Tradeoffs Military planning is filled with tradeoffs and what-ifs. This planning includes the selection of targets and the assignment of weapon systems to the targets, where greater certainty of force protection and mission success comes only with greater costs. Even away from the battle, there are significant tradeoffs in matters such as logistics, where schedules, carrying capacity and routes have to be arranged so that the most equipment and personnel can be deployed in the shortest time with the least effort. In many situations, defending against some contingency increases vulnerability to another.

Risk reduction Military decisions can have enormous consequences. A significant assurance is needed that the best available option is chosen. Aetion's approach is to discover various options by combining plan fragments, to simulate many possible options against a representative set of contingencies, and to present the performance of all the best ones to the decision-maker. Planning can still proceed rapidly because many plans can be simulated in parallel, and the simulation of anticipated situations can run before the situations actually transpire. Situations can even be anticipated automatically by adversarial gaming, where the computer generates and evaluates a range of enemy responses. (This makes sense because running many simulations on a cluster of computer hardware is cheap in comparison to the consequences of making the wrong decision.) The decision-maker is presented with concrete tradeoffs between the best choices, and then given the control to make the best value judgment.

Example Based on previous Air Force work, and with funding from the Army and the Navy, we examined the problem of how to decide what nodes and links to attack in order to maximally disrupt a communications network. The mission objective was to fragment the network into many isolated segments with the major command centers unable to communicate with one another. Complications included a communications routing algorithm that could compensate for loss of a path by using several of the remaining paths simultaneously, and the fact that, due to being located near population centers and being better-defended, a direct attack on a command center would be costly. We applied an evolutionary algorithm that progressively developed increasingly complex attacks and successfully discovered a number of the plans that we expected to be of interest. Our interactive graphical environment clearly showed the expected tradeoffs: for instance, between mission success and expected casualties.