4
One, Two, Three, Four I Declare a Space War

On December 23, 1947, electrical engineer John Bardeen and physicist Walter Brattain demonstrated their newly invented solid-state amplification and switching device, the transistor, to their superiors at Bell Labs. Since the development of the first triodes, invented by Lee De Forest in 1906 but not perfected and put to general use until around 1912, electrical amplification had been achieved through use of the vacuum tube, in which a current is passed through a vacuum-sealed glass cylinder containing a coil from a negatively charged anode to a positively charged cathode.¹ In 1918, British physicists William Eccles and F.W. Jordan created the first electronic flip-flop using vacuum tube triodes, which could transmit information in the binary language of 1’s and 0’s as the current in individual tubes switched on and off. In 1932, another British physicist named Charles Wynn-Williams developed an electronic counter using vacuum tubes. Similar switching and counting devices carried instructions and facilitated mathematical calculations for computers from Colossus to Whirlwind.

While this method produced perfectly functional computers, vacuum tubes came with a host of problems. First and foremost, they generated a lot of heat, which led to increased power consumption and a tendency to burn out over long periods of use. They could also not be miniaturized passed a certain point and had to be spaced relatively far apart from each other due to their thermal output, so a computer built with tubes would always be large and bulky, and the more powerful the computer, the larger and bulkier it would be. The only way to create powerful computers of reasonable size, power consumption, and durability would be to manipulate the flow of electrons at the subatomic level instead.

¹ The first vacuum tube diode was developed by John Fleming in 1904 for use as a rectifier to boost a wireless signal for transatlantic communication. Lee De Forest was the first person to add a grid to a tube to create a triode that could serve as an amplifier, but he believed that a small amount of gas needed to be present in the tube, which hindered its function. Harold Arnold at AT&T corrected this flaw in 1912 so the company could use De Forest’s invention to amplify signals for long-distance telephone calls.

As early as the 1820s, scientists had observed that certain elements like germanium acted as conductors in some situations and insulators in others, but they did not possess the requisite knowledge of atomic structure to explain the behavior of these so-called “semiconductors.” By the 1930s, the related fields of quantum mechanics and solid-state physics had advanced sufficiently to tackle the problem. At Bell Labs, a team led by physicist William Shockley began studying semiconductors in 1936 as part of an effort to replace the relays and vacuum tubes employed in AT&T’s nationwide telephone switching system, but World War II interrupted their research. After the war, Shockley’s team returned to work, leading to Bardeen and Brattain’s breakthrough in 1947.²

Years of refinement followed before the germanium transistor could be mass produced reliably, but by the early 1950s it had reached an appropriate level of sophistication to begin replacing vacuum tubes in electronic products. Incorporating transistors into highly sensitive defense projects like guided missiles proved difficult, however, due to the low melting point and considerable electron leakage of germanium. A different element, silicon, promised both a better semiconducting capability and a higher temperature tolerance, but initially proved impossible to dope with the necessary impurities to transform it into a transistor. The solution to this problem came not from Bell Labs, but from a relative newcomer to the electronics field called Texas Instruments (TI).

Established by Clarence Karcher and Eugene McDermott in 1930 in New Jersey before moving to Dallas four years later, the company that morphed into TI originated as an oil exploration company called Geophysical Services, Inc. (GSI). After a failed attempt to expand into oil production, the founders sold the company to Stanolind Oil and Gas, but McDermott joined with R&D head J. Erik Jonsson, field exploration head Cecil Green, and crew chief H. Bates Peacock to purchase the oil exploration business on December 6, 1941, the day before the United States was dragged into World War II. The war put a halt to GSI’s extensive international oil exploration efforts, but Jonsson realized that the same technology used to locate oil deposits beneath the surface of the earth could be used to locate airplanes above it, and GSI became a major defense contractor by the end of the war.³

² Jeffrey Young, Forbes Greatest Technology Stories: Inspiring Tales of the Entrepreneurs and Inventors Who Revolutionized Modern Business (New York: John Wiley & Sons, 1998), 54–71.

³ T.R. Reid, “The Texas Edison,” Texas Monthly, July 1982, 104.

