Sage Journals: Discover world-class research

Abstract

How should we understand the surveillance state post Snowden? This paper is concerned with the relationship between increased surveillance capacity and state power. The paper begins by analysing two metaphors used in public post Snowden discourse to describe state surveillance practices: the haystack and the panopticon. It argues that these metaphors share a flawed common entailment regarding surveillance, knowledge and power which cannot accurately capture important aspects of state anxiety generated by mass surveillance in an age of big data. The paper shows that the nature of big data itself complicates the power attributed to mass surveillance states by these metaphors and those who use them. Relying heavily on Ezrahi’s distinction between information and knowledge, the paper situates this argument concerning the state and anxiety borne of information overload in the context of literature that concerns the state and information management. Drawing primarily on James Scott’s work on legibility, it argues that the big data born of mass surveillance problematises the concept of information as empowering the state. Instead, understanding mass surveillance in an age of big data requires understanding the relationship between the surveillance state and information in terms of anxiety as well as power.

Keywords

Surveillance intelligence risk terrorism Snowden panopticon

This article is a part of special theme on Data Associations. To see a full list of all articles in this special theme, please click here: http://journals.sagepub.com/page/bds/collections/data-associations.

Transparency is sunshine, corruption is a cancer and our homes are where our hearts are: our conceptual system is metaphorically constructed, and everyday concepts are usefully and regularly understood in terms of other concepts – a type of cognitive scaffolding is at the root of our linguistic structures. Indeed, complex topics, like anti-corruption, are commonly understood via multiple metaphors like, for example, sunshine and cancer. Lakoff and Johnson (1980) argue that although such metaphors are usually incompatible, they combine to make sense of complex concepts by sharing underlying ‘common entailments’ about the meaning of the concept involved and are in fact essential to doing so. In this case, for example, sunshine and cancer arguably share the common entailment: ‘corruption is unseen’. This paper is concerned with two such metaphors which mark popular and political discourse concerning post 9-11 mass surveillance, particularly in the wake of the Snowden revelations: the panopticon and the haystack. The paper begins by tracing the long history of the panopticon metaphor and highlighting its applicability to post 9-11 surveillance. It goes on to analyse the far more recent use of the ‘needle in a haystack’ metaphor by security agencies as a way to justify mass surveillance. It argues that although the metaphors appear inconsistent they share a common entailment, which helps users to make sense of surveillance. Here, that common entailment is the claim that ‘knowledge is power’.

The bulk of the paper situates this common entailment in the context of the information state. Following Aradau and Blanke (2015) the paper shows how the nature of big data itself complicates the power attributed to mass surveillance, and indeed the common entailment of ‘knowledge is power’ which is shared by the haystack and panopticon metaphors. Relying heavily on Ezrahi’s (2004) distinction between information and knowledge, the paper situates its argument regarding the state and anxiety borne of information overload in the context of literature that concerns the state and information management. Drawing primarily on James Scott’s work on legibility, it argues that the big data born of mass surveillance problematises the concept of information as empowering the state. Instead, I argue that understanding mass surveillance in an age of big data requires understanding the relationship between the surveillance state and information in terms of anxiety as well as power. Post 9-11 mass surveillance is driven by the fragility of state power in an age of big data rather than simply the untroubled exercise of that power.

This argument is progressed via the interrogation of the panopticon and haystack metaphors. Metaphors provide a way for users to understand complex concepts by trading in the language of the familiar. They can trade on intuitive understandings to explain the new in older and more familiar terms (Betz and Stevens, 2013: 15). Metaphors do not only facilitate communication: they can shape models of the world and modes of action (Lakoff and Johnson, 1980). This is particularly pertinent in the field of security, where, as Bobrow (1996: 436) contends, the ‘choice of metaphor carries with it implications about contents, causes, expectations, norms, and strategic choices’.

The panopticon metaphor dates from the pre-digital era and is inherently political given that it concerns the way the state manages flows of information. It is also critical, used by activists and others seeking to interrogate government policy. The ‘needle in a haystack’ metaphor is by contrast a pre-existing metaphor which has come to be used in a political sense in the big data era. It has been adopted by government to explain government processes and goals and as a result is also inherently political in the same way as the panopticon metaphor in that it concerns the way the state manages flows of information.¹ For Edelman (1964: 124), the process of using older metaphors to explain new phenomena results in a ‘dulling of critical faculties’. This paper highlights that dulling by showing that neither metaphor adequately depicts the relationship between big data, mass surveillance and the state in the post 9-11 era.

The Snowden revelations: Mass surveillance on an unprecedented scale

The Snowden papers are well traversed, but the details bear repeating here. The papers revealed an astounding level of international surveillance, driven by the US but involving extensive collaboration with intelligence partners and global internet companies. The programmes aim to collect what General Keith Alexander, former chief of the US National Security Agency (NSA), has called ‘the whole haystack’ – meaning an almost limitless range of metadata, which the agency stores for five years, possibly longer.

Such surveillance is a creature of the post 9/11 era, driven by the US’s failure to predict the attacks of 9/11. Such surveillance is not limited to the US: the programmes involve extensive collaboration within the Five Eyes community comprising the UK, the US, Australia, NZ and Canada. In 2013, for example, Le Monde published documents from Edward Snowden’s archive showing that the NSA obtained 70 million French phone metadata records in one month (Greenwald, 2014).

Mass surveillance differs from other forms of surveillance because it relies on the collection and manipulation of massive data sets. A massive technological system, it is conceived of and used as a data set rather than individual data points – as ‘the whole haystack’ rather than, for example, the single street surveilled by a street camera. As above, this data set is comprised of metadata, and the results of such surveillance fulfil Boyd and Crawford's (2012) definition of big data: where the results of the surveillance are made useful to collectors via manipulation of data rather than, or as well as, the data sets themselves.

The big data resulting from mass surveillance is, then, what Ezrahi (2004: 257) describes as ‘information’, as distinct from other forms of understanding about the world. Ezrahi distinguishes information from more generalisable and complex ‘knowledge’ about the world. He describes information as ‘more restricted to the technical practical surface of knowledge’. He argues that ‘when it is represented by information rather than by knowledge, “reality” can be flattened and simplified…’. In contrast, knowledge – a different form of knowing – is more organised and systematic, privileging clarity and rationality and making a sharp distinction between truth and error.

