State Capacity of Secret Surveillance

Abstract

The state capacity-civil society tradeoff model tends to treat the “state” and “civil society” as separate entities who move to constrain one another. However, this modeling technique leaves out the nuances of individual action within a collective setting by treating each as a relative black box. This article explores this balance in the context of the surveillance state that has arisen in the 20th and 21st century. As state capacity in surveillance increases it better allows the state to respond to threats to citizens from citizens. However, the increased capacity also lessens the ability of societal pressure to check authoritarian advances even in a nation with a thriving civil society presence.

Introduction

In 2013, The Guardian¹ and The Washington Post² began publishing information about ongoing National Security Agency (NSA) activities that were discovered through documents leaked by former NSA consultant Edward Snowden. Citizens became aware in more detail of the existence and operations of the Foreign Intelligence Surveillance Court (FISC), of the extent of authority provided under Section 215 of the USA PATRIOT Act, of the PRISM program, and to a lesser degree the expansive use of the 1981 Executive Order 12333 regarding foreign information. These stories, and the leaked documents that spawned them, indicated that the NSA had been conducting surveillance operations of both foreign nationals and U.S. citizens on a massive scale.

Some view this incident as a failure of “state capacity” in the face of the overwhelming strength of civil society, a society that does not allow the state to do as it wishes. Acemoglu and Robinson (2019, pp. 332–336) use the history of the U.S. surveillance state to illustrate their model of the continuous struggle between state authority and individualism. In doing so, they argue that civil society in the U.S. was sufficiently strong, and the state capacity sufficiently weak, that the U.S. government was prohibited from openly engaging in surveillance and had to resort to secretly building the capacity. In their argument, “the shackles and compromises imposed” on the U.S. government “have kept it weak and forced it to come up with innovative, sometimes unusual solutions to deal with new problems and expand its capacity” (Acemoglu and Robinson 2019, p. 332). Thus, it is the weakness of the state that leads to the secretive surveillance of U.S. citizens that is exposed in the 2013 NSA leak. The topic of this trade-off has seen a resurgence of late with economists (Acemoglu and Robinson 2012, 2019, 2023; Besley and Persson 2009, 2010; Johnson and Koyama 2017; Heldring and Robinson 2023) and political scientists (Berwick and Christia 2018; Hanson and Sigman 2021; Lee and Zhang 2017; Lindsey 2021) working in the areas of economic development, political economy, and institutional economics. The role of state capacity, and the ability to constrain it, is argued to be an important factor for economic growth and long-run economic stability.

This current article has three overarching goals: (1) to critique the state capacity literature by illustrating the difficulty civil society has in effectively shackling the state, (2) to provide a history of U.S. state surveillance, and (3) to demonstrate that a political economy approach is a better fit for explaining the evolution of state surveillance than the state capacity model. To achieve goals 1 and 3, we argue that the state capacity-civil society tradeoff model tends to oversimplify the collective action problems³ on each side, as it tends to treat the state and civil society as separate entities who move to constrain one another. When applied to the U.S. security state, this simplified model overlooks the “multiplicity of competing, parochial, and inconsistent interests at work in the American deep state” (Koppl 2018, p. 222). Rather than being separate entities, the military- or surveillance-industrial-complex is more aptly modeled as “entangled” (Koppl 2018, p. 222) due to the intermingling of the public and private sectors within the complex (see Higgs 1987, 2004, 2006, 2007, 2012a, b, Duncan and Coyne 2013a, b, 2015, Hooks 1991; Coyne 2015; Coyne and Duncan 2019; Alshamy et al 2023).

The remainder of the paper will be organized as follows. In “The state capacity model” section more fully critiques the state capacity literature. In “The U.S. surveillance state in the 20th to the 21st century” section provides a history of the U.S. surveillance state in the 20th and 21st century. Section 4 applies a political economy analysis to that history. In “Assessing the constraint” section concludes.

The state capacity model

The core concept of state capacity, and the trade-off between state authority and individual freedom, is not new even if the terminology has shifted over time. It is a study at least as old as Hobbes’s (1651) publication of Leviathan, where he argues the state arises to protect individuals from anarchy. Locke’s (1689) Two Treatises of Government covered similar ground albeit from a rather different perspective, noting that consent of governed is necessary condition. Smith’s ((1776) 1991) publication of The Wealth of Nations similarly investigates these themes with a more benign view of human tendencies. Though there has been a resurgence in the popularity of this debate, perhaps stemming in part from Acemoglu and Robinson’s Why Nations Fail (2012) and The Narrow Corridor (2019), it has not been entirely neglected. It is useful to address, in brief, the current state of the literature before applying the model to the surveillance state.

