At age seventy, the National Security Act of 1947 is now well past the normal retirement age, but, like many other Baby Boomers, it is still working full-time. As the Trump revolution unfolds should we expect to see it junked? Even before Trump there were calls to update the fundamental statutory basis for national security decision and implementation, as young up and comers have looked impatiently at the seemingly outmoded attitudes and waning capacities of elders they wish would step aside. The urge to replace is natural for those who focus on blunders in the substance of national security policy and assume results would have been better if the process of policymaking had been different. Or it seems natural to frustrated toilers inside the Beltway as they focus on barriers to accomplishing the specific priorities of interest to them among the complex mix of purposes in national security.

If we look beyond the intuition stoked by immediate dissatisfactions, the case for major change is unclear. Dissatisfaction is largely about the speed and results of the existing structure and process. Speed is blunted in large part by the conflict of complex competing interests and the checks and balances that protect them. This is the price of democratic government in general. Results, in turn, depend on the wisdom of incumbents who act through the existing process. At some level judgment about the success or failure of institutions is not independent of the policy that emerges from them, but designers can only aim for structures and processes that will enable a proper process or decision; they cannot select the individuals who accede to power over time and inhabit the structures. Office holders come and go, but serviceable institutions must endure for generations. Successful institutions set and bound the conditions in which government operates and provide incentives and opportunities for incumbents to do their best. Institutions must work despite wide variation over time in officials’ ideologies, experience, and sensibility, and through tidal changes in the political, economic, and social conditions that surround them.

A bad structure may hamper the good designs of wise incumbents, but a good structure cannot prevent bad initiatives from foolish ones. In short, it is important to focus on the difference between institutions and outcomes, or, put a bit more specifically, the difference between a design that optimizes one purpose and a design that “satisfices” across the board when many purposes conflict. Fixing particular problems opens the door to other problems.

There are two main managerial issues to be handled over time, horizontal and vertical. The horizontal one is when to add new organs as technology changes and new functions or policy problem areas emerge. In principle, this should also mean when to retire obsolete organs, although this hardly ever happens, so the complexity of the system tends to grow. The vertical issue is where the balance should be struck between the benefits of centralization and decentralization of authority.

Among the tests of how well an institutional system manages these adaptations are whether it works to reveal issues requiring consideration and decision, provides relevant information and options to policymakers, and produces clear and deliberate actions rather than accidental or unrecognized ones. A system must enable these processes even if it cannot force a President and his lieutenants to take proper advantage of them—a big issue in the evolving Trump era. Enabling requires some balance between organizational pluralism and decentralization on one hand, to represent relevant concerns, and integration and centralization on the other, to impose direction and coherence on choices. The National Security Act and the several amendments and additions to it have accomplished these aims imperfectly, but better than the less-developed system before 1947, and probably as well on balance as a system that might be designed differently. In the United States since 1947 failures in national security policy have been due far more to high politics and misjudgments than to the managerial apparatus established by the Act. Process can be a problem, but it has not been the problem.

When attention focuses on the difference between institutions and results, and one considers whatever else is on offer, the National Security Act still looks like a solid foundation. This does not mean that no changes are needed. New problems can sometimes prove revolutionary—for example, the rapid emergence of cyber security as a fundamental priority to all elements of government and the civilian economy. The question is whether needed adaptations cannot be thoroughly accomplished within the limits of existing legislation. No structure remains perfectly suited for a mission over long spans of evolving conditions and challenges, and any successful one must evolve along with its context. But the institutions designed in 1947 and the 1949 amendments to the Act, tweaked occasionally since, have proved remarkably able to perform their functions through changing times.

Strategic Revolution, Institutional Evolution

What was the impetus for the Act? Consensus on peacetime American activism in the world, and a large standing military establishment, is now deeply rooted in American politics but in 1947 it was new. Within government the National Security Act gave support to the direction in which the United States had been firmly headed since Pearl Harbor. The idea of isolationism had always been exaggerated anyway: The United States was deeply involved in Asia since 1898 (or even 1864 if we count Commodore Perry’s “opening” of Japan) and in Latin America even earlier; isolationism applied in effect only to abstention from actively navigating the European balance of power. The events of the 1930s and subsequent global catastrophe discredited that abstention and World War II left Washington with dramatically expanded political, economic, and military roles, both outside and inside the country.

On the outside, the war left two revolutionary changes in world order: economic unipolarity and politico-military bipolarity. With the rest of the West in ruins, both revolutions seemed to demand American leadership. On the inside, all of American government had grown rapidly, first in the New Deal, then in the mobilization for war. Both outside and inside the country these expansions naturally generated efforts to organize and integrate the proliferation of participants and mechanisms of policymaking and implementation. Outside U.S. borders the point was to unite allies in reliable and effective collaboration well before any next war, to avoid, in short, the mistakes of the 1930s. Inside U.S. borders the point was to bring coherence to a complex array of organizations that had emerged haphazardly and grown quickly.