With the war’s conclusion, GSI established a new Laboratory and Manufacturing Division to expand its wartime electronics work in both the military and private sectors and tapped a former Navy procurement officer named Patrick Haggerty to run it. By 1950, Haggerty had grown the division’s annual sales to nearly $10 million a year. With manufacturing now a far more important part of the business than oil exploration, company executives realized the name GSI no longer fit the firm. They decided to change the name to General Instruments, which conjured up visions of the great electronics concerns of the East like General Electric. Unfortunately, there was already a defense contractor by that name, so the Pentagon asked them to pick something else. They chose Texas Instruments.⁴

As Haggerty explored avenues to expand the newly renamed company’s burgeoning electronics business, he quickly settled upon the transistor. Due to anti-trust problems with the federal government, Bell Labs parent AT&T decided not to keep the transistor to itself to avoid further government scrutiny and instead licensed the technology relatively cheaply to any firm willing to research the application of transistors in hearing aids, which had been a personal passion of company founder Alexander Graham Bell. TI took a license in 1952, but the company was far smaller than fellow licensees like Raytheon, Zenith, and RCA and proved unable to compete for clients. Haggerty decided the only way TI could remain competitive was to seize the technological lead, so he resolved to bring the first silicon transistor to market. To accomplish this goal, Haggerty hired the Bell Labs chemist instrumental in developing the doping process for the first germanium transistors, Gordon Teal. Teal created the world’s first silicon transistor in 1954, which propelled TI to the top of the transistor industry.⁵

***

The emergence of fast, reliable silicon transistors paved the way for semiconductors to replace vacuum tubes in computers. A new revolution in electronics was at hand, and Ken Olsen and Wes Clark wanted Lincoln Labs to be at its forefront. Born in Bridgeport, Connecticut, Olsen tinkered with radios as a teenager and received his first formal electronics training through an 11-week Navy course at the tail end of World War II. Matriculating to MIT in 1947, Olsen graduated three years later and pursued a master’s degree from the institution, during which time he joined the Whirlwind team.⁶

⁴ Ibid.

⁵ Young, 1998, 72–77.

⁶ Ken Olsen, interview by David Allison, September 28–29, 1988, National Museum of American History, https://americanhistory.si.edu/comphist/olsen.html.

One of Olsen’s jobs at Whirlwind was to construct a smaller version of the computer called the Memory Test Computer (MTC) in order to experiment with core memory configurations. In this project, he was aided by Clark, a physicist who had recently joined the Whirlwind team after a stint at the Hanford nuclear site in Washington State. Through their work on the MTC, Olsen and Clark became proponents of efficiently designed real-time computer architectures geared toward individual use by scientists and researchers in a lab environment, a new approach to computing different from the giant batch-processing mainframes controlled by a group of trained operators or even the real-time systems like SAGE dedicated to controlling larger systems. The emergence of the transistor, with its potential to be smaller, cheaper, and more reliable than the vacuum tube, appeared poised to transform this vision into reality.⁷

In 1954, Olsen and Clark mapped out the design for a machine they called the TX-1, one of the most ambitious computers conceived to that point in history. Fully transistorized, the TX-1 would have far outperformed the Whirlwind and its progeny at SAGE but would also have taken up nearly half an acre of space and required the largest core memory array ever built. Consequently, Forrester turned the project down as overly ambitious. Olsen and Clark returned to the drawing board and created a simpler design in 1955 to prove the merits of a fully transistorized computer. As this felt like a precursor to their TX-1, they dubbed it the TX-0. Olsen and Clark completed the TX-0 in 1956 and turned their attention back to adapting the technology on a larger scale, resulting in the completion of the TX-2 in 1958.⁸ Having served its purpose, the TX-0 was placed on permanent loan to the MIT electrical engineering department, where it attracted the attention of a newly minted professor named Jack Dennis.

***

An electrical engineer with bachelor’s, master’s, and doctor’s degrees from MIT, Jack Dennis was first exposed to computing in 1954 through a graduate course taught by Charlie Adams, the bouncing ball demo programmer helping Jay Forrester build the Whirlwind. Dennis enjoyed programming on the Whirlwind and soon learned that if he came in at night he could bypass the operators and work directly with the computer himself. Dennis’s experiences with Whirlwind instilled within him a belief that student programming experiments could be just as valuable as official programming tasks and that greater student access to computing should be encouraged.⁹

⁷ Young, 1998, 97–99.

⁸ Ibid, 98–100.

⁹ “The TX-0: Its Past and Present,” The Computer Museum Report, Spring 1984, 9.

In the fall of 1958, Dennis joined the MIT electrical engineering faculty and settled into his new office in Building 26. Uninterested in continuing his dissertation research, Dennis was briefly at a loss for what to do next. He found new purpose when he discovered the recently donated TX-0 being installed just down the hall in the Research Laboratory for Electronics (RLE). Dennis dedicated himself to improving both the computer’s hardware and software, most importantly by creating an assembler for the machine and developing a debugger called FLIT.¹⁰ He also sensed an opportunity to advance his philosophy of providing opportunities for unofficial student experimentation and resolved to recruit interested undergraduates to program on the TX-0.¹¹ Dennis knew just where to go to find students who would welcome a chance to have direct access to a computer, for as a freshman in 1949 he had joined a student organization full of electrical tinkerers called the Tech Model Railroad Club (TMRC).

