Interview 2: Warren Robinett

What was your inspiration for Adventure and what were the roles you played in Adventure's creation.

Well, I created Adventure for the Atari 2600 video game console in 1978-79. I guess this was the first action-adventure video game. I was directly inspired by the original text adventure game (also known as "Adventure") which was created by Don Woods and Willie Crowther. I played it at the Stanford Artificial Intelligence Lab, where my friend Julius Smith was a grad student, in 1978.

I had just finished my first Atari 2600 cartridge when I played Woods' text adventure at the AI Lab. I thought I could do something similar as a video game, but my boss at Atari did not agree. We had only 4K bytes of ROM available in the video game, whereas the text adventure ran on a mainframe with hundred of Kbytes available. But I was stubborn and fixated on the idea, so I ignored him and created a prototype in a period of about a month, and then went on vacation for a month. When I got back, I found out that people in marketing liked it. So I continued working on it. It took about a year from start to finish, although I did another Atari cartridge (Basic Programming) concurrently with Adventure. I sort of got stuck in the design of Adventure, and went off and started Basic Programming, then finished them both at the same time.

I was the sole designer, programmer, artist, sound designer, and tester for Adventure. It may be hard to believe nowadays, but that's the way things were back then. It didn't take all that long to write enough assembler code to fill up 4K. The problem was jamming in enough functionality to do what you wanted to do. And yes, I wasn't that great of an artist -- my dragons looked like ducks.

How did you develop the game Adventure, and Adventure's character design?

Well, it was an adaptation of the text adventure game to video game format. The text game gave text descriptions of the room you were in and was controlled with text commands the player typed in, like "GO SOUTH" or "TAKE WAND". The Atari 2600 video game, on the other hand, could not display text, but could display little movable sprites and a low-res background, and could play sounds. It was controlled with a joystick with a single "Fire" button on it.

So the initial problem was how to do an adventure game, with rooms and objects, given those resources. My idea was to show one room on the screen at a time, use sprites for the objects in the game, and use moving off the edge of the screen to mean Go North, South, East or West. This required a cursor or figure of some kind to indicate your position in the room -- I ended up using just a small square because I saved the higher resolution sprites for the objects. I called this "The Man" or sometimes, the cursor. There had to be a way to pick up and drop objects, and for this I used collision (2D overlap) between the cursor and an object to mean "Pick up". I used the button to mean "Drop the object".

There was an early version in which the player could pick up multiple objects and therefore had an "Inventory" screen, analogous to how things worked in the text adevnture. But I felt it was unsatisfactory and inconsistent to leave the action of the game world to go into Inventory limbo, and relaized that if I restricted the player to carrying one object at a time, it forced him or her to make strategic choices -- do I carry the weapon or the treasure?

Then there needed to be someway that objects could interact, analogous using the Cheerful Singing Bird to drive away the Dragon in Crowther and Woods's Adventure. I again used video overlap to trigger actions -- in my game, touching the sword to the dragon killed it.

Those were the main conceptual hurdles to resolve how one would do an adventure game as a video game. Then of course I had to have a goal, and invent obstacles and objects and creatures.

The game was supposed to be a quest, to retrieve the Holy Grail and bring it back to the castle, but the Atari marketing department sanitized Hoily Grail into Enchanted Chalice.

The dragons that chased you were an early idea that seemed to work pretty well, but that by itself was not enough. It took me quite a while to work out how to do effective mazes, since the Atari hardware forced the left and right halves of the screen to use the same graphics. The dragon required an object which could be used against it -- the sword. If the dragon ate you, the player, then there had to be a way to restart -- hence the reincarnation. Reincarnating any dragons you had killed at the same time as you were reincarnated was a good touch, I thought. It was like being vulnerable in the game of bridge -- you had more to lose battling the third dragon if two were already vanquished.

The bat was added as a confusion factor to spice things up fairly late in the development.

Please describe Adventure's testing process.

I started in the spring of 1978 and released it to manufacturing in June 1979. I showed it to other game designers at Atari and also play tested it at home, at parties, back home in Missouri on Christmas vacation, and so forth. There were no forms to sign, no big secrecy at that time. Things were pretty relaxed.

How did you choose the name Adventure?

I called it Adventure because I was adapting the text adventure game to video game format. It turned out that the name "Adventure" was public domain due to Don Woods intentionally putting his text adventure in the public domain, so that was the name that got used.

