Keywords

1 Intelligence Services as Research Object

At the end of 1998, I commenced my doctoral thesis on the Spanish intelligence services at the University of Barcelona, Spain, although all the fieldwork took place in Madrid at a recently established hybrid university institute somewhere between the National University of Distance Education and the Ministry of Defence: the Instituto Universitario ‘General Gutiérrez Mellado’. The aim of my doctoral thesis was to describe the model of the intelligence services that Spain had adopted after its transition to democracy in 1978, the political and bureaucratic dynamics underlying that model, and the similarities of the Spanish model with respect to other Western ones. Since that initial study, most of my academic career has been centred on intelligence services, branching out to particular research areas on especially sensitive topics.

Oral sources of information have always played a major role in my research, due to the fact that other sources that are extensively used in other research fields, such as governmental reports, statistics, and archives, are inaccessible to the general public and hardly ever disclosed in Spain in view of the legal restrictions of the Law on Official Secrets of 1968, and its development in Law 48/1978, of 7 October. Over the years, I have conversed with former members of the intelligence services from the 1970s and 1980s who claim to have held onto internal documents, either as a protective measure or as part of a somewhat relaxed security culture that prevailed after the democratic transition. In any case, aware of its fragmented and individualistic features, I have never asked for access to that sort of material. However, I have been involved in some attempts to declassify blocks of confidential material, only released after protracted and wearisome administrative procedures.

Although members of the intelligence services might in the first instance appear to be the most obvious research participants, this domain can only be understood with the inputs that other actors can offer, such as prime ministers and parliamentarians, members of law enforcement and military information services, journalists, and diplomats. All these actors represent the multiple mirror-like facets of a polyhedric cube and should never be dismissed. Nonetheless, the overall relevance of oral sources to intelligence studies should be placed in perspective. According to a systematic review of all research articles published between 1986 and 2016 in Intelligence and National Security – the flagship intelligence studies journal – researchers specialized in intelligence and national security studies conducted and referred to interviews in a mere 15% of all publications (Van Puyvelde, 2018).

Interviews aside, it is quite unimaginable for researchers to be given access to the premises of the intelligence services for observation, to conduct surveys there, or merely to gather information on working life (Hammond, 2015). There are nonetheless some very rare examples (Johnston, 2005; Cremades, 2021) where such arrangements have worked: intelligence services, for example, have commissioned research and invited teachers or guest researchers to access the service indirectly via intelligence institutes. In my case, after years of trustful cooperative relations with the Spanish intelligence services, I am perfectly unaware of the location of the Centro Nacional de Inteligencia (CNI) [National Intelligence Centre] canteen, if indeed they even have one, because of the physical restrictions on any movements when visiting its headquarters.

The focus of this chapter, the study of the intelligence services, quite rightly reflects the contributions of intelligence officers and associated agencies. I will not refer to my own experiences with political elites and other such sources, as that topic is better covered in other branches of the scientific literature. Most of my interviewees have been male, military personnel, retired from active service in the intelligence agencies, although the thoughts of both male and female civilians have been reflected in my studies, since completing my Ph.D. dissertation in 2003, as a consequence of the evolving human resources profile of the Spanish intelligence services. These individuals fall into the category of hard-to-reach populations. Researching people whose identities are often unknown and, even when known, are forbidden to discuss their work with outsiders, presents many difficulties. The scientific research methods within this field, therefore, require some adaptations that are detailed in the following section.

2 Phases of Fieldwork

2.1 Accessing the Fieldwork

Intelligence agencies are uncharted territory, whose staff have blurred faces and whose facilities have doors with neither doorbells, knockers, nor doorknobs. Knocking on that sort of door is extremely challenging. Identifying the beginning of a potentially informative thread is time consuming, and the very few available sources mean that potentially useful gatekeepers are difficult to identify. The limited number of available sources, added to their invisibility, is the main difficulty when documenting and designing a fieldwork strategy to access those threads. In my initial research, I knocked at two doors: one formal and another informal. The formal door consisted in contacting the Spanish intelligence services to share details on my proposal for a doctoral thesis and to invite their collaboration. After some phone calls made by the director of the University Institute, at the end of 1998, I received a green light to send a letter to the director of the Centro Superior de Información de la Defensa (CESID) [Higher Centre for Defence Intelligence] – at that time the name of the Spanish intelligence services – later renamed the CNI in 2002.

Following various negotiations, moderated via a prestigious Army general, an interview at the CESID headquarters in Madrid was arranged where they asked me for a copy of my doctoral project – that I had sensibly slipped into my briefcase – and for over 2 hours an intelligence officer succinctly responded to a long list of questions. I never heard anything back from CESID until years later. Some high-ranking Armed Forces officials had made calls on my behalf to move the process on; however, all their conversations came to nothing. It was crystal clear that I was on their radar and that they held little or no interest in cooperating with my research project. In formal organizations, collaboration can be prohibited between personnel and outsiders (Dingwall & Strong, 1985; Wadds, 2022), and I wanted to avoid being vetoed and gaining a reputation as an unreliable and bothersome researcher. Although fruitless, this first formal approach was necessary to pre-empt any subsequent criticism, i.e. that CESID could justify a decision not to cooperate with my study, due to my having rushed into the field without having consulted them first.

