Exit strategies for "chameleon questions"

Question

I'm not sure if there's already an existing term for this, so I'm inventing my own.

(tl;dr: I call them "chameleon questions" because they change every time you submit or edit an answer. If you're already intimately familiar with the phenomenon, please skip past the first set of bullet points to where I ask for recommendations.)

---

I think many of us have experienced this on occasion:

1. You open up a narrow-looking question, usually dealing with some specific error message. The question probably contains a snippet of sloppy but not quite awful code; code indicating that the author, typically a new/low-rep user, possibly a Mort, doesn't really know what they're doing. I'm not trying to be condescending here, they're probably very intelligent and hardworking, but in this specific case, they're clearly in way over their head.

2. However, as a question, it's not really too bad. The goal and problem are both clearly stated. The code is formatted (sort of). Mort was polite and checked their spelling. The question looks answerable, although from a rep perspective^* it's an obvious dead-end, it's had maybe 3 views in the past 30 minutes.

3. So you decide, you know what, this person took the time to write up the question properly, they deserve an answer, put on your good citizen hat and throw them a bone. You slog through the ugly code, find the offending line, direct their attention to it and explain why it's wrong and how it needs to be fixed.

4. Later, they leave a comment saying that yes, that was indeed the problem and your answer helped them fix it (great!). Oh, but, now there's another problem (uh-oh...). Something else is broken. After letting out a long sigh, you read the comment and/or edited question, roll up your sleeves, and write up Take 2, hoping that's the end of it.

5. And of course it's not the end. Now maybe the error is gone, but it's not doing what it's supposed to. If you're anything like me, your patience is now starting to wear thin. It's lousy code and you didn't volunteer to help them rewrite their whole app. And of course, the question still only has 12 views, and nobody has upvoted either the question or your answer, including the person you're helping, so this is turning out to be a genuinely thankless task in all respects. Welcome to "Family Tech Support", Web 2.0 Edition.

_{* Which is of course just an arbitrary number that none of us care about. Not really, anyway. Not very much, at least.}

---

So, here's where I get to my question. At this point, do you:

• Leave a comment, stating as politely as possible that you've done your best to help them answer the original question they asked, and if they have a new problem then they should start a new question? (I did this once, and actually ended up with an accept, but I felt kind of guilty afterwards, I don't think anybody answered their 2nd question).

• Bite the bullet and edit your answer a 3rd, 4th, 5th time, however many times it takes to resolve their issue, knowing full well that they may not even remember to accept the answer once you're done? (I may work for free on SO, but that doesn't mean my time is worthless.)

• Just abandon the question, hoping that it will force Mort to start thinking for himself?

• Something else?

I've tried all three of the above and none of them ever quite feel right. I do want to help, but at the same time there are a lot more "lucrative" questions I could be answering. And I don't mind spending a lot of time on an answer either, if the question is interesting or important to me, but debugging is dull work.

I'd prefer not to reference specific questions here; I'm pretty sure that this is a shared experience, and my objective isn't to point fingers at "misbehaving" members. I'm more looking for some practical advice on how to dig myself out of these pits of despair.

Anybody have any other ideas?

Answer 1

Generally as long as the OP is still trying to solve the original problem or something closely related, i.e., my answer still makes sense, I'll keep helping. Once the topic starts to drift or it's clear that whatever I write the person just isn't getting it, I'll just let the matter drop for awhile and make a mental note to check on it later when I'm less busy. I might make a suggestion a new direction might be better handled as a new question. Generally, I don't feel bad about this as most questions get answers and a new question is an opportunity for them to get more rep. Eventually, the time between check ups increases to the point where we both give up.

Answer 2

Since this was an older question, I feel the need to add a new answer:

Now that we have a real-time chat system, use it! I have found that chat is the perfect solution to questions that require a lot of back-and-forth. I would suggest posting a comment to your answer inviting the user to join you in a chat room. This will encourage new users to chat, but also lets others who are trying to answer know that you're in the chat room.

If the user joins you, ask for details in chat and update your answer with new suggestions, asking the user to refresh your answer. Optionally insert into your answer links to the detail provided by the user in chat, so others can see how you're miraculously learning more details about the user's question.

Here's an example of a question I answered while chatting with the asker, and it worked out very well.

Answer 3

Here is the approach that has been proven to work pretty well for me. ^{1, 2}

As soon as I sense that Oh, but, now there's another problem I reply with comment like this:

Looks like <refer to original problem> is solved isn't it? As for <refer to "next" problem>, that would be a different question, consider posting it separately.

For easy copy-n-paste.

Looks like <refer to original problem> is solved isn't it? As for <refer to "next" problem>, that would be a **[different question](https://meta.stackexchange.com/questions/43478/exit-strategies-for-chameleon-questions "at Stack Exchange sites, “chameleon questions” are not quite welcome")**, consider posting it separately.

