Consensus algorithms deployed in the crash fault tolerant setting chose a leader-based architecture in order to achieve the lowest latency possible. However, when deployed in the wide area they face two key robustness challenges. First, they lose liveness when the network is unreliable because they...