Established in 1946 by 26 members of the MIT student body with interests in electrical engineering, model making, and scenery construction, TMRC maintained an elaborate HO gauge model railroad network in Building 20 on campus, a ramshackle structure built during World War II as temporary quarters for the MIT Radiation Laboratory. While most of its membership concentrated on constructing train cars, erecting buildings, and sculpting scenery, a small group of students called the Signals & Power (S&P) subcommittee developed and maintained the complex maze of wires, switches, and relays running the length and breadth of the railroad that controlled the entire setup.¹²

The members of the S&P subcommittee of TMRC all shared one trait in common: an insatiable desire to understand the inner workings of electrical and electronic technology so they could harness it for their own purposes. For many of them, this was their real reason for being at MIT, far more important than their coursework or their degree programs. As its members bonded under the tables of Building 20, they developed their own culture and their own language. Garbage, for example, was referred to as “cruft,” while an individual more interested in studying or following the rules than exploration and discovery was labeled a “tool.” When a member made a particularly profound discovery or accomplished an especially clever feat of engineering – especially if he had done so just for the hell of it – then his endeavor was designated a “hack.” For the members of S&P, there was no higher calling than that of the “hacker.”¹³

¹⁰ Flit was a popular brand of pesticide in the 1950s, so the name was a pun on the concept of debugging. A version of the debugger was later ported to the PDP-1 as DDT.

¹¹ “The TX-0,” 9.

¹² Steven Levy, Hackers (Sebastopol, CA: O’Reilly Media, 2010), 8–9.

¹³ Ibid, 9–10.

In late 1958, Jack Dennis approached several members of TMRC and offered to show them how to sign up for time on and program the TX-0 in the RLE. These students included future computer chess pioneers Alan Kotok and Bob Wagner; a freshman named Peter Samson, who had once tried to build his own computer from the relays he pried from discarded pinball machines after seeing a documentary on computers on Boston public television; and junior Bob Saunders, the current president of the S&P subcommittee. Many of these students had come to MIT specifically due to an interest in computers but had grown disappointed with their lack of access to the MIT Computation Center, guarded as it was by its white-coated IBM priests.

Jack Dennis arranged for these students to have unfettered access to the TX-0 during any period no one had signed up to do actual work, which usually meant overnight or on weekends. A half dozen or so TMRC members leapt at this chance and could soon be found slaving away over a flexowriter in furious all-night coding sessions, foregoing their studies and all semblance of a social life to play with their new toy.¹⁴ The language of TMRC naturally transferred over to their programming exploits, so they referred to the programs they wrote as hacks and to themselves as computer hackers.

The TMRC members developed several interesting, though largely frivolous, programs for the TX-0. Peter Samson programmed an Arabic to Roman numeral converter and a simple music program using the TX-0’s speaker. Alan Kotok interfaced an FM radio receiver with the computer’s analog-to-digital converter to create what he called the “Expensive Tape Recorder,” while Robert Wagner built a program he called “Expensive Desk Calculator.” They were also introduced to the computer’s capacity as a game machine through a program developed for public demonstrations of the TX-0 by faculty members Doug Ross and John Ward. Called MOUSE and deployed in January 1959, the demo allowed the player to use a light pen to draw a maze on the display and to place three pieces of cheese that a computer-controlled mouse attempts to collect before running out of energy.¹⁵ TMRC’s fascination with the TX-0 lasted until late 1961, when a new computer moved into the room next door, a groundbreaking machine called the PDP-1.

¹⁴ Ibid, 14–16.

¹⁵ John E. Ward, “MOUSE Preliminary Instructions” (memorandum MIT Servomechanisms Laboratory, January 16, 1959).

***

When Ken Olsen completed the TX-0 at Lincoln Labs in 1956, he felt he had proven a new era of computing was at hand, but the rest of the world disagreed. It was one thing, people said, to deploy a prototype real-time system in a controlled academic environment, but quite another to mass produce such a system for use in a real laboratory setting. To prove these naysayers wrong, Olsen left MIT in 1957 with fellow Whirlwind staff member Harlan Anderson to form a new company dedicated to bringing small, interactive real-time computers to market. They planned to call their company the Digital Computer Corporation but were unable to secure financing from investors leery of backing two academics attempting to challenge IBM. Pioneering venture capitalist George Doriot finally agreed to sign on, but only after Olsen and Anderson agreed to limit themselves to building test equipment and components like power supplies. Therefore, they named the company the Digital Equipment Corporation (DEC) instead.¹⁶

