Will Crowther wrote software for two networks that grew far beyond anything he intended. At Bolt, Beranek and Newman (BBN) in 1969 he was one of three programmers who wrote the program for the ARPANET’s packet switches, the machines that switched the traffic of the network that grew into the internet. A few years later he wrote a small FORTRAN game about a cave he had mapped in Kentucky, left it on BBN’s computer, and found on his return from holiday that it had spread across the network. That game was Adventure, and in Don Woods’s expanded version it became the model for the text adventure. “I never distributed anything,” Crowther said.
Fast facts
- Born: 1936.
- Education: MIT, BS in physics, 1958.
- Career: MIT Lincoln Laboratory (about ten years); BBN, including the ARPANET IMP team; Xerox PARC from about 1976; back to BBN in the 1980s.
- Game: Adventure, FORTRAN on the DEC PDP-10, 1975–76; expanded by Don Woods in 1977.
- Caving: cartographer for the Cave Research Foundation; drew a 1975 map of the Bedquilt section of Colossal Cave.
From Lincoln Lab to the ARPANET
Crowther studied physics at MIT, graduating in 1958, at a time when, as he put it, the computer was kept in the physics department “because they were the only people who were going to try and keep it running”. He spent about ten years at MIT’s Lincoln Laboratory writing real-time control programs: one pointed a large antenna, another ran a truck that bounced signals off the moon and passing satellites.
In 1968 BBN bid for the contract to build the Interface Message Processors (IMPs), the small computers that would switch packets between the sites of the ARPA network. Shortly before the proposal was due, Crowther joined the team as its senior software person. Over the first eight months of 1969 he, Bernie Cosell and Dave Walden wrote the IMP program for a modified Honeywell 516: Cosell the debugging and statistics code, Walden the code that talked to the host computers, and Crowther the code that handled the IMPs’ dealings with each other. BBN was promising a switch at least ten times faster than an existing design, and some people told the team it could not be done. “Dave and I sat down, worked out the algorithms, figured out that it was only going to take a hundred and fifty lines of code to process a packet through one of these switches,” Crowther recalled. “We actually sat down and wrote the hundred and fifty lines of code, and counted them, and then we knew.”
Routing, deciding which way each packet should go, was the part he enjoyed most. “The original RFP specified fixed routing. And I looked at that and said, ‘That’s going to be terrible.’ So Dave and I worked out how to do variable routing.” He was a co-author of the 1970 paper that described the IMP to the computing world, and later worked on the Pluribus, a multiprocessor IMP that ARPA wanted for more speed. Its reliability code, “most of which I wrote”, let an engineer “pull out any card or pull out any wire… and the thing would keep running”.
He left BBN for Xerox’s Palo Alto Research Center in about 1976 and later returned to BBN, where in 1990 he was working on multiprocessor designs. He warned his interviewer that year that “my grasp of times and history and such is pretty fuzzy. All these are plus or minus two or three years.”
Caving
Crowther and his first wife, Patricia (Pat) Crowther, were both cavers with the Cave Research Foundation, which explored and surveyed the caves of Kentucky’s Flint Ridge and Mammoth Cave. Pat had squeezed through a tight passage, about 35 feet long, near the end of the known Flint Ridge system. On 9 September 1972 the caver John Wilcox pushed on along that route to Mammoth Cave’s Echo River, within sight of the tourist trail, and so proved that the two systems were one.
Will was the foundation’s cartographer. In 1975 he drew a map of the Bedquilt region of Colossal Cave, a section of the same system. “As part of my role as cartographer, I had produced a large and accurate map of a small but complex part of Mammoth Cave, which featured seven levels, stacked one approximately above the other,” he told Retro Gamer. Caving was not even his first love: “In those days, the thing I cared most about was rock climbing.”
