15

Where do we draw the line between "I'm having the same problem, too" posts and partial answers?

This answer and its comments are what have motivated me to ask this. But I'm hoping for a general answer to this question, too, if it's reasonable to have one.

As Zanna pointed out, our official help page How do I write a good answer? says this:

Have the same problem?

Still no answer to the question, and you have the same problem? Help us find a solution by researching the problem, then contribute the results of your research and anything additional you’ve tried as a partial answer. That way, even if we can’t figure it out, the next person has more to go on. You can also vote up the question or set a bounty on it so the question gets more attention.

(the bold emphasis is mine)

But we frequently delete posts that may at least have been intended to conform to this rule. There seems to have been little awareness (including by me) that this is considered a legitimate use of answers, perhaps because it's rather vague. At what point does a post confirming a problem come to provide enough progress toward a solution that it crosses over into being a partial answer? Or am I asking the wrong question? Is the amount of information provided not the important question after all? Is the detail about what was tried important, even if it hardly approaches a solution? Do we just disagree with the help page and want to get it changed?

For me, the central issue seems to come down to this:

  • Sometimes it's easy to tell that an answer is an "I'm having this problem, too" comment and serves merely to confirm or clarify something small about the problem and should be removed or converted to a comment.
  • Sometimes it's easy to tell that an answer is a partial answer that doesn't solve every aspect of the problem, but still provides a usable solution or workaround for part of the problem or an explanation of the problem.
  • And then there is a huge expanse in between, where it's not clearly one or the other.

I'm not sure what my position on this issue is, because:

  • It seems to me that, right now, we are ignoring an official rule in the help entirely, and any post that doesn't come out, guns a-blazing, screaming "I'm an answer! I'm an answer! I'm an answer and this is the solution!" tends to get flagged and deleted in review. This actually goes beyond answers that don't provide a solution--I often see complete, definitive, fully tested answers flagged for removal because they started with "I have this problem too" or contained a couple of question marks.
  • On the other hand, there seem like good reasons not to allow "I have this problem, and by doing A, B, C, D, ..., I have confirmed it's not W, X, Y, Z, ..." posts to be answers. It places the burden of knowing it's actually the same problem, as well as of judging if the question is written in a way that will ever attract enough attention to solve the problem, on the answerer, who is often new to the site. If they posted a separate question, these problems would be avoided... at the cost of fragmenting what might be the same problem into multiple questions.

Finally, I wonder: Is this part of the help page just a holdover from back when we would close questions as duplicates of other unanswered questions? Should it be replaced with advice to post a new question while linking back to the original for context?

Currently, we do not dupe to unanswered questions. The system does not allow it anymore, except for questions asked by the same user--since users sometimes make multiple questions about the same problem when only one is needed. Questions are considered "unanswered" when they have no answers with a positive score or the OP's accept, and the system no longer allows other authors' questions to be duped to them. This change was met with much criticism at the time (including by me) but seems to have worked out very well.

Thus:

  • Before, a new question with substantially different information and troubleshooting steps that appeared to about the same problem could easily be closed as a duplicate of an older unanswered question without that information. By posting their findings as answers, people could avoid this.
  • Now, such duping is discouraged and largely prevented by the system. Furthermore, suppose someone posted their findings ("it's not W, X, Y, Z, ...") as an answer, and that answer got upvoted. Then other questions could be closed as a duplicate of a question where no solution has been provided.

How do we want to interpret our policy around these issues? Or, alternatively, what policy do we want to have?

  • 2
    I think that advising to post a new question would be a perfect solution. We know that lately there was such an answer posted. At first it was a pure "me too" post. Then after some comments the user made many changes to the answer and it contains some useful information. But this may be a waste of effort in this form. The original question is about an EoL release and may be closed. It is unlikely that many people will read the answer. A new question would be much more useful. – Pilot6 Aug 16 '17 at 9:50
  • Related (but not a duplicate): How can a new user escalate an old unanswered question? – Eliah Kagan Oct 15 '17 at 6:21
5

I would like to note that the post in question has been edited (and deleted), and I think Elder Geek's answer may be more directed towards the latest revision than the one that was current when this question was posted. I believe it was certainly a "me too" post at that point, as it only contained information on what the author had ruled out as causes of the problem through their research.


More or less just rephrasing what you already said, in my view the current situation is something like this...

  • Reviewers consistently delete answer posts like the one you mentioned in your question, which I encountered at revision 6 (here's a screenshot for those who can't access that link). This post does not contain a solution, but it contains lots of useful information that belongs in a question about the issue. Two experienced reviewers advised the OP to ask their own question, and so did I. It's worth noting that the post has now been deleted, with the last vote cast by a mod, even though it now contains a workaround of sorts.

