Failure to detect an unequivocal conflict

[slindley]

HotCRP does a reasonable job of detecting potentially missing and inappropriate conflicts, but I had a recent case where it failed to detect an unequivocal conflict that should have been really easy to detect.

This was for ICFP 2023 (icfp23.hotcrp.com).

• PC Member A submitted Paper X to ICFP 2023 with coauthors S.
• PC Member A was assigned Paper Y to review, which was submitted to ICFP 2023 with authors T.
• The sets of authors S and T are not disjoint.

HotCRP did not, and still does not flag this up as a conflict. Reviewing was lightweight double-blind so it only emerged much later that the conflict existed. PC Member A has a single HotCRP account which is associated with Paper X.

This seems like a clear bug to me!

[kohler]

PC member A did not list coauthors S in their Collaborators & other affiliations list. Also, coauthors S did not list PC member A in their Collaborators & other affiliations lists.

There is a missing feature here. One could imagine automatically supplementing Collaborators & other affiliations lists (which is what's used for conflict detection) with coauthorship reports from the current conference. Even better, since it would also improve conflict detection for other conferences, would be to automatically suggest that PC member A supplement their Collaborators list with the current conference's coauthors.