Trust Mining

Normative

Full Practice · Bond · Cooperating Under Bad Faith

Mechanism

Trust Mining is the practice of building up trust capital specifically in order to spend it. The pattern has two phases. The first is accumulation: the actor passes calibration tests, demonstrates alignment, earns authority through visible competence, patiently, sometimes over years. The second is extraction: at the point where the accumulated authority is high enough to be worth spending, the actor redirects the institution's resources, norms, or mission toward an outcome the framework was never designed to sanction.

What makes the pattern hard to catch is that the accumulation phase is structurally indistinguishable from an ordinary career inside a cooperative system. The behaviors that produce earned trust — passing calibration tests, demonstrating alignment, contributing visibly, building relationships, accumulating credibility through results — are the same behaviors the system was designed to reward. From inside the accumulation phase, the actor looks like a high-functioning member who is doing exactly what the system asks. There is no signature that distinguishes pre-extraction accumulation from ordinary cooperative behavior. The signature only becomes legible at the inflection point, and the inflection point is rarely announced. The turn is usually recognized only in retrospect.

The accumulation phase is the cost of the extraction phase, paid in advance. A long, visible, expensive career inside a cooperative system is the price of admission to a level of authority worth spending. Trust mining is what an iterated-defection strategy looks like when the rounds are years long and the payoff from a single large defection exceeds the discounted value of continued cooperation. The math is Axelrod's, applied to a substrate Axelrod did not model in detail: the substrate of an actor whose time horizon is long enough to absorb the cost of cooperation as a calibration expense.

The institutional form of the same pattern is capture. Senior staff rotate through industry positions and back into the regulatory agency, or through advocacy organizations and back into the institutions those organizations were created to oversee. Each individual career move is defensible. Many are admirable on their own terms. The aggregate effect is that the institution stops producing decisions that constrain the captured interest in meaningful ways, while the formal structure persists for years after the substantive function has been hollowed out. Institutional capture is trust mining run at scale, through legitimate channels, over time. The capture is invisible step by step because step by step nothing unusual is actually happening. It is the pattern, integrated across years, that produces the outcome.

The specific vulnerability the mechanism reveals is structural: calibrated trust, the Bond's own prescription, is gameable by an actor patient enough to pass the calibration. The calibration runs on behavior over time, and a sufficiently patient actor can produce behavior over time. The defense is not to abandon calibrated trust — that returns the framework to either naive openness or suspicion as default — but to add a layer of pattern recognition that operates on the career-arc scale, not on the within-interaction scale where calibration normally runs.

Practice

The diagnostic question is: "Does the alignment-demonstrating behavior I am rewarding fit a pattern consistent with what an accumulator would do, regardless of present intent?"

The question is not whether the actor is currently accumulating trust strategically. That question is unanswerable from outside, and asking it is often the move that turns a legitimate institution paranoid. The question is whether the system's calibration is running on a time horizon short enough that a patient actor could game it without ever producing a signal the calibration can see. Three practices test the calibration itself.

The three practices share a discipline: they operate on the system's calibration architecture rather than on any individual actor's behavior. The shift in object — from "is this person trustworthy" to "is this calibration gameable" — is what keeps the tool from drifting into the failure mode it most resembles. Asking whether a person is currently accumulating trust strategically is paranoia. Asking whether a system's calibration is structurally vulnerable to patient strategic accumulation is institutional design.

In the Wild

A community built on cooperative epistemic norms — rigorous argument, charitable engagement, meritocratic allocation — accepted a major donor whose fluency in both the community's language and its values was unusually high. The accumulation phase ran for years. Calibration tests were passed at every level. Resources were allocated, authority extended, and the donor's apparent alignment was reinforced by visible competence and visible generosity. The inflection point arrived large, and the extraction phase revealed misappropriation at a scale that surprised even the actors who had watched the alignment most closely. The community's post-mortem was less about whether the failure had happened than about how an actor so deeply embedded had operated undetected for so long. The answer pulled together exactly the architecture this tool names: calibration running on the alignment dimension, no calibration running on the inflection-point dimension, and a payoff structure that rewarded a sufficiently patient actor's restraint until the payoff was worth abandoning.

A regulatory agency was created to oversee a powerful industry. Over two decades, senior staff rotated through industry positions and back into the agency. Each individual move was defensible. The aggregate was capture: the agency stopped producing decisions that constrained the industry in meaningful ways. The capture was institutional rather than personal. No one set out to capture the agency. The career incentives, the informational asymmetry, and the social dynamics of repeated interaction did the work the formal structure was designed to prevent. By the time the capture was visible in the outcomes, the people best placed to name what had happened had internalized the institutional norms that made naming it difficult.

