2
Shall We Play a (War) Game?

Between February 4 and June 8, 1952, the RAND Corporation administered a training exercise codenamed “Casey” involving 28 students from the University of California, Los Angeles (UCLA). The exercise served as the first test for an initiative called “Project Simulator” designed to teach the staff of Air Defense Direction Centers (ADCCs) how to identify enemy aircraft on radar displays and scramble interceptor jets to shoot them down. In many ways, Project Simulator differed little from exercises conducted by militaries for more than a century save for one important distinction: the scenario was driven by an IBM 701 computer that simulated the flight paths of the aircraft trainees were required to identify, pinpoint, and interdict. As the computer charted the paths of the planes over the course of two hours, an IBM 407 printer constantly produced new paper readouts that approximated an actual radar display.¹

Project Simulator was developed by a RAND group established in May 1951 called the Systems Research Laboratory (SRL) staffed by a mix of psychologists and mathematicians studying the impact of new technologies on the battlefield by focusing on the “human factors” present in man–machine interaction. SRL gravitated toward early warning systems due to growing paranoia regarding Soviet nuclear strikes using long-range bombers. Casey proved successful, so in early 1953 SRL conducted a second round of trials with Project Simulator codenamed “Cowboy,” this time with active-duty military personnel.²

¹ Willis Ware, RAND and the Information Revolution: A History in Essays and Vignettes (Santa Monica, CA: RAND Corporation, 2008), 94–98.

² F.N. Marzocco, “The Story of SDD,” SD-1094, October 1, 1956, 1–2.

The U.S. Air Force was so pleased with the results of Casey and Cowboy that it deployed Project Simulator across the service and funded upgrades like a high-resolution camera and a CRT display to replace the printer readouts of the original system.³ In September 1955, work on the renamed “System Training Project” (STP) moved to a new division, the System Development Division (SDD), which by May 1956 had installed the STP at seven air divisions.⁴ In December 1957, RAND spun off SDD as its own company, System Development Corporation, which became a major defense contractor in its own right. While the STP was not a fully automated exercise and required a human team to administer the scenario and interpret the results, it represented the first training simulation constructed around a computer.

***

Like the AI researchers located primarily at major universities, the U.S. Armed Forces were quick to realize that computerized games could be a boon to their work, in this case determining effective strategies to thwart the Soviet military if the Cold War turned hot. The computer arrived at the perfect time in the evolution of military strategy, as over the course of World War II the nature of war gaming, which for two centuries had largely focused on tactical encounters between opposing armies and decision-making by field commanders, fundamentally shifted. The new buzzword in the think tanks of the emergent military-industrial complex was “operations research” (OR), a discipline focused on a commander making decisions to overcome an implacable simulation model rather than a fellow military leader. Once the earliest computers started coming online in the 1940s, the OR crowd quickly embraced the technology to run its simulations.

In the late 1950s, top business leaders, particularly those working for defense contractors, took notice of the complex decision-making models the military had developed and realized that they could model business markets and corporate organizations the same way. Thus, computer simulations migrated from the war room to the classroom, and America’s business students became some of the first laymen to challenge each other in computer-mediated competition. As with the chess programs, these military and business simulations were not entertainment products, but the individuals being trained through these programs were among the first in the world to compete through a computer game.

³ Ware, 2008, 97.

⁴ Marzocco, 1956, 8.

Abstract depictions of warfare are nearly as old as civilization. One of the most celebrated of these games is chess, which developed in Medieval Europe on the model of an earlier Indian game called chaturanga that dates back to at least the 700s. While chess and its many variants became popular across Europe, they took on particular significance in the Holy Roman Empire, where in 1616 Duke Augustus the Younger of Braunschweig-Lüneburg wrote a treatise under the pseudonym Gustavus Selenus entitled Das Schach-oder König-Spiel positing chess as not just a game, but as a tool for instruction in war and statecraft.⁵ Over a century of conflict in Germany reinforced the need to train the princes of the empire how to wage war effectively, and chess variants proliferated throughout the period.⁶