Defining State Capacity and Civil Society

The concept of state capacity is one that is difficult to concretely define, as use of the term “varies considerably across the literature in political science and related disciplines” (Hanson and Sigman 2021, p. 1496), with the formal definition being highly dependent on the scope of the research question undertaken (Berwick and Christia 2018).⁴ Use of the term is further complicated by its association with several “closely related attributes of states: strength, fragility, failure, effectiveness, efficiency, quality, legitimacy, autonomy, scope, and so on” (Hanson and Sigman 2021, p. 1496). The most common construction stems from Mann’s (1984, pp. 188–189) use of functional characteristics to define the state (see also: Suryanarayan 2021). Mann differentiates state power into two forms, despotic and infrastructural, with infrastructural power signifying “the capacity of the state to actually penetrate civil society, and to implement logistically political decisions throughout the realm” (Mann 1984, p. 189, emphasis added⁵).

Attempts to measure this capacity have been tied to the state’s fiscal capabilities (Besley and Persson 2010, 2013; Irgoin and Grafe 2013; Dell et al 2018; Johnson and Koyama 2017), to the effectiveness of the existing bureaucratic apparatus (Bruszt and Campos 2019; Dell et al 2018; Evans and Rauch 1999; Knutsen 2013; Acemoglu et al. 2016; see also Heldring and Robinson 2023, p. 403), to attempted measure of leadership and control within society (Sobek 2010; Christensen 2019; Suryanarayan 2021; Heldring and Robinson 2023). Hanson and Sigman (2021, p. 1496) lean into the control aspect to argue that state capacity “embodies state power, as in the ability of one actor [the state] to get another actor [members of society] to do things they would not otherwise do.” Whether the power of the state is used for beneficial or nonbeneficial ends, state capacity is measured by “the state’s ability to accomplish its intended policy action—economic, fiscal or otherwise” (Dincecco 2017, p. 3).

Civil society is similarly a complicated concept. As Skidmore (2001, p. 55) notes, “the term refers to the varied forms of social organisation that lie between the individual and the state.” In Mann’s (1984, p. 188) original differentiation, civil society is defined simply as “power groupings whose base lies outside the state.” Gellner (1994) offers civil society as the antithesis of the state, largely defined into existence based on its ability to counterbalance state activities (see also Kurzman 1995; Greer et al. 2017). Ingram (2020, p. 1) similarly identifies civil society by its disassociation with government, allowing for the inclusion of “schools and universities, advocacy groups, professional associations, churches, and cultural institutions [business sometimes is covered by the term civil society and sometimes not].”⁶

Acemoglu and Robinson (2019, p. 218) follow this trend in their formal model, referring to civil society as “autonomous societal organizations [what is sometimes referred to as “civil society”], capable of articulating demands and making suggestions…” Acemoglu and Robinson (2023, p. 407) reiterate this view of civil society as its own form of capacity, defined as “the ability to organize and engage in collective action. Like state capacity accumulated by elites, greater societal capacity adds to productivity because it allows greater coordination and public good provision.” As Dixit (2021, p. 1365) points out, this usage treats civil society as “a collectivity of individuals unanimous in their desire to provide and protect liberty for all members” as a simplifying assumption for the state capacity model.

Reiterating and Challenging the Basic Model

Despite the difficulty in defining either state capacity or civil society, state capacity literature utilizes a model of each checking the power of the other. Johnson and Koyama (2017, p. 3, emphasis added] argue that economies “governed by strong, cohesive, and constrained states are better able to overcome vested interests and avoid disastrous economic policies, while societies ruled by weak states are prone to rent-seeking, corruption and civil war.” Using “constrained” as an adjective does much work for the argument. For Acemoglu and Robinson (2019, 2023), the constraint arises from the strength of civil society in relation to that of the state. They argue that a “strong state is needed to control violence, enforce laws, and provide public services that are critical for a life in which people are empowered to make and pursue their choices. A strong, mobilized society is needed to control and shackle the strong state” (2019, p. xv). For this balance to maintain, state capacity and civil society capacity must expand simultaneously, with “state and society running fast enough to maintain the balance between them” in what is termed the “Red Queen Effect” (Acemoglu and Robinson 2019, p. 41).⁷

There are inherent issues with this construction. If state capacity is defined by a combination of despotic and infrastructural capabilities and civil society is defined as simply that which keeps the state in check, the model may become true-by-definition yet have little application. As noted, state capacity and civil society are both quite complex concepts. Societies are not single entities with a unified purpose. Rather, they are “almost everywhere and at all times are split by wide and deep crevasses along many dimensions: race, class, income, wealth, economic ideology, nationality or ethnic origin, and most importantly and most disastrously throughout history, religion” (Dixit 2021, p. 1365). Similarly, the state is not a single entity with a unified purpose. Rather, a “more sophisticated” model of state decision-making “would try to identify the interaction of such variables as political property rights, bureaucratic and administrative capacity, and interjurisdictional competition” (Piano 2019, p. 306). Buchanan (1949, p. 498) argues for an “individualistic theory”⁸ of the state, where the state “has its origin in, and depends for its continuance upon, the desires of individuals to fulfil a certain portion of their wants collectively. The state has no ends other than those of its individual members and is not a separate decision-making unit” (see also Coyne 2015; Coyne and Duncan 2019).

