Jan 1, 1996
For reasons we shall elaborate, the single most important task for Secretary of Defense Cohen and his successors over the next two decades is to transform U.S. military forces for adaptiveness in new strategic and operational circumstances. In this paper we present a strategy for doing so. Our suggested strategy is quite consistent with, but more far-reaching than, the course defined by recent Department of Defense (DoD) decisions. The paper is based on our understanding of what has proven successful for other large organizations that have adapted to and harnessed the information revolution. In particular: the strategy we suggest is rooted in compelling strategic objectives; it confronts technological and operational complexity; it stimulates innovation but grounds it in experimental evidence and other forms of realism; and it promotes both tight integration and decentralization—a hallmark of success in the new era.
Our paper is unusual in covering the gamut from grand strategy to mathematical methods for analysis and experimentation. This reflects our belief that a successful transformation strategy will depend on both top-to-bottom coherence and attention to operational and technical details.
The Department of Defense now has a strategy well suited to the modern world. Terminology may change with time and administrations, but the concepts embodied in the "Shape, Respond, and Prepare Now" strategy of the Quadrennial Defense Review (QDR) are sound for the long haul. The strategy recognizes the centrality of environment shaping to avoid strategic vacuums and dangerous regional hegemons, and to facilitate the transitions of Russia and China. The strategy also recognizes the core need for U.S. military forces to be able to respond effectively in both large and small contingencies in diverse operational circumstances.
What is sometimes less well appreciated because of omnipresent here-and-now problems worldwide is the "Prepare Now" portion of the strategy, which corresponds to what we call assuring strategic adaptiveness. This is the ability over the course of a few years to change force structure and force posture as needed to deal with changes, both good and bad, in threat, coalitions and burden sharing, and other aspects of the strategic environment. To put it differently, strategic adaptiveness seeks to assure that the United States will be able to "shape" and "respond" in the future, despite inherent uncertainties. Achieving this strategic adaptiveness will not be easy and—for reasons discussed below—the Department, with further urging by the National Defense Panel, has concluded that it must carry out a transformation of U.S. forces to accomplish this. More normal changes will not be adequate.
Transforming the force will require major changes in both doctrine and organization. History tells us that such transformations must have a strong and widely understood strategic motivation to succeed. In fact, the strategic motivations for transforming U.S. forces are strong and involve both opportunity and necessity.
On the opportunity side, transformation should in particular exploit the U.S. advantage in creating and applying information technology to secure U.S. global interests. U.S. technological strengths—already manifest in superior sensors, weapons, and communications—can, when leveraging high-quality military personnel, enable the United States to project power globally and to prevail decisively in most military contingencies with forces that are smaller but even more effective than today's. Technological enablers, of course, are not enough, and much of this paper is about the need to develop, test, and implement concepts for harnessing technology effectively.
Transformation is also a necessity. Failure to "prepare now" could leave the United States militarily superior on paper but impotent in reality. Military challenges already visible and certain to worsen will very likely make current U.S. forces both vulnerable and unsuitable. The worry here is not a hypothetical peer competitor in the distant future, but already-recognizable regional states availing themselves of commercially available technology, weapon systems on the open market, and, of course, weapons of mass destruction (WMD). As signaled earlier, preparing for these threats will require substantial, not merely routine, changes. Transformation is also needed because the current U.S. force structure cannot be maintained within likely budget levels: Rather, the shift to smaller but more capable forces is necessary to avoid losing capabilities and reducing U.S. ability to shape the international environment.
The stakes are therefore high for the United States—and for those that look to it for leadership in protecting and expanding the transition toward a democratized world. The inexorable spread of dangerous technologies and weapons could undermine the ability and willingness of the United States to intervene in crises and conflicts. This could weaken deterrence and open the possibility of a new era of aggression. And this in turn could both endanger specific U.S. interests and derail the promising trend in world affairs that is so important to more general U.S. interests. Transforming the force, then, is about nothing less than ensuring the continued viability of U.S. global equities, goals, and strategy.
It would be useful to define clearly what constitutes "transformation" or what many call a "Revolution in Military Affairs" (RMA). In practice, this is difficult and many argue about whether what we undergoing is even a revolution at all, rather than merely an impressive evolution. This seems to be scholastic hair-splitting, but we do believe—based on experience in many other sectors—that exploiting information technology requires a revolution in mind-sets and behaviors, even if the revolution turns out to have been the cumulative result of many evolutionary steps along the way.
Reaching a consensus definition of the oncoming RMA may be impossible, but we see its principal features as distributed, ubiquitous, and immensely powerful information technology; precision weapons; and what some call "systems of systems." The most important changes will probably be organizational and doctrinal—especially the networking of forces to permit dispersed yet integrated operations. The current "tokens" of defense capability (e.g., numbers of classic versions of divisions, wings, and carrier battle groups; or, worse, end strength) are rapidly becoming less relevant, while globally netted command, control, and communications (C3) and globally netted reconnaissance, surveillance, tracking, and acquisition (RSTA) are becoming dominant factors in capability. These changes allow smaller, dispersed, but lethal forces to operate synchronously and quickly with excellent situational awareness. To be sure, we will also need new and different platforms and weapons over time: aircraft with much greater range, unmanned combat aerial vehicles, relatively stealthy surface ships operated with small crews, lighter and faster combat vehicles, new varieties of short-take-off-and-landing aircraft, and both stand-off and direct-attack precision weapons, among others. But the most fundamental changes will be integrative developments driven by information technology.
It is useful to distinguish between what we call Era A and Era B and to recognize that we may need RMAs for both. The time periods for these overlap (Figure 1) but we can think of Era A as from now until 2005–2010, with aspects of Era B already recognizable today (e.g., the threat of small chemical attacks) and many more becoming important by 2010 or so.