As core memory began supplanting CRT memory in the late 1950s, DEC experienced a surge in demand for its memory testing equipment and quickly exceeded its profit forecasts. Consequently, Doriot agreed to let the founders return to their computer plans. Designed primarily by Ben Gurley and first unveiled at the Eastern Joint Computer Conference in December 1959, the Programmed Data Processor-1 – more commonly referred to as the PDP-1 – was essentially a more powerful version of the TX-0 that could be equipped with an optional point-plotting display. While not nearly as capable as the latest computers from IBM and its competitors in the mainframe space, the PDP-1 only cost $120,000, a stunningly low price in an era when buying a computer would typically set an organization back a million dollars or more. Although sales only reached roughly 53 units, its arrival presaged a new era in which improving technology coupled with falling prices would lead to improved computer access.¹⁷

¹⁶ Glenn Rifkin and George Harrar, The Ultimate Entrepreneur: The Story of Ken Olsen and Digital Equipment Corporation (Chicago, IL: Contemporary Books, 1988), 9–15.

¹⁷ Ibid, 16–49.

In September 1961, DEC donated one of the first PDP-1 computers to MIT, which it placed next door to the TX-0. Although not installed until the fall, news of its impending arrival swept through the programming community that summer and attracted the attention of a university employee named Martin Graetz. An Omaha, Nebraska, native, Martin Jerrold Graetz, who preferred the moniker J. Martin and was known to his friends as “Shag” after once offering to tell a shaggy dog story as a freshman, originally came to MIT in 1953 to study chemistry, but ultimately failed to graduate.¹⁸ While he was not a particularly adept chemist, he found a new calling in computer programming thanks to his good friend, Wayne Wiitanen.

Raised in the suburbs of Detroit, Wiitanen originally planned to attend the Michigan College of Mines, but stellar grades and a slew of science awards gained him a scholarship to MIT. Wiitanen and Graetz first met as freshman through the MIT Outing Club and bonded over shared interests, spending hours together rock climbing, mountaineering, playing piano duets, and singing in the MIT-Wellesley madrigal group. Like Graetz, Wiitanen failed to graduate from MIT, but he became an experienced computer programmer through several part-time and work-study jobs starting in 1955. Therefore, Wiitanen secured employment in the MIT Meteorology Department in the summer of 1957 programming an IBM 704 computer for weather prediction. That fall, he and Graetz decided to room together and moved into a men’s cooperative called Old Joe Clark’s.¹⁹

In the late 1950s, the military draft was still in full force in the United States, and being of an age, both Wiitanen and Graetz learned in spring 1958 that they could expect to be drafted within 30 days. To forestall active military duty, both men joined an artillery unit as reservists and trained together for six months at Fort Dix, New Jersey, and Fort Sill, Oklahoma. With this training complete, Wiitanen took a new programming job with the MIT Electronic Systems Laboratory, where he worked under Doug Ross on the APT project on the IBM 704 and (later) IBM 709 computers. In 1959, Wiitanen moved on to the Littauer Statistical Laboratory at Harvard, where he worked as a systems programmer on yet another 704 system. By this time, Wiitanen and Graetz had left Old Joe Clark’s for an upstairs apartment at a rundown tenement located at 8 Hingham Street in Cambridge. A friend and fellow MIT alum named Dave Freeman rented the downstairs apartment, and the three decided to give the building the tongue-in-cheek name “The Hingham Institute,” a play on the nickname of their alma mater.²⁰

¹⁸ Martin Graetz, interview with the author, December 3, 2015.

¹⁹ Wayne Wiitanen, email message to author, May 16, 2015.

²⁰ Ibid.

While Wiitanen worked with computers, Graetz struggled to find work as a chemistry lab technician. After brief stints at Massachusetts General Hospital and MIT, Graetz found himself out of work for several months, so Wiitanen helped secure him a job at Littauer as a junior operator helping feed cards into the computer and retyping any that were mangled by the system. At the same time, Graetz immersed himself in the inner workings of the 704 and learned how to program in both assembly language and FORTRAN. In early 1961, Graetz was dismissed from Littauer, so to find new employment he called on Jack Dennis, whom he knew through their time together in the MIT science fiction club, and secured a job working on a diagnostic program for a new magnetic tape unit for the TX-0.²¹