    Suggesting the post author ask their own question is what we usually do, and since we have canned comments for this situation, like:

    If you have a new question, please ask it by clicking the Ask Question button. Include a link to this question if it helps provide context.

    it seems this is what the review system is directing us to do.

  • Experienced reviewers and mods (and I) often defend (by upvoting, editing, voting Looks OK, commenting @ or for the benefit of other reviewers, etc) flagged answer posts that do some combination of these things:

    • explain why what the OP wants can't be done, or why their question represents misconceptions
    • provide an explanation of the problem
    • provide a workaround or alternative way of approaching the problem
    • provide a valid answer framed tentatively
    • start with "I had the same problem..." and go on to describe how it was solved in the author's case
    • include a closing question, perhaps an aside to the main issue, alongside some useful solution

    So I think we consider such posts answers, depending on the details of the case.

  • If someone attempts to edit information like this (details of the same problem on their own system along with research and results) into a question, their suggested edit will almost certainly be rejected as against the author's intent.

These actions tend to promote the creation of new questions.


Considering the type of edge case that raised this issue, where some useful content that belongs in a good question is posted as an answer:

  • If the original question is good and has the requisite details already, then the new post wouldn't be adding anything useful and there would be no reason to keep it
  • If the original question lacks such detail, and hasn't received much or any positive attention, the new post is likely to languish unnoticed, rather than helping anyone find a solution.

Thinking about two groups of people - folks searching for solutions to their problems using search engines, and users here who review, vote on, and answer, it's the latter group who need to see the information posted by the "me too" poster, and the attention of the latter group is mainly on new questions. A new question is, I think, a lot more likely to get an answer than a bumped older one.*

Let's consider two scenarios. In both "I did A, B, C and D and determined that neither W, X, Y nor Z is responsible for this" (the content of a good question) is posted as an answer to a poorish question. In the first scenario, this post is left as it is by reviewers, per the Help Center's advice, and later a good answer is added by someone else, or the "me too" post is later edited to be or include what we would consider (in our current practice) a valid answer. In the second scenario, the "me too" post is deleted, per current site practice, and the author of the deleted post writes a new question, which receives a good answer, from the OP or someone else. The old question can be closed as a duplicate of the new one, if it's still open, providing an additional signpost to the solution. I think the latter scenario is much more likely (I think the old question is less likely to get an answer), and that matters here, but even putting that aside, I think it is the better scenario...

In both cases, we (site users and visitors) got the answer, but only in the second case is the answer attached to a clear, well researched question. Whether (or how much) this matters is debatable, but I consider it broadly a good thing in terms of helping people find and understand the information they need here. Creating links between the old and new posts will assist towards that goal, if the old question isn't misleading or otherwise worthy of closure as anything other than a duplicate.* Also, the person who asked the new question will actually get notified of answers to the question, and is more likely to get some upvotes and might therefore feel good about the site and gain some privileges that encourage them to participate more, which we like, I think... but that brings me to another thought...

If we change our current practice to allow "me too" answers that add useful information (assuming we as a community and particularly as reviewers are able to draw the line on useful information, which is another issue), would we upvote them? The visible criterion for upvoting an answer is "this answer is useful", which allows for extremely broad interpretation...

Upvoting such posts seems wrong to me, because the question still needs an answer, and the system will consider it answered if any of its answers have a score of at least 1. That means it won't show up in filters that some experts use to find questions to answer with particular tags, and so on (I also use unanswered filters to do cleanup, which means giving attention to old posts, and I believe I do some modestly useful things in terms of helping folks find the answer they need that way).

So after thinking about this for a few days, I'm concluding that we should keep deleting "me too" posts even if they add useful information, contrary to the Help Center advice:

TL:DR

  • If a post does nothing but add useful information that belongs in a question, it should be the basis of a question, because in the answers it

    • probably won't be seen by those who need to see it at this stage (answerers)*, unlike a new question
    • might get an overly generous upvote, making a question with no solution appear to be answered, thus making the original question less likely to be found by those who need to see it at this stage (answerers and janitors)*
    • fail to help those looking for solutions

So, we should carry on more or less as we are doing now, but

  • IMHO, as reviewers we should do our best to encourage authors of "useful" "me too" posts to Ask a new question if they have useful information lacking from the original question, and encourage them to link to the old question. If it's appropriate, we might also consider editing some of the information into the original question, or posting it in comments on the original question.