In 1780, Johann Christian Ludwig Hellwig, Master of Pages for the city of Braunschweig, published the rules for a game intended to educate his young charges in the prosecution of war that adopted the basic configuration of chess, but evolved the game in a new direction. Instead of an abstract board, Hellwig’s playfield consisted of squares representing different terrain types with their own rules. He maintained the traditional chess pieces but divided them among the contemporary military branches of infantry, cavalry, and artillery alongside several new pieces. In 1803, Hellwig published a revised version of his game that eliminated traditional chess pieces altogether in favor of contemporary military formations under the name Das Kriegsspiel, which literally translates as “The War Game.”⁷

Unlike chess and similar abstract strategy games, Das Kriegsspiel attempted to simulate real battlefield conditions by taking into account terrain, the facing of units, the varying capabilities of military formations, and other elements of warfare. The game spread slowly, but steadily after its initial publication, and eventually new games based on the same principles appeared as the kriegspiel concept gained adherents across Europe. Its most important convert was one of its first, a minor Prussian noble and civil servant named Georg Leopold von Reiswitz.

The son of a military veteran who settled in Silesia after the War of Austrian Succession, Reiswitz was instructed in both chess and warfare as a youth, but was forced to forego military service after suffering a serious arm injury. He adored military tactics, however, and was introduced to Hellwig’s game by his friend the prince of Anhalt-Pless. Hellwig’s game was so expensive that only members of the upper nobility could afford to own a set, but Reiswitz grew so enamored with it that when he left Silesia to attend university at Halle, he introduced his friends to the rules and worked with them to construct a variant from cheaper materials.

⁵ Translated into English, the title of the work is The Game of Chess or Kings.

⁶ Jon Peterson, Playing at the World: A History of Simulating Wars, People and Fantastic Adventures From Chess to Role-Playing Games (San Diego, CA: Unreason Press, 2012), chap. 3.1.1, Kindle.

⁷ Ibid, chap. 3.1.2.

In 1809, Reiswitz read a critique of Hellwig-style kriegspiel decrying the use of chess-like squares, which inhibited the realism of military maneuvers within these games. Inspired, Reiswitz reunited with one of his university friends the next year to create a new kriegspiel that eliminated a grid in favor of a sand table upon which terrain could be accurately modeled in three dimensions. Reiswitz demonstrated this game to the sons of Prussian King Friedrich Wilhelm III in 1811, who were so taken with it that a demonstration for the king himself followed in 1812, by which time the sand table had been replaced by modular terrain blocks.⁸

Although Reiswitz was initially motivated to create his game due to the expense of the Hellwig game, after its adoption by the Prussian monarchy, his game became more elaborate and far costlier. Therefore, Reiswitz never attempted to prepare a version for commercial distribution and never finished codifying the rules. After the Napoleonic Wars, Reiswitz’s son, a decorated Prussian artillery officer named Georg Heinrich Rudolf Johann, began further developing the ruleset with an eye toward increased battlefield realism. Completed in 1824, the younger Reiswitz’s kriegspiel introduced two important elements that continued to define war games in subsequent centuries: the application of probabilities in resolving combat and the appointment of one player as a “referee” to develop scenarios and arbitrate rule disputes.⁹ Reiswitz demonstrated his new game for General Karl Freiherr von Müffling, chief of the Prussian general staff, who adopted it as a training tool for the Prussian Army.

The popularity of kriegspiel waned in the middle of the nineteenth century as Europe experienced a prolonged period of relative peace, but after the accession of Wilhelm I and the appointment of his ambitious chancellor Otto von Bismarck, warfare became a central policy tool in their push to unify the German states under Prussian rule. Kriegspiel experienced a resurgence as the Prussian military conducted extensive war games while planning the campaigns against Austria and France that resulted in a unified German Empire.¹⁰ The swift and overwhelming success of these campaigns led other nations in the developed world to model many aspects of their own militaries along the Prussian model and spurred adoption of kriegspiel on a large scale.

