Rule proposal: Duplicate closure to roll-up questions is no longer allowed

"Figure out what, in this giant pile of information, answers your question."
That's not useful.

Exactly. Yes, we should all Read The Fine Manual. But if the OP didn't already RTFM, throwing yet another manual at them isn't going to improve matters. And if they did read the manual but are a bit overwhelmed & confused, a giant info-dump is likely to compound the confusion & increase the anxiety.

If the OP asked a clear and specific (i.e., well-scoped) question they need and deserve some clear and specific guidance.

I have no objection to those giant roll-up Q&As, per se, but I've never thought they make good dupe targets. They are useful repositories of more specific questions, and those specific questions should be used by the Subject Matter Experts as the actual dupe targets. The roll-up question can be linked in an "Also See" comment, or added as an additional dupe target by a gold badger or mod. But to use the roll-up as the primary dupe target is just cruel.

If the OP submitted a low quality no-effort question which is not clear and specific, we can close it for other reasons. We don't really want it to survive as a dupe because it's not a useful signpost.

I do think some dup targets that I've seen before (and I'll admit have even used once or twice in a way that I currently question) are kind of collapsing / have collapsed upon themselves. Some that make me think "wait- I thought SO was to make it so people can come with a specific (well scoped)¹ question and not have to sift through a bunch of text to get the answer to their specific question".

To people who think closing as a duplicate of these "catch-all" / "reference" questions is fine, here's a challenge: See also the MSE FAQ post on duplicates, which links to this blog post by Joel Spolsky written all the way back in 2011:

If you're going to close a user's question as a duplicate, it has to be a real duplicate. For example, if a user asks, "What does the IP address 128.0.1.1/24 mean?" it's OK to close that as a duplicate of a more general question like "What do IP addresses of the form a.b.c.d/e mean?" But it's not OK to close it as a duplicate of a twenty-seven page guide to netmasks. That's the moral equivalent of saying "RTFM." Stack Overflow is not meant to be a library of reference manuals. It's supposed to contain the same information as a library of reference manuals, in the form of millions of questions and answers. Combined with Google, that gives us the magical power of a library of reference manuals you never have to read! It's like, you got to the library, and there's a wizard there at the door, and you ask your question, and, instead of being told to read a book, you just got (are you sitting down?) the actual answer!

I've been thinking about this for a while but had been too lazy and scared to bring it up. Actually, I had sort of posted about this kind of topic a year ago, but deleted that post.

The one particular one that comes to mind right now (with no disrespect to the people who have contributed to it) is What is an undefined reference/unresolved external symbol error and how do I fix it?. I think "what is an undefined/unresolved reference" by itself is a fine and extremely useful SO question to have, but "how do I fix it" (or "how do I find out why exactly it's happening in my type of setup") is too much for a single Q&A to handle. There are just too many possible reasons- each that have causes that are common enough that having dedicated Q&A would be more useful. I do think having some sort of aggregation of links to Q&A for specific, common causes is a useful thing, but this specific way of going about it with the links being to individual answer posts under the same question, each for a different possible cause indicates that something is wrong with the way SO is being used here. I'd imagine a better scenario would be to have

• one Q&A for what an undefined reference error is and why it happens in general,
• one community wiki Q&A for what common specific / in-context causes are, with an answer linking to
  • individual Q&A for those specific / in-context causes, and how to fix those individual problems.

An example of the community wiki collection of links to specific Q&A would be PHP parse/syntax errors; and how to solve them. Possibly ironically and sadly, that question is the top-5th duplicate target (counting by duplicates that are non-deleted).

What would one do instead of closing as a duplicate of that "table of contents / list of common reasons" question if one sees a question that is caused by multiple specific problems and one doesn't have a gold badge to close as a duplicate of multiple specific Q&A? Close as needing focus? Answer? Downvote and move on?

