INTELLIGENCE, SECRECY, AND DEMOCRACY

Bruce Berkowitz[1]

Hoover Institution on War, Revolution and Peace, Stanford University

During the past three decades democratic government has spread throughout the world, and especially in the Western Hemisphere. It is a trend of historical importance. But it also presents important questions for anyone interested in the profession of intelligence.

Are intelligence operations compatible with democracy? Or is intelligence a necessary aberration that democracies must tolerate in order to survive in a hostile world? And, how can countries operate an intelligence community with the least compromises in the democratic process?

Freedom House, a New York-based nonpartisan organization, has conducted a nation-by-nation survey of political rights and civil liberties each year since 1978. According to this survey, twenty-five years ago most nations in North and South America were not free, or were only partially free. Today Freedom House reports that most nations on the two continents are free, meaning that their citizens can vote in fair elections and can express their views in public or through the press.

Although there were heated arguments between the supporters of extreme left-wing governments and right-wing governments during the 1970s and 1980s, it is clear that these regimes had more similarities than differences. Both offered reasons why it was necessary to restrict political participation and expression. But the results were the same: Elections were bogus or nonexistent, and controls on information limited the real choices voters enjoyed. These restrictions on information included press censorship, prohibitions on assembly, and controls on the ability of people to obtain information from their own government.

The democratization of the Western Hemisphere is probably the single most important development in the region during the twentieth century. In the future it will be seen that most other economic and social trends depended on whether these governments were free.

For those people in the intelligence profession, who study the intelligence profession, or who make intelligence policy, the development of political freedom in the Western Hemisphere is especially interesting. Many countries in the Americas have had intelligence organizations. Yet few of these organizations have historically been subject to democratic control. Often these controls were lacking or were deficient even in countries that otherwise had functioning democracies.

Even today there are some officials and scholars who argue that democracy is fundamentally incompatible with intelligence operations. At best, they say that intelligence operations are a necessary aberration from democratic government. Even some advocates of a strong intelligence community in their own country accept this argument.

Such assertions are common, but rarely are they subjected to rigorous, logical analysis. So, are intelligence operations and democracy really incompatible? To answer this question, one must first address three issues.

First, what are the essential features of democracy, and what is its purpose? If we do not know what democracy is or what it is supposed to do, it is impossible to determine whether intelligence operations present obstacles to democratic governance.

Second, what are the essential features of an effective intelligence organization? This question must be answered to determine whether there is something intrinsic about intelligence operations that conflict with democratic governance, or whether these conflicts are the result of poorly designed intelligence policies and operations.

 

And, third, if intelligence operations do require special restrictions on the democratic process, are they significantly different from other accommodations we make in democracies? It may be that intelligence presents problems for democracy that are not much different from other kinds of government activities. It may also be that people have learned to limit these problems in other policy areas, and these lessons can be applied to intelligence.

If one does perform such a systematic analysis, one discovers that the accommodations that are necessary to run an effective intelligence community in a democracy consist mainly of two kinds of compromises: Restrictions on participation in decision making, and restrictions on information. However – surprisingly – one also discovers that the requirement for such accommodations is not unique to intelligence. Other kinds of government activities require similar restrictions, and such concessions are routinely made even in governments that no one would claim are undemocratic. In fact, such restrictions often occur as a natural byproduct of democratic politics.

This is an important finding. It suggests that intelligence is not a “necessary evil” that democracies must engage in. Intelligence policies are not fundamentally different from other kinds of policies, and intelligence operations are not inherently different from other kinds of operations democracies carry out.

The central question is how intelligence policies are made, and how intelligence operations are conducted. It is possible to construct rules and procedures that are tailored for the special needs of effective intelligence, but which are not much different from those used to carry out other government activities.

 

Necessary Conditions for Democracy

 The most important condition for a government to be considered democratic is whether it holds fair elections. Ideally, elections should be held at predictable intervals and all adult citizens should have the right to vote (with the possible exception of, say, convicted felons and mental incompetents).

Voting is fundamental to democracy. It is the main mechanism through which citizens can influence the actions of their government. Petitions, public demonstrations, and lobbying are also important. But in a democracy, the thing that gives all these other tools weight is the ability of citizens to remove public officials and replace them with officials who will carry out different policies – in other words, the power of voting.

Beyond this basic condition, there are several secondary conditions that governments must to be considered democratic. Many of these other conditions are necessary for elections to have their intended effect, i.e., the potential to change the existing government and its policies. For example, free speech allows candidates to offer alternate proposals, and permits voters to criticize both current officials and the candidates who hope to replace them. Similarly, the requirement for policies to be made in public is required so voters can link an official to his actions. And so on.

