Which exploration policy finds robustnesscliffs fastest under fixed budgets? We ablate –focus∈ {boundary,runtime,robustness, balanced} atop ouragentic sweep framework [1], measuring (i) time-to-first-cliff,(ii) area-under-uncertainty, and (iii) policy ranking under fixedwall-clock and evaluation budgets. Results show that explicitlyboundary-seeking acquisitions dominate low-budget discovery,while balanced policies win under tight runtime constraintswhen end-to-end system effects (ghost cost [2], scheduleroverhead [3], … Continue reading Balanced vs Boundary-Seeking Policies for Robustness Discovery