. 21
( 53 .)


pers, equated the law with “Britain™s loathsome Of¬cial Secrets Act.”4
The New York Daily News railed against what it called a “Soviet-style
secrecy law.”5 “Other nations have long criminalized the disclosure
of government information,” said former White House counsel John
Dean, “but there™s a crucial difference between them and us: They
lack an equivalent of our First Amendment.”6

Blacked Out

In reality, the distinction between “them and us” was not quite
so simple. It was true that the United States had a long tradition
of resisting attempts to create criminal penalties for leaking classi-
¬ed information. In 1957, for example, Congress rejected propos-
als put forward by two special inquiries “ the Wright Commission
on Government Security and the Coolidge Committee on Classi¬ed
Information “ to establish such penalties.7 The proposals, said the
New York Times™ James Reston at the time, posed great danger to
government accountability and a free press.8
Reston could not have appreciated the extent to which the United
States was, at that very moment, secretly succeeding in efforts to
strengthen criminal penalties for unauthorized disclosure of classi-
¬ed information in other countries. In the spring of 1953, planners
within the North Atlantic Treaty Organization became frustrated by
a series of compromising leaks of information about NATO military
planning. At a secret meeting in Paris in April 1953, the Organization™s
Secretary General, Lord Ismay, told NATO leaders that “the problem
of information leaks” had become grave and that “it was essential to
search out the malefactors and to make examples of those who were
Some of the most frustrating news stories had arisen as a con-
sequence of leaks by American sources to C. L. Sulzberger of the
New York Times. The leaks provoked a three-year effort by the NATO
Security Committee, in closed meetings at NATO headquarters at the
Palais de Chaillot in Paris, to push for stronger antileaking penal-
ties. Prodded by the Security Committee, several NATO countries
strengthened criminal antileaking laws over the next four years.
Ironically, the United States “ the dominant partner in NATO “ was
the only major government that did not respond to the Security Com-
mittee™s call for stronger laws.10
The United Kingdom was not then a bastion of open govern-
ment, and appeals from NATO to maintain a strong antileak law were
unlikely to be ¬ercely resisted. In other respects, however, British pol-
icy on the handling of sensitive information was more liberal than
that of the United States. British of¬cials protested against American
demands for tighter security clearance procedures, arguing that the
procedures were likely to be ineffective and that the American stan-
dards “ which denied clearances based on political af¬liation and
sexual orientation “ went too far in limiting civil liberties. In 1953,

Opaque Networks

the U.S. government made clear it would not share any informa-
tion about its nuclear defense plans until the British adopted tighter
procedures, and the British ¬nally relented.11 After 1955, these U.S.
rules were codi¬ed in secret standards that applied to all NATO
There were also other governments that were troubled by the
secrecy rules imposed through NATO. Norway and Denmark com-
plained privately in the 1950s that NATO™s standards for classifying
information were too broad and vague. Other tensions arose as NATO
member states began adopting national right-to-information laws.
NATO™s practice of subjecting all information ¬‚owing through its
channels “ classi¬ed and unclassi¬ed “ to the “third party rule” (which
gives the originating government absolute control over its distribu-
tion) clashed fundamentally with the ethic of disclosure built into the
new national laws. Sweden, which has a long tradition of open gov-
ernment, felt this con¬‚ict acutely when it began cooperating more
closely on military matters with Western European governments in
the early 1990s.
In the late 1990s, NATO™s in¬‚uence on national openness poli-
cies became evident as it responded to American pressure to incor-
porate former Soviet bloc countries in Central and Eastern Europe.
Following the collapse of the Berlin Wall, countries in this region had
been quick to throw off Soviet-era secrecy laws and adopt right-to-
information laws on the Western model. Soon, however, governments
began adopting new state secrets laws that included rules on the pro-
tection of sensitive information, security clearances, and criminal
penalties for leaking. The motivation was simple: NATO had made
clear that the laws were a prerequisite for joining the alliance.
Nongovernmental organizations throughout the region were frus-
trated by their ˜governments™ rush to adopt new secrecy laws. In
Romania, legislators successfully launched a constitutional challenge
against the government™s ¬rst attempt at a secrecy law, arguing that
it had been compromised by procedural irregularities. (Legislators
were eventually persuaded to agree on a law in 2002. A security of¬-
cial “came down like a storm” on members of a Senate committee,
a Bucharest newspaper reported: “ ˜This morning we have received
signals from Brussels indicating that if the bill on classi¬ed informa-
tion is not passed before 16 April, they cannot exclude adopting a
critical attitude regarding Romania. We agree with any form “ the