In Era A, an RMA would involve exploiting within-reach technology to address understood near- and mid-term problems. The United States should be able to greatly reduce its vulnerability to Achilles' heels in both Persian Gulf and Korean conflicts (e.g., base-access problems and current-generation WMD threats to forward bases). U.S. forces should be able to maintain and even increase their lopsided advantages. And, Era-A changes can begin a reengineering of the force to reduce personnel and related costs while maintaining and even extending functional capabilities. That is, divisions, wings, and battle groups will shrink in personnel or be displaced by different units, but U.S. abilities for forward presence and warfighting should only improve. These changes will require modernized forces but not drastically new platforms. That is, the F-22, Comanche, V-22, and the Navy's planned new cruiser are by no means dinosaurs merely because they are aircraft and ships.
Era B is a different story and an RMA for Era B might not be entirely of our liking. That is, the United States could be the victim rather than the instigator if it fails to make adequate changes. After all, in the long run, the WMD and missile problems will only worsen, both quantitatively and qualitatively. This will endanger forward-operating U.S. forces and could force further changes of both force structure and broad strategy. For example, China may be able to interfere with U.S. operations in East Asia and rogues could pose threats to the U.S. homeland in hopes of undermining American will to intervene against their misconduct. Because of the nature of Era-B threats, Era-B options should include currently "exotic" concepts such as speed-of-light defenses against large cruise-missile attacks, large submarines with land-attack capability and Marines, more long-range bombers and possible "arsenal aircraft," and much more use of space and unmanned combat aircraft. Both regional and national missile defense may be essential. Because of uncertainties, Era B is difficult to fathom and preparing for it is a matter of hedging: developing a robustly diverse set of capabilities that will allow us to adapt strategically as the era comes closer.
The two-era model suggests a two-track approach. Era-A work lends itself to fast evolution driven by well-defined programs (e.g., F-22, F-18E/F, "digitization" of ground-force units, V-22, and moves toward network-centric operations), albeit with many adaptations along the way and with smaller buys than currently planned, and to pathbreaking work on organization and doctrine. Era A's efforts must have considerable visibility because of the need to achieve organizational buy-in and funding, and because of the importance of such changes for the reengineering needed to maintain functional capabilities within a constant-dollar budget. If Era-A investments and innovations are made appropriately, they will be stepping stones into Era B.
While Era-A work must be visible and determined, and can focus on exploitation, Era-B work needs to be more concerned about possible changes that would undercut U.S. strengths and make very different operational concepts necessary. Thus, Era-B work should be motivated in significant part by "worriers" and should encourage "conceivers" to anticipate military difficulties and use emerging technologies to develop new operational concepts to deal with the possible difficulties. Era-B work should encourage competition among operational concepts and should be quite exploratory with multiple paths, multiple knowledge-building experiments (not mere "demonstrations" minimizing risk), and more toleration of "failures" than in Era-A work. The better concepts and supporting systems should be pursued without committing to acquisition, but such work must still be profitable for industry and career-enhancing for military participants. In the past, such work has sometimes been done in so-called "skunk works" protected from more normal forms of review.
Unfortunately, the tracks are in conflict. The "worriers" concerned with Era B tend to see Era A procurements as old-think systems that could squeeze out resources for more innovative changes, while those trying to exploit Era A opportunities see the more radical discussions of Era B RMA as threats to their programs. We see the need for both tracks. A decisive factor in our thinking is the conclusion that even worst-case threats of Era B will probably arise in only one or a very few regions. They will require special counters, but appropriate Era-A systems and their adaptations should be effective for most missions for a very long time. Assuming the base has been laid and the capabilities tested, scaling up the special capabilities needed for Era B should be relatively easy when and if the time comes. This adaptive strategy would be easier, less expensive, and less risky than a more radical tilt now toward the dimly envisioned Era B.
The difficulties of transformation can scarcely be exaggerated, as study of revolutions in business and military affairs demonstrates. Some of the most common keys to transformation are (Table 1) the threat of imminent bankruptcy, a recent debacle, or some other clear-and-present threat such as a fierce business competition. Other keys have been a strong, visionary, long-term leadership insistent on change, having some operational slack to permit experimentation, and having some budgetary slack to support experiments without threatening existing organizations. The Department has none of these factors working in its favor. Indeed, a disinterested observer might well bet against transformation occurring until some disaster occurs.
|Historical Keys to Transformation||DoD's Context|
|Imminent bankruptcy||Comforting budget agreement|
|Recent debacle||Military at top of game|
|Clear and present threats||No peer on horizon|
|Visionary, powerful, long-tenure leadership intent on change||Decentralized power and frequent turnover of leaders|
|Operational slack||Leaders and forces at high operational tempo|
|Budget slack||Underfunded program with problems building|
|Professional motivations||Forward-looking officer corps; technologically ambitious recruits; belief in "learning organization" culture|
There is some good news, however, as indicated in italics at the bottom of Table 1. The United States has the world's finest military establishment with a highly motivated officer corps schooled in the concept of "learning organizations." There is no shortage of good ideas, initiative, and motivation for change. The obstacles to change lie elsewhere. Another positive feature of context is the dramatic testimony to the potential value of change offered by the Gulf War and the achievements of those American companies that have exploited information technology.
There is more good news in the sense that the Department of Defense has many organizations to call upon to help stimulate innovation and change. The Office of Net Assessment took an early lead by sponsoring a series of RMA war games, which affected the thinking of many military officers. Later, Joint Vision (JV) 2010 embraced a strongly forward-looking vision. And there now exist many studies by the Defense Science Board, Naval Studies Board, Air Force Science Advisory Board, Army Science Board, think tanks such as RAND, and the National Defense Panel—albeit most of them stronger on concepts than detail. The defense community is intellectually healthy.
Nonetheless, rhetoric on the RMA and transformation remains far ahead of reality, with many viewgraphs but little depth. Some of the signs of this are
All of this suggests that establishing and implementing transformation strategy will be no mean undertaking.
The question, then, is what DoD strategy would be effective in bringing about the transformation(s) necessary for Era A and Era B. Whatever the framework for transformation strategy, it must balance conflicting considerations. On the one hand (left side of Figure 2), there is need for top-down guidance, because transformation involves painful change that organizations can scarcely be expected to undertake on their own. Also, the United States has civilian control of its military and transforming the force is so important as to demand civilian attention. Moreover, problems are encountered with the multi-Service approach to armed forces. This approach has great advantages, but the dysfunctional aspects are more obvious than in the past because exploiting information both enables and demands "jointness."