These limitations of utilizing the simplistic forms become apparent when applied to capacities in national security. The individualistic view of the state “offers a superior alternative for understanding state-provided defense because it does not assume away that which needs to be explained…Instead, individuals within the system maximize their own well-being subject to the constraints created by the political rules within which they act” (Coyne 2015, p. 379).⁹ Individuals maximizing their own well-being is similarly as important for modeling the behavior of society.

It is in this more comprehensive view of state and civil action that one finds the competition between state security agencies regarding resources and outcomes (Duncan and Coyne 2013a; Coyne and Duncan 2019) as well as the intermingling of state and private activities (Duncan and Coyne 2013b, 2015; Coyne 2013, 2015). The unified versions of each do not allow for divisions within the ranks. As neither the state nor civil society are entirely coherent and unified in their respective purposes, it stands to reason that a model of their interactions should provide for overlap between the two.

Fjelde and Soysa (2009, p. 6) somewhat acknowledge the overlapping nature in their definition of state capacity by separating it into “three conceptually and empirically different notions of state capacity…which can be summarized as coercion, co-optation, or cooperation.” Coercion relates back to the notion of state capacity as the ability to tax and otherwise directly control civil society. The latter two notions, however, offer something a bit different. High levels of state capacity may also allow governments to co-opt or influence civil society’s level of compliance [or the compliance of certain, differentiated parts of civil society] through “redistribution by way of high levels of public spending on political goods” (2009, p. 9). For national security, in particular, the political machinations of state and private entities are well documented and provide explanations for the waste and largess that accrue within the military-industrial complex (Melman 1971, 1985; Higgs 1987, 2006; Hartung 2011; Duncan and Coyne 2013a, b, 2015; Coyne 2015; Coyne and Duncan 2019).

The final notion, cooperation, relates to the level of trust that society places in the state, stemming from the states’ “ability to develop legitimacy and to invoke relations of loyalty and trust to secure the cooperation of societal actors and groups” (Fjelde and Soysa 2009, p. 10). Trust is a key component in Acemoglu and Robinson's (2019) modeling of the rise in U.S. surveillance state capacity. They acknowledge that the Danish state also created a similar surveillance apparatus to that of the U.S., but their citizens “trust that the Danish government will not use their IP addresses, port numbers, session types, and time stamps to snoop on them, suppress their freedom of speech, or imprison them because of their political views” (2019, p. 490). This level of trust allowed the Danish government to operate openly, rather than the seemingly secret way in which the U.S. surveillance state emerged.¹⁰ In Acemoglu and Robinson’s (2019, p. 490–491) argument, the U.S. security state arises only because it is secret, and it is secret because the U.S. civil society is so strong as to prevent a Danish model of operation. However, it is not purely the secrecy that leads to the rise of the surveillance state. Rather, it is a combination of coercion, co-optation, and cooperation from the individuals within the state and civil society that breeds the ground for its emergence. The implication here is that perhaps the state is not so weak, or that perhaps civil society is not so sufficient a check, as the model imagines.

The U.S. surveillance state in the 20th to the 21st century

The surveillance apparatus in the United States is not a new phenomenon. In 1974, the New York Times reported on the Central Intelligence Agency’s domestic surveillance of “the antiwar movement and other dissident groups in the United States” (Hersh 1974; Coyne and Hall 2018, p. 88; see also Acemoglu and Robinson 2019, pp. 334–335), through a variety of illegal means beginning as early as the 1950s (Hersh 1974). Civil organizations such as the ACLU and the Committee for Action/Research on the Intelligence Community [CARIC] were mobilized by this initial report. CARIC began publishing the CounterSpy magazine in an effort to raise public awareness (McCarthy 2008, p. 21). This pressure led to a two-year Congressional investigation of the of intelligence activities within the United State, nominally referred to as the Church Committee. The U.S. Senate Select Committee (1976, p. 169) report concluded that “virtually every element of our society has been subjected to excessive government-ordered intelligence inquiries” stemming from anti-dissident activities spanning the previous decades.

The litany of discovered abuses ran deep (Coyne and Hall 2018, pp. 84–86). The NSA¹¹ had accessed millions of private communications between U.S. and foreign citizens. The FBI¹² covertly surveilled prominent members of the civil rights and antiwar protests as well as those with leftist ideologies using a variety of methods. The CIA engaged in programs¹³ to infiltrate domestic organizations, even to the extent of fostering relationships with “local individuals—including police, political leaders, and college campus administrators” who would aid in their illegal surveillance. While these examples are illustrative, they are not exhaustive of the committee’s findings. The committee report resulted in Congress passing the Foreign Intelligence Surveillance Act [FISA] in 1978 (Taylor 2014; Coyne and Hall 2018, p. 89), whose purpose was to bolster protection against unreasonable search and seizure by creating and empowering the Foreign Intelligence Surveillance Court (FISC) to issue government warrants for domestic surveillance (Taylor 2014).

