• Got a Eurogamer account? Your details will work here too!

  • Need an account?

    Create an account. They're free!

  • Forgotten your login details?

    Recover your account here.

Dave Lebling on the Genesis of the Adventure Game - and the Creation of Zork

Jaz and Jeremy talk to adventure game pioneer Dave Lebling about the early days of Zork, Infocom, sarcastic parsers, and the joy of feelies.

A long time ago, on a video monitor far, far away, there were nothing but glowing green characters, borne from a stream of electrons hitting the phosphor layer on the back of a curved glass screen. Sometimes those characters glowed amber. But either way, there were just characters and darkness.

Colossal Cave Adventure, the first of its genre in its native form.

But out of that void came something new. Through a combination of human ingenuity and imagination, those characters became the building blocks of something never before seen. In a pre-video game world where there was nothing but the darkness of an empty command line, the appearance of the words "You are standing at the end of a road before a small brick building. Around you is a forest. A small stream flows out of the building and down a gully," heralded the birth of a new world. And lo! There was light.

It was 1976, the dark command line was a DEC PDP-10, the sun was Colossal Cave Adventure aka ADVENT aka Colossal Cave aka Adventure, and its world was a maze based on Mammoth Cave in Kentucky, where William Crowther, Colossal Cave Adventure's creator, had previously spelunked. It was, of course, the first ever text adventure.

Since there wasn't much of a games market outside of the Magnavox Odyssey and dedicated Pong-type consoles, Colossal Cave Adventure was shared around the ARPANET mainframe network like many other programs of the day, with little thought that it might have some commercial value. Over the following two years, as programmers began to tweak and modify the game, Colossal Cave Adventure fragmented into a myriad sub-versions – even to the point where it became a commercial product, as was the case with Microsoft's 1980 Apple II Plus and TRS-80 Adventure.

However, the most famous version was created by Don Woods. He discovered the game on a Stanford University computer in late 1976 and, with the blessing of Crowther, modified and streamlined the game map, and added fantasy elements, including elves and a troll, creating what is now considered the definitive 1977 version of the game.

Dave Lebling

It's at this point I'd like to introduce my guest. Dave Lebling is a pioneering interactive fiction games designer best known for his work on Zork, the earliest commercial interactive text adventure. Written between 1977 and 1980, Zork absolutely owes its existence to Colossal Cave Adventure. Or was it Adventure? I ask Dave whether he remembers which version was which?

"It's the same thing, really," says Dave. "I'm not sure what Crowther called his first version. I think Colossal Cave. And then when Don Woods expanded it, he called it Adventure. But maybe it was the other way around. I don't remember. Colossal Cave is actually a real cave, but the geography of Adventure is based on Bed Quilt Cave, which is a different cave, but part of the same system. But yeah, everyone who worked on Zork played it."

"Don did not hold onto the source code the way we held on to the source code of Zork, and so people would get his FORTRAN source, and if they decided they wanted to add new stuff to it, they'd just go in and code the new stuff and compile it and release it."

Zork's snappy tone was influenced by Dungeons and Dragons sessions. [Source]

I ask Dave how he felt the first time he saw Colossal Cave Adventure.

"I was blown away by it," he says. "I was a D&D player at the time, so what I saw it as being was D&D on a computer, where you were interacting with a computer DM to do things using a two-word parser. Plus, if you ever had the privilege of visiting Bed Quilt Cave, you could literally navigate through it using Will Crowther's Adventure, because it's exactly the same. I thought that was pretty cool when I discovered it was true."

"However, a lot of the descriptions referenced objects that didn't exist in the game, which was frustrating. It's like Chekov's gun. If you mention something somewhere, it really ought to have some role in the game. At least you ought to be able to talk about it."

"At the time, I was at MIT Dynamic Modeling Group along with Tim Anderson, Marc Blank, and Bruce Daniel, and we thought we could create a better parser. That way we could make all these things that are referenced exist in the game, and make it feel more real. We thought we could do better. In some ways I think we did a lot better. In other ways I don't think we did."

Deadline Feelies

Jeremy asks Dave what he thinks he could do better today?