At the same time (right side of Figure 2), there are strong reasons for a more bottom-up and distributed approach that exploits the strengths of the four Services. Legally, the Services have the responsibility to recruit, train, and equip the forces. They do this proudly and well, and having too many joint, transformation-related activities could undercut their efforts, e.g., by pulling personnel away from Service-specific training or by further burdening people who already suffer from high operational tempos (Optempos). Such considerations argue for being very cautious about joint experiments and programs. Further, it is the military Services that possess the in-depth competence to examine new ideas for forces and doctrine. One could argue, then, that the DoD should "get out of the way," except for some coordination. This view is intuitively attractive to those familiar with the power of decentralization for innovation, not just to those who resist change. However, one factor seems to us unappreciated and decisive (bottom left of Figure 2). That is the great technical and operational complexity of future military operations at the joint level. This factor suggests the need for more rather than less DoD-level responsibility—but of a special kind that we shall now outline.
Over the course of the last six months we have come to recommend that DoD transformation strategy include a framework strategy as shown in Figure 3, which is largely consistent with the process emerging currently in the Department of Defense (as of August, 1998), but which goes beyond it in ways that we shall describe. We have not attempted to draw sharp comparisons because to do so would be to chase a moving target and because the significant aspects of what we suggest will be issues for many years. Broadly speaking, the process applies to both Era-A and Era-B activities, but most of the specifics we offer are most relevant to Era A. The reason for this is that work for Era B is in its early stages and will require a good deal more research and experimentation before firm assessments and decisions can be made.
Our first managerial device is to suggest that the Secretary formulate tough questions that reflect his role as a national leader concerned with the viability of U.S. global strategy and the security of U.S. global interests. These questions, along with the ability to demand credible answers, are crucial in implanting strategic motivations for all that follows. The questions should highlight his principal worries and lead to what we call operational challenges: managerial devices for focusing attention on activities that will, at a minimum, produce major evolutionary improvements. If correctly stated, the challenges will be decomposable into subchallenges, each of which might be accomplished in a variety of ways, some of them involving new technology and doctrine. There is need for oversight and integration of such activities, as well as for investment in the related scientific and technical knowledge base. The decomposition will permit serious design of a multiyear plan along with competitions among operational concepts for meeting the challenges and sub-challenges. This will require investment in research, analysis, and experiments. All of this can be followed by assessment and, as results come in, decisions. It is not enough to plan interesting experiments. Reengineering needs to begin (and has, in some quarters), although the process will be sequential and iterative with some false starts. Let us now discuss these matters sequentially in more detail.
What questions should the Secretary ask? We can be guided by appreciating that the Secretary's legacy will be capabilities and options available to future CINCs and presidents. Indeed, this is a restrained statement of something more fundamental. The very fabric of the current geostrategic environment, a dynamic world composed increasingly of democratic states and free-market economies, depends on the continued ability of the United States—working with international partners and institutions such as NATO and the United Nations—to maintain the peace. This, however, requires continued forward engagement of military forces and the ability to intervene in crises and conflicts when appropriate. Will this be feasible a decade or two hence?
To put it differently, will the United States still be able to maintain stability in regions such as the Persian Gulf or East Asia through forward presence and militarily credible brokering (as well as political and economic instruments)? Will the United States be able to operate forward and intervene in contingencies? And, as a separate matter, will the United States be able to adapt its force posture if changes occur in Korea and elsewhere that make untenable the particular type of forward presence we now have?
Against this backdrop, Figure 4 summarizes the questions we suggest for the Secretary's focus. What is remarkable about the questions is their tone. They are intentionally blunt—almost rude—because they demand more than glib assurances. Starting at the top, these questions are to be interpreted rigorously: will the United States really be able, quickly, to construct, deploy, and employ joint task forces (JTFs) in a variety of difficult military contingencies? In studies it has been common for the answer to be yes, because in studies the frictions of military operations are swept away. Consider, however, that the United States has never rolled quickly from a peacetime posture into effective warfighting. Even in the Gulf War of 1990, not only was deployment begun a week after the invasion, the deployment was much slower and more problem-ridden than studies and plans had anticipated. To be sure, Desert Shield was a great feat historically and the problems encountered do not detract from the superb performance of the U.S. military, but they should give us pause. It is perhaps relevant to observe that current studies often assume Air Force deployment rates several times faster than in the Gulf War. Similar comments could be made about other capabilities.
Perhaps most clear to us is the irrationality of assuming, as do most studies, that the elements of future JTFs will deploy rapidly with little warning and come together seamlessly into exquisitely effective operations. Most of the questions, then, reinforce the need to ask whether selected key capabilities will be real and robust, even against adversaries that use WMD and adopt "smart" strategies, and even against large regional powers.
Questions 9 and 10 are different in character. The ninth will become increasingly germane as the outyears of the defense plan come closer and the dollars to support the current force—as currently configured—remain elusive. Even if they could be found by raising the defense budget as some suggest, question 10 would apply. It is true, after all, that a single jet fighter can pay for a great many schools.
Given the rude questions, can we now construct a set of challenges to be used in the Defense Planning Guidance (DPG), the Joint Strategic Planning Guidance (JSPD), and other documents as focusing devices? A complete list of problems would be long, but in what follows we focus on a small but representative subset.
The challenges should have a number of features. They should, of course, respond to the questions. In addition, they should be posed at the operational level because that is the nexus between national-level objectives and concerns on the one hand and military art on the other. It does not intrude into the realm of solutions or tactics, but it likewise avoids excessive abstraction.
Although some disagree, we believe that even the first group of operational challenges should honor the diversity of missions and environments. There should be no tilt toward one or another simple-minded view of priorities such as one that highlights only being ready to fight Saddam Hussein in the desert again, or one that highlights only difficult manpower-intensive operations in rough terrain such as urban sprawl.
The challenges should also relate clearly to the ideas of Joint Vision 2010 and the building-block capabilities discussed by the Joint Warfighting Capability Assessment (JWCA) process, but they should be at a higher level and should be useful in constructing programs, joint experiments, and issues for decision. Ultimately, the challenges should be useful forcing functions substituting for traditional impetuses such as losing a war, and transmitting the urgency and direction of the Secretary's questions to the operational and program levels.
