Right now I owe myself quite some sleep, so this likely ends up as an incomplete answer; nevertheless I wanted to put down the thoughts that I had about it.
StackExchage sites are natively excellent for short-term information exchange. The noble and understandable intent to keep AskUbuntu's canonical questions well-maintained, however, forces a second role onto this site: it introduces the role of a documentation. Desire to live up to this role, however, may push mods into patterns that go against the the nature of how a Q/A site tries to work.
Closing questions as duplicates forces users to find their way against the grain of the site (e.g. arguing in comment threads, appealing to reopen, commenting in outdated threads, etc.). Frustration, like "Those 8-9 year old posts don't answer my question" ensue.
Navigating merged posts — with the current implementation at least — may prove to be confusing.
The principle of decaying information + theorizing on ways to cope
Linux and Ubuntu evolve over time. For this reason, static records of information that try to describe them (answers on the site), slowly, but inevitably lose their validity. Surely, some info holds up better, while some age faster. But the tendency is the same. Information recorded in a database is slowly rotting away.
One way to keep information up-to-date is via manual edits. This however is not native to how a forum works; this is how documentations work. Also, it is pretty work-intensive.
Following the aforementioned principle, newer questions are always in a better position to collect up-to-date information — whose shelf-life expectancy is also better.
Allowing deprecated information to get replaced with valid one is aligning well with the grain of how a forum works, and delegates the lion's share of work to the site users.
I'd like to offer an adaptation of Lao Tze's guidance:
Governing a large country is like frying a small fish. You spoil it with too much poking.
The adapted variant being:
Managing large amounts of evolving/alive information is like frying a small fish. You set yourself up to a Herculean task with swimming against the stream. (Plus you may spoil it too.)
In principle, aim to achieve the goal with the least amount of least intrusive interventions, and the least amount of work. Utilize well what the site already delivers instead of over-sculpting / rewriting it.
I would like to suggest to be slower and more considerate on closing "duplicate content". The discipline to discourage redundant-appearing posting could stay, but the enforcement could become more lenient.
I would advise against trying to force several years old threads to remain alive just because they are already there and there are a lot of upvotes. Several of them may be beyond their peak usefulness. Even more of them will be so a year or two later.
The interpretation I'm offering invalidates the very term canonical post. It's not the posts themselves that are canonical: remember, the posts are slowly decaying.
Rather, it's topics, that are evergreen. Or something along that line.
- I suggest relying most on interlinking questions.
- After identifying an evergreen topic, maybe a "hub-question"could be created for it, with the goal of interlinking several organic questions that deal with its premises.
- Then high quality organic questions — that have accumulated a lot of useful information — could be identified, and and marked as "Elevated", or "Staff pick" questions. These could be the ones that could be linked in the hub-questions.
- In this case, aging would mean less of a challenge, because a question marked as Elevated would not attempt to guarantee up-to-dateness; it just testifies, that at one point in time, it was highly helpful.
- The parent hub-question could try to offer an insight on the relative values of Elevated questions linked within: by ordering, displaying (well curated) tags, question age, average answer age, and upvotes (both question upvotes + cumulative, inclusive answers).
- Hub-questions could organize the interlinked questions in chronological order, by default the newer being on the top. Or one could write a script that would update the ordering with the overall number of upvotes also taken in consideration.
- Possibly, if the need arises, the number of Elevated questions linked from a hub-question could be limited, with the least valuable ones getting pushed out.
- Alternatively, visual organization, usage of layouts, grouping, and appropriate typography (headings) could aid the overview of those lists with several items.
- Organic questions would also contain a link to their parent hub-question.
- A question could contain a link to a corresponding hub-question, even if it's not elevated, and is not being included in the hub's list. This way, if someone finds a lower quality (but already hub-link equipped) question via googling, they could quickly navigate towards highly valuable information.
- In this sense, adding the hub link to a question could be analogous to how tags are added. Regular users would be given the right to identify and carry out the association of any question with a relevant hub-page.