# Shall we have tea?

As was recently discussed at Should meta be Q&A?, it's rather strange to have our meta site run on a Q&A basis.

In the past our meta site was a Vanilla forum, and in fact it still exists, renamed to http://mathoverflow.tqft.net/ at the transition to Stack Exchange 2.0. (Interesting trivia --- we had a meta before any other stack exchange sites besides meta.stackoverflow.com!)

Some of us (me, in particular) have always been dissatisfied with the replacement of tea by our current meta. It seems we used to have in depth conversations discussing the community and future of mathoverflow, which are now significantly obstructed by the poor software choice at meta. (But perhaps I'm just being nostalgic; hence this post.)

The folks at Stack Exchange and Discourse have kindly offered to set up a Discourse installation for us, probably to be hosted at http://tea.mathoverflow.org/. Discourse is modern forum software, developed by Jeff Atwood and co, with close links to the Stack Exchange team. There are limits to how far we could integrate it with mathoverflow itself, but here's the rough idea:

• Discourse would support essentially the same set of login methods as mathoverflow, but we would probably not be able to directly link accounts.
• We would add a custom close reason on meta, asking that the conversation be moved to tea, and would try a lot, perhaps even nearly everything (besides bug reports, login problems, and so on, that Stack Exchange staff need to be able to monitor), over to tea.
• We can explore the possibility of using our custom javascript footer allowed by our migration agreement to add notifications on mathoverflow and meta.mathoverflow about new posts on tea. (No promises here, however --- if anyone with some javascript skills wants to help me explore this, contact me directly.)
• I think independent of what the exact format for meta is, it is important that it is integrated well with the main site -- i.e. in particular that no separate account is needed. Also I think meta should support a means of expressing agreement / disagreement in the form of votes as it does now. – Stefan Kohl Dec 1 '14 at 23:31
• I disagree with the need for a voting mechanism, although Discourse does provide a "liking" mechanism. I would be sad if the desire for complete login integration prevented us from even considering other options. It is possible that Discourse's "single-sign on" plugin would allow authenticating via MathOverflow, but I'm not sure yet. – Scott Morrison Dec 2 '14 at 2:41
• IMO login integration is not that important per se, but what is really helpful on the current meta is network-wide reply notification. – Emil Jeřábek 3.0 Dec 2 '14 at 13:20

Although I am not so much involved on meta, I prefer the current Q&A format over the alternatives. Let me try to convince you that our current system is a good one.

The main reason, for me, is that the Q&A format tends to lead to a relatively focused form of discussion, usually concentrated on the main relevant issues. In posting an answer, a participant is forced to articulate the best grounds for a given perspective. A user must make their best case. Thus, we tend to find essentially one or just a few well-written posts for each major position on the issue, and one can sensibly compare them. I find this more useful than the kind of back-and-forth discursive debate that one finds on a discussion board.

The feedback resulting from voting only adds to this, since not only can one see the various main positions, but also one can get a sense of the community on those issues. Thus, the voting feedback is an important secondary reason to keep the current meta.

The old discussion format, in contrast, tended to amplify the views of those who post loudly or often, rather than those who post well. I think that wasn't good for the community, and I personally didn't find it as useful as the current system.

So let's keep the current format, which I feel encourages people to find and articulate the best supporting arguments for their position, and which helps the community stay focused.

I am against the proposal in its current form. I am not against it in total, but I am against one specific point, namely:

We would add a custom close reason on meta, asking that the conversation be moved to tea, and would try a lot, perhaps even nearly everything (besides bug reports, login problems, and so on, that Stack Exchange staff need to be able to monitor), over to tea.

This is in my opinion a problematic idea. It is one more thing that is not complicate but not immediately obvious, and it are such small things that can alienate users and lead to friction. It is already not straight forward to keep a relevant part of the community engaged with meta; this split will make this even harder.

