I've noticed that some people seem to find posts from years ago and moderate them, closing them as off-topic, duplicate or whatever. This seems valuable, since people can search the whole database of questions and improving it is a good goal, but how do people find these questions?
I routinely review every active question.
Old questions become active whenever they're touched by somebody. The "Community" bot often touches old questions that might have some value (because they were upvoted or answered).
I can no longer do this as thoroughly as I used to because there are about 200 new active questions every day. That's why we have multiple moderators!
(There is one class of exceptions: some community members are in the habit of making tiny modifications to large numbers of questions all at once. I recognize those people and the sequences of threads they have touched, and usually just skip over such posts.)
Obviously I am not commenting as a moderator, but I have often contributed to old threads either becoming "active" or entering the review queue.
Essentially it happens when I find an old thread and do something "stronger" to it than merely upvoting or commenting on it - for example if I edit it, answer it, flag it or vote to close it. I am an active editor; if I see a typo, an incorrect tag or poorly typeset math, I am keen to edit. Since many threads that come to my attention contain editable issues, a fairly high percentage of inactive threads that I have browsed to will become active again. If see an old question that has serious issues, I will vote to close it, or if I see an answer with issues that go beyond downvoting, I will flag it; these threads will then enter the review queue.
There are four main ways that I come across such old threads.
- I searched (either site search or an external search engine) for something that led me to the thread. That might be because I was looking for something myself, but is often because I am looking for potential duplicate targets for a recently active thread. In many cases this procedure finds not only a potential target, but also unearths other potential duplicates, which triggers another set of votes-to-close being raised.
- I followed a link in another thread to that target. Often these threads are suggested as "see alsos" and so there is a fairly percentage of duplicate candidates.
- I was searching for a spelling error. In general, if I find a spelling error on one thread (e.g. "nmoral distribution") then not only will I edit to correct it there, I will - if time allows - also perform a site search to identify other places this error has been made. This often brings up threads that have been inactive for years, but I still feel editing is worthwhile, particularly because a thread that I know contains one error generally contains several others!
- I was working through a bunch of threads with a common tag. Sometimes this is with the explicit intention of retagging, and such edits again result in the thread becoming active again.
So none of these four methods are the result of me systematically working through the site's old question starting at number one and working forwards ... but they all had some systematic element in deciding which threads got looked at and which didn't. I would be unlikely, for example, to trawl up an old thread with no inbound links, which used only a very common tag with thousands of entries (i.e. not a category I would likely work through thread-by-thread), which contained no common spelling errors, and which had a low ranking on search engines.
I look at the active list / the front page;
I look at posts that come up in searches (including ones I wasn't looking for);
I look at posts other people link to;
I look at posts in the "Related" list of other posts
(and so forth...)
I see a mix of old and new posts. If I see something that I think should be improved, I generally do something with it, unless I am in a big rush or if it seems like the present state of it should be "grandfathered in".