^{Interesting that above approach "utilizes" your question to deal with the kind of issues you're asking about. Feels pretty meta.}

Answer 4

Really depends on the question, and the mood I'm in. If the questions are closely related and I think I can get a good answer out of it, I'll put the time in; if it's devolving into a hand-holding discussion, I'll probably just walk away (and may even delete my answer).

In at least one instance, I've rolled back the question and asked the author to post a new one.

Answer 5

Another possibility would be to implement a "Create followup question" feature, normalizing the process to some extent.

Answer 6

I give the asker something to do. Make him/her show effort to see if I want to keep showing effort on my side. I say something like

Please expand your question further. Give more detail to clarify your problem more. Add necessary info from your comments into your question...

I am busy in the SQL area and I often tell the user to add a SQLFiddle to demonstrate the problem further. That takes time and many users don't bother to take the effort. But if they do I am willing to help.

If the user shows effort I try to keep up if the question isn't a total different one.

Answer 7

I know exactly what you mean. In fact, these questions typically cause me to add many "Update" portions to my answers, over, and over, and over again. I generally try to stick with it for a few iterations - not everybody can get their point across the first time. Some times it takes some exercising to explain exactly what their problem is. After about 3 or 4 updates, it does start getting a bit fatiguing. If I cannot answer the question in about 4 exchanges, I will likely leave my answer for a reference, and abandon the mission.

Answer 8

Sometimes a question just feels like it's missing something. I can't really describe it in words. It's like they've told you the problem, but there's one minute, seemingly insignificant detail they've omitted that turns the entire problem upside-down.

Whenever I get this feeling about a question, I now hit the Back button immediately. (It took me a few tries before I learned this lesson.)

Answer 9

Perhaps this isn't a viable solution for other sites, but the one Mi Yodeya recommends is discussed here.

Instead, the question edit should be rolled back, regardless of who made it. When rolling back the edit it would be a good idea to leave a comment linking to that Meta policy so that everyone involved understands what is happening, and why.

If the asker really has a different question there is a simple solution: create a new post and ask it. Asking questions is free (except in rare situations) so there is not much to lose. The only issue would be that the asker may be stuck with a question that they don't care about. I don't think this is such a big deal, and it may even cause users to spend an extra minute making sure their question is clear before posting, which (I think) is something desirable.

Simple and straightforward: Rollback the edits, and leave a comment explaining why, with a recommendation to ask their edits instead as a follow-up question.

Answer 10

I think the situation where the question is edited to ask something different is a clear case for why it's often considered best practice (but maybe not on stackoverflow?) to restate the question in your answer.

Say "doing x and y is appropriate", and not "that is appropriate", so that the worst that your answer can become is irrelevant, as opposed to wrong.

Answer 11

Personally, I usually go back to question's edit history, copy the essential part of the old question and add it to the beginning of my answer as a citation, like this.

Often, I also leave a comment ranging from a neutral I've included the original question in my answer to an angry rant, depending on the mood. At least once such rant reversed 3 downvotes I got because my answer didn't match the question anymore.

I don't think that fixing the initial answer to match a new question, or removing an irrelevant answer is a good idea. If the initial answer was good, it might be helpful to someone, so it should stay. Granted, including the original question in the answer is sub-optimal and looks somewhat confusing, but leaving the answer irrelevant is just as confusing (if not more), and I don't see a better way to handle the situation.

Answer 12

With the discomfort in my heart... I am Mort (the line-of-business developer). And here is why I ask Mort looking questions:

• Massive language barrier - not-in-the-same-language-group language barrier. So as an independent not formally educated coder, my knowledge base is mixed from Internet English resources, the resources that are from my native language, resources that are from extremely third-party languages (like it is ironic that a Slavic guy learns to code in English while looking at Indian resources) to resources generated by me through time and experience. This language barrier can go so far that no one can understand me - unless I show my code. Such as I use too foreign terminology (and standards) for native speakers, to uneducated to English speakers and totally at odds with Indian or Chinese speakers.
• Gathered experience via public experience - To explain this further, I am going to list my own example: I've entered serialisation once I discovered bitwise operators (3+ years ago), but I've found out a name for it in past several months. So most of the time, my Mort-like questions are well asking for implementation that isn't widely used or asking for an answer that exists in the right formulated question. This also shows itself in jobs I took on by coincidence (fixing code errors/functionality for others) without actually needing to understand what they need. Code is my native language in this case, and once I look at it I can understand it... but to explain it - that is too far!
• Massive use of too broad flags - I've long passed the level where I need to ask what a specific error code means and am able to understand the compiler when it speaks to me. But projects I am working on have naturally evolved from one-file projects to multiple projects and now to the point where I am writing my own libraries. And in this point the issues I might ask questions for aren't too specific as they can be (especially with right naming), so often my questions are flagged as too broad - with reason. So the only way I can approach the question, without being flagged as too broad, is to shorten it to minimal reproductive code that should encompass multiple libraries and files. And then people answer the simplified version of the question that requires expanding.