On the inside, an anti-statist tradition remained an undercurrent and constrained some aspects of Cold War mobilization.¹ The domestic consensus for sustaining muscular government across the board, however, lasted nearly half a century, from Roosevelt until Reagan. Although that consensus has faltered since, it is useful to remember that none of the three Republican Presidents between 1945 and 1981 challenged the main institutions of the New Deal. In any case the consensus (or to be accurate, the elite consensus) on international activism, and thus for maintaining robust national security institutions, has lasted longer than the public’s favorable view of government in general. Civilian political leaders are rated low in public opinion surveys but the military remains more respected and popular than almost all other domestic institutions. Commitment to forward-leaning foreign policy has been stronger among the American elite than in mass opinion, but it endures, having been able so far to brush off political challenge from losers such as Robert Taft, George McGovern, Pat Buchanan, Ralph Nader, Ron and Rand Paul, and even the one nominee of a major party (Donald Trump) who tilted erratically in the direction of severe retrenchment. The domestic reflection of the global mission at its dawn was the reshaping of the policymaking system in the 1947 Act.

The principal early efforts to organize and integrate in the outside world were the United Nations (UN) and North Atlantic Treaty Organization (NATO), multinational organizations led by the United States. Many had high hopes for the United Nations at its birth and assumed it would play an important material role in peace enforcement. In an exclamation in 1950 that now seems quaintly naive after more than six decades of international violence, Secretary General Trygve Lie responded to Assistant Secretary of State John Hickerson’s news that North Korea had invaded the South in June 1950 by blurting, “My God, Jack, that’s a violation of the United Nations Charter!”² The UN’s utility, however, has remained primarily diplomatic, while NATO in Europe and bilateral alliances elsewhere have been the effective instruments for prevention and prosecution of America’s wars.

In the late 1940s American economic hegemony, represented in the Marshall Plan and its bureaucratic embodiment, the Economic Cooperation Administration (ECA), underwrote NATO’s political and military counterweight to the threat from Moscow. The USSR was ravaged by the war, but left with military forces controlling half of Europe and considerable ideological influence, via Communist parties, within most of the countries of the other half. As time went on the overwhelming American economic dominance eroded as Europe, Japan, and the Soviet Union recovered, but bipolarity endured and intensified—politically, as electoral challenges from western Communist parties were blunted and the two blocs solidified, and militarily, as Moscow developed nuclear weapons and the peacetime fielding of large conventional forces on both sides continued.

NATO was the main event in external integration. As Robert Osgood emphasized more than a half-century ago, it was an unprecedented alliance in two respects. First, it was a peacetime commitment by the United States to immediate involvement in a European war—the “entangling alliance” against which Washington’s Farewell Address had warned, but which the lessons of the 1930s were seen to require. Along with the Marshall Plan that preceded it, the North Atlantic Treaty marked the Republican Party’s enlistment in the Cold War consensus, as Senator Arthur Vandenberg’s support trumped Robert Taft’s opposition.

Osgood’s second emphasis remains NATO’s most significant aspect today: “NATO is unique among peacetime alliances. In fact, it has assumed tasks of multinational planning, decision, and action that few wartime alliances have performed.”³ It evolved quickly from a simple guaranty pact into a highly integrated transnational institution. As with any alliance, members pledged to come to each other’s defense in event of war (although the limits of this Article 5 promise have traditionally been overlooked).⁴ NATO went further by establishing joint command structures, mechanisms, and peacetime deployments, as well as a formal North Atlantic Council (NAC) for regular consultation and deliberation. Institutionalization underwrote the presumption of automaticity in the mutual commitment. In the early days some, especially Dwight Eisenhower, thought the peacetime forward presence of U.S. forces in the integrated defense plan would be temporary as Western Europe recovered and created, in effect, tripolarity.⁵ Instead, permanent peacetime deployment was assumed after the 1950s as Washington pushed the strategic concept of flexible response on the allies.

Fatefully, NATO’s unique institutionalization was not replicated in Asia. There was no NEATO for Northeast Asia. In principle there was multinational military action in Korea, via the UN Command, but in practice, especially after the 1953 truce, deterrence in Korea was a bilateral affair between Washington and Seoul. Starting in 1955, with the French withdrawal from Indochina, there was a Southeast Asia Treaty Organization (SEATO), but it never developed beyond a hollow shell. Indeed, it was all but still born, although it did take twenty years to die officially after proving irrelevant to the main conflict of its time, the second Indochina war.