But to fully understand why meeting these requirements are important to democratic government, one first has to jump ahead and say something why democracy is desirable. This is closely related to what one can expect democracy to achieve.

William Riker once observed that historically there have been three justifications for voting and, by extension, democratic government.^[2] He called the first justification the “Churchillian rationale” for democracy, after arguments that British Prime Minister Winston Churchill once made. According to Churchill,

“Many forms of Government have been tried, and will be tried in this world of sin and woe. No one pretends that democracy is perfect or all wise. Indeed, it has been said that democracy is the worst form of government except all those other forms that have been tried from time to time.”^[3]

 

It is hard to argue with this, but it is close to circular logic. Essentially Sir Winston was saying that voting to make laws and policy is good because voting is the best way to make laws and policy.

The second historical justification for voting and democracy that Riker offered was what he called the “Rousseauian rationale,” after the French philosopher Jean Jacques Rousseau. This justification argues that voting and democracy are good because the results reflect the will of the people. Or, as Rousseau wrote,

“If, when the people, being furnished with adequate information, held its deliberations, the citizens had no communication one with another, the grand total of the small differences would always give the general will, and the decision would always be good.”^[4]

The problem with this justification is that there is only way to enforce the conditions Rousseau assumed -- no communication, no lobbies, and no parties -- is by dictatorship, which more or less defeats the whole objective of democracy. Also, the only objective way to measure the will of the people is through voting, and elections can produce widely varied outcomes, depending entirely upon the procedure that is used. One need only look to the most recent U.S. presidential election to see this.

Kenneth Arrow showed in his “impossibility theorem” that there is no voting method of votes that guarantees a unique outcome, while also meeting the basic conditions of democracy -- giving all voters equal weight, allowing any candidate or platform to a chance to compete, allowing all parties to make deals and form coalitions, and so on.^[5] Richard McKelvey extended Arrow’s theorem by showing that in a democratic system -- that is, one-person/one-vote, majority-rule -- the middle ground is not necessarily the winner. Rather, in a truly democratic system, it is even possible for an extremist to put together a winning coalition.^[6]

In short, policy outcomes in democratic governments are always subject to political maneuver, the particular voting method used, or some other artifact. If this is the case, and voting cannot be relied upon to produce a unique, stable outcome, it is impossible to say that democratic decisions represent some specific, sacrosanct “will of the people,” as Rousseau tried to argue.

According to Riker, democracy can only guarantee two things: That voters dissatisfied with current policies will have an opportunity to propose an alternative at some certain time in the future, and that they will have a fair chance to compete. Riker called this the “Madisonian” justification for democracy, after James Madison, the fourth President of the United States, the chronicler of the American constitutional convention, and one of the authors of the Federalist Papers.

That’s the main principle at work in a democracy, and why democracy is desirable. The primary benefit of democracy is that it prevents bad policies from getting locked into place. In a democracy, citizens can always organize, mobilize, and wheel and deal so that, in time, they can replace the status quo with something else.

This might sound like a weak argument for democracy. But the invention of modern electoral democracy was, in fact, a remarkable breakthrough. Providing a means though which any citizen has the opportunity to change government peacefully offers an alternative to war and revolution. It also offers a political system that can draw on a broad reservoir for new ideas. In all, democracy is an elegant, remarkable device. One can also see why some of the features associated with democracy are necessary for its function. If there were no free speech, people could not offer alternative policies. If there was no transparency, officials could evade responsibility at election time.

 

Necessary Conditions for Effective Intelligence

Other the other hand, once one begins to toss around terms like “free expression” and “transparency,” one can also begin to sense why intelligence might incompatible with democracy. Before leaping to that conclusion, though, one must first think more rigorously about what conditions are necessary for effective intelligence. That analysis, like our analysis for democracy, requires systematic definitions.

Although people often use the term “intelligence” casually, clearly intelligence is different from the other kinds of information policymakers, warmakers, peacemakers, and other officials use. Intelligence is not ordinary information.

At least two properties make intelligence different from ordinary information. One is scarcity. Intelligence is not free for the taking, it is not freely available, and is not in unlimited supply. Quite the contrary; intelligence is usually hard to come by, because it is costly to collect or difficult to analyze, and because often your opponent does not want you to have it.

The other thing that makes intelligence different from other information is that it is what an economist would call a “private good” -- that is, it has some measure of exclusivity. An intelligence consumer can keep others from having it by controlling its distribution.

