this one can. The date on the file is the point. If I am wrong, the record will say so in my own handwriting.
The election is not the test
There is a leadership vote coming, and it will be treated as the verdict on everything written here. It is not.
This file makes no claim about who wins. The reading below holds that the fracture is structural and sits at more than one layer of a federated organization — which means it survives either result. A president who wins comfortably and a president who loses both leave the same question untouched: who is entitled to speak for whom, and does influence travel the route the chart says it does?
If you want to grade this, grade the forecasts. That is what they are for, and they have dates on them.
How I got here, and what I threw away
The first read was the obvious one, and it was wrong.
When this started producing signal, my working hypothesis was the one everybody else was running: authority is over-concentrated at the top and it is distorting direction. It is a reasonable reading. It fits a lot of the record. It is also what every governance critic was already saying — which should have been the tell, and took me longer to notice than it should have.
What broke it was a second complaint, at a different layer, with the same shape. A confederation said a decision had been made without consulting the members it affected. Then a member association said its own confederation had issued a position without consulting it. Same objection, one level down, pointed the other way.
Over-concentration at the top cannot explain that. It predicts one direction of grievance. The record had two.
So the lead read changed — not because the first was disproven, but because a second explanation covered more of the evidence while assuming less.
And one piece of evidence came out.
On 24 August, Reuters published a story reporting that a group of former internationals — World Cup winners and FIFA legends among them — had called for immediate change at the governing body. It supported the read that was developing here. Wire copies of it are still running on syndication partners around the world today.
Reuters withdrew it the same day. The story body is gone from their page, replaced by a notice saying it was withdrawn due to lack of confirmation of some of the players involved in the group.
It came out of the evidence base immediately. Not held pending clarification. Not softened to reportedly. Not tucked in a footnote because it pointed the right way.
I am telling you about it rather than quietly dropping it, which is the only version of this that counts.
The evidence that flatters your reading is the evidence you have to be hardest on. A source that supports your argument and cannot be confirmed is not a weak source. It is not a source. And you find out whether somebody actually works that way by watching what they do on the day it costs them something.
(Exhibit A in the sealed file reproduces the Reuters page as captured on 25 August 2026. At the time of writing, no third-party archival capture of it exists.)
There is a second thing in that episode, and it belongs to the case rather than to me.
The withdrawn story is still live on wire partners across the world. The correction did not travel as far as the claim did. That is not a criticism of Reuters, who pulled it in hours and said why. It is an observation about how information behaves in a federated distribution system: a correction issued at the centre does not automatically reach the edges.
Hold that thought. It is the same shape as the thing this file is about.
Two maps
Every organization has two of them.
One is the org chart. Who reports to whom, who holds which title, where authority formally sits. It is written down, it is approved, and it is the one that gets shown to the board.
The other is where influence actually travels. Who gets asked. Whose correction changes behaviour. Whose sign-off is what makes a piece of work finished.
Most organizations assume these are the same map.
They are not, and the gap between them is measurable. Five things give it away:
One — escalation routing. Where decisions actually go, by category. Not where the policy says they go.
Two — who resolves ambiguity fastest, and whom people choose when speed matters more than protocol.
Three — whose correction changes behaviour, and whose is absorbed without effect. This is the sharpest of the five. Some people can object and the thing stops. Others object and it proceeds anyway, politely.
Four — where the private clarification networks run. The conversations that happen around the formal chain, because the formal chain is slower or riskier or does not answer the question.
Five — whose approval is treated as the completion condition. Who has to nod before people believe the work is done.
An influence model built from titles maps the fiction instead of the system. The map has to be built from what people do.
Now — before you read another word about FIFA, run those five on your own building. Most leaders discover the two maps disagree somewhere they were not expecting, and the disagreement is usually not a problem. It becomes one under load.
What the record measures
FIFA conducts an unusual amount of its routing in public. Statements, withdrawals, resignations, open disputes about who may speak for whom — all observable. That makes it one of the few organizations where these five proxies can be read from the outside at all.
So here is the measurement. What is observable, what it measures, and what would change it.
A proposal was withdrawn after opposition became public. What that measures: correction travelling from outside the formal governance route changed the outcome. This weakens if reporting shows the withdrawal followed an internal process that constrained it before the public pressure arrived.
