Theory

The primer is the lay introduction. The essays develop the framework through cases. The preprint When Is Trust Possible: Toward a Geometry of Coordination argues the full structural claim and is available on SSRN and SocArXiv.

This page sits between them. It is the condensed structural account, written for a reader who wants more than an introduction but less than a forty-page paper. It explains not only what the framework claims but why it has the shape it does, what it competes with, and what it does not yet know.

Trust is not an internal feeling. It is an emergent geometric property of systems. It arises when architectural conditions sustain structurally proportionate relationships across five dimensions — Embodiment, Anticipation, Reciprocity, Feedback, and Tension — under load, among all materially affected participants whose participation remains structurally contestable.

Proportionate means that no dimension expands, contracts, or concentrates beyond the capacity of the remaining dimensions to absorb and adjust to it. When the proportions hold, trust emerges. When they do not, trust collapses, regardless of how competent or well-intentioned the system appears from the outside.

Any agent acting under uncertainty in a coordination system faces five irreducible structural questions before coordinated action is possible. Where do consequences land? When does the outcome arrive relative to the decision that produced it? On what terms does exchange occur, and can those terms be contested? Do signals about system state return to the agents who produced them? And where does the strain generated by coordination go? The five dimensions are the structural answers.

The number is not chosen. It is derived. The five dimensions correspond to five primitive structural features of any inhabited coordination space: the locus of consequence, the timing of outcome, the terms of exchange, the return of information, and the routing of load. These are space, time, exchange, information, and load — the irreducible coordination invariants. A system that answers all five has specified the geometry of coordination at the level below norms, intent, or enforcement.

The decomposition is provisionally complete, but it is also falsifiable. A proposed sixth dimension would qualify only if it satisfied four conditions: it cannot be expressed as a special case of an existing dimension; it produces a recognizable failure pattern that occurs with the other five intact; its repair is categorically different from repairing one of the five; and it matters across multiple domains, not just one institutional setting.

The major candidates have been evaluated against these criteria. Power functions as an upstream shaper of geometry, not a dimension within it; its trust-relevant effects appear as distortions in Embodiment, Reciprocity, Feedback, and Tension. Meaning changes why people participate but not the structural shape of coordination. Identity operates upstream of the geometry as a condition for participation, not as a coordination invariant. Information collapses into Feedback. None of the candidates produces a distinct failure signature with the other five intact.

The four admission criteria are not independently chosen standards. They are the operationalization of the question that generated the five primitives in the first place: is this an irreducible structural question that any coordination system must answer? The same test that admits the five rules out the candidates so far proposed.

Trust behaves as a conjunctive, non-counterbalancing interaction across the five dimensions. The framework represents this structurally as:

The expression is not a calibrated empirical equation. It is a formal hypothesis about the functional structure of trust. It encodes three commitments, each of which follows from the irreducibility of the five dimensions.

Conjunctive necessity. All five dimensions are required. If any approaches structural failure, the product collapses. Addition would imply that strength in one dimension can offset failure in another — and the irreducibility argument forbids that move.

Non-counterbalancing behavior. Excellent Reciprocity cannot rescue zero Embodiment. Fast Feedback cannot counteract displaced Tension. Each dimension does its own structural work; none can do another's.

Interaction fragility. Moderate degradation across multiple dimensions compounds rather than averaging out. Distributed erosion produces sharper decline than isolated weakness.

An alternative formulation — the minimum function, T = min(E, A, R, F, τ) — would encode pure weakest-link dynamics. But it would treat a system with one failing dimension and four intact ones as equivalent to a system in which all five degrade moderately. The multiplicative form preserves weakest-link behavior (as any variable approaches zero, the product approaches zero) while also encoding the compounding fragility that distributed degradation actually produces. Other functional forms — threshold models, logistic models, hybrid forms — remain on the table empirically. Comparison against alternatives is part of the open research agenda.

The five dimensions are analytically distinguishable but not causally isolated. Distortion in one dimension propagates into others through structural mechanisms. Three cascade patterns are theoretically distinct and empirically observable.

Each cascade can become self-reinforcing when propagation reaches a dimension that itself affects the initiating one. The Reciprocity–Tension–Reciprocity loop is the most consistent: asymmetric reciprocity overloads tension; tension failure produces withdrawal; withdrawal further degrades reciprocity; remaining participants absorb the increased load. Each cycle is larger than the last. This is the structural mechanism behind the trust "snap" — the moment when collapse appears sudden but has in fact been accelerating along a self-sustaining loop for some time.

Cascade dynamics explain why localized distortion often produces system-wide failure disproportionate to the initial degradation. Trust geometry is not modular. A repair intervention that targets only the initiating dimension will fail if cascade has already activated a self-reinforcing loop in a secondary dimension.

