 Hi, I'm Nikos. I'm here to talk about the round complexity of Byzantine agreement. This is a joint work with Ranko and Iftachheidner, Matan Norland and Alex Samoordnitsky. So what's a Byzantine agreement? It's the distributed tasks such that every party holds an input bit and the parties are required to agree on the same output bits and there's an additional security requirement that states that at least one party should output its own output bit. So in this work we wanted to bound the probability of halting after one or two rounds and the reason we aimed for one or two rounds is that three rounds are essentially optimal. So what do we show? We show that for BA protocols, if the security threshold is something like n over 3, meaning that at most n over 3 parties are corrupt, then the halting probability after one round is vanishingly small and the halting probability after round two is far away from one. In addition, we show that for most BA protocols and under some combinatorial assumption that conjecture actually that I'm not going to discuss, the halting probability is also vanishingly small after the second round. So to show this result we follow the classic blueprint for deterministic termination, meaning that we considered a sequence of executions such that the input configuration at the ends of these executions gives you different output bits and we show that there is an attack for adjacent executions such that these executions are indistinguishable for the honest parties. However, for randomized protocols this argument fails because the randomness can be used to distinguish between adjacent executions. So to remedy this what we propose as a solution is to abort some parties in order to decouple the randomness from the outputs. This attack gives rise to an isoparametric type inequality. We managed to prove this inequality for some limited cases which reduced to the famous KKL theorem and Frigwood's Janta theorem but the general case we left as an open problem. So the paper is available online. Thank you.