This has been a major issue for us, too. It first became an issue with repeatables: surveyors would duplicate repeatable records to save time, and then forget to edit all the fields that needed editing. So, for example, we would get dozens of trees all reported with the same location coordinates.

In the end I created some JavaScript anti-duplication code: I discovered that if the new-repeatable event code found a value already in a field, this was a good test that the repeatable had been duplicated. (I chose a field that gets auto-populated with a sequence number shortly after creation, so I could rely on it having a value if it had been copied, even from a draft.) So I made this event put a completely invalid value in a read-only field and displayed an error telling them to delete it. (Validation rules would then prevent them saving it.) I expect the same would work for new-record, but it feels like a risky use of an undocumented feature.

For me, having 'duplicate-record' or 'duplicate-repeatable' events would be an adequate alternative to a simple 'disable duplication' option, especially if the event handler code had a way of cancelling the duplication (i.e. preventing it going ahead).