Panopticons, haystacks and the information state

The metaphor of the panopticon is a familiar one and bears little recanvassing here: a short overview will suffice. The metaphor, of course, draws on Bentham’s designs for a prison that reflected his interest in new forms of social control. The design encouraged prisoners to modify their behaviour believing they were permanently visible under an ever-watchful gaze. For Foucault, and in most popular usage, the panopticon is a metaphor for the information state – for the way the state desires, accesses and uses information. The panopticon metaphor portrays the (modern) state as a unitary actor not only hungry for information but able to acquire and use it to inform all-powerful knowledge claims about the world. For Foucault, the panopticon represented the modern state’s powerful ability to discipline the individual to modify their own behaviour in the state’s interests, lessening the drain on the state’s maintenance of its own power. This ability was drawn specifically from the state’s ability to collect information. Foucault (1979: 205) wrote:
The efficiency of power—its constraining force has, in a sense, passed over to the other side—to the side of its application. He who is subjected to a field of visibility and who knows it, assumes responsibility for the constraints of power; he makes them play spontaneously upon himself; he inscribes in himself the power relation in which he simultaneously plays both roles; he becomes principle of his own subjection.²
The panopticon has long dominated scholarly analysis of surveillance (Caluya, 2010; Epstein, 2016; Lyon, 2006: 4): scholars and the broader commentariat use it to frame discussions of an omniscient state and the single, unswerving gaze of state security forces. However, surveillance studies have moved beyond the panopticon metaphor in an effort to account for the myriad sources of surveillance in the modern world, the ways in which they interact and the way in which data they produce flows between them (Ball, 2010; Haggerty and Ericson, 2000). Such scholars argue that the panopticon metaphor cannot account for the nature of surveillance in the late modern era. They argue that new technologies mean that there are no longer discrete systems of surveillance, as new technologies converge and a multiplicity of actors engages in practices previously limited to a powerful, unitary state. These new actors, including non-state actors, may comprise companies, individuals and institutions for example, rather than or alongside the panopticon. Rather than emphasising a centralised power structure such scholars take a Deleuzian approach which emphasises ‘surveillant assemblages’, meaning that surveillance takes a rhizoid form rather than the unitary hierarchical approach of the panopticon. They describe how individuals are disassembled and recomposed into ‘data doubles’ of their physical selves, through fluid network flows. For Haggerty and Ericson, this new form means that power is deconstructed rather than easily demarcated in a hierarchical form. This means in turn that new surveillance technologies enable new forms of resistance, countersurveillance and democratisation (Haggerty, 2006; Haggerty and Ericson, 2000).

Despite this influential scholarly repudiation of the panopticon metaphor, in the years following Edward Snowden’s astounding revelations of mass surveillance by the United States government via the NSA, its allies and commercial partners, the overwhelming metaphor employed in public discourse to describe and understand this phenomenon has been the panopticon.³ Betz and Stevens (2013: 7) point out that a similar process has occurred with the term ‘cyberspace’, which persists in political discourse where it has passed out of favour in academia. Indeed, Snowden himself used the metaphor when describing the projects he had revealed, noting of mass surveillance that: ‘… if a surveillance program produces information of value, it legitimizes it… In one step, we’ve managed to justify the operation of the Panopticon’ (Gellman, 2013).

Surveillance scholars have resisted applying the panopticon metaphor to post Snowden revelations. Ruppert (2012: 118), for example, suggests that mass surveillance via big data techniques is akin to efforts to cut costs in the delivery of services rather than a ‘sinister’ all-seeing manoeuvre by the state. This argument points out that rather than an all-knowing state, such surveillance results only in a state populated by a ‘plethora of partial projects and initiatives’ rather than one, unitary ‘inspector’ in a panopticon. The following paragraphs support this argument by showing that attempts to collect mass surveillance data do not necessarily result in an all-seeing, all-knowing state. However, attempts to effectively downgrade post 9-11 mass surveillance as anything but ‘state attempt(s) to keep close watch on all citizens’ (Lyon, 2014) do not hold if we note the implementation of mass surveillance only after 9-11 via the PATRIOT Act. The timing of the introduction of mass surveillance on the scale Snowden reveals suggests that the data borne of such surveillance is related to the efforts of a centralised state to see all via a unified field of vision, even though, as Lyon points out, the process of collection is in practice undertaken via many partial processes. For Snowden himself and many others, post-2001 mass surveillance recalls the panopticon and the accompanying subjugation of citizens by embodying at the very least an attempt by a centralised security state to engage in a unitary, omniscient gaze over its citizens (Greenwald, 2014; Sanchez, 2014). Although in this case the panopticon is not necessarily effective, it arguably still stands as a metaphor for the intentions of the post 9-11 information state.

We can see how this metaphor has been applied so readily. Snowden’s revelations uncovered a huge network of surveillance that sought to make all individuals concerned permanently ‘visible’ in the sense of collecting as much data as possible, largely indiscriminately. Examples abound of individuals targeted suddenly and without warning, and of the role of mass surveillance in highly efficient targeted drone killings. In this use of ‘panopticon’, the NSA is thus the central tower in the panopticon, seeing its targets as individual data points like backlit prisoners in cells. One might easily argue that mass surveillance also seeks to limit collective behaviour, much like the panopticon because in using data mining algorithms to identify previously unseen connections the system seeks to stop individuals acting together in terrorist plots. In addition, up until Snowden’s revelations the NSA programmes and those of its allies were entirely secret, removing the seen/seer dyad from the equation. As Foucault (1979: 201–202) notes: ‘The panopticon is a machine for dissociating the seeing/being seen dyad: in the peripheric ring, one is totally seen, without ever seeing; in the central tower one sees everything without being seen’. When applied to post 9-11 surveillance as revealed by Snowden, this metaphor resonates with⁴ the anxiety of advocates of free speech who see the mass surveillance as having a chilling effect, for example, or the ire of privacy activists, who see the surveillance as limiting the right to privacy – to an interior life unfettered by the state. The persistence of panopticon imagery is, then, due to the sense that the NSA is omniscient, that it is omnipotent and that it is beyond the resistance of ordinary citizens. As Foucault (1979: 201) notes: ‘The panopticon is a marvellous machine which, whatever use one may wish to put it to, produces homogenous effects of power’.

A second metaphor commonly used to describe post 9-11 mass surveillance practices is that of a ‘needle in the haystack’, where a surveillance target – a terrorist – is a needle in a haystack of data. Unlike the panopticon metaphor, which is used widely by those who object to the power they see as inherent in post 9-11 mass surveillance, the haystack metaphor in its big data is primarily associated with General Keith Alexander: former head of the NSA and architect of post 9-11 mass surveillance and the agency’s ‘collect it all’ approach. General Alexander famously argued that mass surveillance is essential to keeping Americans safe at home and to progressing America’s security goals abroad. He argued that bulk collection of metadata is justified because ‘You need the haystack to find the needle’ (Gellman and Soltani, 2013). A former NSA official said of Alexander’s approach in an early mass surveillance programme in Iraq, for example, that ‘Rather than look for a single needle in the haystack, his approach was, “Let’s collect the whole haystack,… Collect it all, tag it, store it. . . . And whatever it is you want, you go searching for it”’ (Nakashima and Warrick, 2013). The same official said of Alexander that: ‘He is absolutely obsessed and completely driven to take it all, whenever possible' (Nakashima and Warrick, 2013).

