81

Resource requests are off-topic according to the help center

Questions asking us to recommend or find a book, tool, software library, tutorial or other off-site resource are off-topic for Stack Overflow as they tend to attract opinionated answers and spam.

However, some resource requests are tolerated on the site. Shog9 said the following in chat:

I see a lot of folks going after these "what's the canonical download for [tool|library]" posts as resource-requests. I think that's kinda short-sighted; as long as they don't degrade into "ok, screw canonical, just give me a link to a shady host where I can download a malware-clogged zip file" they're unquestionably useful.

Remember, these were originally problematic because folks would clog the answers with competing recommendations without any clear way to suss out which ones were legit. That's absolutely not the case for "where did [Apple] publish [some lib]?"

Which made me wonder: what kind of resource requests aren't off-topic?

Do we allow all resource requests that have a single, well-defined answer, such as requests for official download links, specs and references, since these don't attract spam or competing recommendations?

Or, are these download links the only exception to the rule?

Or, should we ignore that chat message altogether, and are all off-site resource requests off-topic?

Also asking because I want this guidance to be better findable and linkable.

11
  • 11
    I agree that there should be some official guideline or consensus. If it is decided that anything without a competing recommendation is OK, the close message needs to be changed to just say recommend, not "recommend or find". Right now the wording disallows users from asking where to find a specific resource.
    – BSMP
    Commented May 26, 2019 at 20:32
  • 1
    But it also needs to be clear whether this is just an exception for official documentation or any request looking for a specific thing. Can I ask where to find a specific programming book? Where to find the official 64-bit version is of an IDE?
    – BSMP
    Commented May 26, 2019 at 20:34
  • 6
    Shog9 stated it pretty clearly without chiselling it out on a stone tablet, resource requests are often off topic because they generate crappy answers. A resource request that doesn't generate crappy answers is on topic. Everybody can help by not posting crappy answers. Commented May 26, 2019 at 22:26
  • 26
    @HansPassant But then what is the guidance for users who are asking? "Your question is on topic unless/until a spammer makes it off topic by answering it?"
    – BSMP
    Commented May 26, 2019 at 22:48
  • 8
    All resources requests are off topic, you know that. No exception.
    – TGrif
    Commented May 26, 2019 at 23:54
  • 7
    @TGrif That's not quite true. Erik's question even points that out.
    – TylerH
    Commented May 27, 2019 at 0:49
  • 7
    Sorry I don't take chat as a reference @TylerH
    – TGrif
    Commented May 27, 2019 at 8:08
  • 7
    The key concept is "attract opinionated answers and spam" (since this is why we close them), any question formulated so that is the case is off-topic, while questions that does not ask for this can be tolerated according to Shog9 since they may be useful for others programmers. Commented May 27, 2019 at 8:09
  • 12
    @TGrif I don't know that, I tend to take CMs at their word, whenever and wherever they say something, because they're the highest authority we got here. And I'd like us all to follow the same rules, thus I'd like this discussed in a place where others can see it and take it as a reference too. Also, having googled real-world problems and encountered a really helpful but closed SO Q&A more than once, I'm certainly open to accepting these kind of questions
    – Erik A
    Commented May 27, 2019 at 9:16
  • 1
    @TGrif The fact it was said in a chatroom is irrelevant; it's still guidance from a Stack Overflow Community Manager.
    – TylerH
    Commented May 28, 2019 at 13:43
  • Absolutely @TylerH. We all agree it's a very subjective subject. Happily, the community does not apply blindly the moderation rules already. Will try to use chat more...
    – TGrif
    Commented May 28, 2019 at 14:09

4 Answers 4

29

So, I wasn't really trying to establish a new standard with that comment so much as just... encourage common sense. As I said at the time, "resource recommendation" questions were originally targeted not because there's anything inherently wrong with the questions but rather because they have this nasty tendency to encourage terrible answers... So going after a question that hasn't or cannot attract lousy answers is just a waste of everyone's time!

Heck, the close reason itself - which you quote - tries to reinforce this rationale:

as they tend to attract opinionated answers and spam

Is the question more likely to attract opinions and spam than useful information? Shut it down! Has it been sitting around for months or years with only useful, factual answers? Keep it!

There's another option here too, also suggested by the close reason itself:

Instead, describe the problem and what has been done so far to solve it.

If the question can be rephrased to target a specific problem - and thus invalidate spammy or hand-wavy answers that don't attempt to solve the problem - doing this sort of an edit is probably the best of all options: only takes one person, doesn't shut down anyone participating in good faith, but does leave us with ample justification for removing opinionated rants and spam.

All of the above applies to all questions!