Blacked Out

colonel added “ but please, pass it as soon as possible, or we will
be facing huge problems™.”13 ) Protests also delayed the Slovak law,
while the Polish, Czech, and Bulgarian laws provoked unsuccessful
constitutional challenges.
The Bulgarian case provided a vivid illustration of the tensions
between the post-1989 spirit of openness and the new emphasis on
secrecy. The Classi¬ed Information Protection Act adopted in 2002
included a repeal of a 1997 law that allowed citizens to access the
¬les created by Bulgaria™s secret police in the Soviet era. Was this a
step required by NATO, as the Bulgarian government claimed, or had
it simply exploited an opportunity presented by the NATO accession
process? This proved dif¬cult to judge, because NATO itself refused to
release the unclassi¬ed documents that detailed its requirements.14
In September 2003, the Bulgarian government also proposed more
severe criminal penalties for leaking classi¬ed information, including
sanctions against journalists who repeated the information. Whether
NATO had speci¬cally required the measures remained unclear.
Bulgarian newspapers reported in the months before accession that
NATO sources had become concerned about Bulgaria™s capacity to
protect secrets,15 and U.S. Secretary of State Colin Powell, in a meet-
ing with the Bulgarian President, had urged that the government take
further measures to protect classi¬ed information.16

Protecting the network
NATO has been called one of the great military and political alliances
in history. In contemporary terms we would call it a model of “net-
worked governance.” The ten countries that signed the Washington
Treaty in April 1949 faced a common problem that could not be
resolved through action by any one state alone. A system for collective
action by a number of legally autonomous but interdependent states
had to be devised.17 A critical part of the infrastructure built to sus-
tain this new collective security network was intended to encourage
“information sharing” (to use another contemporary phrase) among
military staff and diplomats in the network. As it turned out, build-
ing an effective information-sharing policy was not easy. In fact, the
¬rst ten years of the alliance were punctuated by repeated efforts to
overhaul the policy. But by the end of the ¬rst decade the NATO coun-
tries “ heavily in¬‚uenced by the preferences of the dominant NATO

Opaque Networks

member, the United States “ seemed to reach rough agreement on the
rules for information sharing.
The deepening ¬‚ow of information among NATO governments
undoubtedly enhanced collective security. However, security came at
a price: In the drive to give assurances that shared information would
be strongly protected, each government took steps that restricted the
ability of its own citizens to gain access to the information “ even
unclassi¬ed information “ ¬‚owing through the network. The network
ensured transparency for actors within the network, but opacity for
those without.
That the policy should have put so little emphasis on adequate
transparency for actors outside the network is not surprising. At the
time, the threat posed by the Soviet Union and its satellites seemed
overwhelming, and the revolution in popular expectations about gov-
ernmental transparency had not yet occurred. The dominant actor
in the network “ the United States “ was also indifferent to the costs
that NATO policy imposed on legislators and citizens in other states.
Furthermore, NATO was aided by the secrecy that was imposed on
the policy itself “ it was dif¬cult to protest conditions imposed in
documents that were themselves withheld from public view.
NATO™s own signi¬cance may be in question after the collapse
of the Soviet Union, but the dilemmas posed by its information-
sharing policies have not faded. On the contrary, the restructur-
ing of military and political relationships in the post-Cold War era
have led to a proliferation of bilateral and multilateral partnerships
among national governments that often include equally problematic
rules on the handling of shared information. The burgeoning num-
ber of international networks can be divided into three categories:
those relating to defense cooperation; cooperation between national
intelligence agencies; and cooperation between national police

International defense networks
The U.S.-led expansion of NATO is only the most obvious aspect of the
post-Cold War restructuring of military alliances. The United States
has also undertaken new bilateral partnerships that involve the shar-
ing of information between military and diplomatic of¬cials “ and, as
a consequence, negotiation on terms to protect shared information
from unwanted disclosure.

