With the goal of building a world scientific and technological power, and the transformation of China′s economy from high-speed growth to high-quality development, the process of establishing collaborative network connections in technological innovation is becoming more and more important. The key conditions to implement innovation driven strategy is embedded in the global open innovation network. Bootlegging as the deviant practice process that has no formal organizational support is undertaken with the aim of producing innovations that will benefit the company. Bootleg innovation can help the company to enhance the innovation capability and performance. However, previous persistence studies have mostly focused on the paradox of destructive and constructive innovation, which ignored the function of routines replication that is an important strategy for network organizations′ value creation. Routines replication can help network organization to deconstruct the network routines of existing successful enterprises to improve the ecological position, however, it may cause replication dilemma when network routine is not matching with the environment. Meanwhile，existing studies have not been clarified the root cause of bootleg innovation and the context dependence of routines replication in technology innovation network. To uncover the mechanism of bootleg innovation in technology innovation network and the relationship between routines replication and bootleg innovations are becoming an important unresolved theoretical problem.

In view of practical background and the above research status, to fill these gaps while integrating previous research, this research extends the concept of bootleg innovation from individual level to organization level, and analyze the impact of routines replication on bootleg innovation based on the contingency perspective of paradox integration, and then constructs the theoretical framework of "routines replication-bootleg innovation" based on theoretical analysis and propose hypotheses. Routines replication covers two dimensions: conventional routines replication and flexible routines replication. Meanwhile, this research further examines the moderating effects of network closure and knowledge base, and the interaction effects among them. Then an empirical test was made by the method of questionnaire survey, the survey was based on the R&D intensive industry such as electronics and information technology industry. Snowball sampling was used to construct technological innovation network, and the relevant hypotheses were tested by correlation analysis and multiple regressions model on the survey data with the help of SPSS 22.0.

Our results show the following conclusions: technology innovation network routine replication, network closure and knowledge base are the key conditions for technology innovation network bootleg innovation. Conventional routines replication and flexible routines replication have significant positive effect on technology innovation network organization bootleg innovation. The high network closure positively moderates the effect of conventional routines replication on bootleg innovation, but have no significant effect on flexible routines replication. The high knowledge base negatively moderates the effect of conventional routines replication on technology innovation network organization bootleg innovation, but have no significant effect on flexible routines replication. Technology innovation network closure and knowledge base also have different joint moderate effects, four ego network configurations are obtained by ambidextrous combing the advantages of network closure and knowledge base, such as closed-specialized network, closed-diverse network, open-diverse network and open-specialized network. The joint moderate effects of network closure and knowledge base inhibit the relationship between flexible routines replication and technology innovation network organization bootleg innovation, but have no significant effect on the relationship between conventional routines replication and bootleg innovation.

The conclusions of the study provides a new perspective for further study of technology innovation network organization bootleg innovation, enriches the research contents of routines replication. This research can promote the understanding of enabling condition and the multiple matching mechanisms of bootleg innovation. Following the thought of "poverty leads to change", this research opens the black box of routines replication to overcome the replication dilemma and deconstructs the two paths from routines replication to bootleg innovation, which also cracks the selection preference and context dependence of bootleg innovation in technology innovation network. The conclusions of the study will contribute to routines replication theory, bootleg innovation theory, innovation network theory, innovation network governance theory and innovation ecosystem theory. At the same time, this research also has certain management implications for enterprises. On the one hand, enterprises should pay more attention to the enterprise R&D environment to balance the tension between autonomy and accountability in the generation of innovations. On the other hand, enterprises should imitate the leading enterprises to design their own technical system and abolish the institutionalized law of time to realize the pursuit of underground innovation activities. This research provides useful theoretical support and practical guidance for enterprises to implement bootleg innovation strategic decision-making in technology innovation network so as to construct a moderately loose innovation ecosystem.