In the wake of the last half-century lay several ignominious Navy, and more broadly, Department of Defense acquisition programs that sought to transform contemporary warfighting platforms and paradigms. Although they attempted to meet largely unsubstantiated future threats, they neglected current short-term defense needs—a condition known in some circles as “next-war-itis.” These missteps have come at an unacceptable cost to the taxpayer and have tarnished the reputation of the defense-acquisition community.
Programs that need no introduction are the Seawolf-class submarine, littoral combat ship (LCS), Zumwalt-class destroyer, and the Joint Strike Fighter. The Air Force’s Expeditionary Combat Support System, the F-22 Raptor, the Army’s Comanche helicopter, the Crusader howitzer, the Joint Tactical Radio System, and the Future Combat System are some infamous pariahs among the lot. Though not a branch of the DOD, the Coast Guard’s embattled Integrated Deepwater System program cannot be left out. These foundering systems have been either cancelled or forced through the Integrated Defense Acquisition, Technology, and Logistics Lifecycle Management system to achieve some semblance of success at great cost to quantity of warfighting capability.
While these controversial acquisition programs occupy national news headlines, combatant commanders rely on proven systems that have evolved over time—such as the Los Angeles-, Ohio-, Arleigh Burke-, Spruance-, and Nimitz-classes, Super Hornet, Seahawk, and Tomahawk—to defend the nation and its interests. These platforms have reputations of being reliable, adaptable, capable, and survivable. Each of the DOD’s services maintains this core set of weapon systems, which were developed in parallel with the threats and advancing technologies to maintain U.S. military dominance. As unforeseen menaces arise, such as the improvised explosive device during Operations Enduring Freedom and Iraqi Freedom, the full weight of the military industrial complex is thrown at the problem.
In parallel with the recent history of troubled acquisition programs, numerous public servants have dissected the failings and implemented countless new restrictions and program-development models. The current state of the Joint Strike Fighter and recent developments with the LCS have shown that defense-acquisition reform efforts have produced little except nebulous acronyms and self-named reports. While this bureaucratic mountain does, to some extent, prevent the forward movement of embattled programs, it largely fails to address the fundamental mindset that breeds programs of failure: the perpetual revolution in designing complex systems.
An Unfortunate Precedent
Before the acquisition faux pas of the 20th century, early-American naval strategists and engineers set the standard for schedule and cost overruns. During the Civil War, for example, the immediate success of the USS Monitor design incited fervor among Union admirals who envisioned the impenetrable iron vessels steaming up rivers and through shallow harbors to lead army assaults and provide coastal defense. But these new steamships would need a smaller, lighter draft than the original Monitor designed by John Ericsson, and without a sufficient design history and trial period, knowledge was limited among shipbuilders and naval constructors in the design and construction of these revolutionary vessels.
The program’s chief engineer, Alban Stimers, had previously worked alongside Ericsson on the original Monitor’s construction and the larger, successful, follow-on Passaic class in 1862.1 For the new light-draught, river-patrolling Casco class envisioned by Navy brass, Ericsson again applied his design expertise and produced a simple iron vessel with a draft of six-and-a-half feet, a speed of eight knots, and an estimated construction time of 40 days. Despite the design’s successful beginnings, its fate was sealed with the inception of the Monitor Design Office in Washington, D.C., which gave Stimers and senior Navy line officers bureaucratic control over Ericsson’s design.2
While Ericsson’s initial proposal can be criticized today for its basic naval architecture faults, Stimers’ frequent design alterations and the commercial shipbuilders’ lack of experience with all-iron vessels doomed the effort. The greatest design change, which ultimately led Ericsson to leave the program, was the addition of ballast tanks along the length of the hull to decrease exposed freeboard during engagements. While certainly novel, the changes added substantial machinery and structural weight to the already constrained vessel displacement.
The design office awarded 20 construction contracts for the non-prototyped design to multiple shipyards across the Union, each with varying iron-vessel construction experience (and some having none). As with most troubled acquisition programs, ceaseless design alterations were issued after contracts of $395,000 per vessel were awarded. Stimers’ own signature accompanied 83 drawing changes and 120 letters of deviation to one yard in Boston. When the first of the vessels to be completed, the USS Chimo, was launched in South Boston 14 months after ordered (approximately 980 percent schedule growth from Ericsson’s initial 40 day construction period), the main deck settled three inches above the water. The intended launching draft of six-and-a-half feet was instead eight feet.3 With the waterborne installation of the iron turret and a full complement of ammunition and coal, the draft would grow to 10 feet and fully submerge the main deck.