You contributed the idea and implementation of secret hidden items and rooms within games. The process has proven invaluable to modern game designers. What was your motivation for the initial secrecy, which later evolved into a gameplay mechanic?

I'm glad you think it was invaluable. You could also call it devious, insubordinate, and self-promoting. The hardest thing for me was keeping the secret to myself for a whole year.

The background was that at that time, Atari had a monopoly, and did not choose to give the game designers credit on the published games. Each game cart was designed entirely by one person at that time, but we were anonymous. We also got no royalties, just a salary -- mine was $22,000 per year. So creating the secret room with my signature in it was a way for me to sign my work, and hey if I was going to do it, why not fill the screen with my name in flashing colors? It also fit in with the adventure game theme of rooms and objects. You don't really know what's out there until you go exploring, and some places can be really hard to get to.

My model in creating the secret room was the secret messages hidden in Beatle records ("I buried Paul") in the late Sixties, where you had to play the record backwards to hear the message.

So I released the code for Adventure in June 1979. It contained the Gray Dot, which let you get into the secret room, which contained my signature. Needless to say, I didn't document that part of the code too well. Asembler code is not that easy to casually read, anyway. Atari manufactured several hundred thousand Adventure cartridges, sent them to stores all over the world, and sure enough, some kids here and there did discover the secret room.

Many Atari programmers left Atari and formed third-parties, such as Activision. Many, such as yourself, chose to stay loyal to Atari. Please tell us about this particular 'troup.'

In 1979, there were about a dozen programmers designing games for the Atari 2600 video game console. Four of them left to start Activision, and another three or four started Imagic, both of which sold $50 million worth of video games their first year in business. All those guys were multi-millionaires, and what did that say about those of us who were left? We were out drinking one night, and after about six pitchers of beer, we decided to form the Dumb Shits Club. The requirements for membership were you had to have designed games for Atari and never made any money from it. The idea was conceived by Tom Reuterdahl and myself, with Doug Neubauer (not drinking) observing the spectacle. We later elected Jim Huether, my office mate at Atari, president of the Dumb Shits Club because he stayed there the longest of the original dozen.

At the time of release, what was so innovative about Adventure?

It was great time, because the programmers were free to try all kinds of ideas for games. The game hardware, the controllers, the game ideas were all evolving fast. Some innovations in Adventure were:

A. Multi-screen game world. It may be hard to believe now, but most early Atari 2600 video games (Combat, Basketball, Breakout, Video Olympics) took place on a single screen as the game world. In Adventure, I came up with a way to stitch together many screens to make a much larger game world.

B. Picking up and dropping objects. The idea of rooms and objects came from the text adventure game, but objects as visual icons that could be moved around spatially was rather different from the text version. Making objects do things when they touched other objects was kind of obvious, given the collision detection hardware of the Atari 2600, but it worked well and opened up many possibilities.

C. Autonomous creatures. There were four creatures in Adventure: three dragons and a bat. They were implemented as objects that moved around on their own initiating actions. Each creature had a subroutine associated with it that executed each time through the program's main loop, causing the creature to "decide" which direction to move, testing the nearby game world environment for other objects the creature responded to, and invoking actions, such as the bat stealing an object from the player's cursor, or the dragon eating the cursor.

Using animations to show the state of each creature also worked extremely well. The dragon, for example, had four states (Chasing, Biting, Swallowed Cursor, and Dead). Each state corresponded to a different bitmap image of the dragon.

The creatures were implemented with pure "Behaviorism" (Stimulus-Response) governing their movements. There were only two things a creature could do: go toward an object (chase) or go away from an object (flee). Each creature had a priority list of objects (or other creatures) it responded to, and whether to chase it or flee from it. Each creature responded to the highest priority creature that was in the same room with it (and ignored lower priority objects). This meant you could see creatures "change their mind" as new objects entered a room.

For an object that was chased, something had to happen when it was caught. This was controlled by a collision-detection test on the part of the chaser. In the case of the dragon, if it caught the player's cursor, a collision caused the dragon to go into the Biting state, and if there was still a collision occuring a few frames later, then the player was trapped in the dragon's stomach (and had to reincarnate). If the player had moved away during that short interval, then the dragon went back into Chase mode. Thus a long series of Chase-Bite actions could occur as the player and dragon moved around, ultimately culminating either in the player getting away (the dragon couldn't sense the player across room boundaries, but kept going towards the last sighting in the same room -- so it was possible to get away), in the dragon eating the player, or in the player killing the dragon if the sword was available.

