In this article, I develop two major propositions. First, the future Internet should be polymorphic, and it should conciliate different architectural networking paradigms. The second proposition is that the future Internet should necessarily build on a strong theoretical basis from a networking science that is in the course of development. Particularly, I describe how virtualization makes possible a polymorphic future Internet and provides for easy deployment of new cooperation schemes. Then I attempt to expound on the aspects of security and scalability of the future Internet. This analysis also hopes to account for some justification as to the need for a clean-slate approach to fundamentally tackle the shortcomings of the current Internet and why a polymorphic future Internet is needed to for the coexistence of evolutionary and revolutionary schemes. Admittedly, the future Internet has to be built on strong foundations of networking science. A global theory of cooperation over networks is therefore most needed to foresee the important issues in the design and deployment of the future Internet. Methodologies for experimenting and validating large-scale mechanisms by reproducible large-scale experiments are, moreover, among the expected outcomes of the emerging networking science, which I also discuss in this article.