26

This came out of a discussion in the C++ lounge today with @casperOne regarding the long-term viability of keeping the meta-tag around. Please help me to flesh it out.

The problems:

  • Meta-tags are bad
  • The C++ community, possibly uniquely, utilizes this tag for a multitude of reasons:
  • To easily identify canonical questions that other questions may be closed as duplicates of.
  • Identify collections of "authoritatve answers" as there is a dearth of good, peer-reviewed C++ information.
  • There are too many questions and too many people and we should just get drunk
  • While possible to move the referenced questions into a tag wiki (it would have to be another one, since this tag theoretically would not exist if removed from all the questions) the tag wiki system is not quite up to the task of acting as a reference repository for [allegedly] oft-asked questions, that we instead post as fake questions and tag them so that they do not get closed as Not Constructive.

In order to more constructively create a repository for this sort of information, it was proposed that the tag wiki system be enhanced so that we can use that instead of polluting question-space with long-winded, made up "references". It would need to be searchable, and we would need to be able to close a question by referring the OP to the wiki post somehow, in much the same way that we can presently close as dupe of a post (though, frankly, I don't see this happen that often).

In the meantime, @jalf proposed that perhaps all tags could be treated specially by the SO interface. Any user with a gold badge would be able to cast a vote to FAQ-ify a question.

I was slightly concerned that high rep certainly doesn't necessarily entail expertise (ultimately, what does?), but concluded that it's difficult to get a gold badge simply through picking off low-hanging fruit and rep-whoring. Then, we have the voting system to protect against the odd high-rep, low-expertise person-who-got-really-lucky-with-Reddit from upsetting the delicate balance.

14
  • 11
    C++, possibly uniquely... Not really...I'm sure the PHP and Python users over at SO would love to have a possibility to kick certain recurring questions into the trash can fast and efficient. Feb 20, 2013 at 16:37
  • 1
    Perhaps this has wider implications for the close-as-duplicate system, then. Feb 20, 2013 at 16:41
  • qv recurring canonical discussions
    – AakashM
    Feb 20, 2013 at 16:48
  • 2
    Doesn't the close as duplicate system already show posts that are commonly chosen as duplicates? It's specifically to make it easier to close recurring questions as all duplicates of the same "faq" question Feb 20, 2013 at 16:50
  • Actually, your post is incorrect, it was not about keeping the c++-faq tag around, but the impact it would have if it was removed. The larger issue is other -faq tags showing up and -faq tags generally being meta tags.
    – casperOne
    Feb 20, 2013 at 16:50
  • And for the record, I don't know if voting up on this means I'm ok with -faq tags, or the feature request, or what.
    – casperOne
    Feb 20, 2013 at 16:52
  • 3
    @Sulfurized YES OH YES YES
    – Pekka
    Feb 20, 2013 at 16:53
  • 1
    Wait, that didn't come out right. Anyway, this is a great idea
    – Pekka
    Feb 20, 2013 at 16:54
  • 3
    I'm not sure that overloading the tag system is going to be the best way to handle this, but I'm all for some more system-defined official way to define genuinely frequently asked questions.
    – Charles
    Feb 20, 2013 at 18:06
  • @casperOne: Sorry - my first paragraph wasn't intended to summarise the original intent of your visit into the Lounge. It summarises the discussion as a whole from the users point of view. It doesn't really matter how it started. Thus I think the post can hardly be labelled as "incorrect" on that basis alone. Feb 20, 2013 at 19:50
  • 6
    Might as well put this feature request in the pile of all the others recommended in this regard, and all the feature requests to do with making chat better- the "Will never be accepted, denied, or even commented on" bin.
    – DeadMG
    Feb 21, 2013 at 20:24
  • 1
    @DeadMG: Okay, except it has nothing to do with chat. Feb 21, 2013 at 20:36
  • 2
    I'm having difficulty figuring out what your actual Feature Request is.
    – user102937
    Feb 21, 2013 at 21:18
  • @RobertHarvey: Yeah, so am I. Feb 21, 2013 at 21:40

3 Answers 3

6

Any user with a gold x badge would be able to cast a vote to FAQ-ify a question.

