Update: Find out more about the 50 Years of Text Games book and the revised final version of this article!
a.k.a. MUD1, Essex MUD, British Legends
by Roy Trubshaw and Richard Bartle
Original Version: December 1978
First Full Version Finished: April 1980
Online Until: 1987 (original Essex MUD); 1999 (British Legends)
Language: MACRO-10 (version 1 and 2); BCPL/MUDDL (version 3)
Welcome! By what name shall I call you? *PATHOS This persona already exists - what's the password? *Password Yes! Your last game was today at 6:24:32. Hello again, Pathos the champion! Narrow road between lands. You are stood on a narrow road between The Land and whence you came. To the north and south are the small foothills of a pair of majestic mountains, with a large wall running round. To the west the road continues, where in the distance you can see a thatched cottage opposite an ancient cemetery. The way out is to the east, where a shroud of mist covers the secret pass by which you entered the Land.
“Imagine you are playing an adventure,” asks an article in a popular computing magazine, credited to someone called Endora the Witch:
Let’s say you’re in a room of a house, you have found some treasure, and are now a bit stumped as to how to get it past the bookcase, which you can’t shift but which you’re certain conceals a secret passage. You have tried all sorts of commands to no avail, and are about to give up when up on your screen comes the message: “Tom has just arrived.”
Tom is “not part of the program, but a real, live person,” Endora explains, “sitting possibly hundreds of miles away and exploring the same land as you are.” This is a mind-blowing concept not easy, at the time, to explain. “CB radio in fantasyland,” tries Endora’s editor in the article’s subheadline; another journalist wrote that the game was built atop “quite a sophisticated teleconferencing system.” This adventure with real live people in it had been getting more and more coverage in the gaming press and at enthusiast meet-ups. It sounded incredible. Everyone wanted to try it. For a long time, almost nobody could.
The game was called MUD, which stood for Multi-User Dungeon: the last word used not in a generic sense, but as the alternate title for Zork. During the few months the MIT game had gone by the other name, a popular port had spread far and wide, and one of the places it reached was Essex University in the United Kingdom. Essex was among the first schools to connect to Britain’s answer to the ARPANET, the EPSS (Experimental Packet-Switching System), and in 1978 was one of the few places in Europe that allowed undergraduates to freely access the burgeoning worldwide network of mainframe computers. Some of the bytes that wended their way beneath the Atlantic to Essex described the underground empires of Dungeon and Adventure to a fascinated undergraduate named Roy Trubshaw, who like many others on sundry continents had become enthralled.
Trubshaw and some other Essex hackers had recently discovered a neat trick on the school’s PDP-10 mainframe. A time-sharing system divides its memory and resources between connected users, isolating them from each other to give each the illusion of their own private machine. Admins could use “interprocess communications” to share data between users, though lowly students didn’t have the privileges to do so. But one of Trubshaw’s friends had discovered a back door. Each running program had both a shared “code segment” in memory where all its instructions lived, and an individual “data segment” for each user. It turned out that an obscure and mislabeled monitor call, .SETUWP, actually let an unprivileged user un-set the write-protection bit for a program’s code segment. The friends immediately saw the rich possibilities of this exploit: if some of that memory space could be used to store not code but shared data, and the running program was simulating a world like in those immersive text games from America, everyone connected could see the results of any other user’s changes. The trick enabled the possibility of an unsanctioned but gloriously multi-user Dungeon.
Trubshaw started building a proof of concept, sketching in the core features it would need: rooms, objects, and a parser, like its predecessors, but also notifications for events like another player arriving in or leaving your location, mechanisms to ensure your game couldn’t get out of sync with someone else’s, and commands for players to speak and be heard. Like many future worldbuilders, the first few rooms Trubshaw made were recreations of his childhood home.
By December 1978 he’d finished a basic version of his multiplayer engine, and to build more content enlisted the help of a bright-eyed underclassman enchanted by the game’s potential, Richard Bartle. “Roy was mainly interested in the programming side of things, rather than the design of rooms, puzzles and so on,” Bartle later remembered; so he began to build out the game into something more like the size of a Zork or Adventure. The two continued improving and extending the game through 1979. Eventually the engine began to strain against the limitations of MACRO-10, the PDP-specific assembly language it had been written in, so at the end of the year Trubshaw decided to start over in a new and more powerful language, BCPL: the language which would evolve into C.