Graetz learned about the imminent arrival of the PDP-1 while working for Dennis on the TX-0. Eager to put the new computer through its paces, he convened the Hingham Institute, consisting now of himself, Wiitanen, and fellow Harvard programmer Steve “Slug” Russell, to formulate a plan of action for creating a demo program on the PDP-1. Over tea one afternoon at 8 Hingham Street, the trio discussed the attributes of a truly great demo and determined that it must tax the target computer to its limits, never play out the exact same way twice, and engage the user in a pleasurable activity. Building on these basic concepts, Wiitanen further articulated the need for skilled user input and giving the player direct control over some object displayed on the CRT – perhaps a spaceship – and engaging in a thrilling activity such as exploring, racing, or fighting. All three were science fiction aficionados – Graetz had even been published in a pulp magazine once – and particularly enjoyed the space operas of E.E. Smith, intergalactic tales of war and romance full of melodramatic dialogue, sudden plot twists, and clichéd battles pitting good against evil.²² Therefore, Graetz and Russell honed in on the idea of a conflict between spaceships, and the concept for Spacewar! was born.²³

Despite playing a critical role in conceptualizing Spacewar!, Wayne Wiitanen provided no further development input for the game. As tensions between the United States and the Soviet Union over the future of Germany culminated in the Berlin Wall crisis of October 1961, President John F. Kennedy deployed additional military units to Western Europe and then called up several reserve units to fill their slots back in the United States. On August 10th, Wiitanen was ordered to report to Fort Bragg, North Carolina, on October 31, 1961, to fill a slot as a surveyor for an artillery unit, ending his involvement with the Hingham Institute.²⁴ His friend Graetz remained involved with the game, but he passed on handling programming duties because since finishing his tape drive diagnostic for Jack Dennis over summer 1961, he had been working at the MIT Electronic Systems Laboratory in a different building from the PDP-1. Therefore, the task of turning the plans of the “Hingham Institute Study Group on Space Warfare” into reality fell on its third member, Steve Russell.

²¹ Graetz, 2015.

²² E.E. Smith (1890–1965) was a chemist who wrote science fiction stories on the side. He is best known for his Skylark and Lensman cycles, written in the 1920s and 1930s, respectively.

²³ J.M. Graetz, “The Origin of Spacewar,” Creative Computing, August 1981, 60–61.

²⁴ Wiitanen, 2015.

Steven Rundlet Russell, known since high school as “Slug” for reasons never explained to him, was born in Hartford, Connecticut, but spent his high school years in Washington State after his father, a mechanical engineer, was laid off and turned to farming. Fascinated with trains from an early age, Russell immersed himself in electronics so that he could build more elaborate model railroads. Russell received his introduction to computing on a trip back east to visit his uncle, Harvard professor George Pierce, who took him on a tour of Howard Aiken’s electromechanical Harvard Mark I computer.²⁵ Pierce believed in the importance of a college education and later paid Russell’s tuition so he could attend Dartmouth, where he eventually fell in with AI pioneer John McCarthy.²⁶

McCarthy and Russell barely interacted while McCarthy was at Dartmouth, but after the professor moved to MIT, he contacted his former employer to restore an early electronic AI machine developed by Marvin Minsky called SNARC that had been loaned to some Dartmouth students and subsequently disassembled.²⁷ Russell became involved in the project, impressed McCarthy with his mechanical skill, and not only secured employment as a research assistant at the MIT AI Lab in the summer between his junior and senior years, but also the promise of a permanent position at the lab upon graduation. Russell failed to complete a required senior thesis and therefore did not graduate, but McCarthy still brought him on staff in 1958 to help implement the LISP programming language.²⁸

²⁵ One of the first digital computers ever built, the Automatic Sequence Controlled Calculator, informally known as the Harvard Mark I, was a joint project between Aiken, who articulated the basic design of the computer, and IBM that built it. Constructed between 1937 and 1944, the computer used relays as switching units rather than the vacuum tubes found in Colossus and ENIAC.

²⁶ Steve Russell, interview by Al Kossow, August 9, 2008, Computer History Museum, https://archive.computerhistory.org/resources/access/text/2012/08/102746453-05-01-acc.pdf.

²⁷ Developed in 1951, the Stochastic Neural Analog Reinforcement Calculator, or SNARC, was a neural net simulator designed to approximate a mouse learning the correct path through a maze.

²⁸ Ibid.

In his first two years at MIT, Russell was so focused on LISP that he did not pay attention to the hacking scene emerging on the TX-0. After work on the first version of LISP was largely complete, however, Russell befriended Alan Kotok and other student programmers from TMRC, which he joined in 1960. By 1961, Russell had burned out on LISP and left MIT for Harvard, where he became a programming consultant at Littauer and worked with Graetz and Wiitanen, whom he had previously met while living at Old Joe Clark’s.

