1.1 Interactionism

What does it mean to take an interactionist approach to information control? It means quite simply that we ground our study not in some fixed image like a wall of secrecy or deep state but the dynamism of the interaction and struggles over knowledge. In Hilgartner’s case it is a matter of ethnographic observation: drawing on Goffman, it means one might study the games of actors belonging to different laboratory teams. How do they interact sometimes as rivals, sometimes as collaborators? How do they strategize over what information they are prepared to share with whom and for what purposes and what they seek to withhold and why? But ethnographic observation is not adequate to understand all forms of knowledge control. Other methods, such as documentary and legal analysis, are also necessary to understand, for example, how corporations and states compete in struggles to own genomic knowledge.

We argue an interactionist perspective is especially relevant for understanding the power relations immanent to the field of foi. It alerts us to the games, strategies and moves which actors make in their bid to shape what, when and how knowledge is disclosed or not – whether as information officers working in the Home Office, journalists and academics seeking access to the ‘backstage’ (Walby and Larsen 2012: 39) of governmental activity, or case officers in an ombudsman’s [sic] office who referee the games between these players. Interactionism situates actions in time. Each request builds off previous requests which may have been made by the researcher or other researchers, and one can even submit a meta-request about an authority’s handling of a previous request (Rosenbaum 2023: 34). Likewise, the way in which state actors respond reflects tactics they have developed in relation to previous interactions.¹ This means that rather than treat a given foi outcome in isolation, we consider how foi has a concatenated, web-like structure, unfolding over time and taking pathways that might not have been predicted at the outset. A given disclosure may lead to new requests, just as a denial might spark new requests or appeals.

However, in foregrounding the interactionist character of foi we need to offer three caveats. First, the forms of interaction which Hilgartner foregrounds when investigating information control amongst scientific communities involve significant elements of face-to-face interaction. But with foi one is looking primarily at interactions that are overwhelmingly mediated by textual interactions. That said, it would be wrong to imagine that foi is nothing more than the exchange of messages between distant strangers. Building on Walby and Larsen (2012: 35), we venture that successful foi research will profit from forging other relationships alongside these formalized communications and exchanges. This might be the off-the-record verbal tip from an archivist offering guidance to a researcher as to what document to request and how (Jones 2013: 53). It might be an informal relationship cultivated with a particular information officer who knows how their records are organized and what terms need to be used to solicit specific information (Rosenbaum 2023: 19) or with other foi users, such as investigative journalists. In these ways researchers might navigate the significant imbalance of power and resources built into the foi regime.

This question of imbalance brings us to a second caveat. When studying information control amongst technoscientific workers, Hilgartner (2012) is speaking to games played on a somewhat level playing field. There are different laboratories, different research teams, and they play games of sharing, withholding, leading and perhaps even misleading each other. But interaction in the realm of foi takes place on a field that is not at all level but structured by profound institutional, political and legal asymmetries of power and information (Trudel 2024: 139). These asymmetries manifest in numerous ways, including formal and legal provisions which, in the name of ‘national security’, ‘commercial confidentiality’ or ‘privacy’, might exempt various categories of information from the scope of foi, to the informal tactics which governmental agencies can use, such as stonewalling (Walby and Larsen 2012: 34), delaying and frustrating requests, or, in more extreme cases, hiding, transferring or denying the existence of entire archives (Anderson 2015). Even in these extreme cases, foi can be a valuable tool if it makes visible the absence of an archive and other holes in the record.

This leads to a third caveat. Under certain circumstances the actors engaging in foi can engage in ‘boundary work’ (Hilgartner 2017: 7) by changing the venue where the interaction is staged. This is the case, for example, when referring a request blocked by the authority to the Information Commissioner’s Office (ico) or tribunal for arbitration.