Blacked Out

The conditions that govern the handling of shared information are
laid out in bilateral Security Of Information Agreements, or SOIAs.
SOIAs are not new: The United States has been negotiating them with
its defense partners for a half-century. In the 1950s, many countries
that collaborated with the U.S. through NATO also signed bilateral
SOIAs that provided assurances on the ¬‚ow of information outside
NATO channels. The practice of negotiating SOIAs was formalized
in the United States in 1971 by National Security Decision Mem-
orandum 119, which prohibits the sharing of military information
with a foreign government that has not signed a legally binding SOI
In 1999, the Australian government created consternation within
the U.S. Defense Department when it privately advised that it had
never taken the steps required to make their classi¬ed 1962 SOI
agreement with the U.S. legally binding. U.S. of¬cials began press-
ing the Australian government to adopt an agreement that would be
enforceable under international law. “A gentleman™s agreement is not
enough,” a Defense Department of¬cial said. The State Department
agreed. Its “biggest concern,” an internal e-mail explained, was to
get Australia “on the hook legally.” Hurried negotiations produced a
new, legally binding agreement. Australia™s Foreign Affairs minister
told the public that the new agreement was negotiated to “account
for advances in information technology” since 1962.18
Tracking the number of SOIAs that have been negotiated by the
Department of Defense is dif¬cult, because the very existence of
the agreements themselves may not be acknowledged. For example,
the U.S. does not acknowledge an SOIA with Indonesia. It may be
that no such agreement exists; alternatively, the Indonesian govern-
ment might be reluctant to acknowledge the agreement, to avoid stok-
ing domestic concern about military collaboration with the United
States. In other cases, an agreement may be publicly acknowledged
while its content remains classi¬ed. For forty years, the British and
American governments refused to divulge the content of the bilateral
SOIA signed in 1961. This agreement was ¬nally declassi¬ed in 2001,
but the details of some other agreements “ such as the 1986 U.S.“
Turkey SOIA “ remain secret.
Despite such secrecy, there is evidence that the number of U.S.
SOIAs is growing in the post-Cold War age. The collapse of the Soviet
Union led to the negotiation of bilateral SOIAs with former Warsaw

Opaque Networks

Pact states even before those governments joined NATO. Alliances in
other regions “ such as the intensi¬ed collaboration of the United
States with the Indian government “ also led to the negotiation of
new SOIAs. Roughly one-third of the ¬fty-four bilateral agreements
that have been acknowledged by the United States were negotiated
in the decade following the collapse of the Soviet Union.19
The content of these agreements seldom varies. At minimum, they
entrench the third-party rule, so that a government receiving infor-
mation covered by the agreement is denied any discretion to make
a judgment about the wisdom of releasing shared information with
legislators or citizens, or third countries. SOIAs may also require gov-
ernments to adopt acceptable procedures for security clearances and
physical protection of information as well; agreements with countries
that are also NATO partners may simply apply NATO standards to all
¬‚ows of defense information between the two countries.
The right-to-information laws adopted by national governments
must be trimmed to accommodate the constraints imposed in these
sometimes-secret documents. On the American side, an executive
order creates a presumption that any information received from
foreign governments merits classi¬cation, thus protecting it from
disclosure under the Freedom of Information Act.20 Other coun-
tries may create a similar blanket exclusion within their right-to-
information law, or persuade courts that the release of shared infor-
mation necessarily jeopardizes national security, regardless of the
actual sensitivity of the information that is the subject of an access
In some instances, there is evidence that agreements have been
crafted to anticipate the pressures created by new transparency laws.
In 1982, Australia and Canada were among the ¬rst eight countries to
adopt national right-to-information laws (following the United States,
the Nordic countries, and France). Both laws established a general
presumption that government-held documents should be publicly
accessible and created procedures for an independent tribunal to ref-
eree disputes about access to information. These provisions clashed
with the arbitrary protection provided by SOIAs. In 1996, diplomats
and defense of¬cials in the two countries negotiated a new bilateral
SOIA drafted to counter the new legal restrictions. Each government
promised to “take all steps legally available” to block disclosure of


. 21
( 53 .)