Perhaps even more damning, speed trials revealed the vessel struggled to make three-and-a-half knots, rendering river travel impossible. As the remaining 19 vessels approached their launch dates, it became apparent that the entire class was grossly overweight and embarrassingly underpowered. Per-vessel cost ballooned from $395,000 to $500,000, a 27 percent cost growth. During the subsequent congressional inquiry, shipbuilders openly testified of their refusal to ride the boats into open water, certain they would sink. The truth was inescapable: The entire class of 20 light-draught monitors, all in varying stages of construction, was unsuitable for naval service. The magnitude of the failure is encapsulated in Ericsson’s response to the congressional panel: “I have nothing whatever to do with those twenty monitors, directly or indirectly.”4
Chief Engineer Stimers drove the program to ruin with unchecked design changes and the introduction of revolutionary—and undeveloped—design concepts. Having squandered $10 million of wartime funds (in 1864 dollars), the Navy relieved him from office.5 Though the remaining contracted vessels were launched, none was used in fighting war, many went directly into lay-up, and all were scrapped by 1875.6
A century-and-a-half later, numerous parallels can be drawn between current acquisition programs and the light-draught monitor debacle.7 Weapon platforms have increasingly become systems of systems of systems in the last half-century. Once autonomous actors linked only by visual or line-of-sight communications, technology has reduced naval combatants to mere nodes in a global- defense network. While this paradigm has its warfighting benefits, it comes at the expense of exponentially increasing technical and organizational complexity.
The DOD adopted the systems engineering “V-model,” (see Figure 1), to control complexity in the design process through rigorous testing and verification. The model dictates an iterative cycle of requirements definition, designing, testing, and requirements validation. In theory, it prevents capability growth and promotes total system integration. Though effective in bounding the acquisition problem, recent history has shown that forces outside the ability of systems engineering are at work.
Complex system design can be envisioned as a three-dimensional array of Design Attributes (e.g., hull-form, propulsion plant, manning plan, etc.), Lifecycle Entities (e.g., Program Executive Offices, industry, universities, naval shipyards, regional maintenance centers, Navy training centers, Navy personnel command), and the existing Paradigms, Processes, and Capabilities (PPCs) within each entity (e.g., training pipelines, logistics streams, research and development efforts, production and manufacturing methods). Lifecycle entities (LCEs) are included rather than solely non-recurring design and acquisition entities because, historically, 60 percent of program costs are incurred during the operation and support phase for capital assets.8
As design attributes are selected, LCE are incorporated into the arrays, along with each entity’s unique internal PPCs. Consider the major design attributes of propulsion. Predominant forms in the Navy include combined gas and gas (COGAG), combined diesel and diesel (CODAD), and nuclear fission. Selecting nuclear fission incorporates naval reactors, naval shipyards, the Department of Energy, Knolls-Bettis, Naval Nuclear Power School, and the Hanford naval-reactor disposal site, among many others, into the LCEs. Finally, PPCs are within each of these entities, which have been molded and entrenched over time to meet the past and current Fleet composition. Program schedule and cost risk stem directly from the number of incorporated LCEs, standing up new LCEs, and the disruption of their existing internal PPCs.
To illustrate this, imagine the selection of a more efficient combined diesel and gas (CODAG) propulsion train over a guided-missile destroyer’s COGAG arrangement. Selecting identical gas turbines to previous ship classes and diesel engines, which are adaptations from existing CODAD ships, reduces disruption to the existing PPCs and the parent LCEs while still achieving change. This evolutionary approach reduces the probability of failure, generalized as cost and schedule growth, by capitalizing on existing LCEs and PPCs. If an entirely new propulsion train were selected, such as a CODAG propulsion plant consisting of both gas turbines and diesels from two different manufacturers (both of which have never worked with the U.S. Navy), the ripples sent through the complex system-design matrix would be far reaching and severe, though not necessarily fatal. As further revolutionary design attributes are selected and cause program disruption, however, the overall probability of program success declines. The LCS program is an excellent model of this process. Just how much, or at what rate, does the probability of success fall as disruption is introduced to the LCEs and PPCs by adverse design decisions?