Adventure
After the Crowthers divorced, Will wrote a game for his daughters, Sandy and Laura. He had already written them small programs: a typing tutor, and an addition game that rang the Teletype’s bell for a right answer. At BBN he also played in a months-long Dungeons & Dragons campaign with colleagues. The new game combined the two. “It was always intended to be a game, not a virtual tour. I needed a backdrop, and I was cartographer for the Cave Research Foundation at the time. It seemed a good match. The role playing dungeons of the time were usually set underground, so it seemed right.”
He wrote it in FORTRAN on BBN’s PDP-10 during 1975 and 1976, typing at a remote Teletype over a phone line. The game’s cave follows the real Bedquilt passages; he described each junction and let the player choose a direction, and cavers who had never been to Bedquilt later said they could find their way round it from the game. He did not think of it as “Colossal Cave”: “Bedquilt is a small part of Colossal cave, with perhaps ten miles of passage. The Adventure game is set in Bedquilt… So I would have called it ‘Bedquilt Cave Adventure’. But someone else picked Colossal. Oh well.”
The source files he left behind, dated March 1977 and recovered in 2005, show a finished game rather than a cave simulation: fewer than 80 locations, a snake, a bird, a crystal bridge, axe-throwing dwarves and the magic words XYZZY and PLUGH. “I wanted a unique magic word, and was considering working for Xerox at the time, so words with letters near the end of the alphabet came to mind,” he said of XYZZY. Asked about PLUGH in 2006, he wondered whether Woods had made it up, though it is in his own files. What his version did not have was a score, a way to save the game, or an inventory command. Those came from Woods. “My notion of winning the game was just getting all the treasures back out of the cave,” Crowther said, “whereas Don realised it would be more fun to have a running score. Good idea!”
In 1977 Woods, a graduate student at Stanford, found the program on the computer at the Stanford medical centre. It named “Willie Crowther” as the person to contact, so Woods sent mail to Crowther at every host on the ARPANET until one reached him. Crowther gave him the source, and Woods’s larger version, with its puzzles, treasures and score, is the one that spread. “I think Don did a great job,” Crowther said. “It was very popular and became notorious for bringing university computers to a halt because so many copies of it were running on their main frame!” According to Woods, the two of them received only “some modest royalties” for endorsing a commercial version from the Software Toolworks.
Was it the first?
In 1983 Peter Gerrard called it “the first ever Adventure program” and “the origin of every other adventure game”, and later guides repeated the claim. Its influence is well documented. Scott Adams wrote in BYTE that he played the Crowther and Woods game for almost ten days before he “achieved the coveted score of 350”, and then set out to fit an adventure into a home computer; the Colossal Cave Adventure entry follows the same path to Zork and Roberta Williams’s Mystery House. What the sources support is that Crowther’s program set the form of the text adventure: a place described in prose, commands of one or two words, and a world that changes as the player acts. It was not the first computer game, nor the first program to hold a conversation in English, as the scholar Dennis Jerz has pointed out.
Crowther’s own account added a complication. In 1984 he told the writer Shay Addams that he had written the game in 1967 or 1968, before Dungeons & Dragons existed, and that he had earlier played Mirkwood, a Tolkien-themed game devised by a friend and never sold. The recovered files, his 1975 map and his colleagues’ memories all place the game in 1975–76, which fits his own warning that his dates were plus or minus a few years.
Crowther was matter-of-fact about the game. His daughter Sandy remembered him saying: “You know I’ve done all sorts of wonderful things in my career, it’s funny that the one thing I’m remembered for is ‘Adventure’.” In 2023 Roberta and Ken Williams released a new 3D version, Colossal Cave, meant, Roberta said, to show the world “that Will Crowther and Patricia Crowther and Don Woods had in their minds”.