"Well, the tools that come to mind for me would be better speech recognition, so you don't have to type. Speech recognition is getting better, but it's still not completely natural, as anyone who's ever gotten to a call center phone tree will immediately realize. That's one thing. Another is natural language parsing, as opposed to just recognizing what the words are – knowing what the their intent is. I'm not an expert in that. My software fields are somewhat away from that. But I haven't seen better natural language parsing so much. Maybe Google has it hidden away somewhere."

"But really the hardest problem is expanding the storytelling nature of the game. No one has cracked that. Not yet. People keep trying. There's sort of the pure dungeon crawl kind of thing – sandbox games and whatnot. On the other end, someone will write a small gem of a story that just happens to have an adventure game type of user interface on top of it. You can wander around. You can do this or that. In between those, you've got things like The Last of Us and The Walking Dead. Those are sort of railroad-track experiences. There aren't many ways you can go wrong. You can die every other move, but you can't really get terribly far off course and get stuck in a corner, even if you wanted to."

"There's no other story other than the railroad track story. And that problem has not been cracked. How do you have an open world game that's also a story, where the player really has agency, as opposed to being on a train track to the end of the game? People have done great runs at it, like The Last of Us, which is really a story more than an adventure game. But they haven't melded the open world nature of the original dungeon crawl type adventures with storytelling very well. That's hard. The reason they haven't is it's really hard to do that. Maybe there are people out there who are working on it, and people out there who try different approaches, but it hasn't been done yet. That's when you get the real virtual world kind of thing."

Jeremy wonders how Dave correlates the in-game storytelling experience with traditional storytelling experiences told through games like Dungeons and Dragons.

Bureaucracy in action.

"The kind of D&D I played was sort of a slight twist on regular D&D, which at that point was still in the boxes. It was the old, old, old D&D. The dungeon master who ran our group way downplayed the number parts. It was all about storytelling for him, because he loved to just talk and evoke the environment you were in and all that, instead of, oh, well, you have a +1 and he's got a -2..."

"All that numeric stuff really pushes you away from the story and into the nerdiness, if nothing else. I mean, it's nerdy enough without the numbers, but it gets even nerdier with it. That was a good D&D sort of background in terms of trying to create a story, instead of trying to just be obsessive about the numbers."

That gets me thinking. I ask Dave who he was writing Zork's story for. After all, in those early days, there wasn't exactly much of a consumer audience.

Dave explains, "What happened is that because the machine we were on had a bunch of games on it, people knew that there were making interesting things on our machine. There was no security, and you could just make a name, log in, and look around to see what was going on. They would see us running a thing called Zork every now and then. They searched for it and when they finally found it, they'd run it. So we never had an announcement saying, here's this new game you can play. They just saw it on the machine. Oh. Zork? I wonder what that is. They typed Zork and then bam, up it would come."

Deadline feelie - a coroners note.

That almost feels like a viral game to me. I ask Dave whether he agrees.

"Oh, it was absolutely viral," he nods. "Adventure was also totally viral. It just spread through the ARPAnet in source form. People compiled it."

Jeremy interjects, "At what point did the team realize there was commercial potential in what they were creating?"

"When we began to talk about starting Infocom, which was not originally intended to be a Zork company, a games company... it was intended to be just a company that had a lot of smart people in it, and it was going to do something. We spent a lot of time trying to decide what it was that we were going to do. Zork ultimately became it, because we kept kind of saying, you know, Zork, we could do Zork commercially. And then people would say, but it won't fit on a micro. Yeah, yeah. And then Marc and Joel figured out a way to make a substantial fraction of it, almost half of it, fit on a micro. At that point we said, wow, here's this thing, and it's effectively done. All we have to do is port it to this new language and we can publish it and make some money. So we did."

"However, that first version was on the DEC PDP-11. I think we sold 20 copies. That was not intended to be a big moneymaker. The second was the Trash-80! Third was the Apple II. That came out just a couple months after the Trash-80 version. From then on, just more and more and more. Commodore, Atari. IBM PC when it came out. Macintosh. Bam, bam, bam. Amiga. All the other things."

Leather Goddesses of Phobos was graphically illustrated in what would now be called pixel art style.