Formal access having failed, I focused on informal access to the world of spying. I also discovered that some well-known social research personae, such as the informant and the gatekeeper, also existed in the intelligence world. As in other domains, despite the sound design of an investigation, successful access is still a question of being in the right place at the right time, which implies a certain degree of good luck and fortuitousness (Sarsby, 1984). In my personal experience, I had that sort of good fortune with my three main gatekeepers: a former minister of Defence, a Navy officer and former CESID member in the 1980s, and a journalist specializing in security and defence. However, it is to be noted that the first contacts to come forward in fieldwork may often be marginalized or dissatisfied members within an organization and should be rejected; otherwise, the research risks being skewed, and the researcher may even be considered supportive of the most deviant members under scrutiny. Fortunately, this was not my case, but I cannot trivialize the risks of welcoming absolutely anyone with open arms who promises to unlock the gate because of my own personal frustrations of months and months of work to no avail.

Seen with some perspective, my introductory contacts were the hotchpotch of the three characteristics that, in his influential work, Lee (1993) attributed to key informants and gatekeepers. In the first place, they were the bridge that allowed me to overcome the very often invisible ditch that separates the group from the milieu, in my case, a young civilian from the south of Spain at the University of Barcelona conducting research at a Ministry of Defence building in Madrid. Arriving at the Institute, it was a marvellous day whenever a note or an email with a suggested contact lay on my desk: ‘Telephone 91-398952, he is expecting your call’, which allowed me to cross the river towards the other side of ‘spy town’. The informants and gatekeepers also acted as guides, facilitating a certain freedom of movement within the group by indicating to me what to do in some circumstances, what type of questions to ask, and even to avoid embarrassing situations by sketching out the customs, taboos, and internal dynamics of the group. Comments such as ‘have you already spoken with…?’, ‘When you speak with X, remember to ask him about Y, he’ll understand’, or ‘Don’t even dream of going out for lunch at a military club without wearing a jacket, will you?’ were not at all uncommon. Finally, there were the guardians who offered their protection, in my case, knowing that I was under the academic umbrella of a former Defence Ministry, and warning me: ‘You know who X really is now, don’t you? Be careful with him’.

Another way of gaining informal access to the field has been attendance at conferences, seminars, and forums at which the presence of military personnel, diplomats, and security personnel is plentiful. It implies being continually on the lookout for whoever might be or might have been working or in contact with the Spanish intelligence services, handing out visiting cards, and sending out feelers on work underway, which was I suppose tiring and repetitive for some people. Smiling kindly, one of the frequent attendants to those forums asked ‘Hey, son, still selling your thing?’. Slowly, people started to appear who ‘might know someone, who perhaps knew someone, who might have worked for the intelligence service’. Gradually more doors opened and more potential informants appeared. More recently, social media platforms – mainly Twitter and LinkedIn – have opened up new gateways for accessing former members of the intelligence services who had either retired or left the service for the private sector – as also referred by Cremades (2021) in his Ph.D. dissertation on the Colombian intelligence services. Communication channels that I have hardly used and if so, mainly to stay in touch and to keep communication channels open.

Although secret and hidden from the outside, internal intelligence service members maintain close ties of camaraderie between each other. As an example, a leading director of the service during the years of the transition to democracy and whom I was at a loss to find for years reproached me when I identified him at a conference: ‘Good Lord, you’ve already spoken to all my promotion before me; I thought you weren’t interested in me! Come home and we’ll talk, whenever you want’, which confirmed that good – and also bad – reputations can run rife. The response that I received from the other end of the telephone of an ex-senior CESID manager also serves as an example. After having merely told him my name, he said, ‘Hallo, young man, how is your thesis going?’, which I interpreted as an ostentatious way of demonstrating that he knew perfectly well who I was.

So much time preparing ways of accessing these participants might be considered futile and the best way could indeed be to try direct contact and cold calling. However, I must say that most of my ‘failed interviews’, i.e. those in which I could not obtain any relevant information or in which the cooperative attitude of the interviewee was stifled, were precisely those in which my strategy was not sufficiently case-specific or the relationship between the intermediary and the interviewee was not sufficiently deep. In that regard, I learned – a posteriori, from my own mistakes – that access must be carefully planned, key informants identified and used whenever possible, and reputation maintained intact, as invisible ties connected the group to the eyes of outsiders.

2.2 Data-Collection Technique: Interviews

As summarized by Van Puyvelde (2018), the traditional approach in intelligence studies towards interviewing has been, first of all, the positivist one, serving the interviewers to help fill in the information gaps left by publicly available documentary sources. From this perspective, interviewing insiders can enrich and sometimes contrast the government narrative recounted through archival sources and other public documents. The second approach is the constructivist one, where interviewers coproduce data with their interviewees and not merely to glean information from them. Observing my whole research career, my use of oral sources could mainly be explained as a positivistic approach, seeing my interviewees as data providers, in the absence of other sources, such as archives or official documents. Nonetheless, semi-structured interviews leave room for cosmovision to emerge and motivations that are more identifiable with a constructivist approach, as I was indeed trying to give meaning to such questions as the evolution of the Spanish intelligence services following the end of the dictatorship under Francisco Franco (Díaz-Fernández, 2005) and the role of the Spanish intelligence community in the decision-making process (Díaz-Fernández, 2010).

