1

TL;DR Can the system do something to help us more efficiently handle quality of the content in questions with multiple answers?


Recently I've seen notes from Stack Exchange staff about how it would be nice if site would allow for broader, softer kind questions. This was said by Robert Cartaino here:

The idea that there are hordes of people just waiting flood the site for want of a single precedent was one of those things we worried about before we had anything to back it up. But if you haven't noticed, Stack Exchange has moved on from a lot of those old adages foretelling of widespread doom in ways that simply never materialized in actual use... every question must have exactly one absolutely factual answer. ...

...and by Jon Ericson here:

Given quality of users on this site, there's no reason opinions (backed by research and experience) shouldn't be useful to other programmers...

Thing is though, the system makes it difficult (I would say, unnecessarily difficult) to produce and maintain quality content in questions with many answers. It doesn't provide any means to make it easier and some features even tend to incentivize low quality content in questions of this kind.

Current UI doesn't offer any means to help answerers and readers find / avoid repeating what was stated before. Combined with being very easy to post an answer, this leads to useless answers that don't add anything over what was posted before. This is noticeable even in questions with brief answers which are easy to check (example) and when answers are lengthy, things become even worse.

This penalizes web search visitors, forcing them to read same points over and over and over again. Letting repetitive answers proliferate is a step toward old-fashioned "forums", and quite a substantial one I think.

Particular feature that essentially encourages low quality content is Hot Network Questions.

To set the record straight for readers not familiar with its purpose - it is established pretty firmly and it has nothing to do with content quality. The feature is there only to show network wide audience entertaining / interesting questions. If you're interested, refer MSE for more details on that: What is the Goal of “Hot Network Questions”?

Per se, entertainment aspect of HNQ looks pretty acceptable and I don't think it should necessarily be that it makes a critical obstacle in maintaining quality content. Thing is though, as implemented currently... it really does.

HNQ, the way it is now, make maintaining quality content cumbersome and painful. Problems with content quality in hot questions are well known for a long time (1, 2), and it keeps happening here and now, see for example Recent Trouble With Popularity


Over years, Stack Exchange tried various ways to handle questions with multiple answers but none of these seemed to work.

  • Jeff Atwood's warning was probably the first one,

    Did you read through all the existing answers first to make sure your answer will be contributing something new?

All I can say about it is that it doesn't work, has never worked and is unlikely to ever work.

  • Another thing they tried was automatic conversion to community wiki when there are many answers.

This worked... sorta, in the sense that it made it easier to downvote low quality answers (without incurring rep penalty) and removed incentive to flood popular questions with answers in a (frankly, quite realistic) hope to gain a bit of rep with a random upvote.

The problem with this trick is, it had some very ugly side effects due to involved rep denial from users providing quality content and in the end it was deservedly discarded. Thing is though, nothing was done to compensate loss of features used to counter issues in questions with many answers.

Sure, there's standard rhetoric about voting down low quality answers but tell me, how many users can afford rep loss of voting this way, say, 5-10 of total 20 non-CW answers in a question? how many of them can afford doing this every month? every week? And if you additionally consider "sympathy upvotes" frequently happening to such answers in popular questions (each worth more than 4 votes down rep wise), this starts looking like a wishful thinking.

  • Another approach that was tried about two years ago is randomizing hot questions.

This seemed to work at first but as time passed by it revealed signs of inability to scale. Shuffling 100 hot questions to 100K site visitors works, shuffling same amount to 1 million will break and essentially return us back to old known issues with "snowball effect" when questions stick in the list solely because they were advertised there.

  • Somewhere along the way they implemented special adjustment for Stack Overflow questions, which strongly pushes these from hot list after 7 hours.

This seems to work, in the sense that there aren't very loud complaints about hot questions at MSO anymore (like it was in the past with "croissants"). Still, this doesn't resolve problems at other sites. And still, even SO is having garbage answers problem in questions that acquire popularity and high views organically, not via hot questions.

  • Yet another feature intended to help control answers quality is question protection, which makes it impossible to answer for users having less than 10 on-site reputation.

