Krizan, Liza.
Intelligence Essentials for Everyone.
Joint Military Intelligence College, Occasional Paper No. 6. Washington. D.C.: Government Printing Office, 1999.
CHAPTER 5
COLLECTION AND THE COLLECTION DISCIPLINES
C
OLLECTION IS THE bedrock of intelligence. Intelligence collection has been written about since the biblical references to spies in Numbers, 13-14 and the Book of Joshua. Without collection, intelligence is little more than guesswork—perhaps educated guesswork, but guesswork nonetheless. The United States and several other nations use multiple means of collecting the intelligence they require. The means are driven by two factors: the nature of the intelligence being sought and the ability to acquire it in various ways. In the United States the means of collecting intelligence are sometimes referred to as
collection disciplines
or INTs. This chapter discusses the overarching themes that affect all means of collection, then addresses what the various INTs provide as well as their strengths and weaknesses.
Primarily in the military, collection is sometimes spoken of as ISR: intelligence, surveillance, and reconnaissance. The term covers three different types of activities.
1. Intelligence: a general term for collection
2. Surveillance: the systematic observation of a targeted area or group, usually for an extended period of time
3. Reconnaissance: a mission to acquire information about a target, sometimes meaning a one-time endeavor
OVERARCHING THEMES
Several themes or issues cut across the collection disciplines and tend to drive many of the debates and decisions on intelligence collection. These themes point out that collection involves more than questions like. “What can be collected?” or “Should that be collected?” Collection is a highly complex government activity that requires numerous decisions and has many stress points.
BUDGET. Technical collection systems, many of which are based on satellites, are very expensive. The systems and programs are a major expenditure within the U.S. intelligence budget. Thus, costs always constrain the ability to operate a large number of collection systems at the same time. Moreover, because different types of satellites are employed for different types of collection (imagery versus signals, for example) or may be equipped to carry multiple sensors, policy makers have to make difficult trade-offs. Significant costs are also associated with launching satellites. The larger the satellite, which is driven in large part by the nature of its sensor package and the equipment needed to power the satellite and to transmit the data, the larger the rocket required to put it into orbit. Finally, the costs of processing and exploitation (P&E), without which collection is meaningless, should be factored into the total expense. Builders of collection systems often ignore P&E and launch costs as part of their estimates for collection.
During the cold war, cost issues for technical collection rarely surfaced. The sense of threat, coupled with the fact that no better way existed to collect intelligence on the Soviet Union, tended to support the high costs of the systems. Also, decision makers placed greater emphasis on collection systems than on the processing and exploitation needed to deal with the intelligence collected. In the immediate post-cold war period, given the absence of any large and potentially overwhelming threat, collection costs became more vulnerable politically. The terrorist attacks in 2001 raised additional questions about the utility of these systems, as terrorist targets are less susceptible to collection via technical means and may require greater use of human intelligence.
There have been several recent decisions that underscore the increased difficulty in sustaining the costs of technical collection. In June 2005, Rep. Peter Hoekstra, R.-Mich., chairman of the House Intelligence Committee, argued that too much money was being spent on satellites and not enough on human collectors and on analysts with language skills. Advocates for both views exist, but this is the sort of argument that rarely would have been made during the cold war. One of director of national intelligence (DNI) John Negroponte’s major collection decisions came in September 2005, when he ordered the Boeing Company to stop work on a system known as the Future Imagery Architecture (FIA), widely thought to be the next generation of imagery satellites. FIA had fallen way behind schedule and had also incurred cost overruns. (According to detailed press accounts, FIA had gone from a program bid at $5 billion to more than $18 billion and was still $2 to $3 billion short.) This move was also seen as an attempt by the DNI to have a greater say in satellite decisions, which have customarily been dominated by the Department of Defense (DOD). Two years later, in August 2007, National Reconnaissance Office (NRO) director Donald Kerr testified publicly (during his nomination hearings to be the new principal deputy DNI) that he had recommended terminating two other satellite collection programs because he believed they could not be successfully completed.
An added complication in building future technical collection systems is the shrinking industrial base that occurred in the 1990s. Secretary of Defense William Perry (1994-1997) had urged defense contractors to consolidate, arguing that there were too many firms competing for declining defense dollars. A period of consolidation followed, with firms either merging or acquiring one another. In the late 1990s it became apparent that there were now actually very few firms left, especially in such high specialty areas as technical collection systems. Thus, in the case of FIA there were only two industrial teams bidding on the contract.
The intelligence budget is also important because it is a major means by which Congress influences and even controls intelligence activities. Congress tended to be supportive of collection requirements throughout the cold war, but it was also inclined to support the disparity between collection and the less-favored processing and exploitation. Some changes in emphasis began to appear in the mid-1990s. The House Intelligence Committee, for example, advocated the use of some smaller imagery satellites, both to have greater flexibility and to save on building and launching costs. This committee also tried to redress the collection and P&E balance, emphasizing the importance of TPEDs (tasking, processing, exploitation, and dissemination). However, the TPEDs problem remains and may grow worse as new collection systems are launched, as they will have increased collection capabilities. Indeed, it has become increasingly difficult to get congressional backing for new collection systems without promising to improve the amount of intelligence that is processed and exploited.
LONG LEAD TIMES. All technical collection systems are extremely complex. They have to be able to collect the desired data, perhaps store it, and then send it to a remote location where it can be processed. All systems have to be rugged enough to endure difficult conditions, whether Earth-bound or space-based, although those in space face more austere challenges. No matter how satisfactory current collection capabilities are, there are several impetuses to build new systems: to improve collection capabilities, to take advantage of new technologies, and to respond to changing intelligence priorities.
