For few recent months, I've got a habit of downvoting answers which quality doesn't look OK to me.

http://i.stack.imgur.com/HlC8e.png

These probably can be generally described as low effort and/or these lacking relevance to question asked.

  • Opinionated slogans, claims that are not backed up by appropriate references or by compelling presentation of personal experience, posts that appear to ignore prior answers covering same grounds, stuff like that...

As far as I can tell, many (probably most) of my downvotes go to answers in "hot questions".

While I downvote maybe one of the answers to 5-10 "regular" questions, I noticed that almost every question with views over 2K brings answers that look bad to me.

"Hot garbage waves" in the answers appear to happen once or twice a week on average, frequent enough to feel the connection between these and respective questions making their way into SE collider list.

Is that something to worry about?


My particular concern is the poisonous effect these mis-answers have on questions, making interesting and well presented problems look the same as non-constructive popularity contests.

What is especially depressing is that regular ways to deal with this kind of issues just don't work. It's typically not difficult to edit the question to repel garbage answers, I can easily name a handful of active regulars who can and do just that.

Thing is though, it takes some time to figure how to clean up ambiguous wording while preserving the essence of original. In regular questions this works like a charm, but when editing a hot one, I often find out that when I'm done with edit, someone already posted an answer that invalidates my edit. And answer that exploits another minor ambiguity. And yet another, and so on, until my brain explodes!

  • It feels like all one gets is just like 60 seconds to figure protective edit to cover every word and letter in the question that could possibly be misinterpreted by some random passer-by and exploited for their senseless cheap shots. That's just... impossible. And more, it feels quite unfair to over-police text of such questions: per my observations "hotness algorithm" have been smart enough to pick questions that have reasonably good wording as-is.

The way how things work now helps to attract new contributors but I would appreciate if achieving this important goal 1, 2, 3 would somehow be less damaging for good, highly visible questions.

protect your own users from scale ...human interaction, many to many interaction, doesn't blow up like a balloon... (A Group Is Its Own Worst Enemy )


Here are some additional observations based on discussions that followed posting this question.

For the sake of precision, note that number 1700 above is only indicative since search does not exclude answers turned CW for other reasons. Also note some of the answers could be high-voted "unfairly": per analysis 10+ voted answers in sampled questions (below), this could happen, although rather infrequently.

Background research

For a little data to back up what was written above, I quickly went through few sampled questions with more than 10K views asked for last half year.

Please bear in mind that below list only partially represents the issue: it would be hard to do similar walkthrough for questions with over 2K views since these appear about 10x more frequently.

For the sake of completeness, note that besides 199 answers reviewed above, there are also 19 answers deleted in 16 questions listed above. These answers are visible only to moderators and to users with sufficient privileges. Of these 19, 9 are deleted by moderators, the rest has been deleted by owners. Of 10 answers deleted by owners, I would downvote 7-8.

URL used to get above questions is:
  http://programmers.stackexchange.com/search?tab=newest&q=views%3a10000%20closed%3a0

Note I wrote "I'd downvote" above since I did not really do that to all the answers I checked because of voting limits.

Summing up, I would downvote about 101-118 of 218 answers I reviewed per above research.

update

Some data related to recently added observations on community wiki status.

  • Questions with 2K+ views - about 25% are CW, 307 of total 1247

  • Questions with 10K+ views - about 60% are CW, 62 of total 101

Quoting self, wow. Just... wow.

share
3  
@JimG. hm I'd feel guilty if it turns out that votes I cast for good questions (votes up - the ones that contribute to question "hotness") contribute to something that leads to these questions getting crappy answers –  gnat Jan 2 '13 at 7:26
4  
FWIW, I try to rely upon up-voting the good answers over the bad ones. I save my down votes for the blatantly wrong answers. Down-voting the newbie answers doesn't seem to be all that beneficial since the user may or may not return to the site. I think Rachel's assessment is correct regarding where some of those low-quality answers are coming from. Good observation and question. –  GlenH7 Jan 3 '13 at 15:57
    