This period of societal pushback and the resulting legal changes are used as Acemoglu and Robinson’s (2019, pp. 332–336) shackling of the U.S. surveillance state. However, the constraints were not as secure as one may expect. From its construction, the FISC operated in a non-adversarial manner where only judges and government attorneys are present during cases (Coyne and Hall 2018, p. 89). This modus operandi allows court decisions to remain classified so that even those affected by the surveillance would be unable to appeal, as they would be unaware of the case entirely (Taylor 2014). The introduction of the court placated some of the public concern and allowed the surveillance state to operate under the radar for decades (Taylor 2014; National Constitution Center 2019) but did not prevent the surveillance state from operating.

As early as 1981, less than three years after the presumed shackling, the FBI began a years-long investigation into the Committee in Solidary with the People of El Salvador [CIPSPES], using the pretense to also surveil labor unions and church groups with only tangential connections (Gibbons 2019, p. 19).¹⁴ Though this investigation did lead to the eventual disciplining of six FBI agents once the operations were discovered (Gibbons 2019, p. 19), the FBI’s exploitation of the secretive surveillance court continued in other areas. In 2000, the FISC reviewed “extremely sloppy and questionable work by the FBI” and found “errors in approximately one hundred applications for FISA warrants” including misrepresentation about the investigations when submitting for approval (Bamford 2008, p. 67). Yet these errors did not prevent approval of the warrants as the FISC denied only 12 of the 35,333 surveillance warrant requests received between 1979 and 2013 (Coyne and Hall 2018, p. 89).

After the terrorist attacks on September 11, 2001, intelligence activities received additional governmental attention. The USA Patriot Act of 2001 broadened “the scope of government’s electronic surveillance, search and seizure, and other investigative powers” (Taylor 2014). The continuation and expansion of state surveillance was reported on several times in the subsequent years. In 2005, The New York Times revealed that the NSA had found avenues for circumventing FISC altogether, using a 2002 presidential order to monitor “the international telephone calls and international e-mail messages of hundreds, perhaps thousands, of people inside the United States without warrants over the past three years in an effort to track possible "dirty numbers" linked to Al Qaeda” (Risen and Lichtblau 2005).¹⁵ While ostensibly required to seek FISC approval for accessing domestic communications, the NSA relied on classified legal opinions to assert that the presidential authority allowed for warrantless domestic eavesdropping on U.S. citizens, permanent legal residents, and foreign individuals on U.S. soil (Risen and Lichtblau 2005). Under this rationale, NSA intercept operators¹⁶ were authorized to listen, record, and store the conversations of U.S. citizens such as foreign-based journalists communicating home to other journalists and even family members (Bamford 2008, p. 131). The New York Times report sparked a lawsuit by the ACLU, who battled the constitutionality of the program through the courts. Though the program was initially deemed illegal with an injunction issued by U.S. District Judge Anna Diggs Taylor, that decision was overturned by the Sixth Circuit Court of Appeals. The Supreme Court refused the case in February 2008, leaving the decision to allow continued surveillance in place (Free Speech Center 2009).

In 2013, surveillance became a media focal point again with the information leaks mentioned in the introduction. On June 5, 2013, The Guardian released the first report¹⁷ based on the Snowden documents, publishing a secret FISC order requiring Verizon to provide the NSA “on an ongoing daily basis” with “an electronic copy of the following tangible things: all call detail records or "telephony metadata" created by Verizon for communications (i) between the United States and abroad; or (ii) wholly within the United States, including local telephone calls.” The release of the order confirmed that the surveillance state was using Section 215 of the PATRIOT Act to indiscriminately collect communication records of millions of unsuspecting U.S. citizens who were not under any official investigation (Greenwald 2013).¹⁸ After the Guardian’s initial report, the Washington Post (2013)¹⁹ released a story on the NSA’s PRISM program, under which some of the largest Internet-based companies (Yahoo, Google, Facebook, and Twitter] were compelled to provide users online communications to the NSA. Under this program, the NSA used language in Section 702 of FISA Amendments of 2008, amendments to the 1978 act that supposedly shackled the surveillance state, to justify access to this communication so long as there was a reasonable belief that one party was located outside the United States (Kehl 2014, p. 4; see also Greenwald 2014, p. 73–75).

Public outcry after these reports did yield some changes to the U.S. surveillance state, providing evidence that civil society successfully exerted some pressure on the state to modify its behavior. The effectiveness of these modifications varies significantly, however, with some changes reducing rather than increasing constraints on the state. The USA Freedom Act of 2015 required the presence of an advocate for the public²⁰ during FISC cases. It also officially ended the NSA’s bulk metadata collection, at least when done by the agency itself.

The Freedom Act still permitted the agency access to significant amounts of data, requiring the phone companies to collect and hold the records but to provide copies to the NSA upon request (Savage 2018). While private entities technically hold the records, the state could “query” the providers for information about targets and those in contact with targets. The private entities then run the query through their databases and send the results to the NSA, allowing the agency to access the calling patterns of U.S. citizens with no suspicion of wrongdoing (Franklin 2019). A 2018 Office of Civil Liberties, Privacy, and Transparency (2018, p. 35) report indicates that in 2017, over 534 million phone records were queried, with at least 31,000 of these queries directly targeting U.S. persons.²¹ Human Rights Watch (2020) found that “[b]etween May 23 and December 31, 2018, the National Security Agency queried over 19 million phone numbers and collected over 434 million records under this program.” These records only include the legal queries as the NSA repeatedly failed in compliance checks by accessing data, through errors or otherwise, it is not authorized to receive (Savage 2018, 2019a; Sanchez 2018; ACLU 2019).