1.2 Symmetry

While asymmetry may be the name of the game when it comes to describing the power imbalances facing the foi researcher, a principle of symmetry prevails when we consider what is known, how it is known, and what is not known. Hilgartner rejects the view that publicity is a kind of natural and bounteous state of affairs, a default situation with secrecy understood as an exception requiring effort and explanation (see also Galison 2004). Instead, he places secrecy and publicity on a continuum of knowledge control. Making particular knowledge public requires work such as legal contracts, editing, printing, social media campaigns and hardware like public archives, libraries, servers and websites etc. It raises questions of form: is knowledge materialized in a report, a spreadsheet, a memo, etc. Likewise, making knowledge secret also requires work, be this the work of classification, redaction, secure holding areas, oaths of office or the selective destruction of records. Rather than seeing secrecy and publicity as a binary we are encouraged to see them as co-produced and multiple. Nowhere is this clearer than with the redactions that are a part of so many foi releases (Connelly 2024; David 2007; Gilbert 2016). These black or white shapes cleanse a document of names, places, actions, preparing it for release to a requester, but also, as we explain below, occasionally they fail to obscure the underlying text.

1.3 Technologies, Materials, Infrastructures

Knowledge control is not reducible to the interactions of human actors. It is also mediated by the heterogeneous array of technical forms and devices these actors draw upon in processes of knowledge production. Like ant scholars, Hilgartner (2012, 2017) takes a symmetrical view of actors as well the technologies which mediate the transfer of information, allowing space for the irreducible and sometimes unpredictable contribution to information control made by non-human actors like test tubes, tissue samples, technical reports, password-protected spreadsheets and databases, encrypted documents and much else. He asks how these devices and infrastructures differentially constrain or capacitate the production and circulation of knowledge.²

Materials and devices mediate the way in which information is circulated and the selective flows of knowledge in the realm of foi. This begins at the level of the database. If and how information is coded, indexed, and stored can determine whether it is released in response to an foi request. Although an authority may ‘know’, for example the demographic details of a cohort, unless that information has already been extracted from case files and kept as a statistic, retrieving and collating it will likely exceed the cost-limit which authorities use to govern requests, leading to the request being refused.

Technologies and devices can be set up in a way that confronts researchers with encumbrances and constraints. Sometimes it is a matter of the friction of physical space. For example, today the cia’s database of declassified historical records (the cia Records Search Tool, or crest) is housed in its Electronic Reading Room alongside documents generated by foi requests – all these records are accessible to anyone with an internet connection. But from 2000 to 2017 the only way to access crest was to physically use one of four terminals located in the National Archives and Records Administration (ii) at College Park, Maryland. Furthermore, documents could only be printed; downloading was not permitted (Mistry 2021: 13).

Another example of such encumbrances is the European Border and Coastguard Agency’s Public Access to Documents system which was set up ‘as a flagship effort towards transparency and accountability’ but which remains ‘extremely user-unfriendly’ (Karamanidou in Ambri and Carron, 2024).

Attunement to the mediating work of technologies and infrastructures adds nuance and texture to the way we study foi. For example, in what format are records released to a researcher? Are they digital or paper? Has the authority provided only sparse data which strictly responds to the request or ‘dumped’ a mass of raw data? Has the information been shared in an unlocked spreadsheet that can be readily repurposed, searched, combined, or provided as a data table in a pdf document which the researcher has to then extract data from again? There is clearly a micropolitics of format.

But this point about formats and devices comes with a caution as hinted with our crest example. Are authorities acting malevolently in the choices they make regarding forms, or is this merely happenstance? Are they strictly playing by the rules or bending them? The risk is to engage in ‘dietrology’: imagining bad intentions or at worst a conspiracy operating ‘behind’ scenarios that may or may not warrant them (Bratich 2008: 15; see also Rappert 2012). Perhaps we have to allow for a certain ambiguity.

