Innovation and Learning in High-Reliability Organizations - IEEE Xplore

IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 3, AUGUST 2008

393

Innovation and Learning in High-Reliability Organizations: A Case Study of United States and Russian Nuclear Attack Submarines, 1970–2000 Paul E. Bierly, III, Scott Gallagher, and J. C. Spender

Abstract—Given their complexity and tight coupling, one of the most serious challenges high-reliability organizations (HROs) face is how to innovate, learn, and adapt without upsetting the internal processes that lead to their reliability. This paper describes the success of the United States Navy in using a “platform strategy” to facilitate modular innovation in its attack submarine program while maintaining high reliability. We compare the United States’ submarine development program against that of the Soviets, who innovated by building a number of different types of nuclear attack submarines to test their new design concepts and thereby aggressively push both manufacturing and performance limits. We illustrate that, by adopting a platform strategy, the U.S. development program was able to sustain reliability by controlling factors that derived from four classes of concern: 1) operational; 2) manufacturing and design; 3) resource limitations, and 4) cultural constraints. The use of a platform strategy assists in maximizing system-wide organizational learning, which helps enrich a culture of reliability. However, at the same time, a platform strategy can hinder revolutionary and architectural innovation and reduce operational flexibility. Finally, we consider whether an HRO’s innovation strategy is partially shaped by its decision-making process. Index Terms—High-reliability organizations (HROs), modularity, organizational culture, organizational learning, platform strategy, product innovation.

I. INTRODUCTION IGH-RELIABILITY organizations (HROs) are defined as complex and technologically sophisticated, wherein a system failure may result in a catastrophe [38], [44]. Due to these competing demands and constraints, most HROs must learn how to cope in restrictive and dangerous environments, while being forced to operate error-free. They must maintain their high reliability even though they typically are not able to benefit from experiential or experimental forms of learning, relying instead on rich analyses of small samples [28]. They must overcome major conflicts between divergent goals, especially the tension between the outcomes normally expected of organizations, such as goal attainment and efficiency, while also avoiding a catastrophe and remaining reliable. In the literature, examples of HROs

H

Manuscript received October 1, 2006; revised July 1, 2007 and September 1, 2007. Review of this manuscript was arranged by Department Editor J. Liker. This work was supported by the College of Business and Center for Entrepreneurship, James Madison University. P. E. Bierly, III, and S. Gallagher are with James Madison University, Harrisonburg, VA 22807 USA (e-mail: [email protected]; [email protected]). J. C. Spender is with Queen’s University School of Business, Kingston ON K7L 3N6, Canada (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TEM.2008.922643

include nuclear power plants, chemical plants, electric power grids, space projects, military systems (aircraft carriers, submarines, missile launch facilities, etc.), air traffic control centers, and nuclear waste facilities and a large body of research has developed on how HROs can best maximize reliability while also maintaining efficiency [8], [38], [42]–[44], [62], [69], [72], [74]. In this paper, we focus on innovation in HROs, an area of research that has received less attention despite clearly being important. Almost all organizations considered HROs are “high-tech” and must be able to maintain reliability while they implement new technologies. We make specific contributions to the HRO literature in two areas by 1) increasing our understanding of innovation in HROs and identifying the advantages and disadvantages of using platform and modularity strategies within such sensitive systems, with particular focus on maintaining reliability during innovation implementation and 2) better appreciating the challenges and importance of organizational learning in HROs by contrasting incremental platform-based learning against experiment-driven learning. We use the case study method as outlined by Eisenhardt [14] to illustrate and unpack the nuances of innovation and learning in HROs. Specifically, we analyze the different innovation approaches used by the United States and Soviets in their respective attack submarine programs, covering the design, manufacturing, and operation of the submarines while also considering social and cultural aspects.1 We use the findings of this case study to develop propositions useful to both managers and researchers of HROs. II. HIGH-RELIABILITY ORGANIZATIONS HROs share properties that differentiate them from other organizations. 1) Catastrophic potential—each of these organizations has the potential to create a major catastrophe (i.e., Chernobyl, Bhopal, U.S. space shuttle accidents). This potential increases scrutiny of their actions and sharply increases the employees’ stress level. 2) Tightly coupled systems—most of the subsystems within these organizations are interdependent, with little or no slack between them [38]. A failure in one subsystem may have an immediate impact on the viability of another subsystem that can potentially cascade into a catastrophic system-wide loss. 1 Our study period 1970–2000 spans both the Soviet Union and Russian governments. We refer to Soviets during the period 1975–1990 and to Russians after 1990.

0018-9391/$25.00 © 2008 IEEE

394


3) Interactive complexity—the interdependency of an HRO’s subsystem creates a complexity that results in many unexpected and unfamiliar sequences of events. These may interfere with the operators’ ability to diagnose aberrant behavior [38]. Complexity is also extended because HROs are normally operating near their design performance envelope, hence limiting the operatives’ opportunities to use minor variations as learning opportunities. 4) Accountability-–likewise control in HROs tends to be strictly hierarchical, and employees are held immediately accountable for their areas of responsibility [44]. Improper performance generally results in strict disciplinary action. Little leeway is granted or available. HROs must find ways of handling the combination of catastrophic potential, tightly coupled systems, and interactive complexity. The preferred ways of handling tightly coupled systems and more centralization are generally in conflict with the best ways of handling systems with complex interactions and more decentralization [38]. Here, we see the same contradiction of “proverbs” that triggered Simon’s analysis of bounded rationality [57]. It is difficult to develop the structure, values, and atmosphere that control the interactions successfully. Weick [73] showed that HRO systems become more complex and tightly coupled, and hence, more vulnerable to crises, when normal routines are interrupted and operators experience increased levels of stress. This highlights the operators’ crucial diagnostic role when working in conjunction with these systems. Previous empirical research has addressed a substantial variety of management issues in HROs, mostly by studying catastrophic system failures such as the Challenger [63], [69], the Columbia [62], Three Mile Island [38], and Bhopal [55]. Summarizing, the research indicates HROs are more successful if 1) a learning environment has been established, where empowered members can act and learn during crises as well as during their extensive training; 2) appropriate reward and incentive systems guide the tradeoffs between efficiency and reliability; and 3) there are systems and processes that ensure employees understand the “big picture” and encourage open communications across disciplines [43]. Developing a culture of high reliability interacts with and supports an organizational structure designed to lead to superior performance [8], [72]. A “collective mind” among a system’s actors increases their understanding of how technical and social subsystems interrelate and create more reliability [74]. The organization’s members remain focused on preventing failure, and scanning for and responding to signals of potential problems [75]. Little research has been conducted on innovation strategies in HROs. Technical or design improvements of HROs often fail to consider the need for subsequent changes in organizational structure, management systems, and culture [16], [70]. Incrementally improving or fine-tuning an HRO system can be dangerous due to the lack of appropriate feedback and the creation of a false sense of security [63]. Starbuck and Milliken [63] described how this fine-tuning innovation strategy was a leading contributor to the Challenger disaster. Radical changes to an HRO component are also troublesome to implement because of the difficulty understanding its complex interaction with other

components and the difficulty realizing how the new technology impacts social and cultural aspects of the HRO. The root cause of HRO failures is rarely either a technical failure or a human error—it is usually both [16], [25]. A key issue concerning innovation of HROs, and the primary concern of this research, is how can HROs innovate successfully? Looking at the disparate experiences of the United States and the Soviet/Russian nuclear attack submarines suggests a platform strategy to exploit the many advantages of modularity may be effective. Adopting a platform strategy will not only dictate the innovation path of the organization and the radicalness of innovations implemented, but also will determine how the organization learns. However, before discussing these issues in the context of HROs in general, or our submarine case in particular, we will first summarize the pertinent research in the various interrelated research streams associated with a platform strategy. III. THEORETICAL OVERVIEW—PLATFORMS, DOMINANT DESIGNS, STANDARDS, AND MODULARITY According to Meyer and Lehnerd [30], a platform is defined as “a set of subsystems and interfaces developed to form a common structure from which a stream of derivative products can be efficiently developed and produced.” Simply put, a platform is a complex of intellectual and physical assets that can be shared across multiple products [45]. A platform strategy has the advantages of reducing development cost, increasing product reliability, and promoting component, or modular, innovation. While theoretically appealing, a platform strategy also carries disadvantages by possibly limiting architectural innovation or other breakthrough technological advances, as well as incurring higher upfront costs and potentially constraining how the platform-based product is used [26]. A theoretical argument describing the advantages and disadvantages of a platform strategy can be derived from three specific literatures: on dominant designs, on standards, and on modularity. Each of these positions provides insights about the effectiveness of a platform strategy. A. Dominant Designs Dominant designs appear as a result of an industry’s technological evolution and can be defined as the product architecture used by most surviving members of the industry [1], [9], [61], [65]. The concept is central to the punctuated equilibrium model of industry evolution that describes how industries often follow a pattern of 1) a radical innovation, leading to a period of technological and market competition between rival designs; 2) the appearance of a dominant design; and 3) an extended period of incremental innovation around this dominant design [1], [4], [46], [68]. Different design approaches are usually synthesized into a dominant design as a result of technical competency, collateral assets, industry regulation, government intervention, strategic maneuvering, and feedback from users to producers [68]. It is also important to realize that the concept of a dominant design can be applied to different levels of analysis, such that the components that make up the product architecture of an industry’s dominant design may each be a dominant design in their own industry [2]. For example, there may be a

