@moz @ajft @samhainnight having worked in system administration in vaguely similar spaces it feels very bad to prematurely merge potential duplicate records because they usually cannot be gracefully unmerged if you've made a mistake

I've had fantasies of making a CRM where, instead of a person's record being this monolith and everyone/everything's updates being mixed in irretrievably, all the updates are marked with how the user identifies the subject:

"Joe Bloggs, owner of joebloggs@gmail.example"

"Joe Bloggs, subject of record ID 24601 in the old system"

"Joe Bloggs, resident at 22 West Wallaby Street"

"Joseph Bloggs, resident at 22 West Wallaby Street"

and if it transpires some of these identifiers are for the same person, you can link them and the users see a combined record with all the updates made against all the linked identifiers. but if it turns out you made a mistake you can unlink the identifiers and separate the Joseph Wallaby casenotes from the Joe Wallaby casenotes