What would be the primary audience of this space? The initial proposal made almost 4 years ago seemed to be centered on editors, while the most recent discussion is bringing examples more focused on sysadmins / developers. We could say that ask.wikimedia.org would be for everybody since day 1... but maybe we can get something out of the door easier if we have a specific audience in mind?

Would it make sense to start with developers interested in MediaWiki and the Wikimedia APIs? It's an audience smaller than editors that can be handled in English only (at least for starters). We have a problem of proliferation of support spaces for such small audience (wikitech-l, mediawiki-l, mediawiki.org's Support Desk, stackoverflow, Facebook, more?) We also have #dev.wikimedia.org aka Data & Developer Hub, which could benefit from a good resource for technical support.

Then, whatever tool and workflow we define for this audience could be considered for other audiences, either increasing the scope of a one and only ask.wikimedia.org or cloning the solution for each purpose.

I think github may be added to Quim's list.

I think (looking at mediawiki.org's support desk for example) the main audience is the setup and usage of mediawiki core and other extensions (in a limited way, depends on, if the extension has an LQT talk page). But i think a portal like ask.wikimedia.org (just as an example) could handle anything related to mediawiki, such as topics for sysadmins, usage information and help for editors/administrators/..., help for developers using mediawiki's api or other way's to access mediawiki-data (maybe tool labs, too). Thinking about that, an intelligent structure is essential, based on tags or specials boards or any other way, but a big big pool with all questions without any organization and structure sounds bvery bad ;)

Having this big picture in my mind, stackoverflow is (by far) the worst place to handle it (if it is the only place), at least in the philosophie of the community and the goal of the platform (mostly developer questions handling code issues and problems), the technology, on the other hand, is one of the best for a Q&A system (m ypersonal opinion) and a self moderated community driven portal.

StackExchange prefers to collect questions about the same topic but for a different audience at different sites: StackOverflow for programmers, ProWebmasters for site administration, SuperUser for users etc. So all MediaWiki-related questions fit in somewhere, although the user experience is fractured (improved somewhat by the ability of admins to easily move questions between the sites, but still an annoyance).

We have a problem of proliferation of support spaces for such small audience

Hence the suggestion to close down some of them where we can. :) Whatever we think about it, we can't prevent people from asking stuff on StackExchange.

In T31923#1260218, @Qgil wrote:

What would be the primary audience of this space? The initial proposal made almost 4 years ago seemed to be centered on editors, while the most recent discussion is bringing examples more focused on sysadmins / developers. We could say that ask.wikimedia.org would be for everybody since day 1... but maybe we can get something out of the door easier if we have a specific audience in mind?

Would it make sense to start with developers interested in MediaWiki and the Wikimedia APIs? It's an audience smaller than editors that can be handled in English only (at least for starters). We have a problem of proliferation of support spaces for such small audience (wikitech-l, mediawiki-l, mediawiki.org's Support Desk, stackoverflow, Facebook, more?) We also have #dev.wikimedia.org aka Data & Developer Hub, which could benefit from a good resource for technical support.

wikitech-l is not a support space. It's a place for developers to collaborate on improving the MediaWiki technology. In my opinion, of those you listed, the only two that are both Wikimedia-run and overlapping are mediawiki-l and Project:Support_desk (not sure who started the Facebook).

In my opinion, overlap is not inherently always a problem. It just needs to be managed carefully. For example, Ubuntu has both askubuntu.com and https://lists.ubuntu.com/mailman/listinfo/ubuntu-users . People have different preferences.

http://www.discourse.org/ was mentioned in wikimedia-l, and I wonder if that could be a good candidate for a support site.

@Aubrey, are you aware of cases where Discourse is being used for this purpose?

I thought Discourse is just a nicely-designed general-purpose forum. It seems Flow should be able to serve its use case (for the MediaWiki world).

I don't think Discourse is intended to be a Q&A app ala Stack Overflow.

@Quim Discourse comes from the very people who created StackOverflow, so they know the difference from Q&A and discussion, and Discourse is for the latter. There is no open source real alternative for StackOverflow software, it is aimed to be the "only place" for programmers to talk. I'm on mobile now, but there are interestinng links i can provide.

There is no open source real alternative for StackOverflow software

[citation needed] Please edit https://en.wikipedia.org/wiki/Q%26A_software#Comparison_of_Q.26A_software

Discourse is awesome, and I would love to see us starting to use it, but it's discussion software, not Q&A software (unlike the many alternatives that have been mentioned in earlier comments). See e.g. this or this conversation.

What is the problem that needs to be solved here? There are several mature opensource Q&A products, some even in PHP.

In T31923#1334283, @Tgr wrote:

What is the problem that needs to be solved here?

That is a good question, and at this point I wonder as well. A list of use cases or features would be useful.

There are several mature opensource Q&A products, some even in PHP.

An alternative option being Flow with a system to rank comments and... what else? If we have a list of features, we can compare.