In one sense there was less need in Asia for the unprecedented military integration of NATO because the main asset to be defended, Japan, was protected by a water barrier, unlike Germany and France, and the geography of the Korean peninsula limited the conventional force that Communist powers could bring to bear on that short front. The other military challenges in Asia were primarily revolutionary movements, as in Vietnam, against which maximizing a conventional force seemed not to be the main problem. The absence of a clearly mobilized and articulated multinational deterrent, however, did enable North Korean miscalculation in 1950. Today the absence of a multinational security organization also reflects the continuing hesitancy of countries in the region to commit to united deterrence of Chinese expansion.

Postwar demobilization proceeded even as the 1947 Act was passed—conscription lapsed in that year, although it was renewed when the Berlin Blockade began in 1948—but much of the organization of U.S. military forces to backstop bipolarity was already in place or nascent by the time of the 1947 Act. Forerunners of the system of unified and specified commands that came to implement the global sweep of U.S. military operations in the Cold War grew out of the preceding war and subsequent occupations: the army’s European Command and Far East Command, and the Strategic Air Command formed in the Army Air Force the year before the Act made the service independent. The Joint Chiefs of Staff (JCS), which had grown out of the Joint Board in early 1942, became permanent. Similarly, modest centralization of foreign intelligence functions already occurred with the Central Intelligence Group (CIG) before the Act made it into the Central Intelligence Agency (CIA). Many postwar planners hoped for fundamental changes that would truly unify the armed services and consolidate organizations, but resistance (especially from the navy) kept the final legislation from going that far.⁶ For the institutions to carry out U.S. policy, the National Security Act clarified, codified, and extended the institutional evolution that originated in World War II.

The bumper sticker for the main effects of the 1947 Act would be “coordinating pluralism.” Some officials frustrated by the complexity of overlapping and independent units involved in planning and operations during World War II hoped for dramatic consolidation of functions. After the pulling and hauling that produced the Act this did not happen; duplication and overlaps largely survived, although they became subject to higher level oversight, and over time, some greater discipline. Three new organizations, however, were crucial management innovations. Establishment of the Department of Defense (DoD) did not merge the separate services, but it did create a civilian hierarchy that could arbitrate, choose, or direct among the services more effectively than before.⁷ Establishment of the Director of Central Intelligence (DCI) did not merge the intelligence functions of the military services’ and State Department’s intelligence units, but it did create more coherent mechanisms for getting information and analysis to the top. Establishment of the National Security Council (NSC) did not streamline the bureaucracies, but did formalize and regularize the process of representing them in ultimate decisions.

Organizing for Decision and Direction

For developing policy the 1947 act created novel organs to accomplish coordination that had proved inadequate during the war. The Department of Defense, the biggest and most important of the organizational innovations, has persisted as a giant conglomerate combining large and diverse military forces, technological research and development, complex and expensive procurement programs, and a dense corps of civilian managers. Over time the crucial evolutions have been the growth and clout of staff in agencies of the Office of the Secretary of Defense (OSD), the elaboration and thickening of the system of peacetime combatant commands worldwide (known as “CINCs” until Donald Rumsfeld changed the designation), and a major ratchet in unification of the armed services through a strengthened JCS since the Goldwater-Nichols reform.

The second major institutional creation of the Act was the CIA. The CIA did not take control of separate pre-existing intelligence services, but limited centralization did occur under the CIA Director’s other hat, the more senior position of Director of Central Intelligence (DCI). This second more important hat was consistently misunderstood outside the intelligence community (IC).

One institutional change that was more radical than evolutionary was the warrant for covert political operations abroad that was given to CIA. Even this grew out of wartime—the Office of Strategic Services (OSS)—but it heralded the routinization of a function never before common for the United States in peacetime. Covert action was at best only vaguely authorized in the 1947 legislation, which directed the agency “to perform such other functions and duties related to intelligence affecting the national security as the National Security Council may from time to time direct.”⁸ (More specific authorization for covert action came from the council later in the year via directive NSC 4A and, in 1948, NSC 10/2). At the time, DCI Roscoe Hillenkoetter resisted taking responsibility for the function, which is really a matter of policy execution rather than intelligence collection, but the CIA got stuck with it when Secretary of State George Marshall refused to let the State Department take it.⁹

