Generational backups are a great example. If I do nightly full backups of every change but do not retain historical check-points, the issue you describe is not just likely, it is inevitable.

When you think of *that*, it makes the volume of data that is already huge exponentially difficult to manage. Dedupe helps in a situation like this (where you only have to store the revision CHANGES and can create each checkpoint from a base document) - but it still means that you're storing multiple revisions of each document for archival - and for how long? 10 years from now you get an eDiscovery request for a *particular* revision of a document, and you can only provide the final version, when the pertinent edits have been conveniently removed. You're going to face fines. Just ask RJ Reynolds and Phillip Morris.

Backup is a nightmare.