BIERLY, III et al.: INNOVATION AND LEARNING IN HIGH-RELIABILITY ORGANIZATIONS

dominant design in the aircraft industry, which, in turn, utilizes the dominant design in the aircraft engine industry. B. Standards Whereas the concept of a dominant design has a broad application at the industry level, the concept of a standard is narrower in focus. Standards are especially important in situations requiring some type of interconnectivity, either between systems, e.g., telephones, or between systems and their components, e.g., computers and software [61], [65]. Standards emerge as the interface protocols that allow networks of users [20]. As the network extends, positive externalities arise and organizations can gain an advantage through either the direct utility of interoperability between users themselves (e.g., fax machines) or the supply of complementary products, such as software, which becomes more plentiful as the size of the network grows [5], [17], [24]. Standards, normally an aspect of dominant designs, are necessary for modularity. C. Modularity When a firm has an established system architecture and the interfaces between components are standardized, the benefits of modularity can be realized [7], [27], [50], [52]. Modularity enables the building of complex products by independently designing and improving the components while maintaining standardized interfaces. The degree to which a system’s components may be separated and recombined defines the modularity of a system [31], [49]. Modularity offers many advantages, including economies of scale, as standardized components can be mass produced, but also, economies of scope via product flexibility [31]. Modularity provides firms with the strategic flexibility of frequently changing components while maintaining the system architecture [7], [50]. Modularity allows the rate of innovation at the component level to be maximized by allowing the technological trajectories of the components to evolve at different rates [7]. Sanchez and Mahoney [50] illustrated how modularity in product design creates the information structure that enables the use of a modular organizational design. The focus of design then shifts from the fit between functional areas of an organization toward optimizing the coordination of the various subsystems [56], [59]. There are also disadvantages associated with modularity. Modularity may require a high degree of independence and “loose coupling” between the components, and may not be effective when there is either tight coupling or interactive complexity, both of which are common in HROs. Ethiraj and Levinthal [15] used a simulation model to illustrate that overuse of modularization in complex systems can have destabilizing effects on the organization because it will inhibit their ability to systematically improve and exploit knowledge associated with prior search efforts. However, Salvador [49] explained how complex systems can maintain modularity during design and manufacturing of components, while the subsystems remain tightly coupled during operation. A modularity approach may also lessen architectural innovation since there will be more resistance to changing the existing

395

architecture [21]. Strict interface protocols may severely limit component innovation opportunities and frustrate firms working on a component when they lack control of the architecture [64], or the interface protocols may increase conflict over the interpretation of how ambiguous activities are linked [33]. Schilling and Steensma [53] have described how in certain circumstances components may be able to be recombined in different ways, changing the system architecture, but this is less common in complex systems such as HROs. D. Platforms The concepts of a dominant design, standards, and modularity can be used to explain the effectiveness of a platform strategy. Modularity allows the efficient organization of complex products or processes by decomposing complex tasks into simpler portions while maintaining their integration at key break points governed by standards [31]. The architecture of these module components, break points, and standards is a platform, and when it is adopted by most organizations in an industry, it is also a dominant design. When the architecture is especially robust it is referred to as a scaleable product platform [58]. Components can be changed for different products, innovations at the component level can be implemented quickly, economies of scale can be established to reduce costs, and the reliability of the system can be enhanced as long as the standards at the relevant break points are maintained. However, innovations at the component level sometimes require modifications to the interfaces between components, and so corresponding innovations to the other components. This situation calls for platform leadership and close coordination between the producers of both components [11]. We expect this to be common among HROs, and particularly difficult to manage. All these advantages come at a cost. A platform approach generally requires higher upfront development costs to create the flexible platform. Also, mathematical models have shown that in comparison with a two-distinct-model approach, a platform approach limits an organization’s ability to serve diverse markets and generally results in underdeveloped high-end products and overdeveloped low-end ones [26]. E. Dominant Designs, Standards, and Platforms Applied to Submarine History Most of the world’s navies use some sort of platform strategy when designing their warships. They refer to their platforms as ships of a “class.” Submarines are no exception. As we will illustrate in our detailed discussion of submarine development, there are many advantages and disadvantages associated with different platform strategies. Platforms can increase system-wide learning, improve operational capabilities, and help to create a culture of high reliability, but maybe at the cost of hindering radical innovation. While submarines date back to at least the eighteenth century, submarine attacks on commercial and military shipping began in earnest during World War I (WW I) (1914–1918). Operating from a position of naval inferiority, the Central Powers, especially Germany, made effective use of submarines. However,

396


WW I submarines were primarily surface warships that could submerge for a short time only. Even later World War II (WW II) (1939–1945), submarines were surface craft that surrendered considerable endurance, speed, and detection capabilities by submerging. Despite these limitations, today’s craft embody many of the architectural features of the early submarines, such as a streamlined hull, deep surface draft, sail or conning tower arrangements, rear-mounted propellers, forward-placed diving planes, and air-blown ballast tanks. However, the combinations that comprised the dominant design shifted over time. Three separate dominant designs evolved during the decades before the period of our research. First, early diesel-electric submarines (i.e., WW I) used a periscope and lookouts to locate enemy contacts, and then attacked with torpedoes and guns mounted on the top of the submarine. They had limited underwater attack capabilities. The subsequent generation of diesel-electric submarines (i.e., end of WW II) used active electronics (sonar and radar) for enemy detection while submerged, had a closed propulsion system with a single turbine, had better underwater maneuvering capabilities, and attacked other ships only with torpedoes rather than guns. Nuclear attack submarines (SSN) are an extension of this last generation with new nuclear propulsion technology, advanced sonar and radar systems, and vastly greater underwater capabilities, including virtually unlimited underwater endurance. These advances also enabled the development of intercontinental ballistic missilecarrying submarines, but our focus is on attack submarines— boats designed to attack other submarines and surface ships. Our approach to the differences between the United States and USSR attack submarine programs is to contrast each nation’s different responses to what were more or less similar design challenges. As their designers grappled with a common set of design challenges such as speed, stealth, and attack capabilities, and innovated in response to the capabilities of their rivals, nuclear submarines comprise an ideal study area for innovation in HROs.

IV. FAST-ATTACK NUCLEAR SUBMARINES

the ideas we present might both benefit managers and guide future research in the conclusion. We study the evolution of one type of HRO, the fast-attack nuclear submarines, over an extended period of time, focusing on the consequences of platform and component design changes. The advent of the nuclear submarine made submarines HROs by effectively tightening the coupling between propulsion and diving, allowing a failure of the former to result in a catastrophic loss of ship incident. Specifically, we compare and contrast the different innovation strategies deployed by the United States and USSR/Russia between 1970 and 2000. This period was chosen because, for the United States, it covers 1) the transition from the Sturgeon class (SSN 637) to the Los Angeles class (SSN 688); 2) the building of all 62 Los Angeles class submarines, including many component improvements; and 3) the transition from the highly successful Los Angeles class submarines to the Seawolf class. While the United States predominantly used one platform for attack submarines during this period, the Russians followed a very different innovation strategy, as illustrated in Fig. 1. They experimented by building a large number of different types of nuclear attack submarines including the November, Victor, Alpha, Sierra, Mike, and Akula classes [40]. This allowed them to test many new designs and aggressively push performance limits. For example, the Alpha was a small submarine with a liquid metal reactor and titanium alloy hull that was designed to be faster than any other naval submarine. In this manner, the Russians should be credited with developing a new dominant design for an entirely new type of submarine, the guided missile submarine with its Charlie and Oscar classes. How the United States eventually adapted the Los Angeles class of attack submarine to include this capability, by adding a vertical launch missile system, is an interesting example of the robustness of a well-designed platform. The data for our analysis of attack submarine platform development comes from many secondary sources, government documents, the authors’ experience in the industry, and interviews, including an in-depth interview with Captain David Schubert, Commanding Officer of the U.S. Naval Research Laboratories [54].

