Military networks have a huge range of performance requirements, many of which are continually changing in the face of evolving operations. In such circumstances, ensuring that network performance is adequate, let alone optimal, is a significant and ever more challenging management burden. The panacea to this problem is autonomous self-optimizing networks that continually adapt themselves as situations change. This work looks at some of the benefits and challenges of attempting to design such networks to address this uniquely military problem. Specifically we investigate simplified analytical problems on link state routing protocols, as a first step to gain insight that is more widely applicable. A crucial component of such protocols is the dissemination process by which links periodically broadcast their current state across the network. While this allows nodes to correctly compute routes, it also incurs significant control overhead that diminishes the effective capacity of the network. In this work, we study how nodes can make the best routing decisions as a fundamental tradeoff between link state dissemination cost and the accuracy of route computation. Using a Markovian model for link dynamics and a parameterized model for the state dissemination process, we investigate the impact of selectively sending link state updates on the packet delivery ratio performance in a ightly loaded dynamic network. Our analysis reveals the optimal state dissemination strategy that maximizes the packet delivery ratio given a total budget for the network-wide link state update rate, under sparse information flows. We instantiate our results explicitly for a simple path disjoint topology.