Sampling, recruitment, and interviewing schedules must be clearly aligned with the research question(s) and with the utility we expect from the available types of oral sources. However, beyond any ideal strategy that we might have in mind, we must also consider that availability, convenience, and chance in the field of intelligence can often affect the order in which interviewees will recount their stories (Van Puyvelde, 2018). The main sampling method I used was snowball (non-probability) sampling. It is based on the assumption that a link or a connection exists between the initial sample and others within the population that is under study, thereby building up a series of reference units within a known circle of contacts (Khoury, 2020). Hence, at the end of my interviews, I would usually make the following sorts of requests: ‘Could you put me into contact with the person that you mentioned during the interview?’ or ‘Do you know someone who might be of interest in this study?’. Some interviewees rapidly dismissed some of those requests, but others kept them in mind until, for example, a circle of officers met up for lunch or ran into each other by chance. Patience and optimism are, therefore, important qualities to develop in this field: even though the researcher might sometimes perhaps feel blocked, one never knows when another informant will emerge.

One critique often directed at snowball sampling is linked to the selection of participants and, especially, the connections between them, which can lead to a biased sample and can, therefore, render the research invalid and unreliable. However, although it is a subjective perception, I have never noted any such bias within a specific network of participants. Obviously, this perception may be likened to the tip of an iceberg, at first sight perhaps thought to be floating upon the sea even though 10% is only ever visible. It may perhaps be due to my approach that was focused on the intelligence services as an organization, so I never pursued personal narratives (i.e. interest in joining the service, family life, risk perception…) where I might potentially have faced a wider variety of realities. In 2012, a group of former members of the Spanish Intelligence Services created an association to stay in contact and to participate in some outreach activities: Asociación de Ex-Miembros del Servicio de Inteligencia Español [Association of Ex-Members of the Spanish Intelligence Service] with some of whose members I am still in touch. Its membership of approximately 110 ex-members shows greater interest in sharing some of their experiences and participating in publications, round tables, and TV programmes. It is another niche that can serve as the start of a new thread or to canvass for new informants. Although some of the members with whom I am personally acquainted share a more open view of their work and relationship with society, my perception is that they cover a wide variety in terms of ideology and life experiences.

Sampling is one of the key themes where the size of the group is unknown. A quick look at research published on the topic reveals a wide variety of characteristics and samples sizes, from Riste (1999) on a historical approach to the Norwegian intelligence services to Perry (2021), in his research on the motivations of spies to become involved in espionage, and the 42 participants in Eickelman’s (1988) study, to ten in the study of Ratcliff (2006), and around 400 in the study of Johnson (1996). But if we go beyond the figures, we see that the 42 oral sources of Eickelman in his study of the intelligence service of Oman, before that country gained its independence, supposedly represented 89% of the members of the service: an unquestionable success. Considering the size of the potential universe, the number of interviewees, although reduced, can be considered very relevant. Gill (2000) interviewed 70 elite members of the police information service and O’Brien (2011), in his doctoral thesis on the South African intelligence services, interviewed around 20 officers in charge of national security during the days of apartheid.

The size of the sample is directly linked to the research question. Unlike in the first few years when each interview was as a desert oasis, it was increasingly easy later on to enter into contact with the ‘hidden dimension’ of the State. The decision over when to end fieldwork has, therefore, become more difficult, as new snippets of information and new sources have emerged with greater frequency. Throughout the years, this final stage has always been marked by the well-known ‘saturation point’ (Glaser & Strauss, 1967): when (i) it seems that no new or relevant data are going to appear in relation to that category; (ii) the category is well developed in terms of its properties and dimensions, demonstrating variation; and (iii) the relationship between categories is well established and validated. Therefore, the collection strategy – and here the size and composition of the sample are included – must be clearly aligned with our research question.

Regarding the order in which the interviews were conducted, most of my research has been rooted in interviews with intelligence officers, before moving on towards the spheres of political and managerial influence, in order to check my findings and to test my hypotheses. I must admit that young operative agents have never been the principal target of my research. Firstly because, for obvious reasons, they have a narrower vision of intelligence agency activities and, secondly, their more limited willingness to cooperate with researchers, as I have observed, diminishes with age as the agents become more flexible in terms of the information that they are willing to share with the researcher. Nonetheless, every source has its value, and younger members and those with fewer responsibilities in the service are also interesting sources that can bring new insights to the attention of the researcher for further exploration. The research strategy must, therefore, be based on accepting the different types and the depth of the information that each available source can provide; something that is obvious in the field where the activity is so heavily compartmentalized and where sources cannot provide information they do not hold.

Researchers could employ the triangulation method – i.e. cross-referencing the information obtained via different sources – to interpret the data, due to the scarcity and the fragmentation of the available information (Hammond, 2015), to overcome the barriers of compartmentalization, lack of information, and fragmentation of knowledge. In my personal experience, I have used triangulation on a threefold basis. Firstly, asking the oral sources about the content of some documents, and, secondly, asking the same questions to members of different groups – agents, medium-level managers, and executives – although assuming that they hold a different vision of the topic according to the place they occupy in the organization. I by chance experienced a very particular, rare, and unexpected situation of ‘triangulation’ some two decades ago. An operative agent vividly recounted a bizarre operation to me that had taken place in Morocco where he had been working alongside a second Spanish spy. More recently, chasing up the comments of a former spy at a conference, I was curious and questioned him in private: ‘I think I know who your peer was during that operation: he was X, wasn’t he?’ The individual admitted having accompanied the second spy during that mission, and no other single detail came out of his mouth. His polite, yet extraordinarily succinct explanation taught me two lessons: (i) that operation was exactly how it was described two decades ago by the first agent and (ii) the different characters of spies and the degree of reserve that they can hold towards past operations.