We should note here that this paper is concerned with future fielded forces, not supporting infrastructure or total-force organization. Further, our operational challenges are intended to be focusing devices for special changes, not a comprehensive set of goals for overall defense planning, which includes a great deal of "routine" modernization and change.
Figure 5 shows our recommended set of initial operational challenges, the motivation for which may be obvious. The first challenge plays to long-range precision fires and establishes a stressful joint mission (an early halt) that cannot be accomplished by merely waiting until an advancing enemy army eventually gets to where U.S. ground forces can engage after a week or two. In contrast, the second challenge recognizes that some early halts will not readily be accomplished by long-range precision fires and will require a very different mix of forces, including ground forces. In this context one might think today of Korea, but there are many possible future wars in which the preferred first move of the enemy would be sudden and decisive, and would involve not long desert marches but rather mixed terrain. This problem lends itself to solutions with capable and rapidly deployable Army and Marine forces, packages to assist allied forces in place, air forces, long-range naval gunfire, and missiles.
The third challenge puts squarely on the table the requirement to find out whether the visions of "dominant maneuver" and "information dominance" can be translated into real-world capabilities. If so, then surely a key capability would be the capacity to attack an enemy army early, even in parallel with a "halt operation," and to conduct an early counteroffensive to destroy the enemy army quickly, rather than after months of a massive buildup. Why should the United States plan to allow enemy ground forces to escape for another time and another war?
The last two challenges are oriented toward small-scale contingencies. The first reflects a dilemma of 1991: Serbia's aggression could probably have been stopped, avoiding the resulting calamity, but the political price was too high because of the number of forces that was deemed to be needed and the risk of quagmires and casualties. Will future leaders have better options? The last challenge has obvious importance and need not be elaborated.
Although our creation, these SecDef challenges connect readily with those under study in OSD and in the Joint Staff. This is not accidental. It was not our intention to generate something creatively idiosyncratic, but rather to suggest integrative challenges that would link to familiar concepts but be a pragmatic basis for action.
Each of the challenges is just the outer shell of a multi-layer structure as indicated in Figures 6 and 7. To say that the United States should be able to bring about an early halt is straightforward, but such a halt requires many building-block capabilities as shown (an incomplete depiction). Connecting back to the rude questions, we see on the left, for example, the need for quick and effective command-control and theater missile defense.
Figure 7 shows a similar abbreviated decomposition into building blocks for the early counteroffensive. Some of the building blocks (e.g., suppress and paralyze. . .) are quite different from those for head-on-head attrition warfare. In this figure we also indicate with dashed lines links to crosscutting functions such as those in JV 2010—notably, C3 and RSTA, precision fires, and so on. These are support functions rather than building block operations such as "locate and monitor enemy and friendly forces."
Decompositions such as those in Figures 6 and 7 can and should be used managerially. They are in many respects similar to well known "strategies-to-tasks" breakdowns, which have been used effectively by some of the Services. In our context, however, subordinate operations (and even some of their subordinates) turn out still to be joint operations rather than single-Service tasks. Also, to a greater extent than previously, the operational-level building-block operations depend on a myriad of crosscutting support functions as suggested in Figure 7.
Later in the paper we shall return to the use of decomposition into building blocks—not just for management, but also as the basis of analytical architecture affecting research, analysis, and experimentation. First, however, let us discuss some fundamental managerial issues.
The usual issues raised about transformation are "How do we go from JV 2010 to something more concrete?" and "How do we assess whether the joint experiment program makes sense and is being well managed?" Those asking these questions tend implicitly to accept the JV 2010 vision and to assume that developing related real-world capabilities is mostly a matter of "coordinating" among Services and tidying things up in the experiment program.
A different set of issues stems from the observation that the envisioned future operations are new and exceedingly complex—in both technological and operational terms. The very concept of dispersed but integrated forces and sensors, networked by distributed processing and operating at high speed, suggests unprecedented complexity. The operations involve, for example, highly parallel and decisive operations with relatively small forces and accurate fires, rather than the deliberate concentration of forces practiced in the 20th century. There may be no secure areas anywhere near the battlefield, and no clear-cut lines demarcating friendly and enemy zones. Success may depend less on head-on-head attrition than on outmaneuvering and collapsing the enemy. These operations are unlike what our military leaders have experienced and what is well described in the texts they used in command-and-staff colleges.
If we are correct that transformation will be profoundly difficult and complex, then the Department (and Congress) needs to change the way it "sees" the transformation problem, as suggested in Table 2. Arguably, at least, the current approach sees joint experiments as add-ons or overlays designed to add value without "interfering" with Service prerogatives. Joint experiments come and go, often as demonstrations, without establishing a legacy of rigorous knowledge. No one is particularly responsible for results and the assumption is that the Services can and will supply what is needed. And, finally, there is no pressure for decisions actually to change forces or doctrine.
|Joint activities as "add-ons"||Future warfare will be inherently joint and complex|
|Emphasis on "coordination without interference"||Need robust, highly effective integration|
|Demos without much rigor or foundation laying||Need rigorous development and testing to understand phenomena and minimize risk|
|No one responsible for results||Need clarification of both responsibility and authority|
|Services as only "suppliers" of future capability||Need joint-level suppliers and stronger Joint Staff role|
|Process-oriented road maps with no pressure for near-term decisions on forces||Need change processes with consistent stream of actions to create momentum and foundation|
As indicated on the right of Table 2, we see the need for a highly capable, sustained, technical effort with rigorous integration and testing and with a building momentum for change driven by a series of decisions over time (not just experiments that don't "bother" anyone).
To better appreciate this contrast, consider some past transformational accomplishments:
This list of glorious accomplishments involving complex systems perhaps tells it all. In each case the challenges were seen as fundamentally technical and complex, not as something to be accomplished by mere coordination.