Now, I originally brought this up in the context of folks looking to download a library for a reason: there are a ton of questions on SO from folks just looking for official sources - as package repositories / managers have become more popular, more and more publishers have started skipping normal download links or source control in favor of publishing to these package repositories. That's great if your build system integrates with that flavor of package manager... Not so great if you need a stand-alone build on a disconnected system or something.

In other words, these questions already reflect specific problems, aren't asking for opinions or even recommendations at all. The only problem is that they can look an awful lot like resource-requests, particularly when the asker - likely desperate - throws in a line that suggests non-canonical sources might be accepted if no canonical source exists. In those cases, the solution is to get rid of that line - as long as random links to mega.nz or whatever aren't acceptable as an answer, we're still good.

Keep in mind: if a publisher of a common library isn't making their library easy to download, we're just going to get more questions as long as we keep shutting down new ones... And that just ends up providing a bigger surface area for folks to post misinformation (or shady download links) in. Having one well-linked question that contains safe information and can be used as the target for duplicate-closure of new questions is better for everyone involved, and a heck of a lot easier to manage - so once again, this goes back to common sense: don't make more work for yourselves than is necessary!

3
  • 1
    as long as random links to mega.nz or whatever aren't acceptable as an answer, but what if I want to find out where MegaUpload is hosted?
    – Adriaan
    Commented Jun 12, 2019 at 20:47
  • 19
    Then you're a cop!
    – Shog9
    Commented Jun 12, 2019 at 21:01
  • 6
    Came here from a current case. What I do not like so much about "off-site resource" question is that it encourages link-only answers as well as may get outdated soon, even if it is not opinionated or spam. People don't need answers to "where did [Apple] publish [some lib]?", they probably need answers to "where does [Apple] publish [some lib] now?" and that may be more difficult to keep relevant. On the other hand it could work though. Commented Jul 13, 2020 at 9:05
20

The general rule against off-site resources is targeted at this type of question

Does anyone know of a library to do this thing?

We don't allow questions like that because they invite open ended opinions. You could keep on posting new answers forever. They can also be ways to end-run the link-only rule (and can sometimes be a vehicle for spam). That's why Erno's suggestion fell through

What library is optimized for speed when calculating the average over 10000 decimals in .NET?

That's still an opinion question. Maybe Library A does that. Or Maybe Library B. Wait, that's wrong, it's Library C! It does it so much faster than B, you know... (comment: D is still faster)

The Shog9 exception seems to cover one specific type of resource request: an official source for a resource you already know exists.

That's absolutely not the case for "where did [Apple] publish [some lib]?"

We're not asking for opinions or suggestions here. We already know [company] publishes [library], but they don't have an easy-to-find link for that. The original context of the Shog9 exception was the burnination of [ibm], where IBM is a massive company and has lots of obfuscated places to download official IBM docs or libraries. Yes, it's not a clean exception (I agree with Ross on that), but there's only so many places to officially download something.

18

Currently the policy is that all off-site resource recommendation questions are off-topic. As far as I know guidance made in chat by a community manager doesn't change community policy. While Stack Exchange, Inc. can set policies that we must abide by if want to use their services, I don't think anyone is arguing that the chat message by Shog9 was meant to be the announcement of such a policy. Instead I think Shog9's chat message was addressing how old posts should be handled, ones from before the change in policy to disallow resource requests. Generally the view has been that "good" questions that are no longer on-topic shouldn't be deleted.

On the other hand if it was meant to suggest a change in policy on how new questions are handled, I would disagree. It would require people to make a subjective decision on whether a question was a request for a canonical resource, and thus has only a single well-defined answer. I think it goes against the spirit of Stack Overflow to make moderation decisions based on what the correct answer to question is presumed to be. Off-site recommendation questions with only one valid answer would also suffer from the problem that these answers are pretty much fundamentally are going to be low-quality link-only answers.

The problem with assuming the correct answer shows up all the time with duplicate close votes. Someone sees a new question, decides that it has the same correct answer as an existing question and votes to close without considering all the potential answers. Instead of the community as a whole deciding what the correct answer is, it's just five close voters deciding to disallow the posting of other potential answers. By changing the policy allow requests for "canonical" resources we'd be asking people decide that there is only one possible correct answer that everyone would agree on. Deciding what the correct answer should be is something that should only be done through up/down voting.

The biggest problem with this suggestion though is that even if there's only a single correct answer that everyone agrees on, it'll just be a link only answer and suffer from the same problems they have. The link will go stale eventually making the answer useless until it gets fixed, if it even can be. Such answers also don't really add much value to the site. If everyone actually agrees that there's a single canonical source for a resource, then there's good chance the resource itself will appear in search results ahead of any question asking for its location.

