Executive Summary This study addresses how efficiently our collectors work together ("synergy"), the budgetary balance between collection and "downstream" activities, and ways to reduce collection costs, primarily in the satellite area. Regarding collection synergy, the study concludes that we are only beginning to look at how different forms of technical, human and open collection could be developed, budgeted and operated to work together cohesively and efficiently. If we proceed as now planned, progress will be very slow. Recommendations, therefore, include opting for a "revolutionary" rather than evolutionary approach. We should develop technical work-arounds for existing systems, and through an independent body establish as soon as possible the common standards and protocols to provide for intra- and cross-INT interoperability, based as much as possible on commercial standards. There should be much greater attention to cross-cueing our collection through integrated collection management using improved, common data bases. We must also better manage the balance between crisis and longer-term target priorities. Since the fall of the Berlin Wall and despite the exploitation and dissemination problems revealed during the Gulf War, collection, especially satellite-based collection, is taking an increasing share of the budget. We should be shifting more money into processing, exploitation/analysis and dissemination. This is possible without sacrificing collection capability and even as we make greater efforts to overcome denial and deception, because technology and streamlining offer the potential for large cost savings. Numerous areas, other than synergy, where we could reduce collection costs are listed, and study of the feasibility of a "market" approach to collection budgeting is suggested. COLLECTION SYNERGY Scope This paper is weighted toward satellite collection issues, although it addresses the interaction between satellite, aircraft and other collectors. Issue Summary There is no doubt that U.S. intelligence collection capability far surpasses that of any other country, particularly with respect to technical collection, and that this capability has been the envy of both allies and enemies. Questions regarding collection have focused on whether we could sustain and improve collection capability at greater efficiency and lesser cost, and whether existing trends should be maintained or altered in order to preserve the US collection advantage for the future. The following have been identified as problem areas relating to collection, and will be discussed further in subsequent sections of this paper: 1) Collection management lacks the accessibility, flexibility and dynamism necessary for the post-Cold War period. At present there is an imbalance in collection management priorities favoring near-term crises at the expense of baseline capabilities and future needs. The erosion of regional data bases is expected to accelerate as limited assets are focused mainly on a relatively few top Presidential Decision Directive - 35 (PDD-35) priorities. 2) Collectors work independently and thus at suboptimal efficiency, in separate "stovepipes." 3) There appears to be an imbalance between collection and "downstream" capabilities, especially in projections of the future; regardless, it appears that significant savings could be made in satellite collection without sacrificing capability. 4) The Intelligence Community (IC) appears unable or unwilling to make cross-program, cross-INT budget tradeoffs. Budget priorities and cuts often are not driven by requirements/users. The division of resources between the "INTs" is largely static. 5) Proponents find greater difficulty in funding relatively inexpensive collectors/technology than in funding high-cost programs. 6) Spacecraft and associated systems are becoming ever more costly and consuming more of the intelligence budget. 7) We need more, rather than fewer, spacecraft platforms for better global coverage, more frequent revisit and reduced vulnerability. Demand outstrips capability. Denial and deception problems are increasing and the planned future architecture makes us more vulnerable to them. 8) There are very long lag times in getting technology on orbit. We need to adapt to commercial standards, technology and processes. 9) Unrealistically low spacecraft life calculations exacerbate problems of cost, fielding timely technology and maintaining the industrial base. "Synergistic" or "Fused" Collection At present, collection platforms normally are "stovepiped" to operate independently from other collectors, including completely distinct processing systems, and usually unique exploitation, dissemination and receive systems as well. While in the best cases a coherent "end to end" system is created, usually this involves considerable inefficiencies in collection tasking, and in achieving an "all source" intelligence picture that meets user requirements and that gets to the deployed military user in a timely way. Synergistic or fused collection would make more efficient use of collection assets through timely tipoff, cooperative geolocation, avoidance of duplication, assignment of the most efficient collector for a given task, and through coordinated orbits or collection plans. There seems no doubt that collection assets could work together far more efficiently had they been deliberately designed to do so. However, continual technology advances in key areas also present much greater opportunities for end-to-end synergy than existed previously: broadband communications, data compression, large data base methodologies and data exploitation tools all allow broadened opportunity. Technical and other collection assets could be employed cooperatively rather than independently, tipping off each other with minimal time lags. The aim should be to achieve greater efficiencies and higher quality product through coordinated collection, so that the total product when collectors are working together is greater than would be the sum of their output working separately, as they do today. Such efficiencies might also reduce costs by allowing deployment of fewer collectors to achieve given requirements. It should be possible, for instance, to avoid redundant collection and to select the most effective and least costly collector. Cross-tipoff or "cross-cueing" of technical platforms would allow near-real-time reaction to overcome denial and deception tactics or to capitalize on opportunities. Likewise, key human intelligence (HUMINT) or open-source data should be distributed and rapidly acted upon by other collectors. Coordinated use of satellites and of aircraft-satellite combinations could permit greatly improved tasking and geolocation without deploying additional platforms. During crisis or war, efficient use of collectors becomes particularly important, because there is great competition for limited assets. Historically, very little attention has been accorded to synergy in the collection area. This is partly because each of the INTs developed in its own "stovepipe," with jealous protection of bureaucratic turf. Even within agencies, there was very little cross-cooperation between program managers. Rivalry among National Reconnaissance Office (NRO) components and program managers was legendary. Aircraft and spacecraft architectures usually were developed separately, and service rivalry impeded comprehensive aircraft planning or division of labor. Tasking of and reporting from sensitive CIA/Directorate of Operations (DO) human assets is highly compartmented, as are the existence and operation of other "black" collection programs and many of the sources managed by the National Security Agency (NSA). Open source information often was slighted or belated, and is distributed in separate unclassified channels. The habit of operating in isolation extends from collection through distribution, each INT or program often having developed its own idiosyncratic communication and receive system. As a result, the systems and their collection managers usually cannot "talk" to each other for rapid tipoff or cooperative target geolocation (especially important to overcome denial and deception and in wartime). Individual users receive directly only the data for which they have procured specific