Closing a question on Stack Overflow doesn't delete it or hide it from public view, so the comparison doesn't make sense. Closing an old, popular question only prevents it from receiving new answers and puts a banner at the top. The point is to avoid setting bad examples for new questions. The fact that a question was well received many years ago does not guarantee that it's in agreement with current policy.
Additionally, we generally do not close old questions simply because they're "outdated", e.g. refer to deprecated libraries etc. We recognize that people are often stuck maintaining unsupported legacy systems, effectively indefinitely. We sometimes close questions because they refer to services (especially web APIs) that are no longer available. But overwhelmingly, when old popular questions get closed, it's because they're deemed to be no longer on topic for the site. Since a lot of people will see the question, we don't want them to get the wrong idea about what's topical.
And, of course, it makes perfect sense to downvote things that used to be correct but are now incorrect. Practically speaking, this doesn't happen nearly enough; upvotes have a kind of inertia, and wrong answers are often evaluated by people who don't know they're wrong.
Additionally, we generally do not close old questions simply because they're "outdated", e.g. refer to deprecated libraries etc. We recognize that people are often stuck maintaining unsupported legacy systems, effectively indefinitely. We sometimes close questions because they refer to services (especially web APIs) that are no longer available. But overwhelmingly, when old popular questions get closed, it's because they're deemed to be no longer on topic for the site. Since a lot of people will see the question, we don't want them to get the wrong idea about what's topical.
And, of course, it makes perfect sense to downvote things that used to be correct but are now incorrect. Practically speaking, this doesn't happen nearly enough; upvotes have a kind of inertia, and wrong answers are often evaluated by people who don't know they're wrong.
By the way: about 89% of up/downvotes ever cast on Stack Overflow are up (https://data.stackexchange.com/stackoverflow/query/492368/to...).