As in other STEM disciplines, early computing courses tend to stress individual assignments and discourage collaboration. This can lead to negative learning experiences that compel some students to give up. According to social learning theory, one way to improve students' learning experiences is to help them form and participate actively in vibrant social learning communities. Building on social learning theory, we have designed a set of software interventions (scaffolds and prompts) that leverage automatically collected learning process data to promote increased social interactions and better learning outcomes in individual programming assignments, which are a key component of early undergraduate computing courses. In an empirical study, we found that students' interaction with the interventions was correlated with increased social activity, improved attitudes toward peer learning, more closely coupled social networks, and higher performance on programming assignments. Our work contributes a theoretically motivated technological design for social programming interventions; an understanding of computing students' willingness to interact with the interventions; and insights into how students' interactions with the interventions are associated with their social behaviors, attitudes, connectedness with others in the class, and their course outcomes.