We should be able to close questions as duplicates of any question

Question

Today, the behavior of duplicate close votes was changed so that a question can only be a duplicate close vote target if it has an upvoted or accepted answer. I think that this change is not a good one and that it should be reversed, for the following reasons:

1. The purpose of being able to close questions as duplicates is to aggregate all answers to a question in one place. This new change makes that more difficult because if a duplicate question is asked and the original does not have an answer, neither will be closed. And as @BenBrocka points out:

   what happens when one of the questions gets an answer? The effort to find the duplicate earlier goes wasted and someone else has to find the dupes yet again after one is answered, and know which one is answered just to be able to close.

2. Whether a question is a duplicate of another is entirely independent of whether it has an answer.

3. This change is open to abuse. As @fbueckert points out, one could

   Ask question, get no answer, ask same question with sockpuppet, continue until answered.

   Abusive behavior obviously isn't allowed, but if someone is even a little clever it could take two or three iterations to realize that this is what's happening and that it's not just different people asking the same question around the same time (That happens sometimes on Arqade when new popular games are released).

4. This change provides minimal benefit. It's true that new users have trouble getting old questions attention, and asking another question is often their only recourse. But if a question is old and unanswered and the new question really is identical, it would probably take a bounty to get the question answered anyway. And if the new question is not identical to the old one, this shouldn't be a problem anyway.

Answer 1

I'd like to point out that we tend to have a fairly established process for handling or updating duped questions on Arqade. If we can engage the new user (ie. They don't leave in a huff the instant we close their question), we expect more from our more established users to help them out, mostly in the form of bounties. A question hopefully gets an updated answer, or an answer at all, and everybody's happy.

For updating answers, we can continue this same process.

For unanswered questions, though, you're shortcircuiting our process. We can't close their question, and by leaving the duplicates kicking around, we raise the noise ratio and fragment our effort in answering as well. Like @murgatroid99 said, once we get two questions that ask the same thing, and can't close one of them, the chances of either being closed gets reduced drastically.

Arqade currently does a pretty good job of trying to keep questions together and close dupes. This new rule seems to encourage dupes of unanswered questions, which, if no one can answer the question, means we just gather un-closable questions. I suspect we'll shortly get answers that do nothing but go, "I don't know", and get upvotes, just so we can close them.

The point of new rules is to encourage good behavior, right? What good behavior are we encouraging by this new rule?

Answer 2

I could not agree more.

Whether a question has answers has nothing to do with whether some other question is identical.

Often we see users getting all disgruntled because their question got closed, then re-posting it in an identical fashion just to "get around" the closure. If the question was that bad in the first place that nobody answered (which happens), then we can no longer close the new one as a duplicate. Why?!

This dude agrees:

Whether the original has an accepted answer is not often a criteria used for closing a duplicate.

Answer 3

I predict the following.

1. People start flagging-for-moderator attention to ask for merges in these cases.

2. Moderators groan under the load.

3. Something different gets implemented.

Compare the progression of the delete-vote story.

I fully expect to see a response from Shog9 to the effect that this wouldn't disturb him due to the low volumes.

Answer 4

This new change makes that more difficult because if a duplicate question is asked before the original gets an answer, it substantially decreases the likelihood that either question will be closed even if the other gets an answer.

In practice, this rarely happens anyway. It takes a little bit of time for folks to find duplicates, vote to close, etc...

Whether a question is a duplicate of another is entirely independent of whether it has an answer.

Well, I wouldn't say entirely. See, most duplicates aren't exact, word-for-word duplicates. In fact, the system will stomp on your fingers if you try to post a carbon-copy of an existing question. So what you actually end up having to close is something similar but not identical. Ideally, they're both asking for the same thing though, and if one of them already has answers you can cut right to the chase and verify this.

one could ... Ask question, get no answer, ask same question with sockpuppet, continue until answered.