One additional advantage of foregrounding the mediating work of technologies and devices is that it allows us to see foi not just as the outcome of a prior societal demand for freedom of information, or a sweeping cultural movement for transparency and open government, or policy reform. Instead, we can analyze it as an assemblage of elements and practices that have been combined in a particular way (Collier 2009). This opens the possibility not of a linear, institutional account but a genealogy of foi that materializes this practice and disperses its origins. Scholars typically narrate the story of foi in terms of legislation. Hence, it is often noted that the United States was a pioneer in passing modern foi laws in 1967, Canada was part of a second generation in the 1980s, and a large number of countries followed suit from the 1990s onwards (Gilbert 2023: 188). However, a genealogy of the constitutive practices of foi complicates this picture. For example, we could ask: where and when did a practice of allowing actors to submit requests to access particular closed documents and secretive records emerge? When were expert reviewers first empowered to assess these requests and determine what could be released, in what form, and what knowledge should be kept closed? From where did such an individualized, customized, interest-based mode of governing closed information arise? Here one might look some time before the formal emergence of freedom of information. Instead, we might consider the birth of a policy of ‘declassification’ concerning atomic energy research within the Manhattan Project in 1945. This was an arrangement that included the creation of ‘responsible reviewers’ who worked rather like peer reviewers for scientific journals, declassification guides and standardized processes whose brief was to assess whether and what risk the release of any given document might pose for national security (Donnell 1946; Wellerstein 2021). In short, it is fruitful to think about foi at the level of its constitutive practices and devices, and not just as a law or policy. foi is like any other policy: it is not simply willed into being. Instead, it relies upon the invention and adaptation of particular technologies of government to give effect to its ambitions and the exercise of particular rights and freedoms now associated with it (Rose and Miller 1992).

1.4 The Constitution of Actors

The final analytic we develop from Hilgartner concerns the identification of actors. Hilgartner speaks of the ‘governing frames’ of knowledge-control regimes, and how these frames mark out particular kinds of actor (2017: 11–13). Just as the constitution of a democratic state identifies actors like ‘the executive’, ‘the legislature’, ‘the judiciary’ and ‘citizens’, and accords them particular powers, responsibilities and expectations, then so too do knowledge-control regimes. For example, scientific publication in journals is governed by identities such as ‘manuscript author’, ‘editor’, ‘reviewer’, ‘publisher’ and ‘readers’. The scientist who wants to publish their work in a journal cannot interact with a journal in an ‘arbitrary’ way; they are obliged to align themselves with the expectations and burdens of being an author and respect the powers and responsibilities of these other types of actor.

We find this concern with the constitution of actors and their place within governing frames especially useful for thinking about foi as a regime shaping selective flows of knowledge. For example, in contemporary foi policy we see the concept of the ‘requester’ as a particular kind of actor. Unlike the private library or company archive that wants to know who you are before granting you access to its collection,³ foi is supposed to be applicant blind. There is the key norm that authorities should make the same decision irrespective of the identity or motive of the requester. Release of material is to depend on the nature of the material alone and not the question of who wants to see it. That said, in the UK, one notable exception to the applicant-blind nature of foi can arise if the authority suspects the request to originate from someone acting as part of a campaign or ‘in concert’ with others. In these cases the request can be aggregated with similar requests from other people, and one or all of them refused due to the cost-limit. The authority might also deem the request to be ‘vexatious’ – an unjustified use of the foi procedure. These caveats reinforce the implication that informational governance be depoliticised, even when the information at stake is highly politicised. Nevertheless, we can say that formally, if not always in practice perhaps, the work of control is concentrated almost entirely on the file or document, not on assessing the identity, motives or profile of the requester. Whether it is released or not, how it is released, when it can be released, these decisions depend entirely on the authorities’ assessment of the record in question, not the requester.⁴ Likewise we see other identities taking shape, such as the information officer and the commissioner. They too have certain norms, duties, expectations, responsibilities and powers attached to them.

