Hold duplicate detection has a flaw that allows duplicate hold requests to be placed on the same titles. The detection seems to apply on title and author information stored with the hold request (apparently on the SysHoldRequestSearchTerms table), which stores point-in-time title/author details that may change if the bib record is overlayed or edited.

This idea is for improving the duplicate detection to fix this gap, but also any similar gaps I may be missing where title-related data is stored separately when the hold is created, and referenced by the dupe detection (I'm not seeing separate tables storing elements like format or pub year, but in case there may be a similar dynamic with any other criteria that can fall out of date).