[jauco]

Real archivists (as in people that have archivist as a job description and work at places that have “storing data forever” as a mission statement) tend to store the data in multiple formats. The source + a few derivations. They also store a bunch of copies to ward against bitrot. And they periodically compare the copies.

Real archivists use a lot of data :)

[selfhoster11]

I think part of the job of an "archivist" archivist (as opposed to an amateur archivist), is making information accessible to others. For that, you need derivations, because nobody will necessarily know how to deploy a Mac OS 7 virtual machine, install Claris Works (or whatever it is), load the original file onto the machine, and then navigate the contemporary UI (with it's unusual conventions) to get at the information they wanted. For personal data, I already know how to get an old environment up and running, so I'm happy enough to keep multiple copies of the original and of any software I need to open it.