receive equipment, if indeed the communications capacity is available to distribute that data. Just as we have had difficulty getting data collected by national systems out to the field, often we are unable to transmit collection from tactical assets back to the United States, where it could be integrated with data from other sources and evaluated by more analysts. There have been some initial steps to address these problems, but most are in their infancy. Not only is there a very long way to go, but we should squarely face the choices between fragmented and comprehensive, as well as evolutionary and revolutionary, approaches. Maintenance of adequate security represents another challenge. Fused collection is particularly difficult in the signals intelligence (SIGINT) world, especially when it is to be utilized for geolocation purposes, because collectors operating at vast distances from each other must determine whether they are receiving the same signal at the same precise given time. One of the major impediments to this is synchronizing (signal) time of arrival to a specific portion of a single SIGINT electromagnetic wave. This, in turn, requires that each collector be synchronized to precisely the same "clock" in nanoseconds, to determine the precise receiver location -- a feat difficult in itself, but even harder when each system was developed independently with varying precisions, equipment and methodologies. Ongoing R&D is addressing the timing problem. Even if it is solved, a means of communicating the data between collectors, especially when field-deployed or mobile units are involved, can be a formidable task. And if the communications lines exist, efficient operation requires that data formats be compatible, again problematic when each of the existing systems was developed in isolation. The Defense Airborne Reconnaissance Office's (DARO) Joint Airborne SIGINT Architecture (JASA) attempts to evolve standards, interface protocols, hardware and software to develop coordinated and interoperable airborne SIGINT collectors. The apparently large disconnect between the spacecraft and aircraft architectures should be a matter of high-level concern. The NRO and DARO have executed a memorandum of understanding which provides for common standards, especially in timing clocks. However, in other areas, spacecraft and aircraft will continue to go their separate ways unless further action is taken. Distribution systems, data formats and data bases will not necessarily be interoperable. Each community will develop its own software, although much of this probably could be shared. Developmental work on attacking the most difficult existing and future signals should be better integrated between spaceborne, airborne and ground systems. Indeed, it often appears that cooperative focus on improving performance in core present and future SIGINT competencies has taken a back seat to one of the more difficult and even exotic SIGINT applications, i.e. extremely precise target geolocation. The latter has been driven by the military development of expensive precision-guided weapons which often outstripped the ability of US intelligence to provide highly accurate target positions. In the process, more basic concerns -- such as the less difficult but potentially very productive task of rapid tipoff between collectors and the issue of whether we will even be able to find future signals in order to geolocate them cooperatively -- appear to have been given less priority for collaborative effort. It is also unclear whether the NRO will, in practice, accord increased synergy the priority it has received historically. SIGINT has captured most of the attention regarding synergistic collection, and the reason for this is unclear. Imagery requires less precision and overall, is easier to "fuse." Further, while the NRO likes to advertise its goal of creating a "system of systems," cross-INT collection synergy does not seem to be receiving much attention. As other studies have pointed out, at present there is no structured, consistent Community-wide set of requirements for the collection, processing, exploitation and dissemination of information. Processing includes storage, translation, scanning, formatting, structuring, indexing, cataloging, categorizing and extracting; there are no Community standards in any of these steps. Therefore, tasking systems also must be "stovepiped" according to the platform or the "INT." Archived material must be retrieved through varying procedures, and in some cases, archive retrieval nonetheless has been extremely inefficient. If we could achieve a single workstation for exploitation of all INTs, we could much more easily serve the user, address gaps in the data bases and requirements, evaluate information sources and task collectors. In theory, there seems no reason why this cannot happen. With the move to digitization, "bits are bits," and data consists only of ones and zeros. With coordinated and accepted standards and protocols, compatible automated systems could be built which would be able to exchange data. If these standards and protocols were made as close as possible to commercial standards, various users not only would enjoy independence and flexibility in selection of vendors, but also would experience considerable cost savings both at the outset and for upgrades. Examples such as the cable companies' expansion into various forms of data transmission should be an inspiration for the IC and a partial basis for judging its efforts. Cable companies now are creating systems to accommodate video (IMINT), telephone and fax (COMINT) and computer exchanges. But the revolutions witnessed in the commercial world have been slow transferring to US Intelligence, which will increasingly lag unless it opts immediately for a much more vigorous, ambitious and holistic approach. Further, the problems experienced recently with Joint Deployable Intelligence Support System (JDISS) indicate that serious follow-up enforcement must be part of the plan. Collection Management It has been argued above that collection platforms should be built and operated to function in complementary and coordinated ways, to improve efficiency. Many of the barriers to this goal are cultural, political and institutional rather than technical. At present, each service or "stovepipe" controls its own collectors, subject to the direction of standing requirements committees or, in crisis and war, to the overriding authority of the Joint Task Force Commander or his designee. The Persian Gulf War illustrated the difficulty of achieving centralized control even when one has the putative authority. Theater collection managers found it hard to ascertain what assets were in theater, much less to control them intelligently. With the eventual availability of over 150 types of platforms of varying capability, it was extremely difficult to find anyone with the requisite knowledge to orchestrate them effectively. Military service specialties do not include intelligence collection management, and relatively few analysts take the time to learn the arcane technology and requirements processes. When overwhelmed with duties, one of the first tasks they eliminate is collection management; and if they are assigned to a low priority area, this increasingly is a practical decision, since their submitted requirements often are unlikely to be filled anyway. There are not established lists of people with such competency, so reliance is placed upon a word-of-mouth "old boy" network to find and reassign known experts. As a result of these deficiencies, national collection management experts had to be seconded to the theater, departing at a time when their skills also were most needed at home. The Gulf War allowed a six-month buildup, which was fortunate, because from the intelligence collection viewpoint, the time cushion was desperately needed. Less than 50 intelligence experts initially were allowed in theater. Weapons also were given priority over intelligence collection platforms, in the view that this would best deter the Iraqis from hostile action. Even when intelligence platforms could be imported, those controlling them sometimes were uncooperative, the classic case being Air Force policy regarding the developmental Joint Surveillance Target