2.3 Gaining and Keeping my Credentials

My key informants and my own smooth walk into the fieldwork – perhaps only possible within the framework provided by a four-year Ph.D. grant – were sufficient to gain some credentials. The latter is critical, as access is achieved not so much because of what is explained, but by how the researcher behaves in front of the gatekeepers and key informants (Wax, 1985). Since first starting my fieldwork, I have strictly adhered to some basic rules, thoughtfully established in the first months of the research, in order to gain and, of course, to keep the necessary credentials. Firstly, research objectives – as Peritore (1990) advised – must always be presented with the greatest possible clarity and transparency knowing, moreover, how misleading the intelligence world can be and, more specifically, considering the scandals of political espionage in which CESID was embroiled in the year I began my doctoral project (1998). Secondly, I emphasized my academic affiliation and the seriousness of my work: I was preparing a Ph.D. dissertation, so I could never be confused with a snooping investigative journalist. I always carried my visiting cards bearing the logotype of the institution with which I was affiliated at the time, to emphasize the scientific character of my work (Williams et al., 1992). And thirdly, I decided to never ask about topics that lay outside the object of the investigation, even though the interviewees may have raised them. As Feldman (1991, p. 12) also underlined, I too made it clear to my interviewees that ‘there were things, places and people that I did not want to see’.

I was following these rules so strictly that it was sometimes shocking for some interviewees. This was, among others, the case of the director of the Organización Contrasubversiva Nacional de Franco [National Countersubversive Organization of Franco] (1968–72). Reaching out to him took almost 2 years and four intermediaries before I was guaranteed a meeting. It may be noted that he was convicted for participating in the attempted coup d’état of 1981 and was released from prison in 1986. After some smooth interviews with him, in reaction to my never having asked him about his prominent involvement – as this was not part of my dissertation – he interjected: ‘Shall I tell you about the coup?’, to which I evidently responded in the affirmative as an expression of empathy and rapport.

Part of this reluctance to break these rules was based on my permanent suspicion that some interviewees were sending probes to check my reliability and to find out whether I was really interested in the Spanish intelligence model or whether that was an excuse for another type of work. In that sense, I always doubted whether the initial questions of some of the interviewees were simply asked out of curiosity or whether they were testing my trustfulness. Those questions were – and still are – somewhat scary as they are usually very short, straightforward, and punctilious with little room for escape clauses: ‘Who else have you talked to?’ or ‘What have the others told you?’ They sometimes even attempted to glean my opinions with questions like: ‘You’ve been investigating the matter for some time, what are your conclusions?’, all of which even before the proper beginning of the interview. It was always clear that revealing some of this information implied breaking the commitment to both the anonymity and the confidentiality that I had only just a few minutes ago guaranteed and, potentially, closing the door on any new interviews that they might otherwise have suggested or even ending the flow of information from an interviewee.

The truth is that a researcher has no wish to disappoint an interviewee. Therefore, among the few ways of skirting around, any inquisitive questions from an interviewee are responses that somehow exaggerate the underlying hypothesis of the matter at hand, which have been of great utility in past interviews. It forces the interviewee to rise to the challenge and sympathetically ‘correct’ my distorted view of the matter, to put me out of my ignorance, in such a way that the interviewee is suddenly in charge of the situation. Nevertheless, these situations should hardly be ignored, as such topics, areas of interest, and casual hints that usually crop up – and not in the most politically correct manner at times – can help to contextualize the content, to open up new lines of work, and to construct alternative hypotheses (Warren et al., 2003). Over the years, personal and academic maturity has helped me to answer and to block these questions in a more direct and straightforward way, sometimes quite deftly, using humour.

2.4 Logistics

Where to meet up with the interviewees has never been an inconsequent topic for obvious reasons of privacy and security. In my case, I always offer the interviewees the choice of venue, which on the immense majority of occasions they choose; in any case, it is important to have an alternative in mind, so as not to appear disorganized or amateurish, in case the interviewee is unable to set the place. In my experience, over half of them prefer to meet up at a café or restaurant, and a quarter in their office. I cannot establish a precise rule on the choice of venue as cafés are chosen both by retired and active members or those who continue working for the government or the private sector. On the other hand, some of the people I have interviewed several times have on various occasions chosen different venues for the meetings. The ideal venue for an interview is, in short, wherever the interviewees feel at ease and relaxed. In the case of the intelligence officers, I draw what I call ‘Three golden rules for not bothering a spy’ as I sense that they feel more comfortable controlling the interview setting. These are (i) always let them place themselves in the furthest corner and face the door; (ii) never arrive early for the meeting, and let them be the ones waiting for me so that they could scrutinize the meeting place, and (iii) make sure not to keep any piece of paper on which they may have written or drawn during the interview.

In some cases – exclusively for high-ranking services personnel, prominent politicians, and business leaders – I was able to identify a similar pattern. On most occasions, the interview was interrupted by an assistant who entered the room recalling a recent interview or event. Over time, I interpreted those events as a pre-established escape door in case the interviewee felt uncomfortable during the interview. I am glad to say that all the interviews have always continued and even beyond the pre-established length. My perception has invariably been that the interest of these elite groups always tended to run deep, out of a desire to better understand topics of which they had direct experience, and they were pleased to express their interest in an interview with an outsider whose cross-cutting view of the topic might also give them further insight.