We could extend this point about the constitution of actors by connecting it with Hacking’s (2004) idea of ‘looping effects’. It seems that for many decades foi requesting was disregarded by scholars who saw it as part of the world of investigative journalism and activism but not a respectable or desirable mode of scholarly inquiry (Walby and Larsen 2012). This is changing. The more that scholarly articles, chapters in research handbooks, websites and books (e.g., Rosenbaum 2023) begin to flourish offering best practices and guidelines for conducting foi research, the more that the foi researcher emerges as a new kind of research subject possessing certain skills, habitus and ethics. Perhaps one could even say that foi is becoming a new kind of normativity, a benchmark for what is to count as serious empirical and scholarly inquiry into the affairs of states and other powerful agencies.

2.1 Symmetry

Our analysis of the use of foi in this paper primarily focuses on the use of charter flights to deport or remove persons from the UK. This is a subject which in itself illustrates the principle of symmetry, namely the need to attend equally to practices of revelation and concealment in information control.

Charter flights were initially justified in government statements as an option to remove a large number of deportees in a more efficient and effective manner than deportations on regular commercial passenger flights (Bosworth and Singler 2023: 175). Sometimes they were proposed as a way to create routes to destinations not adequately served by regular flights (Home Office 2021). But deportation charters are not reducible to logistics. For it is their paradoxical visibility that is crucial. They are important for what they hide as well as what they display. They hide violence and abuse that would disrupt a regular flight: on the charter plane there are no regular passengers to bear witness or intervene in your removal (Fekete 2011; Walters, 2024). But there is also the question of what these charter operations reveal. In recent years, the charter flight itself has become a means through which states spectacularly perform immigration enforcement, despite the declining numbers of people placed on board and the periodic cancelled flights. For example, the Operation Esparto flights which removed ‘small boat’ Channel-crossers from the UK to European countries under the Dublin agreement almost always had fewer than 20 people, all asylum-seekers, onboard, and even as few as one (Corporate Watch 2021). Nevertheless, each departure was highlighted in Home Office social media posts to prove the department’s determination to end Channel crossings, while ignoring or obscuring who the people were on the flights as well as the violence necessary to enforce their removal.

Another example of such spectacles of enforcement is the attempt to use charter flights to deport asylum-seekers to Rwanda. Analyzing internal communications about a video promoting arrests and detentions of people for removal to Rwanda (Home Office 2024) in order to ‘mark the milestone’ similarly shows a calculation of revelation and disclosure. In one email civil servants ‘raised some concerns about the optics of the handcuffs shot’ and asked if it was possible to edit this out (Petit 2024). Today, the Rwanda project has been scrapped as government policy. Had the Home Office succeeded in implementing its unlawful policy, any publicity would likely have been similarly concerned with obscuring the detail of violence done to people to get them on the plane, as documented during the first failed flight (Walawalkar, Rose and Dearden 2022).

Freedom of information is a powerful tool to force the disclosure of exactly this kind of information authorities would prefer to have absolute control over, and reintroduces some parity into what information is publicly available about deportation charter flights. Among all the information each flight generates – manifests and itineraries, biographical information of deportees, operational costs, identities of the airlines involved, number of security personnel employed, records of events onboard, etc. – foi practitioners have succeeded in obtaining some limited information such as departure dates and destinations, operational codenames, use of force incident report forms, numbers of ‘foreign criminals’ onboard, as well as internal Home Office communications regarding their preparation of social media posts. Such information can, on its own and in combination with other sources, offer us insights into, and opportunities to unsettle, the control of authorities in selectively choosing what information to disclose, and how.