IMO, minimum features of a Q&A site which are different from generic forum software:

• differentiate questions and answers (most extra features that affect answers ("posts") should not affect the question)
• differentiate answers (replies to the question) and comments (replies to an answer) - comments on questions are nice but optional
• vote up/down answers
• sort answers by votes
• accept an answer
• question tagging

Use cases:

• ask a question
• answer a question
• correct/punish bad answers
• find a question you need answered
• find questions you can answer

I would add this to the list of required features:

• Stick questions/answered questions to the top (of a tag search, if tags exists), that would especially useful for questions, that are much more important temporary as normally
• mark a question as a duplicate of another one (currently we have a lot of questions, which are, in fact, the same as another one, which would duplicate an answer

Apart from that: +1 to all points of @Tgr :)

I don't think it's useful to turn this task into a description of standard Q&A software features. Please edit https://en.wikipedia.org/wiki/Q%26A_software#Comparison_of_Q.26A_software

Wikipedia strives for neutral point of view and referenced information. Here we are discussing which features we want for us, and which tool could be the best for us.

Not really, those bullets points just define what a Q&A site is.

In T31923#1368606, @Nemo_bis wrote:

I don't think it's useful to turn this task into a description of standard Q&A software features.

If we are seriously considering using Flow as a Q&A tool, then it certainly is.

I'm a bit lost in the conversation, but please check these links (especially the first one):

• List of SO clones: http://meta.stackexchange.com/questions/2267/stack-exchange-clones
• Is SO code open source?: http://meta.stackexchange.com/questions/14656/is-the-stack-overflow-source-code-available

In T31923#1368843, @Qgil wrote:

Wikipedia strives for neutral point of view and referenced information. Here we are discussing which features we want for us, and which tool could be the best for us.

Well, StackExchange sites, IMHO, do just that as well. They use CC-BY-SA too.

The (big) only problem is that either we create a StackExchange site too (mediawiki.stackexchange.com?) or we use a similar software and do our own (maybe with AskBot, which seems the best software available, and very similar to SE software).
It would be interesting to know for example how the Ubuntu community decided to go the former way: http://askubuntu.com.

The (big) only problem is that either we create a StackExchange site too (mediawiki.stackexchange.com?)

This proposal was already linked from the task description and failed.
https://web.archive.org/web/20140802055511/http://area51.stackexchange.com/proposals/49276/wikis

or we use a similar software and do our own (maybe with AskBot, which seems the best software available, and very similar to SE software).

Yes, it looks like it is. Please edit https://en.wikipedia.org/wiki/Q%26A_software#Comparison_of_Q.26A_software to confirm. There are some questions I didn't find answers for yet: http://askbot.org/en/users/2418/nemo_bis/

In T31923#1369902, @Tgr wrote:

In T31923#1368606, @Nemo_bis wrote:

I don't think it's useful to turn this task into a description of standard Q&A software features.

If we are seriously considering using Flow as a Q&A tool, then it certainly is.

There are no short-term plans for Flow to support a dedicated Q&A mode. If you think it should, please discuss it with us (the Collaboration team), including the product manager (DannyH).

Looks like this has stalled. What would be the next step step forward? Pick one (or more) Q&A engine and install it on Labs for testing?

If someone wants to give Phabricator's Ponder a shot, we could enable it on a Labs instance...

Ponder is only considered a prototype by its developers, it's not ready for testing. AskBot would be the logical option.

Are these hard requirements?

• Wikimedia login
• Multilingual

PS: Ponder has been available for a long time at https://phab-01.wmflabs.org/ponder/

I would say multilinguality is not, given that none of the existing alternatives support it, no one else does it (Stackoverflow experiments with multilanguage but in a rather limited way) and it's not obvious whether it would do more good than harm (fragmentation of content, duplication of effort etc). Especially if we are primarily thinking about a tech-oriented Q&A site (as opposed to a helpdesk-type thing), it is hard to imagine that reading English would be problematic for a significant fraction of the userbase.

Global login should be a non-negotiable requirement but it is an easy one (pretty much everything supports OAuth these days).

Ponder is clearly not useful at this point (major functionality like commenting is broken, it looks rather ugly and the search sucks horribly even by Phabricator standards).

pretty much everything supports OAuth these days

According to our current information, only AskBot does. Please edit. :)

Let's say the English version is a great success and now the German speakers also want one. We wouldn't be able to offer a single multilingual instance, but we would still allow (encourage?) the creation of new instances in other languages, right?

In T31923#1574494, @Qgil wrote:

but we would still allow (encourage?) the creation of new instances in other languages, right?

I don't see a reason why not. But remember, that the points, @Tgr already mentioned, apply for this solution, too:

• fragmentation of content
• duplication of effort
• more complex structure (as a questioner: Where I have to ask my question, maybe there are different structures/tags in different language versions; where I get the best visibility for my question, in language xxx or yyy?), while a Q&A should be as easy as possible (at least for the questioner to ask a question)

So, I think, that there must be a huge community demand for a Q&A site to create a new language version :)

StackExchange is apparently planning to expand into an unofficial documentation HUB.

