Putting all your eggs in one basket? A mathematical analysis of attack strategies
Аннотация
In fields such as cyber operations, management, medicine, and military strategy, the choice between sequential and parallel strategies is critical for optimizing success rates and minimizing completion time. This study examines a hacker using multiple attack tools to destroy a rival system, assuming that each tool has an equal and independent probability of success. Attack execution is time-consuming, with the required time increasing concavely as more tools are used simultaneously. We analyze two protocols of attack strategies: one beginning with a multi-tool strike followed by single-tool strikes, and the other following the reverse sequence. We study the system’s expected survival time as a function of attack duration and the single-tool success probability, and for each parameter configuration we identify the strategy that minimizes expected survival. Our analysis yields two main results. First, hybrid strategies can be strictly optimal, depending on the success probability of each tool and the concavity of the setup-time function. Second—and more importantly—although one might expect the optimal hybrid policy to depend delicately on the model parameters, we prove a structural invariance result: within the class of strategies studied in this paper, whenever a hybrid strategy is optimal, the simultaneous strike must occur before the sequential phase. This finding provides a unified analytical principle for timing and coordination decisions for structuring resource deployment under uncertainty.