8
Conclusions and Recommendations

In this chapter the committee combines its findings into conclusions and offers recommendations. First, it collects the factual findings presented in Chapters 2-7 into three overarching conclusions concerning the importance of networks and the current state of knowledge about them. Next, it articulates specific conclusions that are directly responsive to Items 1 through 3 of the statement of task. Finally, in response to Item 4, the committee provides its recommendations, including for research initiatives. Box 8-1 summarizes how the report responds to the statement of task.

OVERARCHING CONCLUSIONS

Conclusion 1. Networks are pervasive in all aspects of life: biological, physical, and social. They are indispensable to the workings of a global economy and to the defense of the United States against both conventional military threats and the threat of terrorism.

Conclusion 1 was developed in Chapters 2 and 3 and summarized in Tables 2-1, 2-2, and 3-1 and the discussions surrounding them. It sets the stage for the committee’s inquiry into the state of knowledge about these networks.

Conclusion 2. Fundamental knowledge about the prediction of the properties of complex networks is primitive.

Given the pervasiveness and vital importance of networks, one might assume that a lot is known about them. As documented in Chapters 5 and 6, however, this is not the case. Although the technology for constructing and operating engineered physical networks is sophisticated, critical questions about their robustness, stability, scaling, and performance cannot be answered with confidence without extensive simulation and testing. For large global networks, even simulations are often inadequate. The design and operation of network components (such things as computers, routers, or radios) are based on fundamental knowledge gleaned from physics, chemistry, and materials science. However, there is no comparable fundamental knowledge that allows the a priori prediction of the properties of complex assemblies of these components into networks. Indeed, such networks are expected to exhibit emergent behaviors—that is, behaviors that cannot be predicted or anticipated from the known behaviors of their components. In the case of social and biological networks, even the properties of the components are poorly known. A huge gap exists between the demand for knowledge about the networks on which our lives depend and the availability of that knowledge.

The committee learned that developing predictive models of the behavior of large, complex networks is difficult. There are relatively few rigorous results to describe the scaling of their behaviors with increasing size. Surprisingly, this is true for common engineered networks like the Internet as well as for social and biological networks.

Simulation rather than analysis is the research tool of choice. In the case of social networks, even simulation is vastly complicated by the diversity and complexity of the agents that are the nodes of the networks—humans or groups of humans “in the wild.” Which of their many properties are relevant for developing mathematical models of a particular phenomenon? Existing models of social networks, moreover, represent highly simplified situations and not necessarily ones that are relevant to the Army or network-centric warfare.

Finally, the notion of using network models in biology is relatively new. Controversy swirls around their utility, indeed around that of systems biology itself. In spite of a burgeoning literature on the structure of simple networks, the advancement of the field to allow relating basic scientific results to applications of societal and military interest still lies mostly in the future.

Conclusion 3. Current funding policies and priorities are unlikely to provide adequate fundamental knowledge about large complex networks.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 46
Network Science 8 Conclusions and Recommendations In this chapter the committee combines its findings into conclusions and offers recommendations. First, it collects the factual findings presented in Chapters 2-7 into three overarching conclusions concerning the importance of networks and the current state of knowledge about them. Next, it articulates specific conclusions that are directly responsive to Items 1 through 3 of the statement of task. Finally, in response to Item 4, the committee provides its recommendations, including for research initiatives. Box 8-1 summarizes how the report responds to the statement of task. OVERARCHING CONCLUSIONS Conclusion 1. Networks are pervasive in all aspects of life: biological, physical, and social. They are indispensable to the workings of a global economy and to the defense of the United States against both conventional military threats and the threat of terrorism. Conclusion 1 was developed in Chapters 2 and 3 and summarized in Tables 2-1, 2-2, and 3-1 and the discussions surrounding them. It sets the stage for the committee’s inquiry into the state of knowledge about these networks. Conclusion 2. Fundamental knowledge about the prediction of the properties of complex networks is primitive. Given the pervasiveness and vital importance of networks, one might assume that a lot is known about them. As documented in Chapters 5 and 6, however, this is not the case. Although the technology for constructing and operating engineered physical networks is sophisticated, critical questions about their robustness, stability, scaling, and performance cannot be answered with confidence without extensive simulation and testing. For large global networks, even simulations are often inadequate. The design and operation of network components (such things as computers, routers, or radios) are based on fundamental knowledge gleaned from physics, chemistry, and materials science. However, there is no comparable fundamental knowledge that allows the a priori prediction of the properties of complex assemblies of these components into networks. Indeed, such networks are expected to exhibit emergent behaviors—that is, behaviors that cannot be predicted or anticipated from the known behaviors of their components. In the case of social and biological networks, even the properties of the components are poorly known. A huge gap exists between the demand for knowledge about the networks on which our lives depend and the availability of that knowledge. The committee learned that developing predictive models of the behavior of large, complex networks is difficult. There are relatively few rigorous results to describe the scaling of their behaviors with increasing size. Surprisingly, this is true for common engineered networks like the Internet as well as for social and biological networks. Simulation rather than analysis is the research tool of choice. In the case of social networks, even simulation is vastly complicated by the diversity and complexity of the agents that are the nodes of the networks—humans or groups of humans “in the wild.” Which of their many properties are relevant for developing mathematical models of a particular phenomenon? Existing models of social networks, moreover, represent highly simplified situations and not necessarily ones that are relevant to the Army or network-centric warfare. Finally, the notion of using network models in biology is relatively new. Controversy swirls around their utility, indeed around that of systems biology itself. In spite of a burgeoning literature on the structure of simple networks, the advancement of the field to allow relating basic scientific results to applications of societal and military interest still lies mostly in the future. Conclusion 3. Current funding policies and priorities are unlikely to provide adequate fundamental knowledge about large complex networks.

