 called must the communication graph of NPC protocols be an expander and it's going to be given by, I'm too cool to wear a name tag. Okay, thanks. So this talk is about the GeoMultipart communication where some parties want to compute something together. Now we have many, many beautiful waterfalls from the 80s that show us how to do it. In the good old days, everybody talked to everybody. And when you have many parties like today and nowadays, if everybody talks to everybody, there can be some bottleneck that can have some overhead. And one way to really need this huge overhead, one way is to consider partial graphs of communication. So one way is to consider a fixed partial graph. Well, before we start computing, everybody knows who we can talk to. This is, again, an old model that was studied from the 80s. And corruptions of the parties can be a function of the topology in this case. As a result, we have many hard, low bounds. So even if you just want to broadcast a message, the connectivity must be t plus one, which is number of corruption. And if you have a linear number of corruption, then the communication complexity is quadratic in the number of parties. A more recent partial graph methodology is a dynamic partial graph. Here, everybody can talk to everybody else, but they choose to talk to dynamically during the protocol. And in this model, we can get around many of these low bounds. So we can get secure computation where everybody talks only to polylog many other parties. We have broadcast where the communication complexity linear up to polylog factors. So one interesting question is, when you look at the communication graph that you really get from these protocols from secure computational Byzantine agreement protocols, what are the necessary properties that you really need to get secure computation? In this work, we give a foundational study of the dynamic graph model. And we have a framework to analyze the various graph properties that we get from these protocols. And as an example, we'll look at one property of expansion. So all of the protocols that we know, both in secure computation and in the distributed computing, the underlying communication graph is an expander. And we wanna see whether this is really inherent. Like everything in life, the answer is it depends. So we show that in many settings, we can have protocols that are secure, which are in a very, very strong sense, not expanding. But for some functionalities in some settings, any secure protocol must be an expander. It's a, if you wanna hear more, you have to be patient a few more days and hopefully it will be only pretty soon. So thanks.