ADVENT and Dungeon: The Pirate and the Thief

Introducing ADVENT, the first widely-known text adventure and its famous, more craft-conscious imitator, Zork and/or Dungeon (henceforth to be referred to as Dungeon).

What is ADVENT/Adventure/Colossal Cave Adventure?

I have said that I would like to avoid historical discussion, but no one can outrun their fate. At times, ADVENT and Dungeon seem discussed less as text and more as historical facts. It cannot be denied that all Infocom games are in conversation with both ADVENT and Dungeon to varying degrees, and all serious discussions of Infocom games must, at some point, call forth their spectres. All of Zork I, Infocom’s most commercially successful title, is ported directly from Dungeon, and Dungeon was initially imagined as a way to improve upon ADVENT (sometimes also referred to as Adventure or Colossal Cave Adventure) in terms of both writing and technology. A playthrough of the entire Infocom canon will reveal a genre growing further and further away from its point of origination, even if some limitations of the form were never fully overcome.

Group photo of a group called "The Imp" team. They helped build out the ARPANET. Will Crowther, the second from the right, has a close-cropped beard, a neatly-trimmed beard, and horn-rimmed glasses.
The “Imp” Team: ARPANET Pioneers (Will Crowther second from right)

ADVENT was initially authored by Will Crowther. He was a significant player in the deployment of ARPANET, a technology that would eventually evolve into what we now know as the internet. An avid caver with an interest in Dungeons & Dragons, he combined these interests to create what was–so far as the world knew–a new type of simulation or game. Just as in Dungeons & Dragons, a “dungeon master” provided text descriptions of areas, objects, and events to players, and the player likewise described via text their actions in the described world. In ADVENT‘s case, the program is not nearly as good at interpreting text as a human dungeon master would be, but that is a tough bar to reach. The technology that ADVENT and its descendants use to interpret human commands came to be known as a parser. While ADVENT‘s parser was limited to two words, it remained revelatory.

However, unlike the setting of a typical Dungeon & Dragons campaign, ADVENT‘s geography was based on an actual location: Kentucky’s Bedquilt Cave. This is the oft-discussed simulation element of ADVENT. Many have written enthusiastically about the credibility of this simulated world: “Of this first crop of games, `Adventure’ remains the best, mainly because it has its roots in a simulation” (Nelson 5). Most certainly, if simulation is a defining element of the “best,” then the prize must go to ADVENT.

The cover of Microsoft's port of Adventure, which features a man in a loincloth running across a rocky mountaintop populated by a large snake, two goblins, and a dragon.
Microsoft’s port of Adventure, which apparently never paid a dime in royalties to its authors.

It is worth wondering if, in general, simulations are by their very nature better text adventures. In this blog, I will occasionally praise this game’s or that game’s map for its “organicism.” By this I will mean that is is connected in a natural, believable way based on some unifying principle. Of course, in its originating days, ADVENT would have been considered both the most and least organic IF geography to date, simultaneously the most austere and gonzo.

I believe that the players of the 1980’s, who had no theory or essays with which to inform their experiences, preferred the wild exuberance of Zork‘s design as much as they preferred its technology. Tastes change, genres evolve, and, yes, theory emerges that may contradict the interests of a generation of players. The truth is, audiences then and now have both been right, experts as they all have been with regard to their own tastes.

ADVENT, for its part, is a revelation, building as it does something out of nothing. It (especially the canonical version modified and enhanced by Don Woods) has given tremendous enjoyment to various audiences, specialized and otherwise. Thanks to ARPANET, it was a very successful online entertainment enjoyed by its audience of hackers at universities and large corporations. Later, it was wrongly reproduced and sold by many unscrupulous publishers who never paid Crowther and Woods a dime, thereby reaching a wider and less specialized number of home micro owners. Nowadays, it is enjoyed in a rather “inside baseball” way by contemporary members of the pro-am interactive fiction community.

This is a common enough trajectory for a literary work: production, dissemination, and audience contraction. Eventually, the specialists take over. It happens more quickly in technology circles because the underlying infrastructure evolves rapidly. Instead of taking several decades, ADVENT (and its immediate successors) became “canonical” in twenty years or less. As an example of specialist contraction, imagine trying to find a conversation about Marianne Moore today. Even though she once threw the first pitch at Yankee Stadium, the discourse surrounding her is now narrow, deep, and generally limited to specific, specialized communities.

Despite ADVENT‘S roots as a simulation, its gamified focus is the acquisition of material riches, and exploration is simply a means to gaining more and more wealth. Players find various treasures and drop them on the floor of a specific room, and once a sufficient amount of treasure has pilled up, the endgame–also a celebration of material wealth–is unlocked. The treasures are just as unreal as ADVENT‘S more frequently remarked upon Tolkien-inspired elements, and both presumably originate from Crowther’s interest in Dungeons & Dragons.

Zork/Dungeon: Now With 200% More Mazes!