You don't even need a deviously-clever puppetmaster to get that effect; we've had entire classes descend on SO to all post the same assignment.

But the truth is, dup-closing tends to be a slow, labor-intensive, painful way to approach this problem. New release excitement notwithstanding, Gaming has something like 37 questions closed as duplicates of other questions that didn't meet the new criteria. Stuff that generates a lot of excitement tends to get answered quickly too. Stuff that's abusive tends to draw rather more harsh responses.

This change provides minimal benefit.

It's pretty minimal, yeah. Because... Almost no one does this.

In all honesty, I don't expect it'll make a noticeable difference one way or the other. If it does create a problem, we'll fix it.

For details on why this change was made, and how it fits into the other recent changes to duplicate-handling, see: Changes to "close as duplicate" (part deux)

Here are some recent questions from Stack Overflow. All of these received at least one vote that would've been blocked with the new restrictions:

• .Focus() is not working using jquery or javascript in JQUERY MOBILE
• A strange error in Xcode:Internal error [IRForTarget]: Couldn't rewrite external variable
• Accessing Gmails SPAM Folder from PHP
• Add full link to short link to make it valid using scrapy?
• Alternatives For iOS Development Under Windows
• android client certificate
• Attached property full syntax
• Average days from dates Mysql, PHP
• Best practice of documenting user defined events
• Capture ScreenShot with openGL
• Click on Parent then display Child but both are display. Only display Child at a time how its possible?
• Convert from and to puny code to unicode using c#?
• Could not find rake-10.0.3 in any of the sources (Bundler::GemNotFound)
• Decode a wordpress theme
• decode file $OOO000000=urldecode
• Deleting multi-line comment based on a string
• Deobfuscating functions.php
• Designing/Styles of Javascript Popup Box?
• Detect shape in canvas
• Determine PDF type using iText
• Does bind() change the function reference? | How to set permanently?
• executing drush site install in c#
• Facebook auth.login does not trigger window.location
• Facebook error dialog when login for the second time
• fast suffix tree construction in c++
• Fatal error: Call to a member function query() on a non-object in
• FckEditor converts nbsp; to some diamond characters?
• Fetch a Like-page's events? (FACEBOOK API)
• get me out from this nutch task
• Getting all posts from a certain post type filtered by taxonomy In Wordpress
• Google Maps Coordinates Incorrect
• How can I add contextmenu function for copy & paste?
• How can i keep one button as pressed after click on it?
• how can i parse the response xml string?
• how do i pass pointer ref to native method in JNI
• How do I solve this issue faced in datepicker dialog in android?
• how to define parameters in PHP from XML DOM, TO Create simple php/curl (function/class) using xml Dom
• How to generate a Reed-Solomon generator matrix?
• How to generate seo friendly URLs use .htaccess with PHP?
• How to get direction from my current position to a location in another country
• How to get the Sum from three tables?
• How to know if our app is uninstalled from iOS Device?
• How to Make RichTextBox Text Only?
• How to save a file to a shared folder in a server via WPF and C#
• HTTP GET and POST
• Human voice conversion to Darth vader Voice in ios
• Indicate mode in GNOME Terminal vim
• interpolating a rotation through an angle with numpy
• 'iostream': No such file or directory
• Is C++ context-free or context-sensitive?
• Is it possible to download app from google play from my app - Android
• Is it possible to enable sandboxing a Facebook app for specific devices?
• Is user logged into Google?
• k-d tree implementation tradeoffs
• lnk2019 error when using glew
• Making Enter key to work as tab key for all controls on a form vb 6.0
• MPMoviePlayerController getting paused when app enters in background
• Must like on Facebook to view a webpage
• Only for experts : Facebook API PHP/Javascript - Refreshing page after Ajax call
• Out and ref in TypeScript
• probem while working with ITLSSPProc.dll
• Put RadioButton's text to the left of clickable area
• Qt Creator - How to add a compiler?
• R legend for color density scatterplot produced using smoothScatter
• read from USB the 5v source voltage
• Resizing photos on upload
• Sharing text in facebook with Android share intent
• The impossible task - Format string attack
• undefined reference to `vtable for ...'
• VBS to Delete the folders and files
• VF2 (or other) graph isomorphism implementation in Java
• Why android.os.IServiceManager not resolved in android
• Why isn't .length() a method for arrays in Java?
• Windows 8 - open modal dialog unwanted UI issue appears
• Winforms - UI tools
  Look 'em over...