The proposal reached public rollout before that opposition became visible. What that measures: formal sign-off was not functioning as the gate that finishes work. This weakens if records show the proposal cleared full internal challenge first.
A member association publicly disputed its confederation’s standing to issue a collective position on its behalf. What that measures: a position travelling on a route the member did not authorize. This weakens if the membership formally ratifies the position, making it authorized after all.
Member associations declining to align with confederation leadership; leadership appealing directly to associations. What that measures: influence moving around the formal chain in both directions at once. This weakens if voting behaviour ends up tracking confederation declarations closely.
Resolution arrived from public pressure rather than from internal process. What that measures: the fastest route to a decision runs outside the governance structure. This weakens if a comparable dispute resolves internally, on a shorter clock, before it ever becomes public.
None of those rows says anything about anyone’s motives. Each one is an action that happened and what the action measures.
And here is the result, stated as measurement rather than as accusation. On these five proxies, the influence map and the org chart do not agree.
That by itself is unremarkable. It is the ordinary condition of most large organizations, which is why the instrument exists. What makes this case useful is not that the gap exists.
It is that the gap is being contested out loud, at more than one layer at once.
The same complaint, one layer down
Here is the reading.
The same authority question is repeating at two levels of the same organization, and the actors on each level are making the identical complaint about the level above them.
A confederation says a decision was made without consulting the members it affects. A member association says its confederation issued a position without consulting it. Both complaints have the same shape. Both invoke democratic legitimacy. Both are about who holds standing to speak for whom.
That is not one dispute at two addresses.
In a federated architecture, the question of representational standing is unresolved at every layer. It becomes visible at whichever layer takes load first.
Four reasons this explains more than the alternatives, and none of them is that I like it better.
It explains the dispute persisting after the triggering proposal was withdrawn. An explanation built on the proposal alone predicts the conflict subsides when the trigger is removed. The trigger was removed. It did not subside.
It explains both directions of routing with one mechanism — associations going around confederations, leadership going around confederations — instead of needing a separate story for each.
It explains why every side can invoke legitimacy in good faith. They are all right about their own layer. That is what makes it structural rather than a dispute between people who are being difficult.
And it requires no assumption about anyone’s motives, which not one of the alternatives can say.
Now the part that matters to you, and it is not about football.
Every multi-site, franchise, regional, or matrixed organization is federated. If your regional heads speak for territories that never authorized the position — if your division leads negotiate on behalf of teams that were not asked — you are running this architecture at a smaller scale. It costs nothing while things are calm. It will show up the first time something takes load, and it will not look like a governance problem when it does. It will look like people being unreasonable.
What I expect to see, and by when
A forecast without a date is not a forecast. It can never fail; it is always not yet. So each of these has a horizon, and each will be marked met, missed, or void in the update that follows it.
By 15 October 2026 — at least one further public dispute about standing to speak for others, at any layer. Not necessarily the same two parties. If the question is structural, it does not stay where it started.
By 15 October 2026 — the conflict continuing past the removal of its trigger rather than subsiding with it.
By 31 December 2026 — declared or formal voting behaviour diverging materially from confederation-level declarations. The coalition on paper differing from the coalition in fact.
Before the March 2027 vote — resolution attempts aimed at the proposal or the personnel, rather than at representational standing, failing to settle it. If the fracture is about who may speak for whom, fixing the thing they were speaking about will not close it.
And a counter, which is different from a forecast
Article 25 of the FIFA Statutes requires the Council to convene an Extraordinary Congress when 20% of member associations request it in writing — the meeting to be held within three months, on an agenda that cannot be altered.
FIFA lists 211 member associations. Twenty per cent of 211 is 42.2. So the threshold is 43. Forty-two associations is 19.9% and compels nothing.
I am showing the arithmetic because the number gates the argument, and a counter with a wrong denominator is not a counter. I had it as 42 in my own notes for a week.
Every update will publish the number of member associations that have publicly expressed dissatisfaction, against that threshold. Counted once each, associations only — not confederations, since treating a confederation as a member would import the exact conflation this file is about. Reversals come off the count and the reversal gets published.
This is reported, never forecast. I am not predicting that an Extraordinary Congress will be convened. I am publishing a number that either climbs or does not, and letting you watch it either way.
