Massive open online courses promised education at a scale that in-person teaching can’t reach, but the instructor–student ratio that works in a seminar doesn’t survive the jump to tens of thousands of learners. So much of what makes a class work — feedback, discussion, a sense that anyone is paying attention — has to be reconstructed out of what the learners themselves do for each other.

Over several years with Dilrukshi Gamage, Indika Perera, Shantha Fernando, Thomas Staubitz and others, we studied how to make peer-driven learning actually work. We found that aligning incentives to the quality of feedback, rather than simply its length, produced feedback learners judged more useful L@S’17. We showed that seeding peer assessment groups with trained “introduced peers” improved the quality of subsequent discussion TALE’18. We surveyed the broader state of the field in a systematic literature review of peer assessment in MOOCs Distance Education. And we showed that treating learners as communities of practice — rather than isolated users moving through a course — measurably improves outcomes Asian CHI’21, which was recognized with a best paper award.

The through-line across these projects is that scale doesn’t remove the social structure of learning; it just changes who has to supply it.