The military Services understand analogous issues within their own domains and Service Chiefs are actively interested. For example, the Marines—with close attention from the Commandant—have been conducting experiments involving ambitiously new doctrinal concepts. They stress in lessons-learned briefings how they have been repeatedly humbled and how some of the experiments have "failed" (as would be expected in serious work). The experiments prove useless unless enough time is set aside for participants to learn and assimilate the doctrinal concept being tested. The Marines also have considerable technical support as they design and conduct their tests, which are seen not as "demonstrations," but rather as research experiments to gain knowledge essential to decisions that will determine the very future of the Marine Corps. So also the Army, Navy, and Air Force are deadly serious in their experimental efforts on core competencies.
The problem, ultimately, is that future operations are envisioned as quintessentially joint, but there exists no deep body of technical and organizational joint competence comparable to that enjoyed by the Services. CINCs have deep current-operational competence, but are ill equipped for research, development, and testing. Changing this situation, at least with respect to certain experiments, will soon be the responsibility of the U.S. Atlantic Command, but that will address only part of the problem.
Another related observation is that our goal should be to plan for future strategic and operational adaptiveness. As a result, we are less concerned with tidiness than with the richness of the experience base laid by experiments and other studies.
Some basic issues, then, are three:
The answer to the first issue is that the knowledge base of military science has been neglected for many years: We have been living off intellectual and empirical investments made decades ago. We shall not elaborate here, but rather point the reader toward a study on the subject. One of us (Davis) was the principal author, but the results reflect the collective judgment of a broadly experienced panel of experts in modeling and simulation (M&S). The panel emerged convinced that the problem with military M&S is not software technology but the lack of knowledge of what to embed in that software (and the need for new modeling methods to represent key issues such as uncertainty). The bottom line here is that the Department needs to develop an integrated joint investment strategy in warfare-area research, presumably with leadership and funding from OSD, the Defense Advanced Research Projects Agency (DARPA), and the Joint Staff for matters going beyond what the Services are already doing. Research is also needed on advanced methods for modeling and simulation.
The second issue is who will be the architect of the larger effort to decompose challenges, establish study efforts, and integrate results. Here we have in mind that architecture involves design, not just administration. An important question arises here: If Services are the "suppliers" of task-level capability—specialist organizations that know how to develop and choose among alternative concepts for accomplishing tasks—then who are the suppliers of those joint-level capabilities that are not simple concatenations of Service-component tasks? Who is going to pose the operational challenges and subchallenges, assure a competition among concepts, assess the results, and make decisions that will affect the tool kits available to future CINCs and JTF commanders? Who is going to guarantee that each of the requisite building-block capabilities will be available when needed and will be amenable to efficient JTF operations?
Our answer is that the Joint Staff (and OSD) must play a larger role than heretofore, and at a high level such as that of the Vice-Chairman. At the same time, the Joint Staff will need to delegate the detailed responsibilities for such efforts—probably on an operation-by-operation basis once decompositions as illustrated in Figures 6 and 7 exist. CINC USACOM might represent joint concerns in one case (in addition to being responsible for joint experiments), an office of the Joint Staff might represent them in another, and in still other cases one of the Services would take the lead. In any of these cases, the lead organization could be buttressed technically by federally funded research and development centers (FFRDCs) and university or national laboratories, as well as commercial contractors. In any case, the responsible authority would be concerned not just with coordinating Service activities and creating timelines but with architecture, research, analysis, and experimentation.
To reiterate, we see need for a strong role by someone with the rank and locational clout of the Vice-Chairman, as well as a strong role for CINC USACOM. We do not wish to convey the impression of centralized micro design, however, since we expect that most innovation and problem solving will be distributed, and even bottom-up, in character. Otherwise, innovation will likely be stifled and results unimpressive.
The third issue relates to when and how leaders should force the transition from experiments into controversial and painful changes of force structure. We believe that transformation strategy needs explicit pathways from experimentation to implementation, including measures that will endanger current stakeholders' interests.
DoD's emerging strategy includes delegation of some responsibilities below the Joint Staff, particularly to CINC USACOM. Several issues arise here. First, CINC USACOM has major duties concerned with here-and-now force preparation. It follows that one natural course of events might be to have the Joint Warfighting Center (JWC) take leadership. Expanding the role of the JWC is quite sensible, but to assure a pathway to fielding of new forces, additional high-level individuals need to be involved—e.g., ex-commanders with access and influence. They are needed both to provide guidance and to serve as advocates of change. Organizations focused on modeling and simulation may or may not reflect the mind-sets of wartime commanders and often lack the ear of top decisionmakers. Fortunately, as matters now stand (August, 1998), CINC USACOM plans to take a strong personal role.
The second issue involves assuring continuity, operational expertise, and especially command-level integration. We see a standing JTF command group for experimentation as very important. This would involve scores, not thousands, of people and would therefore not interfere unduly with ongoing operations and Service imperatives—except that top-quality officers should be involved and protected. Occasional experiments with thousands of participants would be quite exceptional. The command group would not be a standing command force, and would support experiments with a variety of command relationships.
The third issue is the need for continuity of in-depth top-notch analytical infrastructure. The need for this should now be evident from our discussion.
Overall, then, we recommend a structure within ACOM that has a standing Joint Task Force command group for experimentation, a senior advisory panel of well-connected retired commanders, and an analytical body with top-flight support from industry. An analogy might be the support provided to the Air Force by TRW and Aerospace during the development of missile and space technology. That support continues to this day. Another would be work during the 1970s for the Joint Chiefs of Staff by the Weapons Systems Evaluation Group (WSEG) and the Institute for Defense Analyses.
We also suggest that the Department plan on moving some leading-edge new-concept capabilities and units into warfighting commands early rather than later. Technology-intensive near-term capabilities could dramatically supplement U.S. Central Command's (USCENTCOM's) ability to deal with certain Achilles' heels/asymmetric strategies such as short-warning attacks. CINC USCENTCOM could build them into his war plan and test their "reality,"—and do so while retaining current capabilities. This might translate into an ability to halt an invasion early, in Kuwait or northern Saudi Arabia. If the new capabilities came up short, he would still have the ability for defense in depth. Our belief is that nothing short of moving new capabilities into the high-visibility warfighting regime will suffice to increase transformation momentum.
