I think Vint gets that, and is speaking to the higher level and using "security" as an abstract generalization.
For example, the web was explicitly developed as a "pull" technology with declarative linking by reference with public visibility. Understanding the impact of that to how you build a security model governing access presents unique challenge. By comparison, Usenet is the opposite. It's essentially a syndicated push technology, more similar to a broadcast publishing method. As a result, the security model for how people gain access to resources, and what talks to what, is handled in a very different way.
Those are just two examples of content on today's general Internet which is an extension of Vint's work. When he talks about the Internet of Things, he doesn't merely mean the fad of sticking a web browser on a toaster. He's talking about the bigger vision of omnipresent computing and direct interaction of common devices to each other. Much like the Internet (specifically TCP/IP and DNS) was conceived as a way for computers to directly talk to each other (not going through a centralized hierarchy for approval and redistribution). We learned a lot of great lessons about how it would be used, the shortcoming, and the security ramifications. Now that we're in the fledgling stages of doing the same thing for a whole new are of automation and computing, there's great opportunity to think about and apply the lessons learned.