27 June 2002
William Odom is a former Director of the National Security Agency (1985-1988).
Senate Governmental Affairs Committee
A Review of the Relationship Between a
Department of Homeland Security and the Intelligence
Wednesday, June 26, 2002, 9:30am
342 Dirksen Senate Office Building
TESTIMONY BEFORE THE SENATE GOVERNMENT AFFAIRS COMMMITTEE
CREATING A DEPARTMENT OF HOMELAND SECURITY
21 June 2002
by William E. Odom
Thank you, Mr. Chairman, and members of the committee, for inviting me to appear before you today and to share some thoughts on the plan to create a new department of homeland security.
I will address my comments to three points. First, the general question of creating this new department; second, how intelligence collection needs can best be accomplished; third, how best to deal with intelligence analysis in support of this department.
Why a Consolidation of Border and Security Responsibilities is Imperative
The creation of a cabinet level department with full responsibility for control of US borders is long over due. A "border management agency" was actually proposed by the Carter administration but met great resistance within the executive branch at the time. The issue has been periodically raised since then, especially where it concerns drug trafficking. I wrote a short op-ed piece in US News & World Report in 1988, not long after I retired, calling for such a department because from the point of view of providing intelligence to support the war on drugs, I had seen interagency competition and the lack of secure communications create all kinds of difficulties. Too often we had good intelligence but ineffective action because of fragmented arrangements within the federal government charged with border control responsibilities. There were also serious limits on the capacities of local law enforcement officials for using intelligence. At the time, I likened the drug war to a football game in which the drug traffickers were an NFL team and government a division III college team. The fragmentation of responsibilities among agencies at least nine with border responsibilities being spread among five different cabinet departments made it impossible to defend US federal borders effectively.
Several years later, when the National Guard and the Marine Corps began to supplement the Border Patrol along our border with Mexico to deal with illegal immigration, the fragmentation problem was made worse. Now we face the terrorist threat, giving border control the greatest urgency. Thus the arguments for creating this new department concern not just terrorism but also immigration, drugs, and other contraband in trade. I mention these other purposes because they are too often lost in todays public debate on the matter.
I do not see, therefore, how any serious arguments can be made against a major re-organization of the kind the administration is now proposing. It is long overdue. We are living with organizations that date back to the 18th and 19th centuries when the federal bureaucracy grew haphazardly, often in response to parochial political interest before genuine needs. The resulting potpourri of fragmented bureaucracies is not surprising, but it is astonishing that we have failed for the last half a century to modernize the federal organization for dealing with the plethora of dysfunctions caused by the fragmentation.
Those who warn that reorganization will not fix the problems are right. Reorganization alone cannot insure improvement, but it will make it possible. As the organizational arrangements stand today, the problems cannot be fixed. Do the critics prefer to keep it impossible to improve the situation? I was involved in the reorganization that created FEMA during the Carter administration. The same arguments were made, and for several years, various parts of FEMA did not work together very well, but over time, it has developed a respectable record of performance, a record that the pre-FEMA organizations could not have matched.
As a final point in favor of reorganization, changing technologies and changing markets have caused more than a few major business firms in the United States to fail. Others have restructured to accommodate the changes and have prospered as a result. Given the dramatic changes in technology for communications and information management over the past several decades, we should assume a priori that we need a major restructuring of both the bureaucracies controlling US borders and of organizational arrangements for domestic security against terrorist attacks. In other words, the burden should be on those who oppose the new department to prove why it is not needed.
Intelligence Collection to Support Homeland Security
In considering intelligence support to this new department, we should break it into two parts: 1) collection of intelligence and 2) analysis of intelligence for decision makers. I shall comment first on collection, then analysis.
The Intelligence Community has been slowly changing toward a system of national managers for the main intelligence collection disciplines, the three being signals (SIGINT), imagery (IMINT), and human intelligence (HUMINT) collection. A fourth, counterintelligence, also needs national management but remains fragmented beyond anyone's control at the national level.
Many cabinet departments and scores of sub-cabinet agencies now receive intelligence from NSA, NIMA, and CIA's Directorate of Operations. If user agencies maintain properly secured communications and storage facilities as well as cleared personnel to handle SIGINT, they are regularly provided with support. I am not sure how far that kind of distribution system has evolved for IMINT and HUMINT, but the technology for it is has been around for at least two decades. There is no reason, therefore, that the Homeland Security Department cannot receive collected intelligence from all three disciplines directly from the components of the Intelligence Community as it is now constituted. The State Department, for example, depends on this method for meeting its intelligence needs. So do several other departments and agencies.
