Corrupted signal, sealed self-reference, structure as veil — and the four stages by which blindness hardens into a shut door. The Great Homecoming research programme · June 2026
Before a system fails, it goes blind: it loses the ability to see its own condition. We find that blindness arrives by exactly three routes. The signal is corrupted — the data itself is bent before it reaches the decision point. The system seals into self-reference — it stops consulting anything but itself. Or the structure becomes the veil — the org chart stands between the leadership and what is actually happening. Whichever route it enters by, the hardening then proceeds through the same four stages, from residue to seal. The last stage has a property most collapse thinking misses: the point of no return is internal. The correction loop is bricked up from the inside, by the system’s own accumulated refusals — so the shock that eventually arrives does not shut the door; it finds it shut. Each route has an observable signature, readable now, in your own organisation, without waiting for a crisis.
Walk back through any large failure and you find the same ordering: long before the system stopped working, it stopped seeing. The information that would have forced a correction existed, and did not arrive where decisions were made — or arrived in a form that could be ignored. The failure of the organs came after the failure of the eyes.
That ordering is what makes blindness worth measuring: the collapse is hard to time, but the blindness is observable in the present tense. What follows is the part of our instrument that reads it — three routes in, one ladder down, and a door.
We have tested whether the three routes are actually distinguishable inside the engine, and an early sealed result indicates they are. The method, the cases, and the honest status of that result are set out in the research update.
The first route runs through the data itself. Nothing is hidden, exactly — it is bent. Bad news is softened at each level it passes through, because each level is graded on the news it delivers. Metrics are gamed, because the metric became the target. By the time the information reaches the decision point it is still information-shaped, which is what makes this route treacherous: the dashboard is full, current, and wrong.
The signature we read for it is divergence between volume and quality: reporting increases while its predictive value falls. Measurement multiplies on this route — more reviews, more metrics — precisely because no single measurement is trusted any more.
The second route does not bend the signal; it closes the intake. The system comes to consult only itself. Leadership cites leadership. Outside voices are reclassified — first as uninformed, then as hostile — and internal dissent is reclassified as disloyalty, which teaches everyone watching what dissent costs. The system still processes enormous amounts of information; all of it now originates inside, or is selected for agreeing with what is inside.
The signature is a rising share of self-reference: in what gets quoted, who gets promoted, which evidence is admissible. A practical probe: trace the last five major decisions and ask what outside input each one actually turned on. In a sealed system the honest answer is none.
The third route is the easiest to miss, because nobody is lying and nobody has closed their ears. The structure itself has become the screen. The org chart, the process map, the reporting lines — built, once, to make reality manageable — now stand in place of reality. Every box reports green because every box is answering its own narrow question correctly, and the failure lives in what no box owns. Procedure is followed; the purpose it served has quietly left.
The signature is the strangest of the three: reforms fail even though the data is clean and intentions are good. Initiatives launch, comply with every process, and change nothing — the inherited form absorbs them. Where you see repeated, well-run, ineffective reform, suspect the veil.
The three routes are doors in; the hardening is one ladder down, the same ladder whichever door was used. The mechanism: every error a system declines to correct leaves a deposit, and the deposits accumulate on exactly the surface that corrections must pass through.
Stage one — residue. Errors are deflected rather than processed, but the layer is thin. Corrections still land; they just cost more than they should. Most organisations live here much of the time, and from here recovery is unremarkable.
Stage two — the divided state. The system half-knows. It can describe its own problem — often eloquently, at its own offsites — and does not act on the description. Correction is acknowledged, then dissolved. The say-do gap becomes the dominant fact about the place.
Stage three — the film over the eyes. Incoming correction is no longer resisted; it is reinterpreted. Criticism confirms the critics’ bias; failure confirms the need for more of the current course. The system’s picture of itself has become self-repairing — evidence bounces off it.
Stage four — the seal. The loop is closed. Correction attempts of any size are absorbed without effect, and — the diagnostic mark — the response no longer varies with the correction’s strength or quality. Nothing said, shown or done from inside reaches the part that steers.
The stage is read from one thing: what happens to correction attempts. Not from sentiment, not from results — results may be excellent throughout, funded by reserves (see “The Two Clocks”) — but from the fate of the last several serious attempts to tell the system it was wrong.
This gives the point of no return a definition that does not depend on knowing how the story ends. It is not an event, and it is not out in the world of shocks. It is the fourth stage: the correction loop bricked up from the inside, brick by brick, each brick a refusal the system chose. No enemy does this to a system. It is self-laid masonry.
So when the shock arrives, it does not shut the door. It finds the door shut. The same blow lands on a neighbour whose loop is open, and the neighbour corrects, adapts, absorbs. And a sealed system does not even need the shock: with nothing renewed and the loop closed, reserves simply run out, and the end comes triggerless, from exhaustion. The shock is where the story gets pinned afterwards; the door was decided earlier, from inside.
The probes are not exotic. Trace the last three pieces of genuinely bad news: did they arrive intact, softened, or not at all — and what happened to the people who carried them? List the outside inputs your last five decisions actually turned on. Count well-run reforms that changed nothing. Then watch one live correction attempt end-to-end and note which stage its fate describes.
The three routes leave different signatures, and the difference is readable in the record before the outcome is known — which is what makes the typology a diagnosis rather than a label. An early sealed test indicates the signatures do separate inside the engine; the method and its status are set out in the research update.
The Great Homecoming is an independent research programme on why systems cohere or fragment. Companion pieces: “Why Systems Don’t Die From Shocks,” “The Two Clocks,” “When Repair Works — and When to Stop.” The instrument described is under live forward test; we describe it that way until its register adjudicates. Contact: Wim Van Laere.