OCR for page 46
Network Science BOX 8-1 Summary of Responses to the Statement of Task The Assistant Secretary of the Army (Acquisition, Logistics, and Technology) has requested the National Research Council (NRC) Board on Army Science and Technology (BAST) conduct a study to define the field of Network Science. The NRC will: Determine whether initiation of a new field of investigation called Network Science would be appropriate to advance knowledge of complex systems and processes that exhibit network behaviors. If yes, how should it be defined? A working definition of network science is the study of network representations of physical, biological, and social phenomena leading to predictive models of these phenomena. Initiation of a field of network science would be appropriate to provide a body of rigorous results that would improve the predictability of the engineering design of complex networks and also speed up basic research in a variety of applications areas (Chapter 4). Identify the fields that should comprise Network Science. What are the key research challenges necessary to enable progress in Network Science? General consensus exists among practitioners of network research in diverse application areas on topics that constitute network science (Chapter 5). There are seven major research challenges (Chapter 6). Identify specific research issues and the theoretical, experimental, and practical challenges to advance the field of Network Science. Consider such things as facilities and equipment that might be needed. Determine investment priority, time frame for realization, and degree of commercial interest. Current military concepts of “net-centricity” are based on applications of computer and information technology that are far removed from likely results of basic research in network science. Table 8-1 lists current areas of network research of interest to the Army, including priority, time frames, and commercial interest (Chapter 3). Current funding policies and priorities are unlikely to provide adequate fundamental knowledge about large complex networks that will advance network-centric operations. Besides the information domain, there are social, cognitive, and physical technology domains in the current conceptual framework for network-centric operations; there is no “biological” domain (Chapters 2–4). A basis for network science is perceived in different ways by the communities concerned with engineered, biological, and social networks at all levels of complexity. Basic research efforts are totally incoherent (Chapters 5 and 6). Options for obtaining value from investments in network science include scenarios ranging from building a base of basic research, to leveraging business practices for market-driven R&D in specific areas of network applications, to creating a robust capability for network-centric operations (Chapter 7). Given limited resources (and likely investments of others), recommend those relevant research areas that the Army should invest in to enable progress toward achieving Network-Centric Warfare capabilities. Recommendations 1, 1a through 1d, 2, and 3 provide the Army with an actionable menu of alternatives that span the opportunities accessible to it. By selecting and implementing appropriate items from this menu, the Army can develop a robust network science to enable the desired progress (Chapter 8). NOTE: The statement of task is in lightface; the summary of responses is in boldface. Fundamental knowledge is created and stockpiled in disciplinary environments, mostly at universities, and then used as required by (vertically integrated) industries to provide the products and services required by customers, including the military. This fundamental knowledge is different in kind from empirical knowledge gleaned during the development of technology and products. You get what you measure. Suppliers of fundamental knowledge measure publications, presentations, students supervised, awards received, and other metrics associated with individual investigators. The knowledge accumulates along traditional disciplinary lines because this is where the rewards are found. Large team activities are