Zork was released at a time when the Atari 2600 was only just on the shelves, and arcade games were still primitive. Comparatively, Zork was exceptionally sophisticated and complex. Jeremy asks Dave why that was the case.

"I don't want to do too much preening," says Dave, "but the fact that we were all literate, voracious readers helped. And we were also consumers of popular culture. One of the expeditions we had in '77, during the period when we were thinking about Zork, was to go see the first showing of Star Wars in Boston. A bunch of us trooped over to see it, because we had seen... I think we had seen a trailer for it on TV, that made it look like it was actually going to be an intelligent science fiction movie, really the first one since 2001: A Space Odyssey."

Jeremy points out something else that Zork had that was new to gaming – a geography that you could somewhat freely traverse, depending on the puzzles. There was persistence, not only of the world, but also the items within the world.

Leather Goddesses of Phobos scratch and sniff feelie. Yep, you read that right.

"Geez. I hadn't really thought about that that much,'" responds Dave. "Part of it is just because, again, all preening aside, we were all very skilled programmers, and skilled software designers. We weren't just programmers. We were software architects or whatever the appropriate phrase is. That meant we could come up with ideas and write."

"We were also working in a high-level language. Lucas started writing 6502 assembler for all their games, and eventually they invented the SCUMM scripting language. But we had a high-level language that was incredibly expressive and powerful and easy to produce code in, compared to more limited 6502 code. So that, I think, helped. If you had an idea, you did not have to write page after page of repetitive 6502 code, or later 8080 code, or any of the other CPUs around then. I think that helped a lot."

"A good computer language, plus of course the knowledge of how to use it properly and the training to use it properly helps an enormous amount in writing these games. That's obvious. It's a truism. But you don't always think about that. I don't always think about that. I've been lucky enough to be surrounded by a lot of incredibly intelligent people in my career, and so I tend to just think that's the norm, that everyone's really smart. And a lot of people are really smart. It's nice to go in with the assumption that the person you're talking to is pretty smart. If it turns out they aren't, well, that's too bad, but if it turns out they are, then you haven't insulted them at least."

But while Dave might not have insulted anyone, I note that Zork did have a certain level of sarcasm.

"No! Really? Really?" says Dave extremely sarcastically.

Hitchhiker's Guide to the Galaxy feelies.

Was that a reflection of the way the team worked together, or something else?

"We weren't that nasty to each other in person, but we were all kind of sarcastic, snarky people. That got into the parser. Although it's not really just the parser that does that, obviously. If you try something, sometimes it'll call you a cretin or just insult you. If you tried to swear, it would say something like, such language from a supposedly refined adventurer! Every time we came up with a vanilla response, we said, ooh, that's not a good place for a vanilla response. Let me put in something interesting. The something interesting turned out, invariably, to be something snarky."

"We were trying to make it a non-mechanical experience. Here you are, sitting in front of this thing, this absolutely mechanical, absolutely predictable – aside from the occasional pseudo-random number – thing. It could be really boring. We wanted to liven it up. We wanted to make it a more interesting experience. The goal of all these games is to have a relationship with the player. You're teaching the player how to play it at the same the player is learning how to play it, and also how to solve the puzzles. And so what you're really trying to do is kind of manipulate the user. It's a terrible thing to say. Manipulate the user into thinking that the game is omniscient, that it can do anything. Now, in fact, what you're really doing is training the user to know the things that it can do. Because the first thing people do when they sit down and see that it takes English sentences is they interpret that as you can type in some long, 20- or 30-word discursive sentence and it'll understand it – but it won't. So you want to train them to do the things that work, and you want to make it an experience that doesn't sound like you're talking to a computer. You're talking to another person. Maybe an unpleasant person that you wouldn't like to know in real life, but a person. That's the idea."

Classic Infocom magazine advertisement.

The game-with-personality concept was a completely new idea in the late 70s and early 80s – a pre GUI/Windows period when computers ran CLI prompts that were almost unfathomable to the layperson. In some of the adventure games, it definitely felt like the game was giving the player prompts that were actually quite specific, helping give them an idea on what kind of input was required at a given time. The classic, "Do you want to go south, north, east, west" kind of thing sounds simple, but was such a huge step up from the "c://", or even just "_" prompts that computers usually presented to users. I ask Dave about the thought processes that drove Zork's more human-sounding prompts.