Despite no longer being an MIT employee, Russell remained close to his former companions in TMRC and at the AI Lab and continued to frequent Building 26 to watch the hacking scene unfold on the PDP-1. During this period, he began discussing the Spacewar! concept with his TMRC friends and insisted that someone should implement it. Russell’s friends agreed the program was a great idea but argued that as the person who brought the idea to their attention, he should be the one to implement it. Russell did not want to put in the work, so he began making excuses for why he was not up to the task. The TMRC hackers continued to press Russell until he offered what became his final excuse: he did not possess the sine and cosines routines he would need to place and move the spaceships around the screen. An exasperated Alan Kotok responded by driving to DEC headquarters in nearby Maynard, locating the routines himself, and plopping them down in front of Russell. Realizing he had run out of excuses, Russell soon began coding.²⁹

On December 29, 1961, the point-plotting display for MIT’s PDP-1 was installed, finally allowing the computer to run graphical programs.³⁰ Coding on Spacewar! commenced in January 1962, and before the end of the month Russell was able to place and move dots around the screen. The transition from dots to fully rendered spaceships appeared impossible at first due to hardware limitations until Russell realized that because the points comprising the spaceship would always remain in the same relative position to each other, he only needed to calculate the angle once per frame and then implement code that rotated the entire grid as necessary. Two spaceships were duly designed that resembled the curvy spaceship of the Buck Rodgers serials, and the long and slender Redstone rocket. They gained the nicknames “Wedge” and “Needle,” respectively. Before the end of February, Russell finished the basic program, in which the two ships could accelerate, rotate clockwise, or rotate counterclockwise when the player flipped the appropriate toggle switch on the PDP-1. Flipping another toggle switch allowed the player to fire torpedoes that destroyed the opposing ship on contact.³¹

²⁹ Ibid.

³⁰ John McKenzie, Deposition, Magnavox Co. v. Bally Manufacturing Corp., No. 74-1030 (N.D. Ill), October 28, 1975, 87.

³¹ Graetz, 1981, 62.

With the basic functions in place, programmers from both the TMRC hacker community and the AI Lab began to play the game regularly and offered suggestions for improvements. Peter Samson disliked the random placement of stars on the monitor and incorporated an accurate star chart he labeled “Expensive Planetarium,” while AI Lab graduate student and TMRC member Dan Edwards felt the game lacked sufficient strategy and introduced a star in the middle of the playfield that exerted a gravitational field affecting the movement of the ships.³² These features were in place by March 25, 1962. The final major addition came from Graetz, who toiled away on his own in the Electronic Systems Laboratory to create a hyperspace function for the game. Conceived during the original brainstorming sessions between Wiitanen, Graetz, and Russell and inspired by E.E. Smith’s stories, hyperspace served as a “panic button” that could be used a limited number of times during a match to take the ship instantly out of danger, but would deposit the player in a random, and potentially even more dangerous, position on the screen. Hyperspace was added in May, as was a scoring mechanic designed to limit individual playing sessions.³³

After roughly four months in development, Spacewar! made its public debut at the annual MIT open house in May 1962, where the PDP-1 was hooked up to a second, larger monitor to facilitate spectator viewing of matches. While the game had originally been played by flicking toggle switches, it now sported custom-made control boxes installed on March 19, 1962³⁴ that featured two levers – one for controlling left and right rotation and the other controlling thrust and hyperspace – and a button for firing torpedoes. These were crafted by Bob Saunders, who scoured the TMRC stores in Building 20 for the necessary components to build the boxes in order to combat the “Space War Elbow” caused by hunching in front of the PDP-1 for hours at a time. Spacewar! proved a sizable hit at that open house and at many that followed as the decade progressed, with throngs of children forming long lines to take turns playing the game.

³² Due to runtime limitations, the torpedoes fired by the players’ craft were not affected by the star’s gravity well. Russell and friends decided these projectiles were “photon torpedoes” unaffected by gravity to explain this physics defect.

³³ Ibid, 62–66.

³⁴ McKenzie, 1975, 147–148.

***

While a handful of games were programmed in the 1950s and early 1960s, few of them achieved notoriety outside of the institution where they were created due to a lack of distribution capability. At the time, there were no widespread network infrastructures across which these games could travel and no easy way to port programs to new hardware systems, as they were generally written in the machine or assembly language unique to their computer of origin rather than in a high-level language like FORTRAN. Spacewar! overcame these difficulties to find a home in computer labs across the United States.