As the small group of the most senior policymakers, the National Security Council (NSC) was to preside over consideration of issues that cut across departmental concerns. The closest thing to the NSC under the old system was the State-War-Navy Coordinating Committee (SWNCC), but that did not operate at the highest level, as the NSC would—in effect, as a subset of the cabinet.¹⁰ The National Security Act itself also mandated the Council “to advise the President with respect to the integration of domestic, foreign, and military policies”—in other words to engage in statecraft in the original and still proper meaning of the term—and “to assess and appraise the objectives, commitments, and risks of the United States in relation to our actual and potential military power.”¹¹ Presidents Truman and Eisenhower added the Secretary of the Treasury to the normal attendees at NSC meetings to help integrate consideration of domestic and military policies, but for most of the past half-century the NSC has not been known as a forum for considering fiscal or other elements of domestic policy, despite their clear relevance to security policy. More recent suggestions to make the Secretary of the Treasury a statutory memory of the NSC have not been taken up. Instead, even as NSC activity in foreign affairs expanded, a separate entity was created to deal with the most essential domain of national security: the Homeland Security Council.

Hardly anyone now remembers, but the original idea for the NSC differed from what most see as its role today. In the mind of James Forrestal, the first Secretary of Defense, who had chafed under what he saw as Franklin Roosevelt’s freewheeling initiatives during the war, the NSC would regularize consultation and constrain and discipline the President. As time passed it would evolve into more of the reverse: an enforcement arm of the President over the other relevant Executive Branch departments.

In the policy process the council proper (meaning the statutory members),¹² the group meant to confer with and influence the President, is not what many think of anymore when they hear “National Security Council.” Most think instead of the NSC staff, which began as primarily a paper-shuffling conduit from the departments but evolved into the President’s own staff for foreign policy. For a while the Obama Administration even took the word “Council” out of the name for the unit, calling it simply the National Security Staff (NSS), implicitly a purely White House organ.

The term “National Security Advisor” did not come into usage until the 1970s and was not in the authorizing legislation, which specified only an executive secretary to direct the staff. This position evolved into Special Assistant (and under Nixon, simply Assistant to the President) for National Security Affairs. (In the rank inflation of recent times, the bulked up staff came to include numerous “Special Assistants” to the President.) All this reflects the extent to which the NSC staff and its head originated as managers of departmental participation in policy formulation but developed instead into policy players in their own right—a development that became controversial with Henry Kissinger and has been periodically so ever since.¹³ The assertive role of the National Security Advisor recedes at some points, especially as it did twice under Brent Scowcroft, but tends to return.

The controversy over whether the NSC staff should shape policy or simply coordinate it went with widening of the staff’s purview. As the NSC evolved it became a forum for discussing all of foreign policy rather than focusing on aspects related to war and peace as originally envisioned. The staff became a collection of mini-departments paralleling the bigger bureaucracies in Defense, State, Treasury, and the intelligence community. From a number around 20 in the early 1960s the NSC’s professional staff grew to about 100 in the second George W. Bush Administration.¹⁴ Under Obama its size reportedly more than doubled, provoking wide resentment within the departments for micromanagement.

How Well Has the System Worked?

The system operates for whoever runs it, and cannot always compensate for other determinants of results: the balance of power in the outside world that enables or prevents the United States from working its will; the limitations of information available to the system, whoever may be in charge; and especially the vagaries of personality and incidence of bad judgment by the individual office holders due to ignorance, ideology, recklessness, vanity, wishful thinking, reticence, or other subjective factors and human frailties.

For example, reticence, encouraged by the awesomeness of presence at Olympus, is an often underestimated subverter of formal process, and sometimes resembles the groupthink syndrome famously elucidated by Irving Janis. Chester Cooper, who had doubts about the war in Vietnam, recounts:

The President, in due course would announce his decision and then poll everyone in the room—[National Security] Council members, their assistants, and members of the White House and NSC staffs. . . . “Mr. X, do you agree?” “I agree, Mr. President.” During the process I would frequently fall into a Walter Mitty-like fantasy: When my turn came I would rise to my feet slowly, look around the room and then directly at the President, and say very quietly, “Mr. President, gentlemen, I most definitely do not agree.” But I was removed from my trance when I heard the President’s voice saying, “Mr. Cooper, do you agree?” And out would come a “Yes, Mr. President, I agree.”¹⁵

This subjective problem is not susceptible to fixing through legislation. At the same time that Cooper admitted his own personal human failure, he also demonstrated that the official formal products mandated by the NSC system—National Intelligence Estimates (NIEs) on Vietnam— were for the most part remarkably accurate.¹⁶