One structural note, and it is about how the architecture is built rather than about anyone’s conduct. A suspended association loses its membership rights, voting included. So the denominator is a variable, and twenty per cent of a variable is a moving threshold. Any federated body whose central council administers the membership roll has this property; it is a design feature, not evidence of anything on its own. I am noting it so the counter gets read correctly, and re-verifying the denominator every update. If the membership count holds steady, this observation has no purchase here at all.
A sequence is a signal
This is the part I have written and rewritten more than any other, because it is the part most easily misread.
A senior figure publicly criticised the proposal and subsequently left. Another resigned in protest. Those are matters of public record and of sequence.
What a sequence measures is what the sequence teaches — and it teaches whatever it teaches regardless of what produced it.
This is the piece people find hardest to accept. A system does not need an intention in order to reinforce something. If the visible pattern is that public disagreement precedes departure, that pattern is doing reinforcement work whether anyone designed it, intended it, or would endorse it if you asked them. The next person deciding whether to object is reading the pattern, not the intent behind it.
I am making no claim about why anyone left. I have no basis for one and neither does any other outsider. The claim is narrower and it is about the organization rather than about any person: an observable sequence carries a signal, and signal is consequence rather than intention.
This weakens on a transparent account establishing that the departures were unrelated to the disagreement, together with evidence that challenge routinely changes decisions without cost. That would show the sequence was coincidence and my reading of it wrong.
What everyone else is saying
Most of what has been written falls into four readings. Authority is over-concentrated. Post-reform safeguards have eroded. Funding dependence is shaping the coalition. It is election politics amplified by a failed proposal.
Each explains part of the record. Each is already in circulation, argued by people far closer to it than I am, and I add nothing by restating them in different vocabulary. If that is all this file did, it would be branding.
What none of them explains is why the same complaint appears at two layers simultaneously, in both directions. Over-concentration predicts one direction. Election politics predicts the conflict subsiding once the trigger is gone. Funding dependence predicts alignment tracking money.
The fractal reading explains all three observations with one mechanism and no assumption about motive. That is why it leads. It is not more certain than the others. It is more economical.
When this file closes
Midnight, 31 March 2027, whatever that produces. Updates continue until then, capped at twelve. If it closes without resolution, that is the outcome and I will say so rather than leave it open.
Once the record thickens, this case comes back as a full retrospective — the account of what happened, written when it can be written properly. The forecasts above get graded either way, in a piece at the end of the year whose title is already set: What Would Make Me Abandon This Framework?
What would make me abandon this read
A federated architecture in which representational standing turns out to be settled — where one layer speaks and the others accept it — and in which the current conflict resolves through the formal governance route on the formal timetable, without further disputes about who may speak for whom.
If that happens, this reading is not weakened. It is wrong.
And it would not mean the framework is wrong about federated organizations. It would mean this organization is not the case I thought it was, and the honest response is to say so and take the read off the board.
What comes next, and why it is not another autopsy
A format that only ever finds fractures is not a diagnostic. It is a thesis with a search function.
So the second open file is already chosen, and it is the opposite case. Coca-Cola — a planned chief-executive succession in which the outgoing CEO continues as executive chairman. On the public record it looks like sound architecture: announced early, sequenced deliberately, with a clear signal about who holds what.
It runs the same five proxies. Not a different instrument for a case I expect to go well. If the instrument only ever produces alarming readings, the instrument is broken.
The question there is narrower than does it work, because does it work cannot be falsified and takes years to answer. It is this: formal authority transfers on a date. Does reinforcement influence transfer with it? Whose correction changes behaviour after the handover. Whose approval finishes the work. Whom analysts address when they want to know the direction. Whether the executive chairman’s public role narrows on a schedule, or persists without one.
The falsifier for that positive read has a precedent. Authority transferred while influence stayed, producing a stretch where neither the organization nor the market could tell who set direction. It has happened before, at companies full of capable people with no bad intentions anywhere in the building — which is the whole reason for reading architecture instead of character.
Two files. One instrument. Opposite expectations.
If both come out the way I expect, that proves little. If the FIFA read holds and the Coca-Cola read fails, that proves a great deal — because it would mean the instrument is telling me things I did not want to hear, which is the only evidence that it is an instrument at all.
Updates to this file are appended below, in order, each with its own timestamp.