topnav

Series Info...Trials, Triumphs & Trivialities #187:

Social Software & Gaming: Dunbar's Number

by Shannon Appelcline
2006-04-06

For the last week I've been spending some time over at the RPGnet forums. They're a big responsibility, and sometimes they're also a big effort to maintain.

RPGnet, you see, hosts one of the top 100 forums on the net, at least according to Big Boards. Well, its size actually depends on how you count it. By posts, a view of longevity, it's #105, but by traffic, a view of activity, it's #80. By total members it's a dismal #838, but by average online users it's #54 and by max online users it's #39. Suffice to say, it's a big board.

Now, back in TT&T #163, Social Software & Gaming: User Content, I talked about several types of social software and in the processes highlighted forums as one of them. At the time I mentioned issues of "Dunbar's Number" and signal-to-noise ratio, and said that they might be subjects of their own articles. This week's work on the RPGnet forum has got me going down that road, so I've decided to discuss the topic further in this article.

Dunbar's Number & User Groups

Christopher Allen writes about social software quite extensively in his blog, Life with Alacrity. Perhaps the topic that he's covered most is "Dunbar's Number". You can find an overview, a discussion on Dunbar's Number & World of Warcraft and a followup to the same in his blog.

The basic idea is that there is a cognitive limit to how many individuals you can have meaningful relationships with. After doing primate research Dunbar predicted that the mean group size for humans, based upon this cognitive limit, was 147.8 members, or about 150. That number has become known as Dunbar's number.

In Christopher's articles he looks at guild sizes in games like Ultima Online and Worlds of Warcraft and finds that they generally conform to this rule. In fact UO showed even lower numbers, with a mean group size of 61 and extremely few groups larger than 150. WoW showed even lower numbers than that, possibly due to differences in reporting.

Since a lot of these examples already center on gaming, I trust the usage in game design is obvious. However as I've written, last week I was thinking about forums, and I'll get back to the Dunbar Number's role in that medium momentarily.

Other Social Problems

A very busy forum can also highlight other issues with social software.

Information overload is probably the largest, or to put it in forum terms "scroll". The problem is that as a forum gets busier, the information content eventually becomes too large for most readers to take in. Topics that they were interested in go by too quickly, swallowed up by everything else that's being discussed. The most serious board readers don't notice a problem, but casual readers surely do.

Some of those casual readers probably see this as a bad signal-to-noise ratio, though that's not it exactly. It could be that all of the information is good, but if a reader is only interested in 10% of it, he finds his interests quickly buried in the other ninety.

Information overload can also be worsened by poor topic overlap. If a forum contains two topics that are sufficiently unlike each other, a casual reader may not be able to understand half of the labeling or subjects in a forum, thereby worsening his apparent signal-to-noise. For example last week I was looking at topics in a general gaming forum all labelled "[Oblivion]". I had no idea if it was a board game I should be interested in, or a computer game that I probably wouldn't be. (It turned out to be neither, but instead a smart looking computer RPG that I'd play if I had the time.) In any case, where a small forum wouldn't have been too impacted, for a large forum where 10 of these mysterious threads might pop up in a day, the problem becomes increasingly large.

The RPGnet Forums & These Issues

Cut to the RPGnet forums. On a good day they'll have 800-900 readers viewing them simultaneously and half of those users can be jammed into a single forum, called "Roleplaying Open", the core forum of the site.

It's not 100% clear how Dunbar's Number applies to a forum environment, where many users are lurkers. However I think it is clear that a large forum (by which I mean an individual "folder", not necessarily the entire forum system) is going to hit that limit. As I write there are 1005 active users on RPGnet, of which 369 are registered users and 636 are guests. 344 are currently viewing RP Open. Assuming that just the registered members are likely to impact community size limits, that means that there are 127 community members at this second.

