Heterogeneity Under Observability: An Acquisition-Cost Framework for Assessing Diversity as a Cyber-Physical Defense

Response to DARPA Special Notice DARPA-SN-27-17, Formal Learning and Abstractions for System Heterogeneity (FLASH). This working paper proposes an acquisition-cost framework for evaluating when system heterogeneity provides meaningful defense value against adaptive adversaries. It separates three adversary classes based on how knowledge is acquired: class-attacking (A1), capture-then-exploit (A2-hybrid), and real-time (A2-pure). Heterogeneity raises adversary burden under A1 and A2-hybrid by a structural multiplier of the form 1 + Hd,variant·(N−1), where Hd,variant measures the fraction of variants with independent cheapest exploit paths. It does not increase burden under pure real-time probing for static or slow-dynamic heterogeneity. The paper specifies conditions under which heterogeneity pays, identifies hidden centralization mechanisms that collapse diversity, and proposes a metric and evaluation protocol. The anchor domain is unmanned surface and undersea vehicles (USV/UUV) operating in contested maritime environments under COLREGs. This is an independent technical response submitted by AionSystem. It is UNCLASSIFIED. The evaluation protocol described has not been executed.