Update: Daniel Beck pointed out a problem with my query, which I've corrected - the proper number of results for Stack Overflow is 75 in the past 30 days, listed above.

Answer 5

Ask Ubuntu and off-topic bugs

On AU we have the policy to close questions/issues about bugs as off-topic. Sometimes information related to the bug is very valuable and questions get updated in the comments or body. Even when it's just a link to the bug report, it's still information you don't want to add to every duplicate question over and over again.

With the new close-as-duplicate system it's impossible to mark these questions as duplicate. Here's an example of question of which we get a few per week of:

green dots on youtbe videos which is a duplicate of Why is Google Chrome displaying artifacts in YouTube?

The latter provides useful information and if I was able to mark it as the duplicate it would be a lot easier to post updates about it. Also, we now have to close all duplicates as off-topic individually which is a bigger turn off for new users (probably) than pointing them to a question with more information.

Note: As Gilles pointed out, I could just upvote one of the answers in this case, but maybe this was not the best example given. If it were to be closed sooner it wouldn't have any answers in the first place.

Answer 6

I agree with the OP but I understand the reasons for doing this. Having unanswered duplicates closed in chain makes it less likely that any will get an answer as a potential answerer has to go through the entire chain to find the correct question to answer.

I think this could work, with one proviso; that it's still possible to vote to close as a duplicate. These "Possible Future Duplicate Votes" (PFDVs) would be stored in the database with both question IDs. If any question in the ensuing network gets a "good" answer, however that is defined, then there should be some mechanism, probably manual, of either merging the questions in that network into the original or closing the entire network as a duplicate of that question, as appropriate.

Otherwise, the original point of SE is, to a certain extent, undermined. There is no single canonical answer to a question that the knowledge can remain spread out over multiple questions and answers.

Answer 7

I disagree with this suggestion.

People come to StackExchange for answers, and it is very frustrating to spend time writing up a question, only to have it get closed as a duplicate to another older question that has no good answers.

It's not just the question-askers that find this frustrating too. There have been times when I find a SO question that exactly matches my situation, but it's closed as a duplicate, so I follow the duplicate link only to find no good answers there, or that the linked question doesn't exactly match my situation so the answers listed don't apply to me.

As Shog said here (emphasis mine):

Additionally, close targets are restricted to answered questions even when specifying the ID or URL directly (with the exception of questions from the same author, and meta posts). Moderators can override this last restriction if necessary

So if someone really wants to take the time to create a bunch of extra accounts to abuse the system and keep re-asking the same question, our "human exception handlers" can step in and take care of the situation.

As for having multiple unanswered questions that ask the same thing, what's the problem with that? There's now a higher chance that the question will actually get an answer, as it increases the chances that one of the questions will cross paths with a user that knows the answer.

To keep all the unanswered "duplicate" questions together, simply link the two via a comment so it shows up in the "Linked" question list on the right side. Once one of them gets a good answer, users can go through the list and close all the duplicates.

And last of all, minimal benefit? I would disagree.

It's so frustrating to post a new question, and have it get closed as a duplicate to an older unanswered question. Older questions do not get nearly the same attention that new questions get, and there's no guarantee the OP of the older question is still around to answer questions about their post, or to provide any input at all towards getting a good answer.

In addition, the OP of the closed questions won't get notified of answers/comments posted on the duplicate question, so they are unlikely to see any new answers that get posted there or to respond to comments asking for clarification.

