A meta-comment about posts following up on old threads

[Mike Lacy]

I rarely find that a posting following up on an old thread is useful, and I wonder if that practice should be discouraged on StataList.

By "old," I mean threads older than (say) a few weeks, and certainly threads older than six months or so. I seem to find myself all too often digging through a long thread from a year ago, where information gained by the new posting's relation to the old ranges from not very helpful to actively irrelevant. Even if relevant, going to the bottom of a mult-posting thread occupying multiple pages is not convenient. To me, a new posting with a linked reference to an old thread would be much eaiser to read than repetition of a long old thread with someone new problem at the end. Of course, there are situations in which a new posting needs to be interspersed with the old conversation, but I think they are relatively rare. Those latter situations typically involve postings by more experienced and knowledgeable participants, and I don't think those are commonly a problem.

However, what I think is more typical is that relatively new participants search the archive of the list, find something they think is similar to their problem, and presume that their new posting is expected to be a followup if it has any terms/concepts in common with an old one.

Do others here have a similar reaction to the typical followup posting (perhaps not)? If so, I'd think that we could include recommendations in the FAQ (don't laugh too hard) not to post one's question as a followup to a posting older than a few weeks. In the extreme, such followups could presumably be made impossible by tweaking something in the forum software settings.

[daniel klein]

However, what I think is more typical is that relatively new participants search the archive of the list, find something they think is similar to their problem, [...]

I find this to be a problem, too -- at times. I suppose participants take the encouragement to search the archives (which is a good thing to do) to mean that they should not open a new thread if their problem -- or what they think is similar to their problem -- has been discussed before. More often than not, the old thread turns out to have little in common with the (new) problem. I think we should encourage starting new threads (with tags, btw.) if a problem cannot be solved completely by reading through old threads. Obviously, linking to the relevant old threads would be great.

The crux of the matter is that the problem arises more frequently with new participants who often do not read the FAQs. Therefore, a change in the forum software that closes old threads after, say, a couple, of weeks (similar to making editing impossible after a couple of minutes) would probably be the only way to really improve on the current situation. Whether that is desirable (i.e., does not have negative side-effects* that outweigh the benefits) and worth the effort, I leave to others to decide.


* such as not being able to announce updates to community-contributed commands in the original post, or being able to add information to older threads, that might help people to solve problems with the additional information, etc.

[Nick Cox]

I don't have an easy answer here. A new question in an old thread is not what is recommended and it can be frustrating to have to read through several previous posts to try to establish what is relevant. In fact, there is a range from the question having been answered already in an earlier post in the same thread to the new question being totally separate. People sometimes just post something in the last thread they used and/or in a thread of someone who was helpful to them earlier.

I have often suggested that people start a new thread with the same question and anybody thinking that would be a good idea is totally welcome to make the same suggestion. (The old thread should be marked with a cross-reference.)

The success of the FAQ is measured silently in posts that benefitted from looking at it and the failure of the FAQ is measured noisily in threads from people who don't seem to have (tried to) read it at all. Then we get flak on social media for reminding people of requests made every time someone posts.

The biggest problems we have in my view remain questions that don't show enough data, code, or results to follow what is happening.

[William Lisowski]

I will add an iconoclastic problem with posts to old threads.

I rarely read the forums directly. I rely on the RSS feeds from the General and Mata forums to alert me in my RSS reader to new threads, which I then either subscribe to or choose to ignore. And when I post on a thread, I subscribe by default. Further, I give in to my inner OCD self and every month or so remove my subscription to threads that have seen no action in a month, precisely to avoid the experience Mike highlights in post #1. Only if I'm called out by name in a thread to which I contributed six months ago will I actually see the new post to the old thread.

As to changes to the FAQ, I'll go with Eric Idle's memorable "A nod's as good as a wink to a blind bat, eh?"