To answer your question, I would do all three points you mentioned (to answerers' desire) with an added caveat. Ask an follow up question before doing anything, if the question still isn't clear: try and deduce what is the main focus of the question and ask if that is what they are searching for.

I have been directed to explore further "serialisation" by a user on the SO platform. And from then I can distinguish what is and what isn't serialisation, and all that from "did you mean serialisation?" with a Wikipedia link. Knowing the true name of something gives you power over it.

So before doing anything else:

• When in doubt, ask a question.

If a Mort like me isn't knowledgeable enough to try to explain what it needs in a way for you to understand it - abandon the question. The Mort will come back once it has enough knowledge to ask or edit the question... as I did a bunch of times. Abandoning the question is ok here. It might be cold, but you can't help a person who doesn't want (or can't) help themselves.

If Mort has some specific implementation of some functionality/practice that goes a tad bit (or all) over the naming functionality, edit your question for the 3rd, 4th, 5th time, etc. and then abandon the question. *The main reason for this is that to answer the question you need to talk about the same topic. If the main question is fruits, you answer in apples and the user asking the question talks about oranges, both of you will get nowhere. So after knowing that you are talking about the same topic, re edit your question, research if need be.* If still no successful answer is made - just cut your losses. There is no shame not knowing something, or not understanding someone.

If a Mort is frustrated by their lack of language skills, suggest related topics and if a Mort is an asshole, downvote and abandon the question. There is no clear functionality for downvotes; people use them however they want. So I see no issue punishing assholes and awarding politeness. No one here has a job to educate the uneducated. So sure, the uneducated are left alone, but the uneducated can gather knowledge through other means - and no path here gets that knowledge through aggression.

---

I've had moderators and users tell me that they don't understand the question or that am using the wrong terminology ... but when I asked them to correct me in comments... very few did. I get frustration and best intentions thrown in the bin. Just use Wikipedia or some other resource and when you have an issue with the question, link it or say it. A Mort like me can easily copy paste and do research for themselves.

EDIT:

Also, I think it is important to mention that as a Mort (the line-of-business developer) I know what I can do, but I really don't know what I know and in which quantity I know it. How much I feel ready or "knowledgable" my knowledge is filled with small holes that are in outside of my consciousness. I know this might be (or is) frustrating to establish communication between a formally educated developer and me, without a wall of text describing the step-by-step idea I have. And understandably I know that no one is entitled to demand knowledge or to give it. In the end it is up to anyone who is reading my questions or answers to judge for themselves if they have necessary patience. Either way, no harm no foul... unless one of us is an asshole.

Answer 13

I am one of the askers you describe. I know this is late, but someone just now told me about this post. I have read your entire question and understand it completely.

I disagree with a word you used: "worthless." Your efforts might cause people to stop contributing to this site, and by extension, to a better world.

We don't answer questions for payment or peacockery. We do it to build a more active community. And we still get credit for our actions. Just look at missionaries and open source community contributors.

But the problem still bores answerers like you, and of course askers like me.

Luckily, we are all developers, and as such we are all familiar with using computers to do repetitive, boring, reusable tasks.

[Everything from here down, except for the last bit about your English, is what you should copy to one or more new feature-request post(s). -Ed.]

I suggest that Stack Overflow add the following features to solve the problem:

1. Implement a more intelligent pre-/post-processor

   • Provide an AI that could tell prospective authors about similar/possible duplicate questions before they submit new posts. It could act like Wolfram Alpha.

   • Provide an AI that could send notifications about new questions to users who might be able to answer them. Allow this AI to mark answers "accepted" when it thinks they're good enough.

2. Add more automatic rules

   • Don't just notify askers of activity. Automatically bump questions with low activity, like my poor unanswered questions.

   • Send notifications through IM clients (IRC/XMPP/&c).

3. Add more actions for users

   • Add a fork action to existing answered questions. This would let users ask related/follow-up questions based on the solved problems, and keep related information together. Visitors would be able to learn from the initial question — and its answers — and then move on to related topics; this would help them more than a single question would.

   • Add a quote and link or split action for sections. Most of the boring questions you mention could be split into many smaller questions. Some or all of the sections might have been solved before, and we could link the sections to the existing solutions. This would help askers and answerers focus their efforts on unsolved sections.

   • Add a merge or join or union action for questions. Askers often post questions that are just parts of their real problems. For example, one person configuring nginx might ask about configuring the nginx proxy module while another might ask about configuring the nginx cache module. If we could put those and other related questions in some kind of "nginx configuration group," it could be a treasure house for all people trying to set nginx up.

Sorry for the poor English in the initial revision, I am Chinese. I welcome comments about solecisms and weird usage that can help me improve my English.