OCR for page 46
Network Science relatively rare (except in medicine and large-scale physics experiments) and are mostly left to the consumers of the fundamental knowledge, who must supplement the fundamental knowledge generously with empirical knowledge to convert it into the goods and services desired by the paying customer. This scheme worked marvelously for more than a half a century, when the United States dominated the world and industries were vertically integrated. With the onset of the global economy in the 1990s, however, the situation began to change dramatically, for a number of reasons. First, knowledge, investment capital, technology, and technical labor are becoming globally available commodities. Second, economic activity, including R&D, is becoming global in scale. Third, these two trends are making the networks on which we depend ever larger and more complex and their susceptibility to disruption ever greater. This traditional scheme does not work well for generating knowledge about global networks, because focused, coordinated efforts are needed. Thus, there is a huge difference between the social and financial arrangements needed to gain fundamental knowledge about large, complex networks in a global environment and the arrangements that worked so well to provide such knowledge for the design and production of smaller, less complex entities in a national environment. Any successful effort to create the knowledge necessary to secure robust, reliable scalable global networks must come to grips with this reality. Overall, the committee is led to a view of networks as pervasive in and vital to modern society, yet understood only as well as the solar system was understood in Ptolemy’s time. The military has made networks the centerpiece of its transformation effort without a methodology to design networks in the physical and information domains in a predictive way for network-centric operations (NCO). Further, according to the DOD Office of Force Transformation, research in the cognitive and social domains has yet to yield advances comparable to the technological developments in the information domain. At the same time, current efforts by academia to describe networks are fragmented and disjointed. Relatively little of the current research on networks promises to create a science of networks that will generate knowledge adequate to meet the demand. In short, there is a massive disconnect between the importance of networks in modern society and military affairs on the one hand and, on the other, the support of coherent R&D activities that would raise current network technologies and capabilities to the next level. The Army alone cannot transform this situation, but it can make a beginning. SPECIFIC CONCLUSIONS Items 1 and 2 in the statement of task inquire into the appropriateness of a field of investigation called network science and its definition, content, and the research challenges that would characterize it. Elements of a field of network science have begun to emerge in different disciplines spanning engineering, biological, and social networks. The emerging field is concerned with the development and analysis of network representations to create predictive models of observed physical, biological, and social phenomena. The remarkable diversity and pervasiveness of network ideas renders the study of network science a highly leveraged topic for both civilian and military investment. The provisional consensus around its core content clearly defines the notion of network science. By making an investment in network science, the Army could forge a single approach to a diverse collection of applications. Conclusion 4. Network science is an emerging field of investigation whose support will address important societal problems, including the Army’s pursuit of network-centric operations capabilities. Although the boundaries of network science are fuzzy, there is broad agreement on key topics that should be included within the field, the types of tools that must be developed, and the research challenges that should be investigated. These were documented in Chapters 3 and 4. Conclusion 5. There is a consensus among the practitioners of research on networks for physical, biological, social, and information applications on the topics that make up network science. Responses to its questionnaire greatly assisted the committee in determining “the key research challenges to enable progress in network science.” These responses establish that there is a fair degree of consensus on these challenges across practitioners in diverse applications areas. Conclusion 6. There are seven major research challenges the surmounting of which will enable progress in network science: Dynamics, spatial location, and information propagation in networks. Better understanding of the relationship between the architecture of a network and its function is needed. Modeling and analysis of very large networks. Tools, abstractions, and approximations are needed that allow reasoning about large-scale networks, as well as techniques for modeling networks characterized by noisy and incomplete data. Design and synthesis of networks. Techniques are needed to design or modify a network to obtain desired properties. Increasing the level of rigor and mathematical structure. Many of the respondents to the questionnaire felt that the current state of the art in network