Scarcity and exclusivity not only define intelligence; they are essential for its effectiveness. The goal of intelligence is to obtain an advantage over your adversary -- to stay one step ahead of him, so that you can elude him, evade him, defeat him, or kill him. The process of collecting, analyzing and using information -- and staying one step ahead of your opponent -- is often referred to as a “decision cycle.”

Like most important concepts, it has a variety of roots. In part, the idea comes from organization and management theory. But the military version of the decision cycle concept is often attributed to John Boyd, an American fighter pilot. Boyd called his version of the decision cycle the “OODA loop,” for the four steps that it incorporates: observe, orientate, decide, and act.

Boyd is something of a cult figure among many military thinkers, and his ideas have been incorporated into American military doctrine. The advocates of “maneuver warfare” in the 1980s were acolytes of Boyd. The U.S. Marines explicitly refer to Boyd in their combat manuals, and Boyd’s ideas about the importance of an information edge is part of the current Revolution in Military Affairs and network centric warfare.^[7]

Today, getting to the end of your decision cycle is more important than ever because weapons today are so deadly that one shot often means one kill. Whoever gets to the end of their decision loop first wins, and intelligence is one of the most important means for achieving this.^[8]

Thus, for intelligence to be effective, it must be not only timely and accurate, but it must more timely and accurate than whatever your adversary has. Or, to put it another way, universally available information puts everyone on a level playing field, while intelligence tilts the field in your favor. This is why intelligence requires secrecy. Intelligence is valuable because of the advantage it offers, and secrecy is often necessary to protect that advantage.

Now the challenge of operating an intelligence organization in a democracy becomes clearer. Simply put: information is the engine that makes democracy work, while the effectiveness of intelligence depends on restricting the flow of information.

Can a government impose the secrecy intelligence requires and still claim legitimately to be a democracy?

 

Secrecy and Democracy

One way to address this question is to ask, how much secrecy usually exists in a democracy? As we shall seen secrecy is more common in democratic government than one might think. In fact, some of this secrecy is necessary for democracies to function.

That is the formula for reconciling intelligence and democracy. We can use the same kinds of devices to permit intelligence organizations to operate secretly when they needs to, while at the same time limiting the impact on the democratic process.

As we have already alluded, direct popular democracy is rare, and especially at the level of the national government. The most obvious reason, of course, is simply that national referenda are usually impractical in the modern state. There are too many people, and too many issues to be decided.

Yet, even if it were possible to overcome these mechanical and logistical hurdles, almost all democracies would still have segmented, indirect democracy. There is a natural tendency in all democracies for people to organize themselves into groups for common purposes – such as, for example, political parties.

At one time, political parties were considered a violation of the democratic process, or a symptom of some deficiency in a democratic government. Madison himself, for example, warned of the “mischief of factions” in the Federalist Papers:^[9]

By a faction, I understand a number of citizens, whether amounting to a majority or a minority of the whole, who are united and actuated by some common impulse of passion, or of interest, adversed to the rights of other citizens, or to the permanent and aggregate interests of the community.

More recently, officials and scholars alike have come to see parties as a normal -- even inevitable -- part of the democratic process.^[10] Indeed, many national constitutions implicitly or explicitly recognize parties as permanent players. For instance, the U.S. Constitution was amended in 1804 so that the President and Vice-President would be elected from the same party, rather than from the winning party and the runner-up, as originally stipulated. Many political commentators in the United States even suggest that our democracy would work better if parties did a better job of pooling and representing common interests.^[11]

In any case, the electorates in all democracies worthy of the name have organized themselves into subgroups. In addition to parties, there are also political action committees, lobbying groups, campaign organizations, trade unions, and industry and professional organizations. A similar process occurs in most legislative bodies. Every parliament and congress in a functioning democracy has caucuses and coalitions.

The point is, democracy almost always consists of a multi-step process, and the bodies that exist at every step have the potential to control information. And many of them do.

Political campaigns plan strategies in secret; in the United States, in fact, there have been several scandals in which one party or candidate stole confidential planning papers from their opponent. Legislative committees often meet in closed session, as do party caucuses.

Within the executive branch, most cabinet meetings are closed to the public. Departments and agencies often operate in private, except for public hearings. Contractors often negotiate the details of agreements with government agencies in private to protect intellectual property. In the judicial branch, often the only parts of a proceeding that are held in public are the oral arguments, and the rest are held in chambers.