Following the ACLU’s 2019 report of further compliance failure, the NSA closed its domestic telephone data collection program. This closure was more the result of never actually thwarting a terrorist attack than the compliance issues (Savage 2019b). After this closure, Congress still approved a temporary extension of Section 215 from November 2019 to March 2020 (Whittaker 2019). As the March reauthorization deadline approached, the House and Senate failed in the reconciliation process over their competing bills. This lack of agreement led to expiration of authorities granted under Section 215, for any investigations not already underway²² (McKinney and Crocker 2020). While not insignificant, the expiration of Section 215 should not be misconstrued as the end of the U.S. surveillance state, as the loss of one authority does not end access to information under FISA or Executive Order 12333.

While Section 702 of the FISA Amendments Act of 2008 technically authorized the government to surveil non-U.S. persons outside the U.S., it has also provided the surveillance state access to U.S. citizen’s internet data. The Snowden leaks revealed the use of 702 has been used to surveil citizens provided one communicating party is outside domestic borders (Greenwald 2014, p. 126–127). This carve-out for “incidental” information was used by the NSA and FBI to surveil Muslim-American civil rights activists, lawyers, academics, and at least one political candidate under a provision intended to target terrorists and spies (Greenwald and Hussan 2014).

Like Section 215 under the Freedom Act, Section 702 had some changes under the FISA Amendments Reauthorization Act of 2017. Amendments in the six-year reauthorization²³ provided some constraints on the FBI’s access and use of collected communications, as well as increased the reporting requirements for the NSA.²⁴ However, the amendments did not end the NSA’s continued surveillance of internet metadata. They instead reinforced the use of many egregious [privacy-wise] practices of the NSA. Firstly, the reauthorization effectively codified the “abouts” collection process, where the NSA is permitted to collect communications not just to and from targets, but any communications that mention key selectors also (Crocker and Ruiz 2018). While this policy was officially deemed illegal and ended by the NSA in April of 2017 (NSA 2017), the reauthorization, officially signed in January 2018, legalized the practice so long as the Attorney General or Director of National Intelligence provides a written notice to the appropriate Congressional committees prior to reinstituting the practice. Secondly, the reauthorization expanded the NSA’s authority to engage in warrantless “backdoor search” to collect and store billions of “incidental” electronic communications of U.S. citizens who are not under formal investigation (Crocker and Ruiz 2018). The FBI is required to have a court order to access these databases, however there several caveat conditions²⁵ that provide the agency access without warrants (Greene 2018). These databases are also held so they may be queried at later points.

At the time of writing, the future of Section 702 is being debated as it nears its expiration date. Critics worry about the overbroad authorities it provides, but the surveillance agencies and administration have signaled strong support for reauthorization (Doshi 2023). However, evidence of abuse abounds. Though legally prevented from doing so, the FBI has been provided warrantless access to citizen’s data for investigations not involving national security in clear violations of the 4^th Amendment (Kaplan 2023). Between 2020 and 2021 alone, the FBI accessed improper information in the database more than 278,000 times, with queries ranging from information on citizens involved in the George Floyd protests to the January 6 Capitol riot (Doshi 2023). Recently released court records implicate the FBI of several counts of improper surveillance in 2022, searches that include information on a U.S. senator, a state lawmaker, and a state judge (Beitsch 2023). These instances highlight the issues with Section 702, but it is the long-running history of non-compliance by the NSA or FBI²⁶ (Baumohl 2023) that speaks to the inability of civil society to properly reign in the surveillance state.

Should Section 702 fail reauthorization or be edited during the process, the surveillance state will continue to have broad authorities under Executive Order 12333. Section 2.3 of the 1981 order permits intelligence agencies to collected and share information, including incidental information, “obtained in the course of a lawful foreign intelligence, counterintelligence, international narcotics or international terrorism investigation.” While intended to authorize surveillance abroad, the order has been interpreted as authorization for domestic surveillance as well. While the media has largely focused on Sections 215 and 702, a 2016 report by the Brennan Center noted that most foreign intelligence surveillance takes place under EO 12333 (Toh et al. 2016; see also Goitein 2023).

Despite the perception that U.S. citizens are unaffected by the order, many American communications are transmitted or stored abroad which makes them vulnerable to the NSA using its non-domestic authorization (Toh et al. 2016). Foreign individuals are not constitutionally protected under U.S. law, and communication records held abroad can be presumed to be owned by foreign individuals even if they include the information from millions of U.S. citizens (Arnbak and Goldberg 2015, p. 321). While data is stored in the U.S. where a secret federal court may grant the NSA access, at least the NSA is required to get a court order (Edgar 2015). That constraint does not necessarily exist once the data is offshore.