Most of my interviews were semi-structured; however, over the years, I have also had the opportunity to share meals, beers, trips… where very relevant information is shared for understanding the workings of the Spanish intelligence community in a less structured manner. In view of the smooth strategic approach detailed above, I am fully informed well in advance on the background of the person to be interviewed or I have, at least, the basic information such as period, area, and organization for which the interviewee worked. As greater familiarity was gained with the topic, holding unstructured interviews – although stressful – was not such a huge challenge as it was in my first years and can yield good results. The problem arises on those rare occasions where all I have is a name, making it difficult to conduct even a semblance of an interview. Many years ago, I was not even aware of who the interviewee would be. Two army colonels insisted on having lunch with a former colleague and I accompanied them in a car following a rough plan – before Google Maps was available to the general public – to a social club on the outskirts of Madrid where we had lunch with a person who, after various hours of conversation, confessed to me that he had been number two in Franco’s secret service. I had references to him – to a certain ‘Captain X’ – in some documents and oral testimonies but imagined him to be more of a phantom than a real person until I had not only met him but also had lunch with him.

Although the telephone interview – or more recently Video Teleconferencing – increases access to potential interviewees, in my experience, neither are communications channels that my interviewees have ever shown much interest in using at all for anything other than arranging a meeting. It is linked to a very persistent doubt that my colleagues have raised with me in fieldwork – both in the past and nowadays: the possibility of recording the interviews. There are a few researchers such as Cremades (2021) who have recorded some of the interviews. However, in my case, I have never asked interviewees for permission to record the conversations assuming that sensitive information could potentially be provided and, whether or not sensitive, that the interviewee might make personal or political comments that they would prefer not to be on the record and far less so recorded.

Perhaps, the choice not to record was determined by the time when I started my Ph.D. research. In 1998, some scandals that revealed spying on King Juan Carlos I (who reigned from 1975 to 2014) and other public figures plus the use of secret funds were in the limelight. It was, therefore, not the most propitious scenario, in the midst of journalistic and legal investigations, in which to ask people whether the interviews could be recorded. That perception was confirmed by one of my first interviewees who made it quite clear with a wry smile: ‘Make a recording? Depends on what you want me to tell you, old boy…’ And here, my experience coincides with other researchers who recognize that although they may have permission to record, much of the richest information is obtained under the condition that they do not record certain parts or that the conversation, if recorded, may be less informative (Tunnell, 1998; Cremades, 2021). We must make a choice and strike a balance between extracting as much information and knowledge as possible without recording the interview, and working with a recorded interview and analysing literal excerpts. This approach ensures we do not rely solely on one’s memory and notes, although potentially can cause us to miss collecting some relevant testimonies from the interviewer.

Suspicions over recording remained for years. Some years ago, during an interview with a member of the Defence Committee of the Spanish Parliament, I inadvertently left my agenda on his desk. At that point in the year, it was already stacked with piles of paper that appeared to camouflage a microphone or camera. His repeated glances led me to remove the agenda from the desk and return it to the backpack; I may also add that hidden-camera TV programmes had become very popular in those years. In more extreme cases, simple gestures of touching the pen in the inside pocket of my jacket have raised suspicions of activating some recording device. In recent years, with the recording capacity of smartphones, every time I start an interview, I make an ostensible act of placing it in my backpack that I purposefully leave to one side, so that at the very least the phone will not be used as a hidden recorder.

Therefore, faced with the impossibility of recording, my practice has always been to take written notes and never even ask about recording the encounter, as solid guarantees of anonymity and confidentiality are given when arranging the interview. I also learnt never to note down what is of interest to me when the interviewee was actually recounting events, but instead some time afterwards. In doing so, my aim is not to attract attention to what is of interest to me and involuntarily condition the testimony of the interviewee. It is true that I might also involuntarily have made the interviewee think that what was being said while I was taking notes was of some relevance when it might not have been. In any case, it always appeared to be a form of controlling the rhythm of the interview, although it might be worth rethinking that aspect in the future.

The mental exhaustion of the strategy of not-recording the interviews almost goes without saying. When an interview is over, I go straight away to a nearby café or park to write up the interview from my raw notes. I followed the same strategy that I also noted later on in Hoffmann (2006), who neither recorded the conversations nor took notes during the meetings, and followed a similar routine. Upon leaving, he recorded keywords to recall as much data and as many impressions as he could; with no further delay, he went home and wrote it all down to review it against what he had recorded before anything was forgotten; a process that took him – like me – between 3 and 5 hours. The overall duration of each interview also varied enormously. In my case, some of them lasted for no longer than 40 minutes and others – with lunch or dinner included – extended over 7 hours. The average, however, was between one-and-a-half and 2 hours, with a mode of 45 minutes.

2.5 The Relationship with the Participants

Most of my interviewees have always been cooperative and willing to help over the years, without placing limits on the interviews in terms of time and topics to be covered. However, that in itself is no indication that they are in the position to provide relevant and voluminous information. Intelligence officers – and I include here medium and high-ranking officers of the service – tend to have very compartmentalized knowledge of their work within the organizational culture and the operational procedures of the intelligence services themselves are based on principles such as ‘need to know’, ‘need to be there’, and ‘need to share’. They can all be essentially summarized by the maxim: ‘what you do not know, you do not talk about’; or as an officer in South America told me: ‘The less you know the lighter the torture!’. During the interviews, almost all the interviewees coincided at some point in underlining that ‘this is my impression’ or ‘that much I do know’. I profoundly understood this feeling of compartmentalization when, after seeing my Ph.D. published by one of the largest Spanish publishers (Díaz-Fernández, 2005), a CESID head of Division confessed to me: ‘Thanks to your book I finally understood things and events that had happened to us; from inside the service, we are not able to grasp an overall view of our own organization’.