A similar argument has been made by other officials, including James Cole (2013), the former deputy Attorney General, who told a congressional committee in 2013: ‘If you’re looking for the needle in the haystack, you have to have the entire haystack to look through’. Jeremy Bash, chief of staff to Leon E. Panetta, the former CIA director and Defence secretary, justified mass surveillance in a similar way in 2013: ‘If you’re looking for a needle in the haystack, you need a haystack’ (Schmitt et al., 2013). In this usage, the power to collect the haystack and find the needle is assumed rather than made explicit. Here, the surveillance system not only should be allowed to collect the ‘whole haystack’ but without a doubt can, and will by doing so be able to track down the ‘terrorist that walks among us’, as Alexander (2013b) argued.

Both the panopticon and the haystack metaphor implicitly assume an omniscient, centralised power behind and benefitting from post 9-11 mass surveillance and the data it produces. In Lakoff and Johnson’s (1980: 205) terminology this is an ‘overlapping entailment’ between the two otherwise inconsistent images which makes them coherent. Importantly, this assumed power is drawn from the information the system collects, that is both metaphors present an image of the information state. In the panopticon metaphor, the information the system collects means the state has the power to target and act on any individual, anywhere. In the haystack metaphor such information means similarly that the state can target and act on any individual. In Ezrahi’s (2004: 303) terms the panopticon and haystack metaphors frame the big data borne of surveillance as ‘knowledge’ rather than information. That is, they see the results of mass surveillance as producing a total organised, rational and systematic understanding of the world – absolute knowledge. It is this understanding from which the system and the state’s power is drawn. The shared entailment between the two metaphors, then, is best encapsulated as ‘absolute knowledge is absolute power’, where knowledge is based on the massive information mass surveillance collects. In these metaphors, the transformation of mere data into big data arguably drives the transformation from mere information into the more systematised, rigorous, and truthful ‘knowledge’ from which the system derives its power.

In their shared entailment the panopticon and haystack metaphors engage a particular understanding of the information state as (a) interested in collecting information and collating knowledge on its citizens and (b) capable of doing so. Arguably the most influential account of the relationship between information and the state is James Scott’s work on legibility and the state. Scott’s concept of legibility describes the ways in which modern states seek to make visible their units of interest – their citizens, their land, their income. Making these units legible means bringing them into being as data points of information, labelling them and categorizing them in such a fashion that they may ‘be centrally recorded and monitored’ (Scott, 1998: 2). Such processes include, for example, the invention of freehold tenure, the creation of permanent last names and the standardisation of city design. For Scott, legibility allows states to intervene in the lives of their citizens in ways that seek to improve the lives of those citizens and also, by extension, buttress the state. At first, the link between the creation of – for example – surnames in 14th-century Great Britain and mass surveillance seems marginal. But both these processes bring within the state’s field of vision that which it might not otherwise see and do so for the purposes of solidifying and improving the position of that state.

An important consideration regarding the link between legibility and mass surveillance is Scott’s concept of legibility and the notion of administrative order. Scott’s analysis posits that state processes of making units legible are linked to desires for administrative ordering: for Scott (1998: 11), documentary techniques for the inscription and registration of data allow the state, through representation, to achieve an ‘overall, aggregate, synoptic view of a selective reality… making possible a high degree of schematic knowledge, control and manipulation’. Scott’s work sees such schematic knowledge and control as fuelling interventions designed to provide specific outcomes such as increased agricultural production or social order. Mass surveillance is also arguably concerned with specific outcomes – the safety of citizens – with predictive algorithms doing the work of ‘ordering’ big data.

The damage done

However, rather than simply demonstrating and delivering absolute power the big data generated by post 9-11 surveillance in many ways problematises such power in a way captured adequately by neither metaphor, and in a manner which destabilises Scott’s framework. This section of the paper proceeds by first showing that mass surveillance has failed to prevent catastrophic terrorist attacks in the post 9-11 era. It then outlines three features of the big data produced by such surveillance which show that the common entailment of ‘absolute knowledge is absolute power’ inherent in both the haystack and the panopticon metaphors is destabilised by the realities of mass surveillance.

Despite high-level protestations as to the utility of the data provided by mass surveillance, the NSA and associated agencies within the US and its allies have failed to present any significant cases which demonstrate the utility of such surveillance in stopping catastrophic terrorist attacks. Initially, the NSA defended its programmes in the wake of the Snowden revelations by arguing that the surveillance had prevented such attacks. In the most infamous of these claims the then head of the NSA General Alexander (2013a) said in Congressional testimony that the NSA’s surveillance programmes had contributed to understanding or disrupting ‘fifty-four different terror-related events’. At a conference earlier that year his presentation had claimed that the NSA’s activities had thwarted 54 attacks, with the metadata programme intended to ‘find the terrorist that walks among us’ (Alexander, 2013b, 2013c). President Obama echoed this in 2013, stating: ‘We know of at least fifty threats that have been averted because of this information’ (Alexander, 2013a, 2013b; Calmes, 2013).⁵

However, at a Senate hearing later in 2013, General Alexander (2013d) admitted that the 54-plot statistic was wrong. He admitted that only one plot had been foiled, and that that was not a plot but rather a case involving a small amount of funds transmitted to a banned terrorist group in Somalia by an American citizen. An investigation by the New America Foundation found that the contribution of bulk surveillance to preventing terrorism was minimal, and that traditional methods of intelligence gathering were far more effective. The report found that bulk surveillance had played a role in initiating at most 1.8% of 225 individuals recruited by AQ or a like-minded group and charged in the US with an act of terrorism since 9/11 (Cahall et al., 2014).

Indeed, almost every major terrorist attack on Western soil in the past 15 years has been committed by people who were already known to law enforcement – that is who were under mass surveillance already. These individuals were not only likely under mass surveillance but are also likely to have been – or should have been – under more particularised surveillance. For example, one of the gunmen involved in the Charlie Hebdo attack in Paris in 2015 had previously been in prison for recruiting jihadist fighters. Umar Farouk Abdulmutallab, the underwear bomber, was arrested and interrogated by the FBI in 2009 and was reported to the CIA by his own father just before he attempted to blow up an airliner over Detroit in 2010 – an attack thwarted by a fellow passenger rather than surveillance.

