This document proposes the concept of "SocialMachines" as a unified paradigm for describing systems that blend computational and social elements. It discusses three visions that contribute to this paradigm: 1) social software systems based on social data from platforms like blogs and social networks, 2) people as computational units through crowdsourcing and human computation, and 3) software with social abilities through agent-based and semantic web services. The author argues that SocialMachines can provide a conceptual model for understanding increasingly interconnected web-oriented systems and promoting discussion among researchers.