" Helping the user was part of it," he replies, "but the reason we were trying to help the user is because we were emulating the behavior of a D&D dungeon master. At least the way my D&D group played it. We did not have a tabletop with little pewter figurines you moved around. No. We were just sitting around a table and the dungeon master would describe the situation. He would tell you that there are exits in certain directions. He would tell you what you see, from a casual observation of the room you were in. He would provide a little color to make it exciting. There are giant splotches of blood on the walls, perhaps, or something like that. Then you say what you do, and he does the same thing again. So it was really an upshot of being a mechanical dungeon master that did that. And of course immediately people would start drawing a map of their own, and that was part of the fun.

Switching the subject, Jeremy moves on to early Infocom games' packaging, which often featured prop-like items pertaining to the game – which Infocom called "feelies." They are the earliest examples of what would now be called "collectors editions," and these days, they really are collectible. Who came up with the idea for doing these?

Zork map

Dave ponders. "There were two stages, sort of, to how it happened. The first stage was just getting away from the original Personal Software Zork edition barbarian cover. The one that had this big guy swinging an axe – it was so cheesy.

"I mean, it's clear that the artist had either played the game or gotten enough of a description to do what he thought was a good job. It just somehow managed to come across as out of character. But in any other circumstance it would be an awesome cover. When we took Zork back from Personal Software, we decided to do our own packaging with the help of Giardini Russell, the ad agency we were using.

The original Personal Software version of Zork featured a cover that didn't quite capture the essence of the game.

"The second stage of getting fancier packages came because of Deadline, which was Marc's game that he was working on at the same time I was working on Zork II. Deadline was inspired by the... I guess you'd call them mystery games, which were produced by a then very famous British author, Dennis Wheatley. He wrote maybe a dozen of those. They were a lot like Deadline turned out to be. In the game box were items that were relevant to the case – printed dossiers and things like that. The one I remember the best was called The Malinsay Massacre. There were dossiers, police reports, all the same kind of thing that ended up in the Deadline folio game."

"This was before the gray boxes that became our trademark. It was a big folio, and it had all the stuff, including the pills found near the body – which were Sweet Tarts – and various other things like that. It was done for two reasons. One was that it seemed like an awesomely cool thing to do. Marc really loved the way that Wheatley's game books, book games, or whatever you want to call them, had been put together. He said, can we do something like that? The reason it was such an attractive prospect is the same reason why we had to write the Z-machine. There just wasn't room in the game. It was all background information."

"When you read or write almost any mystery, you clearly see acres and acres of backstory. The secret story about so and so, his horrible past, all that kind of stuff. There's no room for that in 64K of Z-machine code, so we had to have this extra stuff. The idea of doing it like Wheatley had done it in these books was very attractive, and of course Giardini Russell did a great job of making it look like it came from the 30s and 40s. It worked out really well. Actually, it worked out so well, and so many people loved it – including all of us – that we said, we'll have to do that for all our other games now!"

The Starcross disc feelie. [Source]

"And so we did. We had the Starcross injection-molded flying saucer, and the Suspended injection-molded mask. Enchanter and Sorcerer had die-cut packages with flaps on them. The customers loved them. Now, the stores – there was no such thing as Amazon then, of course – the stores hated them. They liked the fact that our games kept selling, but if you set the Starcrossed saucer up on a shelf, it fell over. When it was brand new they'd hang it from the ceiling and did all kinds of weird things with it and it was wonderful. Same with the Suspended package. But once it became not the latest game, but one that was still selling that they just wanted to put on a shelf, it was a pain for them. They didn't like it. So we settled on the idea of a standard package and the box with the things in it, and the ability to open it so that you could see them, but not play with them. We had that for the rest of the history of Infocom."

Jeremy observes, "As someone kind of on the outside looking in, I was always really jealous of my friends who had computers back at the time. I was a huge fan of the Hitchhiker's Guide to the Galaxy books when they first came out, and when I discovered there was a video game, a friend of mine brought the whole package to school, and I was like, wow, I wish I had a computer that could play this. My family didn't one until the 90s, when I went off to college. I went back and played a bunch of those games then, and 15 years after Zork came out, it was still something that, despite the fact that it was just text, it still was engrossing and really evocative. But I kind of missed out on the feelies at that point, because all the games were all reissues in standard packaging."

