One technical point that I disagree with is his characterization of exponential spread. He is mentally modeling it like every person is the root of their own tree in a societal forest. While the number of infected nodes in an SIR model on a network does grow exponentially in early stages, at some point the growth slows as many of your contacts have already been infected. This may not happen until some constant fraction of the graph has been infected, however. Of course as we all have heard the equilibrium state is "herd immunity" when the infection dies out or remains contained to small components, but even before this I believe the growth rate slows. I should know more about this considering it is what i wrote my doctoral thesis on, but i've made room for other things.
