For many applications it is not enough for a system to perform correctly under ideal circumstances. Systems benefit from robust and survivable designs that can withstand faults and degrade gracefully as faults occur. System topologies that resist component failures are certainly attractive in many applications. We demonstrate the effectiveness of a co-evolutionary approach for discovering survivable sorting network topologies. Our study shows an improvement in survivability as selective pressure is applied, that it is possible to tune the degree of survivability in a sorting network, and the mechanisms by which this happens are general enough that the results should apply to any evolved hardware with serial components.