Surely, not all of those are active posters, but on the other hand there are many other readers who aren't logged in at this second but will be sometime in the next 24 hours. Turning back to Big Board's stats, we find approximately 8,500 unique visitors every day, or an estimated 3,120 registered unique vistors on a daily basis. Even if only 35% of those folks visit RP Open (and evidence would suggest this is an extremely conservative number), that means that we have a community size of just over 1,000 for RP Open on any day. Unless 85% or more of those registered members are lurkers, we're over Dunbar's number, and thus approaching dangerous grounds where communities start to break down.

Looking at the other criteria I examined, the problems of allowing a community to reach this size are very obvious. At the moment 50 posts (a screenful) have been updated in RP Open in the last three hours. An average day easily spans 200 posts on RP Open. The tabletop RPG community is small enough that I suspect most of the tags and subjects make sense to most players; however that may not effect indivudals' perceived signal-to-noise ratio. When a single topic (currently, the role-playing game, Exalted) becomes popular enough, it can suddenly take up half the active threads, giving everyone not interested enough in that topic that much more to weed through.

None of these problems are necessarily show stoppers. Before we started looking at them last week, the RP Open forums did function, and they now continue to do so. However, these types of problems will impact growth, because the relative cost of adding a new member becomes higher than the community can sustain.

In addition they have the long-term danger of isolating the community. If only the most serious community members can participate in the community, and that bar for entry continuously goes up as posts and scroll speed continue to increase, then the community will become that much more serious, and ultimately entirely isolated from more casual participants.

The Problems with Change

The obvious solution to this problem with forums, and similar issues with many other social communities, is to split them, which is to say, take individual folders and divide up the topics they contain. Undoubtedly you'll have some members who will want to continue participating in both split communities, but not all of them will, and thus there's that much more room for growth before you hit Dunbar's Number (again).

Unfortunately the increasing specialization of a community that comes about as it butts up against Dunbar's Number may actively hinder exactly this sort of change. Innately any community is going to be self-selected for those members who like the community as it is. That means that if you ask participants if they want a split, they'll probably say no, because they're the people who can deal with the community as it is. Conversely if you tried to ask people who rejected the community because of these problems if they wanted a split, they'd probably say yes.

Ultimately neither type of person is necessarily going to necessarily be able to give you the "right" answer, because the current community members are unlikely to consider growth and outreach and people who have rejected the community are unlikely to consider what makes it great right now. The problem thus falls squarely on the shoulders of community administrators, who have to try their best to measure the interests of both groups, and thus the potential community as a whole.

Changing RPGnet

Ultimately we punted on an RPGnet split. The possibility got to being discussed in the forums, and it turned into a whole big thing, with everyone spinning out their worst-case scenarios.

Rather than a larger split we ultimately added just one new forum, which we called "Roleplaying Actual Play". It doesn't actually split out much of the content from the original RP Open, and thus isn't likely to drop community size back down to a reasonable number, but it does free up one popular topic of discussion to form its own community, well below Dunbar's Number (though hopefully not so far below that it won't be self-sustaining).

As a result we're probably going to have to revisit this whole topic again, but only after the new subforum suceeds or fails on its own.

One person discussing a possible split asked, "Can you point out an RP forum boards where this particular type of split was successful?" Of course I couldn't, because RPGnet is the leader in the field, outpacing every indie RP board out there.

And when you're the leader in a field, you're going to have to take chances, or else you stagnate and die. The Internet is a shark-eat-shark world; there's always some 15-year-old with a Tandy and a 2400-baud modem ready to start up his own website which will ultimately consume your own.

We didn't take our big chance this time, but we started the gears going to show what a split could look like, and next time we may need to take that step.

Conclusion

Managing communities is tough. Just dealing with a pure social software construct like a forum is hard enough, but when you wrap whole game systems around it, the whole situation becomes even tougher.

When you're running your own games, try and keep in mind the various moving parts. There's the game, but there's the community too, and each may require very different methods for maintenance.

[ <— #186: Online Games & The Law, Part Five: Property Rights | #188: Downtime & The Online Game Designer —> ]

Miniseries ...