I agree with your point of view, it does seem to shift some of the teaching burden from instructors to project maintainers. I generally like the way student performance is assessed, with more emphasis on the technical quality and presentation rather than the binary "was this contribution accepted or not?". However I still think that, depending on the initial choice of the project and the issue to work on, some students may receive a lower grade through no fault of their own, e.g. a PR not getting accepted by deadline, changes required at the last minute, etc. These problems are not present with traditional assignments isolated from the "real world", and I believe those can teach the student just as much as open-source participation, if done right. Take group projects, for instance, where students are expected to collaborate using project management tools, set up unit tests themselves, etc.