Emerging optical technologies enable communication networks to be physically reconfigurable, adapting to current workloads and failures. Optical networks further have the potential of providing higher bandwidth at a lower cost than traditional electronic packet switching, making them a natural choice for next-generation networks. However, these benefits come with new challenges, in particular the need for fundamentally new algorithmic ideas. In this talk, I will present some recent results on how to efficiently leverage such physical reconfiguration, even across wide-area networks, but also discuss underlying algorithmic complexities. Reconfigurable topologies are not a plug and play solution and I will therefore outline future research directions.