The prioritization of selected economic activities through a broad consultative process is often considered a central aspect of research and innovation policies aiming at promoting smart specialization (RIS3). This is seen as an effort to reduce fragmentation and a way to increase the impact of research and innovation investments on regional and national development. This note argues that targeting research and innovation policies may not be always recommendable due to inherent problems of incomplete information and the inevitable representation bias towards incumbent interests. When the economic specialization of a region is not evident from observation of market dynamics, policies should aim at enabling the process of market selection, allowing such specialization to emerge as a result of entry, exit and experimentation. Such an approach would have two immediate implications for the development of RIS3: (i) replacing the emphasis on ex ante definition of activities with a focus on a results-based approach for research and innovation investments; and (ii) fully integrating monitoring and evaluating mechanisms in the development of the strategy, allowing for policy experimentation, structured learning and systematic adjustment of programs and policies towards the pre-defined objectives.