Home | About | Collections | Stories | Help! | News & Links | Friends | Lets Talk! | Events & Visiting | Search

Stories from the Maze War 30 Year Retrospective
David Lebling's Story of Maze at MIT (1974+)

David Lebling's Story of Maze at MIT (1974+)

Having read Steve Colley's and Greg Thompson's contributions, let me add mine.
The Imlac PDS-1s at the MIT Dynamic Modeling group were primarily used as terminals for our ITS PDP-10 machine. They had been modified to permit up to a 50kb transmission speed, and a lot of editing was originally done locally on them and then transmitted back to the "mainframe." Jack Haverty wrote a program called Imedit that coordinated pushing bits of a file back and forth that way (so that to the user it looked a lot like today's editors), and it was the state of the art in graphical editing at DM for a long time.
I first got involved with the Imlacs by modifying the ITS assembler, Midas, to output Imlac binaries. I also wrote some other utility programs for them. Eventually I became "the Imlac guy" at DM.
When Greg Thompson appeared at DM in 1974, he brought with him a wealth of paper tapes containing some of the Imlac games that he had encountered at NASA Ames. Among these was Maze. He asked, "Can we get this working on ITS?" He took the Imlac side as his bailiwick, I took the ITS side. The program used shared memory on ITS to store the moves, and each Maze process was responsible for updating its input buffer and transmitting the other buffers (the other active players) to its Imlac. There could be up to eight independent players. It was theoretically possible to have multiple games going at a time, although one game was enough to significantly load the PDP-10. (Ken Harrenstien later improved the communication protocol significantly, which eased the bandwith requirements.)
The game had robotic as well as human players. I wrote a very simple robot player that wandered through the maze looking for victims. It peeked around corners, decided which turns to take probabilistically, and turned around periodically to check its back. Robots adjusted their play based on their score, slowing down if they were beating you too badly. "You were shot by ROBOT1" was an all-too-common message that resulted.
To say that the game was an immediate hit would be an understatement. We had six or eight Imlacs, and during off-hours they were almost all playing Maze. People would come down from the AI Lab to play, or bring their friends in to play with them. Some of the best people recruited by the DM group were originally attracted there by these midnight Maze sessions.
Two unfortunate results of Maze's popularity was that often the Imlacs would be left in a crashed state, which annoyed our "real" users, and in addition Al Vezza, our group leader, was generally worried about anything that DARPA might take a dim view of, and games were near the top of the list. So, Al tried to get the game removed from the system (which would never work, as it was on the other MIT machines as well), and asked me to write a "Maze Guncher," a daemon program to look for running copies of the game and cause them to crash. This was an amusing waste of time, as the ITS system was totally insecure (by design), so anyone who wanted to play could either disable the Guncher or change the signature in the game which the program looked for. Needless to say, both of these methods of enabling Maze were used almost instantaneously!
It is also worth mentioning that Al Vezza and J. C. R. "Lick" Licklider (the director of MIT LCS) were occasionally observed playing Maze, in spite of their official opposition to it.
Other additions that were made to the Maze universe included Tak To's "Maze Watcher," which displayed a top-down view of an ongoing game on the Evans and Sutherland LDS-1 display computer (people would cluster around it and cheer on their friends), and the Maze Editor, which I wrote to allow some variety in which maze to play in (though most people used the "traditional" one).
Early in its life, we played Maze across the Arpanet with players at USC (where there were also "MIT-style" Imlacs), although the speed of the net in those days was so slow that they were at a huge disadvantage. This may well have been the first "Internet" multi-player game.
When Infocom was founded in 1979 (this was the company that eventually published "Zork" and other text adventures), one of the products we considered building before we settled on adventure games was a version of Maze to be installed (like Pong and such) in bars and video game parlors.
-- Dave Lebling

Return to Maze War 30 Year Retrospective Event Page

Know any more about Maze? Contact us!

Please send site comments to our Webmaster.
Please see our notices about the content of this site and its usage.
(cc) 1998- Digibarn Computer Museum, some rights reserved under this Creative Commons license.