11

What's the policy on old questions whose answers are no-longer current (but were correct at the time).

For example, this question: Seeking QGIS tutorials and web resources? - has a number of answers very highly rated. The problem is that quite a few of the links are pointing to outdated material now and don't relate to the current QGIS.

Is it worth re-asking the question seeking updated/current answers?

Or should we rely on users adding new answers to the old question? The issue here is that they'll never get upvoted and thus are less likely to be seen.

Thoughts?


While @whuber's and @PolyGeo's answers are both very good, I still feel they don't work. I don't know what the "correct" answer here is, but these responses elicit several problems.

  • We can't make the "accepted" answer non-accepted, and the accepted answer is always at the top.

  • There aren't enough active members to down-vote all the upvotes of the now-incorrect leading answers.

  • We would be punishing the users who once had the right answer by downvoting.

  • The current incentive program (badges) is obviously insufficient for addressing this fully. For one thing a lot of them are /hard/ to get, even at bronze level, and require a very significant time commitment (or a lot of luck).

  • Closing them isn't a solution because then it's impossible to get new answers unless re-asking the question is allowed.

  • The site is heavily optimised to encourage new answers rather than substantially editing the content of old.

And probably the biggest issue of all:

  • For various reasons, only a small fraction of users will use/edit old questions and provide new answers. This means the question is only getting a fraction of the exposure it got when it was a new question; thus any new answers are likely to be suboptimal.

  • (This is also a problem with editing an answer).

Maybe the solution is to allow reposting of the question, but close the old one and stick a date on it. That way you've got a historical collection of answers and all are canonical for their period?

  • 1
    Related: meta.gis.stackexchange.com/questions/3635/… – Chris W Oct 10 '14 at 5:01
  • @ChrisW +1 that is related but I would not class it as a duplicate because I think this one is about a Shopping List question from Main. The earlier one was about a nicely focussed question from Main. – PolyGeo Oct 10 '14 at 7:32
  • @ChrisW - good catch, I didn't find that one in my search to see if this had been asked before. It's similar but ironically has the same problem as the question itself raises - is the answer current? :-) – GIS-Jonathan Oct 10 '14 at 13:02
  • 2
    There appear to be some misunderstandings in the edit. The accepted answer should not be considered "correct": it merely is the one highlighted by the OP. You also mischaracterize the badges For instance, many of the bronze badges I mention are extremely easy to get--some require just one modest action. I do (thoroughly) agree with some of your negative comments about downvoting: it is not appropriately applied to answers that originally were correct and has little effect in such cases, anyway. – whuber Oct 10 '14 at 14:31
  • @whuber - I'm aware "accepted" doesn't necessarily mean correct; but as a result of being marked so, it will perpetually remain the topmost answer. – GIS-Jonathan Oct 10 '14 at 14:39
  • @polygeo I agree it's not quite a duplicate which is why I commented rather than voting to close. But I don't think the type of question is really relevant to the question at hand: old question, old answers, how do I raise the issue again (because they are or might be wrong now, out of date, etc.)? Since posts only bump if edited or answered, and then only if looking at the 'active' view, the point about people not going back is valid to me - my own example: gis.stackexchange.com/questions/91537/… – Chris W Oct 10 '14 at 19:22
  • 1
    I think your question, which is trying to deal with a number of issues, could benefit by using the terms accepted and downvoting etc rather than correct (who's to judge?) and downmodding etc (new terms to me). – PolyGeo Oct 10 '14 at 22:20
  • @PolyGeo - Thanks, I'm a little rusty of the local lingo. I've edited to correct all instances. Where "correct" remains, it is used in the context of right/wrong. – GIS-Jonathan Oct 13 '14 at 9:37
6

Please do not re-ask the question: that goes against one of the principal aims of the site, which is to have a well-organized canonical set of answers. Re-asking creates a duplicate. Duplication more than doubles the efforts of those who search the site for answers (including your moderators and many active community members who conduct these searches as a matter of course before answering questions).

If this community cannot or will not curate questions (by editing them and adding new answers as needed), then perhaps we should not be managing questions of this sort at all. Questions of fleeting interest or value and those that collect myriad varied results ought to be particularly suspect. Perhaps some of them should have been closed long ago and maybe they could be closed now?

That leaves us with the option of maintaining old questions. The site offers some incentives to do so, including

Engaged readers might want to take the existence of these badges as a personal challenge: can you earn all the bronze ones? Can you earn any of the silver or gold ones?