@JimG. while it looked merely somewhat worrying when originally posted, the more I learn about it (see updated question), the less "normal" it looks... Just for the record, your personal "CW-damage index" is currently 12 posts with total score 281, pretty much in line with that of three other users sampled so far :( –  gnat Jan 16 '13 at 9:39
    
@gnat: Whay is a CW-damage index? Is that defined somewhere? –  Jim G. Jan 16 '13 at 11:03
    
@JimG. it was "defined" in comments in another question where me and Glen discussed "down-voting the crud answers" :) –  gnat Jan 16 '13 at 11:12
1  
@gnat: Ah. So are you saying that I lost 2810 rep points due to CW? –  Jim G. Jan 16 '13 at 11:15
    
@JimG. something like that... You lost 281 votes - that would be 2810 rep points. This is of course oversimplified: you could get some rep of that before your post became CW, plus if there were many votes a day, you'd be rep-capped... For this reason, I tend to delete it by 2, ie 2810 / 2 ~= 1400, something like that –  gnat Jan 16 '13 at 11:19
1  
It's an SO question and it's not particularly "hot", but that's a good example of a question where you get a flood of crap answers quickly: stackoverflow.com/questions/16670776/… . You'd expect people to delete their own answers once they realize the "first come first serve" effect has taken care of distributing the reward or if their answers are mediocre (might be best to explain to the guy why that fails, and how to improve the code, rather than just fixing it). But nope, most gladly leave that stream of useless answers. –  haylem May 21 '13 at 15:48
1  
You might want to change your view about "auto CW". That's no longer a thing ;) –  Braiam May 18 at 2:46
    
@Braiam thanks! updated to reflect on a recent change –  gnat May 18 at 20:05
    
One thing I would like to note: it takes some time to figure how to clean up ambiguous wording while preserving the essence of original. In regular questions this works like a charm, but when editing a hot one, I often find out that when I'm done with edit, someone already posted an answer that invalidates my edit. - This tells me that your edit was invalid in the first place, as you had no way to know that the OP's intent was your version of the ambiguity and not the posted answer's version. –  Izkata May 26 at 16:56
    
no @Izkata there is a whole "class" of valid edits you seem to be missing: wiping out off-topic resource requests from otherwise good questions, editing clarifications from asker's comments into the question, editing titles to closer reflect question text just to name a few –  gnat May 26 at 17:05
    
@gnat Resource requests and title fixes aren't ambiguities. Clarifications based on the asker's comments also wouldn't be invalid, because it is what the asker intended (although ideally should be edited in by the asker themselves). –  Izkata May 26 at 17:14
    
@Izkata resource requests and titles are ambiguities for fastest-gun answerers. They skip everything what the question is really about and drop what sticks - links to resources, answers "to title" totally ignoring question text - you can find fascinating example of the latter in the git question referred in my prior comment "There is only one reason. Linus Torvalds..." In one of more recent hot questions, I even seen an explicit comment from the answerer that they really ignored question text and only addressed the title –  gnat May 26 at 17:20
    
for the record: recent "live demo" at MSE here / screen shot –  gnat Aug 22 at 12:46

2 Answers 2

Most of the "hot" questions with a large number of views have been advertised on social sites outside of SE (reddit, hackernews, etc), so the majority of the views come from Non-Programmers.SE or Non-StackExchange users.

Often these users don't know about SE rules or guidelines for answers, and think they can add something to the "discussion" by posting their own thoughts on the matter.

I looked at a few of your example questions, and a lot of low-quality answers are by low-rep users who had accounts created at roughly the time they posted their answer.

