I’ve been talking to maintainers of a few projects I contribute to, and the same complaint keeps coming up: GSoC and similar mentorship programs are starting to run into what reviewers everywhere are dealing with, students submitting code they can’t explain when you ask/leave a follow-up comments on their submissions.
Two blog posts I read this week get at the same problem from different perspectives:
Jiri Eischmann writes about the maintainer side of this in How AI Is Changing Open Source.
throwing it over the wall to the maintainers.
He describes reviewing a 9,000-line PR (?!) for macOS support that turned out to be mostly AI output where thousands of lines were just single quotes swapped for double quotes, parts of the code unrelated to the actual problem got touched, and the contributor never responded once he raised issues. An hour of review for a PR nobody would later come back to…
Mara Averick’s post Do you WANT contributors? (a nice read) argues that not every project actually wants to grow its contributor base, and pretending otherwise is what burns maintainers out (among other things). She also references the SQLite’s contributing page: “Open-Source, not Open-Contribution.”
How’s everyone here handling this on their own projects?