By design, protection is intended to handle only cases of extreme abuse. It doesn't provide any means against low quality answers posted by users with over 10 rep. Also, due to some implementation specifics, even very limited functionality it offers frequently turns out useless in hot questions.


Can the system do something to help us more efficiently handle quality of the content in questions with multiple answers?

  • @tchrist "system" is Stack Exchange software, see eg section of the question "tried various ways..." and this answer – gnat Nov 29 '15 at 23:34
  • @tchrist meta.programmers.stackexchange.com/posts/7770/revisions (I forgot about this funny limitation on suggested edits, sorry) – gnat Nov 30 '15 at 0:10
  • 1
    As someone who values reading multiple answers to a question, I think the voting mechanism is enough for me. Answers worth my time reading are at the top, and if I am seeking caveats or alternatives there are other posts to go through. Also personally, seeing a single upvoted answer is often not enough for me. I like to see alternatives that answer is compared against, and don't even mind seeing similar answers that reaffirm the top-voted one. The key that makes SE so useful to me is all of them are Answers to the Question, and they don't stray off-target. – Rachel Nov 30 '15 at 15:23
  • @Rachel your perspective seems to be much different from mine because I use different sort order (active). Would be really helpful to know stats for what kind order is most preferred but unfortunately my request for that hangs ignored: How many votes and flags are cast by users preferring particular answer order (active / oldest / votes)? – gnat Nov 30 '15 at 16:00
  • Do you have any examples of recent (~post NPR) questions that were overwhelmed by multiple poor answers? – yannis Dec 1 '15 at 8:29
  • @Yannis with current site norms, answer quality issues look pretty manageable to me (multiple answers still are troublesome in off-topic questions though). Post is intended to address recent ideas about relaxing these norms: "notes from Stack Exchange staff about how it would be nice if site would allow for broader, softer kind questions" – gnat Dec 1 '15 at 9:33
  • ...guess at the end of this lengthy post one can simply forget what was written in the beginning :) – gnat Dec 1 '15 at 9:46
  • I'm not entirely sure what problem you are trying to solve here. You talk about repetitive answers, yet I don't even remember the last time I've seen a question on Programmers being overwhelmed by repetitive or otherwise poor answers. Sure, there's the occasional noob answer that doesn't add anything new, but is that a problem worthy of this long diatribe? Why does the system needs to be adjusted in any way? Where exactly is it failing? – yannis Dec 1 '15 at 9:46
  • To be more specific: How do you think the system should have handled these popular questions with multiple answers? – yannis Dec 1 '15 at 9:52
  • @Yannis I am not sure I understand what you don't understand. With current norms (settled more or less firmly 2-3 years ago), everything looks under control, just like you say. Post is not about what we have now, but about suggestions to change it – gnat Dec 1 '15 at 9:55
  • I don't see any suggestion to change our scope or norms in your quotes. Programmers was never a "one answer to rule them all" site, and opinions (backed by research and experience) have always been acceptable answers here. The quotes may just be a subtle reminder that we have taken a good idea to a ridiculous extreme.... – yannis Dec 1 '15 at 10:01
  • ridiculous or not, this way we have answer quality issues under control, as your own observations suggest. I can understand those willing to try softer approach but want us to realise that the system is designed sub-optimally (softly speaking) to handle this – gnat Dec 1 '15 at 10:30
  • 1
    I have no idea what this is asking. You just sort of launch into trying to back up your proposal without actually clearly stating what it is. What does "handle quality of the content" mean? And what's not "efficient" about how it's currently performed? And what in heck does "question has multiple answers" have to do with anything? – Lightness Races with Monica Dec 11 '15 at 23:57
  • @LightnessRacesinOrbit it took you how long, 5 minutes to read whole post and check the referred links with examples and related discussions? no wonder then – gnat Dec 12 '15 at 0:08
  • "...becoming worse at it" – gnat Feb 27 '16 at 20:52
1

As an example, here are some ideas of what can be considered:

You must log in to answer this question.

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