A. Method of Analysis We used a traditional case study method in this paper to track the developments of the two main attack submarine rivals, following the case study procedures outlined by Eisenhardt [14]. Guided by our broad research question concerning innovation in HROs, we selected the rival case experiences of the United States and Soviet/Russian attack submarines. We utilized multiple data collection approaches, including qualitative and quantitative data on the parameters we identified a priori as being important—reliability, as well as SSN performance characteristics such as speed, stealth (diving ability and quietness), and weapon capabilities. We focused not only on identifying certain relationships, but also on developing an understanding of their underlying dynamics. Consistent with Eisenhardt [14], we shaped our propositions (see Section V) based on a rich, interactive comparison between the different experiences of the two navies. We further discuss the literature, limitations, and how

B. U.S. Submarine Design In the United States, submarines are designed by multifunctional teams drawn from throughout the shipbuilding community to optimize performance, production efficiency, and to ensure quality [13]. The Naval Sea Systems Command (NAVSEA) and the U.S. shipyards work together throughout the various design and build stages. In contrast with the arrangements for producing earlier nuclear submarines, the shipyards involved in the construction of attack submarines were limited to Electric Boat (EB) in Groton, CT, and Newport News in Norfolk, VA. The United States approach to nuclear submarine design and manufacture has its roots in the engineering philosophy developed by Admiral Hyman Rickover, the founder of the Naval Nuclear Propulsion Program (NNPP). As a young naval electrical engineer assigned to the Manhattan Project in the early 1940s, he was the first to grasp the potential of nuclear power in ships


Fig. 1.

Soviet and U.S. attack submarine classes and building dates.

397

and was instrumental in developing the first nuclear-powered submarine, the Nautilus, launched in 1955. Since oxygen was not required to burn its fuel, nuclear power allowed a submarine to stay underwater for months at a time, hugely enhancing its range and the difficulty of discovering its position. Rickover ruled the NNPP from its inception until 1982. His 63 years of service in the Navy was longer than that of any other officer in U.S. naval history. Throughout this time, he strove to maintain tight control over the design, building, and operation of the nuclear power program, understanding well the interrelatedness between these areas and the importance of managing the entire HRO system. While his formal control focused on the engine room of the submarine, he had tremendous influence over all aspects of the U.S. submarine operations. Rickover was fanatical in stressing safety and reliability as the principal objectives of the submarine program. He realized that the political implications of a reactor accident would be devastating and seriously affect the viability of his entire program. He outlined the basic principles of his engineering philosophy concerning building and operating effective naval nuclear propulsion plants in a 1982 talk to the Congress. r Avoid committing ships and crews to highly developmental and untried systems and concepts. r Ensure adequate redundancy in design so that the plant can accommodate, without damage to ship or crew, the equipment and system failures that will inevitably occur. r Simplify system design so as to be able to rely primarily on direct operator control rather than on automatic control. r Require suppliers to conduct extensive accelerated life testing of critical reactor systems components to ensure design adequacy prior to operational use. r Test new reactor designs by use of a land-based prototype of the same design as the shipboard plant. r Confirm reactor and equipment design through extensive analyses, full-scale mockups, and tests. r Concentrate on designing, building, and operating the plants so as to prevent accidents, not just cope with accidents that could occur [12]. One of the reasons for Rickover’s profound concern over safety and reliability was the Navy’s equivalent of the Challenger disaster, the loss of the Thresher on April 10, 1963, with all 112 crew and an additional 17 civilians [54]. This loss was traced to an unforeseen reactor shut down that deprived the submarine of the propulsion necessary to surface during a fire. It plunged below its crush depth and imploded. In the newly designed nuclear submarines, the propulsion system had become much more tightly coupled to the survival of the ship. Unlike WW II era submarines, the new nuclear-powered submarines could not just blow and empty their ballast tanks to surface, they needed power to “drive” up to the surface. Rickover was an outspoken critic of the problems that had led to the loss of the Thresher, and its loss gave his views considerable exposure and prominence. After the loss of the Thresher, he championed better emergency design and the SUBSAFE construction program to prevent its recurrence. The Three Mile Island (TMI) nuclear accident in 1979 reinforced the need for safety and reliability to be the dominant

398


TABLE I U.S. ATTACK SUBMARINES

values in the U.S. Nuclear Navy. This event occurred while some members of the submarine community were openly questioning the appropriateness of Rickover’s value system. They believed his extreme focus on safety and reliability blocked the incorporation of many new, cutting-edge technologies. Many joked that the mission of the U.S. submarines was no longer to be an effective combatant but to support the operation of Rickover’s reactors. The political fallout from TMI, which stopped the building of commercial nuclear reactors in the United States, illustrated why it was imperative that the U.S. Nuclear Navy maintain safe reactors. C. Los Angeles Class Submarine For two decades, the Los Angeles class submarine was the United States’ primary attack submarine and it offers an exceptional example of how to use a platform for an extended period of time. Consistent with a sound platform approach, some components, such as the nuclear reactor (a General Electric S6G pressurized water reactor) were common to all ships, while other components, such as the combat and radar system, were changed. The former could not be changed without reengineering the entire platform, while the latter could be changed without such a consequence. But its modular design also facilitated some types of change, such as the use of an open systems

architecture (OSA) for rapid software and hardware upgrades to weapons control and sonar systems. However, the approach to modular innovations was conservative, especially when the improved subsystem was tightly coupled with other subsystems. The design of the Los Angeles class submarine began under Rickover’s guidance in 1964. The submarine’s primary mission was antisubmarine warfare and to be an escort for a carrier task group. Sixty-two submarines numbered SSN 688–773 (skipping 726–749) were built, with the first being completed in 1976 and the last in 1996. The design was significantly better than previous U.S. submarines along all performance parameters. The Los Angeles class was a technological leap above the previous, and much slower, Sturgeon class, producing a fast, quiet, large submarine with superior weaponry and improved sonar and communication capabilities (see Table I). Weapons included MK-48 torpedoes, Harpoon antiship mimmiles, (range—130 km), and Tomahawk cruise missiles, launched through the boat’s torpedo tubes, for use against land (range—2500 km) or ship (range— 450 km) targets. Many modular and minor improvements were made to the class over the years, e.g., improvements to weapons (a switch to the MK-48 Advanced Capability [ADCAP] torpedo), new sonar systems, and noise reduction and engine room upgrades. There were two major variations in the class. First, the 30 ships built after SSN718 had their hulls lengthened to accommodate


12 vertical launch tubes for additional Tomahawk missiles. This enabled a submarine to launch a total of 20 Tomahawks, rather than the eight it could launch through its torpedo tubes. Second, a series of upgrades were made to all submarines after SSN 750, creating the Improved Los Angeles subclass (SSN688I). These final 23 submarines were designed to operate in the Artic with a stronger sail and with their diving planes moved from the sail to the bow giving superior under-ice capability. A variety of other improvements also made them much quieter, gave them mine-laying capability via their torpedo tubes, and upgraded their sonar suite combat system. D. Transition to a New U.S. Submarine Design From the mid-1970s onward, Rickover and others argued against relying on a single platform, such as the Los Angeles class, for an extended period. He pressed for the development of a faster and larger attack submarine with double the propulsive power of the Los Angeles that could launch many more cruise missiles. Many Navy experts and politicians argued that more low-cost, relatively simple submarines with heavy firepower were needed; others argued the United States should complement the nuclear submarine fleet with less expensive diesel attack submarines increasingly popular in other nations’ fleets [39]. Rickover believed that relying solely on the Los Angeles class submarines for an extended period would inhibit the U.S. from innovating and thereby maintaining its lead over the Soviets [39]. He warned often that the Soviets were innovating at a faster rate than the United States and would soon surpass the Americans on many performance parameters. Concerning nuclear submarines he once said: There has not been an arms race; the Soviets have been running at full speed all by themselves . . . Weapons systems, speed, depth, detection devices, quietness of operation, and crew performance all make a significant contribution to the effectiveness of a submarine force. From what we have been able to learn during the past year, the Soviets have attained equality in a number of these characteristics, and superiority in some [39].

Some of the improvements Rickover and others viewed as essential were added to the Los Angeles class incrementally, such as the vertical launch tubes, but many other potential improvements were forgone, primarily to ensure reliability and cost control. The 20-year dominance of the Los Angeles class ended in 1997 with the completion of the first Seawolf class submarine. This class is a dramatic improvement to the Los Angeles class, in that the boat is larger, quieter, faster, and can carry many more missiles and torpedoes (see Table I). Its pressurized water reactor is similar to but larger than the Los Angeles class reactors. However, the cost of a Seawolf class submarine approaches $3 billion, more than double that of a Los Angeles class submarine. Modular design was used extensively for major systems of the Seawolf to allow for easy upgrades. Additionally, an open database system was used to enable digital data transfer between the Navy designers, the two shipyards, and the other suppliers [13].