As we’ve seen, foi is particularly useful in making visible the information control and public relations strategies of government officials (Walby and Larsen 2012: 32). To counter the symbolic portrayal of deportation charters we requested the correspondence of civil servants discussing how charter flights were publicized in 2023. The mundane correspondence disclosed primarily consisted of officials in the Communications Directorate asking their operational colleagues for the most up-to-date numbers to include in the ‘drumbeat content’ meant to powerfully display the Home Office’s commitment to removing Albanians and so-called foreign national criminals. Each week the Communications Directorate would recycle a Tweet template text and graphic, but with the numbers of people, list of their crimes and overall prison sentences changed to reflect that week’s deportation. Illustrating the propagandistic purpose of these posts, the response obtained by another requester (who inspired ours) for a flight to Nigeria in 2022 reveals one civil servant referring to ‘foreign national offenders’ and stating ‘we want the biggest number possible’ (Abraham 2022). A colleague was asked to amend the original draft which specified ‘11 dangerous foreign criminals were removed on charter flights … 10 immigration offenders were also removed’. The final version combined the two numbers, blurring the difference between foreign criminals and immigration offenders, and emphasizing the aggregate of criminal sentences as being in excess of 64 years ‘for crimes such as rape & sexual offences against children’. In this way the final version demonized all deportees as criminals and sexual offenders, a common trope through which foreigners are made into ‘monsters’ against whom violent state practices are necessary to protect the public (de Noronha 2018: 253). Freedom of information facilitates uncovering what information is obscured and withheld as related information is disclosed and publicized. A request for the hidden breakdown of statistics displayed in plain sight can provide insight into how selections for revelation and concealment are made by authorities.

Similarly, activists in the UK have used foi requests to help identify the private airlines carrying out deportation charters, and pressure them to refuse to collaborate in future flights (Corporate Watch 2024). Sometimes these campaigns are successful, as when multiple charter airlines, including Privilege Style, agreed not to fly a deportation charter to Rwanda for the UK government (Ward 2022). Acquiring this information required careful tactics. Any request for a list of private airlines collaborating in deportation charter flights would be refused by the Home Office based on the section 43 prejudicing commercial interest exemption. However, the dates and destinations of flights cannot be refused. This data can then be combined with other contextual sources to deduce the airlines involved despite the Home Office’s efforts to withhold the information. foi is a powerful tool: it has its limits which should be recognized but also, sometimes, can be overcome.

2.2 Technological Mediation

Taking seriously the role of technologies and devices is important because it offers a more precise account of the work of foi. But it also better explains situations where unintentional disclosures happen; where foi takes unexpected turns. For example, consider the rare but sometimes significant phenomenon of failed redactions emerging from software glitches or human error. These are uniquely electronic artifacts which occur when an authority’s information officer covers over passages of text to make them unreadable, but fails to delete the underlying computer readable characters invisible to the human eye produced through optical character recognition (ocr). (Readers may have performed this process when scanning a book into an electronic pdf format to make the text searchable.) pdf editors come with redaction functions that should remove ocr text while also covering the plain text font, but in the hands of an inexperienced operator, or subject to a glitch, the information can still slip out. Only once did we encounter redactions which did not purge the underlying ocr text. Clicking and dragging the mouse pointer over the section revealed the supposedly sensitive information of how a government agency handles alerts regarding irregularized migrants for all to clearly view, copy and paste. The question for us then became whether to reveal the information and its source, drawing the authority’s attention to their poor practices, or remain silent, not share the disclosure publicly, and hope to receive more failed redactions in the future. The games of concealment and revelation are thus not only for authorities to play. Requesters too will strategically avoid revealing the extent of their knowledge.

This case of a failed redaction illustrates why attention to the technologies through which data is collected, stored, becomes searchable, cleaned and perhaps eventually disclosed is crucial to understanding the information games played under foi rules. Knowing the architecture of a database, what information is stored within it and how, can increase the chances of a large disclosure by allowing for very specific questions to be asked. When first beginning an foi investigation it is recommended to start by asking questions about the authority’s data practices and information management system to better guide follow-up requests rather than ask direct questions for which authorities might have to perform manual trawls of case files and can therefore refuse based on the cost limit (Rosenbaum 2023: 56).