Another problem is that the answer will likely change over time, possibly so there's no longer a single well-defined answer. A new version of the resource could get a new link, while the old link for the old version remains valid. Maybe the resource is abandoned by original maintainers and there's no longer an official source for it. Maybe there will be new links for different regions or different languages.

Finally allowing canonical resource questions as exception also creates weird situations for moderation. If multiple answers disagreeing on link to the canonical resource get posted, what should happen? If the answers aren't all bad, should that be taken as a sign that there it isn't a canonical resource question and the question closed? That would create the perverse situation where a question should be closed because its receiving good answers. If the question generates a lot of spam answers, what the existing close reason is meant to prevent, what should happen? What if a question generates answers giving unofficial links that work behind the Great Firewall or bypass geo-blocking or paywalls?

It's easy to say to just leave it to up/down voting handle situations like these, but that apparently didn't work with off-site resource questions generally back when they were permitted. These canonical resource questions can still easily generate spam, and even opinionated answers from people who think some alternative link is better.

I don't think having a single well-defined answer works as a justification for allowing an exception to the off-site resource recommendation question rule. Arguably such a justification would justify locking canonical resource questions after someone posts the presumed valid answer, since any further answers could only be redundant, wrong or worse. This is why I think such a justification goes against the spirit of Stack Overflow. It's bad enough that most of the "good" off-site resource questions from before the policy changed have had to locked or closed to prevent them from attracting more spam and junk answers.


Cody Gray's comment to Emo de Weerd's answer shows how answers giving links to resources, canonical or not, can and do work well on Stack Overflow, even if questions asking for them do not. Broadening the question so it doesn't assume a single answer both allows for competing solutions, including ones the original poster might not of thought of, and requires that good answers giving links to resources provide justification, making them not simple link-only answers. Questions like these fit Stack Overflow much better, and often require just a small change in phrasing from the off-topic version.

1
  • 3
    I agree with this answer. Link-only answers are bad. Questions that invite link-only answers are bad by association. If a repo is hard to find, it is not SO's responsibility to play "traffic router". Opinion-based answers can have value to researchers, but SO is clear about not wanting them, so they should be closed. Let's make moderating resource requests equally clear cut. Why try to confuse the user base with minutiae? If you are going to make allowances, then you need to change the text in the close reasons UI so that people don't do the wrong thing. Commented Jan 27, 2021 at 11:23
-67

I would argue that a request for a resource is perfectly fine as long as the request is specific.

Not: What is the best library to calculate the average?

But: What library is optimized for speed when calculating the average over 10000 decimals in .NET?

The answers should follow the regular rules for answers (specific, no spam, full disclosure) and, ideally, a link to a repository of the sources or binaries on a well-known host (github, nuget.org, ...)

6
  • 57
    Best: How do I quickly calculate the average of over 10000 decimals in .NET? Library recommendations would, of course, be acceptable answers, as would answers recommending algorithms. (Ideally, "quickly" would even be defined, but I dare not ask for so much at once.) Commented May 27, 2019 at 6:03
  • 4
    @CodyGray such a question would probably be closed as "too broad"... Commented May 28, 2019 at 18:26
  • 5
    If that sentence is all the question consisted of, then sure, @Cassio. But if there was some actual substance to the question, then it should not be closed. It is a practical programming problem fully within the scope of SO. There are good ways and bad ways to ask essentially the same question. Commented May 28, 2019 at 20:05
  • 1
    @CodyGray I don't see what kind of substance could be added that the title doesn't already convey, at least for this specific example. "quickly" could be defined, as you assert yourself, but the question would still be considered too broad. Unless the asker added some code (show your work and all of that), but then library recommendations wouldn't really fit as answers anymore. My point is: If the best answer for a question is a library recommendation, then it probably will be closed as too broad, if we consider the current community standards. Commented May 28, 2019 at 20:34
  • 1
    BTW, I'm just drawing my conclusions on what I see the current community standards are. This doesn't mean that I agree with them. Namely: More specific questions (even too specific) are generally more accepted than generic ones (unless they are a "canonical" posted by a high-rep user), even when (IMHO) the latter is generally more helpful than the former. Commented May 28, 2019 at 20:40
  • 6
    Your interpretation of the current community standards doesn’t match with mine, @Cassio. This is not just a “debug my code” site. All practical programming problems of sufficiently narrow scope are on-topic here, and the question under discussion here strikes me as a perfect example. Yes, more specific questions are preferred. The specifics narrow the scope enough to make it answerable in our format. I do not believe rep is a factor at all. It isn’t for me. Closing a question like this as “too broad” is a misuse of the close feature, as far as I’m concerned. Feel free to flag if you see it. Commented May 28, 2019 at 21:04

You must log in to answer this question.

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