It is reasonable to assume that U.S. national security policy during the Cold War would have been worse in the absence of the integrating reforms of the 1947 Act, because the underdeveloped system of the pre-1940 era couldn’t handle the load. That earlier system simply afforded less availability of information, fewer specialized organs, and inadequate top-level coordination of the departments. No one suggests returning to the less articulated and less centralized system of a century ago. In the course of the past seventy years, moreover, needed additions and expansions could be accomplished through simple Executive Order or specific legislative authorizations that did not reshape the overall structure—for example, creation of the National Security Agency (NSA) in 1952, the Defense Intelligence Agency (DIA) and Arms Control and Disarmament Agency (ACDA) in 1961, or the replacement of the Atomic Energy Commission (AEC) by the Nuclear Regulatory Commission (NRC) and Energy Research and Development Administration (ERDA) in 1974.¹⁷ It is hard to see how a system of enhanced coordination that is markedly different in form from what has evolved since 1947 would necessarily have made policy better. Among the failures in Cold War policy, which would have been reversed if organizations different from DoD, CIA, and NSC had been created?

It would be hard to ascribe much of the responsibility for the two biggest disasters in the time since the 1947 Act—the Vietnam War and the second war against Iraq—to the institutions in place from the Act and its additions. There were many fateful mistakes in deciding for both wars and prosecuting them, but few that could be attributed to the faulty design of policymaking organs. Contrary to common assumptions, American leaders did not stumble into Vietnam with no idea of what they were getting into; the commitment was due to the conviction at the highest level that there was no acceptable alternative to preventing a Communist victory.¹⁸

As for Iraq, organizational dynamics of the legislatively formalized system did not push Bush and company to decide for war, because the incumbents did not take full advantage of the system. Indeed, there apparently was never any formal examination of the question of whether to go to war or any formal NSC meeting to debate the decision. Legislation can establish structure but it cannot enforce process.

President Trump would do well to contemplate the lesson from this case: policymakers ignored or overrode the checks and balances on their judgment that the system afforded. Secretary of Defense Donald Rumsfeld was well aware that most professionals in the U.S. Army thought the lean invasion force he wanted would be too small to stabilize the country, and he was well aware that the State Department was pushing more planning for the aftermath of combat. He suppressed the first because he wanted to show that the “Revolution in Military Affairs” enabled military success with lean forces, and suppressed the second in an unrealistic attempt to avert postwar entanglement in the country. These were decisions made not because of the organization of the policymaking process but because of Rumsfeld’s personal judgment and confidence. In the same vein, Secretary of State Colin Powell’s reservations about the plan for war were not suppressed by the apparatus of decision-making but were soft-pedaled by Powell himself and, such as they were, simply rejected by President Bush in return for agreeing to support Powell’s effort to gather more multilateral diplomatic license for the war.

A partial exception to exoneration of process was the October 2002 National Intelligence Estimate (NIE) on Iraq’s weapons of mass destruction. Although it did not cause the war, since Bush’s decision preceded the estimate, it did provide the excuse for it, when a more accurate estimate would have made it difficult to sell the decision for war to Congress and the public. But even here the main problem was the individual judgment of intelligence managers about how to package the presentation of the analysis more than the information or analysis itself that came out of the system. The actual evidence behind the estimate turned out to be meager, which might be an indictment of the quality of the institutions responsible for collecting intelligence, but on balance it is fair to say that the estimate’s conclusion was wrong for the right reason.¹⁹ It is very difficult psychologically and politically to differentiate judgment of process and results. Reviewing the criticism of the 2002 policy process by the Senate Select Committee on Intelligence Robert Jervis points out:

We like to think that bad outcomes are to be explained by bad processes, but this needs to be demonstrated rather than assumed. . . . If I am right that although some serious (and correctable) errors were made, the processes were not as flawed as the [Senate] reports claim, this is good news in that the system was not horribly broken, but bad news in that there are few fixes that will produce more than marginal (but still significant) improvements. . . . We need to avoid equating being wrong with having made avoidable and blameworthy errors. SSCI [the Senate Select Committee on Intelligence] in particular falls into this trap, as is shown by the fact that it almost always equates reasonable, well-grounded inferences with those that proved to be correct. We can see the problem by asking the obvious counterfactual: would the same report have been written if the estimates had turned out to be correct? This is implausible, yet it is what SSCI implies.²⁰

It is not clear that, before the decision to assault Iraq, intelligence collection would have been improved by any difference in the structure of the intelligence community that could have been established by legislation, which must design the community for all seasons.

Should We Change the System?

If the structure of the national security policy system is out of date, how is it out of date? The main points of the redesign of the system in 1947 were consolidation, centralization, and coordination. The first of these has not been a persistent priority. Outside generalists usually promote efforts to streamline functions they consider redundant, but this often leads to piling a new coordinating mechanism on top of the old units rather than eliminating them. (For example, Secretary of Defense Robert McNamara had hoped that the creation of DIA could absorb most of the three service intelligence agencies, but the latter regenerated and remained about as large as ever.) Continued interest in centralization and coordination has been more prominent.

