Curtis will appear on Wednesday's episode of "The Screen Savers" to talk about the viability of online communities.

First, some definitions

• What is a MUD?
• What is LambdaMOO?
• How do I enter LambdaMOO? Click here.

A computer scientist trips and falls into the MUD

• I came to the Xerox Palo Alto Research Center (PARC) in the summer of 1983, as a summer intern working on adding static type checking to the Smalltalk programming language. By the time the summer of 1990 rolled along, I had started and was leading the SchemeXerox project in the Computer Science Laboratory at PARC. The project ran into some bureaucratic snafus, and I began to spend a lot more time reading NetNews than I had before; I was even willing to consider reading completely off-the-wall new groups whose charters were obscure at best.

• New newsgroup alt.mud; subscribe? (yes/no) yes

  As I read through the first day's traffic on this new group, a few things quickly became clear:

  1. The discussion was neither about wrestling nor wet dirt.
  2. Everyone who was contributing to the group already understood the subject matter so well that they saw no point in making it clear to anyone else.
  3. Whatever this "MUD" stuff was, approximately half the contributors thought it was a game; the other half vehemently and heatedly disagreed.

• Over the next few days, I ascertained that MUDs were some kind of software, that people used them together somehow, and that they owed something to older programs like Adventure and Zork. Remembering the purely textual but evocative puzzles of Zork, clearly the product of wicked, twisted imaginations that I admired very highly, I became even more interested in tracking down these descendent MUD programs. It took, I think, over a week before someone on the newsgroup finally let slip enough information for me to connect to one of these worlds.

• Apparently, there were other people using the program at the same time, and it was possible for them to talk to one another! You can easily imagine my surprise the first time someone "teleported" into the same "room" and started talking with me; I almost embarrassed myself further by mistaking them for a cleverly programmed "robot"! This program, this place, this virtual world was created by the very same people who were currently visiting it, along with a great number of others. Anybody who came there and explored for long enough was allowed to create more!

• After disconnecting from the MUD, I started tracking down and exploring the FTP sites that contained documentation and source code for the many kinds of MUD servers that existed. As I searched and dissected these sites, I discovered to my disappointment that MUDs were, at least in the main, not really programmable at all; there were various commands for building and for setting the messages on the objects, but pretty much all of the actual behavior was fixed and unchangable by the users.

• I finally came across two exceptions to this rule, two MUD programs that were almost entirely programmable by the users. The first of these was very disappointing from a designer's point of view, being very error-prone and difficult to use reliably. The second one, though, was different; it was reasonably straightforward and reliable, and it had a simple but powerful object-oriented programming model. There was this student from Waterloo, one Stephen White, who had created it; he called it "MOO."

Next page: From MOO to LambdaMOO in Four Easy Months

From MOO to LambdaMOO in Four Easy Months

• It was late in September 1990 when I found the "alpha test MOO" running on a machine named "belch" in Berkeley. Stephen (aka Ghondahrl aka ghond) had set up this site so that he and others could test out his creation as it was being written and refined. It wasn't perfect, of course (what language is?), but he seemed to have made many fewer of the mistakes made by most language-design amateurs. I wrote few simple MOO programs of my own, trying to recreate in MOO some of my favorite puzzles from Zork, and decided after a few days that it would be fun to work with Stephen on the future evolution of his language.

• By the end of October, I had added a number of new features, fixed a number of bugs, and gotten hungry to try out my new version of the program on others. For my earlier MUD visits, I had used the name "Lambda" for myself; it was a major keyword in the Scheme programming language that I had been working in for so long and so it naturally sprang to mind when first I was prompted for a character name. I now decided to use "LambdaMOO" as the name for my revision of Stephen's MOO server, since it was "Lambda's MOO".

• I remember, in particular, one evening's conversation early in my tenure there; I chatted for quite some time with three other folks and it became clear after a while that two of them, named Gemba and Gary_Severn, must know each other outside of this electronic context. I was shocked to discover that Gemba and Gary were actually connected from the Australian National University and that the third other person was sitting at a terminal in Israel! It was only then that the geographic scope of MUD participation began to dawn on me.

• During the earliest days of LambdaMOO, through the beginning of 1991, everything was fascinating every day. The technical work was fascinating as Gary, Gemba, and I tried to build the core libraries of MOO programming. The collaborative feel of it was fascinating as we worked closely together from our separate offices thousands of miles apart. The creativity was fascinating as I laid out the core geography of the LambdaMOO mansion (based on the layout of my real-life house). But most fascinating of all was what was happening all around and through me: a community was forming inside this computer program, a community with hundreds of people, all learning about LambdaMOO by word of mouth and coming to see what it was about.