In the case of the bat, when the bat caught the object it was chasing, it grabbed it (even if the player had been carrying it), and flew away with it, dumping the object it had been carrying.

D. No timer. This is not exactly an innovation, but more of a design choice. I wanted to emphasize exploration as opposed to a frantic race against the clock. I also hoped that this choice would make the game more appealing to girls.

E. Game simulation independent of what the player sees. The simulation code governing the dragons, bat, and other objects proceeded each frame independent of which room the player was viewing. This is in contrast to many games, then and now, that randomly generate objects near the player. In Adventure, there were always three dragons and one bat, and they were always somewhere in the game world -- no objects blinked in and out of existence when you weren't looking. This gave the creatures more complex and interesting behavior -- for example, the bat could get stuck in an isolated part of the game world ferrying two objects back and forth between two rooms. If you hadn't seen the bat for a long time, you might go searching and discover this pattern.

F. Random placement of objects at the start of the game. I did this only in level 3, the final level. The first level was intended to be easy, so people could get started. Level 2 was hard, but it had the same object placement every time you played it. I used the random object placement in level 3 for variety. I didn't want it to be like a puzzle, where once you've solved it, it's not very interesting to do it again, and I wanted to avoid that. The bat was also added as a confusion factor, to move objects around a bit, so that the game wasn't too predictable. (I did make a mistake in my random object placement code, and there is a 1 in 18 chance that the yellow key will start out in the yellow castle, making the game unwinnable. This only happens in level 3.)

As you may gather from the above, I think that randomness in a game is very strong medicine, and must be very carefully controlled.

How does one get to the secret room in Adventure, and how did the secret get termed 'Easter Egg?'

To get into the secret room, there was a 1-pixel object (the smallest possible little insignificant looking dot) that was the small color (gray) as the background. The Gray Dot was hidden in a little chamber that was surrounded by walls, and could only gotten to by using the Bridge object to cross the walls. Since it was the same color as the background, you couldn't see it until you ran into it and then saw it against the walls of the maze. If you took this object around throughout the game world, you might eventually discover that it let you get through one of the side walls near the Yellow Castle and into the secret room which contained my signature.

By the time the existence of the secret room became known to Atari, I no longer worked there. I figured they would expunge the secret room when they found out about it, but the manager of the Atari game designers at that time, Steve Wright I think it was, said that he thought it was kind of cool to have little hidden surprises in video games, like searching for Easter Eggs on Easter Sunday. So it was not expunged, possibly influenced by the fact it would have cost $10,000 to make a new mask for the ROM chip used in the Adventure cartridge. They did assign a programmer to track down where in the code the signature was -- he later told me that if he had been asked remove it, he would have replaced "Created by Warren Robinett" with "Fixed by Brad Stewart."

Describe the layout creation process for Adventure. Did technical limitations of the time dictate design?

The game world layout just evolved, based on figuring out how to make an adventure game work on the Atari 2600 hardware, and then figuring out interesting objects and creatures. As I remember, I first had the player's cursor and room-to-room motion working and a dragon that chased you. But with no walls, the network of rooms was sort of pointless, or vacuous. I guess I made a castle fairly early, and had the open rooms below the castle. One problem was that, with the hardware requiring the left and right halves of the screen to use the same graphics, that meant if you could get in from the left you could keep going off the right sid e, and then the same would happen in the next room, and the next, etc. Then I figured out I could use the two one-pixel-wide "Missile" sprites for narrow walls on the left and right sides of the screen. This made it possible to make reasonable regions in the game world that were made up of several screens, but had boundaries. (It also provided a clue for what the Gray Dot was for, because these walls changed color to match the objects brought into the room.)

It took me quite a while to figure out that I could make globally asymmetrical mazes from rooms which were all symmetrical. Once this became clear, then I made the four different mazes pretty quickly. The bridge object (for crossing maze walls) was a logical construct once the mazes existed. Once the bridge existed, then making mazes consisting of disjoint regions became a cool idea. You came in the front door of a maze, and if it had some disconnected regions, you could only get into them by using the bridge. So the Blue Maze was your garden-variety introductory maze. The Red Maze in the White Castle, which had two disjoint regions, needed the bridge to get to all of it.