In fact, one might make an argument that, left to their own devices, participants in the democratic process choose for secrecy at least as often as they choose openness. That is why many state assemblies in the United States have passed “government in the sunshine legislation” and, at the federal level, Congress passed the Freedom of Information Act.

People accept a certain amount of secrecy because, in many cases, a person’s right to privacy seems to trump the public’s right to know. In some cases, secrecy is necessary because government administration would otherwise be too chaotic. In some cases, there is general agreement that parties or candidates should be allowed to plan and organize in secret.

But in every case, there is an underlying issue that must be addressed: Does secrecy defeat the democratic process? In most of the cases described above, people agree that they can tolerate considerable amounts of secrecy without violating the democratic process too much (assuming that certain conditions are met, which we will discuss in a moment). This fact is important because it suggests that the secrecy that intelligence requires is not unique; secrecy is often tolerated as part of government and politics.

Moreover, secrecy in itself is not inherently incompatible with democratic government. Rather, there are many ways to control the impact of secrecy on democracy. The key question is how to limit the effects of secrecy, and that brings us back to why democracy is a desirable form of government. If we assume that the purpose of democracy is simply to allow citizens to cause their government to adopt a set of policies different from the status quo, the remedies for the effects of secrecy become clearer.

Whenever government agencies or other organizations constrain the flow of information,

 •It should still be possible to link officials to their policy decisions;

 •It should be possible to make this linkage in a timely fashion -- at a minimum, policies should not go so far that unacceptable damage occurs before officials are held to account;

 •Whenever government organizations limit the flow of information, there should be built-in checks that limit the actions they can take; and

 •Whenever a government organization is permitted to limit the flow of information, there should be mechanisms that periodically turn over their membership, and, if possible, ensure that a diverse range of people take part;

Some of the practical measures that governments can adopt to ensure that these conditions are met include:

 •Limiting the amount of time facts can be kept secret;

 •Establishing specific guidelines defining when information can be classified and withheld.

 •Providing for limited terms for both legislators serving on oversight committees and top officials.

 •Providing both executive branch officials and legislative bodies the authority to release secrets under well-defined procedures that holds both accountable for their decisions.

 •Providing citizens the opportunity to petition for the release of secrets, and establish a mechanism in which an impartial party can decide whether to grant such a petition in a timely fashion according to an objective set of criteria that are set by elected officials.

The oversight system for intelligence in the United States includes all of these elements. Their effectiveness in practice has been uneven, but, in principle, they provide an approach for reconciling democracy, secrecy, and, thus, intelligence. Also, shortcomings in current procedures suggest where authorities should focus their efforts for reform. This would reduce the impact of secrecy on democracy -- and would result in an intelligence system better suited to the threats democracies face today.

 

Secrecy and the Intelligence Profession

As we have seen, the intelligence profession has a special interest in secrecy because effective intelligence usually depends on having an information edge over your opponent. Unfortunately, the traditional approach intelligence organizations have used to protect secrets not only compromises democratic government in certain avoidable ways; it is also rapidly becoming ill suited to today’s threats and technology. This approach usually consists of three components:

 •A formal definition of classified information. Most classification systems have several separate levels of classification and individual compartments within each level. Levels of classification are defined by the sensitivity of the information. Compartments are defined according to the source of the information or its intended use.

 •A formal set of procedures for clearing personnel who are to be given access to classified information. The screening procedures are intended to be more stringent for individuals with access to more-sensitive information. The screening procedures are also designed to weed out enemy agents and people who might be untrustworthy or susceptible to compromise by foreign intelligence services. These criteria change over time.

 •Technical and operational procedures for protecting classified information. These range from requirements that visitors must sign in before entering a secure facility to specifications for locks and safes to ciphers for electronic transmission of data.

The current approach to protecting secrets is, in short, a classic bureaucratic process: formal rules combined with a hierarchical organization for implementing them. For example, in the United States rules for classifying information are set forth in Executive Order 12356, issued in 1982. The order defines three levels of classification: “Top Secret,” “Secret,” and “Confidential.” The definitions of these levels, given in Section 1.1(a) of the order, are as follows:^[12]

 •"Top Secret” information which, if disclosed could reasonably “be expected to cause exceptionally grave damage to the national security.”

 •"Secret" information which, if disclosed could reasonably “be expected to cause serious damage to the national security.”

 •"Confidential" information which, if disclosed could reasonably “be expected to cause damage to the national security.”

EO 12356 also establishes authorities and a chain of command for implementing these rules. However, this approach -- which is typical of systems used worldwide -- presents problems for both democracy and, oddly enough, the intelligence profession itself.