Similarly, the leader of the London 7/7 bombers had been under surveillance for some time, and the planners of the 2008 Mumbai attacks were under surveillance by the United States, the United Kingdom and India. One of the brothers accused of bombing the Boston Marathon was the subject of both an FBI threat assessment and a warning from Russian intelligence. An individual who beheaded a priest in France in July 2016 was known to authorities and was wearing an ankle bracelet at the time of the murder. In each of these cases, authorities were not lacking in data: authorities simply failed to use the data they already had.

How to explain then, a system which both metaphors – the panopticon and the haystack – frame as all-consuming and all powerful, but which actually fails to prevent terrorist attacks? The following paragraphs argue that it is the nature of the data the system so powerfully collects which problematises that power.

What is a needle?

Post 9-11 mass surveillance aims to stop individuals from committing acts of terror. In doing so, the system, via its designers, makes decisions about what sort of person it is interested in, and what sorts of events: in the giant haystack in which it searches it makes a decision about what a ‘needle’ is. This means creating an event/person profile which tunes the algorithms used to interrogate the data collected. Profiling depends on the categorisation of social groups and statistical computation of risk.

The problem with post 9-11 mass surveillance, however, is that although it engages in profiling of a sort, this profiling is post hoc. The bulk collection of metadata means that everyone is a potential terrorist: the terrorist does not mark him or herself as a ‘needle’ via a particular profile beyond the most obvious. Indeed, the 9-11 attackers themselves went undetected, while the Home Office report on the 2005 London bombings found that nothing marked out the four men involved in the attacks. They were all ‘unexceptional’. This means that the system is doomed to collect as much information as it possibly can to detect a profile it does not yet know for certain exists. As one former NSA analyst puts it: ‘If you target everything, there’s no target’ (Schwartz, 2015). Even Snowden admits to this failing of mass surveillance, noting: ‘The problem is that when you collect it all, when you monitor everyone, you understand nothing’ (Maas, 2015).

If everyone and everything is potentially a terrorist, then, but terrorist attacks are themselves rare, the system suffers from a problem of statistical analysis in huge data sets known as the base rate fallacy. The base rate fallacy applies in many fields other than counterterrorism surveillance, such as the search for rare medical conditions. It means that the potential for false positives is extremely high because of the importance and yet rarity of the event itself, and the profile of the elements involved (see Amoore and De Goede, 2005; Levi and Wall, 2004). Importantly, however, the cost of such false positives in counterterrorism is higher than in other fields. It may induce panic in a population, reveal the extent to which government is infringing its citizens’ civil liberties and reveal the operation of otherwise secret surveillance mechanisms. Furnas (2012) explains the problem as follows in discussing FAST, an NSA programme revealed by Snowden as designed to identify people in airports who are ‘intending to commit a terrorist act’, by tracking changes in their physiology:
… let’s assume for a moment that 1 in 1,000,000 people is a terrorist about to commit a crime. Terrorists are actually probably much much more rare, or we would have a whole lot more acts of terrorism, given the daily throughput of the global transportation system. Now let’s imagine the FAST algorithm correctly classifies 99.99 per cent of observations–an incredibly high rate of accuracy for any big data-based predictive model. Even with this unbelievable level of accuracy, the system would still falsely accuse 99 people of being terrorists for every one terrorist it finds. Given that none of these people would have actually committed a terrorist act yet distinguishing the innocent false positives from the guilty might be a non-trivial, and invasive task.⁶
The problem of false positives intensifies as a result of tools used to interrogate large volumes of data. As Aradau and Blanke (2015: 8) note: ‘…the error in data mining methods actually rises if more and more data is collected, which is contrary to the belief that an “ocean of data” makes it possible to identify “gossamer contacts” with more certainty’. Indeed, leaked details of the NSA’s SKYNET programme suggest that the problem of false positives is intensified when data analytics are driven by machine learning, as is increasingly the case when analysts are faced with an ‘ocean of data’ (Tutt, 2016), especially in the context of predicting terrorist acts (Grothoff and Porup, 2016). Here, statistical overfitting of results is an almost inevitable result of the nature of the data used to train the machine learning algorithms in question. Because the actual incidence of terrorism in a population is extremely low, the number of real life examples available to train an algorithm to learn to predict the presence of a terrorist accurately – the ‘ground truths’ – is limited, which embeds a bias in the algorithm’s learning processes, which is then itself almost impossible to detect given the complexity of the calculations involved (Mittelstadt et al., 2016).

The sheer amount of data fed into a big data system is a hallmark of post 9-11 mass surveillance. Indeed, as the New York Times reported (Bergman et al., 2006), in the months following 9/11 NSA computers spat out thousands of tips per month and every one of them turned out to be a false alarm. At the time, FBI agents complained to the NSA that the data was swamping agents, and many saw it as a pointless invasion of citizen privacy. Similarly, Boston’s Joint Terrorism Task Force (JTTF) assessed one of the bombers two years before the Boston bombing and determined that he did not pose a threat. This was one of about a thousand assessments that the Boston JTTF had conducted that year, nearly double the assessments of the previous two years (Schwartz, 2015). Across the US, by 2013 the Justice Department had trained nearly 300,000 law enforcement officers in how to file ‘suspicious-activity reports’. In 2010, a central database held about 3000 such reports and by 2012 it had grown to almost 28,000 (Schwartz, 2015).

Similarly, a report leaked by Snowden revealed a British intelligence agency warning it could ‘currently collect (whether itself or through partners…) significantly more than it is able to exploit fully’, creating a ‘real risk of “intelligence failure” i.e from the Service being unable to access potentially life-saving intelligence from data that it has already collected’ (Gallagher, 2016). An NSA analyst writing in papers released by Snowden encapsulated the problem as follows: ‘We in the agency are at risk of a similar, collective paralysis in the face of a dizzying array of choices every single day’, the analyst wrote in 2011.
‘Analysis paralysis’ isn’t only a cute rhyme. It’s the term for what happens when you spend so much time analyzing a situation that you ultimately stymie any outcome…. It’s what happens in SIGINT [signals intelligence] when we have access to endless possibilities, but we struggle to prioritize, narrow, and exploit the best ones. (Maas, 2015)
The sheer size of the data encountered in post 9-11 mass surveillance induces a related problem in the issue of algorithm design. As the examples above suggest, the amount of data encountered is beyond human analytic capacities, meaning algorithmic interrogation drives the data analytics which underpin the utility of mass surveillance in the post 9-11 world. In practice, this means information collected by mass surveillance is manipulated into knowledge – a map of the world as the system sees it – by algorithms. But like all technology, algorithms are not neutral, reflecting the concerns and values of their designers (see Bijker et al., 1987; Jasanoff, 2004b). As Mittelstadt et al. (2016: 7) note, this may simply mean a decision about which outcomes are the most efficient, or preferable. In the case of mass surveillance, such a decision means pre-empting which profile or pattern is the preferred ‘needle’, thereby limiting the range of possible outcomes by design. Moreover, the problem of bias is arguably intensified in the context of machine learning (Tutt, 2016), as biases may become harder to detect by human users given the complexity of the system. This means that decisions as to what a needle is and so whether it exists or not are not neutral. They are embedded in the algorithms which perform data analytics, meaning they are not necessarily visible or even – in the case of machine learning – necessarily comprehensible by human agents.