Acquisition Radar System (JSTARS) aircraft. Jointness and cooperation were enforced by placing intelligence experts from different venues side-by-side with each other and with operators, to overcome historical barriers to cooperation. Deconfliction of requirements became a delicate assignment, for instance sorting out the Army and Marine desire to focus JSTARS on moving targets across their lines and the Air Force demand for focus on deep strike targets for the air campaign. With requirements far exceeding capabilities, collection managers sought to utilize non-traditional sensors, which sometimes could be useful for tactical reconnaissance. They had great difficulty finding out about these sensor capabilities and then in finding out where these systems were deployed on the battlefield. Even five years later, an inventory of such supplemental sensor capabilities apparently has not been made. At the national level, collection management has become increasingly contentious, even before the number of satellites on orbit is slashed within the future architecture. With requirements always far exceeding collection capabilities, some argue that program managers are largely free to pick and choose which targets they will pursue. These targets, it is said, often are those that will make their own INT's performance look good and give them visibility in the crisis of the day. They are not necessarily those that are the most difficult "enduring challenges" or those most uniquely accessible by their particular "INT" or collection system, it is argued, and indeed, they may not know what others are collecting, especially in the case of highly compartmented HUMINT or technical programs. The current system is criticized because the stovepipes essentially control their own budget size and allocations within that budget, although in reality they have little idea how their requirements and capabilities should be prioritized compared to others. And finally, the program managers write their own "report card", with little oversight or review by others. A persuasive argument can be made that the best potential requirements and collection managers are not the program managers or INT-based requirements committees, but rather all-source analysts with expertise in the specific mission areas who have access to all associated collection compartments and data. Some argue that not only should such analysts be responsible for day-to-day collection management, but also that they should have more say in allocating funds for new collection platforms. Taking this last point further, some believe it would be useful to give such issue managers discretionary funds to develop relatively inexpensive collection techniques to fill gaps in their respective areas. On the collection management side, the Counterproliferation Center (CPC) has negotiated agreements whereby some of the INTs have passed much tasking responsibility to the CPC; the result is said to be improved collection and a reduced need for duplicative analytic capability within the INTs, plus a freeing of the program managers from this onus, so they can concentrate on other responsibilities. A contrary view recently was presented by the Intelligence Capabilities Task Force, however, which found a high degree of agreement between analysts and collectors that somehow system program managers left to their own devices have managed to build the right system and collect the right material. The Task Force does concede that there exist many "enduring challenges" or gaps, as well as a growing denial and deception problem which has not been acknowledged by most analysts. Just as there is often little control over disparate theater operations unless a Commander-in-Chief (CINC) effectively exercises his options during crisis, at the national level there is no centralized collection management looking across all the INTs and deciding which can most effectively pursue a given target. This deficit arguably has become more problematic since the end of the Cold War. The Soviet targets on which most of our collection previously was focused were largely predictable and slow to change. Most US intelligence players had a fairly set role, and relatively infrequent differences at the margins were adjudicated at a high level rather than on a daily working basis. Now, however, targets are dispersed worldwide and far less predictable, and the strain on resources is greater. Yet we tend still to concentrate on management of static target decks, even as the need grows for far more flexible, ad hoc, rapid reaction to changing circumstances and opportunities -- for support of the military balanced against enduring requirements, for overcoming denial and deception, and for effecting synergy through rapid response to tipoff. The new strain on collection management is especially exemplified by the dilemmas arising from the recent development of simultaneous military involvements in various areas of the globe. Partly because US political culture has evolved to intolerance for even a low level of casualties, military and political leaders are inclined to throw all available intelligence resources against these sensitive situations, even though their marginal contribution there may be far less than if they were collecting in a non-crisis area. Hence the foundation of the widespread complaint among top civilian analysts that collection has been excessively skewed to support for current military operations, to the fundamental detriment of maintaining an intelligence base on non-crisis areas and issues more fundamental to long-term U.S. security. While support for military operations (SMO) is seen as the culprit, however, in reality this is not a "national versus military" dichotomy, but rather a near-term or crisis focus at the expense of medium- to long-term requirements, the latter including SMO. This is true for two reasons: first, the top "national" leadership and users are clamoring for crisis coverage as much as is the military leadership, since military involvement and setbacks in such spots have considerable political as well as military implications. Second, those areas from which collection has been drawn off are also extremely important to the military. Indeed, since military interventions have been occurring in unpredicted areas of the Third World, failure to maintain an adequate base probably will affect most severely our future capability to support military operations. When requirements outstrip capability, prioritization obviously is needed. However, PDD-35, which established a "tier" system for U.S. Intelligence, in some ways appears to have worsened the problem. Analysts believe the tier system is being imposed too rigidly. As a result, the top five or six requirements receive the great majority of the resources so that we do them exceedingly well, but those below, especially those beneath the top tier level, languish with leftovers at best. While this would not become a major issue if intensive intelligence support for interventions or crises lasted only for a few months, prolonged involvements have become increasingly common and have intensified collection management conflicts. Critics of such diversions argue that decisions such as these often have reflected a lack of appreciation for balancing requirements, for longer-term US priorities and needs, and for the fact that piling on additional collection may bring only marginal value added, but at considerable opportunity cost. Such acrimony can only be expected to increase dramatically in the future, if we implement plans to reduce greatly the number of satellite collectors. And the accumulation of diverse capabilities on huge satellites means that whatever such a satellite's theoretical collection capabilities, in reality, severe tasking conflicts often will develop; pursuit of one task may have to be accomplished by excluding use of another capability, or the attempt to execute both over a given area and time may cause inefficiencies. What Share for Collection ? During the 1980s, critics argued that US intelligence had a largely peacetime orientation toward arms control and other "national" issues, and that it was not designed to serve the warfighter well. With an orientation on collection and a focus on distribution to national users