The best you can do in these cases is to downvote the low-quality ones and upvote the high quality ones so the best answers float to the top, and leave a friendly comment explaining why you downvoted so the user knows what's wrong with their answer, and doesn't make the same mistake again.

share
    
downvote and comment... how is that supposed to work in the hot questions? I mean, I've seen it sometimes work in the regular questions and I think I understand the "mechanics" - as you wrote, "the user knows what's wrong with their answer, and doesn't make the same mistake again" - indeed. But in high traffic questions these seem to simply sink under heaps of sympathy upvotes and supportive comments –  gnat Jan 2 '13 at 17:53
    
@gnat Usually "sympathy upvotes" are for negatively scored questions, so if something has a lot of up votes then perhaps its not quite as low-quality as one would think. Do you have some specific examples in mind? –  Rachel Jan 2 '13 at 17:56
    
Sympathy upvotes in answers look routine to me. I think "Git submitters" had a couple positively voted answers deleted by moderators for blatantly ignoring the notice; with 10K+ rep you can take a look. Another example... –  gnat Jan 2 '13 at 18:03
    
...another example is the one I recently highlighted in Whiteboard; I explained in quite lengthy comments what I find problematic in its original revision - despite that it got 2 upvotes. (it's quite a mystery to me why it didn't get more upvotes after much improvement done in 2nd revision but oh well) –  gnat Jan 2 '13 at 18:08
    
@gnat "Git submitters" does have a few positively scored answers deleted by a mod, but they are greatly outscored by far better answers, so a few extra votes on answers to a question with 16k views isn't that unusual (+9, +4, and -1, and I thought the +9 and +4 actually did have some value, although one was a link-only answer and the other would be better off as a comment). Your other example has less than 1k views, so I'm not sure if that's considered a "hot" question, however it appears your comments pointed out the flaws in that answer and it got fixed, which is how it should be :) –  Rachel Jan 2 '13 at 19:34
up vote 2 down vote accepted

Most likely reason for the issues with answers quality is a particular bug in a hotness formula ("fake popularity"). This bug causes some of the questions with many answers to be placed unreasonably close to the top of the collider.

Issues caused by this are explained in details in a post about particularly troublesome ("sticky") kind of questions affected by the bug in formula:

One problem with hot questions seems to be that there is no way for "hotness algorithm" to differentiate genuine popularity from fake one, that is from popularity introduced by the algorithm itself... the algorithm simply is incapable to detect cases when popularity is its own byproduct...

For questions that are not too hot... quality is mostly maintained by site / tag regulars (business as usual), collider brings moderate amount of interested newcomers from other sites with their views, votes and fresh perspective, everything is nice and fair...

But when question gets closer to the top of collider, things change... it attracts a lot of visitors whose desire to answer is not balanced by local community norms. At this point "intended life-cycle of a hot question" breaks.

Hotness algorithm assumes the question gets bumped mostly by good answers and that infrequent deviations are corrected by local community through voting and comments. Thing is though, this does not work on top of the collider...


  • Post further explains why community moderation ("quality control" through voting, comments and editing by site regulars) tends to break on questions impacted by unnatural hotness score:

    ...There are just too many new visitors to keep things under local community control, and there are just too many new voters and commenters to get things going as designed... So comes next and next and next round of garbage answers, bumping the question, bringing mindless upvotes, over and over and over again.

    Hotness algorithm calculates this as genuine popularity, keeps it accordingly high at collider, bringing more visitors that are breaking things further and so on and so on. Note it's not limited to newcomers only, everyone is invited to the party. Community regulars can clearly see that usual quality norms are broken... and that they are free to follow the New Order just like anyone else.

    That's how fake popularity makes shit stick to the ceiling and helps to keep it there.

  • It also explains why there are so much community wiki (triggered by too-many-answers) among questions affected by hotness formula bug:

    Regular open questions ("appropriate" ones) tend to have much less than 30 answers indeed.

    But when we look at "sticky" questions, things seem to be different. These questions aren't "inappropriate", otherwise these would be closed long time ago. Still, 10 examples of sticky questions have more than 270 answers total, about 27 average per question...

    ...at least 2/3 answers in sticky questions are just a garbage that somehow leaked in (probably because "fake hotness" gave question an unnaturally high exposure in collider).


Here are feature requests to address this:

share

You must log in to answer this question.

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