From the perspective of democratic governance, these definitions are simply too vague. The order provides no definition of “damage,” “serious damage,” or “exceptionally grave damage” or method for measuring them. As a result, whenever political discourse happens to address information that is classified, view people really knows the rationale for why the information cannot be discussed openly (even apart from knowing the actual information). Historically, this has been a formula for political demagoguery, and often debates end up focusing on the sensitivity of the information and who might or might not have compromised it, rather than the issue at hand.

Worse, there is no requirement to compare of the costs of restricting the information to the costs of not disclosing it. In effect, an official who has the authority to classifying information can exercise his or her own judgment, and there is no opportunity for anyone to offer counter-arguments or question their subjectivity.

From the perspective of intelligence professionals, this approach also presents problems. Excessive secrecy -- the inevitable result with guidelines as broad and vague as these -- has probably caused as many intelligence failures as lax security. This was likely the case in the failure of U.S. intelligence and its counterparts in other countries to anticipate the terrorist attacks of September 11th.

One of the main causes of the September 11th intelligence failure was that organizations did not communicate effectively. Part of the problem was technology -- communications systems did not link with each other -- and part of the problem was organization -- people in different agencies had been trained not to speak with each other. But in both cases, the underlying cause was too much compartmentation, which is to say excessive secrecy.

The traditional approach to secrecy was probably effective with older information technology. The hardware, software, communications links, and databases did not change much or as often as they do today. Also, the threat did not change much, so the people who had a “need to know” was a stable population, too.

Today, of course, conditions are significantly different. Technology evolves rapidly, the threats intelligence organizations must cover can change significantly, and the data and expertise that are needed for intelligence vary greatly, too. That is why the intelligence profession needs to consider new approaches to protecting secrets.

Intelligence professionals take our current methods of protecting secrets for granted. But many organizations that have requirements for protecting secrets use a very different approach, and many of them protect secrets as effectively as intelligence agencies -- or more. For example:

 •Corporations must often protect proprietary data and intellectual property, even as they market products to mass audiences and collaborate with other companies.

 •News organizations must protect sources -- even as they disseminate the information from their sources to audiences consisting of millions of people.

 •Professional fields such as medicine and law routinely protect secrets in doctor-patient and lawyer-client relationships. Religious organizations protect the sanctity of the confessional, and many ethnic groups, cults, fraternities, and other social groups protect the secrets of their order.

 •Organized crime and terrorist organizations have -- unfortunately -- a remarkable record in protecting secrets.

True, these examples can only go so far. We cannot run intelligence organizations like a cult group, and the Cosa Nostra uses certain tools for enforcing secrecy and security that no democracy would accept. Still, intelligence organizations can learn some useful lessons by studying these other organizations. Some of the features of these alternative approaches include:

 •Promoting professional norms and standards to protect information (and disclose information when appropriate), and relying less on overly broad and rigid systems of rules and regulations;

 •Encouraging intelligence organizations to develop reputations for honesty, objectivity, and minimizing secrecy so that they do not need to disclose as many details;

 •Accepting that there will always be some leakage of sensitive information in the intelligence, so that when leaks do occur, officials do not over-react;

 •Realizing that there is a cost in being too cautious in sharing information -- both in the effectiveness of the information product, and in explaining the mission of the organization.

Secrecy is a tool of the intelligence professional, just as, say, firearms are tools of law enforcement and drugs are tools of the medical profession. In the future, measures for protecting secrets will need to be built more into tradecraft and less into regulation. Instead of trying to keep large programs moderately secret for indefinite periods of times, intelligence officers will need to learn how to use secrecy in a more focused fashion -- the approach now used with many “special access programs.”

In this approach, only small programs or just sensitive operational details will be kept secret. But these secrets will be more tightly held. Intelligence officers and officials will need to take greater personal responsibility for security -- rather than blindly obeying standard procedures. The effective protection of secrets will need to be a larger part of an intelligence professional’s tradecraft.

 

Conclusion

The question is, can democracies maintain effective oversight over intelligence if intelligence organizations take this new approach to secrecy?

There is reason to be optimistic. As we have seen, current approaches to protecting intelligence secrets are ill suited for today. And, because they are so cumbersome and vague, they present their own problems for democratic oversight. So almost anything if likely to be an improvement.

As we have also seen, democracies are not strangers to secrets. Protecting secrets where appropriate, disclosing secrets when proper, and managing secrecy are all normal parts of the democratic process. The same principles that are used to strike a balance among competing interests can be used to oversee intelligence secrets as well.