Surveys, cohesion indices and values studies aren't the wrong answer — they're the wrong instrument for this object. The structural reasons they miss the turn, and what we read instead. The Great Homecoming research programme · June 2026
A survey asks people how they feel, averages the answers, and reads one side of the story, late. For the question of whether a system is cohering or quietly coming apart, that is not a wrong answer — it is the wrong instrument. This piece sets out the structural reasons the conventional cohesion and values instruments cannot see the turn: they read the spent level rather than the moving structure; they arrive a generation late; they cannot tell genuine cohesion from manufactured; they cannot tell healthy cohesion from hardened; and they cannot add up from the individual to the whole. We read the other side instead — the effective side, from conduct, as a profile, across levels. It is a superiority of reading — we see structure the other instruments discard, and earlier — demonstrated on cases, not yet a validated forecast.
Cohesion radars, life-satisfaction scores and values surveys are careful, useful instruments. The point here is not that they are inaccurate; it is that they are pointed at the wrong object. They capture what people say and feel — the apparent side of a system — and they capture it as an average. What actually carries a system's direction is what it does, read structurally, across levels. No amount of refinement to a questionnaire reaches that, because the thing being measured is not available to self-report: no one can tell you, on a five-point scale, whether their society is binding or hollowing. What follows are the structural reasons the gap is unbridgeable from the survey side — and why a different kind of read is needed.
Trust and cohesion levels are the spent echo of a turn that already happened in conduct — often a generation earlier. Stated values lag the real shift by years, because people report the world they were formed in, not the one their behaviour has already moved to. By the time a change of direction surfaces in how people answer a questionnaire, the reserves that were masking it are nearly gone, and the room to act has mostly closed. The instrument is not reading the present; it is reading a slow, lagging awareness of a past the structure has already left. We catch the turn in the conduct, while there is still time to act.
Real, bottom-up cohesion and manufactured, top-down or mass-movement cohesion produce the same survey number. A community bound by genuine, earned trust and a crowd bound by fervour both score "high." A society can register strong cohesion because it is genuinely well-bound, or because it has been swept into a unifying enthusiasm that will not hold — and the questionnaire cannot separate the two, because both feel like belonging from the inside. That difference — earned versus manufactured consent — is not a detail at the margin; for many of the systems that matter, it is the whole question. Only a structural read, taken from conduct rather than sentiment, can tell them apart.
Even setting aside genuine versus manufactured, a single cohesion score cannot distinguish healthy, generative cohesion from defensive or even pathological cohesion. A society can score "highly cohesive" precisely because it has hardened into hostile, internally-loyal camps — bonding fiercely within while the bridges between disappear. The instrument reads a symptom of stress as a sign of health, and reports the hardening as strength. A structural read separates generative cohesion (binding that builds shared direction) from defensive and pathological cohesion (binding that is really fracture in disguise).
Cohesion does not sum. A country can be richly bound in its families and neighbourhoods and still fail to hold together as a whole — the binding at one level simply does not carry to the level above. Averaging individual responses upward hides exactly this: it reports a healthy average over a system that is not binding at the level that matters. The number looks reassuring precisely because the failure lives in the aggregation the method performs. Reading across levels — person, family, community, institution, state — is the only way to see when genuine cohesion at the bottom is failing to carry upward.
Underneath all four of these sits the same structural fact: a survey holds only the apparent side. It captures what is professed and felt; it has no access to what is enacted, and therefore no access to the gap between the two — the say-do gap — which is itself one of the sharpest signals of where a system is heading. And what it does measure, it often mis-attributes: life-satisfaction scores track income and health, not meaning; meaning scores track a personal resilience-buffer, not whether a society's worldview still holds together. None of these is a measure of structural coherence, and none can become one by being asked more carefully.
We read the effective side — from conduct and structural footprints rather than from sentiment: rule-of-law, fragmentation, how fast institutions form and decay, whether what a system professes matches what it does. We report it as a profile, not a single score — where coherence binds, where it breaks, and how fast each is moving — because the single number is exactly what lets a fatal weakness hide behind strong averages. And we read it across levels, so that genuine binding at the bottom failing to carry upward becomes visible rather than averaged away. That locational, conduct-based, multi-level read is the position no survey or index occupies — not because they are done badly, but because it is not the object they are built to measure.
This is a superiority of reading: we see structure the other instruments discard, and we see it earlier — demonstrated on cases, not yet a validated forecast. The distinctive predictive claim rides on the forward tests still in progress, and we keep that distinction in plain view. "Sees what they miss; validation in progress" is the honest and defensible line, and we hold ourselves to it: the instrument is under live forward test, and we describe it that way until its register adjudicates.
The Great Homecoming is an independent research programme on why systems cohere or fragment. Companion pieces: "The Three Ways Systems Go Blind," "The Two Clocks," "Why Systems Don't Die From Shocks." The instrument described is under live forward test; we describe it that way until its register adjudicates. Contact: Wim Van Laere.