Earlier, we mentioned Era-B RMA, but much of our discussion has implicitly focused on Era-A work. Here let us reiterate that Era-B activities are different in kind. It is inherently difficult for organizations to spawn radically different concepts or products. History tells us that this is best accomplished by creating small protected sub-organizations or spinoffs with charters for daring exploration. To some extent the Services do this now in their Battle Labs, which suggests the need for a Joint Battle Lab, although such an organization might well be more "virtual" than physical—linking Service activities through distributed modeling and simulation, including occasional joint experiments. There is considerable interest in such activities among Service innovators and in the Department, but they have not yet been funded.
Let us now turn from matters of strategy, organization, and management to matters more technical and analytical.
Although much current discussion of how to pursue the images of Joint Vision 2010 focuses on joint experiments, it is more appropriate to recognize that for each operational challenge there are building-block operations that must be studied in depth. As mentioned above, this will require DoD-level investments—above and beyond normal Service investments—in physical, behavioral, and mathematical research, analysis, and experimentation, including critical joint experiments. The tail, however, should not wag the dog. The joint experiments should be seen as a crucial element in something larger. Otherwise, they may be little more than one-time demonstrations. With this in mind, let us now describe what we see as core principles for these endeavors—whether they are thought of as developing building-block operational capabilities, the "systems of systems" that make them possible, or both.
One core principle for building an information-age systems of systems is designing with rigorous hierarchical and network-centric decomposition. Successful complex systems work because they have the right "building blocks," which can be linked adaptively to operate effectively together. Most building blocks are themselves comprised of building blocks. Systems comprising such building blocks are typically less efficient for a specific job than is a custom-tailored system, but far more effective overall because of their adaptiveness. So it is, for example, that computer systems are increasingly designed for plug-and-play. Further, they are designed to exploit networks—to the extent that it is now routine to work simultaneously with netted programs and databases resident on computers separated by thousands of miles.
In such systems experimentation and testing are done at all levels. The overwhelming majority of experiments and tests should be at the building-block or module level, as in large-scale software development. Further, how the modular functions are achieved should be largely decided bottom-up with all the advantages that entails. However, it is essential to define and enforce—with Draconian rigor—the modules' performance requirements and interfaces with the rest of the system. To speak more plainly in the transformation context, it is not sufficient to let the Services define, develop, and test their separate building-block competencies and then to have occasional polite information-sharing meetings. Those building-block competencies and their interfaces must be defined in rigorous detail within a coherent top-to-bottom framework. Again, then, we see need for joint-level requirement setting and operational specialization.
This is not a mere "techie" matter of bytes and bandwidth. Suppose, for example, that the Army and Marines developed superb capability for forced entry and securing of installations, but neither planned for rapid no-notice joint operations using real-time network-centric information sharing and intra-operation adaptation in coordination with the Air Force and national assets. The future JTF commander, then, might lack a suitable quick-seizure option if he saw critical air or sea bases being endangered. As many officers observed in our research discussions, "worriers" with a JTF perspective often see different needs from those seen from typical Service perspectives. Thus, the building-block "requirements" must be defined at joint level, and in more rigorous detail than in normal joint doctrine.
Another core principle is that even with good design, complex systems do not simply fall together perfectly. Integration experiments are crucial and must be stressful to be useful. For this and other reasons, we believe that the central focus of joint transformation planning and experiments should be defining the building-block requirements and stressful integrative testing of C3 and RSTA. Again, we believe that a standing JTF command group dedicated to such functions (not current warfighting) would be quite powerful for this process because of the continuity of expertise it would permit.
A third core principle is that systems of systems can seldom be fully tested "live." Instead, assessments and decisions must depend on analysis using modeling and simulation. Aside from its demonstrational and motivational values, live testing has the purpose of refining and spot-testing the models, rather than "proving" what needs to be believed to make decisions! Realizing this can be sobering for those who recognize the limitations of current higher-level models and simulations.
This centrality of integration testing and dependence on models and simulation is old hat to engineers. In the engineering and software communities, the phrase often used is that of a "model-test-model approach." That is, one builds models of the system (actually a family of models), analyzes, tests, analyzes, and revises models, and so on, iteratively. When it comes time to act, it is the model that one must depend on because the testing cannot be sufficiently extensive to cover all the operational circumstances.
The models alluded to in Figure 8 need to be highly varied: there must be no notion of "one shoe fits all." There is need for quantitative models and simulations—with diverse resolutions, scopes, and perspectives—but there is also need for more qualitative structured models incorporating expert judgment. There is need for live, virtual, and constructive simulations, and for multi-resolution model families. And there is need for decision-support tools. Finally, if the family of models is appropriately integrated, it should be possible to study particular issues at the appropriate level of detail and then use the conclusions of that work for cross-calibration.
To elaborate on scope, consider the landscape of analytical methods indicated in Figure 9. We have three axes here. The x (horizontal) axis describes increasing level and breadth incorporated in a method. For example, a narrow model might deal only with weapon-on-weapon engagements, while a broad model or method might not only deal with entire campaigns, but allow users to consider other factors such as shaping and hedging. The y (vertical) axis describes decreasing resolution or the increasing scope of the uncertainties and trade-offs that the model can deal with. High resolution precludes broad treatment of uncertainty. The z axis (into the paper) describes increasing integrativeness for decision support involving economic costs and political constraints.
Figure 9 includes some examples. First, there are familiar existing models such as Janus (bottom left), which has high resolution (entity level), but narrow scope (e.g., a battalion versus regiment battle). Farther up the y axis is the familiar radar equation of textbooks. This is narrow in integrative scope (it deals only with radar detection range), but it can be used for a broad range of uncertainty and trade-off analyses because it is simple. TACWAR and JICM are theater-level (or, in the case of JICM, a multi-theater model) and are therefore more militarily integrative. JWARS and JSIMS are currently in development. They will have moderately high resolution. They may in some respects be rich in their treatment of campaign dynamics, including C3, RSTA, and perception-reality distinctions. They will likely be ill suited, however, for higher-level decision support under uncertainty. Moreover, JWARS has been seen as a single model, not a family. It may therefore prove too large and ponderous to do anything well, a common model problem. And, to make things worse, its development has not emphasized human gaming, which in our experience is critical in developing innovative and adaptive strategies that can then be represented in models. More generally, none of the models in Figure 9 have been designed together or integrated. Without special attention on this matter, DoD will not have good model families for supporting adequately transformation research.
