Byzantine agreement in polynomial expected time: [extended abstract]

In the classic asynchronous Byzantine agreement problem, communication is via asynchronous message-passing and the adversary is adaptive with full information. In particular, the adversary can adaptively determine which processors to corrupt and what strategy these processors should use as the algorithm proceeds; the scheduling of the delivery of messages is set by the adversary, so that the delays are unpredictable to the algorithm; and the adversary knows the states of all processors at any time, and is assumed to be computationally unbounded. Such an adversary is also known as "strong". We present a polynomial expected time algorithm to solve asynchronous Byzantine Agreement with a strong adversary that controls up to a constant fraction of the processors. This is the first improvement in running time for this problem since Ben-Or's exponential expected time solution in 1983. Our algorithm tolerates an adversary that controls up to a $1/500$ fraction of the processors.