⁸ Ibid, chap. 3.1.3.

⁹ Ibid.

¹⁰ Ibid, chap. 3.1.4.

Battlefield war games remained a popular planning and training tool through the end of World War II, but afterwards experienced a decline. Just as the association of war games with the Prussian kriegsspiel drove their adoption in the aftermath of German reunification in the 1870s, the central villainous role of the Third Reich in World War II discredited the war game amid accusations that war gamers were also warmongers. Additionally, some thought the dawn of the nuclear age rendered these exercises obsolete, as any major conflict that devolved into the use of nuclear weapons would wreak devastation over a larger area than any war -gaming table could capture.¹¹

Perhaps the leading cause in the decline of the traditional war game was the rise of OR in military planning during World War II. Unlike kriegspiel, which attempt to predict future success based on past decision-making, OR simulations rely on mathematical models to determine the most efficient strategies or tactics in a given situation. Rather than pitting two commanders against each other, OR games work from the premise that a commander encounters a fixed problem that must be overcome, is constrained in his decision-making by the rules of the game, and must run these decisions through a precise mathematical framework to determine success or failure.¹² The advent of the earliest computers only reinforced the military strategy community’s fascination with OR simulations, as these mainframes were ideal for running the mathematical models crucial to these games. Although OR as a discipline evolved out of radar research conducted by the British in the late 1930s, the rise of the United States as the central hub for computer research after World War II positioned that country as the hotbed for this new style of war gaming in the 1950s.

***

The emergence of the United States as the primary war-gaming innovator of the late twentieth century is somewhat surprising, as the U.S. military had previously paid little attention to the discipline. Before World War II, several American authors dabbled in war games, but none of them exerted much influence outside the United States. The first such games were introduced to the U.S. military in 1867 and were based on existing German kriegspiel.¹³ The earliest American treatise on war gaming, American Kriegsspiel by Major William Livermore first circulated in 1879 and divided war gaming into five disciplines with detailed rules for tactical encounters arbitrated by an umpire.¹⁴

¹¹ Ibid, chap. 3.1.7.

¹² Francis J. McHugh. U.S. Navy Fundamentals of War Gaming (New York: Skyhorse Publishing, 2013), chap. 2, Kindle.

¹³ Ibid.

¹⁴ Ibid. These disciplines were the tactical game, the grand tactical game, the strategic game, the fortification game, and the naval game. Livermore spent most of the book detailing the first two games and did not include rules for naval gaming at all.

In 1880, army officer Charles Totten published an influential American war gaming system called Strategos. Feeling that European kriegspiel were overly complicated and time-consuming for a U.S. military staffed by only a small core of professionals spread across a vast frontier, Totten fashioned his game into a series of escalating encounters designed to gradually acclimate players to the command of an entire army. Totten adhered closer to chess than many of his contemporaries, but in 1908 Captain Farrand Sayre published an influential tome called Map Maneuvers and Tactical Rides that replaced chess-like squares with maps. He also extolled the use of one-sided war games, in which the referee controls the opposing forces in addition to enforcing the rules, a forerunner of the OR approach that would gain popularity half a century later.¹⁵

Before World War II, the U.S. military primarily employed war gaming for training at staff colleges rather than for planning, and war games were especially valued by the U.S. Navy.¹⁶ Indeed, the Navy proposed the first computerized war game in 1945, the Navy Electronic Warfare Simulator (NEWS). The need for an electronic system arose from the evolving nature of naval warfare during World War II, in which carrier battle groups were displacing battleships at the heart of naval engagements. Constantly updating game boards to account for the latest changes in tactics proved difficult, so when staff at the Naval War College learned of a combat information center trainer being developed at the University of California, they petitioned the Chief of Naval Operations in June 1945 to develop a training program around it.¹⁷

The Navy Electronics Laboratory was inaugurated in San Diego, California, in November 1945 to begin work on NEWS. Initially, the Navy envisioned an electromechanical system that would speed up the process of moving ships and adjudicating enemy contact but would not perform any calculations. The NEWS team soon realized that adding a computer to the system would allow it to process complicated damage calculations but would require an extended period of research and construction. Consequently, while NEWS was the first computer-assisted war game conceived, it was not the first one to be completed and did not enter service at the Naval War College until 1957.¹⁸

¹⁵ Ibid.