  • A good answer, but I feel there are a few issues with the proposed solution - see my question edit. I'm also not keen on disallowing such questions - they're topical for users, and a whole host of questions would then arguably be invalid (datasets, tutorials, help, software & tool related, etc.) – GIS-Jonathan Oct 10 '14 at 13:11
  • 1
    Cool, new badges! Note to self: find something to edit before answering anything. ;) On a more serious note, I agree with a path being set here. I think some questions, such as the example given, are good for the site but limited in time. I wouldn't call them fleeting per se, but at some point they need significant overhaul or just to be closed (protected?) and dated. If software constantly changes and evolves, at some point every software question is going to fall into this 'fleeting' category. Though often an old, outdated answer can help direct one to find a newer, working one. – Chris W Oct 10 '14 at 19:31
  • I do agree curating is a desirable route to go as well, but an issue I see there is modifying the original question so much it's basically a new question - perhaps all at once, or perhaps over time. Such questions or topics seem like Community Wiki candidates to me - which I note the example given is. – Chris W Oct 10 '14 at 19:34
  • 1
    @Chris Thank you for pointing out a potential misunderstanding. What needs curation most of all are the answers, if those go out of date. If the question becomes so out of date that it is no longer answerable or relevant to this site, then don't you think it should just be closed outright? (Remember, closed threads are still accessible to people who really look for them, as opposed to threads which have been deleted: those can be seen only by mods and very high-rep users.) Remember, CW status is not a cure for bad material and should not be used to avoid maintaining our site. – whuber Oct 10 '14 at 20:25
  • The problem with talking about hypotheticals is it's so hard to come up with specific examples for or against if you don't have one in mind. :) Yes, I agree in principle. But taking the example question, how would you curate a 28 upvote answer consisting of links (and summary!) to a video series that vanishes? Replace it with links to a whole other video series (hence, answer completely changed but has all the credit rating of the old one)? How likely is it new answers (say the new series instead of an edit) will ever approach that many votes? Perhaps as you say the problem is in the question. – Chris W Oct 10 '14 at 20:49
  • 1
    @Chris If those 28 votes were deserved (I will continue to speak hypothetically because it is not my intention to critically evaluate any particular answer), that would have been by virtue of the quality and thoroughness of the summary. Eliminating the links ought to solve that problem without making any material change in the answer at all. This is precisely why we insist that answers be able to stand on their own! – whuber Oct 10 '14 at 20:55
  • Again, I agree with you in principle. But I note the top voted answer on that example question is pretty much a link list, albeit a maintained one. Or maybe just the titles alone, similar to referring to books, are sufficient (ie if the link doesn't work, search for it on your own). And it could just be the nature of the question (asking for other/outside resources) we're discussing as an example. As your own comment to PolyGeo's answer suggests, it seems CW is the cure/approach there. – Chris W Oct 10 '14 at 21:06
  • @ChrisW - I agree with your points. At the risk of continuing to disagree with whuber (sorry!), I'd point out that the top answers here are link-only, but some questions, such as this one must be link only. What could you write for an answer to this question (and others like it) that wasn't a link yet still warranted upmods? Even if the answer contained a detailed review of each resource, it would still be the links that actually make the answer useful. – GIS-Jonathan Oct 13 '14 at 9:45
  • 1
    This actually is a good argument--and the one accepted throughout most of SE--for why such list-of-link questions should be considered off topic: we just cannot curate them well. Being popular is not a consideration: some of the most popular questions on SE have still been closed for this reason, such as stackoverflow.com/questions/1711. They are left undeleted "for historical significance" and locked to prevent further editing. – whuber Oct 13 '14 at 16:16
3

I think this answer (which I am now updating) is a good place to address this part of your question:

  • For various reasons, only a small fraction of users will use/edit old questions and provide new answers. This means the question is only getting a fraction of the exposure it got when it was a new question; thus any new answers are likely to be suboptimal.

I don't know of any statistics to show how views of questions tail off over time if no editing is performed, but I suspect that would happen to lower profile questions, which is why I think Editing is Essential to keep returning any valuable Q&As to the active queue in a slightly (preferably significantly) better state each time.

I think this is especially true of any Shopping List questions with high profile, like the one you cite, because it is currently #6 on the self-assembling FAQ for QGIS here. In other words, I think it may still be getting a lot more exposure than you think.

I would encourage everyone to use their editing privileges to improve answers by adding content which is new (or consolidated from other brief answers to the same question).

Questions such as this one draw many visitors and future users to our site, so I think it is important that they have a well constructed question with fewer great answers rather than a long list of answers that includes some that might now be considered ordinary.

How aggressively should we maintain and improve very popular questions? looks like it may be worth reading here too.

If you are wondering how to identify the "very popular" questions, which I think are also those most likely to be the duplicates you seek when you read a question and think "I'm sure that's been asked lots of times before" then I recommend becoming familiar with the frequent tab to highlight them in the results of any tag search you do. The FAQ for QGIS is one example, and another is FAQ of ArcGIS for Desktop OR ArcPy.

