Structural diagnostics under deep uncertainty

Position

MARTRO is not a generic advisory wrapper. It is a decision-support layer built on structural interpretation, explicit assumptions, and inspectable logic. It does not predict outcomes. It measures preserved optionality, decision irreversibility, and structural fragility — in time to act.

The system is white-box by design: every output is traceable to the evidence that generated it, every assumption is documented, and every limitation is stated explicitly. The framework is designed to be contested and improved through evidence, not defended through authority.

Framework architecture

The MARTRO–NORN stack is a two-layer system. MARTRO is a constrained stochastic dynamical model representing the organisation as a state vector s = (x, z, G, m), where x captures five structural invariants, z encodes accumulated hysteresis (irreversible commitments), G drives long-run capacity, and m indexes the current market regime.

The five structural invariants — predictability (p), cognitive margin (c), transferability (τ), resilience (r), and inspectability (i) — are the primitive layer. They are not KPIs. They are the conditions under which any strategy remains executable over time.

NORN is the diagnostic interface: a dual-channel instrument combining a multi-role structured survey (Owner/Leadership, Staff, Operators) across ten macro-areas with a canonized interview method. NORN outputs seed the MARTRO initial state via an explicit, auditable weight matrix W. Structural fractures and role gaps become the initial hysteresis proxy z₀. The system is instrumented from evidence, not calibrated to expectations.

Operating posture

NORN is the operational interface where instruments are run and interpreted. During this first publication phase, workspace and admin layers remain intentionally constrained. The sequencing is deliberate: methodological integrity before commercial exposure, calibrated growth over rapid scaling.

The framework does not substitute professional judgment. It does not promise ROI or make causal inferences. It produces structural intelligence under deep uncertainty — legible, contestable, and bounded.

Project origin

Martro Observatory originated to address a recurrent structural blind spot: critical decisions are often made from utilization and outcome metrics that do not reveal underlying capacity decay. By the time the damage is visible in financial KPIs, the decisions that caused it are 24–36 months in the past and no longer reversible.

The project grew from ten years of direct micro-enterprise management in the wood-furniture supply chain, two years observing risk governance in a world-leading luxury company, and a methodological formalization process that produced the MARTRO–NORN framework. The theoretical foundations draw on Peters's non-ergodicity framework, Shannon's entropy measure, Teece's dynamic capabilities, Ashby's law of requisite variety, Aubin's viability theory, and Lempert's robust decision-making.

Current state — March 2026

The methodological construction phase is substantially complete. The system is now in the convergence phase between validation, go-to-market, and demonstrable commercial traction, with June 2026 as the current phase milestone.

On the product side: one active Clew pilot with a real client is at 75% completion; the Weft stack simulator is in development at 40%; pitch materials for seed and commercial partners are complete and in circulation. A European sales advisor with a travel/risk sector network has confirmed strong interest in a Weft go-to-market partnership. A commercial agent is operationally active for acquiring paying pilots.

On the validation side: multi-batch out-of-sample retrospective validation is underway. Metric A (directional coherence) and Metric C (fragility overlap) are stable and robust across cases. Metric B (risk band accuracy) contains residual mismatches concentrated on intermediate and transitional cases — interpreted as a credible analytical result, not a failure. An ex-post validation case on Meta Platforms 2023 has been completed and documented.

On the academic side: a three-paper publication strategy is mapped. The EOS and bootstrap methodology paper is the most immediately submittable target. The SVT theoretical paper is ready for SSRN deposit to establish originality claims. The integrated system paper requires 18–24 months of additional empirical work. Formal parameter identifiability and systematic sensitivity analysis on the quality functional remain open areas, acknowledged explicitly as known gaps.

Demonstrated properties

• Out-of-sample directional coherence (Metric A) validated across multi-batch retrospective dataset. The signal is stable across structurally distinct cases and did not require post-hoc adjustment.
• Optionality trajectory predictability emerged as an unoptimized structural consequence of the invariants-based architecture — validated as an unintended result, not an optimized fit. This is treated as a significant finding.
• White-box framework: inspectable assumptions, contestable outputs, explicitly documented limitations including stipulated weight matrix W and formal parameter identifiability gap. The distinction between parameter transparency and formal identification is acknowledged and open.
• Dual-channel diagnostics (survey and interview) with explicit integration logic, quality controls aligned to ISO 10667-1 and ISO 20252 delivery references, and documented reconciliation protocol between internal NORN-derived state and external EOS-derived state.

Product suite

Clew is the accessible SME entry point: a structured diagnostic delivering a scorecard, structural fractures, and prioritized recommendations within a 30-day engagement. It is designed for micro and small enterprises at CMMI maturity levels 1–2, where outcome metrics are insufficient and structural debt is accumulating invisibly.

Weft is the structural intelligence layer for financial and risk professionals, strategically oriented enterprises, and institutional advisors. It is not an upsell of Clew — it serves a distinct buyer persona and is resourced and positioned independently.

Loom provides the standardization layer for consolidating SME portfolios. Knitty extends the stack into adjacent diagnostic territory currently in definition.