located primarily within the Washington beltway, it did not demonstrate the agility, rapid data fusion or dissemination to far-flung areas which was needed to support field operations efficiently. Although the Gulf war was a far less stressing scenario than we might one day face, and although US intelligence performed well overall, the legitimacy of these critiques largely was confirmed in 1990-1991. The need for more investment in processing and exploitation has deepened as collectors are being designed to amass far larger volumes of data. Critics also long have contended that expensive satellites are not being used efficiently, especially during the early deployment phase of new and upgraded systems, because requisite processing and exploitation capability on the ground are given short shrift and developed only belatedly and sometimes halfheartedly. As a result, billions of dollars routinely are spent on collection systems that have for long periods of time been used suboptimally. The data available to date have indicated that the tendency to favor collection has grown stronger rather than weaker. Since 1992, the budgetary priority and dominance of collection apparently has increased rather than decreased. As the intelligence budget has declined, collection has taken fewer cuts within both Tactical Intelligence and Related Activities (TIARA) and National Foreign Intelligence Program (NFIP) budgets, and hence consumes a larger share of available resources than previously. The NFIP collection budget is dominated by the National Reconnaissance Office, whose budget has climbed fairly steadily and is projected to continue doing so. The requested National Reconnaissance Program (NRP) share of the NFIP, therefore should continue to rise within a static or declining overall NFIP budget. Satellites and associated ground facilities also were taking more of the reduced collection portion of NFIP funds. Nonetheless, the overall collection budget has been faring better than other portions of the NFIP. The TIARA budget is weighted less toward collection, probably in part because many intelligence dissemination systems must be financed within the services. Comparison of 1989-91 figures with 1995-97 projections also show that collection has fared well within TIARA. With respect to TIARA, it should also be noted that unmanned aerial vehicles currently developed as prototypes under Advanced Concept Technology Demonstration (ACTD) programs are not funded for production, and collection budget increments for this purpose might be necessary beginning in FY 1998-2000. Likewise, there is a potentially large unfunded processing, exploitation and dissemination bill for these systems; attention and funding to date usually has concentrated on the collection portion, despite historical neglect and inadequacies in other areas. Overall, TIARA investment in imagery collection has been increasing, but imagery processing and dissemination admittedly are not funded adequately under current TIARA projections. Many in both the Executive Branch and Congress, including this Committee, increasingly have objected to the traditional budgetary dominance of collection and believe we could achieve more value for the marginal dollar by shifting funds to processing, exploitation, analysis and dissemination. This consensus has grown since DESERT SHIELD/DESERT STORM highlighted deficiencies in "downstream" activities, notably dissemination. The aforementioned Intelligence Capabilities Task Force also has provided a dissenting note on this issue, however, finding that collection and production/analytical capabilities have been pretty well balanced, and that if anything a slightly greater emphasis on collection may be needed. It should be noted, however, that at present we often collect significantly less than our capability, since platforms are built with capacity excess to projected normal operating requirements to allow for surge capacity. Regardless whether collection and downstream capabilities other than dissemination were well balanced in the past, many would argue that there will be a future imbalance favoring collection if action is not taken. They fear that it will be difficult to make efficient use of large prospective increases in data, to be collected by technical platforms now planned or under development as well as by "open source" methods. Indeed, some top analysts believe the community already fails to exploit adequately the imagery and signals data currently being collected and processed. While inevitably we will always collect significantly more data than we use, some wonder whether we can continue to explain or rationalize the collection of large excesses, especially since only a very small part of what is collected is actionable. Prominent experts have voiced to the Committee worries that in the future it will become more difficult to separate the wheat from the chaff, and that we could become overwhelmed with data and unable to reduce it to the information we really need. Some have wondered whether we will need a new class of data sorters, to cull information to forward to data users. On the other hand, however, users -- and builders -- sometimes have been loathe to reduce collection platform requirements, which might in turn reduce costs. Some also note that arguments over intelligence assessments usually are resolved definitively only by acquiring more data, not by more analysis. The Chairman of the House Permanent Select Committee on Intelligence (HPSCI) has adopted a position that fundamentally transcends this argument about whether there is an imbalance between collection and downstream activities. It is his view that satellite collection and ground systems, which as noted above account for approximately half the NFIP collection budget, probably could be accomplished for far less money, thus freeing up large sums of money for more innovative collection schemes, for greater investment in downstream activities, and/or for reductions to the intelligence budget. This reduces us to the proposition that we can do it smarter, that technology allows the future NRP to collect as much as or more than now planned, for much less expenditure. The aim should be to reduce substantially the cost of some or most "baseline" NRO systems in order to free up money for other purposes. Moreover, we should attempt simultaneously to decrease satellite system vulnerability and increase our capability to counter denial and deception. In its FY 96 authorization bill, the Committee advocated immediate and aggressive development of prototype small spacecraft imagery alternatives, including associated rapid acquisition practices and perhaps completely modernized ground facilities. The authorization conference referred this proposal to an independent panel established by the Director of Central Intelligence, which is to report back this spring. Potential savings could contribute greatly to containment of collection costs, with the added benefit of providing more platforms, thus decreased vulnerability and greater coverage or revisit. While small satellite applications have to date concentrated on imagery platforms, their potential for SIGINT and communications applications also should be accorded high priority. Regardless whether the panel decides to proceed with development now, we believe that smaller and cheaper satellites are the technological wave of the future, and that the IC also will adopt them eventually, if belatedly. Secondly, the Committee initiative already has spurred the admission that far lighter and less expensive "medium satellites" could be built, confirming our view that considerable reductions could be made to the NRP spacecraft budget. To date, there has been less study and movement regarding ground systems. Thus far, the NRO's reaction to rising costs has been the opposite of what we have recommended. Acknowledging that space system costs were becoming prohibitively expensive, the NRO accepted the recommendations of a 1992 panel to reduce the number of spacecraft on orbit by nearly half, compensating for this by loading up still more investment and capabilities on the remaining upgraded platforms. The theory behind this was that after initial investments, constellation costs would come down. Instead, however, it appears