I met Dungeon in 1981, working with computers when working with computers meant something different than it does now.
Dungeon was a bigger, stranger descendant of a game called Adventure, also known as Colossal Cave. Adventure ran on the giant TOPS-20 machines that lived in universities and cost more than a house. Adventure was primitive and short and wonderful.
Dungeon built on it into something longer and meaner. A text world of two hundred rooms. A troll, a cyclops, a thief, a grue in the dark. And a white house at the edge of a forest with a locked trophy case waiting to be filled.
There were no graphics. You typed. Go north. Take sword. Kill troll with sword. That was the whole interface, and it was enough, because the real magic of the thing was the writing and the jokes. I never even finished it. I got maybe four hundred of the five hundred points, lost interest, had work to do. But it got into me.
Why novelize a computer game from 1979?
Because I never really put it down. When the VMS Professional magazine ran a map of the whole dungeon on its cover around 1983, that was it for me. I took the FORTRAN source and wrote my own Dungeon compiler in assembly language.
That’s another way of saying I loved the thing enough to take it apart and rebuild it. When my old workplace was getting rid of a MicroVAX, a powerful machine for its day, I asked if I could have it. They said sure.
When the VMS Professional magazine ran a map of the whole dungeon on its cover around 1983, that was it for me.Share on X
I brought it home and kept it running for years for one reason: to play Dungeon.
I eventually threw it out. Keeping an obsolete machine alive for no purpose but a game got to be too much. Later I found the game running in a browser and walked through it again for the pleasure of it.
So this was forty years of a thing I kept coming back to. One day the memory returned, clear and warm, and I thought I’d finally do something with it.
Forty years of coming back to one text game, from the MIT original to the novel.
What was early computing culture like?
A culture that’s gone and won’t come back.
In those days you were intimate with the machine. You understood it down to the bits and bytes because you had to. There were no code generators, no AI, none of the scaffolding we lean on now. You had a keyboard and you typed code, sometimes assembly language. You were controlling the machine directly. The monitor, the disk drive, all of it.
On top of that sat a high level language that wasn’t far from the metal, because the compiler turned it into assembly anyway. I loved that closeness. When something broke, it was yours, and you could go find it down in the bits.
Around all of that grew a community, mostly people out of universities who were in love with the technology. These were the people who went on to build the internet and the foundations of everything we use now. And some of them spent their time building a computer game about a troll and a trophy case. Dungeon is a monument to those people, and to a naive, almost idealistic love of the machine that the field has largely traded away.
I’m not mourning it, and I won’t claim it was better. Today has its own culture around AI, and I’m deep in it, a believer in what it makes possible. The old spirit was less commercial, more in love with the thing for its own sake. The new one is more about building the next product and becoming the next millionaire. Neither is good or bad. It’s just different, and Dungeon remembers the older way.
The book: Dungeon is a novelization of the 1979 game that became Zork, built from the original FORTRAN source, with two invented companions and a deserter at its center. A monument to a computing culture that no longer exists.
The culture: The people who built Dungeon knew their machines down to the bits and went on to build the internet. Dungeon is a monument to that lost, idealistic love of the machine.
How do you build a novel from source code?
Literally from the source. I still had the FORTRAN, so I handed all of it to an AI and asked it to produce an outline.
It did, but the process was a fight. The AI kept wanting to cut scenes and shorten things. I kept wanting them in. Getting the full outline built the way I wanted it was the hardest part of the project. AI wants to do what it wants to do, and you have to push back to get your book instead of its book.
There’s a neat symmetry there that I don’t find sad at all. I lived the era where you controlled the machine by hand, down to the disk drive. Then I built a monument to that era using the tool that ended it. That doesn’t trouble me, because I believe in what AI does. It takes the routine and the mundane off people. The turning of the same screw all day. The filling of the same spreadsheet. That frees them for something that matters.
AI wants to do what it wants to do, and you have to push back to get your book instead of its book.
I used it the way I’d use any powerful tool, and then I fought it where it needed fighting, and the book came out mine. Writers who hand a tool like that the wheel and accept whatever comes back are asking for trouble. The AI’s version of Dungeon was shorter and tidier and dead on the page, and if I’d let it win I’d have published its book with my name on the cover.
Why do first drafts trapped in one point of view fail?
It was boring. Dry. Trapped in one man’s head. The problem is baked into the source, because a game character has no personality. You supply the personality when you play, so the adventurer is just a point of view with a lamp and a sword. That works fine in a game, where you’re the center of it. In a novel it’s death.
I tried making the character more introspective, having him explain the world as he moved through it. All that produced was pages of telling, with no life in them and none of the humor.
I hated those pages. A man explaining his surroundings to nobody is the fastest way I know to lose a reader, and I was losing interest in him myself.