The technological challenges alone are daunting and are a significant factor in the time required to build and launch a new system. From the point that a decision is made to acquire such new technology to the actual launch can be as long as ten to fifteen years. Reaching the decision to build a new system involves additional time (sometimes several years) as intelligence agencies and their policy customers debate which intelligence needs should take priority, which technologies should be pursued, and what trade-offs should be made among competing systems in an always constrained budget. Getting congressional approval can also take several years, especially if there is disagreement on which systems should have funding priority. DNI Mike McConnell has expressed his frustration with the satellite acquisition system, comparing the U.S. system with that of Europe, where a satellite can be developed in five years and cost less than $1 billion. But McConnell also admits that U.S. satellites are built to collect against a more diverse set of targets and that there is now a higher degree of risk aversion prevalent in the U.S. system. This last point is important. Collection satellites are extremely complex to build, orbit, and manage, and launching them into a proper orbit really is rocket science. It is interesting to contrast the risk-averse atmosphere that DNI McConnell notes with the early history of U.S. intelligence satellites. According to the NRO, there were twelve CORONA satellite launches in 1959-1960 before the first successful recovery and thirteen before the first image taken in space.
The net result of the lead times involved (not even taking into account the decision time) is that, when a system is launched, its technology may be dated and a whole new set of intelligence priorities may have emerged that the system was not designed to address. There are no shortcuts in system development if a commitment has been made to improving capabilities on a regular basis, which remains the best choice.
COLLECTION SYNERGY. One of the major advantages of having multiple means of collection is that one system or discipline can provide tips or clues that can be used to guide collection by other systems. For major requirements, more than one type of collection is used; the collectors are designed to be cooperative when the system is working correctly. The goal of the U.S. intelligence community is to produce
all-source intelligence,
or fusion intelligence—in other words, intelligence based on as many collection sources as possible to compensate for the shortcomings of each and to profit from their combined strength. Under the 2004 IRTPA, the DNI is responsible for ensuring that “finished intelligence [is] based upon all sources of available intelligence.” This is a somewhat odd provision, akin to a DNI collection seal of approval. It is also ambiguous, as it can be interpreted to mean all sources that should be brought to bear on an issue or all the sources that are available, taking into account other priorities as well. All-source intelligence reflects collection in depth. At the same time, the diverse array allows collection managers to increase collection in breadth, that is, to increase the number of issues being covered, albeit with less depth for a particular issue.
An excellent example of collection synergy is the Cuban Missile Crisis of 1962. Although analysts were slow to understand that Soviet premier Nikita Khrushchev was willing to make such a risky move as deploying medium- and intermediate-range missiles in Cuba, the intelligence community brought a variety of collection means to bear. Anti-Fidel Castro Cubans still on the island provided some of the first reliable evidence that missiles were being deployed. A human source provided the data that targeted the U-2 flights over a trapezoid-shaped area bounded by four towns in western Cuba. Imagery then provided crucial intelligence about the status of the missile sites and the approximate time before completion, as did Soviet technical manuals turned over to the United States by Soviet colonel Oleg Penkovsky, a spy in the employ of the United States and Britain. Imagery and naval units gave the locations of Soviet ships bringing the missiles to the almost-completed sites. Finally, Penkovsky provided the United States with excellent authoritative information on the state of Soviet strategic forces, which indicated overwhelming U.S. superiority.
THE VACUUM CLEANER PROBLEM. Those familiar with U.S. technical collection systems often note that they have more in common with vacuum cleaners than they do with microscopes. In other words, collectors sweep up a great deal of information, within which may be the intelligence being sought. This problem is sometimes also referred to as
wheat versus chaff.
Roberta Wohlstetter, in her classic study
Pearl Harbor: Warning and Decision,
refers to the problem as
noise versus signals,
noting that the signals one wishes to receive and to know are often embedded in a great deal of surrounding noise.
No matter which metaphor one uses, the issue is the same: Technical collection is less than precise. The problem underscores the importance of processing and exploitation.
The issue then becomes how to extract the desired intelligence from the mountain of information. One answer would be to increase the number of analysts who deal with the incoming intelligence, but that raises additional demands on the budget. Another possible response, even less palatable, would be to collect less. But, even then, there would be no assurance that the “wheat” could be found within the smaller volume being collected.
THE PROCESSING AND EXPLOITATION IMBALANCE. A large imbalance exists between the amount of images or signals that are collected and the amount that are processed and exploited. This reflects, in part, the sheer amount that is collected. It also reflects years of budget choices by the intelligence community and Congress that have favored new collection systems over improving P&E capabilities. According to DOD, for example, the National Security Agency (NSA) records 650 million events daily, which culminates in ten thousand reports. Although methodologies are in place to ensure that the most important intelligence is processed and exploited, an important image or message could be overlooked. DOD considered posting all collected intelligence in a single repository and then processing those items selected by analysts. This would, in theory, ensure that only the intelligence that was needed would be processed and analyzed, but it would also increase the burden on analysts to find the intelligence they needed instead of having it sent to them. The Central Intelligence Agency (CIA) is evaluating technology that would automatically examine digital images or video clips to look for details (such as a car) that are the same as those stored in an imagery library. Neither of these suggestions gets at the central issue—that P&E requires more manpower and more funding if it is to have a better chance of getting the necessary intelligence out of the vast amount of information that is collected.