Proxy-Outcome Divergence Score (PODS)

GMI-AGR-020Agriculture and Farm OperationsIntegration-RequiredPattern O-30: Proxy-Objective Divergence

Definition

The rank correlation, trended over time, between the proxy each automation optimizes (sensor-measured grain loss, NDVI or biomass index, fruit color class, soil moisture setpoint, ripening room firmness) and the outcome the operation is paid for (measured loss by pan drop, margin per acre, packout grade and shelf life, crop water stress, buyer acceptance).

The Measurement Gap

There is no metric that tells you how well the proxy an automation optimizes (sensor-measured loss, NDVI, fruit color, soil moisture setpoint) still tracks the outcome the operation cares about (actual loss, profit, packout grade, crop stress), which means the automation improves its score while the outcome drifts.

Formula

PODS = 1 − ρ_s(proxy, outcome) over a rolling window per automation, where ρ_s is the Spearman rank correlation on paired observations; companion: Δρ_s against the baseline window

Proposed formula, pending validation pilots.

At a Glance

Unit
Statistic (0 to 2)
Trend Direction
Lower; approaching 0
Category
Autonomy and Machine Behavior

Decision Supported

Whether an automation's improving score still means an improving result, and when to re-anchor the automation to the outcome before the divergence costs a season.

← Field Machine Duty Cycle Deviation (FMDCD)Weed Detection Mode Coverage (WDMC) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment