When a person in 2017 is looking at an answer that may have been written in 2010 (say), it's not that person's responsibility to try to figure out if the answer might have been correct at some time many years previously.
Either the answer is correct right now or it isn't. The correct actions when faced with an incorrect answer are already laid out by StackExchange -- comments and downvotes.
Ultimately StackExchange is meant to be a source of help to people who are searching for answers. The overwhelming number of people helped by StackExchange are not those who come to have their questions answered -- it's those whose questions already have good answers and so they never even need to ask. [StackExchange is not primarily intended for just answering the question people have at the time they ask -- that's merely the source of the questions for the larger purpose.]
This emphasis on people being able to find correct answers via search engines has been explicit in the design since the earliest days - since before any of us were writing answers for CrossValidated.
As a result, old answers that are no longer correct cannot stand "as is"; if the old answer isn't modified when the answer changes, we need a way for currently correct answers to get higher reputation so that when people search for and find an answer, they can identify which one the users on site think is the best one.
Wrong answers accruing downvotes is StackExchange working as intended.
I've been answering questions since August 7, 2010 (though my earliest answer is currently deleted). A lot of my answers are pretty old.
If I gave an answer that was correct at the time of posting that had accrued upvotes, but facts had changed since the answer had been written, then I would sincerely hope that my answer that contained incorrect information would not simply be left alone.
Ideally any outdated answer should be fixed by its author (or at the very least edited with a very clear note that it was correct at the time of writing but facts have since altered), but of course if you have several hundred answers it can be impossible to simply notice and remember. So authors will need a nudge -- something to draw their attention.
As I mentioned, there are two main mechanisms within the stackexchange software for doing so, comments and downvotes. So the person that finds an old answer of mine that's now wrong should choose at least one of those mechanisms to try to redress the problem (that new people searching will see an incorrect answer).
Which mechanism to use?
Comments are good because they might prompt the author to fix their answer. A correct answer is better than a wrong one. They also - at least temporarily (since comments might not stick around forever) - give some indication of what the problem is, but I know from experience* that readers often ignore the comments, especially with longer comment threads.
Downvotes are good because they change the ordering of answers (when sorting by votes at least), and make it easier for users to choose between competing answers. They also work when the author isn't currently active (which is true for the authors of many of our old answers). Sometimes a downvote will come with a comment (after all, if it's wrong, it's wrong), but it's not required.
So both mechanisms are useful but downvotes have the advantage of not forcing the person choosing between the two actions to simply rely on the original author being able and willing to act.
There is a third alternative for people with sufficient privilege, which is to edit old answers, particularly if the original authors appear to be inactive. While we should avoid altering someone's answer substantively in general, it would often be reasonable to add line that the answer may have been correct at the time of writing but which may now be outdated. Such a note would help people avoid relying on presently incorrect answers; additional detail could be added in comments if you're really unwilling to fix the answer.
* specifically I have more than once (twice, perhaps three times) answered questions on other forums which relied on well-upvoted but actually incorrect answers from CrossValidated where the mistake was noted in comments (posted long after). In both cases that I clearly recall at present, the poster of the new question that was relying on a wrong answer didn't read all the comments under the answer (and fair enough, posts are meant to stand on their own and the comment threads were long) or had seen them but concluded they must be wrong (the answer was heavily upvoted after all, a couple of voices to the contrary wouldn't convince someone who didn't understand the explanation of why the answer is wrong). In both cases the author of the wrong answer had not responded to the comments explaining the problem, presumably due to inactivity. In that case the responsible thing to do is downvote the incorrect answer (and it may need a lot of downvotes not to be among the top answers so that people stop relying on it) and if possible, also post a new, better answer.