What is hay?

Mass surveillance in the post 9-11 era aims to collect as much data – or ‘hay’ – as possible. However, some ‘hay’ is simply uncollectable by the system. For example, the NSA collects phone records, flight records and social media postings, but it does not – as far as we know – collect the secret dreams of young men. It can only collect such dreams in the traces such young men leave behind, and even then only certain traces – those that leave a legible mark.⁷ For example, it cannot collect decisions that are not communicated electronically, or connect dots which are not drawn. It cannot surveil that which it does not see.⁸

This is especially problematic where actors of interest communicate in ways that cannot be ‘seen’ by mass surveillance. In Yemen, for example, a country of great interest to the NSA, only approximately 40% of the country is on the cell phone network (WB, 2015). This means those on the network are easier to ‘see’ but that only makes more stark the absence of the 60% who are not. Sometimes this darkness is a deliberate choice by actors of interest to the NSA. Geotargeting programmes using cell phone towers – the GILGAMESH programme revealed by Snowden – will not work if the target turns off his phone or exchanges SMS cards often with his colleagues, a routine practice amongst terrorist cells (Scahill and Greenwald, 2014).⁹ Indeed, the growing use of personal encryption programmes is proving a barrier to surveillance by US law enforcement agencies. A recently leaked document on the problems of terrorist targets intentionally ‘going dark’ shows the FBI urging its officers to focus on paid informants as well as relying on friends and family of potential terrorists to spot signs of radicalisation (Winter and Weinberger, 2015).

Targets may also exploit the problems of false positives outlined above, and the cost of such false positives to authorities. As a former data scientist at the NSA notes: ‘The most dangerous adversaries will be the ones who most successfully disguise their individual transactions to appear normal, reasonable, and legitimate’. Al-Qaeda (AQ) propagandist Anwar al-Awlaki, for example, called explicitly for ‘cleanskins’ to carry out AQ attacks as they would avoid either surveillance or, if not surveillance, they would avoid suspicion (Hoffmann, 2010). Umar Farouk Abdulmutallab, the failed Detroit bomber of 2009 was one of the results of this call, as was Faisal Shazad, the failed Times Square bomber of 2010. Older studies of individuals evading surveillance point to the same problem of those who understand how the system works avoiding it (Ball, 2010). However, the concept that some individuals may not be within the field of surveillance simply because they do not exist within that field is arguably new, and specific to new forms of global surveillance which have a global aim and a global reach – at least in theory (Gilliom, 2001).

The issue of surveillance evasion is linked to a broader problem of the relationship between the act of surveillance and surveillance technologies themselves. A decision about what data is of interest and what is not must be made by the collector of the ‘hay’, which in this case is the surveillance technology in use. But design of the technology or system itself, especially the purposes for which it was designed, will shape the data which is then collected: what is defined as hay (see Bijker et al., 1987; Jasanoff, 2004b). This problem is arguably amplified in situations of mass surveillance with multiple collection mechanisms and systems – the rhizomic mode which Lyon et al. identify as having replaced centralised systems of collection. Atran et al. (2017) point to this problem when they highlight that much of the data on which the NSA relies is collected by commercial entities with their own, profit-driven interests in collecting data. It is the interests, then, of vendors such as Verizon and AT&T, which were revealed in the Snowden papers as complicit in NSA technology, which shape and drive the nature of the data they collect, not necessarily the NSA itself. Analytics drawing on this data are drawing on a set of information about the world which is shaped by interests not necessarily aligned with the NSA’s.

Ultimately, the data which is collected by post 9-11 mass surveillance is co-created by the technology and those who design it as well as the targets of such surveillance (Jasanoff, 2004a). This means that metaphors which focus only on one aspect of the creation of such data – the incredible technological capacities of the NSA – miss an important aspect of these capacities, namely that they depend on the traces left by targets themselves and the way or indeed whether those traces are harvested.

What is a haystack?

The omniscience of NSA surveillance is also challenged by the problem of defining the limits of such surveillance, especially when the failure to surveil accurately means the state risks an existential threat. The goal of post 9-11 mass surveillance is to take a snapshot of reality in order to predict the future and avoid as yet unknown and unknowable catastrophic events.¹⁰ We know the broad class of events the system wishes to prevent – unexpected catastrophic terrorist events – but by definition such events are unknowable in advance. Aradau and Van Munster (2007: 101) label the risk of such events as ‘precautionary risk¹¹ – a type of risk which demands that we act under scientific and causal uncertainty’. This maps onto Kerr and Earle’s (2013) identification of pre-emptive surveillance as an emerging trend in big data-driven surveillance. Drawing on Ewald’s (2002: 283–284) argument that
pre-emptive risk is marked by ‘scientific uncertainty on the one hand and serious and irreversible damage on the other’, they show that post 9-11 terrorism is a ‘risk beyond risk’ of which we do not have nor cannot have the knowledge or the measure but which must be avoided at all costs. (Aradau and Van Munster, 2007: 100)
The ensuing representation is of terrorism as ‘doubly infinite in its catastrophic effects and the uncertainty of occurrence’ (Aradau and Van Munster, 2007: 105).¹² Indeed, terrorism strikes at the ontological security of the modern nation state.¹³

Intelligence failures prior to 9/11 are key here. NSA mass surveillance is undoubtedly a creature of a post 9-11 world, a world scarred by a devastating terrorist attack which took place in the heart of the most powerful nation on earth and went undetected until its execution. Unsurprisingly, General Keith Alexander named the intelligence failures leading up to 9/11 as one of the animating forces behind his ‘collect it all’ strategy, stating before a congressional hearing: ‘Let me start by saying that I would much rather be here today debating this point than trying to explain how we failed to prevent another 9/11’(Gellman, 2013).