Three interrelated factors aided the initial spread of Spacewar!. First, due to Ken Olsen and Harlan Anderson’s ties to MIT, DEC practically served as a branch of the school, with MIT students coming and going constantly as they took internships, found employment after graduation, or just dropped by to hang out with the DEC staff. With its close ties to MIT, DEC learned about Spacewar! soon after its creation, and by 1963 the company had created an official brochure highlighting the computing power required to run Spacewar! while also providing demonstrations of the game to entice potential buyers of the PDP-1.³⁵

Second, Spacewar! was one of the most technically impressive programs yet created for the PDP-1 and used nearly every last instruction and every last ounce of processing power the machine could muster, so it garnered attention as a showcase for the PDP-1 beyond those interested in playing the game. Furthermore, as Spacewar! used virtually the entire PDP-1 instruction set, it proved a great final diagnostic program to ensure the computer was in working order and would therefore ship to new buyers already resident in memory.³⁶

Third, DEC followed a hardware-first approach and created little software of its own, so PDP-1 users were always on the lookout for new programs and formed networks to trade software. This led to the creation of the Digital Equipment Computer Users’ Society (DECUS) to support organizations that owned a DEC computer by staging conferences and publishing a monthly newsletter called DECUSCOPE featuring program libraries for DEC computers. Graetz presented a paper entitled “SPACEWAR! Real-Time Capability of the PDP-1” in May 1962 at the inaugural meeting of DECUS and encouraged interested parties to contact Steve Russell for a copy of the game.³⁷

³⁵ “PDP-1 Computer and Spacewar” (brochure, Digital Equipment Corporation, c. 1963).

³⁶ The story of Spacewar! being used as a diagnostic program has been widely reported in the secondary literature and claimed by both Steve Russell and Martin Graetz in interviews. However, there has yet to be confirmation of this practice from DEC primary sources, and it may well be apocryphal.

³⁷ J.M. Graetz, “Spacewar!: Real-Time Capability of the PDP-1,” DECUS Proceedings 1962: Papers and Presentations of the Digital Equipment Computer Users Society, 1962, 37–39.

While the efforts of both DEC and DECUS played a role in publicizing Spacewar!, in truth they were not primary factors in the game’s emerging popularity across the United States. Fewer than half of the 53 PDP-1 computers sold were equipped with a display, and not all those computers were as readily accessible as the computer at MIT. Furthermore, the mere availability of a computer with the game installed did not automatically equate to excitement for playing it. At the University of Michigan, the sole PDP-1 was housed in the basement of a physics building, and only a handful of programmers were even aware the game existed. Likewise, the PDP-1 at Harvard was used in conjunction with an electron accelerator in constant operation and was therefore rarely available to play the game. Still other PDP-1 systems were installed in secure government facilities, where security protocols would have prevented anyone playing the game even if they wanted to.³⁸ Ultimately, it was a small group of evangelists from the MIT hacker community itself that took the lead in establishing the game at computer labs around the country.

One of the first Spacewar! hotspots outside of MIT was defense contractor Bolt, Beranek, & Newman (BBN), where it was evangelized by multiple MIT alumni. The game proved so popular at the defense contractor that staff were forced to create custom control boxes because they were constantly breaking the switches on the PDP-1 during playing sessions.³⁹ Another hotspot was the University of Minnesota, where a 1963 MIT graduate named Albert Kuhfeld adapted the game. While Minnesota did not possess a PDP-1, Kuhfeld missed his Spacewar! playing days so much that he recreated the game on a CDC 3100 in 1967–1968. While his version was largely faithful to the original, Kuhfeld added a few features of his own, including limited ammunition and a cloaking device that would temporarily render the ship invisible, though both its thrust and torpedoes could still give away its position.⁴⁰ Kuhfeld submitted an article to science fiction magazine Analog describing his game, which was published in 1971 and represented one of the first instances of national exposure for Spacewar!.⁴¹ While Kuhfeld’s article brought the game some attention, however, the most important Spacewar! hub of all, even more prominent than MIT itself, was the Stanford Artificial Intelligence Laboratory (SAIL).

³⁸ Devin Monnens and Martin Goldberg, “Space Odyssey: The Long Journey of Spacewar! from MIT to Computer Labs Around the World,” Kinephanos, Cultural History of Video Games Special Issue, June 2015, https://www.kinephanos.ca/Revue_files/2015_Monnens_Goldberg.pdf.

³⁹ Ibid.

⁴⁰ Albert W. Kuhfeld, “Spacewar,” Analog Science Fiction Science Fact Magazine, July 1971, 67–79.

⁴¹ The first known national coverage of Spacewar! occurred on January 26, 1969 when Walter Cronkite played the game at the Air Force Cambridge Research Laboratory for an episode of a program called 21st Century (Monnens and Goldberg, 2015).