Anyone can already cast a vote to FAQ-ify a question: vote to close another question as a duplicate of it. If something's really frequently-asked, then prove it.

Right now, the site-generated C++ FAQ counts all links as a "vote". This actually works pretty well for your purposes - note that 25 of the top 30 questions by links are questions. If we only look at duplicate links, then this drops to 19 of 30...

I tend to think giving more weight to duplicate links is the right way to go here: like it or not, these reflect the questions that are actually asked a lot. There are 32 questions in that've never had another question closed as a duplicate of them - kinda hard to argue these are really FAQs. Questions like this or this aren't frequently-asked; there's information there that can be useful to reference when answering questions that are frequently-asked, but that's not a good reason to clutter up a FAQ with them - if anything, you should have a separate list (maybe in a tag wiki...) of useful "background info" posts for use by authors.

I was slightly concerned that high rep certainly doesn't necessarily entail expertise (ultimately, what does?), but concluded that it's difficult to get a gold badge simply through picking off low-hanging fruit and rep-whoring.

It's difficult to get a gold badge by identifying questions that've already been answered too. Let's face it: there's already one system in place for "voting" on FAQs, and not enough people use it - adding a second one ain't gonna make that better.

9
  • Yeah, this makes sense. Feb 21, 2013 at 23:27
  • 3
    The reason those FAQ questions are so prominent in the generated list is that the C++ FAQ-tagging works pretty well. When we started this effort, the site-generated list was more or less useless. So if anything, the correlation proves that the tag was necessary, not that the site-generated list is helpful.
    – sbi
    Feb 22, 2013 at 9:02
  • Regarding the two questions you linked to as "hard to argue they are FAQs": Those are questions few newbies would ever ask, but which should be asked by them — if only they knew enough to ask them. If they hadn't been used as duplicated questions so far, then this might well be because this tag system we're using isn't perfect, still makes it hard to find duplicated questions, and could do well with more support from the site.
    – sbi
    Feb 22, 2013 at 9:05
  • You built a list of questions you thought would be useful, and folks used them, @sbi. That's great! But that's also an argument against building a system that requires folks to vote questions into the list - if you did that, you'd just have to create another list to feed the first one.
    – Shog9
    Feb 23, 2013 at 20:41
  • 1
    As for the "folks would ask this if they knew enough to ask this" argument... This "frequently answered questions" disease tends to infect just about every hand-curated FAQ. You might as well complain that folks should read books instead of asking questions on The Internet, for all the good it'll do. Adding even more support to the system for finding questions no one's asking isn't going to make finding duplicates easier.
    – Shog9
    Feb 23, 2013 at 20:45
  • 1
    @Shog9: You were arguing that those auto-generated list works pretty well because it mimics our manually maintained FAQ questions. I was countering that it only works so well because we put the effort into maintaining those questions. I have no idea what your "don't create another list" argument is supposed to counter in that.
    – sbi
    Feb 27, 2013 at 9:22
  • 1
    @Shog9: Before you look down onto those questions on our list that are not asked by newbies, I challenge you to catch all "what's the result of ++i+i++?"-like questions and explain sequence points thrice a day for a month, in constant good quality. Once you managed that, and convinced a bunch of SO users in the C++ tag who are good enough to also constantly give good answers to those questions, I am willing to discuss the deletion of those question. Of course, alternatively you could propose a third way to deal with that stuff. From what I heard, though, don't count on SO for any support.
    – sbi
    Feb 27, 2013 at 9:28
  • I'm starting to think you're not actually arguing in favor of this feature-request, @sbi
    – Shog9
    Feb 27, 2013 at 13:48
  • No, the site generated FAQ does not work well, because it seems to count "often linked-to" questions rather than just closed-as-duplicate ones. Meaning that things that aren't frequently used as duplicates but "general good advise not related to the core of the question" get very high rank. These may be good questions but are usually not canonical-duplicate material - you can't use them for close-as-duplicate. In addition you also get various off-topic, historical trash questions like subjective book or tool recommendation lists.
    – Lundin
    May 16, 2017 at 8:53
3

I personally don't have a problem with the tag, because:

  1. It is a useful categorization of the questions tagged with it, and
  2. It is actively maintained by the C++ community.