Predicting ‘doubly infinite events’ via mass surveillance is an inherently anxiety-inducing exercise. It means searching for such a needle in an ever-expanding, infinite haystack. It means that NSA surveillance is ‘building the haystack’, rather than simply surveilling it. In doing so, surveillance technologies (or those who design them) necessarily make decisions about where the haystack begins and ends. The unpredictable nature of terrorist events and the catastrophic uncertainty engendered by 9-11 mean that post 9-11 mass surveillance must engage with fundamental uncertainty about the limits of knowledge. It is impossible for mass surveillance to take in all data points which exist at all points in time: to take in any data the system must make epistemological decisions about the type and limit of information it takes in. This means that even as the NSA’s data capacity expands,¹⁴ the problem of deciding where the haystack begins and ends – not just its size – is an anxiety-inducing exercise, and one that can never end.

Aradau and Van Munster (2007) describe terrorism as a situation where the scientific technologies for ‘representing’ the world find themselves surpassed by reality itself. They note that ‘as the underestimation of intelligence and knowledge is considered irresponsible from the viewpoint of precautionary risk, the scope and field of intelligence needs to be enlarged accordingly. Yet, at the limit of knowledge, intelligence becomes itself insufficient’ (p. 101). The existential threat posed by catastrophic terrorist attacks means that the power on which both the panopticon and haystack metaphors rest is fundamentally anxious. Here, although the NSA theoretically has the power to surveil almost everything, this capacity itself arguably induces an anxiety that undermines this power.

The anxious information state

Post 9-11 mass surveillance is an unequivocal demonstration of state power, but this article has problematised the implication in the panopticon and haystack metaphors that mass surveillance is all powerful, rather than simply a demonstration of massive but still fallible state power. It has argued that the system is compromised by its very nature – by the fact that it is a product of the complex social world in which it exists. The knowledge that the system of mass surveillance seeks to produce is co-produced with the world it inhabits, meaning the biases inherent in the system and the historical and social context in which it operates determine its outputs. As a result, the shared entailment of the two metaphors – that ‘absolute knowledge is absolute power’ – is flawed and neither helps us understand the implication of mass surveillance as big data. In this case, using Ezrahi’s terminology, the system produces information, meaning a flattened, limited form of knowing rather than the omniscient form the common entailment suggests. Information alone provides only a limited understanding in comparison to knowledge, tending to ‘… conceal the interpretive layers and normative commitments underlying its structures and uses’ (Ezrahi, 2004: 305): it maps well onto the reification of (big) datafication identified by Boyd and Crawford (2012). Mass surveillance must struggle with the difference between ‘knowledge’ and ‘information’ to use Ezrahi’s terminology: the system collects information, but prediction demands knowledge.

The surveillance state is anxious, then, because it cannot protect its citizens via acquisition of information alone. This violates the fundamental contract of the surveillance state with its citizens – that the state may surveil as long as it uses it in the identifiable service of its citizens (see Higgs, 2004). If a state collects more than ever before, in secret, but still cannot protect its citizens, then the contract is weakened, adding to the ontological instability underlying Western states in a post 9-11 era and to the anxiety this instability generates. Mass surveillance states, then, are states that have a particular relationship with information using Ezrahi’s terminology.¹⁵ Surveillance produces information in the service of the state and it is this product of surveillance – information in the form of big data – which, this article has argued, underlies the anxiety at the heart of the modern information state.

As above, the most persistent analytic framework for the information state is James Scott’s (1998, 2009) work on legibility. Much like mass surveillance states, for the states which form the basis of Scott’s analysis the problem is that which they cannot make legible: the ‘hay’ they cannot see.¹⁶ But Scott’s analysis does not quite capture the anxiety outlined in the previous sections of this paper. This anxiety, it is argued, is driven by the volume of information mass surveillance collects. For Scott, information in itself buttresses the state and the state has the capacity to ‘see’ all that it makes legible: as discussed above, mass surveillance states cannot ‘see’ all which they make legible. Key to this legibility for Scott is the state’s centralised, omniscient gaze. His work cannot easily account for the anxiety the post 9-11 surveillance state encounters at the fact of bulk collection and cannot easily account for the way such a state can miss the information it possesses of threats to its very existence.

Hull’s (2012) work on the information and the bureaucracy of paper in Islamabad’s land title office makes this point. His work challenges Scott’s account by presenting a case – the history of Islamabad land expropriation – in which illegibility and opacity have been produced by the very instruments of legibility designed and implemented by a state (Hull, 2012: 166). This illegibility is engendered by the masses of information and the particular form it takes – in this case reams of paper – which the state requires citizens to produce and legitimise in order to claim registration of land. The mountains of paper required overwhelm the state which designs and implements the system in the sense that the system is ineffectual at best and corruptible at worst, undermining its contribution to state stability and prosperity. Mathur’s (2015) work on the implications of anticorruption regimes for the implementation of a groundbreaking welfare scheme in India points to a similar problem of systematic information overwhelm: in an attempt to make the state legible, this scheme requires bureaucrats to produce information which makes clear to citizens the activities of their government – to make the state legible to its citizens. However, in doing so, Mathur finds that the onus of information capture and management overwhelms the system to the extent that said funds are not implemented effectively.

The examples outlined above seem a far cry from the technologically sophisticated, multilayered forms of mass surveillance as big data. But they speak to the problem of information production and management as a state concern and, particularly in Hull’s work, to information as weakening rather than strengthening the state. Neither, however, speaks directly to the underlying anxiety marking the mass surveillance state: an anxiety which drives information overload in the quest to secure the state from those who wish to commit violence against it. The information overload in this sense is driven not only by the state’s quest to secure itself, but by the nature of the data it seeks to collect and the problem it seeks to address. This anxiety drives mass surveillance and is consumed by it: it must roll on endlessly.

Conclusion

This article has interrogated two metaphors widely used to discuss and explain big data in the post Snowden era. Neither the panopticon nor the haystack adequately describes the phenomenon of big data in a post 9-11 mass surveillance state. Indeed, they share a faulty common entailment which presumes both the capacity of absolute power by a state and the link between absolute state power and absolute knowledge. In doing so, they share an image of the post 9-11 information state which fails to capture the inherent anxiety of mass surveillance in the current era. This article has argued that the underlying driver for post 9-11 mass surveillance is ontological insecurity in the face of terrorist attacks. This insecurity begets an anxiety which can never be fully quelled because of the nature of the data involved. Drawing on Ezrahi’s distinction between knowledge and information, the article shows that the information state in this context is driven by the quest for information but equally damaged by the very nature of that information. Ultimately, metaphors as conceptual frameworks provide a way for new and complex concepts to be understood and communicated. The panopticon and the haystack are diminished in their analytic power because their vision of the information state is inadequate in an era of mass surveillance and big data. However, their inadequacy is in itself a conceptual framework for understanding the relationship between big data, mass surveillance and the information state in a post 9-11, post Snowden world.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Alexander K (2013a) Evidence before the United States Senate Select Committee on Intelligence. In: Hearing on Foreign Intelligence Surveillance Act (FISA) Legislation, Washington DC, USA, 26 September.