*My reasoning might be faulty because I'm focusing excessively on my own experience of the site and flawed assumptions about who else it's useful to and how they use it. In particular, I may not be adequately considering new users who are inspired to answer a question they found unanswered, because they subsequently found a solution. I don't think it should be harder for these people to find a new question with all the relevant information than the original question with extra information added provided the new question has the right search terms, but it's worth considering that page views are what push a question higher up search engine results. This may be an argument in favour of allowing "useful" "me too" posts on the off-chance that they will help the question to be answered by casual visitors. If we decide to keep deleting those posts, then the care of reviewers to link the posts if the original question is useful is needed to avoid this potential disadvantage.

  • I for one agree with you. You are absolutely correct regarding the timeline which I somehow neglected to consider when answering the question. Perhaps I should delete my answer. – Elder Geek Aug 20 '17 at 21:32
  • It would have been difficult to determine the state of the post at the time Eliah Kagan asked this question from the timeline @ElderGeek - I just happened to be there! I'm not upvoting your answer because I think what EK is really asking is "should we delete these posts?", and I think we should, but possibly your answer works to justify the defence of posts that do provide a partial answer, and I agree with defending such posts in general (although I agree with the deletion of this post since the suggested workaround is basically "if it doesn't work, don't use it") as I wrote in my answer... – Zanna Aug 20 '17 at 22:07
  • Difficult to determine the state of the post at the time the question was asked? I thank you for that, but it certainly wasn't impossible to check the timestamps, so my fail. Just prior to deletion there was some work being done in an attempt to improve the answer and while choosing the other GPU isn't an ideal solution, the workaround could feasibly be useful to someone experiencing the same issue, the poster was obviously working on the problem and may well have provided an excellent solution given time. Or not. We'll likely never know. Thank you for clarifying your perspective. – Elder Geek Aug 20 '17 at 23:28
  • 1
    This is quite complete in terms of the various considerations for reviewers and possible outcomes for post authors. At some point we might have to revisit the wording in the help... but I don't know how much general discussion about the issue can really be fostered here. I only had one example, and without others to help explore other aspects of the issue, it would be hard to figure out if any new policy consensus or feature request ought to come out of it. I'm accepting this answer. If the issue arises again in connection with other posts, someone could post on meta and link here for context. – Eliah Kagan Aug 23 '17 at 20:14
1

When I posted this, This was a partial answer. And yes, it would likely have been better as a new question. Be that as it may, The aforementioned post clearly shows a commitment to improve it as the edit record makes abundantly clear to the casual observer. edits

In this case, it's unclear whether the answer related to the version of Ubuntu asked about or a more recent version, but that information could have likely been easily obtained (considering the obvious commitment to improve the answer). The Nvidia driver versions mentioned all appear to have been released recently with the oldest version released about 8 months ago.

Since the Answer the question section of https://askubuntu.com/help/how-to-answer states clearly that "Any answer that gets the asker going in the right direction is helpful, but do try to mention any limitations, assumptions or simplifications in your answer. Brevity is acceptable, but fuller explanations are better." I think it's safe to say that this was an honest attempt to help on it's way to becoming a complete answer, and I hope we didn't drive away someone who was Truly Eager To Help.

It might even be fair to say that all answers are partial answers as they can be edited and improved when new information becomes available, updated to cover new versions, etc. I do this fairly often myself.

Edit:

Guidelines are an important part of maintaining a cohesive policy and it's important that we are all at least trying to accomplish the same thing (Which I think is improvement of the site as a whole)

This is a great question and if as you surmise, this a holdover from back when we would close questions as duplicates of other unanswered questions (before my time here, so I wouldn't know) and we aren't going to accept partial answers that contribute the results of your research and anything additional you’ve tried then clearly the guidelines should be re-written to reflect this so that users attempting to follow the guidelines don't get blind sided by an unwritten policy.

Clarity of instructions is required in order to get desired results so lets be clear about what we want so that we can obtain it.

0

An answer is an answer if it answers the question completely.

I know it sounds ridiculously easy, but if an answer does not answer the question nor shows that the question simply cannot be answered, it should be not be posted as an answer. The information should go elsewhere, as part of an answer that does fix the issue, in the comments, in the chat, but not in an answer. We do not deal in partial answers. I know that's counterintuitive, and I've posted partial answers myself, but have been criticised for it.

In the case of partial answers, this often indicates that more work on the question is possible. It could be split in easier questions, for instance. But not duplicated. It might also indicate that the question is too localised. We don't have to solve all the Ubuntu problems in the world.

I might simplify this too much.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .