The producer and the consumer

Two curves are computed for every scenario. The producer curve measures the enterprise’s own capability throughput against installed capacity: the pipelines ran, the endpoint returned, the checks passed. The consumer curve measures delivered performance against committed thresholds: whether the customer received what was promised, in the form the customer relied on. The two diverge, and the divergence is where liability lives. A producer can pass every check while the consumer’s report is wrong; a provider can degrade a model while every dashboard stays green.

Phases and measures

Delivered performance is plotted against time through a disruption. The curve passes through five phases, prepare, resist, absorb, endure and recover, and is measured in eleven ways. Six measure the capacity to absorb: the time until performance begins to fall, the duration of the fall, its depth, the performance retained at the bottom, the rate of fall, and the area between the curve of a resilient design and the curve of the present estate over the fall. Five measure the capacity to adapt and recover: the time at the bottom before recovery begins, the duration of recovery, the performance restored, the rate of recovery, and the area between the two curves over the recovery. The eleven were drawn from a 2024 review that reduced 395 published formulations to seventeen, and the specification computes all seventeen.

Why the duration measures matter

In the seed study the four duration measures could not distinguish one design from another, because the model gave recovery a weight of zero. An enterprise prices recovery in responder hours, cutover cost and degraded operation, and it is the one term the destination domain cannot set to zero. The duration measures are therefore promoted to primary here.

What the industry measures instead

The one time-integrated measure the data industry uses, incidents multiplied by time to detect and repair, treats every incident as a total outage that ends instantly. A feed that ran below its accuracy floor for eleven days and a warehouse that was unavailable for thirty hours are different shapes with different consequences for different commitments, and only a curve tells them apart.

Status

The curve computation is specified as a scenario simulator over the net, with one complete implementation in the firm’s register. It runs on industry composites in the inferred state and on the enterprise’s own capacities and values in the corrected state.