The contemporary participatory media ecologies within which connected learners work and change themselves may be only loosely organized by disciplinary norms and categories. In such loosely disciplined contexts, learner identity, traditionally understood as an object of disciplinary power from a standard sociocultural (CoP) perspective, may be better understood as an ongoing dialogic project of discursive positioning. This study demonstrates a computational approach to modeling identity in an open, online network of ed-tech innovators. Latent Dirichlet Allocation (LDA) is combined with post hoc qualitative analysis to represent and make sense of online identity profiles as composites of seven latent network discourses. The study operationalizes useful theoretical notions of polyphony and positioning for the study of connected learning and identity in loosely disciplined participatory networks.