399

E. Soviet/Russian Attack Submarines Throughout the Cold War, the Russians viewed the submarine as the most important component of their navy and spent vast resources on its development. Unlike the Americans, they did not view large aircraft carrier groups as strategically efficient. According to the famous Russian submarine designer Igor Spassky, Chief Designer in the prestigious Rubin Central Design Office, the evolution of Russian submarine designs involved many political and technological tradeoffs [60]. The design process followed the following steps: 1) the national maritime policy, established by political party leaders, dictated certain objectives to the navy; 2) the navy formulated specific design requirements for the industry; and 3) the researchers, submarine designers, builders at the shipyards, and equipment suppliers worked together to develop specific technologies [60]. The process was more authoritarian and top-down than the American approach with the designers and manufacturers having less influence on the political and naval leaders above. For example, submarines had to be delivered when promised in their contract, completed or not [37]. Submarines were built at five different shipyards during this time and the design and construction involved many thousands of people. The Russians produced a wide variety of attack submarines that together performed the duties of the U.S. Los Angeles class submarine (see Tables II and III). They fell into three broad types, designated SSNs, SSGNs, and diesel submarines. As such, they embodied three different dominant designs. The SSNs are smaller nuclear attack submarines whose primary mission is antisubmarine warfare, but embody essentially the same design considerations as the Los Angeles class. SSGNs are larger nuclear attack submarines that possess a large number of cruise missiles, in order to attack large concentrations of surface ships, such as U.S. aircraft carrier groups. Finally, the Russians maintained significant numbers of coastal diesel submarines, harking back to the earlier late WW II dominant design, to protect their coasts and sea lanes from other submarines and surface ships. 1) Victor Class: This was Russia’s primary platform for SSN attack submarines. The Victor (I–III) class comprised 49 submarines built between 1968 and 1992, about half the nuclear attack submarines built in this period. The Victor I was the first Soviet submarine developed for antisubmarine warfare and the first with a low-frequency sonar. The class had a double-hull configuration, two pressurized-water reactors, and a high-speed hull design. The Victor II incorporated incremental changes from the initial Victor I design, but included enlarging the ship to allow for new torpedoes and missiles, an improved propeller, and an improved fire-control system. The Victor III involved major improvements to the design and became the most numerous submarine in the Russian fleet. Some of the advances are believed to have been copied from U.S. designs, probably with the assistance of the Walker spy ring [40]. The Victor III was significantly quieter than any previous Russian submarine, with some experts estimating that it was as quiet as the U.S. Sturgeon class, the precursor to the Los Angeles class [41]. In addition to major acoustical improvements, the Victor III class also incorporated improved electronics, navigation systems, and radio

400


TABLE II SOVIET AND RUSSIAN NUCLEAR ATTACK SUBMARINES (SSN)a

and satellite systems. The class also had the ability to launch strategic cruise missiles. 2) Alpha Class: The Russians frequently experimented with breakthrough technologies developing new classes of submarines. Probably the most famous of these are the Alpha class nuclear attack submarines, which were built from 1979 to 1982. These revolutionary submarines were the fastest (>40 knots), deepest diving (>2000 ft) submarines of their time. Their remarkable performance was a result of their exceptionally strong titanium alloy hulls and a revolutionary reactor design that was cooled by liquid sodium metal. These reactors were more powerful than the U.S. pressurized water reactors. Additionally, the Alpha submarines were much more automated than other Soviet submarines, requiring a smaller crew. Unfortunately, despite successfully responding to the stealth and speed design challenges, the Alpha class submarines proved extremely unreliable. Four of the seven submarines in the class had major reactor accidents shortly after being commissioned, resulting in the reactor being destroyed. Additionally, the highly automated systems frequently failed. All the ships encountered major engineering problems. Eventually, they were considered so unreliable and dangerous that all were decommissioned within ten years after launch.

3) Sierra Class: Clearly, the Russians learned from the Alpha experience and some of the technological advances were refined and used in subsequent classes. The Sierra class SSN submarine is generally considered one of the best-designed and most-capable Soviet submarines, albeit at a high cost to build and operate. Only four of these titanium-hulled submarines were built between 1984 and 1993 (two designated Sierra Is, and two slightly improved Sierra IIs) before production stopped because of their exorbitant cost. The Sierra was larger than previous Soviet submarines (requiring significantly more titanium than the Alpha) and quite capable with better weapons than the Victors. It was powered by a more reliable pressurized water reactor but was still able to travel fast and deep. It also incorporated some breakthrough technologies such as a nonacoustic detection system (infrared sensors used to detect thermal gradients), and an advanced countermeasure system that was considered superior to the equivalent U.S. systems. It was considered comparable to early U.S. Los Angeles class submarines on most performance parameters and had the clear advantage of being able to dive deeper. However, all four Sierras were decommissioned in 1997 because of high operating costs. 4) Mike Class: Another revolutionary Soviet design was the Mike class, though only a single ship, the Komsomolets was


401

TABLE III RUSSIAN ATTACK SUBMARINE (SSN) SAFETY AND DESIGN HIGHLIGHTS

commissioned in 1984.2 It was built to test a dozen or so new technologies. It had an improved titanium pressure hull, was highly automated, and could go extremely deep (>3000 ft). It is still considered the deepest diving attack submarine ever made. However, on April 7, 1989, a high-pressure air line connected to the primary ballast tank burst its seal and caused a spray of oil to hit a heated surface resulting in a fire [34]. The fire raced through wiring conducts and forced the engineer to shut down the reactor to avoid a meltdown. This left the ship powerless. Fortunately, Soviet subs have considerable “reserve” buoyancy and the submarine managed to surface before sinking. Forty-two sailors were killed, with one sailor surviving by using the submarine’s special escape capsule—a feature not found on U.S. submarines—while the rest were able to leave the submarine when it surfaced [34]. An investigation into the accident concluded the causes to be 1) construction flaws and 2) improper 2 In addition to it being a unique class, the Komsomolets was also unusual because the Soviets named it. Most Soviet submarines were only known by their hull numbers, e.g., K-19, but in honor of breaking a diving record, K-278, was named Komsomolets, meaning “member of the Young Communist League.”

training [19]. Today, the submarine rests off the coast of Norway in water a mile deep. 5) Akula Class: The Akula class submarine was the premier steel-hulled Soviet attack submarine. Thirteen were built between 1985 and 1995. The initial design was an extension of the Victor class but modified to include many successful features of the Sierra class and characteristics of a cruise missile attack submarine [32]. The result was a multipurpose, fast (35 knot) submarine that was the quietest submarine the Russians ever built, though still at an acoustic disadvantage to the Improved Los Angeles class. Its sonar capabilities were also inferior to the Los Angeles class. The Akula’s are potent, with Sampson (SS-N-21) cruise missiles (range 3000 km), antiship missiles (SS-N-15 Starfish and SS-N-16 Stallion), torpedoes, mines, and even an air defense system. None of the Akula class submarines have experienced a major accident. 6) SSGN (cruise missile attack submarines): During the period of our study, the Russians also built two classes of cruise missile attack submarines, Charlie and Oscar, which were designed specifically to attack enemy aircraft carrier battle groups and convoys (see Table IV). Seventeen Charlie class submarines

402


TABLE IV SOVIET/RUSSIAN CRUISE MISSILE ATTACK SUBMARINES (SSGN)a

were built between 1968 and 1982. These were the first Russian submarines that could launch antiship missiles submerged. They were the only Russian submarines at the time that had only a single nuclear reactor, the configuration favored by the U.S. With its limited propulsive power, Charlies could only reach 24 knots, not sufficient to keep up with U.S. carrier groups. Three Charlies suffered major accidents, indicating poor reliability for the class. One, the K429, sank in 1983. It was raised and refitted but earned the dubious distinction of being one of the few warships to sink twice, sinking again in 1985 alongside a jetty, in an accident that killed 16 sailors. A second submarine, the K320, suffered a major reactor accident in 1970, and a third, the K313, suffered a major primary coolant leak in 1985. The Oscar class included 13 submarines built between 1982 and 1997. The Oscar is a very large and capable submarine, with many dramatic improvements over the Charlie class. Oscars were approximately three times larger than Charlies (length about 150 m and submerged displacement about 20 000 t), were significantly faster (approximately 32 knots maximum speed), and could rapidly launch 24 cruise missiles in addition to torpedoes. However, the safety record for Oscars was not very good. In 1998, an Oscar submarine experienced a major cooling system accident, killing one person. In 2000, another Oscar submarine, the Kursk, sank after a torpedo room explosion. 7) Diesel-Electric Submarines: The Russians also aggressively pursued design and construction of diesel-electric submarines during this period. However, these cheaper and less