***

As John McCarthy pondered the creation of an intelligent machine at MIT in 1957, he realized that interacting with an intelligent program would leave the user with a great deal of idle time, as the basic idea was to run the program, review its operation, and then make improvements so it could reason more effectively. During the long periods when the programmer would be dissecting the performance of the AI, the computer could theoretically be performing tasks for other people, but practically speaking this was not possible because a computer could only handle one user and one task at a time. Out of these ruminations, McCarthy developed the basic idea for time-sharing, in which multiple users, each equipped with a personal terminal, could feed problems to the computer concurrently.⁴² Since none of the users would actually be exercising the processor continuously, McCarthy reasoned, it should be possible to allocate processing time to give each user the illusion that the computer is working on his task the entire time when the processor is actually rapidly switching between dozens, if not hundreds, of users. Though this would cause some inevitable slowdowns due to the need to juggle so many different processes at once, the greater efficiency of allowing multiple simultaneous users would more than make up for this deficiency.⁴³

In January 1959, McCarthy drafted a memo outlining his idea for time-sharing, which served as the basis for a feasibility prototype, the Compatible Time-Sharing System (CTSS), built by Fernando Corbató and operational on the IBM 709 in the Computation Center before the end of 1961. Based on the success of this experiment, a committee formed to discuss upgrades to MIT’s computing capability chaired by McCarthy recommended the Institute’s next computer be a time-shared system, but the university proved reluctant based on the cost of the project and ordered a feasibility study. Annoyed at the university’s recalcitrance, McCarthy opted in September 1962 to accept the offer of a full professorship from Stanford University, where he established SAIL.⁴⁴ Meanwhile, Steve Russell, who lost his draft deferment when he left the AI Lab for Harvard, spent six months in the Army Reserve in 1962 and returned to Littauer only to discover it had come under new management. Unimpressed with his new boss and ready for a change, Russell accepted an offer from McCarthy to join him at Stanford, and he brought Spacewar! with him.⁴⁵

⁴² McCarthy, 2007.

⁴³ McCarthy was not the first computer scientist to publicly propose a time-sharing scheme, that was Bob Bemer at IBM, but McCarthy can take credit for being the earliest theorist on the subject to contribute to a working system.

⁴⁴ Ibid.

⁴⁵ Russell, 2008.

Once he arrived at Stanford, McCarthy wasted no time in securing a grant from the Advanced Research Projects Agency (ARPA) to implement a time-shared PDP-1 system to form the heart of his new AI lab. Thanks to Russell, Spacewar! was one of the first programs up and running on this system, and by the end of 1963 the game was so popular with the programmers at SAIL that an edict had to be issued banning play during regular business hours.⁴⁶ In 1966, the lab moved to the D.C. Power Laboratory in the rolling hills above Stanford and upgraded to a PDP-6, so Spacewar! was adapted to run on the new machine as well. The next year, the SAIL programmers created a special time-sharing setup called “Spacewar Mode” that optimized the operation of the PDP-6 so it devoted just enough processing power for Spacewar! to run smoothly without disrupting any research being conducted on the computer concurrently.⁴⁷

In 1971, a graduate student at SAIL named Ralph Gorin began modifying Spacewar!, now running on a combined PDP-6/PDP-10 setup, so that five players could play at the same time. He also added new features like mines and the ability for ships to survive multiple hits.⁴⁸ This version attracted the attention of noted counterculture figure and computer evangelist Stewart Brand, then working on a feature story on the potential of computing for Rolling Stone magazine. Brand convinced SAIL to hold a competition on October 19, 1972, called the “Intergalactic Spacewar Olympics” using the Gorin version of the game, most likely the first organized video game tournament. Brand published an account of the proceedings in the December 7, 1972, issue of Rolling Stone, bringing computer gaming into the cultural zeitgeist for the first time.⁴⁹

⁴⁶ Edward K. Yasaki, “Computing at Stanford,” Datamation, November 1963, 43–45.

⁴⁷ Monnens and Goldberg, 2015.

⁴⁹ Stewart Brand, “Spacewar: Fanatic Life and Symbolic Death among the Computer Bums,” Rolling Stone, December 7, 1972.

⁴⁸ Ralph Gorin, interview with the author, September 19, 2017.

***

When Steve Russell and the TMRC hackers first created Spacewar! in 1962, they did it purely for the challenge and thrill of taxing the PDP-1 to its limits, but once the game proved popular, they were not blind to the potential commercial implications of their hack. For a heartbeat, they considered how they might profit from the game, but quickly realized there was no way to turn a program that ran on a $120,000 computer into a consumer product. For almost a decade, the game remained trapped in the computer labs of MIT, BBN, Minnesota, and a host of other institutions until in 1970, it was discovered at SAIL by a recent electrical engineering graduate named Nolan Bushnell.