Definition
Degree to which simultaneously operating AI systems produce mutually coherent recommendations for the same vessel.
The Measurement Gap
What This Metric Measures
MSOCI measures the degree to which simultaneously operating AI optimization systems produce mutually coherent recommendations for the same vessel over the same time horizon. It counts the proportion of planning periods free from inter-system conflicts: where the voyage optimizer, CII optimizer, predictive maintenance scheduler, and port berth optimizer all produce recommendations that can be simultaneously executed without contradiction.
The Decision It Supports
Are the fleet's AI systems working together or at cross-purposes, and when they conflict, which system's recommendation should take priority? A vessel might simultaneously receive instructions to slow down (CII optimizer), speed up (port optimizer to hit a berth window), defer maintenance (PdM system says healthy), and schedule maintenance (adaptive PMS says interval is due). The master must resolve these conflicts with no visibility into which system has the strongest basis for its recommendation. MSOCI quantifies the conflict burden.
Why Legacy Metrics Cannot See This
Traditional fleet management evaluates each system independently. The voyage optimizer is measured on fuel savings, the CII system on compliance scores, the PdM system on prediction accuracy. Each can show excellent individual performance while collectively producing impossible instructions. No legacy metric measures whether the outputs of separate systems are operationally compatible when applied to the same vessel at the same time. MSOCI exposes the coherence gap that emerges when multiple AI systems optimize independently.
Formula
Primary Formula
MSOCI = 1 - (Periods_with_conflicts / Total_planning_periods)
Variable Definitions
- Periods_with_conflicts = planning periods (daily recommended) where two or more AI systems produce conflicting recommendations for the same vessel
- Total_planning_periods = total planning periods in the measurement window
- Conflict = two systems recommend actions for the same vessel that are operationally incompatible within the same time window (e.g., one recommends reduced speed while another recommends increased speed)
Units:
0 to 1. A value of 1.0 means perfect coherence with no inter-system conflicts. A value of 0 means every planning period contains at least one conflict. The metric is computed per vessel and can be aggregated fleet-wide.
Edge Cases
Defining conflict requires clear incompatibility criteria: two systems recommending slightly different speeds is not a conflict, but one recommending reduced speed and another recommending increased speed is. Vessels with only one active AI system should be excluded because coherence is only measurable with two or more systems operating simultaneously. Minimum sample of 30 planning periods per vessel is required. Track which system pairs conflict most frequently to identify where integration or hierarchy is needed.
Proposed formula, pending validation pilots.