• Of course, not everyone was so nice or so constructive, though it took quite a while before this became clear. I think it was at least a few months into LambdaMOO's existence before the "Penn State a**holes" came calling. Two connections arrived nearly simultaneously, both from computers at Penn State University, and the log showed the player names these two new users had chosen for themselves: vulgar terms for parts of the female anatomy, words I wouldn't have repeated to my mother. By the time I joined them in the "Living Room," they were both typing "F*** YOU" over and over again, to the annoyance of all present. Later that day, in a black mood, I wrote the first version of the LambdaMOO "@toad" command, for permanently destroying a user's character and all of its possessions. Some of the shine was off the apple, never to return.

Next page: LambdaMOO Faces the World, and Vice Versa

LambdaMOO Faces the World, and Vice Versa

• By the beginning of February 1991, I finally felt ready to make a public announcement of the existence of LambdaMOO; we had a small but thriving community, enough structure in the virtual world to give newcomers a starting point, a reference manual worth reading, and enough online "help" texts to answer most questions we thought a new user would ask. We got a fairly satisfying response, with lots of new users coming by to see what we'd been up to. We stopped being impressed when there were ten users online at the same time and started more often seeing numbers like 25 and 30. I thought we were huge beyond belief.

• With an increase in population and popularity, though, also came an increase in problems; we had built up a set of tacit rules for gracious living on LambdaMOO and had never really noticed ourselves doing so. These newcomers didn't know our rules, didn't know our style, and didn't know the lessons we'd learned over the course of four months of birthing.

• A player approached me one day to complain about how several of his objects had been moved about by other players without his permission; he asked me to find a way to let all the new players know what was considered acceptable and unacceptable behavior on LambdaMOO. The idea seemed pretty daunting to me, but he had a good point, so I wandered all over the MOO asking old-timers for examples of unwritten rules that ought to get writ. After a few days, I wrote the first draft of the "help manners" text that for a long time was the only written "law" that LambdaMOO had.

• The "Case of the Lookalike Puppet." One character and another had had a falling out and one of them had, afterwards, created a MOO "puppet" with the same name and description as the other player. A puppet is a MOO object that acts much like a player (saying things, performing actions, etc.) but that is actually owned by and under the control of some other player. The player who was being imitated in this fashion was most upset that there was this doppelganger of them wandering the halls of LambdaMOO. They wanted the puppet renamed at the least, and preferably outright destroyed.

• I agonized over the situation, I had long talks with both players, I took the problem home to my wife and argued both sides of it with her for hours, it seemed. I eventually worked out some sort of compromise (I forget the details now), but the case got me to thinking. What was going on here? What was happening to me? This was certainly not the kind of thing I had been used to doing as a programming language researcher. I was a hacker and not a judge, wasn't I?

• In January of 1992, I had a memorable conversation with my lab manager. He suggested that after I'd fulfilled my commitments to the SchemeXerox project that I should consider going into "this MUD stuff" full time! It was clear to Mark, and becoming clear to me, that the work I was doing had a good chance of being important and, more significantly, it looked like there weren't any other computer scientists working in this area.

• On the morning of April 1, 1992, when I first got to work, I checked out the transcript of my perpetual connection to LambdaMOO. Amid the usual paged questions and the like, there was a cryptic little message about how a major fire had just swept through the house. Curious, I began wandering around the core of LambdaHouse; it was marvelous. Clearly, some of my staff of wizards had been very busy preparing for this wonderful April Fool's Day hack.

• At some point in my wanderings, a worried player paged me to say that it really, truly wasn't his fault, but he seemed suddenly to be a wizard! I didn't believe it, of course, but I checked it out just the same and discovered to my shock that it was true; when I inspected his player object, it clearly had the "wizard" bit on! He pointed to the latest article in the LambdaMOO newspaper; that article, written by my wizards, described the fire and said that, in order to hasten the repairs, all players had been made into wizards so that they could help out. I was utterly aghast.

Next page: Things Fall Apart; the Center Cannot Hold

Things Fall Apart; the Center Cannot Hold

• With the increased attention being paid to us, the population of LambdaMOO grew at an alarming pace. This, in turn, put an ever-increasing amount of pressure on the wizards; we were spending more and more of our time just keeping the place running, dealing with inter-player disputes, and judging the things that players had built when they wanted permission to build even more. None of us were being paid to operate LambdaMOO, but it was taking an increasing toll; we started to create institutions and automated procedures to lighten the load.

• I created a committee of long-time LambdaMOO players to take over the job of judging other player's work (according to a recently-announced quota on the amount of objects players can create). I named them after the kinds of groups I'd seen in various RL home-owner's associations: the Architecture Review Board (ARB). It was clear that this group would need some slightly privileged tools, so that they could make a sufficiently complete judgement of other players' work.