The governing thesis carries a boundary condition. Trust is only evaluable among participants whose participation remains structurally contestable — that is, whose refusal is a real option. Where refusal is not available, what looks like trust is something else.

Contestability fails when exit would mean durable loss of livelihood, legal standing, bodily safety, or functional participation capacity. A system that runs smoothly only because the people inside it cannot leave is not a trust system. It is a coercive one. The framework distinguishes them to prevent a common mistake: reading stability as trust when the stability is enforced by the absence of alternatives.

Contestability is structural, not motivational. The test is not whether participants choose to stay but whether they could meaningfully leave, refuse, or challenge the terms. A system can be peaceful, productive, and entirely uncontestable. That system is reliable. It is not trustworthy in the framework's sense.

Reliability and trust are separable properties. A system can be reliable without being trustworthy. Reliable means its behavior is predictable. It does not mean the behavior is one you can survive, lean on, or meaningfully participate in.

The Reliability Trap is the pattern in which a system becomes more predictable as its geometry becomes more extractive — feedback is suppressed, tension is displaced onto those who cannot push back, reciprocity narrows to one-way flow. Performance metrics improve. The system appears to be working. And then it fails suddenly, because the proportions that make correction possible have been optimized away.

The framework distinguishes carefully between what it claims as structural argument and what it identifies as open empirical work.

Settled by argument. The five dimensions are irreducible coordination invariants. The decomposition is derived, not assumed. The multiplicative form follows from the conjunctive necessity of each dimension. Cascade patterns describe propagation mechanisms that follow from coupling. The boundary condition — that trust is only evaluable under contestable participation — is a definitional commitment of the framework, not a hypothesis.

Hypothesized, awaiting empirical work. The multiplicative formalism specifies functional structure but is not yet parameterized. Whether the multiplicative form, a threshold model, a logistic form, or a hybrid best describes observed trust dynamics across domains is an empirical question. Domain-specific weightings (the exponents) are similarly hypothesized, not estimated. The full empirical agenda runs to five mechanistic hypotheses developed in the preprint.

Genuine open problems. Two terms remain less specified than the framework would like. Proportionate is defined in terms of the remaining dimensions' capacity to absorb and adjust — which depends on a definition of capacity that the framework does not yet ground independently. Social proprioception — the framework's account of how embodied agents perceive geometric distortion before they can name it — is presented as a transduction hypothesis, not as established perceptual science. Both are flagged as such in the preprint and treated as research agenda rather than as load-bearing claims.

The framework is falsifiable. It would be refuted by a system demonstrating sustained coordination without defensive monitoring, proportionate participation across all materially affected parties, stable adaptation under load, and structurally suppressed EARFT in one or more dimensions. If such a system exists, the necessity claim fails.

The Geometry of Trust is a structural precondition theory. It does not replace existing trust models. It operates one level below them, asking under what conditions the things they describe become possible.

Mayer, Davis, and Schoorman model trust as a willingness to be vulnerable based on perceived ability, benevolence, and integrity. The framework does not contest this. It addresses a prior question: under what systemic conditions can such judgments stabilize over time? Ability, benevolence, and integrity are content dimensions of trust evaluation. The geometry describes the structural conditions that make evaluation coherent across time and scale.

Niklas Luhmann conceptualizes trust as a mechanism for reducing social complexity. The framework is compatible with this functional account but shifts analytical focus. Luhmann explains what trust does for systems. The geometry examines when systems possess the structural conditions necessary for trust to emerge in the first place.

Guido Möllering reframes trust as a process of interpretation and suspension — the leap that bridges uncertainty. The framework does not model the cognitive leap directly. It asks what structural conditions render such suspension reasonable rather than reckless.

Elinor Ostrom identifies institutional design principles that enable cooperation in common-pool resource systems. The framework does not reject these principles. It compresses them. Ostrom's design principles can be read as EARFT-compatible structural arrangements: shared rule-making is Reciprocity; monitoring is Feedback; graduated sanctions are Tension metabolization. The geometry explains why those arrangements work when they work, and why functionally similar arrangements fail when the geometry around them does not hold.

The framework's contribution is the precondition claim itself — that the conditions under which trust becomes possible have a specifiable structural shape, and that the shape is the same across domains. The lineage is mapped in more detail in the knowledge tree and developed at length in the preprint.

The full structural argument is developed in the preprint When Is Trust Possible: Toward a Geometry of Coordination, hosted on SSRN and SocArXiv. The preprint contains the formal argument, the dimensional independence proofs, the case illustrations, the diagnostic indicator tables, and the extended empirical agenda. Direct links will be posted here once available.

The framework is in active development. The essays continue. The empirical agenda is open. Researchers, practitioners, and serious readers with cases the framework should be tested against are welcome to get in touch through the about page. Pushback is welcomed and engaged with substantively.