I think it might also be useful to spell things out clearly for some of the top duplicate targets. I wrote a SEDE query a year ago that could be useful: https://data.stackexchange.com/stackoverflow/query/1673806/top-dup-targets. Here are some particular ones I'm interested in:

• PHP parse/syntax errors; and how to solve them
• What is an undefined reference/unresolved external symbol error and how do I fix it?
• Reference - What does this error mean in PHP?
• Reference - What does this regex mean? (already explicitly used as poster child in proposal post)
• Why does jQuery or a DOM method such as getElementById not find the element?
• What does a "Cannot find symbol" or "Cannot resolve symbol" error mean?
• What are the basic rules and idioms for operator overloading?
• Reference Guide: What does this symbol mean in PHP? (PHP Syntax) (already explicitly used as poster child in proposal post)
• Reference: mod_rewrite, URL rewriting and "pretty links" explained
• What does this symbol mean in JavaScript?
• Learning Regular Expressions

^{1: I always find the word "specific" tricky. It's ambiguous between at least two meanings: "well-scoped" vs. "localized". My general understanding has been "well-scoped" (Ex. I spent some time debugging my problem in my project to create a MRE) => good!, "very localized" (here's a code dump of my project- where's the bug?) => not so good. In this post, when I say "specific", I am referring to "well-scoped".}

Good to see that y'all are looking into this now. And it's good to see that it's being applied in a more broad standard, rather than just in a singular instance.

This is the first step though; people will still want some kind of canonical to catch all of the whack-a-mole questions they get from a subject that has a lot of surface area, like regexes. The next step would be to get some buy-in on what constitutes a sensible canonical for questions like this.

Not all (uses of) catch-all duplicate targets are bad. They are sometimes abused, but a blanket ban is too much. If we want to regulate their use, we need a more involved rule than just "catch-all duplicates bad". But I'm not yet sure what the exact rule should be.

Some thoughts:

• Ok for trivial questions. If a user clearly didn't bother reading the manual, using a manual-like duplicate target is fine.

  For example, if a user asks "what does ^ do in regex", I don't see an issue with Reference - What does this regex mean? being used as a dupe target.

  This is arguably better than finding a more specific dupe target, because we want them to be able to find this information on their own. We don't want them to come back and ask about *, {}, [], and every other symbol they see next (for us to find duplicates for them); we want them to go read the manual, be it our manual-like dupe target or something else.

• Bad for complex questions.

  I sometimes see c++ What is an undefined reference/unresolved external symbol error and how do I fix it? being abused like this.

  This error can be caused by a basic lack of understanding of the compilation process, then this duplicate target is fine.

  But it can also be caused by hard-to-diagnose issues (like messing up the compiler flag order in a subtle way, or downloading libraries for a different flavor of the compiler), then the duplicate target is completely useless. The asker has near zero chances of finding the correct solution among the 48 answers to that question.

• Is the dupe target easy to navigate?

  If the user asks about basic regex syntax, they can easily find the symbol they're interested in in Reference - What does this regex mean?.

  If the user asks about an undefined reference error, it can be difficult for them to navigate What is an undefined reference/unresolved external symbol error and how do I fix it?, because this error has several different causes, all of which produce very similar error messages, and the huge dupe target lacks any organization.

  If they're dealing with the most common case, good, they'll find the explanation at the top of the first answer. If not, they're likely screwed.

• "List of books" as a dupe target is passive-aggressive if OP didn't ask about books.

  This probably goes without saying, but using The Definitive C++ Book Guide and List as the duplicate target for basic C++ questions (as opposed to requests for books) is outright passive-aggressive, and shouldn't be allowed.

Will closure of any question as duplicate of mentioned roll-ups be prohibited?

And if yes, based on mentioned Reference - What does this regex mean?, what should happen to questions that look something like following?

I was given this legacy project, and it has this regex that I can't understand

<some regex>

Also, roll-up seems to be somewhat questionable term. In the interest of transparency, if this rule should be accepted, there must be clearly defined list of roll-ups not-to-be-used for hammering.

No, the problem is that different gold-badge users in a specific tag are having question closure/reopening fights. This is trivially fixed by having a nice chat with the one who is violating rules around question quality by reopening rubbish questions, and if they don't desist, giving them a vacation from the site. That's a moderator job, not a curator job.

Duplicate closure like this is basically throwing a volume of information at users and telling them "Figure out what, in this giant pile of information, answers your question." That's not useful.

