Large scale online citizen science projects are often designed to minimize the need for volunteers to be knowledgeable or learn about the scientific content of a project as a prerequisite for enrolling and maintaining mass participation. Despite this feature, most large scale online citizen science projects have stated educational goals where volunteer learning is characterised as a key benefit of their participation. However, the specific knowledges volunteers bring with them when they join or develop as they participate are largely hidden and outside the control of project owners. With this background many, but not all, projects aiming for scientific output have an instance in the scientific process where volunteers are constructed as being in some way on par with scientists. These instances are usually enacted by scientific protocols that harness a tightly constrained ability of the crowd, usually perception, which makes volunteer contributions valid for scientific work. Outside these tightly constrained contribution opportunities, however, many citizen science projects also work to uphold boundaries between citizens and scientists. Intuitively, this may not seem necessary as scientists by virtue of their professional training generally have skills beyond those of volunteer contributors. In practice, however, such boundaries are not so clear. In this session, we will explore when and how such boundaries are challenged as volunteers develop unexpected knowledges while participating in citizen science projects. The purpose is to illuminate the relationship between the citizen scientist as conceived of as a contributor to science with specific, but static qualities, and the development of contributors over time. Data consists of interactions among researchers and volunteers on the discussion forums of citizen science projects.