10
$\begingroup$

Friends, consider the following two questions:

In the US, what traffic control sign has eight sides?

and

In the US, what traffic control sign is red?

In both cases, the answer to the question is "a stop sign", but the fact that the two questions have the same answer does not make them duplicate questions.

I am of the opinion that recently there has been to many questions closed as "duplicate questions" because they had the same answer as an earlier question, when a close examination of the question shows that they were not duplicates, but two different questions which happened to have the same answer.

This can be a particular issue when one is dealing with novices who may not know the terminology to use in their question so that a search brings up the earlier question with the answer to their problem, or in cases where the problem might not seem to be the same because of the diverse nature of the meshes referenced in the two questions.

Particularly with new users, I think questions which have answers in another question should not just have the "This question already has an answer here", but rather that they should have an answer to their question that not only points to the question, but shows how the answer pointed to applies to their question.

As I demonstrated at the beginning of this proposal, identical answers do not automatically mean that the questions to which they are made are duplicate.

$\endgroup$
  • 1
    $\begingroup$ Do you have some examples of questions on the site which you feel were closed incorrectly? (related: meta.blender.stackexchange.com/q/549/599) $\endgroup$ – gandalf3 Sep 21 '15 at 18:29
  • $\begingroup$ brasshat .... you wrote ,,, [novices who may not know the terminology to use in their question]. I totally agree. I might even like a check box on the question ..[We changed your question to standard Blender terminology, to get you quicker and better answers]. I find it a bit chilling to read [marked as duplicate by X1, X2, X3, X4]. I would like an extra step to see the names. I feel a bit chilled even when I did not pose the question. Some more friendly verbiage such as [Congratulations! Luckily we have found a good answer to your question in our archive. No waiting ... its ready now ...] $\endgroup$ – atomicbezierslinger Sep 22 '15 at 17:00
  • $\begingroup$ maybe there should be a new category beside duplicate.. like 'exhibits strong similarity with another question' , because I think I agree with the term duplicate loosing meaning otherwise. But I also can't stand seeing people answering with the same content just because the question comes from a different perspective, that's silly. $\endgroup$ – zeffii Sep 22 '15 at 19:45
  • $\begingroup$ @gandalf3 maybe you know what percentage of total questions get flagged as duplicate and what percentage stay that way ? $\endgroup$ – zeffii Sep 23 '15 at 11:25
  • $\begingroup$ @zeffii Regarding your first comment, I usually link "exhibits strong similarity" questions to each other with "related" links in the comments, which IMO is a good-enough solution. Regarding your second comment, unfortunately I can't share details of mod-only analytics.. $\endgroup$ – gandalf3 Sep 23 '15 at 18:15
  • $\begingroup$ @brasshat . As new blender versions arrive, new answers may be possible to incorrectly marked duplicate questions, and thus I further agree with your statements above. $\endgroup$ – atomicbezierslinger Sep 23 '15 at 23:48
  • $\begingroup$ @gandalf3 the percentage closed as exact duplicate is just above 9% of all questions. $\endgroup$ – zeffii Sep 24 '15 at 10:21
5
$\begingroup$

Interesting read about duplicates.. : dr-strangedupe-or-how-i-learned-to-stop-worrying-and-love-duplication

I also think we need to improve this area, it can be friendlier. The label 'duplicate' could be reserved for questions which are obviously covered by the suggested duplicate question (in letter and spirit). We can't expect everyone to know if questions already exist somewhere. This goes for both those who ask and those who answer.

We should also make it clearer that the 'duplicate' or 'similar question exists' label is not bad but helps the search feature by adding new terms to the query-matching algorithms.

"a close examination of the question shows that they were not duplicates, but two different questions which happened to have the same answer."

We already have language in place which asks if the answers to the suggested 'duplicate' addressed their question adequately, and if it doesn't then the OP is asked to elaborate. When the OP takes time to come back and clarify why it doesn't address the question, we listen and often agree. Sometimes the original state of a question lacks enough detail to consider it anything other than a duplicate.

But what I, and perhaps others, don't want to see are Questions being treated as unique just because they happen to be a bit more specific in some arbitrary way. There are times when the specifics justify a unique answer, but other times the existing answered duplicate will have enough information to be helpful - that's a perfect opportunity to educate people and show them that we think the question already has an answer.

it's not uncommon for people to comment that ' yes that other question solves my problem, thank you '.


Important to note that when we flag something as duplicate it's a notification to others to check the suggested duplicate before answering. They don't have to agree, and can always say that in the comment and still submit their own answer,.. but then popular vote takes over.


Some snooping around led me to return to the Database schema for General stack exchange

Query: http://data.stackexchange.com/blender/revision/368037/474449/percentage-closed-as-exact-duplicates-of-total result:

enter image description here

This is publicly available data (As far as i'm aware... i didn't need to be logged in to do the query) . This percentage doesn't include all the questions that get unclosed.

$\endgroup$
  • $\begingroup$ Indeed it is open data. Anyone can see and use the data explorer. $\endgroup$ – David Sep 26 '15 at 17:17
1
$\begingroup$

Experienced Stack Exchange users will spot many questions from newbies (me) that are close duplicates for existing, well-answered questions. The question was probably re-asked because:

  1. the new user is either inexperienced/lazy with searching;
  2. searched, but wasn't using a key term;
  3. Wants a case-specific answer... Because an existing answer is overwhelmingly complex for their level of comprehension.

I suggest that the ANSWERS aren't as important as the process of CONVERSATION, or a current, live dialogue between parties.

@zeffii comes close to identifying a modified feature: a search that comes up with 'Related Topics' to the question (which are basically the duplicate questions). S.E. already comes up with potential searches based on what the questioner is typing in, to 'head off the question at the pass'.

But is this the point? Are we trying to foster the growth of a reference manual, or 'exchange' ideas as we need them?

There will be regularly asked questions from very new people. I think they should be allowed the chance to be answered by new people, so they can improve their own reputation. And when new people ask questions, they will find an audience with intermediate users. And intermediate level questions will find help with experienced users, and so on. I suspect this process will help develop community. This doesn't stop experienced people answering newbie questions, but often, experienced people won't answer simple questions because... well, it's been done. Leave it for someone else.

Pointing out that users have asked duplicates might not be a positive move. It can imply the questioner didn't have the courtesy to do a search first before 'bothering' the S.E. hive mind. This puts the new user on guard, and could put pressure on them to research a question far in excess of the kind of answer they're after. "Good", I imagine some might cry, "that's exactly what they SHOULD be doing!"

Nope - That's people projecting their own superiority of knowledge over others, conveniently forgetting that a large portion of their own knowledge came from the generosity of others when they hit an inevitable block.

Rather than bill them as 'duplicates', I think answering the question and adding a link to 'related questions' or 'simil-asks' ( (cc) 2015, Oro), might be a more courteous way forward. Stack Exchange's biggest power is its immediacy and specificity of rewarding people in a constructive, specific, personal conversation, and embracing this POV reduces a need to be pedantic over the vast library of knowledge accumulated over the years.

My 5 cents.

$\endgroup$
  • $\begingroup$ Dupes can be classed in two ways. 1) obvious duplicates 2 ) non-obvious duplicates. Non obvious dupes exist on a gradient, some require more blender / python knowledge to make the link than the questioner exhibits. I have definitely mellowed on this latter class. $\endgroup$ – zeffii Nov 6 '15 at 10:01

You must log in to answer this question.

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