The other two mazes were the kind I called "Catacombs." You could only see the walls of the maze in a little square around the cursor. This was my attempt to transliterate the Lamp from the text adventure game to the video game medium. You had to pick up and light the lamp in the text game or you couldn't see anything. I really wanted a circle of radiance thrown out by the lamp, but circles were hard to do with the blocky 2600 graphics, so I settled for a "square of radiance." Also, it was confusing to have to always bring a lamp object to those mazes (and remember, I had restricted the player to carrying only one object), so I made the lamp automatic in those mazes. So it wasn't really an object. I implemented the square of radiance with a video priority trick with the 2600 hardware by making the walls and background both gray, and putting a quad-size square orange sprite in between. This put the radiance behind, rather than on top, of the maze walls and gave me the effect I wanted.

So the maze inside the Black Castle was both a catacomb and had two disjoint parts. Since you could only see a piece of the maze at any one time, it wasn't obvious that there were places you couldn't get to. I made a very small chamber as the part that was hard to get to, and this is where I hid the Gray Dot.

I invented the Magnet as a solution to the problem that stuff you needed could get stuck in the walls if you dropped an object in the wrong place. No problem, just go get the magnet and suck that puppy out of the wall. (The bat could also retrieve objects stuck in the wall.)

What do you consider your greatest accomplishment concerning Adventure?

The whole thing, really. Doing it within the limits of 4096 bytes of ROM for the game program (including graphics and sound) and 128 bytes of RAM for the machine stack, program variables, and game state -- this was very challenging.

Do you have a particular memory concerning Adventure?

Riding my bicycle home 13 miles from Sunnyvale to Menlo Park at midnight, after hacking on Adventure for 12 hours, my mind going faster than my legs, trying to figure out how to solve the next problem in implementing the game.

The dragons in Adventure have names, but the bat does not. Why is this?

I named the Red, Green, and Yellow Dragons "Rhindle", "Grindle", and "Yorgle". The guy who wrote the manual, Steve Harding, liked those names and used them. I named the bat "Knubberrub," a name suggested by a German friend, Walburg Kicia, but I guess he thought that name was too weird, so the bat had no name.

In addition to Adventure, what other projects have you contributed in making?

I did the Slot Racers and Basic Programming cartridges also for the Atari 2600 console.

Then I got involved with some educators and we started the Learning Company in 1980, where I did Rocky's Boots, which was an educational game.

Rocky's Boots inherited quite a few ideas from Adventure, like room-to-room motion, and objects that could be picked up by the cursor.

Atari sold over 1,000,000 Adventure game cartridges. Did you ever think that you would be the creator of an item desired by so many people?

I was pretty excited about what I was doing when I created Adventure. I was aware that getting it published and distributed nationwide (or worldwide) was pretty cool, since my first game, Slot Racers, was already in the stores when I was working on Adventure. I didn't think about the sales numbers, specifically. Atari management did not ever tell the designers sales numbers when I was there.

Adventure is still played by countless numbers over two decades after it's production run had long ended? Is there anything that you'd to say to those faithful and appreciative of your hard work of so many years ago?

Is it really still played very much? Until I started hearing from people via the Internet a few years ago, I didn't think many people even remembered it, much less played it. I do have a 2600 emulator and a ROM image of Adventure on my PC nowadays, so I do know about that way of playing it.

Is there anything that you'd to say to those faithful and appreciative of your hard work of so many years ago?

Well, now that you mention it, yeah.

1. Thanks for the strokes. Any creative person likes to hear that his or her work was appreciated.

2. Would you like a sequel? I always wanted to do another one, but various things intervened. I'm not sure it makes much sense now, since the game business has changed so much. (I feel a bit like Homer, dressed in a toga, walking into a New York publishing house, asking if they'd like a follow-on to the Iliad and Odyssey. "Well Mr. Homer, there's not much of a market for oral ballads in Greek, nowadays. Do you do screenplays?")

3. My boss at Atari told me not to do it, but I started on Adventure anyway. That seems to have been a good decision. Maybe *your* boss is a numbskull, too.

One thing that you might be interested in is that I wrote a book about the design of Adventure and Rocky's Boots in 1983-84, but my publisher went out of business as part of the collapse of the video game industry. So I still have this manuscript from that era laying around. It was called Imaginary Worlds, but nowadays I would probably retitle it "Inventing the Adventure Game."