I'm personally fine with it and would moderately encourage commenting on old posts. It actually serves a few useful purposes:
1. If the question/comment is actually useful to other people, then it's useful in object level terms to other people on it's own terms.
2. I think asking questions is often helpful for clarifying your own thinking, so even if it's mostly rough notes to yourself it still seems cheap and pretty fine.
3. It's actually kinda nice to periodically give older posts a bit more time on the spotlight, just to expose them to newer users or remind people of them.
4. I think it's specifically useful for LW to move in a direction where content is maintained and improved over the longterm.
There are two categories of posts (for this purpose): Some are comments or reactions to something going on in the rationalsphere, or relatively lightweight explorations of a topic.
Some are summaries or exploration of topics that will remain useful for many years.
LessWrong is doing great work in bringing to the forefront the latter ones - sequence posts, highly-useful commentary, etc. Relevant comments and questions only add value to these posts - they make them better next time they get referenced, and they add a little signal that someone finds them relevant today.
For the more ephemeral topics, there's probably not much harm in commenting, but you may not get any answers.
Is there a good way to know if an AI safety post is "ephemeral" in the sense that it's no longer relevant to the current state of the discussion?
I really like the accepted practice of actively (if infrequently) maintaining questions and answers on Stack Overflow and the other Stack Exchange sites. It's great that people continually update them with new information, e.g. "This is unnecessary as-of version x.", and I think doing the same here is similarly awesome.
I reread a variety of online material, including the sequences, and I think it's great when I can contribute, if only in a small way, long after the material was first made available.
I've recently started working through AI safety posts written on LessWrong 1-3 years ago; in doing so I occasionally have questions/comments about the material. Is it considered good practice/in line with LW norms to write these as comments on the original, old posts? One hand I can see why "necro-ing" old posts would be frowned on, but I'm not sure where else to bring it up. You can look at my comment history for examples of what I mean (before I realized it might not be a good idea)