Do you know what else is not useful? Users who post low-quality questions that fail to do absolutely trivial research. Those questions have zero value to Stack Overflow as a concept of a high-quality repository of questions and answers.

this is why we can't have nice things.

Asinine policies like this allow rubbish questions to stay open for no good reason. That is why we can't have nice things. That is why curators are no longer curating. That is why Stack Overflow is dying.

Stop trying to make it more difficult to get rid of the rubbish. Stop making this site into a one-stop shop for help vampires.

I hesitantly support the spirit of this proposal, but I think it needs nuance. I hope I can offer yet another perspective.

I use CronJob not running as a canonical for "how do I troubleshoot cron jobs". (I don't have a gold badge in the cron tag, but I do have one in several related tags, such as bash, linux, and python). This may not be a common FAQ as such, but a large number of "halp, my cron job is not working" questions do not have any details, and so pointing them to a troubleshooting guideline seems useful, especially when the new question is not so much.

I generally try to find the time to also add a comment to tell them that their question can be reopened if they can update their question to include enough details to make it useful and well-defined.

The canonical I found is dubious for several reasons, though.

• It's about a fairly specific problem, and tangentially about Python.
• It's slightly hard to find in search because the title uses the odd and ugly intercap compound "CronJob" as a single word (I mean I can find it, but it's perhaps not reasonable to expect the OP to have found it before asking).
• It has obviously grown organically over time, so the exposition is not particularly good as a guideline.
• Like many popular questions, it has multiple low-quality answers which worked for a single user or a small group of users, but do not explain why it worked or under what circumstances it would.

I'm providing this background to suggest that the specifics of the canonical can also play a role in deciding when it is acceptable to hammer. Especially if the warts of this canonical could be removed, I think it makes sense to continue to hammer questions which are easily solved by these instructions, complex as they may be. Having one canonical for each case seems excessive (especially when many of the error cases involve system administration rather than programming, and would thus be off-topic for Stack Overflow in isolation).

If I understand correctly the question can be boiled down to:

(...) using a single roll-up question as the duplicate target.

A prime example would be this regex roll-up

I think it's reasonable to ask a dupehammer close voter to choose a specific question from the list instead of just closing against the roll-up thread itself. So yes I think this rule should be enforced.

I agree with this initiative in principle.

When I hammer a page that has been answered by one or more users, I often ask answerers to transfer their insights to the dupe target because The fundamental goal of closing duplicate questions is to help people find the right answer by getting all of those answers in one place.

When a narrow question is closed by a broad dupe target, often it is non-trivial/unsuitable to transfer the narrow insights. Adding insights to a rollup can be less attractive because it might be an unrewarding wiki-edit, or be an answer which gets lost in the flood of other answers.

• (asker win) The asker and future readers receive narrow resolving advice. The argument of returning the same low-effort advice for a low-effort question ignores the fact that future readers will be leveraging this page. We must always give our best advice no matter how lame the question.

• (answerer win) Contributors who wish to share their insights will be able to see if they have unique insights to add to Stack Overflow and they will see where to post them. Contributors are not silenced.

Most of the (PHP) pages that I close (which were posted after 2015) are usually closable by at least 5 different pages. It takes more effort to hunt for multiple duplicates, but I believe this to be the most generous form of closure that there is and everybody wins.

Multi-closing:

• Reduces the likelihood of close/reopen wars because reopen voters have a harder time disproving all of the dupe targets' validity.

• Helps to link related content, and simplifies the job of nominating new canonicals.

• Indicates to the asker that they should have searched for other keywords or, more likely, just searched more.

• Is definitive proof that the page is a good candidate for deletion because it is an nth dupe on a given topic.

I agree with this - it's better to find a more specific duplicate where possible (vs. what amounts to a RTFM closure, which was no longer a valid reason after the "Lacks Minimal Understanding" close reason went away). However, the two cases where I think that this should be allowed are:

• Where the OP clearly put forth zero prior debugging or problem-solving effort, or where it's trivially solvable on their own if they use the right technique. For example, if the problem could be trivially solvable by stepping through the code with a debugger, I might point them to this.
• Where the same or very similar questions have been asked and answered many times before, such as debugging a NullPointerException. In those cases, it's hard to imagine what we could possibly add to the literally thousands of times it's been asked before.