[note: thanks to Henrik for pointing out that Advent 350 had, in fact, two mazes]

Dungeon originated as an effort to improve upon ADVENT (or was it simply one-upmanship?) in terms of both technology and puzzles. Its original authors Tim Anderson, Dave Lebling, Marc Blank, and Bruce Daniels were by all accounts exceptional and gifted graduate students at MIT, a prestigious and exclusive university specializing in science and technology. These four students had, like the rest of the computing world, stumbled across the Crowther and Woods ADVENT courtesy of ARPANET, and had, so the legend goes, lost a week of productivity while they worked around the clock, solving all of its problems save one. They overcame the final puzzle by using a machine-language debugger.

EDIT: according to Jason Dyer, the Implementers played an early, incomplete version of ADVENT, which leaves me wondering: which puzzle was the debugger for? Get the whole story (and Lebling’s hand-drawn map) here.

An 80's-style printout of monotype lettering on yellow stock. It first reads "ZORK" in large letters. The subtitle reads "Zork: The Great Underground Empire-Part I (PDP-11 Version). It is signed by Marc Blank, Dave Lebling, and Joel Berez.
Signed cover for the commercial release of Zork I, PDP-11 version

Afterwards, they wondered if they might be able to write a better parser, and, on their second attempt, they did. These Implementers (as they began to call themselves) then began to build out a map with puzzles. They wrote a game that they, a team of four MIT students, would find challenging and interesting. It’s no surprise that many Zork players find themselves hitting above their weight–only a fortunate few could create the Implementers’ gameplay conditions, and fewer still would be able to debug assembly code.

There will be much more to say about the commercial releases of Zork in posts to come, but for now it is sufficient to say that Dungeon, despite its many improvements, is highly derivative of ADVENT. Its attempts to “top” elements of ADVENT sometimes yield miserable results. As one example, where Crowther and Woods put one genuinely unpleasant maze in their game, the implementers put three in theirs. In all fairness, there was neither accumulated player experience nor craft knowledge that could lead the authors of either game to conclude that mazes simply weren’t, for lack of a better term, fun. Nobody really knew, back then, what made a text game “fun.”

Dungeon reasonably assumes that its audience of PDP over ARPANET gamers would be familiar with ADVENT. It is a fan-made sequel in constant conversation with its predecessor, for better and for worse. ADVENT‘s appeal forty-something years ago as a new and engaging type of puzzle box is what interests contemporary audiences the least. Dungeon is a better, bigger, and more cruel sort of mousetrap, but it lacks ADVENT‘s (mostly) credible and organic geography. I find neither particularly enjoyable to play. They remind me a bit of reading The Mayor of Casterbridge for an undergraduate seminar: educationally and culturally fortifying, to be certain, but a source of something less than pleasure.

When two of the four implementers, along with other members of the MIT community, later formed a company called “Infocom,” there was not yet a clear idea of what they might make or sell. There was only a sense that such intelligent people would come up with something good. When fellow alumni Marc Blank and Joel Berez figured out a way to fit Zork on microcomputers, that particular “something good” marked the beginning of a golden age in the history of agentic reading.

Next up–Zork I: A Critical Introduction

Further Reading:

Graham Nelson’s “The Craft of Adventure”

Craft.Of.Adventure.pdf (

Jimmy Maher’s excellent historical analysis for both ADVENT and Dungeon can be found here:

Support Gold Machine on Patreon

6 thoughts on “ADVENT and Dungeon: The Pirate and the Thief

  1. Cool blog so far! Certainly it has been tempting to cut out all the cruft and go straight for the Infocoms.

    re: the story of the Zork devs playing Adventure, I should mention one of the bits most retellings get wrong is that Adventure wasn’t even “finished” when they played it. Based on Dave Lebling’s map they were playing the “250 point” version which didn’t have the troll bridge area yet nor the endgame, which I wrote about here.

    1. Fascinating. Had no idea! I always assumed they used the debugger for the endgame (I never figured it out on my own), but that must not be the case.

      I’m surprised by how many games I never knew about, despite my C64 and many pirate friends. Really enjoying your blog for this and other reasons.

  2. In the first part of The History of Zork, printed in the Winter 1985 edition of The New Zork Times, it says that

    “I was present when Bruce Daniels, one of the DM’ers, figured out how to get the last point in Adventure by examining the game with a machine-language debugger. There was no other way to do it.”

    While it doesn’t say, I always took this to mean the “last lousy point” you apparently get from leaving the Spelunker Today magazines at Witt’s End. Assuming, of course, that this “puzzle” was in the version they played. The map does mention both the magazines and Witt’s End, though.

    1. That’s must be it! I read History of Zork when I first imagined this project long ago, and really ought to have returned to it for this essay.

  3. I’m a bit to the party… but there are sctuslly two mazes in the 350 point version of Adventure. There is the “rooms all alike” and the “rooms all different”, the former is mapped by dropping objects and the latter can be mapped by observing small differences in the room descriptions.

    1. Ah yes, you’re right! That’s what I get for posting from memory instead of replaying. I’ll note your correction in the post.

Leave a Reply