There again, elites – whether in politics or the intelligence services – are easier to locate – the latter less so – and are almost always more predisposed and willing to participate once they have left office. Despite being in a privileged position, they do not always remember everything or they may wish to convey a biased image of their political work in an area as sensitive as state intelligence. I recall no one willing to state mistruths nor blatant attempts to cover up responsibilities or the like. Only once, could I clearly detect that a person who was involved in a dirty-tricks operation was giving me very inaccurate information. In my research what I certainly did find was a ‘retrospective rationality’ (Wright & Decker, 1997). Some interviewees, at the time of telling their stories, either consciously or subconsciously transmuted what ‘was’ to confuse it with what ‘should have been’. Some tried to redirect the investigation towards different topics; however, I would explain any such diversions to the fragmented knowledge they held of each topic that was under discussion, as well as the effect of time on the clarity of memory.

Generating rapport is always essential when talking with research participants. However, it is even more necessary when interviewees are, by training or by nature, mistrustful and sceptical. Some of them are trained in lie detection and are good at masking their emotions and feelings and hiding details; they are charming and psychological seductors (Hammond, 2015), which, therefore, makes it more difficult to know whether the researcher is indeed breaking the ice or whether it is simple sophistry employed to elicit information from him. I must confess that over the years, I was very careful with my comments, and when unknown people were around, I tended to move to the role of observer and keep my mouth firmly closed. Perhaps that behaviour had no meaning at all; however, assimilating some of the routines and the behaviours of my interviewees was somehow the way that I followed as a junior researcher to approach such an intangible research object. My ‘watch and remain silent’ strategy has been even more intense with the arrival of social networks. Potential interviewees can scan your personal and political opinions, trips, and interests and, of course, have access to your whole professional curricula and thereby establish more or less your opinion and approach towards a research topic (Saltmarsh, 2013).

Very early on, my understanding was that any rapport should be based on a proper understanding of the motivations of the interviewee to collaborate in the research. The reasons to cooperate that I have identified are as follows. In my first years, some interviewees showed a willingness to help out a young researcher. Others helped me simply because somebody had suggested an interview to them and they agreed to make a commitment. Some others, having already surpassed the sixth decade of their lives – and others close to their 80th birthday – wished to share their version of what and how certain events had taken place. The idea that it would be set down for posterity represented a degree of liberation for them. I have listened on more than one occasion to the affirmation that ‘I am going to tell the truth about what happened’. And, some others were very curious to talk with a researcher with a wider understanding of the topic, which reminded me that Rubin and Rubin (1995, p. 76) defended the interviewees who ‘are more willing to talk in depth, if they conclude that you are familiar with and sympathetic to their world’. Finally, I have no recollection of any testimonies based on a wish for revenge against the intelligence services. Researchers should never be so naïve as to think that no other people in the study will be reaping some benefits from the interview.

3 Ethics, Secrets, and Protection of the Sources

None of the universities nor research centres with which I have been affiliated over the past two decades had an Ethics Research Committee, although that situation recently changed at the University of Cadiz that now has one to which I was appointed as a member and remain so at the time of writing this contribution. Over the years, the absence of these committees made it impossible for me to submit any protocol or informed consent for validation. There is no doubt that it would have been of enormous help for my first steps in research; anyway, it was precisely the lack of guidance that led me to look more closely at the ethical research challenges and the protection of my interviewees. However, the lack of any Ethics Research Committee meant that I never had to face up to certain questions that, even today, have yet to be clearly resolved in my mind.

The use of declassified information is not usually a problem. However, it is when the researcher ‘accidentally’ gains access to classified information. It can happen whenever a document has been incorrectly classified as open access in a classified file, or when a person has access to classified information that has been leaked in the media or on the Internet, or when people with access to classified documents share information, either voluntarily, by mistake, or – as happened to me more than once – to be sure if that was the information that the researcher was really interested in declassifying. At least in Spain, all citizens have the duty not to reveal any classified information they have accessed. The dilemma is what to do with the information your eyes have already seen and your brain cannot erase. I have used that information most of the time to guide the research or to question other interviewees without declaring the initial source. On other occasions, the interviewees guided me towards news on the media or towards other sources such as the memoirs of agents or politicians that, knowledgeable of its reliability, could be quoted. Without a doubt, quite a lot of content gathered throughout the investigation is lost, but the protection of sources comes at a cost and not revealing classified information to which one may have had access is a priority.

The second dilemma comes with the oral sources. Obtaining information from politicians, diplomats, and journalists hardly represents a major problem as the information they hold could be sensitive, but it is only classified information on rare occasions. However, information that intelligence officers share is a much more complicated situation, as they have a legal obligation not to disclose any information to which they have access during their career at the agency. I refer to my experience based on the Spanish legislation that represents the complexity of this issue quite well. The 2013 Estatuto del personal del Centro Nacional de Inteligencia [Articles of Incorporation of National Intelligence Centre Personnel] is crystal clear in article 75.1: ‘CNI personnel will be obliged to maintain professional secrecy with regard to CNI activities, its organization, and internal structure, resources, and procedures, personnel, installations, bases and data centres, information sources, and with regard to information or data that can lead to knowledge of the earlier material, in accordance with their classification as secret under article 5.1 of Law 11/2002 of 6 May, in regulation of the CNI, and information on the existence and the content of documents, identities, objects, and elements related with the above points, of which they may be aware. They will neither disclose this information, nor communicate it to anyone, nor keep it in their power on any media other than under the circumstances established in the legislation in regulation of Official Secrets’. Article 75.2 reads as follows: ‘Likewise, CNI personnel will abide by the duty of confidentiality and professional secrecy with regard to those facts or information that are unclassified of which they may have learnt in the course of their duties or by reason of their position, without making use of the information for personal benefit or for the benefit of third-parties, or in detriment to the public interest’.