sophisticated submarines are generally not considered HROs, and are not included in this study. Nevertheless, it is interesting to note that the 75 Foxtrot (1958–1983), 20 Tango (1972–1982), and 30 Kilo (1982–1998) class diesel-electric submarines built have had a much better safety record than Soviet/Russian nuclear attack submarines with only three being lost. A Kilo class submarine suffered a major fire in 1989, and two Foxtrot class submarines were lost at sea, one in 1962 and the other in 1991. F. Performance Measures Fortunately, the ultimate performance measure of the U.S. and Soviet submarines, their ability to defeat each other in war, remains untested. However, data from a wide variety of sources suggest how they compared along several dimensions. Nuclear attack submarines are primarily a design response to three challenges: speed, stealth, and attack capability. However, reliability is what keeps the ships operational and their expensive and highly trained crews safe. Our history of the Russian building program notes the improvements made in their submarines over time. Concerning stealth, the Russians went from their Victor III equaling the earlier U.S. Sturgeon class to the Sierra and Akula class boats that were at least a match for the early Los Angles class (but not the Improved Los Angles class) boats [40]. The Russians also closed the gap in speed, their Alpha being the fastest attack submarine (though loud), and the Sierra and Akula also being


very fast. The Russians also closed the gap in attack capabilities with their later submarines sporting a full assortment of cruise, antiship, and torpedo armaments. However, it is important not to overrate the Russians’ successes. The performance of the Los Angeles class submarine has been exemplary. When it was introduced, it was clearly the most capable submarine in the world. While one class of Soviet submarine may have been superior to the Los Angeles class on a single performance parameter, such as speed or depth, this gain was always at the expense of other performance parameters, such as reliability or stealth. When all of these tactical performance measures are bundled together, almost all experts agree the Los Angeles class submarine maintained an overall advantage over all types of Soviet ships, with the Seawolf class extending this lead [40]. Yet, the initially large performance gap between the first Los Angeles class submarines and Soviet subs narrowed significantly until the launch of the Seawolf. The two dimensions on which the Los Angeles class were particularly outstanding were safety and operational efficiency. The safety record of the Naval Nuclear Propulsion Program (NNPP) was impressive and is frequently used as an exemplar against which other HROs are measured [8]. In spite of operating for millions of hours, there has never been a major accident on a Los Angeles class submarine that has either rendered it inoperable or led to radiation leakage. The only accidents involving these submarines were collisions with other submarines or surface ships, such as in 1992 when the American submarine Baton Rouge (SSN689) collided with a Soviet Sierra I submarine. However, given the dangerous cat-and-mouse games played by submarines of both sides during the Cold War, collisions were inevitable. According to Polmar and Moore [40], 20–40 minor collisions between U.S. and Soviet submarines occurred during the Cold War. Yet, each time a Los Angeles class submarine was involved in one of these incidents, it remained operational. This contrasts with the Russian record of 15 accidents involving either its nuclear reactor or the total loss of the submarine. Therefore, the Russians closed some important performance gaps with specific submarine classes but usually at the expense of reliability. It is interesting to note that their most reliable submarines, the Sierras, were relatively less advanced, coming after both the ill-fated Alpha and Mike classes. V. THEORETICAL ARGUMENTS TO EXPLAIN PERFORMANCE DIFFERENCES The reasons for these performance differences lie in theoretical arguments related to innovation. We start each explanation with broad propositions applicable to HROs in general. The first set proposes rationales for the positive relationship between the use of a platform strategy and system reliability. Four propositions explain this relationship from four different perspectives: 1) operational; 2) manufacturing and design; 3) organizational resources; and 4) cultural. Proposition 1a: A platform strategy by HROs enables increased system-wide learning, which increases reliability. The complexity and tight coupling of the systems, along with the potential for a catastrophe, force HROs to prohibit experimen-

403

tation [38]. Experiential learning in HROs is impossible because trial-and-error approaches are not feasible and may easily lead to catastrophic loss [8], [72]. Instead, HROs must analyze a small sample of critical events for insights on how to improve [28]. For example, the Challenger inquiry changed management practices throughout National Aeronautics and Space Administration (NASA) and its whole complex of subcontractors [63]. A platform approach provides a large total system with more opportunities to learn as a community. Reliability is enhanced as more detailed, structured operating procedures evolve over time, learning from a few critical events and many minor incidents or near-misses. A platform strategy helps overcome the “stickiness” most organizations encounter as they try to transfer best practices between different parts of an organization. Szulanski [66] illustrates how the internal stickiness limiting the transfer of best practices is caused by the recipient’s lack of absorptive capacity, causal ambiguity, and an arduous relationship between the different parts of the organization. Each of these barriers is reduced when a platform strategy is used, because the different organizational entities have similar contexts, routines, and understanding of the technology. For example, every submarine in the Los Angeles class is quickly able to implement changes in best practice initiated on other submarines. In addition, a culture has been created wherein intersubmarine learning is not only encouraged but mandated. The receiving ship, because of the identical technical and operating systems and frequent rotation of sailors to different submarines of the same class, eliminates the causal ambiguity that other organizations encounter. A U.S. submarine shares learning opportunities with its class through two mechanisms: incident reports and crew rotation. Incidents and near-accidents, their root causes, and lessons learned are summarized in incident reports and circulated to all submarines in the class. These must be signed by all officers and the enlisted personnel in the pertinent area. Frequent crew rotation between submarines also aids learning. Typically, officers and enlisted personnel are attached to a submarine for about three years, then rotate to a land-based duty such as a school, then rotate to a different submarine, usually of the same class. This promotes consistency and reliability across all the submarines in the class. Additionally, the transfer of personnel is an effective way to transfer tacit operational knowledge across the class. The large number of submarines in the class provides sufficient variety for the collective, enabling it to evaluate more events and to facilitate organizational learning. According to Weick [71], “The. . .group enacts equivocal raw talk, the talk is viewed retrospectively, sense is made of it, and then this sense is stored as knowledge in the retention process.” Rickover and other submarine leaders also created a test system that demanded all submariners be actively involved in the system-wide learning processes. Each year, a team of highly qualified submarine experts conduct an Operational Reactor Safeguards Examination (ORSE) or operational test aboard each submarine. Initially, Rickover led each ORSE and ensured it would be comprehensive and difficult. He also ensured that there were severe repercussions for those who failed an ORSE. Crews spent hundreds of hours conducting drills, practicing

404


procedures, and reading the latest technical materials to prepare for these dreadful tests. However unpleasant, these tests forced submariners to embrace a learning culture, to learn continually about new techniques, and to understand truly the technologies used. The Russians, whose submarines were clearly less reliable, had a more difficult time encouraging system-wide learning. With fewer submarines in each class, they had fewer opportunities for crew rotation among submarines of the same platform. They also had more difficulty maintaining an organizational memory due to their higher turnover, which was driven by their use of conscription, many sailors were forced to serve on submarines, and the fact that naval service was not viewed as being as prestigious as it was in the United States. This view was shaped by history. Russia was primarily a land power with a limited naval tradition, so the Red Fleet was accorded less status and prestige than the Red Army. Proposition 1b: A platform strategy by HROs enables builders and designers to learn through experience (the learning curve effect), which increases system reliability. The shipbuilders learn through experience and improve manufacturing processes by carefully evaluating project effectiveness after each submarine is completed [6], [47], [77]. Building a larger number of submarines enables feedback from operators and manufacturers, and back to the designers so that they can make improvements on subsequent ships in the class. Likewise, the number of yards building the ships matters since, transferring the tacit knowledge associated with the building becomes more difficult with multiple shipyards. The Americans maximized this learning through experience by building a single class, the Los Angeles class, over an extended period of time and at only two yards. The Russians built many classes of submarines in five different shipyards, which limited their ability to benefit from tacit knowledge gained through their manufacturing experiences. Proposition 1c: A platform strategy by HROs enables an organization to spend more resources on training methods of operators, which increases system reliability. A platform strategy facilitates organizational learning by creating economies of scale for investments in training, since training costs can be amortized across all the units of the platform. With a lot of boats in a class, it is easier to justify spending significant resources on sophisticated and expensive training methods. For example, Los Angeles class operators were trained on advanced simulators developed to introduce them to many adverse conditions and dangerous situations. Full-scale control room simulators are used to practice using the submarine’s sonar and fire control equipment, and similar engine room simulators are used to practice combating floods and other potential problems. Experiential learning is not possible on an actual submarine but such expensive training methods are easier to justify given the large number of ships (63) in the Los Angeles class. Scale effects also justify other large investments in training since more sailors benefit, so more extensive classroom training could be developed. Additionally, it would make sense to invest more in each operator learning particular subsystems in-depth given the operator would likely serve later on another submarine of the same class rather than be transferred to a ship of a different class.