The new rule of only closing as a duplicate to answered questions is much more helpful to people seeking answers, which is primarily who the StackExchange sites are for.

Answer 8

Merge by linking to one "master" question.

Since the user/admin has already located duplicates, the messy and error-prone work is done. Merging would only require...

• Close the "slave" questions for edit and additions and add a "Merged to" link at the top (bottom?) of those questions pointing to the chosen "best" question which becomes the "master."
• Only the "master" question would continue to allow edits and additions and "Merged from" links would be added to the bottom of it pointing to each of the "slave" questions.

This would allow the frustrated, relatively new, user to post a new question (which may very well contain additional aspects to the issue) and more experienced users to later "merge it" with the "master".

The inexperienced users would not be driven away by perhaps their first question being unceremoniously closed without answer, while at the same time providing a potential path to a solution. At the very least, the related questions would likely provide additional keywords for research.

edit...
Proposed Experiment
We could get some useful empirical data by initiating a practice of creating a "Duplicate Of" answer on duplicate questions. It would contain just links to the original and other duplicates. No need to establish a "master" per-se. The new duplicate would not be closed yet so by practice, the "Duplicate Of" answer could continue to be edited with additional links to potential answers. The better the answers on those other questions, the higher the "Duplicate Of" answer would get voted up.

If this proves useful, the first coded enhancement might be to automatically vote up the question being linked-to as well, like Google does for search hits. Another might be to only allow editing of the "Duplicate Of" answer with the rest of the question closed.

Answer 9

Could there be some way to automatically merge the questions when the prior near-identical one has no answers?

Answer 10

The two biggest problems I see with this change is that

1. it promotes asking an unanswered question and answering it yourself (rather than answering the old question, stealing the original askers rep) and
2. questions like this one are being closed as dupe of a different question, though there exists another question, which is the same as the first one, but is not being closed as a dupe.

I don't see the benefits outweighing all of the cons mentioned so far in the answers on this question.

Answer 11

I wish StackExchange would properly merge all the answers from linked duplicates. This would:

• ensure that I get to see every answer to the question, no matter where it got posted
• make the best answer of all the answers float to the top due to upvotes
• prevent people closing similar questions as duplicates, when in fact the existing answers are to a slightly different question.

Answer 12

When we cannot pick an original to the duplicate with zero upvotes / acceptance, we cannot encourage re-use on low-activity tags where the majority of users are first-timers who neither upvote or mark as accepted (i.e., no netiquette).

Example:

• The duplicate (2021-04-03): Remove successive duplications from a list in SML using pattern matching
• An original answer of 3-4 full screens of explanation, code, hand-written evaluations (2020-12-15): Removing elements from list in sml

Many questions in the SML tag relate to university coursework, so we see an abundance of duplicates, and an abundance of new users who don't come back to mark their answers, but an excessive amount of re-asking the same question without looking for existing answers. There is also only a handful of people who answer questions, and if nobody is around to +1 your post, you can't actually mark it.

In this scenario, the "mark as duplicate" feature is practically removed with no gain.

This disincentivises giving good answers where there is a market for them.

Answer 13

There are 2 scenarios to see the message “This question does not have an upvoted or accepted answer” -

• when current question has a good answer,
• and both questions do not have good answers.

For the first case the current behavior is good. Just add the instruction

“Consider to close the referenced question as a duplicate of current one”.

If both questions do not have good answers, I like the idea of the deferred close from Changes to "close as duplicate" (part deux). Instead of current disallowing to close change the message to something like

“Neither current nor referenced question do not have an upvoted or accepted answer. We will record your vote, but only after one of the questions will have a good answer, other one will be considered to be closed”.

At the same time the standard “possible duplicate” comment should be added ( may be to both questions ?).

The vote shouldn’t go to “Close” queue, but wait until any of the questions will have upvoted/accepted answer.