Zhongguancun Science-based Park (ZSP) has made an ambitious plan to build a 'first-class science-based park.' How to reach the goal? Where is the gap between ZSP and Silicon Valley (SV), Hsinchu Science-based Park (HSP)? Why is there no significant innovation in ZSP, where 60 universities, 130 national research institutes and 300 thousand scientists gather? By putting them in the background of globalization and knowledge-based economy and with the concept of 'learning machinery,' the author attempt to analyze the differences of innovative ability among the three districts and the reasons for lack of innovation in ZSP in the perspectives of industrial organization and social relations.First, the author describes the background under which ZSP is being built - globalization and the emerging knowledge-based economy. Globalization and localization are two sides of a coin. The emerging knowledge-based economy has made learning and innovation more important to development of an economy. Regional collective learning is significant to innovation. Social capital, or networks of social relations, is beneficial to formation of collective learning machinery. It is important for a country to build regional advantages in front of fierce competition. Second, the author analyze the importance of the collective learning machinery, which is shaped through a network-based, decentralized system of industry and a network of personal relations, to SV's ability to innovate in succession. The role of the industrial system of 'flexible specialization' for HSP's relative competence in innovation and competition is also described. Third, the author discuss the current situation of innovation in ZSP and analyze why it is short of innovation and the way it can has a leap development by building an network-based industrial organization and bringing the role of the transnational community of Chinese engineers into play.By comparison of the knowledge flows among the three districts, the author analyze the barriers to formation of learning machinery in ZSP and the way to solve the problem. The conclusion is that ZSP has to break the boundaries between organizations of R&D, production and marketing - barriers to knowledge flows. Only then could it make successive and significant innovations. The best industrial organization in place of the present one with barriers to knowledge is a network-based and flexible system. In the age of globalization with advanced IT, face to face interaction is still critical for knowledge flow. Tacit knowledge, which is critical to innovation, has to be transferred through network of personal relations. Successive innovation can only take place in a decentralized and flexible industrial system with division of labor, collaborations and competition. Such asystem is maintained by trust. Venture capital can work only in the places where there is trust relation. For a science-based park, it is as important as training skilled people, building infrastructure to build up social capital. The widest gap between ZSP to SV is lack of social capital in ZSP. It is urgent for ZSP to cultivate trust among scientists, engineers, and entrepreneurs and to build networks of personal relations and the consensus on the goal of the community. Finally, Chinese engineers and scientists in SV and other dense knowledge areas in developed countries are precious resources for China to develop high-tech industries. They have the knowledge, especially tacit knowledge, about the newest technologies. ZSP has now paid not enough attention to them. It is the quickest way for ZSP to become a first-class science-based park of the world to build a transnational community of Chinese engineers and scientists and form a transnational learning machinery. An environment suitable for successful start-up is more important than other incentives to absorb overseas Chinese engineers to come back.