But Trubshaw, at some point, was also meant to be graduating. By the end of the 1980 spring term he’d reluctantly handed over the MUD codebase to Bartle, who would continue improving it in one form or another for nearly a decade: through the rest of his undergraduate and postgraduate careers at Essex, then as a lecturer there, and finally at a start-up he’d found to commercialize its revolutionary multiplayer potential. But as he first took the reins of the project, his immediate concern was how to turn the prototype engine and unfinished map into something more stable: a solid game, but also a place, a persistent universe you could visit and return to, made from words and sentences.
The core of MUD was not unlike its predecessors. Players explored a map of rooms filled with mazes, monsters, and puzzles, collecting treasures which could be returned to a central location to score points. Bartle broke from most earlier games by setting most of his world aboveground, naming it “The Land” and filling it with magical glades, soft pine forests, misty graveyards, and ancient ruins. While geography was relatively easy, recreating the puzzles of single-player games proved far more challenging. At first Bartle tried making puzzles that required multiple people to solve, like the bookcase too heavy for one person to move, or a special area that could only be reached by enlisting multiple players to meditate, boosting your spiritual power enough to reach it:
*southeast Shrine. You are inside a small yet sacrosanct shrine. A sense of deep respectfulness fills this modest room. The way out, into a pine forest, is to the northwest. It is obvious that the shrine was meant to be used for quiet meditation, like similar chambers. *shout Does anyone want to meditate with me? *quickwho Pathos the champion Blatch the enchantress Maria the sorceress Blink Mugs the necromancer Aphrodite the heroine Jethro Gobble the legend * A male voice in the distance shouts "Gobble does" * Aphrodite the heroine tells you "I will" * A female voice in the distance shouts "yep i do blink." *sh Ok, Im in the shrine. The rest of you get ready! * A male voice in the distance shouts "Gobble in cave" * Aphrodite the heroine tells you "ready?" *sh Ok, lets try now!!! *meditate You feal a great tranquility filling your being, and when you cease your meditation, you are in a strange place... Outer sanctum. A golem of solid iron stands here as guardian of the inner sanctum. *sh It worked! *sh Ummm.. is this golem pretty tough? A male voice in the distance shouts "Idiot. Yes, pretty tough. you got a weapon I hope???"
But MUD’s creators soon realized that multiple players break a Dungeon-like experience. Object-based puzzles become problematic: if you need a lamp to go underground, the first person to grab it prevents anyone else from following. And once all puzzles have been solved and treasures claimed, what would new arrivals have left to do? Trubshaw and Bartle had hit on the notion of a world that would frequently reset itself, moving items back to their original locations and randomizing some positions and room connections; treasures, when collected, would be dropped in a swamp, removing them from the world until its next reset so only one player could score those points each time around. But this also proved unsatisfactory. Experienced players would simply rush out to sweep up all the treasures immediately after a reset, leaving novices confused and clueless about what to do.
Bartle came to realize that a dungeon and a multi-user dungeon were fundamentally different things. “The puzzle-based, narratively constrained format of adventure games couldn’t work in the setting of a multi-player game,” he later wrote: “the world had to assume dominance, not the problem-solving.” Dungeon was about puzzles, but “MUD was about freedom,” the joy of co-existing with other people in a simulated, dynamic, living world. He began adding more elements that made the game’s world feel alive, like occasional rainfall that prevented certain actions but enabled others, or creatures that roamed around exhibiting behaviors he called “instincts,” like a cat that attacked mice on sight. Some creatures had an instinct to attack players; killing them became another way of scoring points. Bartle named the creatures “mobiles,” needing a single-word variable name for something that was neither a player nor a static part of the environment. The term is the origin of the word “mob” still used for enemies in online games today.
Combat, either with mobiles or other players, became the key way of acquiring points and power in MUD. Because typing and connection speeds could vary, melee meant typing a single command and watching the results scroll by in real-time, the odds for each hit or miss dependent on the relative stats of each character. Of course, reaction time still mattered: a losing player might choose to retreat with the command FLEE OUT (which could be abbreviated F O, when even more urgency was needed).