Proposition 1d: A platform strategy by HROs enables the organization to have a strong organizational culture of high reliability, which increases system reliability. In HROs, a culture that interacts with and supports the formal structure can provide additional control and be an important source of reliability [8], [72]. Wilkins and Ouchi [76] suggested a cultural mode of control is more effective when there is 1) long and stable membership; 2) absence of institutional alternatives; and 3) interaction among members. A platform strategy provides all three of these conditions, enhancing cultural control. The Los Angeles class was dominant for 30 years, allowing many officers and enlisted personnel the opportunity to spend their entire careers within this single service group. As we have described earlier, there is considerable interaction among all of the operators of this class of submarine, through both formal and informal communication networks. Bierly and Spender [8] illustrated that this strong submarine culture of reliability, created by Rickover, superseded a naval culture that encouraged more engineering ingenuity and risk taking. Among the Russians, there does not appear to be a strong culture of reliability because 1) the lack of powerful champions such as Rickover; 2) less stable group membership and interaction among sailors on the wide variety of types of submarines; and 3) political subordination of reliability in favor of other goals. Proposition 2: A platform strategy by HROs hinders revolutionary and architectural innovation, but encourages modular innovation. A successful platform, such as the Los Angeles class submarine, can last a long time and inhibit breakthrough changes-–especially with an HRO where the emphasis is on safety. Its operational success, reduced cost, design quality, and other scale economies led decision makers to resist replacement. Eventually, in 1982, it was decided to build a replacement, but the process was much delayed as subsequent administrations and admirals disagreed on its merits. Critics argued the high cost of the new submarine at $3 billion each, twice as much as a Los Angeles class submarine, was not worth the performance improvement. The Los Angeles class submarines were more economical as a result of the accrued system-wide learning and experience curve benefits. Similar cost reductions could not be realized for the Seawolf class until a large number were built, at a huge cost. These considerations-–plus inherent conservatism when dealing with HROs—created a major barrier to change. Similarly, the success of a dominant design acts as a barrier to radical changes in the way components are configured, demanding an architectural innovation [21]. However, the use of a platform strategy promotes modular innovation because technological breakthroughs at a component level can be implemented without waiting for other components to evolve or a new platform design be developed. Additionally, the standardized platform provides a form of tacit control by providing contractors stable boundaries around their product. An additional scale effect is that contractors, knowing the platform will be used for an extended period of time, will be motivated to invest resources to improve their component within that design. Proposition 3: A platform approach to HROs can reduce operational flexibility. By the mid-1990s, the U.S. had a submarine force almost entirely made up of only one type of attack


submarine, the Los Angeles class, and this was a very successful and capable submarine. However, a disadvantage is a loss of flexibility. Other studies have illustrated that using a platform and modularity can actually increase product flexibility since the firm can frequently change components while maintaining the system architecture [7], [50]. However, this was not the case in our research. Whenever the U.S. made an upgrade to a subsystem, all subsequent submarines also had that upgrade. The resulting submarine was very good at doing the many tasks for which it was designed, but left the U.S. without certain options. For instance, they did not have a very large ship, like the Russians’ Oscar class SSGN, that could rapidly fire 24 cruise missiles. Likewise, the U.S. also did not have small, fast submarines, such as the ill-fated Russian Alphas, which could be very useful for certain duties, such as gathering intelligence. This lack of flexibility became a major problem for the U.S. later on as the world order, and the demands of the military, changed. Shortly after the U.S. began building the very capable, but expensive Seawolf submarines, the focus shifted to the war on terrorism. As a result, they needed submarines that could operate in shallow waters (known as the littoral waters), excel at gathering intelligence, and provide “special forces” delivery and support. They stopped production of the Seawolf class and started production of the Virginia class, commissioning the first in 2004. Proposition 4: Innovation strategies are shaped by the decision-making process. The U.S. Navy’s new product development process involves many participants and stakeholders, including Congress, Department of Defense budget process, Navy leaders, academics, shipyard, defense contractors, consultants, etc. The politics around selecting a submarine design is complex and includes debates between parties intent on maximizing different performance parameters, along with others more focused on minimizing cost. After negotiation and compromise, the result is often a conservative design intended mainly to satisfy the largest number of stakeholders [3]. The process is unlikely to lead to a radical design that successfully challenges established routines and technologies. Once adopted, a successful design, such as the Los Angeles class design, will be maintained for a long time since it is very reliable and clearly “good enough.” In contrast, the Soviet decision-making process was autocratic and did not involve as much discussion. It was driven by the Party and the Navy leaders, without much internal debate. One man in particular, Igor Spassky, head of the Rubin Design Bureau for many years, dominated the decision-making process concerning Soviet submarine design and construction [40]. Given the support of the power-holders, this type of decisionmaking process is more open to radical ideas, regardless of concerns from less-powerful opponents and those operating the ships produced. In the case of the Soviet leaders, they were quick to test breakthrough technologies and appeared to be less concerned with reliability. This increased tolerance for risk may have been the result of these leaders knowing their fleet was operating with inferior submarines. We do not think this history is merely a story of Rickover’s success and Spassky’s failure. Rickover was operating from a position of considerable advantage. Spassky took design risks in an effort to improve the

405

performance of Russian submarines very quickly, and in this, he succeeded for the gap was closed, albeit at the cost of losing some 15 Russian submarines. It is also unfair to suggest the Soviets were indifferent to the fate of their submariners. As we noted, when discussing the loss of the Komsomolets, Soviet submarines had higher reserve buoyancies and escape chambers not possessed by U.S. submarines. However, when threatened by Russian improvements, the U.S. Navy was usually able to extend or regain its performance advantage via modular improvements that did not threaten reliability or operational success. VI. CONCLUSION HROs are comprised of complex, tightly coupled systems involving human judgments and errors, where failure results in a catastrophe. The loss of the Thresher illustrates the interactive complexity of nuclear attack submarines and so positions them as an HRO. But even the most resilient teamwork cannot compensate for major design failure. Research suggests HROs will be more successful if they facilitate a learning environment, manage the tension between efficiency and reliability, and clearly articulate the organization’s purpose. This paper has illustrated how the Los Angeles class program seems to support those three key considerations. What makes the Los Angeles class program so remarkable as an HRO was how it proved possible to maintain its advantages over a more innovative rival. We are disposed to think that new is better, and that innovation and dynamism must triumph over stability and reliability. But, in the realm of HROs, things may be more complicated. While the innovative and dynamic Soviets were able to erode the design advantages held by the Los Angeles class, they never fully matched the Improved Los Angeles class ships, nor did they achieve anything like the same level of operational performance. Soviet gains in design performance parameters such as speed and diving depth were achieved at high costs in less reliability and stealth. The numerous documented mechanical failures and mishaps that occurred to Soviet submarines mentioned in this paper occurred during relatively benign peacetime activities, such as lying alongside a pier, rather than in combat. It is not a great leap to suggest that, had swords been crossed, the United States’ operational advantages would have asserted themselves in an even more dramatic fashion. As literature on platform strategies has become more sophisticated, greater discussion has emerged about when not to use platforms, such as when a market is either extremely diverse or homogeneous [26]. When considering HROs and their potential to create a catastrophe, reliability is probably the most important consideration that a platform strategy can help us address. This means distinguishing clearly between the stable platform and the dynamism of the components. By designing interfaces that facilitated keeping the platform’s major component, the reactor, stable, the Los Angeles class was able to meet performance gains by the Soviets while remaining reliable. The Los Angeles class platform performed so well because it segmented the critical reliability-defining components, e.g., nuclear reactor, from the performance components, e.g., weapons

406


systems. For example, after the Soviet’s guided missile submarines appeared, the Los Angeles class was lengthened to incorporate additional cruise missile launch tubes. While this required an adjustment to something as fundamental as the external dimensions of the ship, it left critical reliability components such as the reactor, fire control, and emergency surfacing procedures unaltered. Furthermore, it did not radically alter the interface standards between the submarine, its human crew, and its other components. Had this modification only occurred in an entirely new class of submarines (i.e., a new platform), issues surrounding reactor safety and emergency procedures would need to be revisited and crews retrained. We see many commercial possibilities for improving a product platform while insuring stability for its reliability components. Boeing successfully “stretched” its existing line of airliners for decades, especially the 737 and 747s. Similarly, Lockheed’s C-130 transport aircraft was modified for a variety of missions over several decades, from a simple military transport into an ambulance and gunship. These platform approaches left many critical reliability-influencing components, such as cockpit controls, unmodified. Looking forward, the United States seems to be regaining an interest in nuclear power for electricity generation. Given the balkanized nature of electric utilities in the United States, active involvement by the Nuclear Regulatory Commission to insure that data and experiences are shared across a mandated standard design and control system for nuclear power plants would seem to be a key suggestion from this paper. We believe a better understanding of the success of the Los Angeles program can serve as a model for HRO firms attempting to balance critical reliability and performance tradeoffs. We agree that adopting a platform strategy does not necessarily make a HRO more stable or reliable. The success of the Los Angeles class program was facilitated by its stability but would not have happened without efforts to develop learning throughout the man–machine operational complex. The Navy’s experience must be contrasted with NASA’s experience with losing two of its five orbiter shuttles in spite of their stable design and interface standards. The Columbia Accident Investigation Board concluded that the Columbia was lost partially because the lessons of the Challenger catastrophe were lost to NASA. NASA lost this key knowledge due to layoffs and outsourcing as its budget was cut 40% [10].3 Vaughn [70] likewise provides an interesting analysis for why NASA’s institutional learning was so poor as a result of their inability to identify and correct the social problems in their systems. We believe the contrast between the NASA and Naval Nuclear Propulsion Program further highlights the importance of platforms and standards as mediators of operator and system learning. Of course, there are many limits of our research. The nature of HROs means that they are relatively rare, and fortunately, deviances from their reliable operation are even rarer. This generally necessitates a case- or cross-case-based approach for studying them, e.g., [38]. This does constrain formal statistical hypothesis testing and does not allow us to rule out all possible 3 These conclusions are from p. 118 in Chapter 5 of the report that specifically tracks the lessons from the loss of the Challenger to the loss of the Columbia.

