Paper long abstract:

Studies of innovation point out the key role of the generalization process through which major impact are generated. Without generalization, novelty is limited to a small set of first users and its benefits remain low. In the literature, this process is related to two complementary mechanisms:

- standardization of the technology, which allows to exploit increasing returns related to learning and economies of scale;

- standardization of the users' milieu, which involves construction of markets and change of users' habits, and which may imply creative destruction (which is why we don't use the term 'diffusion').

In recent years, public authorities have shown a strong interest for social innovation (SI). Highlighting the need of SI to face big challenges, major institutions generally underline that SI hardly generalizes and that its use is very often limited to the small collective that generated the novelty. Hence, they identify barriers to the diffusion of social innovation and suggest out to get rid of them (BEPA 2010).

In this paper, we argue that referring to the processes and the semantic of innovation to deal with social innovation is misleading. We argue that the mechanisms of generalization do not lie in the standardization of technologies/users but in the dissemination of the capacity of collective experimentation. We thus suggest that social innovation leads to rethink the production and use of knowledge in society.