Negotiating the form in which information is to be released is also an important aspect of the game of freedom of information requesting. How easy it is to analyze the data one obtains – compare with requests from other authorities, aggregate it with historical data, understand patterns, manipulate it to produce tables of visualizations etc. – depends in large part on how it has been packaged and sent. Under the UK’s Freedom of Information Act 2000, a requester can specify the format in which they wish to obtain the information, but the authority does not always abide by the request. In almost all the responses we analyzed information was provided in the form of tables within large pdf text documents, even when spreadsheets were requested. Constructing a table of all deportation charter flights carried out by the UK since 2010 thus first required the lengthy work of extracting the data in the table from the text and placing it in a spreadsheet for transformation and analysis.

After months of delaying and denying a request, an authority will sometimes make a different move. They will issue a torrent of disclosure in which the most relevant information is submerged in the overwhelming noise. Combined with a particularly onerous data format, this can be an effective ploy to continue suppressing information from the public despite appearing to offer it in abundance. An acute case of this appears in the Home Office’s response to requests for use of force reports against deportees (Petit 2023; Walton 2024). After initially delaying the disclosure to conduct a public interest test, the Home Office then couriered hundreds of printed pages of report forms across London, stating it would be too onerous to publish the information electronically. This not only impacted the ability to analyze the disclosures (requiring requesters to further transport the heavy bundle and scan its entirety with a photocopier), but has prevented other members of the public accessing the information through the platform WhatDoTheyKnow.com (where the requests were made). Unfortunately, this website only displays information provided in emails and attachments, and does not allow for files to be uploaded by requesters. The same issue exists with Frontex’s new Public Access to Documents (pad) register which means that information provided through it is not visible on the AskTheEU.org portal (which also runs the Alaveteli software) – for example, see the recent request about deportation flights coordinated by the Agency made by Baerwaldt (2024). Additionally, information released through the pad is only in the form of a temporary file hosted on the authority’s file server along with a password. The requester will be able to access and download the files within the time limit, but no longer be able to see it once the file has expired. Thus, despite the overall increased transparency of information facilitated by the online platforms, the ways in which data is encoded, formatted and transferred by authorities allows them to continue to render it opaque even when appearing to comply with requirements for transparency.

foi requests today are an increasingly crowd-sourced endeavour thanks to websites like FragDenStaat in Germany, MuckRock in the United States or WhatDoTheyKnow (the UK’s implementation of the Alavateli software). Users are ‘more than human’, to use a popular sts phrase. These platforms assist people in writing and sharing foi requests, and exist separately for different countries to better negotiate the specificity of freedom of information laws across jurisdictions. These sites facilitate writing requests by collecting authorities’ foi office addresses to make them easier to contact and suggesting phrases to include in one’s correspondence to counter specific refusals. They also work as an archive of all requests and responses made through the sites and a clearing house of expertise. Not only does this furnish ideas for future requests, but shows what tactics have worked or failed for others users in the past. Especially for inexperienced requesters, the examples on these sites provide valuable lessons obtained through the trial and error of thousands of people who have come before. The list of decision notices and enforcement actions on the Information Commissioner Office’s (ico) website works similarly. Citing these judgments in one’s requests can powerfully demonstrate to an authority that the jurisprudence shows the information should be released without delay. However, the texts of these rulings also provide valuable insights into how the ico appraises complaints which come before it. A skilled requester might use these previous examples to formulate their complaints and improve the strength of their arguments in the cases they refer for arbitration.

The wealth of requests, responses and disclosed information contained within the ever growing online data repository constitutes not just a ‘live archive’ (Luscombe and Walby 2017) but what Gilbert (2016) has called a ‘counter archive’ (see also Reynolds et al 2023). The counter-archive contains not only the abundance of previously concealed information produced by requests which may be of use to journalists, activists and researchers seeking to understand the governance of immigration enforcement. Crucially it also demonstrates how such information can be effectively gleaned from obstructive authorities. It is both data and method – Petri dish and microscope – added to and focused over many iterations of request, response, internal review and arbitration.