A small team operating with high mutual trust caught its own drift. One member's contributions had been received as correct by default for months while another member's were increasingly examined carefully. The asymmetry had nothing to do with the quality of the work. It was about accumulated authority. The team had quietly started treating visible alignment as evidence of competence. They surfaced the pattern before it could institutionalize, and corrected it by making the basis for their assessments explicit and by routing the assessment through structural criteria rather than through the calibration the alignment had quietly captured. Trust mining at the small-group scale, caught early, before the inflection point arrived.

The defense against trust mining is not vigilance about individual actors. It is calibration architecture that operates on horizons long enough to detect the pattern and that intensifies its monitoring as authority approaches the inflection point. The Bond's commitment to calibrated trust is not weakened by adopting the tool; the tool operationalizes the commitment at the scale where calibration has to work against a patient adversary.

What the tool asks of the practitioner is to treat the calibration as a designed object rather than as a posture. A posture cannot defend against a multi-year accumulation strategy. A calibration architecture that runs on the right time horizon and intensifies where the stakes intensify can. This is what calibrated trust requires once the practitioner accepts that some of the actors the framework will encounter have come prepared.

Lineage

Robert Axelrod's The Evolution of Cooperation (1984) established the game-theoretic foundation. Tournament results showed that cooperative strategies outperform defection across iterated interaction under specific conditions: a long shadow of the future, reliable reputation, and the possibility of retaliation against defection. Trust mining is the defection strategy that exploits the conditions themselves. The actor pays the cost of cooperation for long enough to make the payoff from a single large defection exceed the discounted value of continued cooperation, and then defects in a domain or at a scale where retaliation cannot be calibrated against the accumulated trust.

Mancur Olson's The Logic of Collective Action (1965) and Elinor Ostrom's Governing the Commons (1990) developed the institutional economics of the vulnerability. Olson's work on free-riding named the structural exposure: when the benefits of cooperation are diffuse and the costs of monitoring are high, actors who defect while appearing to cooperate can extract substantial value before detection. Ostrom's work on commons governance mapped the institutional defenses — graduated sanctions, monitoring designed into the structure, clearly defined membership — that the practices on this page descend from.

George Stigler's The Theory of Economic Regulation (1971) documented the institutional-capture form empirically. Regulatory bodies systematically come to serve the industries they regulate, not through conspiracy but through career incentives, informational asymmetry, and the social dynamics of repeated interaction. The capture proceeds through the legitimate channels the system was designed to use, which is what makes it hard to surface from inside while it is happening.

The behavioral-economics work on patience and time-discounting underlies the structural claim about adversarial time horizons. Actors with longer effective time horizons can absorb larger calibration expenses, which means systems with payoff structures that grow over time without proportionate monitoring increases are systems that select for patient adversaries.

Cross-references

Within the category. Trust Mining operates under the lens of Adversarial Dynamics — the category's architectural framing of the two adversaries and the diagnostic question of whether the other party is accountable to the same norms they are invoking. Trust Mining and The Cooperative Vulnerability are structural siblings: Trust Mining names the temporal architecture of strategic accumulation; The Cooperative Vulnerability names the rhetorical architecture of weaponized cooperative practice. The two run in tandem in many real cases, with accumulated trust providing the platform that the asymmetric invocation of cooperative norms operates from.

Across the Workshop. Calibrating Trust to Behavior is the Bond category Trust Mining most directly games — calibrated trust is the Bond's prescription, and Trust Mining is the strategy designed to pass the calibration. The relationship is structural rather than oppositional: Trust Mining does not refute calibrated trust as a discipline; it specifies the calibration architecture calibrated trust requires once a patient adversary is in the population. The reframing the lens performs on Calibrating Trust to Behavior — adding the inflection-point and career-arc dimensions to the standard within-interaction calibration — is what the lens delivers to that category's practice.

Limitations. Two worth naming. First, Trust Mining analysis can drift into reading every long career as a potential accumulation strategy, which produces exactly the suspicion-as-default failure the category is built to refuse. The defense is the shift in object the Practice section names: the question is the calibration architecture, not the actor's intent. Second, the tool's structural focus may underweight cases where the accumulation-and-extraction pattern emerges without strategic design — an actor who genuinely intended cooperative practice and drifted into extraction under structural pressure, which is the structural adversary's form rather than the intentional adversary's. The same defenses work on both, but the diagnostic vocabulary should not collapse them into a single account.