Great analysis. By the way, are the number of attendees in the audience externally auditable or verifiable in some way? Like, can people proxy attendance by checking a time series of how many nodes the Twitter servers were firing packages to?