I'm not so sure that it requires any additional system support, however.

4
  • [obligatory comment about anonymous downvote not being helpful to anyone]
    – user102937
    Feb 21, 2013 at 21:37
  • 2
    It gets abused in some situations as a close as dupe magnet featuring Q/A's that are incomprehensible to end users.
    – user7116
    Feb 21, 2013 at 21:58
  • 2
    @sixlettervariables If you want to bring up a case about it being abused, please find one within the past year rather than 2010. That question you linked to hasn't been closed from the look of it and it was argued and discussed properly.
    – Rapptz
    Feb 21, 2013 at 22:12
  • @Rapptz: helps to read the timeline. If jalf wasn't so aloof, the question would never have been reopened.
    – user7116
    Feb 21, 2013 at 22:18
0

While I might be biased, as in I believe that removing the tag can cause more harm than good, I came about an idea when this was being discussed in the Lounge.

We currently have a(n) FAQ tab along with a tag, and the way something is currently designated as a(n) "FAQ" seems to be purely on views. Which in my opinion is an incorrect way of measuring the frequency of something being asked, rather it seems like a way to gauge popularity, if anything. A(n) FAQ should have two components that make it a well executed list, which include searchability and authoritativeness decision upon its usefulness.

So my execution would be to make the FAQ system revamped a bit, based on a previous merit allow users who use the tag to vote upon a question to make it into the FAQ tab. This merit could be anything, just like @jalf suggested which is to have gold owners vote to FAQ-ify a question (though I suggest this being lowered, maybe silver) or a reputation privilege (though I don't particularly recommend this approach see: review queue).

However this point alone doesn't improve upon the searchability that the had, and I'm unsure how many people vote to close on a question, the way I usually do it is through a query similar to this when searching:

[c++-faq] external linkage

Which would lead upon this question pretty easily.

So in order to impose similar searchability, I propose making the FAQ tab itself searchable through a query similar to this one:

[c++:faq] external linkage

In which the semicolon separates which tab is being accessed -- that is [c++:faq] would access the FAQ tab of the C++ tag. I believe this would improve the searchability and address both issues.

10
  • It's become clear to me that gold badge holders don't necessarily know anything at all about the thing in which they have a gold badge. I'm hoping that the voting part will still account for that. Feb 21, 2013 at 20:37
  • Also I don't understand the practical difference between [c++-faq] external linkage and [c++:faq] external linkage as supported search terms for the c++-faq tag, beyond the fact that the latter would require special handling. Feb 21, 2013 at 20:38
  • @LightnessRacesinOrbit [c++-faq] would access the tag itself, while [c++:faq] would access the FAQ tab of the C++ tag.
    – Rapptz
    Feb 21, 2013 at 20:39
  • Oh, after we fixed the FAQ tab? Yeah, okay. Feb 21, 2013 at 20:40
  • We currently have a(n) FAQ tab along with a tag, and the way something is currently designated as a(n) "FAQ" seems to be purely on views. The tooltip on the faq tab reads "questions with the most links" so its not views. This is pretty easy to confirm as the # of views is displayed below the score and # of answers, and its not in descending order. Feb 21, 2013 at 20:54
  • @SomeHelpfulCommenter Thanks, either way I don't find the current FAQ system helpful as the c++-faq tag.
    – Rapptz
    Feb 21, 2013 at 21:12
  • @Rapptz then why not say that rather than something that it easily disputed. Feb 21, 2013 at 21:26
  • @SomeHelpfulCommenter You have a pretty serious typo in your username. Feb 21, 2013 at 21:40
  • @LightnessRacesinOrbit blame sbi Feb 21, 2013 at 21:44
  • @SomeHelpfulCommenter: Ahaha :P Feb 21, 2013 at 22:04

You must log in to answer this question.

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