that, at best, expenditures would level out at higher levels that previously. In effect, we have roughly doubled our costs per spacecraft, as well as increasing our vulnerability to denial and deception and to accident or attack. Technology Allows More Capability at Less Cost Two Committee IC21 hearings on technology trends reinforce our conclusion that commercial technology and practices hold the key to relatively painless reductions in collection costs. Witnesses agreed that commercial technology is much cheaper, is widely available, leads government R&D in many areas, and is characterized by rapid (six to 24 month) generational turnover. The challenge for government, they said, will be to concentrate government R&D in key niche areas with little commercial use or interest, and to change radically our acquisition philosophy and processes. Success will be dictated by our ability to concentrate on swift application and fielding of commercial standards and the latest commercial technology, allowing us to maintain a qualitative and cost advantage over adversaries. This will also permit a more robust, competitive and easily maintained industrial base. Of all the technology advances, perhaps the most important is in processing and microelectronics, or "information technology." Rapid generational advances in this area, with turnover every six to 18 months, have important applications throughout the intelligence spectrum, from "upstream" collection through "downstream" processing, exploitation and dissemination. These continuing revolutions in processing capability, for instance, help permit fielding of spacecraft that are not only lighter and cheaper but also smarter, allowing greater on-board processing of information. The latter, in turn, could permit direct dissemination to the field and communication between satellites. For some applications, eventually "micro-satellites" deployed in "clouds" and communicating with each other and possibly with a larger mother satellite might feature distributed collection and division of labor, thus allowing inexpensive reconstitution or selective parts replacement. Rather than embracing the advancing technology, however, the NRO opted to continue making very large satellites, which are very costly in themselves and also are extremely expensive to launch. Partly, these decisions traced to an assumption that we could not get all intelligence assets off the TITAN IV, and if we could not do so, we might as well put a lot of NRO spacecraft on TITAN IV in order to avoid increasing the already enormous costs per launch. Therefore, for example, despite major advances in composites and lightweight materials, spacecraft bus often remain very heavy. Similarly, electronics often are much heavier than current technology allows. Examples of major technology advances which could be incorporated to reduce spacecraft size and cost while retaining capability include: gimballed or phased array antennae; high efficiency solar arrays and high density batteries; high performance computers and digital commercial DRAMs; and more advanced attitude control systems such as Inertial Measurement Units (IMUs), Star Trackers and Global Positioning System (GPS) receivers. Even where the NRO has pioneered new technology, its baseline programs have not always moved to put it on orbit quickly. In processing, too, better adaptation to commercial standards and rapid technology advances should revolutionize the way the NRO and others do business. In the NRO, ground processing policy often has mirrored the approach to associated satellites. Usually we have resorted to very expensive upgrades of custom-built, vendor- specific, old and inefficient technology. This is one reason why ground processing now can represent two-thirds of space system costs. With dramatically improved processing power and software based as much as possible on commercial standards, tremendous efficiencies and cost savings are possible. This is why some of the small satellite proposals advocate redesigning processing systems with "a clean sheet of paper" approach. Because individual satellite programs currently use different contractors with system- and proprietary-unique processing, this must be changed before we can fully acquire cross-platform, cross-INT collection synergy. This also reinforces the need to integrate ground facilities based on common standards and protocols and on commercial technology to the fullest extent possible. Smaller satellites could potentially feature life cycle costs less than half those of some current satellites, freeing up billions of dollars. Often, smaller satellites also offer important advantages other than financial savings; one major point is that we could put more platforms on orbit, allowing better revisit time, more flexible worldwide coverage, decreased vulnerability and more a efficient industrial base. Advanced technologies such as those allowing increased processing aboard even lighter weight spacecraft now render it possible to disseminate selected data direct from the satellite to simplified, distributed ground stations. This might gratify users by sending some data directly to the field, and it could also reduce our vulnerabilities due to chokepoints in these systems. And, once again, it is commercial technology which has led the way in developing concepts for direct dissemination to individual users. There has developed a belief that "direct" or "global" broadcast is a better option than direct download, since it allows processing and fusion of material in the US and distribution of culled information to military units that might otherwise be overwhelmed. However, it appears that global broadcast and direct downlink (DDL) from collection platforms should be considered complementary rather than competing alternatives, so long as DDL is executed in a cost effective manner. Field ground units could collect from tactical assets and broadcast processed information up to satellites for transmission back to the US. They could task and collect from satellites via direct downlink only the most important data for their purposes, and would have only themselves to blame if they got too much to handle. DDL would ensure their timely receipt of the most important data, the ability to view high priority "raw" product fully, protection against possible communications interruptions or priority problems, and provision of a minimum backup against satellite system vulnerabilities. In general, this study argues that the NRO should eschew a policy of extremely expensive, evolutionary upgrades and instead seek revolutionary leapfrog technology based mostly on commercial technology wherever feasible and prudent. However, affordability also will require a change in acquisition philosophy similar to what others have urged for Department of Defense (DoD) programs. Systems will have to be produced quickly, competitively, and in larger quantities, in order to control costs and get technology on orbit promptly. DoD directives to minimize military specifications on existing and planned systems will have to be taken seriously. Management superstructure should be minimized, and personnel reduced to the minimum needed. This is contrary to current trends. Further, NRO "base" or support costs constitute fully one-third of the NRP, and have not been delineated well for outside or Congressional scrutiny. Streamlined acquisition philosophy also focuses on requirements rather than contract specifications, allowing the contractor to determine how to meet those requirements. Fixed price contracts should replace cost plus contracts wherever feasible. In the past, NRO contractors were incentivized primarily to extend satellite life, with profits increasing accordingly. Hence, intelligence satellites have become very long-lived. This philosophy, too, probably should be reconsidered, because as technology advances ever more rapidly, it has complicated efforts to get new technology on orbit. Despite these advances in longevity, the NRO continues to resist altering artificially low "mean mission duration" (MMD) estimates, according to which acquisition schedules are planned. The result has been inefficient procurement stretch-outs, belated cancellations, high satellite storage and team maintenance costs, constant disruption