8

I'm trying to figure out how to locate a canonical answer which I know does or should exist in order to flag a duplicate when the question isn't canonical.

Example: Problem with projection in meters with ArcGIS 10.1 was asked and it was the third question in the last week or so that appeared to be related to misuse of Define vs Project. It immediately came to mind that there ought to be a canonical question and answer for this issue (or FAQ really), so I started looking into how to create one - beginning with Community Wiki research.

In the meantime someone else flagged it as a duplicate of Layers with same coordinate system should align/overlap but do not? which just so happens to have a great canonical answer for the issue already. But there is no way I would have found it if I hadn't already known of it, because the question is one of half a dozen possible ways to get at the same root issue.

I saw Is it appropriate to change the title of a question to reflect the definitive answer? and more particularly blah238's answer which is the opposite of my issue. Yes, (some of) the dupe questions should remain because there are so many ways to ask a question that the root of the problem is one issue - but how do you find that answer amongst all those questions? Age, quality, and votes are all discussed below as far as which should be canonical, but nothing about how to find it.

My first thought was you create a Community Wiki question that is based on the answer - ie "Incorrect use of Define vs Project" - and the canonical answer could be copied to this question. This is based on what I had read at:

I also saw other meta questions on different SEs that were basically 'link questions here that should be considered cannonical' so mods could do some arranging. But I haven't seen that here on GIS.SE or really people creating CW questions as described (maybe I've missed it). Should I go ahead and create a Community Wiki question? And also the matching CW answer, or flag it for a mod to copy the canonical answer to?

My second thought was how about a canonical tag? I love the canon-fodder and loose-canon idea from the 'Top 5' Question above.

  • Not only can essentially the same question be asked many different ways, its answer can be stated in many different ways. The "projections and units confusion" is labeled here as Define vs Project, but that's in Esri-speak. In Postgis-speak it's Assign vs Transform. My point being we do have a problem and i'm unsure of the solution. – Martin F May 29 '14 at 16:02
6

There was a canonical voting feature suggested and shot down a while back on the main SE meta.

I also don't think applying a tag to questions on the main site is appropriate either, as it is clearly a meta tag.

I like the idea Devdatta Tengshe mentioned which was a meta topic whose purpose is to list canonical questions, with the Server Fault canonical questions list as an example.

Here is a hastily-categorized list of highly-voted questions (certainly not the only criteria but an easy one to look for) as candidates for a future canonical questions list to start things off:

Maps / Cartography

Frequently-requested Data

Software Recommendations / Comparisons / Lists

Concepts and General Knowledge

Data Formats / Conversion / Interoperability

Data Management / Database Administration

ArcGIS-specific

QGIS-specific

R-specific

Education / Learning Resources / Tutorials

Stopping here because it's starting to get long, but please discuss what you think of these examples and whether we should have an "official" canonical questions list.

Also making this CW in case anyone wants to add to or modify the list.

  • Having now read about meta tags I can see that won't fly. I still say it's a viable option in this case, as some sort of mod-only 'sixth' tag. We of GIS.SE more than anyone should know that meta data does have a use and isn't always subjective. ;) I see your title edit. Do you think it would be more appropriate as 'list of questions with canonical answers'? Per my comments on PolyGeo's post maybe we don't need canonical questions, just a way to find our canonical answers. Such a list would serve nicely for that. – Chris W May 24 '14 at 1:43
  • As for process, high voted questions is one approach. Maybe high voted answers as well, and those with lots of linked duplicates (per one of my citations). Again, my approach is answers - 'define/project' answers dozens of questions; when I look at a list I want to see that, not 'why are my coordinates off'. Maybe another level to your list - broad headings, then answer headings, and under that different questions solved by that answer to be whittled down to one with the answer. Would take some research and perhaps effort to find and flag dupes once a suitable canonical answer was found. – Chris W May 24 '14 at 1:55
  • 1
    I don't think we need to differentiate between canonical questions and canonical answers. It's the Q&A pair that's canonical, but it's easier to just refer to them as canonical questions [that may have one or more canonical answers]. It's certainly valid for one question to have multiple great answers. – blah238 May 24 '14 at 3:13
  • 1
    Regarding questions with lots of linked duplicates, the Frequent tab shows the questions with the most links (including duplicates). You could also play around with the Stack Exchange Data Explorer to query out some of this info, e.g.: data.stackexchange.com/gis/query/78743/… -- although surprisingly few of those look canonical to me. – blah238 May 24 '14 at 3:39
  • +1 Frequent tab does seem to provide good list of candidates for what could become self maintaining Canonical Q&As via further curation. I suspect I would use that sort rather than Meta Question list. – PolyGeo May 24 '14 at 21:54
  • 1
    That is a great list of very common questions @blah238 I am certain that quite a few duplicate questions, and some badly written questions, could be avoided by having a FAQ section with the format you have created. It would have to be up to a moderator (or group of moderators) to post them there of course. Just because something is frequent doesn't mean that it's foundational, just because something is up-voted a lot doesn't make it canonical; it's not easy sometimes to find the basic truths so it would be good to have a page devoted to foundational, basic, truth. – Michael Stimson May 25 '14 at 8:54
2