OCR for page 46
Network Science science did not have an appropriately rigorous mathematical basis. Abstracting common concepts across fields. The disparate disciplines need common concepts defined across network science. Better experiments and measurements of network structure. Current data sets on large-scale networks tend to be sparse, and tools for investigating their structure and function are limited. Robustness and security of networks. Finally, there is a clear need to better understand and design networked systems that are both robust to variations in the components (including localized failures) and secure against hostile intent. These challenges are elaborated in terms of specific research issues and their theoretical, experimental, and practical difficulties in Chapter 7 and Appendix E within the framework of exploring various investment scenarios. The scenarios respond to Item 3 in the statement of task. Although all the military services have a vision of the future in which engineered communications networks play a fundamental role, there is no methodology for ensuring that these networks are scalable, reliable, robust, and secure. Of particular importance is the ability to design networks whose behaviors are predictable in their intended domains of applications. This also is true in the commercial sphere. Creation of such a methodology is an especially pressing task because global commercial networks can also be exploited by criminal and terrorist social networks. Conclusion 7. The high value attached to the efficient and failure-free operation of global engineered networks makes their design, scaling, and operation a national priority. RECOMMENDATIONS The statement of task requests investment recommendations from the committee. Options for these recommendations are explored in Chapter 7 and Appendix E. The committee documents in Chapters 2 and 3 that the impact of networks on society transcends their impact on military applications, although both are vital aspects of the total picture. Chapters 3 and 4 explain that the current state of knowledge about networks does not support the design and operation of complex global networks for current military, political, and economic applications. Advances in network science are essential to developing adequate knowledge for these applications. Recommendation 1. The federal government should initiate a focused program of research and development to close the gap between currently available knowledge about networks and the knowledge required to characterize and sustain the complex global networks on which the well-being of the United States has come to depend. This recommendation is buttressed by centuries of evidence that disruptive social networks (e.g., terrorists, criminals) learn to exploit evolving infrastructure networks (e.g., communications or transportation) in ways that the creators of these networks did not anticipate. The global war on terrorism, which is a main driver of military transformation, is only one recent manifestation of this general pattern. Society has the same need in other areas, such as control of criminal activities perpetrated using the global airline and information infrastructures. Addressing problems resulting from the interaction of social and engineered networks is an example of a compelling national issue that transcends the transformation of the military and that is largely untouched by current research on networks. Within this broad context, Recommendations 1a, 1b, and 1c provide the Army with three options: Recommendation 1a. The Army, in coordination with other federal agencies, should underwrite a broad network research initiative that includes substantial resources for both military and nonmilitary applications that would address military, economic, criminal, and terrorist threats. The Army can lead the country in creating a base of network knowledge that can support applications for both the Army and the country at large. Maximum impact could be obtained by a coordinated effort across a variety of federal agencies, including the DOD and the Department of Homeland Security, to create a focused national program of network research that would develop applications to support not only NCO but also countermeasures against international terrorist and criminal threats. Alternatively, if the Army is restricted to working just with the DOD, it should initiate a focused program to create an achievable vision of NCO capabilities across all the services. Recommendation 1b. If the Army wants to exploit fully applications in the information domain for military operations in a reasonable time frame and at an affordable cost, it should champion the initiation of a high-priority, focused DOD effort to create a realizable vision of the associated capabilities and to lay out a trajectory for its realization. Finally, if the Army elects to apply the insight from the committee primarily to its own operations, it can still provide leadership in network science research.