The extent to which agencies have used this loophole is unclear due to the secretive nature of foreign surveillance and the lack of reporting requirements regarding the use of EO 12333. However, Senator Wyden (2022) pressed for declassification of a Privacy and Civil Liberties Oversight Board report indicating the CIA had used EO 12333’s authority to secretly create its own bulk data collection program beyond the oversight of congress, the judiciary, and even the executive branch inherent in using the FISA authority. The nature of these records remains classified, but investigative reporting indicates the program has collected data in bulk over a lengthy period (Associated Press 2022; Oshin 2022). The reality is that despite amendments, reauthorizations, or even non-reauthorizations of surveillance laws, the surveillance state continues largely undisturbed.

Assessing the constraint

While certainly not a complete profile of state activities, the preceding section illustrates the extent of U.S. state capacity for domestic surveillance and the failure of civil society to restrain it. Acemoglu and Robinson (2019, p. 491) argue that “the U.S. program [of surveillance] was secret and kept expanding without oversight” and that “what was problematic in the response by the NSA and the CIA to new security threats wasn’t their expanding responsibilities and activities, but the secret and unmonitored ways in which they went about organizing them.” Secrecy does limit society’s capacity for checking state abuses, yet the history of surveillance provides numerous examples of media coverage bringing those abuses into public view. Media plays an important role in civil society capacity, but such coverage has not been sufficient in generating true constraints. The FBI alone has been repeatedly shown to violate civil liberties and constitutional rights, particularly during the civil rights movement and the post-9/11 years, yet “the bureau’s propensity for the policing of political dissent has remained largely unchallenged” (Speri 2019; see also Gibbons 2019).

Considering only the post-9/11 surveillance state, programs began in 2001,²⁷ were reported in 2005 (Risen and Lichtblau 2005), were comprehensively dissected in a 2008 book (Bamford 2008) and received extensive media coverage following the 2013 Snowden²⁸ leaks. Still, the court upheld them after 2005 despite pressure from a notable civil society organization like the ACLU. The programs were then largely reaffirmed by Congress in 2017. It took seven to 15 years²⁹ after the revelations for the Congress to fail to reauthorize one section of surveillance authority. The failure to reauthorize was also due more to 1) Congressional discord on which final version to approve, 2) other avenues for surveillance remained available, and 3) the bulk telephone data was ineffective at stopping terrorism. With the loss of Section 215, other similar programs continue and are routinely covered by the media. Even after the secrecy has been removed, civil society has been unwilling or unable to serve as a constraint. This failure, in part, stems from two related areas. The first is that civil society is often swayed towards increasing state strength during a crisis. The second is that civil society is not a single unit. At times, some factions of society find it beneficial to work with the state rather than against it, even if, or perhaps particularly if, that benefit comes at the expense of other factions.

Civil Society During Crises

Several significant expansions in state surveillance happen during periods of crisis in the U.S. Project SHAMROCK started during World War II, COINTELPRO during the U.S. Red Scare,³⁰ and CHAOS during the Vietnam War (Coyne and Hall 2018, p. 83–85). Section 215, 702, and EO 12333 were formed or given new interpretations following the 9/11 attacks. For each crisis, the state claimed new authorities and has not relinquished them following the crisis. Higgs (1987, 2004, 2006, 2007, 2012a) describes this phenomenon as the “ratchet effect,” where “after each major crisis, the size of government, though smaller than during the crisis, remained larger than it would have been had the precrisis rate of growth persisted during the interval occupied by the crisis” (Higgs 1987, p. 30).

It is not simply that the state expands during this time. During a time of crisis, the growth in state capacity becomes more tolerable to the population. Civil society tends to welcome the increase to “solve” the crisis, and the increased capacity becomes relatively acceptable over time. As the crisis subsides, the “government actions no longer startled the citizenry; indeed, many Americans, including highly regarded intellectuals and top policymakers, had come to regard them as desirable” (Higgs 2004, p. 171; see also Higgs 2008). Crises provide a moment of weakness for civil society. The state expands. The expansion of activities yields cultural changes, further altering society’s perception in favor of the state. Such a shift following 9/11 is seen in a variety of forms, including the acceptance of the TSA, the initial acceptance of the wars in Afghanistan and Iraq, and, here, the initial acceptance of surveillance to prevent another attack. Even as organizations such as the ACLU (2001) and the Electronic Frontier Foundation (2001) argued for restraint regarding and warned against the then-new passage of the Patriot Act, public approval of Bush-era policies remained high for several years (55–60% answered the polices were “about right”) (Gallup 2003).

Matters of national security are particularly prone to this type of ratchet. This susceptibility, as Acemoglu and Robinson (2019) note, is in part due to the secretive nature of the national security apparatus. It is not necessarily because the system grows without oversight as they indicate but stems from the fact that the public is not privy to real-time information on national security threats. Information asymmetries are quite prevalent in national security matters, reducing the effectiveness of feedback that can be provided by civil society (Koppl 2018, p. 230; Coyne and Hall 2021, pp. 33–34). Rather than society directing level and direction of security efforts, the public must rely on national security experts to provide information regarding these potential threats and to make choices on society’s behalf with only minimal input (Koppl 2018, p. 230). These experts, however, are rarely unbiased in either their evaluation or their choices to counter such threats.

