Is SE 2.0 scaling well, or do we have too many mods?

Question

So I just want to take a look at a simple and basic question, but one that is nonetheless very, very important: How well is the SE 2.0 model scaling?

I don't mean scaling on a technical level—that's fine AFAIK. What I mean is—SOFU were all successes in their own right. Are the new sites getting there as well on the community level? More specifically, do we have too many mods?

Currently, SE has 216 mods, and that number is growing. Since each site gets ~3 new mods, and there's no site limit, that number is going to continue to grow and grow and grow. That's a lot of mods.

Stack Overflow currently has 10 moderators and is in the process of electing 4 more, for a total of 14 mods, all of whom together will handle:

• 859k users
• 2.7m visits per day
• 4k questions per day (not posts)

• 1k flags per day

The most successful sites, depending on various metrics, are Ask Ubuntu and Programmers. The latter is special, so we'll look at AU, which is a more typical SE site.

• 26k users
• 61k visits per day
• 126 questions per day (not posts)

Mathematically, AU is about 3% of SO, yet has 21% the number of mods. If we were to look at this purely by the numbers, either AU only needs 1 mod or SO needs 100 mods.

I don't have AU flag data, but I can say EL&U would find even 30 flags a day to be high.

Obviously, neither is true—no site can survive on 1 mod simply because every individual mod will be away for various lengths of time (sleep, travel, etc.). With three mods, if one is away, the other two are there to deal with issues. As far as SO is concerned, it should be readily apparent that SO doesn't need 100 mods.

But that doesn't mean that we need 216 mods for 71 sites. We shouldn't even have 71 sites—I've posted an answer below to explain why, but it's really a tangent. I suppose the reason why I dislike having too many mods is that we're going to reach a point where every active SE user is a mod on some site, which means we're all policing each other.

Proposal: This isn't fully fleshed out, but why not (from here on out) elect cross-site mods for the smaller SE 2.0 sites? Say 6 mods across 3 related sites—the flag load can be easily handled, mods are on more (a single interface would be nice), mods would be more cognizant of the scope of related sites, and then we don't have this huge number of mods.

I won't specify what a related site is, but Lit + SFF come to mind as being sites that could use cross-mods, maybe even EL&U + Writers. Programmers and Gardening obviously aren't related.

Or is having this many mods not a problem? Or are the logistics going to be a problem (sites don't graduate at the same time)? Or anything else?

N.B. I don't think it's really relevant, but I suppose I should insert a disclaimer that I'm a mod for EL&U.

Answer 1

The idea that three to five people per site moderating is too much is patently absurd.

On the most popular sites—Ask Ubuntu, Gaming, Apple, Android, and Programmers—moderators account for less than a few hundredths of a percent of the total user base. On a site like Programmers, 45 times more new avid users (200+ rep) then there are active moderators are added every two weeks. Even on our least performing sites (e.g. Literature), regular users outnumber moderators 197 to 1.

A site like Programmers also generates 1,000 flags a month: about 10 flags a day for each active moderator. Combined with staying active on the site's meta, TL, and here, there's a lot of time spent per day for something that's entirely volunteer. Throw in the massive amount of effort by Stack Exchange to streamline and improve moderator tools for Stack Overflow and the other Stack Exchange sites because the moderator load has been too high, and the conclusion one ought to reach is that we need more moderators, not less.

And beyond that, each moderator is expected to be completely versed in their own site's activity and policies and be trusted members of the community: we're entrusted to act with a large amount autonomy and privileges because of that.

You say the solution is to start combining moderators: beyond the work each moderator does on their own site, they need to do it on other sites where they aren't trusted or possibly even active. How do you expect that to go? Everyone gets double the work and half the trust they have now, for what? Because you think 200 moderators sounds like a lot of people?

The overhead for having a few moderators per site is offset many times over by how it makes it so much easier for everyone that there are people willing to take the time to get heavily invested in a specific community to help guide and moderate it. I don't know why you'd want to look a gift horse in the mouth like that.

As to the suggestion that there are too many sites: on what basis? Have you been to Area 51 or any of the other sites? Each site was created because there was enough critical mass to create a specific audience about a topic. Most sites have been able to get traction precisely because the experts for those sites would not or could not contribute to a more general site. Most sites on the network are active and have found successful niches that are working for them with hundreds of questions and thousands of users.

