This just means that functionality and interoperability criteria are more important than usability.

Sometimes yes. Usually no, for most people. If you make a word document in an older version of office, it'll still work fine. If you use LibreOffice with the oldest-looking UI, it'll still work. 99% of people don't use the extremely niche features that have been added in recent years.

But people by and large don't do that. They typically use the newest version.

This is the opposite of confirming your argument about UI\UX, because this means that UI\UX are order of magnitude less important in making the decision.

No it isn't.

How is using software with modern interfaces actually a confirmation that people actually prefer older UX?

That's simply because they "theme their system" to look as they wish and they don't have to stop with Win98 or Win2K.

Exactly. And almost nobody themes their system to look like the supposedly superior in UI/UX Win95/98/2000. Indicating that maybe people don't actually want a UI from that era, despite Reddit and Lemmy insisting that everybody does.

Ergonomics is not a matter of opinions, there's plenty of research

Exactly. And that research has lead to where we are now.

Padding controls and indicators with space can be a good thing,

Is a good thing.

They've all heard something of it, but haven't learned the actual thing.

No, they've generally improved it, and listened to actual UX usability studies.

Older UIs were usually (often, but not always) made with respect to ergonomics.

They almost never were. Seriously. Go back and try some 90s software. Most of it was a cluttered mess, ugly, really weirdly laid out, and had zero considering for anybody with disabilities.

Our ideas of all three things seem to be diametrically opposite. For me older UIs seem ordered, compact and correctly accented

And that's fine. You can think differently. But most would disagree with you, outside the Redditor/Lemmy bubble.