*kill skeleton The viciousness of a whack by the skeleton sends you sideways. Dazedly you pull through, and press forward into the contest. Your mis-timed return blow at the skeleton is effortlessly shrugged off. You easily evade a poor swing from the skeleton. You bash the skeleton with a punishing forehand! You comfortably shrug off a feeble thump by the skeleton. You wallop the skeleton with a crushing whack! Your last swing took the life of the skeleton! You are victorious - this time...
Combat messages made heavy use of templated text. You wallop the skeleton with a crushing whack! came from the base:
"You :r :p with a :r :r!"
The first word might become wallop, but could also be thrash, take aim at, smite, bash, or any of a dozen other possibilities. In this way each such message could be rendered thousands of different ways to prevent dull repetition, an early example of procedural text in games.
But MUD’s true dynamism came from other people. Even at the remove of a primitive terminal, interacting with humans was endlessly fascinating, and Bartle soon began focusing his energies on game mechanics that maximized the pleasure of player-to-player interaction. He added powerful mobiles like the dragon which could only be defeated by a large team of powerful players, giving rise to some of online gaming’s first raids. Commands were created that had no game effect except nonverbal communication, like LAUGH: the earliest emotes. A WRITE command let players leave messages in certain parts of the world, such as a Captain’s log book:
Captain's cabin. An inky quill-pen has been left here. The log of Captain Oliver is here. In it is transcribed the following: "Hack and Slay, Hack and Slay, Hack and Slay!" "Shadow's still lengthening, in a 2060 stylee!" "Gail was here all alone and without Richard (sob)..." "Anana was here with the 5 zombies and 1 skeleton!" "Was he? So that makes 6 zombies and a skeleton all together?" "Duncan the wizard played from California... All by himself!!! "
Just as MUD’s multiplayer foundations were growing stronger, Essex University had begun allowing outside users to dial in via modem to its mainframe. While MUD’s original players had all been local to Essex, folks from around the country and, indeed, the world started logging in. The game became so popular that the university’s meager allotment of phone lines became swamped, and Bartle was forced to restrict MUD access to the bleary-eyed hours between 2am and 7am, when all but the most dedicated legitimate users would be asleep. This proved to be no deterrent whatsoever. “It was usual for players to grab a line at about midnight,” one admin later recalled, “and sit there typing ‘HELP’ every 5 minutes until 2am so they didn’t get disconnected.” At first there were only sixteen outside lines available, four of them restricted to a painfully slow 300 baud (at which rate a single line of text could take several seconds to download). MUD’s descriptions were supposedly only a few lines long because “any more and you’d get killed while reading them.” More phone lines would eventually be added, but the game’s architecture could only support a maximum of 36 simultaneous players, a hard limit related to the PDP-10’s 36-bit word size.
The game’s small but hardcore player base, staying up all night to play and yawning through classes or day jobs, began to define a culture for their new online world. Like many first stabs at culture, it was based almost exclusively on bloodshed and power. Defeating another player in combat netted you a fraction of their points, which in turn controlled your stats and rank; so the fastest way to advance became a well-timed slaughter. While Adventure and Zork had offered named ranks to players based on points, this was purely decorative: in MUD, ranks became a visible pecking order, with a Champion likely to kick the ass of a Warrior but prone to being slaughtered in turn by a Warlock.
Playing the game could be brutal. “It allows you to behave in a way which would be totally unacceptable in real life,” one guide wrote: “MUD is a very violent game, and pacifists never live long.” In a passage that works equally well to describe a post-apocalyptic dystopia, it added that “A gang of players is quite difficult to overcome, and players often find it beneficial to form one of their own.” One player recalled their typical newbie experience:
I worked out how to login (not easy), I worked out how to load the game and... I got killed. I tried again, I got killed again and eventually, when people were bored of killing me, I tried to talk to a wizard about how the game worked and was told to bugger off and then killed again. ...This was most people’s initial experience of MUD.
Wizard, or wiz, was the highest rank, and came with a special form of ultimate power. Trubshaw had built debugging verbs into the game that allowed for actions like manipulating objects no matter what room they were in, printing the location of any item or player, or viewing the output of any user’s terminal. Bartle had the ingenious idea that these verbs could be made available to high-ranking players if they were rebranded as magic spells. Suddenly there was a tremendously appealing reward for working your way up through the brutal ranks and achieving the maximum level possible: ultimate power over the virtual world that mere mortals lacked. While wiz spells were touted as a responsibility given to experienced players to help enforce order and assist novitiates, surviving anecdotes make it clear they were often used for anything but.