In turn, the Spanish Criminal Code, under article 598, points out that prison sentences will be handed down to ‘Whoever, without intending to favour a foreign power, obtains, reveals, forges or erases information legally classified as reserved or secret, related to national security or national defence, […]’. The criminal offense, therefore, would be in ‘revealing’, which is a much more disagreeable concept than ‘using’. In this way, an intelligence official could ‘speak’ about aspects that may not necessarily imply revealing a matter that is formally classified as secret. The researcher can employ these conversations to for example, confirm a hypothesis without it implying the use of classified information or the consequent revelation by the intelligence official. A separate point is that some of the interviewee’s motu propio did not consider that the use that I could make of it might endanger national security; although that is somewhat irrelevant because the interviewees cannot grant themselves the capacity or the competency to declassify documentation and, thereby, decide what is sensitive or not for national security.

In 1999, and after a few months from the start of my research fieldwork in Madrid, I began to harbour serious concerns about the information that I was obtaining from members and former members of the intelligence services. I shared my doubts with a very relevant judge of the Military Court with experience in cases of espionage and classified information leakages. His statement was clear and calming: if I was not legally responsible for the custody of that information and that information objectively implied no risk to the national security of Spain, he would be unable to pursue a criminal case against me. As much as it helped, the pressure proved excessive for a Ph.D. candidate who might potentially face police questioning or be summoned to a Public Sector inquiry.

The following example illustrates how ‘talk’ will not necessarily imply the disclosure of classified information. I have quite a few times found myself in an almost identical situation to ‘Deep Throat’, the key informant in the Watergate scandal that brought down the Nixon presidency, which is masterfully narrated in All the President’s Men, filmed in 1976. With no need to go to a garage, the practice of some key informants has over the years consisted in confirming or denying the information laid before them – collected from the media, old documents, and some interviews. If they confirmed the veracity of the information, the interviewees excused themselves insofar as I already knew that information, and they were, therefore, not revealing it to me. In those cases when they considered that the information was not sufficiently accurate, they indicated as much, but without giving any further guidance, they insisted on meeting up again once I had more to show. It all took up a lot of time because looking for new leads and obtaining new sources for specific details is no simple task.

Directly connected with this legal provision is the need to ask the interviewees for their informed consent. As referred to above, talking with an intelligence officer is not directly synonymous with the disclosure of official secrets; however, a mere request to sign this document could, potentially, have administrative consequences, as the disclosure of secret information during the interview might appear rather dubious to the agency. That is the reason why I have never asked any of my interviewees to sign an informed consent form, making clear the terms of our conversation when the meeting was agreed, and that was explicitly explained at the beginning of my conversations. It could have been requested using a fake name or a code, however, in my opinion, it would not have changed the engagement that had been orally reached between the interviewees and the researcher. I cannot imagine any other purpose for signing such a document than to protect our universities from legal or ethical controversies.

Protecting the anonymity of the interviewees has always been an overriding goal of my research. The protection measures I adopt are spread through the collection, filing, and publication stages of the research. Firstly, I never recorded the interviews, thereby avoiding any risk of interviewee voices being recognized in case of non-authorized access. Secondly, I codified the names of the interviewees that were never written down in my fieldnotes and that, consequently, led me in the past to create an imaginative way of recalling whoever was behind each code; for example, they were codified by the name of a plant or tree according to a letter in their names – whether the first or the last or any other will not be disclosed here. And, thirdly, at the publication stage, I altered some data in the profile of the interviewee to forestall deductive disclosures, although without losing relevant information that could help the reader to understand the position from which that interviewee was giving his information and statements. Needless to say, although this technique can diminish the risk of a deductive disclosure, it is not as easy as may be thought in some areas where only a few people share very specific profiles.

4 Positionality and Emotions in a World of Secrets

My positionality has evolved with the passing of time and is directly linked with my emotions and how interviewees may perceive me. Anxiety arises over insufficient access to interviewees and materials and, therefore, to relevant information. Insecurity is triggered whenever not enough is known about a complex topic and faced with a feeling that the fieldwork will never be well enough prepared, due to a lack of open sources. Likewise, scary situations when trying to complete the Ph.D. dissertation, doubts, and jokes from my colleagues and peers about my real work in Madrid: spy or researcher? Data overload when all the information gathered in each new interview appears overwhelming as names, places, and events… rain down in the words of the interviewees and where everything is always new and always unknown; bewildered and lost within a world of secrecy.

These varied emotions explain – and are explained by – how my positioning has evolved with respect to my research object: an evolution of over 25 years that began with my Ph.D. It is assumed that credentials legitimize the entry and permanence of a researcher within a particular field. However, after some time, roles and positions with respect to research changed. During the first months of my doctoral process, I was simply a junior researcher from the University of Barcelona in the academic-military environment of the General Gutiérrez Mellado University Institute of Madrid. Running on intuition – and to be honest, lacking deep knowledge – I kept my eyes and ears open but my mouth shut; good advice, confirmed in the work of Polsky (2006). Sometime later, after extensive reading and some more experience in the field of intelligence services, I moved on to the role that Lofland and Lofland (1984, p. 70) defined as the ‘non-expert’ or the strategy of the ‘man from Mars’, which is characterized by adopting the role of a novice or apprentice.