  • I recommend the editing part of your answer, but it should be stated clearly in the text that the link has changed since the original answer was given. – AndreJ Oct 10 '14 at 5:19
  • I did not mean the links you gave in your answer, but (related to the question) dead links in old answers. I would not suggest to simply exchange those links, but extent the answer saying "Update: those links are dead now, please follow this new link." I don't like downvoting those old answers just because the link has changed. – AndreJ Oct 10 '14 at 8:13
  • @AndreJoost (and Poly) - I've edited my question to raise a few issues I see with the solutions proposed. – GIS-Jonathan Oct 10 '14 at 13:08
  • 1
    @PolyGeo Have updated the post you are referring to gis.stackexchange.com/questions/3651/… - it is quite hard to maintain as websites come and go. – Mapperz Oct 10 '14 at 13:42
  • 2
    I would ask you to reconsider your downvote policy. Downvotes are a tool for encouraging people to modify their answers as well as a way to signal problems. However, many originators of old answers will no longer be active. It is better--and much more friendly--to edit the post yourself when you detect such a chance to improve it. With community wiki (CW) threads, anybody can make those edits; you don't need a high reputation. (If answers in a thread rely on rapidly changing material, links to other Web sites, and lists of possibilities, then that thread should be CW in the first place.) – whuber Oct 10 '14 at 14:25
  • 3
    I agree with the others that downvoting isn't a way to go here. I have my own issues with downvotes and a recent increase in their use, but I'll get into that elsewhere. Now if an answer is truly wrong or incorrect or whatever, by all means downvote. But just because it's old or outdated doesn't justify a downvote to me - at least by an experienced site editor. I agree with others it's better to find the updated link and fix it, or edit the post to remove the link and state it's deprecated and no replacement found (assuming it's not link-only with no summary whatsoever). – Chris W Oct 10 '14 at 19:41
  • I think I will remove the downvote aspect from this answer with a view to either adding a new answer specific to it or more likely a new question. I fear it is distracting from the key point of the other answer which is that simply re-asking a question is counter-productive. – PolyGeo Oct 10 '14 at 20:59
  • @Mapperz Thanks for doing that - we can only fix 'em as we find 'em - my downvote has changed to an upvote, and hopefully your answer will start to float towards the top. – PolyGeo Oct 10 '14 at 22:03
  • 1
    @whuber In quick response to it being friendlier to "edit the post yourself" than to downvote. Of course, editing is always my first choice, but this was an occasion where I knew the answer no longer had value in that state, but I did not have the knowledge (or interest to acquire that particular knowledge) to edit it into a better answer myself. – PolyGeo Oct 10 '14 at 22:11
  • An excellent clarification and good point about the lack of statistics. I was basing that comment on the fact that the newer answers have only a fraction of the votes of the older ones. Given the number of new-but-lower-rated answers to that question, it's clear that editing the top answer isn't something that people want to do. Further, to me at least it's also something that seems ... wrong. You're subverting an answer the community accepted with new content that may not be worthy of the +50 upvotes. Do we really want to encourage that? – GIS-Jonathan Oct 13 '14 at 9:57
  • To me it "seems ... wrong" is precisely where I think Stack Exchange is different from all other Q&A, Discussion Forum. Bulletin Board, etc sites. I think we really do want to encourage the view that no Question and/or Answer is ever perfect, and if a change is made that takes it further from "perfect" then the expected outcome would be a rollback. – PolyGeo Oct 13 '14 at 10:07
3

To address this dot point:

  • We can't make the "accepted" answer non-accepted, and the accepted answer is always at the top.

A minor point is that the accepted answer is always at the top, unless an asker accepts an answer of their own - in which case its position is determined by its votes alone.

On the main point, I am not too concerned if the highest voted answer sometimes appears second when sorted on votes, because it never appears lower than that.

However, if there is a clearly incorrect answer accepted at the top then you could:

  1. Make a comment to that effect
  2. Edit to correct it - but be ready for a rollback if someone disagrees
  3. Downvote it (a recommended action to any incorrect answer by the two top voted answers to What is the etiquette for correcting old questions with incorrect answers?)
  4. Vote to delete it - but you need to be a trusted user (20,000+ rep) and for the answer to already be at -1; and
  5. as a last resort, a flag to a moderator may lead to it being assessed for deletion, but if it is from a topic where they lack expertise, then without a preponderance of negative comments and downvotes already in place to guide their decision, it is most likely that it will remain.

By coincidence, I just came across an example where a Nice Question (with 10 votes) has an "incorrect"/"ill-advised" accepted answer with -14 votes at its top (ahead of an answer with 8 votes): Consuming Google Maps as background map through ArcGIS Server? I have added it here to show that it is not just hypotheticals being discussed.

Can we exempt downvoted accepted answers from getting the top spot? is worth a read too.

There is a feature request at Meta Stack Exchange that may also help in this situation: Keeping special status for Accepted Answers without sticking them to top forever?

You must log in to answer this question.

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