Sources
- An Interview with William Crowther, OH 184, conducted by Judy O’Neill, 12 March 1990, Charles Babbage Institute, University of Minnesota (interview, web): MIT and the BS in physics, 1958; Lincoln Laboratory; the hundred and fifty lines of code; variable routing; the claim that it could not be done; the Pluribus; leaving BBN for Xerox PARC in about 1976 and returning to BBN; his warning about dates. https://hdl.handle.net/11299/107235
- The IMP Software Guys, “The ARPANET IMP Program: Retrospective and Resurrection”, draft of 2 December 2013 (web, secondary): Crowther added to BBN’s proposal team as a senior software person; Cosell, Crowther and Walden writing the IMP program over the first eight months of 1969, and their division of the work; bibliography of Crowther’s ARPANET papers. https://walden-family.com/bbn/imp-code.pdf
- F. E. Heart, R. E. Kahn, S. M. Ornstein, W. R. Crowther and D. C. Walden, “The Interface Message Processor for the ARPA Computer Network”, AFIPS Conference Proceedings vol. 36, Spring Joint Computer Conference, May 1970, pp. 551–567 (technical paper). https://www.walden-family.com/public/1970-imp-afips.pdf
- Retro Gamer issue 31, 2006, pp. 54–57: the making of Adventure, interviews with Will Crowther and Don Woods (the programs for his children, the Cave Research Foundation, the Bedquilt map and its seven levels, “a game, not a virtual tour”, the Teletype and “I never distributed anything”, “Bedquilt Cave Adventure”, Woods finding the game and mailing every ARPANET host, the score, the royalties, XYZZY and PLUGH).
- Dennis G. Jerz, “Somewhere Nearby is Colossal Cave: Examining Will Crowther’s Original ‘Adventure’ in Code and in Kentucky”, Digital Humanities Quarterly 1.2, Summer 2007 (web, secondary): BBN and the ARPANET; the Cave Research Foundation; the 1975 Bedquilt map; the divorce and his daughters Sandy and Laura; the BBN Dungeons & Dragons campaign; the 1975–76 dates and the month’s holiday; the rock-climbing and “Adventure” quotations; the 2005 recovery of the March 1977 files; no score, save or inventory command in Crowther’s version; “neither the first computer game nor the first program to emulate conversation”. https://dhq.digitalhumanities.org/vol/1/2/000009/000009.html
- Will Crowther, Adventure data file and FORTRAN source dated 31 March 1977, as recovered in 2005 (source code, web): the locations, the snake, bird and crystal bridge, the dwarves, XYZZY and PLUGH. https://jerz.setonhill.edu/intfic/colossal-cave-adventure-will-crowthers-original-source-code/
- Joshua Rapp Learn, “50 Years Ago, Cavers Connected Mammoth Cave and Flint Ridge”, Discover, 12 August 2022 (web, secondary): Patricia Crowther and the tight passage; John Wilcox’s 9 September 1972 connection to Echo River. https://www.discovermagazine.com/50-years-ago-cavers-connected-mammoth-cave-and-flint-ridge-43971
- Shay Addams, letter, Computer Gaming World issue 21 (vol. 5 no. 2), April 1985, p. 37: Crowther’s account of writing the game in 1967 or 1968, Mirkwood, and writing it “to entertain my children”.
- Peter Gerrard, Exploring Adventures on the Commodore 64 (Duckworth, 1983), pp. 2–5: “the first ever Adventure program” and “the origin of every other adventure game”.
- Bob Redrup, The Adventure Gamer’s Manual (Sigma Press, 1992), chapter 1: Colossal Cave as “the very first adventure game”.
- Scott Adams, “Pirate’s Adventure”, BYTE vol. 5 no. 12, December 1980, p. 192: ten days to “the coveted score of 350”.
- Ken and Roberta Williams, interviewed in “Q&A with Roberta & Ken Williams, developers of reimagined Colossal Cave, out today”, PlayStation Blog, 19 January 2023 (web, secondary): Roberta’s quotation. https://blog.playstation.com/2023/01/19/qa-with-roberta-ken-williams-developers-of-reimagined-colossal-cave-out-today/