Martro Observatory is an independent applied research project. Its object is structural diagnostics for organizations operating under deep uncertainty — conditions where the probability distribution over future states is either unknown or non-stationary, and where forecasting-based decision frameworks systematically underestimate irreversibility.
The dominant failure mode in organizational strategy is not inaccurate forecasting. It is structural ignorance: hidden coupling between decisions, delayed irreversibility, and capacity decay that becomes visible in financial metrics only 24–36 months after the commitments that caused it. The observatory addresses this gap directly.
The appropriate response to deep uncertainty is not prediction-enhancement but structure-redesign: preserving optionality, minimizing premature irreversibility, and building governance that learns. The observatory does not prescribe actions or optimize an objective function. It maps irreversibilities, fragility, and future room for maneuver — before strategic commitments become irreversible.
Observed utilization does not demonstrate structural capacity. An organization running at high utilization on degraded invariants is consuming option value faster than it accumulates it. This distinction — between throughput capacity Y and realized production q·Y — is not a modeling convention. It is the structural condition that makes irreversibility measurable and door-closing decisions detectable before they compound.
The observatory's theoretical foundations draw on Peters's non-ergodicity framework (time-average versus ensemble-average economics), Shannon's entropy as a measure of dispersed futures, Teece's dynamic capabilities, Ashby's law of requisite variety, Aubin's viability theory, and Lempert's robust decision-making under deep uncertainty.
The organization is modeled as a constrained stochastic dynamical system with state vector s = (x, z, G, m). Strategic optionality is defined formally as the entropy of good future states — the variety of structurally viable future configurations weighted by a regime-dependent quality functional — estimated through multi-scenario simulation rather than market prediction. This yields a quantitative, testable notion of door-closing irreversibility and robust viability under non-stationary environments.
The framework's validity claim is narrow and stronger for it: given explicit structural assumptions, it provides a formal, falsifiable, and operational way to detect door-closing decisions and quantify preserved optionality. It does not claim a universal objective function, causal inference from survey data, or truth independent of its assumptions.
The theoretical frame is operationalized through three instruments with distinct observation scopes and data sources, designed to be run jointly or independently depending on access constraints.
Internal observation instrument. Multi-role structured survey across ten macro-areas, producing five structural invariants (p, c, τ, r, i), role gap analysis, fracture coefficients, and the initial hysteresis proxy z₀. Seeds the MARTRO dynamic model via an explicit, auditable weight matrix W.
External observation instrument. Evidence-based screening tool that derives structural invariant estimates from publicly observable signals — without internal access. Enables out-of-sample validation and cross-case structural comparison under limited data conditions.
Optionality propagation model. Integrates NORN-derived internal state and EOS-derived external state into the MARTRO dynamic model, runs multi-scenario simulation, and produces ranked optionality trajectories with explicit sensitivity bounds.
Every component of the framework — the five invariants, the weight matrix W, the hysteresis parameters, the viability thresholds — is a stipulation made explicit, documented, and revisable. Parameters are transparent; formal identifiability is acknowledged as an open gap, not a resolved claim. Empirical calibration of W is planned post-pilot via sensitivity analysis protocols consistent with Saltelli et al. (2004).
Falsifiability is structural, not predictive: the framework is evaluated through backtesting on historical cases, ranking stability under parameter and scenario variation, and correlation between rising hysteresis signals and observed constraint escalation. Instability in rankings is treated as structural information, not noise to be averaged away.
