# Meta-meta discussion, and voting conventions

Yes, this is a discussion about question on meta :)

We have a lot of discussion questions, but it's not always clear whether the discussion has congealed into an actual policy, or if people just got tired arguing their positions. These discussions tend to live on in limbo, and I guess as a moderator, I'd like to know what the resolution of the issue is without having to wade through long discussions.

Over on meta.SO, there's an interesting and somewhat alarming policy in place in which questions votes are used to express agreement or disagreement, rather than opinions about the well-formedness of the question. Since meta votes don't count except for badges, I was thinking that we could adopt this convention here, and so if you want to express your disagreement with the question, just downvote it.

It's a little disconcerting because on the main site we're used to interpreting negative scores for a question as a major slam on the question, whereas here it would just imply disagreement. But it would provide a quick 'view of community' on individual questions.

• And now we just upvote this question and you have no idea how to interpret it. Dec 7 '10 at 23:57
• @Jukka: :) I am preparing an answer to explain my upvote. Dec 8 '10 at 0:01
• @Suresh: a minor issue: some of us are new to meta, and maybe we do not fully understood the policy about the different meanings of upvotes in parent site and meta. So there must be a noise in the overall voting procedure. Since the votes in meta are rather few, this kind of signal may affect the policy decision. Dec 8 '10 at 1:20
• I usually dowvote answers with which I don't agree (on Meta). I upvote answers I agree with. I upvote questions if they're worth discussing on Meta. Questions don't always propose a solution, so it's not clear what an upvote would mean for such questions. Dec 8 '10 at 1:36
• I've been doing what @Robin has been doing. I have upvoted questions even if I think they are proposing something incorrect, because I'm glad someone at least put something on the table. Dec 8 '10 at 1:53
• I think at some point we should write a meta-faq. :) Dec 11 '10 at 19:30
• We should start a discussion on how we should treat meta-meta questions. And that will be the first meta-meta-meta question. Dec 11 '10 at 19:40
• We lock the policy posts after they are accepted/rejected to preserve the votes and the discussion. If you want to start a discussion about changing a policy please post a new meta-question linking to the original discussion about the policy. Nov 16 '11 at 23:39

Discussion on SE software sucks ! I much prefer the old MO SE 1.0 model for discussion.

Having said that, I think the suggestions about limiting this interpretation of voting to policy questions is a good one. I also like the idea of explicitly requesting votes in favor and against, so that the default state for a question remains as before, and ONLY posts with explicit requests for votes (usually policy posts) have this proviso.

From a moderator perspective, my goal is to act as "policeman", but have the rest of the community write the laws. So getting resolution on policy is helpful to me. It is also helpful to users, because then we can cite specific policies, so that decisions don't seem arbitrary.

As a matter of perspective, I've had essentially zero problems with users. So the problems, if at all, are minor. But people do get confused and disgruntled when policies are not clear.

To conclude, I'll go with the policy outlined above when I'd like a question of policy to be resolved, and people can vote on it. I'll also mark it CW, so that policy changes as per discussion can be made.

• By the way, what was so much better in the MO SE 1.0 model of discussions? If you sort answers by age (click "newest") here on meta, it is a lot like MO meta, except that you can also add comments like this. Dec 8 '10 at 13:48
• the MO model is "flatter": just a thread of comments. The answer+comment mode is not great. Ideally, what a meta site needs is proper discussion threading, or a regular forum. I think this is the case of the SE folks getting carried away a bit. Dec 8 '10 at 19:06

In my opinion, the major problem is caused by the ambiguity of interpreting the upvotes in meta, which may have the following two different meanings:

• the question itself is a good one, we should have discussion on it; or
• I agree on the policy OP suggested, and by upvoting I show my agreement.

If these are the major (if not total) possibility of what an upvote means, then we can separate them by an additional answer added by OP him/herself to his/her own question: by upvoting the original question it means we should have discussion on the issue; by upvoting the separated answer we show our agreement/disagreement to the policy OP suggested.

I'm not sure that this is going to work if there is a third kind of meaning to the upvotes. And we should NOT apply this rule to all of the posts on meta; maybe only on posts with [policy] tags on them.

