# How can we prove Karlin and Yao’s probability bound on randomized Byzantine Broadcast protocols?

Can we concretely prove that with 3 nodes (one of which is faulty), a randomized Byzantine broadcast protocol will have a 2/3 probability of success?