I've hollered about software preservation as a "thing we need to make anything work at all". Even at the Old Job in the 90s we were already dealing with tons of that shit. I ended up at one point rewriting some DOS control software for an outdoor LED display in Perl for a client because the original software was on a failed hard drive (in a failed 286, in 1999) and was just, not available anywhere.
i've seen people pass up or throw away stuff like System Restore CDs for ancient laptops, because they're "useless" or not as interesting to them as some lost media computer game or whatever, and I have to just metaphorically slap them across their faces. You cannot know everybody's use case. Someone, somewhere, has that exact model of laptop, that they're relying on for something just so damn specific for some critical or at least semi-important task. And wouldn't it be awfully nice if they had the peace of mind that, if it ever all went to shit, they could use the correct restore CD to get the thing back in working order, instead of having to deep-dive through ancient forum posts and FTP servers that might or might not still be running, deciphering the 8-character file names of a hundred some WinZip Self Extractors in hopes that maybe one will be the one that enables the PCMCIA slots in their bizarre, manufacturer-customized version of Windows NT 3.51.
This isn't even me exaggerating. In 2015, one of Paris's busiest airports had to shut down temporarily because their air traffic control computers, some of which were still running Windows 3.1, had failed. In a lot of cases, systems like this are so important that the time cannot be spared to take them offline even to upgrade them to newer hardware or software. And if it fails? To say nothing of whether anybody even knows how to fix it, even a skilled maintenance engineer would have trouble repairing a computer without the correct software. And the clock would very much be ticking, if it was not already.
okay but actually as someone who has spent a great deal of time writing virtual machines i've also spent a great deal of time thinking about this. did you know that, most of the time, writing an interpreter for bytecode (which is effectively an emulator which is a VM) is fairly trivial? if you're a company selling $250,000 microscopes i'm not sure why you're not writing your software to target a light VM (with a lot of host pass-through functionality for performance), preferably an open source one, so that your software can literally just run on anything anywhere forever
well, actually i do know. because if you're that company then if you forcibly obsolete your microscopes then you can entice at least some institutions with more money than sense into just buying another very expensive microscope. but my point is that we have the tools to prevent this kinda thing and nobody's really using them
EDIT: this includes games by the way. like, it is more or less impossible to maintain a unity game past like 1-2 years, but if you write a game for the gameboy or the sega genesis that fucker will be playable until the end of time
this is part of the reason why i use FNA and not say, monogame. FNA has the explicitly stated goal of supporting XNA games which are no longer maintained, it intends to be a drop-in replacement for XNA to the point where you can do it from the outside. that's still not perfect, and i started doing this before i knew much about VMs, but it's a damn sight better than using something that will bit rot if you look at it funny