¹⁶ Ibid.

¹⁷ Ibid.

¹⁸ Ibid.

The Navy also conducted the first OR war game in the United States at its Naval Ordnance Laboratory in 1942 under the auspices of Dr. Ellis Johnson, chief of the countermeasures section of the laboratory.¹⁹ The other military branches subsequently formed their own OR groups, which fostered a high degree of cooperation between the military, academia, and industry in the research and application of new technology. The effectiveness of these programs in developing new strategies encouraged the military to continue OR activities after the war’s conclusion via “think tanks” staffed by a mixture of civilian and military personnel.

Dr. Johnson established one of the earliest post-war think tanks in 1948, the Operations Research Office (ORO), which primarily served the U.S. Army and worked in conjunction with Johns Hopkins University.²⁰ From its inception, ORO desired to run simulations, though these were generally computer-assisted rather than fully automated programs due to the slow speed of analog and digital computers at the time. ORO began developing models for computer simulations as early as 1948, but complete programs did not emerge for several more years.

One of the earliest finished programs was a theater-level war game called HUTSPIEL first played in 1955 on an analog computer developed by the Goodyear Aircraft Corporation called the Goodyear Electronic Differential Analyzer (GEDA). Intended to study the use of tactical nuclear weapons and conventional air support in Western Europe in the event of a Soviet invasion, the game places one player in control of NATO forces in France, Belgium, and West Germany, while the other directs a Soviet force attempting to penetrate the region across a frontage of roughly 150 miles.²¹ At the start of the game, each player allocates his forces across the sectors he controls and chooses targets for his planes and nukes, which can consist of enemy troops, airfields, supply depots, and transportation facilities. GEDA then determines the results. In the original version of the game, the simulation would continue without human intervention until a player paused to issue new orders, but in subsequent versions play was divided into turns of fixed time increments.

¹⁹ Alfred H. Hausrath, Venture Simulation in War, Business, and Politics (New York: McGraw-Hill, 1971), 39.

²⁰ Ibid, 64

²¹ Joseph O. Harrison, Jr., “Computer-Aided Information Systems for Gaming,” RAC-TP-133, September 1964, 19–20.

In 1957, ORO completed a second computer-assisted game to model air defense called ZIGSPIEL. In 1961, the Army terminated the ORO and replaced it with a new organization dubbed the Research Analysis Corporation (RAC) that continued the work of ORO in computer-assisted war gaming and military simulation. In the early 1960s, RAC promulgated two additional computer-assisted war games along the model of HUTSPIEL called TACSPIEL and THEATERSPIEL. TACSPIEL evolved out of an ORO game called INDIGO I first played in 1958 and models ground warfare at the division level and lower over a time frame of roughly eight hours. The state of the battlefield is conveyed by maps covered with pieces of various shapes, sizes, and colors representing individual units, while the computer calculates the results of individual encounters. THEATERSPIEL also focuses on ground combat, but as a theater-level simulation it incorporates air operations and logistics into its computer modeling to simulate a complete contained ground war.²²

ORO also deployed what is likely the first military simulation designed to run unassisted on a digital computer. This game evolved out of work performed by Dr. George Gamow, a physicist at George Washington University and participant in the Manhattan Project at Los Alamos who began working with Monte Carlo simulations, so-called because they contain a considerable element of random chance as in a casino game, during his work on the atomic bomb. After he became a consultant for ORO in 1950, he began applying these same methods to warfare and developed a simple manual game called Tin Soldier in which two players assumed the role of commanders of opposing tank formations depicted on a contour map.²³

A fellow Los Alamos alum at ORO, Richard Zimmerman, began working on a computerized version of Tin Solider around 1953, which became operational on an ERA 1101 scientific computer around 1954 as the Maximum Complexity Computer Battle.²⁴ The next year, ORO installed an ERA 1103 computer, and Zimmerman began adapting his test program into a complete simulation. This game became fully operational in 1959 as CARMONETTE, which stood for “Computerized Monte Carlo Simulation.”

