The paper starts from the increasing spatial and functional fragmentation of value-added chains, global deregulation and dis-embedding of "markets", and interdependencies among the Net-based digital technologies. It develops a socioeconomic setting with ubiquitous direct interdependencies and interactions, Net-externalities, "strategic" strong uncertainty, and omnipresent collective-good and social-dilemma problems. These entail coordination failures, either in the form of conventional market failure (i.e. collective blockages of action) or of "wrong" or outmoded institutional coordination and, thus, widespread technological "lock-ins" that are indicative of insufficient ability of collective action. This is particularly true for de-regulated, individualistic cultures. In contrast, sustainable innovation, used in a broad, i.e. technological and institutional, sense, requires an effective collective action competence. This, in turn, requires a new and increased coordination. Against this background, the global corporate economy has spontaneously developed private individualist substitute arrangements to cope with the new complexity, such as local clusters and hub-and-spoke networks, which all have severe shortcomings. With reference to what we call the "Linux" paradigm, the paper discusses the possibility of a spontaneous evolutionary, i.e. collectively learned, institutional coordination through emergent collective action and networks with "good" governance. The paper argues that only a hybrid system that consists of "well-governed" networks and a new approach towards more comprehensive and deliberate "interactive" and "institutional" public policy, supporting collective learning and emergent institutional coordination , is capable of solving the complexity and coordination problems of the "new" economy by increasing certainty, stability and more continuous and comprehensive innovation. This new policy approach is outlined at the end.