Reasoning about the success of distributed computations in the face of node and communication failures is an ever hard issue. This is particularly true in highly dynamic ad hoc networks where communication between two nodes can often take place only via intermediate nodes, and nodes additionnally can be mobile. Rather than imposing a rigid deterministic system model on such dynamic ad hoc networks in attempt of obtaining ``exactly once'' reliability guarantees for distributed computations taking place among nodes, we propose to embrace the non-deterministic nature of these settings, and work with probabilities and hence notions of partial success.