The first version of CARMONETTE pitted two opposing tank companies designated Red and Blue against each other in a simulated encounter replayed upwards of 50 times to develop a casualty distribution for both sides. By tweaking the characteristics of the tanks between runs and comparing results, military planners could optimize tank designs to minimize losses on the battlefield. The game became more complex as time went on, with CARMONETTE II adding infantry in 1960, CARMONETTE III incorporating helicopter support in 1966, and CARMONETTE IV adding communications and night vision into the mix in 1969. CARMONETTE V and CARMONETTE VI followed in 1970 and 1972, respectively. As the Army continued updating the game, it also moved to more powerful computers, first an IBM 7040 in 1965 and then a CDC 6400 in 1969.²⁵ While deemed a useful tool for military research, CARMONETTE demonstrated the limits of early computers, as it could only simulate reasonable approximations of combat conditions and needed to be supplemented by extensive field testing.

²² Hausrath, 1971, 172–187.

²³ Ibid, 64–67.

²⁴ Ibid, 67–68.

²⁵ “Carmonette Volume I: General Description,” CAA-D-74-11, prepared for General Research Corporation, November 1974, 69–71.

***

Like the Army and Navy, the U.S. Air Force embraced military simulations in the 1950s, which were initially run largely through its own think tank, the RAND Corporation. Originally established as Project RAND in December 1945, the organization – whose name is short for “Research & Development” – started as a collaboration between the Douglas Aircraft Company and the United States Army Air Forces (USAAF) and evolved out of advanced weapons research conducted during World War II. In 1948, Douglas ended its involvement with the project, and RAND spun out as its own corporation.²⁶

From its inception, RAND’s goal was to bring together researchers across disciplines to tackle problems related to national security. Perhaps the most famous policy to come out of this research was the mutual assured destruction approach to nuclear deterrence that defined the Cold War. It also played a critical role in the space race with the Soviet Union. To develop its many doctrines and strategies, the organization relied heavily on OR simulations.

RAND’s earliest war game, a strategic-level manual game called SAW, was conducted in 1948. A less complex version called STRAW followed in 1953, and its successor, SWAP, became a regular part of the U.S. Air Force Academy curriculum after its introduction in 1959. War gaming became so integral to the RAND mission that in 1954 the organization constructed a war gaming facility called Project SIERRA dedicated to exploring Air Force strategy in a limited war. Within a decade, over 50 war games had been conducted at the facility.²⁷

²⁶ Ware, 2008, 5–9.

²⁷ Sharon Ghamari-Tabrizi, “U.S. Wargaming Grows Up: A Short History of the Diffusion of Wargaming in the Armed Forces and Industry in the Postwar Period up to 1964,” StrategyWorld.com, accessed May 27, 2019, https://www.strategypage.com/articles/default.asp?target=wgappen.htm.

RAND placed itself at the forefront of computer simulation as well, beginning with its pioneering Project Simulator in 1952. The most complex computer game developed by the Corporation in the 1950s was the Air Battle Model I (ABM), which simulated a nuclear war by modeling the deployment of missiles, bombers, and refueling tankers and assessing their effectiveness against ground targets. First tested in 1955 at the Air War College, the program was completed by a civilian contractor, Technical Operations Research, Inc., in 1957.²⁸

RAND also focused on supply issues through its Logistic Simulation Laboratory, which developed a series of computer-assisted games beginning with Laboratory Problem 1 (LP-1) in 1957 that trained personnel in the management of aspects of the Air Force Supply system.²⁹ The lab also developed several manual logistics simulations, and the most celebrated of these, MONOPOLOGS, helped spur the expansion of simulation games out of the military and into the realms of business and academia.