Then I did the thing that saved the book.
I added people. I added Troi, a seamstress from Miren who had wandered into the dungeon by accident. Not a romantic interest. Another human being to talk to. And I added Spot, a pewter-gray reptile of no clear species, because a man alone underground for three weeks needs something to not talk to. The moment they arrived, the book came alive.
Why did two invented characters fix everything?
Because it turned a monologue into a buddy story. Think of the buddy cop movies, the Lethal Weapon shape, where the whole engine of the thing is two different personalities rubbing against each other. That’s what Troi and Spot gave me. Suddenly there were people to talk to, temperaments to play off, friction to generate warmth and comedy and stakes. Three of them working their way through the dungeon’s dangers is a story. One man narrating his own inventory isn’t.
Spot does something extra, which is carry the humor, and Dungeon was full of humor. The catch is that it was humor of its time. MIT developers wrote those jokes in the late 1970s, steeped in the culture of that exact place and moment. If you don’t know the era, a lot of them sail past you.
The fix: A game character has no personality, and that kills a novel. Adding Troi and Spot turned a solitary monologue into a buddy story, and that’s the decision that made Dungeon come alive.
How do you keep jokes that are forty-five years old?
It’s the same problem as Monty Python’s Flying Circus.
So much of that comedy is wired into the British politics and culture of its day. Without the references you watch something bounce along and never learn why it’s funny. Dungeon’s humor is like that, locked to a place and a decade most readers never lived in.
I made a choice. I kept the original jokes, even knowing some readers won’t get them, and let Spot and the modern layer supply new humor that works today. My hope is that immersion solves it. A reader falls far enough into the book that the period jokes become atmosphere instead of a wall. And older readers who were near that era get a reward the younger ones won’t. It’s a decision with a cost, and I made it on purpose.
Why give a treasure hunter a war behind him?
Because every good novel gives its characters a backstory, and the game gives its adventurer none.
He’s a point of view who wants treasure. That’s a fine premise for a game and a thin one for a book, so I built him a person. He’s Corvin, a Legion soldier who refused an order at the siege of Halvenmark. He walked away and became a wanted man months before he ever found the dungeon.
I toyed with the obvious version first, a man who hears a rumor of treasure and goes looking, and it was dry and cliche. The deserter had more interest, and it did practical work too.
Ten years in the Legion is why this particular man can survive a place designed to kill people, and the irreversible thing he did before the story starts gives him the interior weight that makes him worth following into the dark. The desertion earns both his skill and his soul.
| Problem | Decision | Result |
|---|---|---|
| Building the outline | Handed the original FORTRAN source to an AI | The AI kept cutting scenes; pushing back kept the book his |
| A narrator with no personality | Tried making the adventurer more introspective | Pages of telling with no life and none of the humor |
| One man alone underground | Added Troi, a seamstress from Miren, and Spot, a reptile | A monologue became a buddy story |
| Jokes from late 1970s MIT | Kept the originals and let Spot add modern humor | Period jokes become atmosphere; older readers get a reward |
| An adventurer with no backstory | Made Corvin a Legion deserter from the siege of Halvenmark | Ten years of soldiering explains his survival and gives him weight |
The man: Corvin, a Legion deserter who refused an order at a siege, gives Dungeon the backstory the game never had. Ten years of soldiering is why he survives the dark.
My take
Game characters make terrible narrators. Dungeon didn’t work until Troi and Spot turned a monologue into a buddy story. If you’re adapting anything built for play, give your hero someone to talk to. I kept the old MIT jokes even though plenty of readers will miss them. That was a cost I chose on purpose. Readers who were near that era get a reward, and the rest get atmosphere if they fall far enough in.
Who is Dungeon written for?
Myself, the way I write all my fiction. My novels are a hobby with little commercial intent. I write the weird ideas I want to write, from berserkers that destroy galaxies to living galaxies that think and go to war with one another, across scales that make no sense, because I get to control what happens and decide how it comes out and experiment with whatever style I feel like.
Collision With Andromeda reaches across a hundred million years for the same reason, and The Eternal War marches an army through Hell for the same reason. I wrote them because I wanted to.
Dungeon is that impulse pointed at a warm old memory. A man walking back into a dungeon he first entered in 1981, and building it into a novel because it was fun and he wanted to. If you were there for that era, this book is a homecoming. If you weren’t, I hope it’s simply a good adventure with a deserter, a seamstress, a strange gray reptile, and a great deal of darkness. Either way it’s waiting in my library.
Dungeon is the novel that grew out of that old text game. A deserter named Corvin Cray walks east into a hole under a white house, and two hundred rooms wait below.
Frequently Asked Questions
- The History of Zork, The New Zork Times (Infocom), via the Interactive Fiction Archive
- Zork, Wikipedia