Alexander K (2013b) Keynote presentation. In: Black Hat USA 2013, Las Vegas, USA, 27 July–1 August. Available at: https://www.nsa.gov/news-features/speeches-testimonies/speeches/transcript-of-gen-alexander-black-hat-speech-31july2013.shtml (accessed 1 October 2015).

Alexander K (2013c) Keynote presentation slides. In: Black Hat USA 2013, Las Vegas, USA, 27 July–1 August. Available at: https://assets.documentcloud.org/documents/802262/us-13-alexander-keynote.pdf (accessed 1 October 2015).

Alexander K (2013d) Evidence before the United States Senate Committee on the Judiciary. In: Continued Oversight of the Foreign Intelligence Surveillance Act, Washington DC, USA, 2 October.

Amoore

De Goede

(2005) Governance, risk and dataveillance in the war on terror. Crime, Law and Social Change Vol 43(2–3): 149–173.

Aradau

Blanke

(2015) The (big) data-security assemblage: Knowledge and critique. Big Data & Society. 1–12.

Aradau

Van Munster

(2007) Governing terrorism through risk: Taking precautions, (un)knowing the Future. European Journal of International Relations 13(1): 89–115.

Atran

Axelrod

Davis

et al. (2017) Challenges in researching terrorism from the field. Science 355(6323): 352–354.

Ball

(2010) Workplace surveillance: An overview. Labor History 51(1): 87–106.

10.

Bergman L, Lightblau E, Shane S, et al. (2006) Spy agency after September 11 led FBI to dead ends. The New York Times, 17 January. Available at: http://www.nytimes.com/2006/01/17/us/front%20page/domestic-surveillance-the-program-spy-agency-data-after-sept.html (accessed 2 February 2017).

11.

Betz

Stevens

(2013) Analogical reasoning and cyber security. Security Dialogue 44(2): 147–164.

12.

Bijker

Hughes

Pinch

(1987) The Social Construction of Technological Systems: New Directions in the Sociology and History of Technology, 2nd ed. Cambridge: MIT Press.

13.

Bobrow

(1996) Complex insecurity: Implications of a sobering metaphor: 1996 presidential address. International Studies Quarterly 40(4): 435–450.

14.

Boyd

Crawford

(2012) Critical questions for big data: Provocations for a cultural, technological, and scholarly phenomenon. Information, Communication and Society 15(5): 662–679.

15.

Cahall B, Schneider E, and Bergen P (2014) Do NSA’s Bulk Surveillance Programs Stop Terrorists? Report, The New America Foundation, Washington, DC, January.

16.

Calmes J (2013) Obama says surveillance helped case in Germany. The New York Times, 19 June. Available at: http://www.nytimes.com/2013/06/20/world/europe/obama-in-germany.html?mcubz=1 (accessed 1 January 2016).

17.

Caluya

(2010) The post-panoptic society? Reassessing Foucault in surveillance studies. Social Identities 16(5): 621–633.

18.

Cisco (2015) The Zettabyte Era: Trends and Analysis. Report, Cisco Systems. Report for Cisco Systems, May. Available at: http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/VNI_Hyperconnectivity_WP.pdf (accessed 20 January 2016).

19.

Cole J (2013) Evidence before the United States House of Representatives Committee on the Judiciary. In: Hearing on the Administration's Use of FISA Authorities, Washington DC, USA, 17 July.

20.

Davies

(2014) Why society is not made of documents. In: Bacchini

Cauto

Dell’Utri

(eds) Metaphysics and Ontology Without Myths, Newcastle: Cambridge Scholars Press, pp. 125–150.

21.

Edelman M (1964) The Symbolic Uses of Politics. Urbana: University of Illinois Press.

22.

Epstein

(2016) Surveillance, privacy and the making of the modern subject: Habeus, what kind of corpus? Body and Society 22(2): 28–57.

23.

Ewald

(2002) The return of Descartes’ malicious demon: An outline of a philosophy of precaution. In: Baker

Simon

(eds) Embracing Risk – The Changing Culture of Insurance and Responsibility, Chicago, IL: University of Chicago Press, pp. 273–301.

24.

Ezrahi

(2004) Science and the political imagination in contemporary democracies. In: Jasanoff

(ed.) States of Knowledge: The Co-production of Science and the Social Order, New York: Routledge, pp. 254–274.

25.

Ferrari

(2007) Metaphor at work in the analysis of political discourse: Investigating a ‘preventive war’ persuasion strategy. Discourse and Society 18(5): 603–625.

26.

Ferraris

(2012) Perspectives of documentality. Phenomenology and Mind 2: 34–40.

27.

Foucault M (1979) Discipline and Punish: the Birth of the Prison. New York: Vintage Books.

28.

Furnas A (2012) Homeland Security’s pre-crime screening will never work. The Atlantic, 17 April. Available at: https://www.theatlantic.com/technology/archive/2012/04/homeland-securitys-pre-crime-screening-will-never-work/255971/ (accessed 1 January 2016).

29.

Gallagher R (2016) Facing data deluge, secret UK spying report warned of Intelligence failure. In: The Intercept. Available at: https://theintercept.com/2016/06/07/mi5-gchq-digint-surveillance-data-deluge/ (accessed 1 October 2016).

30.

Gellman B (2013) Code name ‘Verax’: Snowden, in exchange with post reporter, made it clear he knew risks. The Washington Post, 9 June. Available at: https://www.washingtonpost.com/world/national-security/code-name-verax-snowden-in-exchanges-with-post-reporter-made-clear-he-knew-risks/2013/06/09/c9a25b54-d14c-11e2-9f1a-1a7cdee20287_story.html?utm_term=.cbfd207b9417 (accessed 1 October 2016).

31.

Gellman B and Soltani A (2013) NSA collects millions of email address books globally. The Washington Post, 14 October. Available at: https://www.washingtonpost.com/world/national-security/nsa-collects-millions-of-e-mail-address-books-globally/2013/10/14/8e58b5be-34f9-11e3-80c6-7e6dd8d22d8f_story.html?utm_term=.5c8e37bfb045 (accessed 1 October 2016).

32.

Gilliom

(2001) Overseers of the Poor: Surveillance, Resistance, and the Limits of Privacy, Chicago, IL: University of Chicago Press.

33.

Greenwald

(2014) No Place to Hide: Edward Snowden, the NSA and the US Surveillance State, New York: Metropolitan.

34.

Grothoff C and Porup J (2016) The NSA’s Skynet program may be killing thousands of innocent people. In: Arstechnica. 16 February. Available at: https://arstechnica.co.uk/security/2016/02/the-nsas-skynet-program-may-be-killing-thousands-of-innocent-people/ (accessed 1 January 2017).

