Detaching from the parent task as this proposal does not block T18659, and not vice versa either.

Moving around query parameters is for the most part trivial to do. There are specific urls, such as those for action=raw&ctype=type/javascript (and same for CSS) that have been stable and canonical for a long time. Those should not be changed so as to ensure correct automatic purging of those imported script urls.

However, for other arbitrary query parameters, I think we could certainly change and standardise the order to prefer title-first.

The permalink (and likely any others that this is applicable to) is generated by calling $this->getTitle()->getLocalURL( "oldid=$revid" ), which relies on Title::getLocalURL. There, the relevant order is specified with $url = "{$wgScript}?title={$dbkey}&{$query}";

Revisiting this separate from T18659, I think we should decline this. It's only been reported once and that was 10 years ago with no other interest being shown.

This is technically easy to do, but I'm skeptical about its usefulness. Our links are already consistent, and they prefer title-first. For the vast majority of links users see and care about, the title is also the only part of the url. As such, making it last won't actually change anything for most use cases (such as shown in bookmarks and search results from DuckDuckGo etc.)

Can we quantify a problem with title-first vs title-last? A sense of confusion or user experience? I think we could probably make an argument both ways. For the most part, our urls are structured to have the most detailed part at the end. E.g. title=Foo, title=Foo&action=edit, title=Foo&action=edit&oldid=123, title=Foo&action=edit&oldid=123&section=2 etc.