Consider the person who looks at the price of an economy car and sees $9,000. He says, "Whoa, that's a lot of money! Why does it cost so much? I don't want to spend that much." Costs so much compared to what? Other new cars? No, that's on the low end. Other modes of transportation? Sure, but you don't get the benefits of the car. You want to pay less, you don't get the car.

In the same vein, 200 moderators and 70 sites sounds like a lot when they're taken completely out of context. But if you look at what's been built over the past 2 years, it's a small price to pay for what we've gotten. You don't want this many sites and this many moderators? You clearly don't want what makes Stack Exchange Stack Exchange: a network of niche sites full of topic-specific experts.

Answer 2

If you take a look at Mark's excellent list of moderators you will see that there are 28 moderators who moderate on 2 sites and 2 who moderate on 3.

Looking at the nominations for Stack Overflow elections I can see that 3 of the candidates are already moderators on other sites.

Therefore there's no need to artificially limit the number of moderators - it's happening organically. No one has told anyone that people should be moderator only only one site, so people will volunteer to do the job on other sites if they feel they are able to do it.

Answer 3

The reason why a number of sites are small right now is because the site hasn't grown beyond programmers. The majority of users have a SO account, or friends of a SO user. The fact is, if SE originated under a different topic (such as English language or religion) you would be asking, "Why is the programmers site so small?" Give it time and the sites will grow.

Merging sites isn't going to work. How many Christians are also Muslims? A much smaller percentage than C programmers who know Pascal. There also wouldn't be any more activity going on, only an illusion.

Sites need a minimum number of moderators. This is not because there is more moderation to be done, but rather it is good to have a mod online at all times of the day. It certainly discourages, "Mods are asleep ask subjective questions!" Also keep in mind that users do the majority of moderation work - 10000 users is potentially 10000 editors, flaggers, and reviewers.

I suppose the reason why I dislike having too many mods is that we're going to reach a point where every active SE user is a mod on some site, which means we're all policing each other.

What's wrong with this? The number of people suitable for moderating is much higher than the number of mods; it's not like we're running low on quality.

Answer 4

This post picks up a tangent of the main question. I wrote it up before realizing that it's really its own thing.

For that matter, we don't need 71 sites. Let's take a quick look at some of the merges that are theoretically possible.

The following lists are only hypothetical. If we were to consider a merge, I believe that a referendum should be put to all affected sites, let alone any problems that we may face with each particular set of sites.

• Religion and Spirituality, incorporating Christianity, Jewish Life and Learning, Islam (A51), Hinduism (A51), Buddhism (A51), the closed Atheism, and all other religions
  • Why? Think of why SO succeeded—C and Pascal have little to do with each other; it's the fact that we let all things code related onto SO that SO succeeded.
  • Downsides: Religions tend to be more controversial, thus leading us to need more mods + people are more likely to ragequit; also, do we admit atheism/agnosticism;
• Languages, incorporating EL&U, French, German, Japanese, and all other languages
  • A lot of languages share roots
  • Might allow limited translation to thrive
  • Downsides: Do we need mods in every language? We're trying to lower the number of needed mods, or at least cap it.
  • Where would Lingustics & Writers fit in, if anywhere?
  • EL&U is a graduated site—does that change things?
• A bunch of the computing-related sites
  • Especially the programming related (we have SO, Programmers, Theo. CS, Code Review, Code Golf)
• Literature + SFF
• Answers.OnStartups + Project Management
• Math + Stat

I'm not arguing that these merges need to take place, or that they even should (some of them would severely redefine site scopes—what I'm trying to make evident here is that fact that we seem to be creating a bunch of tiny sites rather a few large sites, which is why SE worked in the first place (SO allowed all programming languages, regardless of popularity; SF & SU allowed all platforms).

I'm also not saying that we need to be doing something, but we should definitely consider more site integration. And if any of the sites get near failing, osmosis into another could be a way of rescuing them. Anyway, that's outside the scope of this question.

Of course, if we were to do these merges, than the main question wouldn't really be a problem. So the two topics are linked after all.