The NSC staff was barely foreseen in the National Security Act, emerged gradually as a coordinating unit, and has drifted fitfully toward the centralizing mission. This trend has probably gone too far in recent times with the radical expansion of the NSC staff. Nixon and Kissinger’s extreme centralization of foreign policy in the White House was controversial long ago, but it was accomplished more by ignoring or circumventing the professional bureaucracies in State and Defense than by duplicating them. It is hard to see how an NSC staff now numbering in the hundreds is not a parallel bureaucracy. If reducing the staff is desirable, however, this should ideally be done by executive direction within existing legislation; a hard and fast numerical cap in law could prove problematic under unforeseeable future circumstances.

The original impetus for the 1947 Act was unification of the armed services and greater integration remained a continuing concern long after its passage. It is hard to make the case, however, for significant further moves in that direction. The Goldwater-Nichols reforms late in the Cold War accomplished nearly all the additional integration that should be desired. There are some changes at the margin that might still make sense—for example, merging the several separate service war colleges into the National War College, an option considered at the time of the original act but rejected because of service resistance. A similar case was made for the merger of the three nuclear weapons laboratories after the Cold War, and it was rejected then, too. These two particular consolidations would be incidental for defense policy today, and it is uncertain that reducing the variety of educational institutions would be desirable anyway. When recommended economization is a miniscule portion of defense spending, vested interests and their rationales, which may even be correct, trump the recommendation.

Consolidation of military services was a principal goal of the 1947 act and its amendments. What about the possible need for new additional services? Important changes in the milieu of military action have arisen since 1947. One is the use of outer space for intelligence and combat functions. Going back to the 1990s some in the Air Force and elsewhere thought that the nature of new space operations was radically different from traditional operations on land, sea, and in air and that a separate Space Force was the logical adaptation to the military results of scientific progress. Support for such radical change did not grow, however, and Space Command with the U.S. Air Force appears to be performing adequately.

There is one issue area that, intuition suggests, should logically require major reorganization: cybersecurity. First, in a short time cyber concerns have gone from incidental matters to fundamental challenges for the economic prosperity and politico-military security—indeed, potentially, the survival—of contemporary civilization. Both domestic and foreign policies are inevitably putting a far higher premium on dealing with emerging problems in cyberspace.

Second, this whole realm of technology and society’s utter dependence on it was completely unforeseen in the National Security Act or any of the legislative adjustments of it. As with intelligence, all departments and government functions have a stake in cybersecurity, so centralization in a single new agency would not remove the other departments from involvement. Nevertheless, the position of the National Security Agency and its twin, Cybercommand, as the dominant institutions in the business may not be the optimal organizational solution to the question. As yet there is no vigorous movement or coherent rationale being promoted for any radical reorganization of units and functions concerned with cyber conflict, but change of some sort—whether it be a central agency for cybersecurity of something else—is one of the few obvious candidates for consideration.

Otherwise, sensible proposals for change in the national security system should be evolutionary rather than revolutionary. All sensible observers realize that in practice any system will fall down sometimes, if only because officials who design and operate it are fallible. Some, however, believe that in principle all good things go together and hence that policy could be more or less perfected if the ideal system for organizing interests, priorities, and consideration of options could be discovered and engineered. This is an illusion, because in real life all good things in any general category of human endeavor hardly ever go together. All too often optimizing one function detracts from another. Consolidation reduces inefficiency but penalizes concerns whose representative units are merged or streamlined. Centralization improves consistency of purpose but reduces checks and balances. Checks and balances reduce the odds of precipitous action and mistakes from zealotry or ignorance, but also erect obstacles to innovation, initiative, speed, and reform. In these tradeoffs reorganizations that tilt in favor of one interest against another eventually are questioned as mistakes due to shortchanging the first interest emerge.

Does Change Always Help?

Most mechanisms or norms have two edges. For example, should military leaders simply salute and keep their mouths shut when civilian leaders direct them to undertake actions they believe will fail, or should they be expected to let Congress, or even the press and public, know their professional judgment and disagreement? Those concerned with keeping the military in their proper place to protect democracy from men on horseback tend to believe the former, others concerned with maximizing awareness of representatives and citizens in a democracy may want the latter. Civilian officials in the Obama Administration reportedly felt undermined by leaks from military leaders when making decisions about level of effort in Afghanistan.²¹ But was the behavior of the Joint Chiefs of Staff in 1965 preferable, when their doubt that Johnson Administration plans for war in Vietnam would succeed were not made known to the public?²²