to an incorrectly sized industrial base, and attendant high overhead costs which are passed along to the government. In addition to these inefficiencies, stubborn adherence to artificially low MMDs has driven us to numerous policies that otherwise would be considered illogical, if not downright silly. Apportioning the Collection Budget Regardless how they are operationally used, there is widespread agreement that there is little logic in the process for deciding which collection capabilities we most need and should acquire in the first place. Not only are there few means for trading off the value of one potential platform against another, but there is little mechanism for trading off collection against other priorities. It is striking, for instance, that the division of resources among the INTs has remained largely static over the years, especially within the NFIP, which is less volatile as a whole than is TIARA/Joint Military Intelligence Program (JMIP). This static -- or stagnant -- status persists despite vast changes in world politics, targets, and technology. Measurement and signatures intelligence (MASINT) also presents a perplexing case history. Difficult to understand and often without an established constituency, under the current budget allocation system, it will have a hard time coming to its own due to declining budgets. Indeed, MASINT budgets have shrunk we rushed to shut down traditional radar collectors on the theory that they no longer were needed for the post-Cold-War period. Yet many believe that MASINT collection could become the most exciting future intelligence technology if properly managed, and if these and other potential new initiatives were not considered primarily as threats to the financial viability of expensive existing programs. Non-technical collection capabilities considered relatively cost effective sometimes also have had difficulty maintaining and increasing budget share. HUMINT, for example, sometimes has been cited as potentially far less expensive than technical platforms as a means of collecting the most highly focused and sought-after intelligence requirements, e.g., on enemy leadership and intentions. This could be particularly true if civilian and military HUMINT collectors undergo the cultural change of realizing that their future is brightest if they wholeheartedly marry HUMINT operatives to technical collection, something now made possible by the advance of technology and miniaturization. Open source intelligence traditionally also has had a difficult time increasing market share commensurate with its potential. The growth of open source material should allow a further refinement of collection strategies and an ability to concentrate the limited number of technical collectors on the truly "hard targets." However, the burgeoning availability of open sources has complicated the IC's ability to manage the amounts of data now available. In addition, there is a bias among some in the intelligence and policy communities against open sources, stemming from the erroneous belief that no information that is valuable is likely to be easily accessible or unclassified. This prejudice severely undercuts the utility of open sources and can only be overcome through positive action. Moreover, the under-utilization of open sources -- and HUMINT -- may be due partly to a lack of understanding among users about their potential and how to use them. The IC has been addressing these problems for the past several years and should devote more resources to them, given the savings this may create in terms of overall collection costs. Such collection budget allocation problems apparently derive partly from the observation above that each stovepipe or program determines its own budget and writes its own report card. There is little mechanism at the top level for judging between them, and some argue that it would be virtually impossible to maintain in one decision-maker or centralized location the detailed knowledge of all the diverse intelligence programs and capabilities that would be needed to inform centralized management over a sustained period. The only current institutional mechanism for effecting such trades within NFIP has been the Community Management Staff (CMS), which sometimes has been directed not to interfere with program managers. Moreover, program element monitors within CMS are detailed from elsewhere in the Community and eventually must return to their old positions, so are in a poor position to issue judgments which might be unpopular with their parent organizations. Some argue that both collection management and program trades at the margins can best be effected by the all-source analysts located in centers, by task forces or by issue management teams. These persons are read into most or all relevant collection programs, know their capabilities, access and current operations, and can judge past performance and cooperation compared to other collectors. One suggestion is that these groups be given some "seed money" of their own, so they can pursue low-cost collection programs which now languish as large, expensive programs receive the attention and money. It can be confirmed that on Capitol Hill as well, allocations of a few million dollars often are scrutinized far more carefully than large programs, although their sums amount to less than the rounding errors of the latter. These seemingly intractable problems regarding allocation of the collection budget might be approached in a novel way by considering development of a "market" approach to apportioning collection monies, rather than the current system. The market approach would seek to avoid the problems of the "command economy" alternative most often considered; for objective, long-term expertise in these many and complex programs probably is at best fleetingly achievable in an all-powerful DCI or collection "czar" or centralized staff. A market system might also present numerous other advantages, although implementation could be difficult, at least initially. The following exemplifies the outlines of such a system, which requires further thought and development of detail. One way in which a market system might be implemented would be to apportion among intelligence users money or monetary "chits" for the coming and out years, which they could divide and allocate among potential collection systems that appear able to meet their future requirements most cost-effectively. Those most successful in allocating their money wisely would not be punished by taking away savings, but rather would be free to use those savings for additional collection benefiting themselves. Under this example, a method would have to be devised for fairly apportioning money or monetary "chits," representing non- baseline dollars, among users/consumers, with flexibility for changes in perceptions of need/fairness and in national security priorities over the years. On the military side, for instance, consumers could include the Defense Intelligence Agency (DIA) all-source analysts, CINCs, services, joint staff and the Director of Military Intelligence; on the civilian side, they might include the DCI, departments and agencies, the National Security Council (NSC) and CIA all-source analysts and centers. If necessary, a means could be found to weight a portion of these votes towards "enduring challenges" or long-term gaps and for collection to overcome denial and deception, e.g., by requiring individual users suffering from such gaps to expend a percentage of their chits in this area or by setting aside a bloc of DCI and DMI chits for this purpose. Core or "baseline" capabilities would be determined and maintained for program stability, but would be thoroughly and critically reviewed both initially and yearly thereafter for cost effectiveness and operational responsiveness to consumers. Any questions or discontent surfaced by either an independent staff permanently assigned to a CMS-style organization or by Congress and consumers would be aired thoroughly and periodically reviewed by the consumers, with budgetary adjustments made accordingly. An accumulation of enough "chits" could either finance a fully designed and costed system as presented to users or, in planning and requirements stages, represent the cost and requirements/users for