[MUD-Dev] Raph's collection of MUD design Laws
Kristen L. Koster
koster at eden.com
Sat Feb 12 23:24:49 New Zealand Daylight Time 2000
on 3/13/2000 1:45 AM, Greg Underwood wrote:
> Raph must have put up his page-o-laws whilst I had traffic turned off, so
> the first I saw the page was just this week.
Wow, you must have been without traffic for a while, considering how often
they get mentioned... :) In any case, as an FYI to the group--I am once
again giving a talk on the Laws at the Game Developer's Conference this
year. It's March 9th through 12th in San Jose. It's far from the only
mud-related thing at GDC this year, either--off the top of my head, I can
* an Everquest case study by Brad McQuaid
* A tutorial on online community building. I'm on it, Mike Sellers is on it,
so is Jonathan Baron, and Brad & also Toby Ragaini of Asheron's Call, and
Amy Bruckman of MediaMOO fame. Organized by Amy Jo Kim.
* Jonathan is also doing a followup to the "Glory and Shame" thing this
year, on how to inject community-generating conflict into your online games
* A study of the economics of UO by Zach Simpson (referenced on this list in
the past, now the data will be public, or more of it anyway)
* A thing on managing ethics in games moderated by Bernie Yee with Warren
Spector (Ultima Underworld, the upcoming Deus Ex, and the original System
Shock) and Toby and I think Doug Church (Ultima Underworlds, Thief, both
System Shocks). I was asked to do this one but it conflicts with...
* A thing on the development process for massively multiplayer games that I
am doing with Rich Vogel (UO, M59).
There's more, I just can't think of it all. The conference website is at
http://www.gdconf.com. It's expensive, but if you go, maybe we can get
> On the whole, I think the
> comment at the top about viewing them as challenges and not rules is most
> excellent. I know I found myself saying "yeah, but if you try this," or
> "well, yeah, but looking at it this way..." an awful lot. ;)
Thank Ola, not me, for that one. :)
> Anyway, forgive me if I'm restating a previously discussed topic, but I
> just had to comment on what seems to me to be an obvious correlation
> between a couple of the points. Specifically:
> A corollary to Elmqvist's Law
> In general, adding features to an online game that prevent
> people from playing together is a bad idea.
> A caveat to the corollary to Elmqvist's Law
> The exception would be features that enhance the sense of
> identity of groups of players, such as player languages.
> Community size
> Ideal community size is no larger than 250. Past that, you
> really get subcommunities.
> Does it not make sense, then, to say that you ought to severly limit (or at
> least very carefully consider) the number of 'group identity enhancing'
> features in your game, until you reach that level of roughly 250 players?
Well, the thing to consider is that the subcommunities often form way way
sooner than 250. They can start forming with just a dozen or so people. I
don't think there's anything wrong with providing tools for even small
groups to make their group identity clear.
That law is really more about the scope for hierarchy in groups of small
size. 250 is the point at which communal consensus forms of policing seem to
fall apart and the position of "manager of community force" needs to be
established (eg, when you need a boss for the group). This seems to be tied
to the number of close acquaintances one can really have (apparently, and
don't ask me where I read this, a given individual generally has close
personal ties and even solid acquaintanceships with only up to a hundred
people or something. Hence forms of social pressure that depend on "family
or friend intervention" such as communal policing break down, because large
groups tend not to be able to produce a friend on the spot when one is
needed to talk or shame the aggressor out of his actions).
> In answer to my own question (and probably what got me thinking along these
> lines as I read those points), I have to offer up my experiences in running
> EoD. Early on, we didn't have too many guilds (the stock 4 or so that come
> with most DIKU's, IIRC). However, we thought it'd be keen to add some
> more, give more people a chance to find their niche, not to mention fit
> more closely with the story (Wheel Of Time) that we were supposedly based on.
But did these guilds prevent the players from playing together? That's
really what the law is referencing--stuff that actually keeps players apart.
A good example would be putting racial enmities that somehow automatically
forced fights or damaged reputations or something when the opposing groups
came into contact. With too low a playerbase, this would be unsustainable.
With enough critical mass, it would be great fun.
> Once we boosted the number of guilds, we ran in to a few problems which, in
> retrospect, seem to be what the corollary to Elmqvist's law is trying to
> prevent. We had about 150 people online during the peak hours. Early on,
> people tended to split out pretty evenly amongst the guilds. After a
> while, once we had a few pretty charismatic leader types in charge of the
> guilds, people tended to group together under the strongest guild, leaving
> the rest as vastly underpopulated. Which guild was "the guild to be in"
> depended on a lot of things... which charater class was perceived to be
> strongest (class didn't determine guild, but the nature of the guilds based
> on the story from the book did tend to determine the class of the members).
Charismatic leader types make the size of the potential tribal structure
larger. But even charismatic tribal leadership has a limit; at some point
the leader must delegate because the size of the tribe he manages becomes
too large for him to personally know and therefore influence enough of the
population under his control. That's what the law is getting at; the
subgroups that form may be around common interests within the umbrella
organization, or may be geographical in nature as is the case in the real
> Now, granted, any community will want to divide into sub-groups, but what
> you call those groups, as well as how many you provide can have a great
> affect on how much flavor they add vs. how much they push at the seams of
> what holds the community together.
The problem is that in muds, your total population generally falls below the
magic number anyway, in terms of constant online presence (by that, I mean
that a playerbase of 250 doesn't count unless they are all on every other
day or so... they are just too inconstant a presence to make them
significant participants). Some interesting examples to look at in games
where the population is so large that this phenomenon is easily observable:
- EverQuest does a nice job of providing many different, overlapping
subgroups. Between classes, factions, and races, there's a shifting network
of subgroups within each given community. Unfortunately, there are not
enough tools for leaders to grow their groups beyond the "tribal" stage.
- In UO we had some success with tying groups to locations, which led to
overlapping groups sharing territory that was of common value. We allowed
free formaltion of guilds and subgroups at any time--there are currently
over 21,000 of them, most pf which are ephemeral and very small. But the
largest exceed the 250 limit handily, via forming multiple guilds with the
same rules (such as a guild for new members, a guild for the "guards," and
guild for the advanced members, etc). Tying people to a location (and also
having them able to run NPC vendors, which made them active participants in
e economy and society even when not logged on, increasing their "constancy")
made a big difference.
- And of course Asheron's Call formally attempts to codify the hierarchy
with its allegiance system, but I don't know how successful it has been at
creating larger-scale communities, or whether the hierarchy is disconnected
enough that members of a common tree do not feel a group identity. It also
affects constancy, of course--you are actively affecting the stats of people
in your allegiance tree even if they are not logged on, so the community as
an entity does not require the constant participation of all involved.
Pretty much all the massively multiplayer commercial muds got slammed by
players saying, "It's obvious that you would get this level of antisocial
behavior!" But I've come to think one reason why it caught us by surprise
was because we were conditioned by the typical behavior in playerbases where
the magic number was not exceeded, and in which therefore there was more of
a curb on behavior based on peer pressure and community ties. Ostracism and
public shaming, the traditional tools for community policing on muds, work
best in small groups. They failed utterly when faced with several thousand
players. Predictable, but only obvious in hindsight.
MUD-Dev maillist - MUD-Dev at kanga.nu
More information about the MUD-Dev