rival explanations for the disparate results seen between the two navies. We have attributed the reliability problems the Soviets had to contrasts with known U.S. Navy procedures. However, the U.S. Navy procedures are more clearly known to us than Soviet/Russian ones are. Therefore, some admittedly unforeseen alternative explanation may yet emerge to explain the variances in reliability we see between the two nations’ navies. We should reiterate that we do not believe the Soviets were indifferent to their submarine crews; for example, the USN still does not possess escape chambers such as those found on Russian submarines. However, we remain confident in the strong value a platform approach has in facilitating a HRO’s reliability during the innovation process, as illustrated here. An unfortunate characteristic of HRO research is its focus on failures and disasters, such as Challenger and Bhopal, rather than on successes. In thinking about future research, we hope that our analysis will encourage additional work on the numerous HRO successes. Because failure results in such dramatic consequences, like the maxim that bad news sells more newspapers, HROs normally receive attention only when things go wrong. Yet, these increasingly prevalent and complex systems typically do an amazing job day after day meeting the critical needs of our increasingly complex society. While we have offered some propositions explaining why a platform strategy may provide reliability while also enabling flexibility, we have not identified which of these factors has a stronger impact. Is system-wide learning across submarines more important than the shipbuilder’s learning curve? Is operator training more important than having a culture of reliability? Under what conditions does a platform approach to HROs increase operational flexibility, and under what conditions does a platform approach decrease operational flexibility? These types of questions cannot be answered in a single study such as ours. We encourage future research on the success of HROs and their ability to innovate. More studies in different contexts will enable us to better understand these more complex issues associated with HROs. REFERENCES [1] W. J. Abernathy and J. M. Utterback, “Patterns of industrial innovation,” Technol. Rev., vol. 80, no. 7, pp. 40–47, 1978. [2] A. Afuah, Innovation Management: Strategies, Implementation, and Profits. New York: Oxford Univ. Press, 1998. [3] G. T. Allison, Essence of Decision Making: Explaining the Cuban Missile Crisis. Boston, MA: Little Brown, 1971. [4] P. Anderson and M. L. Tushman, “Technological discontinuities and dominant designs: A cyclical model of technological change,” Adm. Sci. Quart., vol. 35, no. 4, pp. 604–633, 1990. [5] W. B. Arthur, “Competing technologies, increasing returns, and lock-in by historical events,” Econ. J., vol. 99, pp. 116–131, 1989. [6] C. Baden-Fuller, “The implications of the learning curve for firm strategy and public policy,” Appl. Econ., vol. 15, pp. 541–551, 1983. [7] C. Y. Baldwin and K. B. Clark, Design Rules, vol. 1, The Power of Modularity. Cambridge, MA: MIT Press, 2000. [8] P. E. Bierly and J.-C. Spender, “Culture and high reliability organizations: The case of the nuclear submarine,” J. Manage., vol. 21, no. 4, pp. 639– 656, 1995. [9] C. M. Christensen, F. F. Suarez, and J. M. Utterback, “Strategies for survival in fast-changing industries,” Manag. Sci., vol. 44, pp. 207–220, 1998. [10] “Columbia Accident Investigation Board,” vol. 1, Superintendent Doc., U.S. Gov. Printing Office, Aug. 2003.


[11] M. A. Cusumano and A. Gawer, “The elements of platform leadership,” Sloan Manag. Rev., vol. 43, no. 3, pp. 51–58, 2002. [12] F. Duncan, Rickover and the Nuclear Navy: The Discipline of Technology. Annapolis, MD: Naval Inst. Press, 1990. [13] T. J. Eccles and H. S. Marcus, “Integrating design and production: A case study of the naval submarine program,” Int. J. Prod. Manag., vol. 28, pp. 107–126, 1992. [14] K. M. Eisenhardt, “Building theories from case study research,” Acad. Manag. Rev., vol. 14, no. 4, pp. 532–550, 1989. [15] S. K. Ethiraj and D. Levinthal, “Modularity and innovation in complex systems,” Manage. Sci., vol. 50, no. 2, pp. 159–173, 2004. [16] W. M. Evan and M. Manion, Minding the Machines: Preventing Technological Disasters. Upper Saddle River, NJ: Prentice-Hall, 2002. [17] J. Farrell and G. Saloner, “Installed base and compatibility: Innovation, product preannouncements, and predation,” Amer. Econ. Rev., vol. 76, pp. 940–955, 1986. [18] Federation of American Scientists [Online]. Available: http://www.fas.org (accessed Sep. 2007). [19] Federation of American Scientists. (2003). Project 685 Plavnik, Mike class [Online]. Available: http://www.fas.org/man/dod-101/sys/ship/ row/rus/685.htm [20] S. Gallagher, “The complementary role of dominant designs and standards,” IEEE Trans. Eng. Manag., vol. 54, pp. 371–379, 2007. [21] R. M. Henderson and K. B. Clark, “Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms,” Adm. Sci. Quart., vol. 35, no. 1, pp. 9–30, 1990. [22] C. W. L. Hill, “Establishing a standard: Competitive strategy and technological standards in winner take all industries,” Acad. Manag. Exec., vol. 11, no. 2, pp. 7–26, 1997. [23] R. Hutchinson, Jane’s Submarines: War Beneath the Waves. London, U.K.: Harper Collins, 2001. [24] M. L. Katz and C. Shapiro, “Network externalities, competition, and compatibility,” Amer. Econ. Rev., vol. 75, no. 3, pp. 424–440, 1985. [25] E. H. Kessler, P. E. Bierly, and S. Gopalakrishnan, “Vasa syndrome: Insights from a 17th-century new-product disaster,” Acad. Manag. Exec., vol. 15, no. 3, pp. 80–91, 2001. [26] V. Krishnan and S. Gupta, “Appropriateness and impact of platformbased product development,” Manag. Sci., vol. 47, no. 1, pp. 52–68, 2001. [27] R. N. Langlois, “Modularity in technology and organization,” J. Econ. Behav. Org., vol. 49, pp. 19–37, 2002. [28] J. G. March, L. S. Sproull, and M. Tamuz, “Learning from samples of one or fewer,” Org. Sci., vol. 2, no. 1, pp. 1–13, 1991. [29] M. H. Meyer and D. Dalal, “Managing platform architectures and manufacturing processes for nonassembled products,” J. Prod. Innov. Manag., vol. 19, pp. 277–293, 2002. [30] M. H. Meyer and A. P. Lehnerd, The Power of Product Platforms: Building Value and Cost Leadership. New York: Free, 1997. [31] J. Mikkola, “Capturing the degree of modularity embedded in product architectures,” J. Prod. Innov. Manag., vol. 23, pp. 128–146, 2006. [32] Military.com [Online]. Available: http://www.military.com (accessed Sep. 2007). [33] M. Miozzo and D. Grimshaw, “Modularity and innovation in knowledgeintensive business services: IT outsourcing in Germany and the UK,” Res. Policy, vol. 34, pp. 1419–1439, 2005. [34] G. Montgomery, “The Komsomolets disaster,” Cent. Intell. Agency, CSI Publ., vol. 38, no. 5, 1995. [35] C. R. Morris and C. H. Ferguson, “How architecture wins technology wars,” Harv. Bus. Rev., vol. 71, no. 2, pp. 86–96, 1993. [36] Navy Office of Information [Online]. Available: http://www.chinfo. navy.mil/ (accessed Sep. 2007). [37] T. Nilsen, I. Kudrik, and A. Nikitin, “The Russian northern fleet nuclear submarine accidents,” in The Russian Northern Fleet: Sources of Radioactive Contamination. Murmansk, Russia: Bellona Foundation, 1996. [38] C. Perrow, Normal Accidents. New York: Basic Books, 1984. [39] N. Polmar and T. B. Allen, Rickover: A Biography. New York: Simon and Schuster, 1985. [40] N. Polmar and K. J. Moore, Cold War Submarines: The Design and Construction of U.S. and Soviet Submarines. Washington, DC: Brassey’s, 2004. [41] N. Polmar and J. Noot, Submarines of the Russian and Soviet Navies, 1718–1990. Annapolis, MD: Naval Inst. Press, 1991. [42] K. H. Roberts, “Managing high reliability systems,” Calif. Manag. Rev., vol. 32, no. 4, pp. 101–113, 1990.