Following the ‘non-expert’ phase, I progressively and consciously adopted the role that Leverentz (2010) referred to as ‘acceptable incompetence’ to generate greater confidence and gain access to more and more detailed information. That game and the progressive changes in my positionality were made possible by my youth – starting my Ph.D. at 26 years old. However, in the analysis of Seldon and Pappworth (1983), age has effects on the level of information that interviewees may share with a researcher. Interviewees can consider that the researcher is too young to grasp a historical understanding of very distant aspects and lacks the personal context to assimilate them. That is the reason why, on occasions, I had to reiterate that my knowledge and capabilities – while throwing some data on the table – were sufficient to understand the topics about which they were talking.

However, in all honesty, with my youth and not having lived through all those events, everything almost always appeared reasonable and credible. Something that was exemplified by a warning that an old senator whispered in my ear. Despite my naïveté, his words helped me contextualize a specific situation: ‘that pleasant old man with whom you had some refreshments only yesterday was the fearsome director of Franco’s secret service’. Otherwise, I must confess that I never felt – or at least I do not remember it as such – what is labelled as ‘ethnographic seduction’. Ethnographic seduction places the researcher in a pleasant situation; it creates a feeling that the researcher has reached a profound knowledge and understanding of both the circumstances and the interviewees, that harmony exists, and that there is a fluid and sincere exchange of information (Robben, 1996). At that time, I started feeling that my positionality was changing in the field, however I still had to gain a much clearer notion of the terrain over which I was travelling.

Clarity descended the very day I understood that I had accumulated excessive knowledge, my sources were too distant from the initial contacts of the first snowball sample, and I also knew the inner workings of the organization too well. A former highly placed civil servant of the presidency of the government drew this turning point to my attention towards the end of the year 2002. In response to the question of whether he could tell me something about Operation Fénix (an internal restructuring programme of CESID), he smartly answered: ‘Well, you do know a lot…! There are people who have ended face down in the Manzanares river for knowing less’. Without a doubt, it was a joke for the initiated, funny after many years, but a bit shocking when you are in your 20s. That day I understood that the interviewees saw me with different eyes and I had to accept my new positionality in the field.

Various years had passed since the start of my research on intelligence, and the transition from novice to expert researcher took place, a transposition that was by no means simple. In my case, the role of expert researcher placed me in a position too close to an insider. So close that even some colleagues and relatives doubt that I am really a university professor; but being branded by the research topic is part of the game, especially with sensitive topics (Moon, 2012). An insider is not so much a formal member of the group, but someone who shares values and culture with the population under investigation (Sherif, 2017) and, when the group considers that an insider – one of their own – is at the door, access is easier. It has affected my positionality as it brings about what Mitchell Jr. (1991) called the ‘paradox of intimacy’. There are things that I am supposed to know and that I evidently never knew nor know nowadays, but I cannot ask for elucidation, because I am almost perceived as an insider. Frequently, I am in a position to need to fill in the existing gap between my real knowledge and a new piece of information that has been disclosed, thanks to ‘my trustworthiness’.

After working for so long with people trained to deceive, to hide things, feelings, and emotions, to make another person feel amazing, in order to generate trust, and to obtain information, I have become more suspicious and that arises every time that I receive a relevant piece of information and I wonder: Why me? Why now? What is the purpose?... That suspicious character of spies can worm its way into the researcher’s own personality who will assume as a fixed starting point that not everything is as it seems. This ‘on-alert’ mode is of greater complexity when I have had access to the family circles of some agents and have developed an intermittent relationship with them. To know personal information about their children’s studies, future projects, health issues… and, at the same time, to be sure that their real names are not the ones they have been using over the past few years can be shocking. I do not question that a sincere and honest relationship can be established with some interviewees; however, effective intelligence officers will never be content merely acting as information sources as, like the confession of the scorpion to the frog in the fable, it is in their nature, at least to gather information.

5 Lessons Learnt and Methodological Perspectives

The lessons I learnt are perhaps not numerous, but they are very illustrative of how to work with intelligence officers. The first one is honesty; it is essential to keep your promises and to concentrate, so as to stay within the limits of the topics proposed to the interviewee. Spies are a highly suspicious group of individuals and moving beyond the focal point of the interview could in their mind mean that you are not a reliable researcher and that you hold a secret agenda. In such reduced groups, a bad reputation can jeopardize any future interviews, leaving your research to flounder.

Second, this topic has a potential impact on junior researchers; being in frequent and close contact with spies can transform the researcher into a distrustful person. This does not happen overnight. Those tiny but recurrent routines, tests, tics, and behaviours such as always confirming the restaurant at which to meet at the last minute, not receiving – beyond a smile – any acknowledgement of your comments, or warnings such as ‘please, do not send .pdf documents, only Word files, they are more secure’ are elements which have drop by drop – as with the Chinese water torture – undoubtedly impacted on the personality and behaviour of this researcher.

Third, triangulation must remain one of the research objectives. Although access could be granted to a sufficient sample of interviewees, the different stories that are illustrated in the documents are substantial enough to extract conclusions. Like the tip of the iceberg, 90% is hidden from our eyes; however, the tip is so visible and the part beneath the water is so hidden that we may often think that the tip is all we need to know and to understand.

The sensitive topics that are investigated lead most researchers to agree that some of the guidelines of our Ethics Research committees cannot be applied to them in full, due to their special features. Research into the intelligence services has, undoubtedly, made sense of the legal obligation to abide by the duty of secrecy. It necessarily affects the way we agree to the terms under which the interviewees participate, which may include the use of informed consent, recorded interviews, and the protection of their anonymity during the fieldwork and at the time of publishing our research.