which a system should be designed. Program managers would have to market their proposed product among potential users/payers/voters. A truly independent CMS (not using agency detailees) could serve not as the DCI's resource to grade and prioritize programs, but as a "truth in marketing" organization for technology risk and cost estimates, to which users could refer (cf. Intelligence Community Management staff study). If high-cost but necessary systems could not achieve funding "critical mass," a "runoff" system might have to be developed. Such a "market" system would appear to have the advantages of: naturally eliminating unnecessary redundancy; favoring lower cost systems; forcing users to prioritize their requirements more carefully, since users would have only a limited amount of money to spend for their particular needs and would be truly paying the bill; forcing a debate over requirements priorities, both when distributing and when expending chits; and presenting incentives for cross-service, cross-TIARA/JMIP/NFIP investments, depending upon which option would meet needs at lowest cost, since the user would be able to retain savings for other purposes. Program managers would be incentivized to minimize compartmentation and program costs, and both they and users would be motivated to form groups of multiple users who might share the bill. Once the system was operational, collection management would be geared to satisfy those who had paid the bills, in order to sustain their support for the existing system and maintain consumer trust for future budget decisions; utilization for other unforeseen customers could be directed by the DCI or his collection deputy. As in a true market system, the DCI and other users would be free to trade informally some of their own chits/votes, as they saw fit. The system would become more free-wheeling, and aspects of it might seem undesirable to some. Consumers would have to become far more educated on the range of collection systems and opportunities than most are now, and inevitably would make some errors. Political infighting and wheeler-dealing would continue to flourish, especially over consumer "chit" allocations. Expert marketing or salesmanship could become a program commodity as valued as substantive expertise. However, consumers primarily voting their own self-interest ultimately should produce a more rational, efficient, fair and flexible system than we have now or than could be achieved and maintained under "command economies" overseen by the DCI/CMS, DMI and individual services. Recommendations Collection Synergy and Collection Management 1) Interoperability should be effected through a high-priority revolutionary approach rather than through the evolutionary methods now contemplated; the latter would delay achievement of extensive synergy for a generation. This revolutionary approach would accept more short-term risk and disruption in exchange for much larger and quicker pay-off. - For the near term, universal translators should be developed and fielded to put headers on data coming from "legacy" collectors using diverse protocols and standards, thus providing a conversion factor for all pulse description words. - Over the next five years or so, comprehensive standards and protocols (for timing, ephemerus, frequency, geodesy, etc.) should be developed and enforced for new systems, similar to the multi-layered standards set for the computer science industry by an international standards organization. - Synergy thus should be maximized from collection through processing, exploitation and dissemination. The number of unique systems and components should be minimized, and use of commercial off the shelf components maximized. With digitization and proper standards, we should eventually be able to disseminate, exchange and exploit all data within a common transmission/receive system, just as the commercial world now is leading the way in routing voice, video, computer and fax over the same lines. 2) An independent DCI/Secretary of Defense (SECDEF) level board should be established which sets and enforces all necessary standards, protocols, etc., for intra- and cross-INT interoperability from collection through dissemination and exploitation, basing them as much as possible on commercial standards. (cf. Intelligence Community Management staff study and its discussion of the Infrastructure Support Office (ISO)). 3) While we should be effecting a shift from single system geolocation to collaborative geolocation, too much of the initial focus of fused collection has been on what might be the most demanding of fusion problems, i.e., the achievement of extremely precise geolocations. Much greater effort should be devoted now to cross-cueing and integrated collection management, with high priority on cross-INT aspects. 4) All-source analysts extensively trained in collection management and with access to data from all collectors relevant to their mission area should select and task the collectors most suited to their problems. (cf. Intelligence Community Management staff study on CMS collection management and electronic connections with analysts and collectors.) A concerted effort must be made to develop and sustain this expertise at both the national and tactical levels, through improved, centralized cross-INT collection management training and utilization programs. 5) It seems necessary to centralize collection management in order to: reduce duplication; effect cross-INT trades and use the most efficient collectors; achieve desired collection synergy and counter-denial and deception (D&D) capability; and provide improved collection dexterity and responsiveness suited to the post-cold war world. (cf. Intelligence Community Management staff study.) - With improved communications and computer programming and graphics, and with a transformation to "bits are bits" synergism, multiple centers could exist with independent capability and full interoperability. For instance, there could be a national collection management center as well as tactical command and control/information centers in each major regional command, plus ad hoc hoc teams for local crises or operations. - Computer programs could depict all available assets and their tracks, and automatically compute the most accessible and cost-effective collection solutions. Interoperable dissemination could bring all requested data from any source down to a single point -- with digitization, "bits are bits." 6) Improved, common data bases with easy retrieval by those at remote locations are essential for synergism in both tasking and exploitation. (cf. Intelligence Community Management staff study.) 7) The Intelligence Community must find a better way to manage and balance near- and longer-term priorities, which recently have become too weighted toward support for current crises and interventions. Collection-Downstream Balance 8) The NFIP/TIARA budget should be broken out within the five cross-program categories of collection, processing, exploitation/analysis, communications/dissemination and infrastructure. The purpose of these groupings would be to focus policy and budgetary attention on the relationships and trends between the five components. At minimum, overall figures with accompanying tables of component line items should be presented in overview books/portions of the Congressional Budget Justification Books (CBJBs)/Congressional Justification Books (CJBs) for FY 98 and beyond. This approach could be compatible with and complementary to mission-based budgeting. If detailed mission based budgeting does not prove practicable, these five divisions could form the basis for building the budget and for organization of all CJBs/CBJBs, and could be a vehicle for forcing competition for decreasing funds within and between the five divisions. Categorizing the budget in this way should also incentivize programs to reduce costs (see below). - The collection category should include the platform command and control portion of the ground infrastructure, but there should be further study of whether any initial ground processing should be included within the collection category, and, if so, to what level. - TIARA, JMIP and NFIP activities should be budgeted and operated cohesively, since the distinctions between them