407

[43] K. H. Roberts and R. Bea, “Must accidents happen? Lessons from highreliability organizations,” Acad. Manag. Exec., vol. 15, no. 3, pp. 70–78, 2001. [44] K. H. Roberts and D. M. Rousseau, “Research in nearly failure-free, highreliability organizations: Having the bubble,” IEEE Trans. Eng. Manag., vol. 26, no. 2, pp. 132–139, 1989. [45] D. Robertson and K. Ulrich, “Planning product platforms,” Sloan Manag. Rev., vol. 39, no. 4, pp. 19–31, 1998. [46] E. Romanelli and M. L. Tushman, “Organizational transformation as punctuated equilibrium: An empirical test,” Acad. Manag. J., vol. 37, no. 5, pp. 1141–1167, 1994. [47] D. Ross, “Learning to dominate,” J. Ind. Econ., vol. 34, pp. 337–353, 1986. [48] Russian Navy Weapons Catalog [Online]. Available: http://www. militarism.navy.ru (accessed Sep. 2007). [49] F. Salvador, “Toward a product system modularity construct: Literature review and reconceptualization,” IEEE Trans. Eng. Manag., vol. 54, no. 2, pp. 219–240, 2007. [50] R. Sanchez and J. T. Mahoney, “Modularity, flexibility, and knowledge management in product and organization design,” Strateg. Manag. J., vol. 17, pp. 63–76, 1996. [51] C. Shapiro and H. R. Varian, Information Rules: A Strategic Guide to the Network Economy. Boston, MA: Harvard Business School Press, 1999. [52] M. A. Schilling, “Toward a general modular systems theory and its application to interfirm product modularity,” Acad. Manag. Rev., vol. 25, no. 2, pp. 312–334, 2000. [53] M. A. Schilling and K. Steensma, “The use of modular organizational forms: An industry level analysis,” Acad. Manag. J., vol. 44, pp. 1149– 1169, 2001. [54] D. Schubert, Commanding Officer Naval Research Laboratory. Washington, DC, private communication, Oct. 18, 2005. [55] P. Shrivastava, Bhopal: The Anatomy of a Crisis. Cambridge, MA: Ballinger, 1986. [56] N. Siggelkow and D. Levinthal, “Temporarily divide to conquer: Centralized, decentralized, and reintegrated organizational approaches to exploration and adaptation,” Org. Sci., vol. 14, pp. 650–669, 2003. [57] H. A. Simon, Administrative Behavior. New York: Macmillan, 1947. [58] T. W. Simpson, “Product platform design and customization: Status and promise,” Artif. Intell. Eng. Des. Anal. Manuf., vol. 18, no. 1, pp. 3–20, 2004. [59] K. K. Sinha and A. H. Van de Ven, “Designing work within and between organizations,” Org. Sci., vol. 16, pp. 389–408, 2005. [60] I. Spassky, Five Colors of Time. Rubin, Russia: Central Design Bureau for Marine Engineering, 2002. [61] R. Srinivasan, G. L. Lilien, and A. Rangaswamy, “The emergence of dominant designs,” J. Mark., vol. 70, pp. 1–17, 2006. [62] W. H. Starbuck and M. Farjoun, Eds., Organization at the Limit: NASA and the Columbia Disaster. Malden, MA: Blackwell, 2005. [63] W. H. Starbuck and F. J. Milliken, “Challenger: Fine-tuning the odds until something breaks,” J. Manag. Stud., vol. 25, no. 4, pp. 319–340, 1988. [64] N. Staudenmayer, M. Tripsas, and C. L. Tucci, “Interfirm modularity and its implications for product development,” J. Prod. Innov. Manag., vol. 22, pp. 303–321, 2005. [65] F. F. Su´arez and J. M. Utterback, “Dominant designs and the survival of firms,” Strateg. Manag. J., vol. 16, no. 6, pp. 415–430, 1995. [66] G. Szulanski, “Exploring internal stickiness: Impediments to the transfer of best practices within the firm,” Strateg. Manag. J., vol. 17, pp. 27–43, 1996. [67] United States Naval Institute [Online]. Available: http://www.usni.org (accessed Sep. 2007). [68] J. M. Utterback, Mastering the Dynamics of Innovation. Boston, MA: Harvard Business School Press, 1994. [69] D. Vaughn, The Challenger Launch decision: Risky Technology, Culture, and Deviance at NASA. Chicago, IL: Univ. of Chicago Press, 1996. [70] D. Vaughn, “System effects: On slippery slopes, repeating negative patterns, and learning from mistakes,” in Organization at the Limit: NASA and the Columbia Disaster, W. H. Starbuck and M. Farjoun, Eds. Malden, MA: Blackwell, 2005. [71] K. E. Weick, The Social Psychology of Organizing, 2nd ed. Reading, MA: Addison-Wesley, 1979. [72] K. E. Weick, “Organizational culture as a source of high reliability,” Calif. Manag. Rev., vol. 24, no. 2, pp. 112–127, 1987. [73] K. E. Weick, “The vulnerable system: An analysis of the Tenerife air disaster,” J. Manag., vol. 16, pp. 571–593, 1990.

408


[74] K. E. Weick and K. H. Roberts, “Collective mind and organizational reliability: The case of flight operations on an aircraft carrier deck,” Adm. Sci. Quart., vol. 38, pp. 357–381, 1993. [75] K. E. Weick, “Making sense of blurred images: Mindful organizing in mission STS-107,” in Organization at the Limit: NASA and the Columbia Disaster, W. H. Starbuck and M. Farjoun, Eds. Malden, MA: Blackwell, 2005. [76] A. Wilkins and W. G. Ouchi, “Efficient cultures: Exploring the relationship between culture and organizational performance,” Adm. Sci. Quart., vol. 28, pp. 468–481, 1983. [77] L. E. Yelle, “The learning curve: Historical review and comprehensive survey,” Decis. Sci., vol. 10, pp. 302–328, 1979.

Paul E. Bierly, III, received the B.S. degree from Wharton School, University of Pennsylvania, Philadelphia, the B.A.S. degree from the Engineering School, University of Pennsylvania, both in 1983, and the MBA and Ph.D. degrees from Rutgers University, New Brunswick, NJ, in 1992 and 1995, respectively. He is currently the Zane Showker Professor of Entrepreneurship and Director of MBA Programs at James Madison University, Harrisonburg, VA. Previously, he was an Officer on a fast-attack nuclear submarine in the U.S. Navy’s Nuclear Power Program, a Manager at Johnson & Johnson, and a Consultant for Princeton Economic Research, Inc. He is the author or coauthor of more than 35 papers published in the Strategic Management Journal, Long Range Planning, Academy of Management Executive, Journal of Management, the IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, Journal of Engineering and Technology Management, R&D Management, and numerous other management journals and books. His current research interests include management of technology, innovation, knowledge management, and strategic alliances. Prof. Bierly is currently a Department Editor for the IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT.

Scott Gallagher received the B.B.A. degree from the University of Texas at Austin, Austin, in 1989, the M.P.P. degree from John F. Kennedy School of Government, Harvard University, Cambridge, MA, in 1991, and the Ph.D. degree from Rutgers University, New Brunswick, NJ, in 2000. He was an Administrator at Lamar University, Beaumont, TX, and then joined the faculty of James Madison University, Harrisonburg, VA, in 2000. His current research interests include management of innovation, standards, and strategic alliances.

J. C. Spender was educated in the U.K. at Oxford University (Engineering) and Manchester Business School (strategy), and received the B.A. and M.A. degrees from Oxford, in 1960 and 1965, respectively, and Ph.D. degree from Manchester Business School, Manchester, U.K., in 1980. He has been involved in service of experimental submarines, nuclear submarine power-plant design and test facility building, large-scale computing, merchant banking, and consulting. He has served as Chair of Entrepreneurship and Small Business at Rutgers University, New Brunswick, NJ, and has been on the faculties of University of California, Los Angeles (UCLA), City University, London, U.K., and the University of Glasgow, Glasgow, U.K. He retired in 2003 but retains Visiting Research Professorships at Lund University, ESADE (Spain), Cranfield, Leeds, and Open Universities (U.K.). He is engaged in researching a theory of organization and managerial rhetoric. He is the 2007–2008 Fulbright Visiting Professor of Knowledge Management at Queen’s University, Kingston, ON, Canada. He is the author or coauthor of numerous articles on knowledge management and Industry Recipes (Basil Blackwell, 1989). Prof. Spender is a member of the Royal Society of Arts, U.K.