It is not enough, however, that the SIGINT, IMINT, and HUMINT are delivered to a single office in the headquarters of the Homeland Security Department. Several of its sub-departments and agencies will also need to receive relevant intelligence directly, often on a "time sensitive" basis, for their own analysis and. When the Intelligence Community agencies supply intelligence to a major command within the Department of Defense, for example, the Central Command (CENTCOM), they do not dump it all at CENTCOM headquarters. Much of it is sent directly to tactical units in the field. The new Homeland Security Department will need to develop analogous systems of direct intelligence distribution based on the particular needs of the various organizations within the department. They will have to have secure communications and storage facilities, and they must provide "cleared" personnel to receive and process it at every level of the department where time-sensitive intelligence is used.
To use a familiar metaphor news services the Security Department's various subunits will have to subscribe to intelligence collection services just like customers subscribe to AP, UPI, Reuters, CNN, FOX, and others. The Intelligence Community will need to make more progress in getting the CIA/DO and NIMA to act as news services. NSA is more advanced in this approach for historical and technical reasons, dating back to World War II when distribution of SIGINT (such as "ULTRA") was compartmented and handled in special communications channels. Advances in imagining technologies and broadband communications have made the same approach feasible for IMINT. The technology to distribute HUMINT is simple enough, but the security problems are different. Still, the general approach can be adapted for all three collection disciplines.
Another kind of intelligence collection will be especially important within this new department. It is analogous to what the military calls "reconnaissance" or "tactical" intelligence reporting by non-intelligence units. Rifle platoons, for example, report all kinds of enemy actions and locations as a result of their direct encounters with enemy forces. The results of combat actions immediately become "intelligence" although they are not collected and reported by intelligence personnel. Air force pilots and ship crews at sea do analogous collecting and reporting of enemy activities and capabilities.
In the Homeland Security Department, the same kind of tactical reporting by all subunits and field operators will be extremely important where useful information can be observed. This will be the responsibility of the department's internal intelligence processes. Learning that all intelligence does not come from the Intelligence Community will be critically important for the department's success.
I have not yet mentioned counterintelligence (CI) support, which is terribly important for homeland security. To be clear about CI, it is like ordinary intelligence except that it focuses only on hostile intelligence services, their collection capabilities, agents, knowledge of the United States, etc. Because terrorists have much in common with spies, operating clandestinely, CI must also include counter-terrorism (CT) intelligence, both domestically and abroad. Until CI is better organized within the Intelligence Community, CI support to homeland security will be poor. If a "National CI Service" were created to provide "national management" for all CI and counter-terrorist intelligence removing it from the FBI, then CI/CT could be handled just like SIGINT, IMINT, and HUMINT. Putting the FBI within the new department would not provide better CI/CT, but rather worse. Moreover, it would make CI/CT cooperation with the CIA, Army, Navy, and Air Force virtually impossible.
If such a change were made, relieving the FBI of CI/CT responsibilities, it and all local law enforcement agencies would still need to share and distribute information about all criminal activities that bear on terrorists and their activities. The new National Counterintelligence Service would takeover the task of collecting and producing CI/CT both abroad and within the United States. Coming under the Director of Central Intelligence, it would have the same, perhaps more, oversight than the rest of the Intelligence Community, and its budget and tasking priorities would be provided by the DCI in line with the intelligence requirements that he gathers each year from all agencies within the executive branch needing intelligence support.
Intelligence Analysis for Homeland Security
Thus far I have discussed intelligence collection and distribution. "Analysis" and "production" of so-called "finished" intelligence, or "all-source" intelligence, are different from "collection" of intelligence. Normally done by different people from intelligence "collectors," they consist of the processes of integrating intelligence, making sense of it, and using it to answer questions that users of intelligence have for their particular operational and policy goals. I emphasize this difference between collection and analysis because the public debate about intelligence support for the new Homeland Security Department often confuses the two as one and the same. They are not.
The new department must have its own analysis and production capabilities. They will have to give their special collection requirements to the Intelligence Community agencies according to the DCI's guidance, and those agencies must collect and sort information to answer the requirements. The Homeland Security intelligence analysts will have to put it all together from the various reports supplied by the collection agencies. Because these analysts will be within the Homeland Security Department, they will be properly located to know its intelligence needs and how to shape them, both to steer collection efforts and to produce answers that are timely and truly needed.