First played in 1955, MONOPLOGS evolved out of a RAND study to model the entire Air Force supply system mathematically and placed the player in the role of a spare-parts inventory manager. While ostensibly tailored to the Air Force system, the game had general utility for any organization that needed to determine the best way to move goods quickly through a distribution network at low cost and attracted the attention of the American Management Association (AMA). The AMA invited members of the Logistic Simulation Laboratory to join a team it assembled in 1956 to create a computer game derived from the decision-making models present in military simulations to train top executives.³⁰

Working alongside business consultants and IBM programmers, the RAND staff helped develop a business management simulation called The Top Management Decision Simulation. Programmed on an IBM 650 and delivered in May 1957, the game challenges five teams of three to five players to run single-product firms in direct competition with each other. The game plays out in turns representing a quarter of a year, during which each team decides how much to spend on marketing and R&D, determines a rate of production for its product, chooses whether or not to increase manufacturing capacity, and considers whether or not to purchase marketing research on competitors. The computer then calculates the market share of each company at the end of the quarter based on these decisions.³¹

²⁸ Ibid.

²⁹ R. M. Rauner, “Laboratory Evaluation of Supply and Procurement Policies: The First Experiment of the Logistics Systems Laboratory,” R-323, July 1958, 1–5.

³⁰ Ghamari-Tabrizi, “U.S. Wargaming Grows Up.”

³¹ Kalman J. Cohen and Eric Rhenman, “The Role of Management Games in Education and Research,” Management Science 7, no. 2 (1961): 135–136.

The AMA incorporated a version of The Top Management Decision Simulation into its Management Development Seminar and its Executive Decision Making Program, and by March 1958 it had already been played by 350 corporate executives and 50 scientists and business school professors.³² The positive reception of the AMA game led to the adoption of similar games by both corporations and university business schools. The first known business simulation used in a university classroom was the Top Management Decision Game created at the University of Washington in 1957, which was a direct adaptation of the AMA game. UCLA also adapted the game as the UCLA Executive Decision Game No. 2 and added the wrinkle of allowing companies to borrow money.³³ This game was further revised as the UCLA Executive Game No. 3, in which companies could sell up to three different products.³⁴

Business games reached a new level of complexity in 1958 with the development of The Management Game by Kalman Cohen, Richard Cyert, and William Dill at the Carnegie Institute of Technology in Pittsburgh – renamed Carnegie Mellon University in 1967 after a merger with the Mellon Institute of Industrial Research. Still used by the university to this day after a major overhaul in 1986, the game runs over the course of two semesters as the player takes control of one of three competing detergent companies and makes as many as 100-300 decisions a month over a 3-year period regarding everything from R&D to marketing. As the complexity and utility of computer business simulations continued to increase, they were widely adopted across both the business and academic worlds.³⁵ By 1961, over 100 business games had been created, and over 30,000 executives had played them.³⁶

³² Ghamari-Tabrizi, “U.S. Wargaming Grows Up.”

³³ As the title implies, this game obsoleted a short-lived UCLA Executive Decision Game No. 1.

³⁴ Hugh J. Watson, Computer Simulation in Business (New York: John Wiley & Sons, 1981), 238.

³⁵ “Management Game: More Difficult than Chess, and More Fun to Watch,” Carnegie Mellon University, 2007, http://cmu.edu/corporate/news/2007/features/mgmt game.shtml.

³⁶ A.J. Faria and Ray Nulsen, “Business Simulation Games: Current Usage Levels, A Ten Year Update,” Developments in Business Simulation & Experiential Exercises 23 (1996): 22.

Like the AI research projects undertaken by select universities and corporations, the simulations developed by the military-industrial complex were not intended as entertainment programs and were played only by small groups of students, military recruits, and executives. Furthermore, many of these programs were limited merely to calculating the results of player action and required manual administering and/or scoring alongside additional play aides such as maps or markers. As such, these programs exerted virtually no impact on the subsequent expansion of computer gaming to the world at large. At the same time these simulations were being conducted, however, computer games were slowly being introduced to the general public through a variety of fairs, trade shows, and open houses.