I have to agree in principle around the idea that a game should probably not be remade or remastered at all. Short of making it work on existing hardware. The trouble lies in that emulation only gets you about 90% accuracy at most, and remaking the engine requires a lot of work, for a game that people already played and there’s no secrets to discover.
Hence I’m more inclined to let a “remake” pass if the spirit of the original game is intact, and it’s not simply being remade to monetize it.
One of the issues with using something like DOSBOX to emulate hardware is that it doesn’t machine-cycle accurately emulate a legacy 8088/80286 PC, it emulates a DOS environment with only some upper-bounds on the CPU emulation. So you will NEVER get an accurate emulation of any PC game on DOSBOX, and in most cases, you don’t want to, since that would also mean having the floppy/cd-rom/mechanical hard drive loading speed of a legacy system that was never even a consideration when those games were made. So getting a game to work on DOSBOX still involves just as much finagling with configurations that a legacy PC would have, if not more if you want to make the game look and sound like it was designed, and not simply PC Speaker and/or Adlib music.
SCUMMVM on the other hand takes the “remake the engine” approach, and this works for some games, it means that it has to be aware of every bug and work-around in the original engine. So for many games (especially SIERRA games), this means that you have to be aware of which bugs exist in certain games/platforms, and either keep the bug in for accuracy, or try to fix it (if it would normally crash the game.)
Neither are really an appropriate way to play the game like it was intended, hence it may be better to start a new game from scratch based on the original game’s content, maybe even replicating the look and feel of the original game, but not re-using the original game assets at all. At that point you may as well expand upon the original game since disk space is not as much of a premium as the original game was. In many old floppy-disk era games, content was cut or dummied out to fit a disk replication budget. Even when CD-ROM came along, most of that space was filled with speech, and often at a compression and sampling rate that is embarrassing by todays standards. If something like Ogg Vorbis or Mpeg I Layer 3 Audio was available then (which MP3 was in fact available in the late 90’s) the software implementations of the time were simply not up to task and required a high end system to play just the audio.
The same can be said of texture compression now as well. Early 3D cards didn’t have it, then every card produced since 1998 pretty much required it and the patents on that only expired this year. Now GPU’s have 1-8GB of RAM and compression is a bad idea for UI Widgets and 2D style content. Likewise 8-bit textures haven’t been a thing for quite a while and attempting to use them simply generates a 32-bit texture anyways, so trying to take advantage of 8-bit tricks like palette cycling simply isn’t possible on current hardware. Some other trick like a 1-dimensional texture with a shader, must be used to do that, or streaming them like they were tiny videos.
At any rate I think the larger problem is that a lot of proposals for remaking a game are really a bad-faith argument to monetize an old IP, and use that IP’s built in audience. You see this kind of thing with art/comic anthologies right now, where the anthology with a big name artist, results in a lot of sales because of that artists fanbase. Do you really want the same three games remade every 7 years like what happens with films and TV shows when the IP holder changes hands, and the content platform changes too much to run the previous one? I don’t know about everyone else, but I’m not willing to buy the same game for 3 different consoles and a PC every 7 years because the console changed or Windows/MacOS changed and it can’t play the old one.