The national security state is among the most consequential institutional innovations in American governance since the New Deal. It has also produced one of the most persistent patterns of governance failure in the post-war era.
What the National Security State Is
The national security state — the constellation of agencies, authorities, budget programmes, and institutional relationships assembled in the decade after the Second World War and expanded continuously since — is among the largest and most secretive components of the federal government. Its core institutions — the CIA, the NSA, the DIA, the military service intelligence agencies, and the network of contractors and federally funded research centres that support them — operate under legal authorities and oversight frameworks that differ fundamentally from those that govern the rest of the executive branch. The national security state's defining characteristic is the national security exception: the claim that the specific demands of intelligence collection, covert action, and military planning justify a governance framework that is less transparent, less accountable, and less subject to normal legal constraints than civilian government. Formal oversight structures exist — congressional intelligence committees, an inspector general system, a dedicated surveillance court — but the historical record is that these structures have repeatedly proven deferential to, or dependent on, the institutions they were built to check.
The justification for the national security exception is genuine in its core: some intelligence and military activities require secrecy and speed that normal democratic accountability processes cannot provide without compromising effectiveness. The institutional problem is that the national security exception, once established, generates its own expansion: the definition of what counts as a national security matter expands to accommodate the institutional interests of the agencies that benefit from the exception, the oversight mechanisms that were supposed to constrain the exception become deferential to or dependent on the institutions they oversee, and the secrecy that protects genuine national security information also protects the institutional failures, the legal violations, and the policy decisions that would not survive public scrutiny.
The national security state's governance failure is structural: it claims the authority to be exempt from the accountability mechanisms that prevent other government institutions from abusing their authority, on the grounds that accountability would compromise its effectiveness. The claim is sometimes true. It is also the claim that every institution seeking to avoid accountability makes, regardless of whether accountability would actually compromise effectiveness.
The Record
The mechanism is not hypothetical; it has a documented history of reform followed by relapse. In the mid-1970s, prompted by reporting on CIA assassination plots and an Army domestic surveillance programme, the committee launched a series of investigations into American intelligence agencies including the CIA, National Security Agency (NSA), and the Federal Bureau of Investigation (FBI). What the Church Committee found was not a handful of isolated abuses but a settled institutional practice: the committee exposed COINTELPRO, MKUltra, mass surveillance, and assassination plots, and established that the NSA had been conducting warrantless surveillance of Americans for decades, something the government had consistently denied was possible or legal. The final report ran to six volumes and totaled 2,702 pages.
The reforms that followed were real. President Ford issued Executive Order 11905 banning political assassinations, and Congress created standing oversight bodies: the Senate Select Committee on Intelligence (1976) and House Permanent Select Committee on Intelligence (1977) were created to provide ongoing congressional oversight. The Foreign Intelligence Surveillance Act of 1978 created a dedicated court to review domestic surveillance warrant requests, replacing unilateral executive authorization with judicial sign-off. This is the reform cycle working as designed: exposure, embarrassment, new institutional constraint.
What happened to that constraint illustrates the second half of the mechanism — capture. Between 1979 and 2012, federal agencies submitted 33,900 ex parte requests to the court. The judges denied eleven, and approved the great bulk of the remainder — though not uniformly as filed, since a further, unquantified number were withdrawn, resubmitted, or granted only in modified form rather than granted outright on first request. The headline figure this produces, a 99.97% rate of approval, measures outright denials against total requests; it does not by itself capture how many applications were altered before they were approved. The court itself has argued this figure is misleading rather than damning: for the period 2001 through 2012, it has said that it demanded changes to 24.4 percent of those applications before granting final approval, meaning the low denial rate partly reflects a pre-filing negotiation process rather than uncritical rubber-stamping. Both things can be true at once — a court that pushes back in private while almost never saying no in public — and that ambiguity is itself characteristic of an oversight body whose proceedings are classified and hears only from the government it is reviewing.
The clearest instance of oversight capture running the other direction — an official misleading the overseers rather than the overseers deferring to the official — came in 2013, when the Director of National Intelligence was asked directly by a senator whether the NSA collected data on millions of Americans. On March 12, 2013, National Intelligence Director James Clapper told Congress that the federal government didn't collect large amounts of data on millions of Americans: "No, sir," and, "Not wittingly." Months later, leaked documents showed the U.S. intelligence community obtained secret court orders forcing phone companies to turn over millions of U.S. call records on an "ongoing, daily basis." No prosecution followed; Clapper was poised to avoid charges for allegedly lying to Congress following years of apparent inaction by the Justice Department. The episode is a near-literal illustration of the portable law: the body charged with providing truthful answers to the overseers determined, in the moment, that the demands of secrecy outranked the demands of candor, and paid no institutional price for that determination.
None of this proves that the specific programmes exposed were unnecessary, or that the officials involved acted in bad faith at every turn. It proves something narrower and more durable: that the accountability structures built after the last cycle of exposure — a surveillance court, standing intelligence committees, an inspector general regime — did not prevent the next cycle of concealment. The exception did not stay bounded to the cases that justified it; it expanded to cover the institution's discomfort with scrutiny generally, and the mechanisms meant to police that expansion were either too deferential to catch it or too easily misled to matter. That is the structural claim this essay makes, and it is not falsified by the fact that some fraction of what the national security state classifies is, in fact, genuinely secret.
Discussion