OCR for page 46
Network Science TABLE 8-1 Network Research Areas Research Area Key Objective Time Frame Commercial Interest Priority for Army Investment Modeling, simulating, testing, and prototyping very large networks Practical deployment tool sets Mid term High High Command and control of joint/combined networked forces Networked properties of connected heterogeneous systems Mid term Medium High Impact of network structure on organizational behavior Dynamics of networked organizational behavior Mid term Medium High Security and information assurance of networks Properties of networks that enhance survival Near term High High Relationship of network structure to scalability and reliability Characteristics of robust or dominant networks Mid term Medium Medium Managing network complexity Properties of networks that promote simplicity and connectivity Near term High High Improving shared situational awareness of networked elements Self-synchronization of networks Mid term Medium High Enhanced network-centric mission effectiveness Individual and organizational training designs Far term Medium Medium Advanced network-based sensor fusion Impact of control systems theory Mid term High Medium Hunter-prey relationships Algorithms and models for adversary behaviors Mid term Low High Swarming behavior Self-organizing UAV/UGV; self-healing Mid term Low Medium Metabolic and gene expression networks Soldier performance enhancement Near term Medium Medium Recommendation 1c. The Army should support an aggressive program of both basic and applied research to improve its NCO capabilities. Specific areas of research of interest to the Army are shown in Table 8-1. This table expresses the committee’s assessment of the relative priorities for these areas, the time frames in which one might reasonably expect them to be consummated as actionable technology investment options, and the degree of commercial interest in exploiting promising options. Specific research problems and sample projects are given in Appendix E. The committee notes that both trained personnel and promising research problems exist in many of these areas, so the Army should be able to create a productive program readily. By selecting from among Recommendations 1a through 1c an option that is ambitious yet achievable, the Army can lead the country in creating a base of knowledge emanating from network science that is adequate to support applications on which both the Army and the country at large depend. Regardless of which option (or options) are adopted, Army initiatives in network science should be grounded in basic research. Recommendation 1d. The initiatives recommended in 1, 1a, 1b, and 1c should include not only theoretical studies but also the experimental testing of new ideas in settings sufficiently realistic to verify or disprove their use for intended applications. Recommendations 1, 1a, 1b, and 1c span only part of the investment opportunity space—namely, those segments of the space described in Scenarios 2 and 3 in Chapter 7 and Appendix E. They will involve substantial changes in how the Army invests its R&D dollars and in how it plans and manages these investments. The Army also has the opportunity associated with Scenario 1 in Chapter 7, which involves funding a small program of basic research in network science. This investment of relatively small amounts of Army risk capital funds would create a base of knowledge and personnel from which the Army could launch an attack on practical problems that arise as it tries to provide NCO capabilities. Investments in basic (6.1) research in network science can generate significant value; however, the committee wants to be crystal clear that such investments have no immediate prospects of impacting the design, testing, evaluation, or sourcing of NCO capabilities. They would create additional knowledge that builds the core content of network science, and they would train researchers who could also be recruited by the Army for later efforts. While the knowledge generated would probably be less valuable than in the case

OCR for page 46
Network Science of Scenarios 2 and 3, the cost is less and implementation can be immediate. If the Army elects to exploit Scenario 1, the committee offers the following two further recommendations: Recommendation 2. The Army should make a modest investment of at least $10 million per year to support a diverse portfolio of basic (6.1) network research that promises high leverage for the dollars invested and is clearly different from existing investments by other federal agencies like the National Science Foundation (NSF), the Department of Energy (DOE), and the National Institutes of Health (NIH). This modest level of investment is compatible with the Army’s current R&D portfolio. There is an adequate supply of promising research topics and talented researchers to make this investment productive. Additionally, it can be implemented within the Army’s current R&D management work processes, although some enhancements along the lines noted in Chapter 7 and Appendix E would improve the return on this investment. To identify the topics in basic network science research that would bring the most value to NCO, the committee recalls that the open system architectures for computer networks consist of layers, each of which performs a special function regarded as a “service” by the layers above. It is useful to distinguish among the lower (physical and transport) layers of this architecture, the higher (applications) layers that are built on top of them to offer services to the people, and the cognitive and social networks that are built higher still, on top of the services-to-humans layers. Research on the lower layers of the network architecture is relatively mature. Improving the services offered at these levels is more of an engineering problem than one requiring basic research. The most immediate payoffs from network science are likely to result from research associated with the upper levels of the network architecture and the social networks that are built at an even higher level upon their outputs. This is where the committee thinks that Army investments are most likely to create the greatest value. An area of particular promise that has little or no current investment is the social implications of NCO for the organizational structure and command and control. Basic research could provide valuable insight into how military personnel use advanced information exchange capabilities to improve combat effectiveness. For example, one might study how troops in combat could use these capabilities to make better decisions. Additional basic research in the core content of network science might help to determine how the Army can most productively utilize the capabilities of its advanced information infrastructure. Recommendation 3. The Army should fund a basic research program to explore the interaction between information networks and the social networks that utilize them. The Army can implement Recommendations 2 and 3 within the confines of its present policies and procedures. They require neither substantial replanning nor the orchestration of joint Army/university/industry research projects. They create significant value and are actionable immediately. The committee’s Recommendations 1, 1a through 1d, 2, and 3 give the Army an actionable menu of options that span the opportunity space available. By selecting and implementing appropriate items from this menu, the Army can develop a robust network science to “enable progress toward achieving Network-Centric Warfare capabilities,” as requested in the statement of task.

OCR for page 46
Network Science This page intentionally left blank.