35.

Haggerty

(2006) Tear down the walls: On demolishing the panopticon. In: Lyon

(ed.) Theorizing Surveillance: The Panopticon and Beyond, New York: Routledge, pp. 23–45.

36.

Haggerty

Ericson

(2000) The surveillant assemblage. The British Journal of Sociology 51(4): 605–622.

37.

Higgs

(2004) The Information State in England: The Central Collection of Information on Citizens Since 1500, London: Palgrave Macmillan.

38.

Hoffmann B (2010) Evidence before the US House of Representatives Committee on Homeland Security. In: Internet terror and recruitment and tradecraft: How can we address an evolving tool while protecting free speech? Washington DC, USA, 20 May.

39.

Hook

(1984) The nuclearization of language: Nuclear allergy as political metaphor. Journal of Peace Research 21(3): 259–275.

40.

Hull

(2012) Government of Paper: The Materiality of Bureaucracy in Urban Pakistan, Los Angeles: University of California Press.

41.

Jasanoff

(2004a) The idiom of co-production. In: Jasanoff

(ed.) States of Knowledge: The Co-production of Science and the Social Order, New York: Routledge, pp. 14–27.

42.

Jasanoff

(2004b) Ordering knowledge, ordering society. In: Jasanoff

(ed.) States of Knowledge: The Co-production of Science and the Social Order, New York: Routledge, pp. 27–53.

43.

Kerr

Earle

(2013) Prediction, pre-emption, and presumption: How big data threatens big picture privacy. Stanford Law Review 66: 65–78.

44.

Kessler

(2010) Risk. In: Burgess

(ed.) Handbook of New Security Studies, London: Routledge, pp. 17–27.

45.

Lakoff

Johnson

(1980) Metaphors We Live By, Chicago, IL: University of Chicago Press.

46.

Levi

Wall

(2004) Technologies, security, and privacy in the post-9/11 European information society. Journal of Law and Society 31(2): 194–220.

47.

Lyon

(2006) Theorizing Surveillance, London: Routledge.

48.

Lyon

(2014) Surveillance, Snowden and big data: Capacities, consequences, critique. Big Data and Society 1(2): 1–13.

49.

Maas P (2015) Inside NSA, officials privately criticize ‘collect it all’ surveillance. In: The Intercept. Available at: https://theintercept.com/2015/05/28/nsa-officials-privately-criticize-collect-it-all-surveillance/ (accessed 1 October 2015).

50.

Mathur

(2015) Paper Tiger: Law, Bureaucracy, and the Developmental State in Himalayan India, New Delhi: Cambridge University Press.

51.

McMullan T (2015) What does the panopticon mean in the age of digital surveillance? The Guardian, 23 July. Available at: https://www.theguardian.com/technology/2015/jul/23/panopticon-digital-surveillance-jeremy-bentham (accessed 1 January 2016).

52.

Mittelstadt

Allo

Taddeo

et al. (2016) The ethics of algorithms: Mapping the debate. Big Data and Society 3(2): 1–21.

53.

Nakashima E and Warrick J (2013) For NSA chief, terrorist threat drives passion to collect it all, observers say. The Washington Post, 14 July. Available at: https://www.washingtonpost.com/world/national-security/for-nsa-chief-terrorist-threat-drives-passion-to-collect-it-all/2013/07/14/3d26ef80-ea49-11e2-a301-ea5a8116d211_story.html?utm_term=.c32e25ad4dc6 (accessed 1 January 2016).

54.

Ruppert

(2012) The governmental topologies of database devices. Theory, Culture and Society 29(4–5): 116–136.

55.

Sanchez J (2014) Snowden showed us just how big the panopticon really was. Now it’s up to us. The Guardian, 6 June. Available at: https://www.theguardian.com/commentisfree/2014/jun/05/edward-snowden-one-year-surveillance-debate-begins-future-privacy (accessed 1 October 2015).

56.

Scahill J and Greenwald G (2014) The NSA’s secret role in the US assassination program. In: The Intercept. Available at: https://theintercept.com/2014/02/10/the-nsas-secret-role (accessed 1 October 2015).

57.

Schmitt E, Sanger D and Savage C (2013) Administration says mining of data is crucial to fighting terror. The New York Times, 7 June. Available at: http://www.nytimes.com/2013/06/08/us/mining-of-data-is-called-crucial-to-fight-terror.html (accessed 1 October 2015).

58.

Schwartz M (2015) The whole haystack. The New Yorker, 26 January. Available at: http://www.newyorker.com/magazine/2015/01/26/whole-haystack (access 1 October 2015).

59.

Scott

(1985) Weapons of the Weak: Everyday Forms of Peasant Resistance, New Haven, CT: Yale University Press.

60.

Scott

(1998) Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed, New Haven, CT: Yale University Press.

61.

Scott

(2009) The Art of Not Being Governed: An Anarchist History of Upland Asia, New Haven, CT: Yale University Press.

62.

Simpson E (2013) The panopticon paradox. Foreign Policy, 24 December. Available at: http://foreignpolicy.com/2013/12/24/the-panopticon-paradox/ (accessed 1 October 2915).

63.

Tutt A (2016) An FDA for algorithms social science research network. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2747994 (accessed 1 January 2017).

64.

Wai S, Meyer T and Elliott J (2013) How the NSA’s claim on thwarted terrorist plots has spread. In: ProPublica. Available at: http://projects.propublica.org/graphics/nsa-54-cases (accessed 1 March 2017).

65.

Weinberger S (2011) Terrorist ‘pre-crime’ detector field tested in US. Nature, 27 May. Available at: http://www.nature.com/news/2011/110527/full/news.2011.323.html (accessed 1 October 2015).

66.

Wight

(2015) Rethinking Terrorism: Terrorism, Violence and the State, London: Palgrave MacMillan.

67.

Winter J and Weinberger S (2015) Facing growing encryption, law enforcement recommends more informants. In: The Intercept. Available at: https://theintercept.com/2015/11/03/facing-growing-encryption-law-enforcement-recommends-more-informants/ (accessed 5 November 2015).

68.

World Bank (WB) (2015) World Bank statistical database. Available at: http://data.worldbank.org/ (accessed 1 October 2015).

The needle and the damage done: Of haystacks and anxious panopticons

Abstract

Keywords

The Snowden revelations: Mass surveillance on an unprecedented scale

Panopticons, haystacks and the information state

The damage done

What is a needle?

What is hay?

What is a haystack?

The anxious information state

Conclusion

Footnotes

Declaration of conflicting interests

Funding

Notes

References