Wizzes, not to put too fine a point on it, were often right bastards. They could move monsters to any room they liked to sic them on unfortunate players, perhaps dropping a deep-sea shark on an unsuspecting hero wandering the forest; or even make objects come alive and attack, like one wiz who enjoyed making the river mobile and granting it tremendous combat skills before setting it loose on confused and hapless mortals. (“Oh no!” he’d shout, “It’s that Killer River again!”) One reviewer noted that it was “not unknown for a wizard to remove the cliff from the beach and put it down at the exit so that anyone entering the game falls to their death.” If it amused certain wizzes to kill players for no good reason, it amused them even more to torment them first. A secret storeroom accessible only to wizzes was stocked with items designed to frustrate players, including a replica of the game’s most valuable treasure which looked exactly like the real thing but would score zero points when dropped in the swamp. Some wiz traditions were more kindly: the storehouse also held items made specially for the winter holidays, including reindeer, snow, and a mobile Santa. But since the holidays were also a time to stage mass battle-royales, the efficacy of these items for encouraging goodwill towards men was questionable. “Christmas in the Land of Mud is a time for thanksgiving,” Bartle once noted, “and mass slaughter.”
Despite the brutal milieu, a culture did sprout up on MUD. A dialect called “mudspeke” appeared, where T meant treasure, SNIF meant sadness, and countless in-jokes became enshrined in shorthand terms and local slang. While the game could be brutal, real friendships also sprouted between adventurers up way past their bedtimes, and soon some of the wizzes were elevated to co-maintainers, working along with Bartle to police, bugfix, and revise the game.
MUD had been built to be easy to extend. Trubshaw had based its data format on the one in Adventure, which stored message texts and map details in external files, and he and Bartle extended the format so nearly anything about a world could be defined without touching the core engine’s source code. They named the format MUDDL, for MUD Definition Language—confusingly, totally unrelated to the Muddle used to make Zork—and Bartle boasted it allowed you to add a new room to the game in two minutes, without needing to recompile or even, in some cases, restart: the data files were large enough they were often read directly from disk, not stored in memory. Creating a new room meant adding only two new text blocks to the active MUDDL file, one to the *ROOMS section defining the location itself:
You are on the bank of a fast flowing river with pasture to the north and forest to the south.
And another to the *TRAVEL section, defining the exits:
river1 n wfall w
n river2 e
n clifff nw
n wpstre n ne out
boat spstre s se sw
rain spstre s se sw
120 s se sw
The code above goes some way toward demonstrating the power of MUDDL syntax. The first column holds a condition that might gate access to an exit, with “n” meaning none; if an object or a class is listed as a condition, it must be in the room with a certain property or carried by the player for the exit to work; and a number means to print that particular message rather than allow movement in the given direction. The last three lines above define that the river to the south can only be crossed if the player has a boat or it isn’t raining; if neither is true, message #120 will be shown (“The weather has swollen the river and you cannot cross!”). Messages could themselves be defined in their own section of a MUDDL file, *TEXT.
Other parts of a MUDDL file could define nearly any particulars of a given MUD game: *VOCABULARY, *COMBAT, *LEVELS, *OBJECTS, and even minutia like the *HOURS the game should be available to play. An admin could even create new action responses through a format allowing for references to objects, functions, values, and messages. For instance, the following four lines:
action drop .insert anything flame null null 781
drop torch container destroy destroy second 771
drop something stream move null bwfall 835
drop something stream ifweighs null 2001 836
...succinctly define four special-case behaviors for disposing of objects: putting anything into something flame-like will destroy it, putting any torch into any container destroys both items, and dropping something into the stream will carry it away to the room behind the waterfall, unless it weighs more than 2 kilograms. MUDDL allowed for an extraordinary flexibility in extending and incrementally revising a simulated world, and its flexibility paved the way for thousands of future games directly or indirectly built on its model.
Success for MUDs would be some time coming, however, largely because it would take a while for infrastructure to catch up to demand. The difficulty of jockeying for a spot on one of Essex University’s late-night phone lines meant that for years, very few of the people who heard about MUD could actually play it. As late as 1985, a somewhat grumpy games journalist called it “the game you’ve always wanted to play but have never been able to log on to.” The high cost of long-distance connections was also a big factor: “If British Telecom could only be persuaded to reduce its rates for data transmission,” another journalist wrote, “I believe that this type of game is going to prove far more popular than anyone now suspects.”
