11. Scaling Metrics

Scalability is measured both by processing volume and by the system’s ability to maintain correlation accuracy and adversarial resilience. Each metric is defined formally to support reproducibility and to ensure that scaling does not introduce drift. All metrics are reported as global baselines. When available, additional slices are reported without altering definitions or formulas.

Optional stratifications

• By ecosystem vertical
• By actor sophistication tier (superficial, structured, targeted)
• By acquisition profile (baseline, geo-variant, challenge-aware)
• By pipeline stage (acquisition, extraction, enrichment, correlation)

Windows
All metrics specify the observation window (e.g., weekly, monthly) and the summary operator (e.g., P50/P90)

11.1 Throughput

Definition
Number of entities fully processed per unit time across the acquisition → extraction → correlation pipeline.

Formal definition

$$T = \frac{N_{\mathrm{processed}}}{\Delta t}$$

• $N_{\mathrm{processed}}$: total entities processed
• $\Delta t$: elapsed processing time

Rationale
Throughput establishes feasibility for internet-scale coverage while maintaining evidentiary integrity.

11.2 Processing Efficiency

Definition
Volume of entities processed per unit of compute time.

Formal definition

$$\eta = \frac{N_{\mathrm{processed}}}{H_{\mathrm{compute}}}$$

• $H_{\mathrm{compute}}$: compute hours consumed

Rationale
Tracks optimization gains in acquisition, enrichment, and correlation pipelines.

11.3 Signal Diversity Index (SDI)

Definition
Measures the distributional balance of signal families extracted per entity.

Formal definition

$$SDI = 1 - \sum_i p_i^2$$

• $p_i$: proportion of signals of family $i$

Rationale
High diversity corresponds to richer pivot graphs and resilience against collapse when any one family is suppressed.

11.4 Correlation Graph Density (CGD)

Definition
Proportion of realized edges relative to the maximum possible in the correlation graph.

Formal definition

$$CGD = \frac{2 \, |E|}{|V| (|V| - 1)}$$

• $|E|$: admitted edges
• $|V|$: total vertices

Rationale
Density calibration prevents both fragmentation (too sparse) and inflation (too dense).

11.5 Adversarial Coverage Retention (ACR)

Definition
Preservation of detection coverage across adversary sophistication tiers when scaling from pilot volume to census scale.

Formal definition

$$ACR = \frac{C_{\mathrm{scale}}}{C_{\mathrm{pilot}}}$$

• $C_{\mathrm{scale}}$: coverage rate at scale
• $C_{\mathrm{pilot}}$: coverage rate at pilot

Rationale
Ensures that scaling preserves detection capability across actor sophistication levels, preventing erosion of coverage when workloads increase.

11.6 Render Fidelity Retention

Definition
Proportion of entities where the rendered DOM and dynamic assets are captured without interference from defensive mechanisms.

Formal definition

$$R_{\mathrm{fidelity}} = \frac{N_{\mathrm{full}}}{N_{\mathrm{attempts}}}$$

• $N_{\mathrm{full}}$: successful full renderings
• $N_{\mathrm{attempts}}$: total render attempts

Rationale
Render fidelity preserves downstream signal extraction and prevents structural blind spots.

11.7 Reprocess and Error Rate

Definition
Fraction of entities requiring repeat acquisition or correlation due to upstream failures.

Formal definition

$$E_{\mathrm{rate}} = \frac{N_{\mathrm{reprocess}}}{N_{\mathrm{processed}}}$$

• $N_{\mathrm{reprocess}}$: failed or reprocessed entities
• $N_{\mathrm{processed}}$: total processed entities

Rationale
Error containment is critical for scale; uncontrolled reprocess rates negate throughput and efficiency gains.

11.8 Metric Map

Visual grouping of scaling metrics into operational categories (volume, fidelity, diversity, resilience, error control) to support quick audit and coverage discussions.