The asymmetric nature of the information allows for the selective release of information to the public, a release that may be selected not based upon transparency but on the intention to generate a specific, strategic response from the public (Coyne and Hall 2021). Information may be withheld under the flawed assumption that is in the best interest of the public (Coyne and Hall 2021, pp. 39–41), but it may also be withheld for more nefarious reasons as well. Those who are supposedly in the best position to assess the risk often have a direct incentive³¹ to overinflate it (Meuller 2006, pp. 33–43). As a result of these incentives, members of the surveillance state often provide the public with the worst-case scenarios—touting scenarios that are “not impossible”—rather than those more probabilistic—downplaying the “likelihood” aspect of the scenarios (Meuller 2006, pp. 43–47). As part of these strategic releases of information, estimates provided by experts and internalized by civil society often suffer from probability neglect—where the likelihood of attack and capacities of the enemy are often overinflated—and cost neglect—where the cost of increasing security is often underestimated (Meuller and Stewart 2011, pp. 13–23). The TSA is a prime example of these biased estimates. Despite the relative safety of flying, the agency has repeatedly sold the public on its own necessity, inflating its own effectiveness while downplaying the cost it imposes on the budget and the public (Meuller and Stewart 2011, pp. 132–158; Coyne and Hall 2021, pp. 107–139). Potentially similar hyperbole has been used in the argument to renew Section 702. The intelligence community credits Section 702 with effectively protecting Americans from threats ranging from China and Russia to fentanyl to ransomware while downplaying the community’s repeated abuse of the authority (Iyengar and Detsch 2023).

When society is overestimating the threat and underestimating the costs of protection, it leads to a scenario where civil society allows for a relatively large ratchet during a national security crisis. Such a ratchet is particularly likely while society is still in the shock of the crisis and is more willing to accept the strategically released information as truthful. Chanley (2002, p. 469) notes that after 9/11, trust³² in the U.S. government “rose to a level not seen since the mid-1960s.” It is not surprising that similar surveillance state capacities get built out in that period as the ones in the 1960s. Though the level of trust may decline over the years that follow the crisis, the state capacity has already become entrenched in the system.

Collective Action is Not Unanimous Consent

The preceding section begins to illustrate the idea that the “state” is not a single entity with a unified interest. This section elaborates on that point for both sides of the state capacity model. Though modeled as distinctive units, the state and civil society often blur the lines between one another. It is more appropriate to model both the state and civil society as an “entangled” mixture of heterogeneous interests locked into a continuous struggle of cooperation and conflict (Koppl 2018, p. 228; see also Wagner 2010). An accurate description of state activity will include “the inherent messiness of collective action” (Coyne and Duncan 2019, p. 240). Similarly, an accurate description of civil society will include an understanding that democratic systems and politics³³ may lead to “incoherent, chaotic, or dictatorial” results rather than a unified interest (Lemieux 2021, p. 49). In many instances, the state may be “actively [and] strategically trying to disrupt society’s collective action effort, or to form coalitions with one subset of society to favor themselves and that subset while harming others” (Dixit 2021, p. 1369). Rather than serving as a unified block, some portions of society may find it beneficial to work with the state. The availability of non-state interests willing to aid the national security state has been well documented (Melman 1971, 1985; Friedman and Friedman 1984; Higgs 1987, 2004; Bacevich 2007, 2010; Duncan and Coyne 2013a, b, 2015; Coyne and Duncan 2019). The ability for the state to “penetrate civil society” (Mann 1984, p. 189) through co-optation or cooperation (Fjelde and Soysa 2009) depends on how easily individual sections of society may be co-opted. Given that society is susceptible to these overtures during a crisis for the reasons above, the ease varies with the conditions faced.

Heldring and Robinson (2023) illustrate an extreme example with the Rwandan state’s co-optation of the citizens’ social networks to conduct the 1994 genocide. Later, the state would use a similar co-optation to implement agricultural reforms. In both cases, the state utilized existing networks. This enabled leaders to use civil society to develop “more state capacity because it was more connected to society and was able to use and activate social networks more effectively” (Heldring and Robinson 2023, p. 402). Instead of checking state power, “these networks allow the state to capture society and this is what creates state capacity” (Heldring and Robinson 2023, p. 404, emphasis added). This concept of state capacity as the state’s level of embeddedness within society speaks directly to society’s ability to serve as a check against state power. If the state is embedded within society, it may be working with rather than against society.

Less extreme than genocide, the U.S. surveillance state has also found willing partners within society to increase its capacity. Several portions of the private sector found it advantageous to help build the state’s surveillance capacity. Recall from the surveillance state history that 1) ITT, RCA, and Western Union aided the NSA with SHAMROCK, 2) political leaders and college administrators, who would be connected to and likely key figures in civil society,³⁴ aided the CIA with CHAOS, 3) AT&T allowed the NSA to build its own offices at the company’s key internet traffic hubs³⁵, 4) Verizon provided the NSA daily telecommunications information, and 5) Yahoo, Google, Facebook, and Twitter provided similar information about online communication.

