Stakeholder Stratification

Large infrastructure projects are governed by distributed decision systems: configurations in which multiple independent stakeholders—each with distinct mandates, evaluation frameworks, and definitions of acceptable risk—must simultaneously reach acceptable outcomes within their own domains. No single actor controls the result. No single intervention is sufficient to produce success. Analysts routinely document projects that exceed cost and schedule estimates, that collapse after years of apparent progress, or that succeed technically but fail commercially or politically. Project failure patterns are not explained by individual factors alone, which are primarily symptoms of the structure in which those factors operate.

This report corrects several specific misunderstandings and misframings that have been adopted broadly. Authority in large infrastructure is not centralized in any one institution. Progress in one domain does not advance the system unless other domains converge. Partial progress is the expected pattern, not an anomaly.

Progress, and failure, are the result of a more interconnected and interdependent structure, and need a different framework. Project realization depends on non-fungible approvals—decisions across legal, commercial, financial, technical, political, and social domains that do not substitute for one another—and outcomes are determined by joint satisfiability, the requirement that all approval conditions hold simultaneously.

In parallel, feasibility is bounded by constraints that no single actor controls. Some are genuinely external (i.e., exogenous), such as macroeconomic conditions. Others are quasi-exogenous: supply chain capacity and infrastructure limits produced by the system's own institutional structure and demand uncertainty, yet experienced by each participant as external conditions they did not create and cannot individually resolve. Still others are internal, manufactured by the system's own risk-allocation decisions. Even when alignment is achieved, execution depends on conditions that can change after approvals have been secured.

Importantly, the concept of “risk” is an institution-specific construct: stakeholders may not disagree about the magnitude of a common risk, but evaluate different categories of uncertainty entirely.

The implications are structural. Single-domain interventions displace constraints rather than resolving them. Stakeholder conflict reflects different risk definitions, not different facts. Demand-side acceleration exposes bottlenecks more quickly; it does not remove them. Effective action requires understanding not only where a constraint appears, but how it is connected to other domains and where it will re-emerge when displaced.

Nuclear energy projects are the primary case in this report because the distributed decision dynamics are most visible there: the number of independent approval domains is large, the time horizons are long, and the consequences of domain-level failure are well-documented. However, the mechanisms are not nuclear-specific.

This report does not attempt to quantify the probability of any specific project’s success, prescribe particular interventions, or rank alternative approaches. It develops a structural framework intended to support more accurate analysis, more robust project design, and more disciplined evaluation of policy and institutional interventions.