I have a concern about interpreting highly voted questions/answers on Meta as a consensus because the intent of votes is ambiguous. For example, I have voted up this question because I believe that it has raised a very good point. However, if it is interpreted as a sign of agreement, I have to state that at this moment, my upvote does not mean that I agree to adopt the policy on Meta Stack Overflow of using votes to express agreement and disagreement.

Another issue is that sometimes I agree to someone’s opinion first and change my mind later. Or it can be the other way around. I will feel uneasy if I am aware that I voted the wrong way and my voted is counted as it is. (As you can see, this is not a problem with using votes to decide the policy, but a problem with the current voting system itself, but it might be more serious if we use votes to decide the policy.)

However, I understand that lack of summary and/or conclusion of discussions makes the job of moderators much more difficult. If there is no good alternative, I am fine with accepting the voting convention on Meta Stack Overflow here.

First, I agree.

Now, this makes sense if the question is a policy proposal, but most questions are not. We can use the [policy] tag to formally distinguish discussions on policy/guidelines from other discussions.

In some cases there is no consensus, sometime people have opposing opinions even after discussion, so it might be more reasonable to consider those cases where there is many votes up and down votes as disagreement rather and adoption of no policy rather than adoption of one of proposed ideas, i.e. adopt a proposed policy only when there are enough up votes (above some threshold) AND not many negative votes.

Addition: To address the point raised by Hsien-Chih Chang, I think we can explicitly state in the question that the votes for that question or its answers mean, something like:

This is a policy question. The proposed policy in the question (or answers) will be adopted in the case that after W days it receives total vote count X AND the number of negative votes is not more than Y.

I think something like 4 for X and Y and 14 for W are reasonable numbers (at least for now).

• Explicitly state the convention in [policy] post is nice. And the numbers are reasonable to me. Dec 8 '10 at 1:41

Here is what Suresh added to the meta discussion “Policy: Questions about topic for research project” started by Aaron Sterling:

This is a policy question. The proposed policy in the question (or answers) will be adopted in the case that after 3 days it received at least 5 votes with a majority in favor. Please upvote or downvote this question based on your opinion of the policy statement as presented (not based on whether the premise of the question is reasonable)

While I agree on the general principle, I do not agree on the implementation because “at least 5 votes with a majority in favor” is too weak. In my opinion, most policy changes will not be urgent, and therefore we should adopt a new policy only when it is highly supported. A proposal with 3 upvotes and 2 downvotes would not be highly supported. Therefore, I think that we should require at least 5 upvotes in net, that is, (# of upvotes) − (# of downvotes) ≥ 5. If 5 sounds too large, it is negotiable.

• I think something like at least five upvotes, and a 2/3 majority of upvotes to downvotes would be even more appropriate. I think we only want policies that are very close to consensus. Otherwise 49% of the active users just won't enforce them, and grouse. Imagine if the discussion on homework had been split 55-45. Bleck. Dec 12 '10 at 3:44
• I'll amend it accordingly. I was a bit confused myself about the original wording. Dec 12 '10 at 3:53

We should separate discussion and vote even if we have to use the same software for both. Currently, we have discussions via answers and comments. While not ideal, it can work.

I propose that after a time of discussion (i.e. when no more new views are added) somebody, ideally a moderator, creates $k$ new answers for $k$ possible outcomes and closes the question. People are then to vote on those for aggreement (upvote), indifference (no vote) or disagreement (downvote) in a given timeframe.

One could move this vote to a new question (title prefixed with [Vote]), providing a clean statement of question and answers well as an RSS push. In this case, people who are not interested in discussing can immediately see a vote coming up. Important votes could also be linked in a notification box on the main site.

I'd like to add another addendum. once a policy has been sanctioned, it cannot be revoked for at least X days, where X should be at least 7 or 14. This is to allow a cooling off period before the disgruntled minority attempts to reverse policy. Thoughts ? or is this too excessive.

• You are very optimistic about future participation on meta! :-) Dec 12 '10 at 20:17
• More to the point, I think if there's a significant disgruntled minority about a policy (one or two people who are loud is not what I mean; a chunk of respected users who are likely to leave and never come back is what I mean), then something went wrong somewhere. I'm not opposed to your suggestion, but I think, if the situation were to arise, it would be like putting a band-aid on a problem instead of addressing it. Dec 12 '10 at 20:20
• fair enough. I was a little concerned about that myself. Dec 12 '10 at 20:41