Also striking is the vacuum that exists currently at the top, right, and rear (low resolution and highly integrative). To fill this vacuum we and RAND colleagues have been using relatively simple operational-level models (e.g., spreadsheet models used for the halt problem) and a decision support tool for resource allocation called DynaRank.
We show this information to emphasize the range of tools needed, the many gaps, and the many uncertainties that exist regarding models in development. And yet such models will be central to the eventual success and quality of transformation and transformation-related decisions, and—even more frightening—to the eventual success of future-warfare operations such as those sketched in Joint Vision 2010. Models and simulations will be thoroughly embedded in future command and control systems.
Perhaps the most important single technical point to be made, if one accepts our assertion that transformation strategy involves complex operations and supporting systems of systems, is that much work should focus on assuring robust effectiveness, not optimizing "nominal" performance. The reasons should be obvious from personal experience. For example, most of us are relatively indifferent to a 10 percent increase in the operating-system speed of our personal computers, but we are exceedingly concerned about avoiding crashes. Similarly, in command and control systems, what matters most is success (i.e., not failing), rather than the second decimal of performance in nominal cases. This is familiar theology to people experienced, for example, in the problems of cold-war strategic command and control, but less intuitive to others.
Our theme here is that technical analysis must address uncertainty and risk—and even dwell on it. Doing so will involve a level of scientific and technical sophistication that may be uncomfortable to military officers who have not previously found it necessary. Why have they not found it necessary? The reason is that U.S. military doctrine today is rooted in decades of experience. Doctrine now handles risk and uncertainty so that many aspects are well understood in nontechnical terms. "Everyone knows" that to have good odds of success a corps commander may want a 6:1 force ratio at the point of attack.
In contrast, when discussing new military operations we do not have the benefits of decades of research and experience. The issues of risk and uncertainty must be taken up afresh. Understanding and reducing risk may be difficult, but corner-cutting could mean disasters.
How does one address risk and uncertainty? The general principles here are two: work to reduce risk and uncertainty, and work to assure adaptiveness. Both of these depend heavily on leveraging information, some of it at-the-time information for commanders.
Risk and uncertainty are not just for operational commanders to worry about, however. They should be central themes in the planning, programming, and budgeting system (PPBS). Consider, for example, the QDR decision to reduce the size of the JSTARS (Joint Surveillance and Targeting System) fleet. The JSTARS aircraft is undeniably expensive and there were reasons for the decision, but the decision was reportedly influenced by rules of thumb that counted up the number of JSTARS nominally needed without adequately considering the need in real wars and other sustained operations for substantial redundancy, much less the need for greater-than-nominally-needed numbers in a two-crisis situation with high stakes. Our analysis suggests that being short some JSTARS would be a good deal worse than being short a division or wing. When critics complain that the PPBS system still reflects old mind-sets, the JSTARS decision is an example. So also are examples that use standard and unstressful scenarios to test the need for at least some advanced stealthy fighters and bombers and at least some expensive long-range standoff munitions. Such analysis discounts risk and uncertainty.
To illustrate the kind of analysis we believe might become routine, and to demonstrate briefly some new methods we have been developing, let us imagine the problems faced by a future Joint Task Force commander who contemplates using small, lightly armed ground forces relatively far forward in a mixed-terrain theater while attempting to halt the invading army with long-range precision fires. The ground forces would be scouts and sensors, would call in fires, and would perhaps attack rear-area targets. Versions of this concept have been proposed by the Defense Science Board, Marines, Army, and RAND among others. If force planners are indeed to provide future commanders with needed capabilities, we believe they need to analyze problems with those commanders' operational needs in mind. Focusing on nominal scenarios and standard measures is not good enough.
Our imaginary JTF commander knows the number of attacking divisions and the numbers of his long-range and tactical fighter aircraft. He also has missile capability in the form of Army and Navy TACMs. And air forces are reinforcing at a predictable rate per day. Despite this knowledge, the JTF commander is worried about uncertainties. His effort is not a mere study, but a real-world operation with live human beings and national needs. He asks tough questions, perhaps like these:
Commander: "How long will it take to do SEAD?"
Staff: "Two days, sir."
Commander: "How sure are you?"
Staff: "Well . . . we don't know what we don't know, but if things go right, two days!"
Commander: "What if they don't?"
Staff: "Well, the worst case would be about eight days. If they have obtained the SAM-XXX, then, yes, it would take about that long. We haven't actually seen the SAM-XXX here yet, but they have been seeking it and they could have it and be hiding it."
How should we go about dealing with the kinds of uncertainty highlighted in the dialogue? Our answer involves "exploratory analysis," which bears some explanation, after which we will return to our commander example. The first point is that it is not sufficient to do standard sensitivity testing because, in typical complex operations, uncertainties are large, base cases are not meaningful best estimates, and effects are nonlinear. We need to look at combinations of uncertainties—indeed to explore the relevant "scenario space" (Figure 10). That is, a given name-level scenario such as Iraq invading the Saudi peninsula corresponds to an infinite number of more fully defined analytical scenarios that differ, e.g., in actionable warning time, alliances, forces, and so on, including assumptions about force effectiveness. There are usually large uncertainties in all the dimensions shown in Figure 10, and the concept of a best-estimate case is quite misleading.
Exploratory assessment across scenario space is a key to the capabilities-analysis approach to defense planning that we have long stressed. However, there are two quite different ways of accomplishing exploratory analysis. We and RAND colleagues (notably Steve Bankes) have presented the first in earlier research, which we now call "parametric exploratory analysis." Figure 11 illustrates the form of results, although a full survey takes a number of pages. The measure of outcome here is the distance penetrated by the attacker before he is halted. Light is "good"; dark is "bad." The numbers in the boxes indicate distance in kilometers. We have discussed our work in preparation for the QDR elsewhere.