The Probability of Success
Because analysis without numbers is only an opinion, reviewing the laws of probability underscores the effects of revolutionary thinking in designing complex systems. Consider again the selection of propulsion plant for a new class of small surface combatants. The overall success of the program relies on the success of each of its LCEs. In general, the probability of success of each effort at the lowest tier of the program hierarchy either is or is not statistically dependent on the success of other efforts in the level. For example, the success of research efforts at a university does not affect the success of the naval training center tasked with building a new curriculum for sailors, though the research may impact the work and general success of Naval Ship Systems Engineering Stations. Likewise, the success of the university does not preclude the success or failure of the training center; they are not mutually exclusive. However, the success of the propulsion train throughout its lifecycle relies on the combined probability of success of all of these entities.
The probability of success of the propulsion plant example over the program lifecycle follows the multiplication rule, not the theorem of total probability, due to the lack of mutual exclusivity. Quantitatively, the impact of increasing and altering LCEs and their PPCs, driven by the high-level design attributes, is exponential. As complex tasks are broken down into more manageable parts across more entities, the number of statistically interdependent LCEs naturally grows as well. The impact on the combined success is therefore doubly affected by this inherent dependency across some of the LCEs. Additionally, the individual probabilities of success for newer innovative systems in development phases will naturally be lower. The probability of overall success decays exponentially as more LCEs are incorporated (or are flat-out created) to control the additional complexity caused by high-level, revolutionary design attributes. This paradox is inescapable, and the response of “contract it out” or “divide and conquer” only compounds the problem.
The Program-Complexity Paradox
The commercial world, though generally more economically risk averse, is not immune to the program- complexity paradox. Boeing’s 787 Dreamliner sought to change large-body commercial aircraft design and construction at all levels of the design hierarchy. A one-piece, composite barrel-section fuselage, new engines, and a truly revolutionary supply and production model are among its openly discussed and heavily criticized program attributes. The delivery of the first plane slipped three years from the intended delivery date as the company struggled to meet the intended aircraft gross weight and maintain the flow of its globally distributed supply chain.9
Many other mechanical and electrical issues, aircraft groundings, and delayed deliveries have been well publicized. The space shuttle program (STS) that began in 1972, though boasting many accomplishments and technological firsts, was a programmatic disaster for NASA. A gigantic leap from the Apollo program, STS required new support infrastructure, training, and logistics—think LCEs and their PPCs—aside from the obvious design departures from previous space vehicles.10 Early program goals included 50 missions (launch and landing) a year at a cost of $54 million each with a non-recurring cost of $43 billion. Through 2011 when the program ended, the total cost broke $196 billion with a conservative cost of $450 million per mission (all in approximate Fiscal Year 2011 dollars).11 The 135 STS flights divided over the program’s 30-year lifecycle equals 4.5 flights per year. Both flights per year and cost per mission were an entire order of magnitude from the initial program goals.
Despite this glum history, and the ever-increasing complexity and interoperability of defense systems, continual failure to meet program cost, schedule, and capabilities is not guaranteed. The paradox and its effects can be mitigated by limiting revolutionary ideas to the lowest tier of the hierarchy: paradigms, processes and capabilities. Above this, program managers must adhere to evolutionary engineering principles in making high-level design decisions. As new concepts are developed, tested, and perfected by subject matter experts in the lowest tier, they will percolate up and across the design hierarchy in a natural progression as their defense need is validated. A limited amount of revolutionary movements may be injected at the design attributes level, but only when they meet a well-defined and verified defense requirement. The disturbance to existing lifecycle entities and their paradigms, processes, and capabilities must be mapped and fully understood prior to incorporating the design attributes.
Identifying a “Revolution”
The modular mission warship is not a new or untested concept. The Royal Danish Navy first implemented modular mission payloads, coined “standard flex” (STANFLEX), on board Flyvefisken-class patrol vessels in the early 1980s. Swapping the modules allowed for the small combatants to perform roles in surveillance and pollution mitigation, surface engagement, mine countermeasures/mine hunting, or mine laying. STANFLEX was so successful that the Royal Danish Navy incorporated it into nine ship classes, including its air-defense frigates (the Iver Huitfeldt class) and its combat command-and-support frigates (the Absalon class).12
The key difference between the seamless STANFLEX implementation and troubled LCS mission modules is obvious. STANFLEX modularized existing weapons systems while the LCS program, with the one exception of the 30-mm Bushmaster IIs, sought to modularize weapons and combat systems still in their developmental phase or systems new to the U.S. Navy. The success of the mission packages, and the LCS’s combat effectiveness, depends not only on successful modularization, but also on a new logistics supply chain, new maintenance plans, new sailor training plans and facilities, and new manufacturers—or, in the context of this article, countless new LCEs and entirely new PPCs. Just as dictators squash a revolution at its roots for fear of compounding disruption to the higher echelons of government, program managers must battle attempts to introduce revolutionary concepts at the design attributes and lifecycle entity levels and contain them within the program’s PPCs to prevent the exponential decay of their probability of success.