Within the department, there will undoubtedly have to be a decentralized system of analysis in some cases, centralized in others. With modern communications, however, the mix of both approaches is easy to establish. For example, central data bases can be built, allowing dispersed access and use in decentralized analysis units.
What I am suggesting here, of course, is the intelligence analysis model found in the military services. Every operational command has a staff intelligence section, which can request and receive SIGINT, HUMINT, IMINT, and CI. This intelligence staff section is responsible to the operations staff section for finished intelligence analysis. At the joint level, this section is known as the J-2 and the J-3 is the operations section. At the corps and division levels, we find the G-2 supporting the G-3, and so on, down to S-2s and S-3s in brigades and battalions. In the air force and navy, analogous staff intelligence sections are found with different names.
We can call this the "distributed processing" model of intelligence as opposed to the "central processing" model. The latter model is analogous to a big mainframe computer that serves many users who have only "dumb" terminals to provide access to the central processor. The distributed processing model is analogous to small microprocessors in PCs and laptop computers located in many analytic units, widely dispersed with different needs and demands, performing analysis tailored for local use. The distributed processing model removes the queues, the lines in which users may have to wait for responses. And it emphasizes particularizing analysis for the familiar user immediately at hand, not a faceless one far away. Only analysis done locally, where the intelligence officers are in immediate contact with operators who use intelligence to make decisions, can accurately perceive what kinds of intelligence analysis is useful. A large intelligence analysis center far away from the users will never have the sensitivity to local operations to provide effective support. This has always been the weakness of much of the all-source intelligence produced by CIA/DI and DIA. Both can produce useful products for some purposes, but they cannot provide comprehensive and "time sensitive" support to all users within a single department, much less the entire government.
When we understand the difference between 1) intelligence collection and 2) intelligence analysis and production, it becomes clear why it makes no sense to put collection agencies from the Intelligence Community inside of the Homeland Security Department. That will makes things worse, not better, for intelligence support, unless the Intelligence Community agencies work only for that department. In that case, additional and similar agencies would have to be created to prove support to the military services, the State Department, and dozens of other executive branch intelligence users.
Another kind of problem arises, however. How do the intelligence collectors SIGINT, IMINT, HUMINT, and CI decide whom to give priority? The Defense Department? The State Department? Homeland Security? Treasury? The President has to set the priorities, and his executive for doing that is the Director of Central Intelligence.
Like any other resource, intelligence is costly and scarce. All users cannot have all they want all the time. Their requirements have to be prioritized, giving some preference over others as situations and needs change. This prioritization task is not new. The Defense Department has dealt with it for a long time, quite successfully in most cases. Among departments, especially its largest users, State and Defense, the DCI has also developed a reasonably effective record of handling the prioritization task according to the demands of the department secretaries and the Joint Chiefs of Staff. The new Homeland Security Department should look to that experience in handling its own internal priorities and learning how to get the most from the Intelligence Community's collection agencies and centralized analysis and data bases.
The complexities surrounding both the practical matter of creating a new Homeland Security Department and of providing it with adequate intelligence support are enormous. In my remarks I have merely sketched the broad outlines, but I hope that they explain these conclusions:
First, the wisdom of creating the new department is beyond doubt. It will not be easy, and mistakes will be made in the process, but they can be ironed out by trial and error. This reorganization process is absolutely essential if there is to be any serious improvement in the control of our borders and in defense against terrorist operations within the United States.
Second, terrorism is not the only reason for creating this new department. It has long been needed for dealing with immigration, drugs, illegal trade, and several lesser matters. If it is properly organized and tasked, it can help reduce the rising transaction costs these issues are placing on the US economy.
Third, intelligence support for the new department is not a matter of putting the FBI, CIA, or some other intelligence organization within its domain. It is a matter of creating its own intelligence analysis elements and insuring that they have access to the major collection capabilities in the Intelligence Community. That approach works well for the Intelligence Community's support to many cabinet departments and lesser agencies today.
There is one exception. The Intelligence Community is poorly organized to provide "counterintelligence" (CI). CI must also take the lead in providing counter-terrorism intelligence (CT) as well. And its collection and production cannot be the responsibility of criminal law enforcement agencies, not just in the case of the FBI but also in the military services. As the Intelligence Community is now organized, it simply cannot provide a comprehensive CI and CT picture to anyone, not to mention the Homeland Security Department.
Thank you for your attention, and I am prepared to answer questions.