Let us now focus on the second method, which is unusual in force planning. We call it "probabilistic exploratory analysis." It complements Method One nicely and is now feasible with desktop tools. In this approach one characterizes the uncertainty of key variables with "probability distributions." We use quotes here because these are primarily subjective probabilities related to uncertainty, rather than randomness of process (although random effects can also be reflected). Given such distributions, we can use Monte Carlo methods and a spreadsheet-level halt-phase model to generate outcome "probabilities" for the problem considered above. This folds together the consequences of uncertainties in many dimensions. After all, some uncertainties work to the U.S. advantage (e.g., the Iraqi army may have low morale and may break early). Others work to the U.S. disadvantage (e.g., if RSTA is not quite as effective as hoped). The probabilistic approach connects these in a kind of net assessment.
If the net effect of uncertainty is substantial and worrisome, then we need to find out what contributes most to the outcome uncertainty. Thus we may revert to parametric exploration. In any case, upon learning the source of problems, we would then focus on ways to reduce risk.
Returning now to the example of our JTF commander, we can imagine that instead of accepting standard planning factors such as SEAD (suppression of enemy defense) shall take precisely two days, or the enemy column will move at precisely 60 kilometers/day, the JTF commander works with his subordinate commanders and staff to characterize uncertainty mathematically.
Figure 12 shows schematically some input assumed for an illustrative Monte Carlo analysis to support our hypothetical commander—all performed quickly on a personal computer. The SEAD time distribution (top left) represents a bimodal assumption as in our earlier dialog. If the enemy has obtained and is hiding SAM-XXX batteries, then SEAD will take much longer than expected.
The distribution for post-SEAD air-to-ground effectiveness of aircraft (top right) has a long tail to the right. The reasoning here is that air-to-ground effectiveness could be extremely high if the attacker has trouble dispersing his vehicles or completing road marches quickly, but there are many ways for effectiveness to be much lower. For example, real-world C3, RSTA, and fusion might be less effective than it "should be" because of traffic-handling saturation, technical problems, terrain, and inexperience.
The third case (bottom left) represents the number of armored vehicles that have to be killed to bring about a halt. If morale or dedication is low, a halt might come about much sooner than predicted by a conservative planning factor such as 50 percent. Finally, on the bottom right, the distribution for attacker's movement speed is described as triangular. For our purposes, these assumptions are merely illustrative. What matters is that one can characterize the uncertainties. Indeed, senior commanders (and peacetime decisionmakers) routinely deal with issues far more complex than characterizing roughly the nature of uncertainty of key factors. Technology now makes it possible to reflect those judgments readily in simulation.
Figure 13, then, is the outcome distribution generated by hundreds of Monte Carlo simulations. This is the "probability distribution" for the distance penetrated by the attacker before being halted. For the illustrative assumptions we used, a wide range of outcomes is possible and the result is not at all like a normal, or Gaussian, distribution. Indeed, there is a bad tail on the right side indicating that in a sizable number of cases the attacker reached his objective and stopped (588 kilometers is the distance from Iraq to Dhahran). If our mythical commander were contemplating putting small ground forces far forward in northern Saudi Arabia (e.g., somewhere in the range of 200–400 km from Iraq), he might be quite sobered by the analysis. Even putting them at the northern end of the Iraqi coastal approach (e.g., 480–500 km) might end in a debacle for the troops involved.
Also of interest here is that the mean value of probabilistic analysis may be quite different from what one would obtain using a deterministic simulation such as TACWAR or JICM with ordinary planning factors. As analysts all know but often forget, the mean (or so-called expected value) of a stochastic simulation is not equal to the deterministic value obtained by replacing the stochastic variables by "nominal" values, which typically are a bizarre mix of best estimates, optimistic estimates, and worst-case estimates.
We can use the same spreadsheet tools to generate sensitivities (Figure 14) of the sort the commander would ask about. These are not one-at-a-time sensitivity tests as in normal analysis, however, but rather measures of how each variable's uncertainty affects overall outcome uncertainty viewed across the range of values of the other variables. Here we see that air-to-ground and missile effectiveness dominate, but SEAD time and the size of the force that must be destroyed (a function of enemy morale and competence) are also important.
Some of the most sensitive variables will often depend on C3 and RSTA. Further, the appropriate responses will be to reduce uncertainty in factors that the United States controls by, e.g., avoiding common-mode failures, assuring redundancy of capability, and building in the traffic-handling ability to service large numbers of shooters and targets. Network-centric warfare can be powerful here, so long as the network has no critical node.
Finally, let us observe that to do this kind of analysis one needs to work with relatively simple models. Detailed models are essential for in-depth understanding of phenomena—and even for identifying the right aggregate variables, but understanding consequences of uncertainty typically requires low-resolution models such as those we have been developing. They should be explicitly related to more-detailed models by multiresolution modeling.
This concludes our survey of a strategy for transformation, one that began with issues of grand strategy and ended with mathematical methods for assessing risk and uncertainty.
Perhaps the most important recommendation involves a change of perspective. Because transformation involves creating the capability for large complex operations, supported by large and complex systems of systems, success will require far more than good top-level documents, some joint doctrine, and the loose coordination of Service efforts.
A number of concrete steps would get the transformation process off on the right foot. Some of these are under way, based on recent important DoD decisions such as the decision to have CINC US Atlantic Command manage joint warfare experiments. Others are yet to emerge. The steps we emphasize are
Let us end by observing that most of the needed transformation can occur with business-as-usual activities by the Services, many of them bottom-up in character. The important proviso is that there be a coherent overall framework, that the necessary information system be developed, and—most difficult—that the crucial integrative issues be dealt with effectively, as we believe will be possible with the measures we have suggested. This will require a stronger joint role than heretofore.
This paper derives from research sponsored by the advisory board of RAND's National Defense Research Institute, a federally funded research and development center (FFRDC) sponsored by the Office of the Secretary of Defense, the Joint Staff, the defense agencies, and the unified commands. Our work benefited from numerous discussions with senior officials and military officers working on transformation issues. We also appreciate reviews by colleagues Glenn Kent and Michael Hix.