Yeah it’s an interesting one.
The archivist in me thinks that fixed versions of games should be versions of games, and not a pseudo version that sort of arises from the emulator/tool massaging the implementation.
For example there are numerous editions of Monkey Island 1 & 2 that were released for various platforms with many small variations. Should the tool be smushing those variations into one definitive version, or letting the differences stand — like a missing prop description, or minor dialogue fixes?
If so, then what about bugs? Different versions of classic games have different bugs because of how patches didn’t exist back then. Should the tool also smush those fixes together?
I think for casual players or those seeking the ‘best’ experience then it does make sense to do the above. However, few casual players are going to be using ancient versions of games and trying to run them on modern systems — they’ll just pick up the Special Editions.
It’s going to be enthusiasts playing these old versions, for whom I feel the tool’s primary job is to preserve the original game exactly as it was, and if there is any ‘fixing’ that is done very transparently in a toggleable way. Otherwise, what if you do want to play SOMI with its bugs intact — whether it be for the experience, or to check something (i.e. research)?
With that said, I think the ship has sailed as far as ScummVM being a serious archival tool goes. It seems the only real avenue for that is using something like DOSBox, or another way of simulating the original runtime conditions, so that there is no interpretation or decision making happening.
BTW this wall of text isn’t meant to savage your contribution Thunderpeel. It is more a general concern I have about the scope and direction of ScummVM, which is admirably volunteer-run but also constitutes the only official way of playing a lot of games, and the primary way people play them unofficially.