Lurking Horror and its feelies

"You were playing Lost Treasures of Infocom or something," says Dave. "Or getting them somewhere else. Yeah. The nice thing about the all text format is that text is moderately timeless. If you have a graphics game, a game from five or ten or fifteen years ago seems like caveman scratches on a wall compared to the latest graphics. That, in a perverse kind of way, has helped keep our games alive. That, plus the fact that if you write an interpreter for something – an iPhone, an Android, Linux, a watch – you can play them today. The fact that they're text makes them easy to do. So that's been kind of fun, just seeing them come out. Every now and then I'll get an email from one of the other Infocom types saying, did you see this? They did it on this silly machine!"

Infocom created a number of adventure trilogies. Was that a tech decision, or a creative one?

"There was a core of stuff," says Dave. "There was the parser, the syntaxes, all the verbs. There was a bunch of code that was the normal handling of each verb. There was just a few other things like that. The code that handled timing and things burning out, like your candle burns out, your light burns out. And all of that tended to be pulled forward into the next game."

"Typically, the way it worked is that it was the next game of the same genre. So all three Zorks, pull, pull, pull. Deadline became a different taxonomy branch, and then the mystery games tended to draw from Deadline, because there had been some code that Marc put in Deadline that was a really good start to some of the things we wanted to do in mysteries. He did a cut at it in Deadline, and I did a cut at it in Suspect. Stu Galley did a cut at it in The Witness. We would just push that forward each time."

Hitchhiker's Guide to the Galaxy feelie - a council order to destroy Arthur Dent's house.

"The three Enchanter games? Boom, boom, boom – all the same substrate. We didn't have shared libraries. It wasn't like, say, Java or C++ where you could have a library and derive new classes from it or whatever. Partly because we had no room to do that. So what tended to happen is that people would start from the substrate, and then they would tweak the default responses to be appropriate for the game they were writing. Some default responses you don't touch. It's like, you can't take that. But other times you would put in something that was appropriate for the game that you were writing. If you were floating in space, it might say something like, "It's too far away." In some ways, that was better than having libraries, because it meant each game had its own personality. In fact, a really careful examination will show that sometimes the responses were appropriate for a previous game, but inappropriate for the game they are in, because whoever had pulled it forward to that next game had not correctly themed the tailored responses from the previous game in that genre."

French Leather Goddesses of Phobos magazine ad.

Text adventures were exceptionally popular during the 80's, but were eventually superseded by graphic adventures. Towards the end of their heyday, there were highly ambitious games like A Mind Forever Voyaging, which was almost a concept adventure. It was very high end, and delivered strong political messaging. How did Infocom end up there?

Dave explains, "Steve Meretzky had, and still has, a reputation as the funnyman of these sorts of things. I guess, like many comedians, he would really like to play Hamlet – and A Mind Forever Voyaging was Steve's Hamlet. It's a fun game to play. It expresses his political views, which are not the same as mine, but I'm sure other people thought, wow, this is really hitting them where they live. It was a great thing to try."

"The best thing about Infocom, before the very end of course, was that we were able to do this stuff. Being your own publisher gives you an enormous amount of freedom. "The suits," the marketing guys, the bosses and stuff, would come in and tell us, you really don't want to do that. They'd make us change the title of a game or something like that. But compared to today, we had incredible creative freedom. That was one of the things that made it enormously fun."

"We had a pretty extensive development process. Where possible, all the other writers would play the games. The game would have to be sold to the writers first. And then to marketing. The marketing people of course would come in from the other direction. You know, we have this possible licensing deal with so and so. Is anyone interested in it? And if everyone said no, then we didn't do it. Of course, when Douglas Adams was interested, that was a different matter. We got really excited by that."

With our time running out, I ask one last question to Dave. He'd mentioned in his GDC Postmortem that when the guys came together to create Zork, they didn't really know what game design was...

"And someday we'll learn," says Dave with a smile, before I can even finish the sentence.

Tags: Analysis

8 comments

Comments

Close