Indeed, I feel the tighter integration of the current meta with the main site is really helpful.
The old meta did confuse some users, and in later days of MO 1.0 it was quite idle, and if not idle, the number of active participants was really low.

I could imagine to have a second board to complement the current meta, indeed, we have one (the old one), and it was essentially said when we moved into the SE network, we will continue to use it for some things (and I argued in favor of thus), but just never did. However, this should not interfere or replace the current meta, but rather be something in addition.

If somebody wants to start a discussion on tea, they are welcome to do so. But, discussions should not be moved (with force) from here to there. To inform users of new discussions there we could simply have a meta question: "What are current discussions on tea?" and a new discussion can be announced via an answer. This is very easy to do; why not simply try it.

Whether we should change the software from Vanilla to Discourse, I do not know. Maybe let us see with the board we have if activity there regains some traction and let us optimize the technicalities if or when it works. (Also, what about the current "tea"; it seems we would still need to preserve it.)

Finally, if this is mainly for communication moderators to community, a moderators' blog could be more apt (this idea came up frequently). A mailing-list as proposed in comments could also work, though it seems a bit old-fashioned to me.

• I'd like to add to your paragraph 4 in that the polished user experience of the current integration between main and meta probably goes a long way into more participants at meta, not the least of which is the same software being used on both. – Jason Polak Dec 3 '14 at 15:38

I also miss tea and I would like to have it back for discussions. However, tea was really bad for creating authoritative statements of policy that we could link to later, and meta is really good for this. Also, as JDH says, it was bad that it amplified the loud-mouthed (like me).

Here is a proposed compromise. Keep tea running and find a way to promote it more visibly, but only as a forum for loose conversations. Once a concrete policy has been drafted, bring it over to meta to create something specific to be voted on and linked to.

• Do you consider the following as insufficient to promote the other board (if so, why)? 'To inform users of new discussions there we could simply have a meta question: "What are current discussions on tea?" and a new discussion can be announced via an answer.' [Sorry for the second comment I thought the joking start might have overshadowed the question.] – user9072 Dec 4 '14 at 14:08
• @quid Just that life overshadowed. I think that is very good solution. – David E Speyer Dec 4 '14 at 14:14
• That's great. And sorry if I was somewhat impatient here; let me also blame life for it. :-) – user9072 Dec 4 '14 at 16:47

I want to respond to Joel's post, but I'm unclear whether I should be doing that as a comment or an answer. :-)

My main concern is that while we used to have conversations about the future of mathoverflow in public (i.e. at the old meta), they now largely happen on the private moderators' mailing list.

I would like to open these up to the community again.