are decreasing or disappearing. - Congressional budgetary oversight would best be organized along these five budget categories as well. 9) The DCI and Secretary of Defense should determine percentage allocation goals among these five components, which would redistribute resources over a defined period of years to a more rational and less collection-heavy budget. - Exploitation/analysis should receive highest priority for improvements, especially automated exploitation/data screening; an attempt should be made to quantify the extent to which automated exploitation improvements are needed to cope with increased data flow and to quantify how increases in collected and processed data and improvements in automated exploitation should affect analytic manpower levels. Dissemination also is a very high priority, but more rational, cross-INT, common dissemination of digitized information might eventually reduce funding requirements in this area. In the processing area, SIGINT requirements could become so financially and technically demanding that we should now reappraise the long-term cost-effectiveness and viability of current approaches. Processing should be sized and financed to ensure efficient use of new or upgraded collection systems from Initial Operating Capability (IOC) through Final Operating Capability (FOC), including in these calculations the use of likely "residual" or partially operational systems. 10) Overcoming denial and deception which we have experienced or to which we have known vulnerabilities should be a major factor in establishing requirements and budgetary priorities, for both collection and downstream activities. - The collection community should be shifting a significant portion of its resources toward unwarned/unexpected collection, and downstream investment and analytical resources should be specifically devoted to means of overcoming denial and deception. Reducing Collection Costs 11) The following is considered a finding rather than a recommendation, which should be further studied for feasibility and implementation details. We should try to devise a system whereby all types of collection, including TIARA/JMIP as well as NFIP, human and open-source as well as technical, are forced to compete for money from a common, reduced pot of collection money. A "market" approach, rather than the current system or the alternative "command economy" approach, should be developed, in which intelligence users/consumers individually and collectively decide which collection systems might best meet their needs. 12) Costs should be delineated as thoroughly for "baseline" collection and other programs as for non-baseline programs. The NFIP practice of maintaining an undelineated intelligence "base" should be banished, both to promote needed transparency for users and Congress, and as a logical fall-out of dividing the intelligence budget into five parts with separate lines for each, including infrastructure. 13) Congressional Budget Justification Books (CJBs, CBJBs) should be written to elucidate clearly the costs, limitations and mission applications of existing or proposed collection systems. If the above "market" system of budget allocations were implemented, these books would serve as the basic reference documents for users as well as for Capitol Hill in assessing individual programs. 14) Planned NRO funding levels should be reduced, and there should be an immediate shift in direction toward rapid deployment of more, smaller and cheaper satellites wherever this is practicable, with appropriate measures to maintain large satellites in these respective areas so long as reasonably necessary to hedge technology and development risk. 15) We should move to supplement broad area and multispectral collection with commercial satellite sources, maintaining a minimum core capability but relying heavily on commercial adjuncts and surge capability. Modernized ground stations should be made compatible with commercial standards and capabilities. 16) Especially if the NRO does not move toward a far more distributed, robust architecture than now is planned, the military should consider developing inexpensive and possibly reusable "tactical satellites" to supplement national collection over denied areas during crises. 17) NRO ground systems should be modernized as required, using a "clean sheet of paper" approach and employing commercially based, interoperable technology to the greatest extent practicable, except for necessary specialized applications. This should allow meaningful and continued contractor competition, drastically cut both initial and upgrade costs, and be designed to maximize synergy between collection systems and associated ground stations. A systems integrator should be hired to study the best way to effect these goals, and we should consider the possibility of maintaining updated, cohesive ground stations by contracting out to a systems integrator (cf. Intelligence Community Management staff study). 18) On-board processing and partial data transfer through direct downlink should be pursued as a means of better serving customers, reducing satellite system vulnerability and potentially reducing costs. System vulnerability and chokepoints should be addressed as a matter of intense concern, especially if the prospect of information warfare is taken seriously. 19) The current method of gearing acquisition strategy to an artificially low calculation of expected satellite life should be altered to reflect actual experience and more realistic expectations. Spacecraft program managers should consider elimination of a specified mean mission duration in contract requirements and contract incentive rewards, allowing this to remain as a "bonus" factor in evaluating contract competition. 20) Platforms and sensors built for purposes other than intelligence collection should be used routinely for intelligence purposes when this is possible, needed or cost effective. Sensors built for other purposes, but which might provide data useful for intelligence purposes, should be surveyed, inventoried and utilized, for both strategic and tactical collection purposes. 21) Especially in the space area, the focus should be on technology leaps with maximum utilization of commercial developments, rather than on numerous expensive block changes and system upgrades. 22) The NRO's industrial base policy should be closely scrutinized. Expenditures for this purpose should be minimized in coordination with the drive to maximize use of commercial technology. Policies for selection, especially non-competitive selection, of those companies which will survive, become "centers of excellence," or receive all future NRO business, should be revealed and externally examined for both fairness and long-term financial sense. The industrial base problems associated with building and upgrading few complex satellites with long design lifes should be examined. This approach should be weighed against the advantages and disadvantages of building many more and cheaper satellites quickly and in larger numbers, with competitive procurement of leapfrog technology for space and ground segments, rather than relying on expensive block changes and partial upgrades to old technology. 23) A much cheaper system of reliable spacecraft launch should be developed (cf. Collection: Launch staff study.) 24) Program managers building intelligence platforms, especially spacecraft, should immediately embrace the Secretary of Defense's directive to adopt commercial standards for existing and new contracts, minimizing use of military specifications and standards. 25) Acquisition timelines, personnel and paperwork must be reduced considerably, to get available new technology on line rapidly and to reduce costs. 26) There should be a concerted effort to educate users on the utility of lower cost open source and HUMINT information, and this material (with proper safeguards for sensitive clandestine HUMINT material) should be rapidly communicable over the same dissemination system used by other collectors. 27) The burgeoning availability of open source material presents both problems and opportunities. In order to take full advantage of open sources, the IC must continue to develop improved means of collecting, exploiting and processing open source information.