The devotion of MUD’s players had long since convinced Bartle of the commercial potential of multiplayer fantasy gaming, and by the mid-1980s he had struck a deal with Compunet, an early British dial-up network provider, to add the game to their service. By 1987 it was also running on CompuServe in the United States under the name British Legends. A trickle of other MUDs had started to appear, with titles like Shades, Gods, and MirrorWorld, in part because despite his commercial ambitions Bartle had been generous in sharing the game’s source code and encouraging others to expand on it:
We could have clamped some intellectual property on it, but the reason that Roy and I wrote MUD wasn’t to make money. It was because we wanted to make the real world a better place, and the way to do that isn’t by clamping down on intellectual property and stopping anybody else from making it. The way to do that is to give it away for free and to let other people do what they want with it. So that’s what we did. (source)
In early 1989, an engine called AberMUD debuted, written a year earlier by a group based at the University of Wales, Aberystwyth and then ported to C to run on Unix-based systems: this caused an explosion of new MUDs and MUD engines. By the early 1990s there may have been something like a thousand active MUDs running around the world, with hundreds of thousands of players exploring virtual worlds and building friendships (and bitter rivalries). A branch of MUDs would break away from their dungeon-crawl origins and become serious spaces for online socializing and creation, the MOOs and MUSHes; the main line would evolve by the late 1990s into graphical MMORPGs, with the designers of games like Ultima Online and EverQuest coming in with years of experience running their textual predecessors.
Bartle acknowledges that the contribution he and Trubshaw made was as much an accident of timing as of pure innovation. He likes to tell a story about how many times the game of golf had been independently invented through history before a particular version from Scotland became the dominant model, and he’s tracked down at least five other multiplayer fantasy games being developed contemporaneously with the original MUD, each unaware of the others. But he also likes to suggest that perhaps the game he and Trubshaw made succeeded because it was built on something more than just technical innovation:
Okay. So it’s 1978, and you are a student at the University of Essex, and you’re studying computer science. You’re not supposed to be there. You’re not supposed to be at university: no one in your family’s ever been to university before. ...[Roy and I,] we weren’t rich. My parents—my father was a gas fitter, you know. He spent all day installing cookers in people’s houses. My mother was a school meals cook... I’ve got a Northern accent, [and] Roy comes from Wolverhampton, and he’s got a West Midlands accent. ...It doesn’t matter, being really smart, if people, as soon as you show up, as soon as you open your mouth, they’ve pigeon-holed you because you’re working class.
And we wanted a place where we could go where none of this mattered, where who you were was based on your strength of character, on who you were as a person, and it didn’t matter what sex, gender, class, whatever you were, you could just go there and be and become yourself. And we never really discussed it at the time, Roy and I. We just sort of—we implicitly understood that this is what we wanted to do. And so we did it. We made a world.
Though it’s ironic that MUD’s player-versus-player design spawned a new kind of digitally-enabled class system, the tyranny of the wizzes was at least in theory a meritocracy—a dictatorship that any connected citizen could aspire to hold. And it’s also true that multiplayer fantasy games would open up a generation to the possibilities of a life free from constraints that had once seemed unbreakable: of social class, of physical ability, of gender, of distance. Online games would change the world, even if only the geeks, at first, could see it. “MUD is not a one-off occurrence,” its creators predicted in 1985, far more accurately than they had any right to. “Instead, it’s just the first of a new generation of computer game. ...What we see in the Multi-User Dungeons running at various universities... can only be called the beginning.”
Next week: The experimental text game where players interacted not by typing commands, but writing their own florid dialogue.
Richard Bartle has collected a tremendous number of writings about MUD by himself and others, providing many key sources for this article. You can use telnet on the command line to connect to a server running the original MUD code, and a version of the source code from 1986 is preserved on Github. Excerpts are from the 1985 book An Introduction to MUD by Duncan Howard, somewhat cleaned up and modified; other major sources include The Digital Antiquarian and Richard Bartle’s books “Designing Virtual Worlds” and “MMOs from the Inside Out.”