• From the very beginning of the ARB, there were players suspicious of it. How was it formed? Who chose those particular people and why? How do they make their decisions? What is said in the Star Chamber? What can't we go in there? It wasn't (at least at first) that anyone knew of anything bad actually happening around the ARB; its very existence, and the way it was created, were enough to worry some players.

• The third major burden on the wizards during this period was inter-player disputes. The wizards were the police, the judges, and the executioners; we had set ourselves up for this back when "help manners" was drafted, when we had claimed "Vengence is ours, sayeth the wizards." In retrospect, I think the last half of 1992 was almost entirely characterized for the wizards by our weariness and stress. Something had to give, and it was us.

• On December 9, 1992, I posted a pivotal message to LambdaMOO's *Social-Issues mailing list; I titled it "On to the next stage...," but somehow history has indelibly tagged it "LambdaMOO Takes A New Direction," or "LTAND." In that message, I announced the abdication of the wizards from the "discipline/manners/arbitration business;" we would no longer be making what I glibly termed "social decisions."

• LambdaMOO slowly became a rougher place after LTAND. It's hard to say how much LTAND accelerated a process that was already in place, but surely it didn't help to hold it back. The level of inter-player strife and harassment rose and rose, slowly but inexorably. The crisis point came about four months after LTAND, in the infamous rape in cyberspace case.

Next Page: Direct Democracy: LambdaMOO Is to Sysiphus as...

Direct Democracy: LambdaMOO Is to Sysiphus as...

• My primary inspiration for the design of the petitions and ballots system was the voter-sponsored initiative process here in California. It was simple in outline: any LambdaMOO citizen could create a petition proposing that the wizards take some action; if it got enough signatures, it became a public ballot measure that passed on a two-thirds majority vote. Before a petition could become a ballot, though, it was necessary for the wizards to "vet" it, to ensure that the proposal was (a) clear, (b) feasible, (c) appropriate, (d) legal, and (e) secure.

• Overall, though, I'd have to say that the petition system has failed on LambdaMOO. It has, by and large, failed to be the jumping off point I hoped for; we have not seen it used successfully to move LambdaMOO to a working, stable form of self government. There were long periods, indeed, where many petitions reached ballot stage and none of them passed; it seems to me now that the voting population could never agree on anything of real substance. I think that this is the real lesson of LambdaMOO's experiment with direct democracy.

• Deep in its very structure, LambdaMOO depends on the wizards and on the owner of its machine. These are not and cannot be purely technical considerations. Social policy permeates nearly every aspect of LambdaMOO's operations, and only the wizards can carry out those operations. As a result, the wizards were at every turn forced to make social decisions. Every time we made one, it seemed, someone took offense, someone believed that we had done the wrong thing, someone accused us of awful ulterior motives. It felt a bit like the laws of thermodynamics: you can't win, you can't even break even, and you can't get out of the game.

• Throughout the entire month of April, the wizards' private mailing list was pulsing with activity; we were drafting, arguing about, and re-drafting a new fiat, my third pivotal message to *Social-Issues. Finally, on May 16th, we all agreed on a draft and I posted LTAD: LambdaMOO Takes Another Direction. In it, we formally repudiated my earlier theory of a social/technical dichotomy; we explained how impossible that fiction was and declared our intent to cease apologizing for our failures to make it reality. It was, in a way, a wizardly coup d'etat; out with the old order, in with the new.

Conclusion: Can We Get Where From Here?

Conclusion: Can We Get Where From Here?

Will it work this time? Can we achieve peace on LambdaMOO and a tolerable job for the wizards? It's best not to set one's sights too high for a place like LambdaMOO. Trapped in a fundamentally untenable economic and political situation, perhaps it is enough to maintain, to keep the place together for another year or two or three, until it stops providing something that its thousands of regular users can't get anywhere else, or until something better comes along.

What would "something better" look like? On the top of my personal list is a vastly more equitable distribution of both power and economics. If these online "virtual" communities are to have the same robustness as the more physically oriented ones, they must become pliant and mutable under the same forces that cause RL communities to grow and change. It must be possible for incompatible sub-communities to separate and grow apart, thereby relieving the kinds of stresses that constantly tear at LambdaMOO. It must be possible for individuals to completely control their own creations, without the specter of an all-powerful wizard looming in the background, distorting all natural social interactions.

When these things happen, when online communities can form and break apart as fluidly and naturally as do face-to-face ones, something dramatic will be taking place in the history of the human race. If you look closely at the Internet today, you can begin to see it happening already, in newsgroups and in the World Wide Web. It's still not here in the area of more synchronous interactions, but I know for a fact that it's coming; I'm working hard for it every day. We have only to hold on, to keep dreaming the dream; more than at any other time in human history, dreaming is nearly enough to make it happen.

Pavel Curtis is the founder of LambdaMOO, still home to more than 150 virtual communities and 8,000 members. Curtis is also the founder of the Web-conferencing company PlaceWare.