More recent examples also illustrate how the state has used existing social and communications networks increase its own capacity, where “private data firms become appendages of the national-security surveillance state” (Alshamy et al. 2023, p. 26). In the world of “surveillance capitalism,” the information available for access continues to grow. Though social media is viewed as a power force for building the capacity of civil society (Shirky 2011), it does not come without risks of state co-optation. Social media organizations routinely track the geolocation information from their users to generate marketing and other data sources. While the “companies that store and transmit user data are generally prohibited from “knowingly” sharing those records with the government” under the Stored Communications Act of 1986, they are allowed to sell the collected information to third party data brokers where the legal restriction is weaker (Edelman 2020). In 2022, the Electronic Frontier Foundation (EFF) reported on the Fog Data Science’s “Fog Reveal” application that has collected “billions of data points about ‘over 250 million’ devices and that its data can be used to learn about where its subjects work, live, and associate” and sells this “raw location data about individual people to federal, state, and local law enforcement agencies” (Cyphers 2022). EFF was able to access records indicating that “Fog has past or ongoing contractual relationships with at least 18 local, state, and federal law enforcement clients; several other agencies took advantage of free trials of Fog’s service” (Cyphers 2022). An interesting feature of this type of arrangement with private entities is the legal loophole it provides. While there may exist some protections against government attempts to “acquire sensitive personal data like your location without a warrant, sketchy data brokers can and do sell this information directly to the government” over the open market where it may be purchased with little legal oversight (Kovacevich 2023). In 2023, TechCrunch provided information regarding the sale of privately collected geolocation data to the National Guard, Immigration and Customs Enforcement, the “FBI, the military, and several other government and regulatory agencies” (Kovacevich 2023). It is not just the data used either. In 2023, the House Judiciary (2023) revealed that the FBI had again attempted to infiltrate “local religious organizations as ‘new avenues for tripwire and source development’…contacting so-called ‘mainline Catholic parishes’ and the local ‘diocesan leadership’” to increase surveillance capacity.

The extent to which this engagement between the state and society is cooperative or co-optation varies and is often difficult to distinguish between the coercive instances. Society may be eager to cooperate, as with AT&T’s partnership with the NSA (Bamford 2008, p. 181; Harvard Law Review 2018, p. 1725) or Fog Data Science’s brokerage to government agencies (Cyphers 2022). It may also be less than willing. The extent to which Google and Facebook complied willingly with the intelligence agencies is unclear, while there is evidence that “at least some tech companies—including Yahoo and Twitter—challenged national security–related requests long before the Snowden disclosures” (Harvard Law Review 2018, p. 1728). Yet whether the state can encourage or incentivize through patriotic duty or financial gain, or whether it compels society, the result is often the same. In either case, civil society is actively helping to increase state capacity rather than constrain its use.

Conclusion

The preceding sections are intended to make clear the relative ineffectiveness that civil society has in constraining state surveillance capacity. The U.S. is noted to have an exceptionally strong, vibrant civil society, and yet it has repeatedly failed to shackle the state, particularly in matters related to national security. This ineffectiveness is noteworthy for the general model of the state capacity-civil society tradeoff. Firstly, reliance on a strong state is an inherently risky proposition. This failure from a purportedly strong civil society raises the question of whether any civil society would be sufficient to check a strong state. It may, in fact, be dangerous to encourage increasing state capacity under the assumption that civil society will provide a valid check to state predation. Such a strategy may yield negative results for both economic growth and civil liberties when society falls short of those lofty goals.

Secondly, even if a stable balance is momentarily achieved, an exogenous shock may destabilize it. Momentary snapshots that indicate state capacity is more positive than negative may not hold after a crisis. These crises may take many forms—war, terrorist attacks, sharp economic downturns, pandemics—that are difficult to anticipate. When they occur, the strength of the civil society may do less to constrain the state’s predation as it utilizes that strength to aid the state during the shock. Once done, these shifts between constraint and ally are difficult to undo.

Thirdly, modeling this balancing act requires more nuance to understand the players in the game. Using too simplistic a definition loses sight of a wide variety of incentives, and entanglements do not fit so neatly into the black boxes. Neither civil society nor the state should be treated as a monolith. Treating either as monolithic entails obscuring the heterogeneous cooperative and antagonistic relationships that different people and groups within civil society have with the state. This paper highlights efforts that various groups in civil society have taken to resist and restrain the surveillance state as well as instances when actors in civil society empowered the surveillance state. Future research on state capacity must not take this heterogeneity for granted. Many scholars outside of economics have done detailed work on the interactions between civil society groups and the surveillance state (e.g., Robé 2023; Lehoucq and Tarrow 2020). Economists can use rational choice theory and political economy to build on this work, thereby developing more realistic accounts of the relationship between civil society and the surveillance state.