Digital technologies have transformed how firms and individuals collaborate while giving rise to the phenomenon of open collaboration. Open collaboration, which originated in open source software, has spread beyond software. Open collaboration communities usually have three types of actors who collaborate by selectively revealing problems and solutions. These actors, that propose problems and solution to the community, are: 1) independent users, 2) firm users, and 3) firm producers. The proposed problems can be either path dependent, which build on prior problems, or path creating problems, which are novel functionalities or features. This article theorizes about the characteristics of the structure of collaboration in open collaboration, by drawing from a theory of organizational path constitution. The proposed theoretical framework argues that actors within a community do not only solve problems by themselves (i.e. independent problem solving), but other actors also solve other actors´ problems (i.e. shared problem solving). Furthermore, the proposed theoretical framework allows understanding the importance of who revealed the problem and what type of problem was revealed in independent and shared problem solving. We perform a multinomial logit model analysis using a two year data from a successful open source software development in Open Stack, in which all three types of actors identified more than 5,000 problems. Our results provide novel empirical insights into the structure of open collaboration. In independent problem solving, our findings suggest that any actor solves novel problems rather than focusing on path dependent ones. However, in shared problem solving, firm actors focus only on novel problems revealed by independent or firm users. Moreover, independent users focus indiscriminately on improving prior problems and novel problems from firm users.