As a direct Answer to your Question title (i.e. Do we need a list of canonical questions?), I think we already have what amounts to a list of canonical questions available as the Frequent Questions list (which was brought to my attention by @blah238 in a Comment on another Answer).

I don't think creating yet another duplicate, even if it is intended to be canonical, is a good idea. Also, I do not think that a Canonical Q&A has to be Community Wiki, although that would certainly be appropriate for some.

However, I do think you have raised a very valuable discussion and have made an excellent choice of something for which there is a strong need to have a Canonical Q&A available and easy to find.

I think the steps to creating a Canonical Q&A when faced with multiple duplicates of a recurring Question should be along the lines of:

  1. Think carefully about what the "perfect" (canonical) Question and Answer should be - perhaps write out, but do not post, a draft.
  2. Look for the oldest Question that can be reasonably interpreted as having asked the Q part of that draft Q&A.
  3. Try to edit that Question to become the "perfect" (canonical) Question, and perhaps include text like "This Question has been edited with the intention of its Q&A becoming canonical" (with a link to what Canonical Q&As are). This text could later be removed.
  4. Start working on the Answers to that Question to float the perfect one to the top (if it is not already there) and any poor ones to the bottom by up and down voting as appropriate. If there is no Answer that approaches "perfect", then consider writing a new one. If you are concerned that this may be copying someone else's hard work and getting the glory yourself then there is the option to suggest to them via a Comment that they should add their Answer to the new Canonical Question.
  5. Once there is a Canonical Q&A that you are proud of, do what you can to get all later duplicates linking to it. I suspect flagging some of them for moderators to change where they point may be needed and/or act as an expedient.
  6. Your Canonical Q&A should now be well on its way to appearing on the self-maintaining Frequent Questions list.
  • Agree another question might not be a great approach, but editing the oldest/question with the best answer to be the best question doesn't seem right either as it loses the original phrasing (rollback option noted). And if there is a list as blah suggests, maybe having an actual canonical question doesn't matter. I'm approaching the problem as trying to find the answer hidden behind one of many possible questions. I need something to identify with minimal effort the question I want to point my flag at. As a footnote only mods can mark questions as CW, something I originally missed. – Chris W May 24 '14 at 1:25
  • Further on my approach is that I am presuming there is already an extremely good candidate (with many upvotes and needing minimal editing) for a canonical answer out there if a question has been asked enough times to be thinking about it being canonical in the first place. We've got the answers, but maybe not the questions because there are so many ways to ask or actually different questions with the same solution. Hence my original question title of canonical questions vs canonical answers. – Chris W May 24 '14 at 1:32
  • 1
    +1 from me, only disagreement here is about the canonical meta tag being used on main. I'd rather handle it with a list of canonical questions on meta.gis.se since this doesn't require a new tag or any changes to the SE system. – blah238 May 24 '14 at 3:17
  • 1
    One other thing I would add/clarify is that I think if there is no existing Q&A which addresses a topic, it is perfectly acceptable (and encouraged) to create a new Q&A that is intended to be canonical. Again, referring to the ServerFault canonical Q&A list, in addition to the list of existing Q&As, they also have a list of topics which are not yet addressed by a canonical Q&A, which is something I think we could benefit from as well. – blah238 May 24 '14 at 3:31

You must log in to answer this question.

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