Just wanted to leave a pointer to Wiki Edu's Q&A page (based on Askbot).

I think, the first and best step would be to talk in a session and find out the top 3 or max. 4 Q&A software, that would fit our needs and after that, install a demo system for these 3 products and test them to find out, what seems to best fit our community,. needs, and technical effort we're willing to spend :)

What about organizing a Tech Talk where drivers of this task would explain the idea and advocates of each option would demo the software? If you find this idea useful, please create a subtask about the Tech Talk and we will help organizing it.

Related: even if we said that Discourse is not a Q&A system, maybe it is worth noticing that as a result of a discussion in wikimedia-l, now there is a Discourse test instance in http://discourse.wmflabs.org/

If talking is a way to extract information from people allergic to writing, sure. However I'd expect us to rely on software whose functionality is written down somewhere, rather than based on grapevine.

In T31923#1959888, @Qgil wrote:

Related: even if we said that Discourse is not a Q&A system, maybe it is worth noticing that as a result of a discussion in wikimedia-l, now there is a Discourse test instance in https://discourse.wmflabs.org/

Gràcies Quim. Issues with the **https://discourse.wmflabs.org/** installation are tracked in T124690.

@DannyH , you added this task to #community-wishlist-survey but I cannot find a related entry at https://meta.wikimedia.org/wiki/2015_Community_Wishlist_Survey/Results. Could you or someone else update the description with the information, please?

Ah, thank you. And yes, I should just look up the task number.

I don't see this Q&A space having a clear relation to "a better way to capture ideas from the community for things like this document [the Community Wishlist survey] and RFCs". Q&A is Q&A, we are having a problem of scope already, and if we try to include RfCs and surveys too then probably this task will become more of a mess.

I think the connection between this task and that CW proposal should be removed (and anyway, it got zero votes).

@Qgil:

This may have been one of the proposals where I created a ticket, and somebody else told me to connect it to this ticket instead. If it's not actually connected, then I agree, let's disconnect them. :)

I don't see this Q&A space having a clear relation to "a better way to capture ideas

The relation is that Q&A systems are forum systems which typically have a feature to upvote questions, similar to IdeaTorrent which is the platform we tested some years ago.

Leaving a note here just to find http://ieeexplore.ieee.org/document/7202966/ again:

The analysis shows that for all four software projects in this study, two of the desired quality indicators, developer participation and response time, did show improvements on Stack Overflow as compared to mailing lists and forums. However, we also found several projects that moved back from Stack Overflow, despite achieving these desired improvements.

I was thinking about this just today. I was wondering if it's even possible to think to create a StackExchange website (you can propose one): it won't be on our servers, but AFAIU there is a precedent, with the AskUbuntu community. Alexa ratings for stackexchange sites are good, and we could even think of having a multilingual portal. In my small community (italian wikisource) we don't have time, competence and will to develop good Help pages: our documentation is messy, obsolete and newbies totally get lost. They don't understand at the beginning that the social rule is to ask other users: they get frustrated and don't have the courage to ask in Village pumps, publicly, what they may think as "dumb questions". Maybe a Q&A website is what we need in our community, and could help also different language community to understand each others tools, workflows and practices.

Wiki Ed has been running a Quora / StackExchange-type Q&A site using AskBot for the last year or so at http://ask.wikiedu.org

It's far from perfect, but it's served our needs pretty well and it works with Wikimedia OAuth. https://github.com/WikiEducationFoundation/ask.wikiedu.org

As long as the description says

Target audiences

To be defined (...)

progress on this discussion will be difficult. The description focuses on editors, so I am removing Developer-Advocacy. If someone is interested in discussing this feature to support developers and other technical contributors, then we can bring it back.

In T31923#1657275, @Tgr wrote:

StackExchange is apparently planning to expand into an unofficial documentation HUB.

Which apparently did not work out.

This ticket seems to be a split between T29852 and T155678 and I don't see any aspects not already covered by those two other tickets instead. (Someone feel free to disagree.) So which task(s) to merge into which task(s), to have discussion more centralized? :)

This is a (somewhat) specific implementation proposal for T155678: Provide an easy to use support system for contributors to ask technical questions (I don't see much relation with T29852, Q&A and issue reporting are different things). Arguably a bit premature, but having subtasks for specific proposals helps structure the discussion IMO.

I remember the posts above saying that Discourse is not a Q&A software. Still, the more I know about it the more I think we could use it as such. Discourse's features out of the box already allow for decent user support (selection of a reply as answer to the topic question, conversion of first posts into wikis, highlighting most popular replies in long threads). I haven't investigated plugins, maybe the multiple demands one can see to bring Discourse closer to a Q&A system have resulted in some code (or we could write/commission a plugin if we need some feature so badly).

Discourse has a potential to solve other problems to other people in Wikimedia. Just like the choice of Phabricator was more than a simple replacement for Bugzilla, the choice of Discourse (and the investment in Discourse knowledge for various purposes) can be more useful than choosing a specialized Q&A tool only useful as a Q&A tool (see related T155678#3728874).