• This is a bit misleading since very little happens on the so-called "private moderator mailing list". – François G. Dorais Dec 2 '14 at 13:19
• I deleted an earlier more playful comment (mainly as a half-sentence of mine was based on a misunderstanding), but I would still like to record the specific question: why exactly do you feel that you cannot post these thing now, neither here nor on the old meta, but believe that then on the third meta it will be good? – user9072 Dec 2 '14 at 13:56
• @quid, let me guess: Discourse has not just a discussion thread feature and a search feature and a login feature like Vanilla, but also has one or two features in common with this meta that support such discussions. Hopefully Scott will make those features clear. – The Masked Avenger Dec 2 '14 at 16:30
• @TheMaskedAvenger according to OP 'we used to have in depth conversations discussing the community and future of mathoverflow, which are now significantly obstructed by the poor software choice at meta.' These discussions happened on that very Vanilla board. I do not see any reason why they could not continue there. – user9072 Dec 2 '14 at 16:51
• I also think the discussions could continue on Vanilla. However, Scott's point is that the tie between the Vanilla board and the MathOverflow forum is weak, and there would be more of a challenge to notify users of changes on Vanilla, and less so (or differently and in a better way) using Discourse. However, I could see a determined group using Vanilla and updating members of this forum. Gerhard "Has Feet In Both Camps" Paseman, 2014.12.02 – Gerhard Paseman Dec 2 '14 at 16:58
• @GerhardPaseman in my answer I propose a very straightforward way to notify users of new posts that I assume will be somewhat rare. As I said in a comment that I then deleted as it was somewhat redundant with JDH's answer the lack of discussion is in my mind not mainly due to the format. Anyway, there is nothing much to be said against the board using a more advanced software. However, if noone has anything to discuss it will be a bit pointless. What would you like to discuss? (I am sorry to say but most of the more recent general debates did not really lead anywhere AFAICT.) – user9072 Dec 2 '14 at 17:14
• @quid: in brief: additional promotion efforts, outreach to mathematics in developing countries, considering an adjunct to WDML, and most importantly what to do to grow the current community. Also, some previous discussions on archiving and content need to be continued. – Gerhard Paseman Dec 2 '14 at 17:22
• Okay @GerhardPaseman, what prevented you from starting the discussions here or on the current tea? – user9072 Dec 2 '14 at 17:23
• Partly trying to fulfill some earlier committments. I am willing to attempt it here because I have a sense of who is here to read it. I might try it on tea or a new forum, but I will talk to an empty room only so long. As commented elsewhere, it is important to know who (or at least how many) is listening. Gerhard "Running Short On Signatures Today" Paseman, 2014.12.02 – Gerhard Paseman Dec 2 '14 at 17:27

Who else is going to be there?

I think the current meta (I call it 2.0) for short is well suited for a number of issues pertaining to day-to-day operations and to the SE interface, and 2.0 is ill-suited for the long term discussions that need to take place to keep MathOverflow the idea going. However, if there is only a small group interested in using Discourse, then I recommend using the old Vanilla forum, with someone updating 2.0 periodically with a summary of goings on.

EDIT Having seen the Discourse page, I can see where it could lead to the kind of brainstorming and collaboration needed for the future of MathOverflow. Again, I think it is important to know how many will participate. The forum publishing.mathforge.org of Andrew Stacey was and still is a good idea, but it seems to have very little traffic nowadays. MathOverflow cannot enjoy its current success without continual efforts to moderate, promote, evolve, and discuss. I know I don't want attempt such discussions on this meta.

• Part of my concern is that the only place those "long term discussions" take place anymore is on the private moderator mailing list. I think this is unfortunate for several reasons: 1) We often have bad ideas, which would be shot down more quickly in a wider forum. 2) It makes it hard to recruit new people who take an interest in long term planning for mathoverflow, because the conversations that are actually happening are invisible. 3) It's just not transparent. – Scott Morrison Dec 2 '14 at 2:35

Thank you @Scott for letting me (and others, I guess) know about http://www.discourse.org/ (discourse). I glanced at it, it looks attractive, seem to have new features, etc., however I am not an expert. Thus I want to propose an intermediate project MetMO_tst--based software-wise on discourse--which would be a stepping stone possibly leading toward a better Meta called ultimately MetMO. Thus MetMO_tst would be a simulation of the future MetMO:

1. A team of $2$ to $6$ people would create, own and manage MetMO_tst based on discourse, which would simulate the possible future MetMO.
2. For the sake of the software and otherwise simulation, the team members would perhaps have more than user ids, to imitate having more users. Etc.
3. There would be some other supporters who would simulate regular users to assist the test project. Call these other participants consultants. Consultants would be admitted only when invited by the managers.
4. The simulation, when the time is ripe, would allow the managers to decide about starting or not MetMO. Perhaps first a discussion can take place on the present Meta and Tea, and MetMO can be open for reading only, to present the results of the simulation, and the ideas for the near future.

DISCLAIMER   Just in case, I am not going to be a MetMO_tst manager. Possibly, I could be among consultants but I am not too much interested one way or another. However, I really would like some people, including @ScottMorrison, to create the simulation--MetMO_tst.