These revolutions are easily identifiable. In the embryonic stages of a program, descriptors such as “clean-slate, ground-up, and transformative” should send chills up the spine of a program manager. Concept drawings can capture the imaginations of an audience and build enthusiasm, but they also serve as visual cues to underlying revolutionary intentions.
With such clear indicators, why do these program failures occur? Simply put, it’s a botched vision, the founding principle of any program. Promises of increased propulsion efficiency or greater warfighting capability at a reduced or acceptable cost become tag lines surrounding artist conceptions of what the eventual product should look like. They exploit our first-world obsession with the future instead of respecting the past and present. Though this disposition allows us to foresee oncoming conflicts, we have become unconsciously bored of the present state and wholly ignorant of the past.
Unfortunately, as many DOD leaders have recognized, we are 100 percent successful in misjudging the location and nature of our next engagement.13 A maxim attributed to Georg von Tiesenhausen, a founding father of rocketry in the United States, sums up the problem. “If you want to have a maximum effect on the design of a new engineering system, learn to draw. Engineers always wind up designing the vehicle to look like the initial artist’s rendition.” At the lowest level, fiction is turned to fact instead of fact being rationally turned to new, improved fact. A vision should not promise anything, either verbally or through a fantastical computer-aided design rendering. Eliminating program visions and adopting a common DOD or branch-specific acquisition vision may alleviate the pitfalls of otherwise grandiose promises.
Fiscal constraints are nothing new and should be welcomed with open arms for the future of the defense establishment. Program managers must act as gatekeepers fending off the forces of revolution led by bureaucratic “forward-thinkers,” self-interested lobbyists, and overzealous senior officers (both unrestricted and restricted line). This period of scarcity is an opportunity to consolidate lifecycle entities, refine their paradigms, processes and capabilities, and recapitalize the organic, broad technical workforce in the systems commands that has atrophied over the past few decades. Whether the defense acquisition establishment evolves to meet these budgetary limitations and breaks two centuries of revolution is uncertain, but entirely possible. After all, revolutions are costly, bloody affairs, and evolution is seamless.
2. Thirty-Eighth Congress of the United States, Report of the Joint Committee on the Conduct of War (Washington, DC: Government Printing Office), 1865.
6. Dictionary of American Naval Fighting Ships, Navy History and Heritage Command, http://www.history.navy.mil/danfs/index.html.
7. Thirty-Eighth Congress of the United States, Report of the Joint Committee on the Conduct of War.
8. Engineering Duty Officer School, “Lecture 3-1-6 Cost Estimating,” October 2012.
9. Boeing, “Boeing, ANA Celebrate First 787 Dreamliner Delivery,” http://boeing.mediaroom.com/index.php?s=20295&item=1939, 26 September 2011.
10. Jon Ostrower, “Boeing Confirms 787 Weight Issues,” www.flightglobal.com/news/articles/boeing-confirms-787-weight-issues-326087, 7 May 2009. Thomas Heppenheimer, “The Space Shuttle Decision, NASA’s Search for a Reusable Space Vehicle” (Washington, DC: NASA History Office, Government Office of Policy and Plans), 1999.
11. Radford Byerly and Roger Pielke Jr., “Shuttle Program Lifetime Cost,” Nature, www.nature.com/nature/journal/v472/n7341/full/472038d.html, 6 April 2011.
12. NASA, Kennedy Space Center FAQs, www.nasa.gov/centers/kennedy/about/information/shuttle_faq.html#10, 2013. Leonard David, “Total Tally of Shuttle Fleet Costs Exceeds Initial Estimates,” SPACE, www.space.com/791-total-tally-shuttle-fleet-costs-exceed-initial-estimates.html. Naval Technology, “Flyvefisken Class (SF 300), Denmark,” www.naval-technology.com/projects/fly, 2011.
13. Micah Zenko, “100% Right 0% of the Time: Why the U.S. Military Can’t Predict the Next War,” Foreign Policy, www.foreignpolicy.com/articles/2012/10/16/why_the_military_cant_predict_the_next_war, 16 October 2012.