Basic reforms of the system are, and should be, rare. There are significant short-term costs in disruption of work and confusion of results when reorganization changes missions and lines of authority, so the longer the term between necessary reorganizations, and the more limited the necessary changes, the better. Depending on what is counted there have been numerous reorganizations but only a few major structural changes in the national security policy system. For seven decades since the National Security Act one might cite only the 1949 amendments (which for practical purposes might be lumped with the 1947 Act), the 1958 defense reorganization, the 1986 Goldwater-Nichols legislation, creation of the Department of Homeland Security in 2002, and the Intelligence Reform and Terrorism Protection Act (IRTPA) of 2004. Establishment of permanent intelligence oversight committees in Congress in the mid-1970s was comparable in importance, but that reform was not related to the Executive Branch institutions established by the 1947 Act.

None of this means other major changes of practice have not occurred. For example, the size and power of agencies in the Office of the Secretary of Defense are radically greater than in the 1950s. The burgeoning of OSD provoked resistance in the military and criticism in Congress, especially as Secretary of Defense McNamara used the Office of Systems Analysis to challenge positions of the services and JCS. The evolution of OSD, however, was immanent in the establishment of DoD by the 1949 amendments.

The few major reorganizations since 1947 have smoothed out lines of authority and increased integration. On balance the benefits probably dominate, but there are also costs to such progress. Centralization increases the capacity for quick and decisive action. Pluralism or redundancy retards that capacity, but it also reduces errors due to submerging particular perspectives or risks. The complexities and controversies of national security issues preclude clear verdicts on how much major reorganizations improve policy outcomes, or which if any should be judged ineffective or counterproductive.

To reflect on the balance of risks in reform consider one that might be a case where benefits do not outweigh costs: the 2004 IRTPA. Its main provisions were the legal requirement to break down “stovepipes” and maximize the sharing of intelligence among the various agencies of the IC, and the establishment of a new Director of National Intelligence (DNI), both to avert failures of coordination seen as contributing to the September 11, 2001 surprise.

The first, improved sharing of intelligence, has a second edge, which seemed less apparent in the aftermath of September 11 but which comes back periodically: heightened risk of breaches in secrecy. The 9/11 Commission and legislators in 2004 focused on the breakdown in collaboration between CIA and the Federal Bureau of Investigation (FBI) before September 11. They did not focus on the danger represented by the Aldrich Ames espionage scandal of a few years earlier—the hemorrhage of sensitive secrets to Russia, which had prompted complaints in Congress about having allowed one individual such wide access to sensitive information. A decade after the IRTPA the second edge came back again with Bradley/Chelsea Manning’s wikileaks and, yet again, with Edward Snowden’s revelation of sensitive intelligence collection activities of the National Security Agency (NSA). These incidents create natural pressures for reversion to more compartmentation in order to limit damage when the custodians of secrecy go bad.

The second edge of the DNI reform has been the introduction of a new layer of bureaucracy—the unit known as Office of the Director of National Intelligence (ODNI)— in an intelligence system that was already beset by bloat, and in which the function of the DNI already existed in the role of the DCI.²³ The DNI issue must be understood in the context of the prevalent misunderstanding mentioned earlier: Almost no one among journalists, public commentators, and even anyone in government outside the intelligence community ever understood that the old DCI, the office established by the 1947 Act, could have accomplished what the DNI was designed to do if only his titular responsibilities and budget authority had been properly aligned. That is because few understood that the DCI was a two-hatted job. Except among professionals in the IC he was invariably referred to as “director of the CIA,” never by his proper title of Director of Central Intelligence, which covered the additional responsibility of coordinating the rest of the intelligence community. The higher coordination function was not accomplished as thoroughly as it might have been because, as already suggested, the DCI’s responsibility was never matched by sufficient authority due to resistance from the Defense Department. The seldom-recognized fact remains that the old DCI did not accomplish everything the new DNI was designed to do because the President did not overrule the other departments in order to invest the DCI with more direct control of their intelligence functions. As it turned out, the new DNI also wound up circumscribed and limited in authority for some of the same reasons that the old DCI had been. More decisive centralization would probably require making the DNI into the cabinet Secretary of Intelligence.

So an alleged benefit of the 2004 IRTPA was the separation of the old DCI’s two jobs. Many believed that combining the CIA director and community coordinating functions in one person (the old DCI) was too big a job, or that it represented a conflict of interest, privileging CIA over other agencies in the community. These arguments were simply wrong. First, the Secretary of Defense has a far larger and more diverse set of responsibilities than the DNI, yet there is no lobby arguing to break that position up. Indeed, the point of creating the Secretary of Defense in the first place was to establish unitary control over disparate organizational leviathans.

