Degree-based attacks are not optimal for desynchronization in general networks.

Suppressing harmful synchronization in complex networked systems is receiving increasing interest in various contexts including epileptic seizure and the Internet and traffic congestion. It is traditionally believed that by disrupting the most important nodes (i.e., those nodes of high degrees) synchronization can be effectively mitigated. However, this approach can perform poorly in more general situations such as homogeneous (random) or small world networks. In this article, we investigate how topological properties, such as the heterogeneity of the network can have an impact on its desynchronization. In particular, we propose a topology-aware scheme that chooses topologically sufficiently diversified contrarian nodes by taking into account the patterns of connectivity among them. Hence, a maximal number of nodes in the network can be influenced, leading to a more global impact and greater disruption to synchronization. Our scheme demonstrates significantly improved performance on various networked systems including homogeneous networks, small world, and even scale-free networks.