Second, if the Central Intelligence Agency established in 1947 was not supposed to be truly central, first above equals rather than among them, what was the point of the agency’s name? The cost to separating the jobs is that the top coordinator’s ability to implement centralizing missions is limited by loss of his own troops in CIA, the practical assets for shepherding initiatives when the other departments succeed in resisting allocation of enough personnel, or their best, to the DNI. The separation of the two hats has probably encouraged some of the expansion in the ODNI, which comes at the cost of adding further bureaucratic complexity. Of course there are benefits from the specifics of the DNI legislation, but there is no consensus in the intelligence community that they outweigh the costs of more bureaucratization and layering.

Whether or how the system that has evolved under the National Security Act should be changed substantially depends on many considerations, but more than any other one thing, on opinions about the relative costs and benefits of centralization and decentralization. Such opinions vary with economic ideology and policymaking experience. In contrast to much else in national policy in our prolonged season of polarization, opinions on these managerial questions do not vary as obviously or consistently with party or ideological divisions. On other matters, Republicans have tended to favor less centralization than Democrats, but national security is the one area of policy where Republicans favor strong government. In a Republican administration Donald Rumsfeld turned out to be the most dictatorial Secretary of Defense since Robert McNamara under the Democrats. Opinion on the proper balance is more likely to vary with the level of perspective: bureaucrats in particular agencies that have interests that conflict with other agencies may be more likely to see the virtues of decentralization, while highest-level policymakers overseeing the whole national security landscape are more likely to value tighter control from the top.

Policy, Strategy, and Organization

The National Security Act was spurred by rapid and radical changes in U.S. foreign policy, military activity, and commitments. Just over four decades later almost as radical a change occurred in the strategic milieu: the end of the Cold War and, at least for awhile, of great power competition altogether. If policy objectives and military requirements change drastically, doesn’t that naturally imply that the policy system for decision and implementation warrant drastic change too?

American military activity did not change drastically, however, when both the Soviet threat and worldwide Marxist-Leninist movements collapsed. Indeed, in the fifteen years after the Berlin Wall fell the United States undertook more than twice as many wars as it did in the forty-odd years of the Cold War: Iraq I, Kosovo, Afghanistan, and Iraq II compared to just Korea and Vietnam in the Cold War era. The American military has been engaged in combat continually since 2001. While the frequency of combat is greater than in the Cold War, the scale so far is not, but great power conflict has also reemerged with conflict over Ukraine and the islands of the South China Sea. Military challenges are not yet as big as they were from World War II to 1989, but they have not reverted to what they were before 1941 either.

One might ask, finally, whether radical change in the policymaking system should be undertaken if the United States ever comes to radically change its basic national security policy, such as shedding the role of globocop and sugar-daddy for cheap-riding allies, which would ratchet down substantially the degree of America’s global military activism. After all, the National Security Act was prompted by the change in the other direction: the radical expansion of commitments. Some changes in managerial particulars would of course go naturally with any big reduction in activity and responsibilities, but no general reversion to the scale of military activity before 1941 is in the cards. Trump’s “America First” instincts point in different directions, toward restraint on humanitarian intervention but belligerence on counterterrorism. In any event, the rationales for organizing the system depend primarily on the size and complexity of the system itself, not the particulars of who the foreign adversaries of the moment are or even what exactly the system is called upon to do.

The case for moving toward a mobilization strategy, as opposed to maintaining a high readiness characteristic of the times since 1950, should have been persuasive for a long time after the Cold War ended. This would involve reorienting the roles of active and reserve forces, military personnel policy in general, research and development and military procurement, organization of the defense industrial and production base, and other functions.²⁴ If that were ever undertaken seriously it would indeed involve major changes in military personnel, procurement, research and development, and other functions. The idea never had any serious constituency, however, and it has been harder to contemplate a mobilization strategy in the past several years, when Chinese and Russian activities have raised the prospect of great power conflict again. It is improbable that national leaders will adopt the logic of the case for reducing U.S. burdens to a budget of 2.5 percent of GDP.²⁵ Trump aims to spend more on the military, not less. If leaders were to do so at some point, however, the American profile in international security would still remain far greater than it was before the mid-20^th century, and the case for dismantling many of the organizations legislated in 1947 would be weak.

The world has changed a lot since 1947 but it has not become simpler. Nor have technological and economic functions become simpler; rather they have grown ever more complex. Only a radical simplification in the context and substance of national security policy could sustain a case for reducing the complexity of the system for policymaking and implementation. At some point technical changes might mandate wholesale reorganization, but so far the centralizing and coordinating reforms of the 1947 act and subsequent piecemeal reorganizations have worked at least as well as any radically new system plausibly would. It’s not broke, so let’s not fix it.