It should be emphasized that the intentions, politics and agendas of actors who contribute to this ever-growing foi counter-archive are not necessarily aligned (Walters and Luscombe 2020). There are likely just as many requests for information seeking to uncover details of the government’s migration policy which might be put to use by anti-migrant groups as those from groups with an interest in furthering struggles for the freedom of movement. Information can also be reframed and repurposed regardless of the requester’s original intent to obtain it. Nevertheless, all requests, and the information they generate, exist within the same informational field of the counter-archive, are produced through the same techniques, and provide useful insights into state practices despite differing motivations.

2.3 Actors and Interactions

We consider the foi websites just discussed to be sociodigital technologies; that is, they are digital technologies which shape while being shaped by social processes. They are a form of social media platform in which participants interact, share content and aggregate resources to redress power imbalances in the playing field with the authority who has its own institutional knowledge of foi and employs experts in the field who often attempt to deny, delay and delude rather than reveal.

Just as Hilgartner so carefully observes in studying the making of scientific information, the use of foi is fundamentally a social activity. This may not be obvious at first. Interactions take the form of carefully crafted letters between requesters and authorities, often constructed from templates and containing long passages of boilerplate. Responses appear to be assembled mechanically, with canned paragraphs copied and pasted in. For example, one description of a public interest test from the Home Office we reviewed referenced asylum accommodation. This was not the actual subject of the request for information on asylum applications that the Home Office was treating as withdrawn (Petit 2023). Similarly, requesters may take suggested response paragraphs disputing the use of a specific exemption from an foi portal or previous request when preparing an internal review or complaint to the ico. The tone of correspondence is most often one of stern politeness, contributing to the depersonalization of all parties. However, requesters are best served by remembering that information generated through use of the Act comes through their interactions with case officers, internal reviewers and occasionally tribunal judges. These are people who work in bureaucracies, but retain their discretion and appreciate requesters’ assistance to make their work easier.

Before submitting a request it can often be useful to try and contact someone at the authority who understands what information is stored (Rosenbaum 2023: 19). Time is the currency of foi, and a small initial investment to understand what request may or may not be successful can avoid disappointment following a month-long wait, and delayed progress of the investigation. Case officers may also themselves be unfamiliar with the information they hold and how it can be retrieved within the cost limit. While Home Office case officers have proven themselves capable of searching the email inboxes of its civil servants for relevant correspondence, journalism students who conducted an foi audit – comparing responses and practices to the same request across a large number of authorities – described the necessity of providing clear instructions to information officers on how to export emails into portable formats like .mbox and .pst in their requests (Thuy Vo, 2024). In other words, this was a case of users tutoring the officials. As alternative messaging applications like WhatsApp have been increasingly used by civil servants and policy makers to conduct governance (Smith 2024), these should be explicitly mentioned in information requests to ensure communications are not missed by information officers who may not be aware of their use or who may choose not to search them if possible.

foi officers also have a pedagogic role despite their agonistic position. Section 16 of the UK’s Freedom of Information Act places a duty on authorities to advise and assist requesters, meaning that even refused requests will contain suggestions for how the information might be able to be obtained, illustrating how all freedom of information requests are a dialectic of concealment and revelation taking place over many interactions. For example, in our experience we learned that a request for all correspondence which related to deportation charter flights over a given period of time may be refused as it would require an officer to analyze all emails to see which ones would be respondent. However, providing a list of search terms would yield disclosure as it would allow the search work to be performed quickly by the email client. This search term tactic helps the officer fulfill the request by relieving them of the task of deciding how to parse it in exchange for the requester risking that some relevant information will be missed which does not contain the keywords.

Finally, reviewers within the authority, as well as arbiters in the ico and tribunals, must be appreciated for the role they play in the game of concealment and disclosure. Despite the presumption of transparency and the duty to disclose which is the foundation of freedom of information laws, our experience is that authorities will look to refuse, if possible, through any of the exemptions they have recourse to. The referee’s job is to take another look at the existing correspondence and see if the authority is playing by the rules, and has responded in line with best practices. Their scrutiny is focused on the technical application of exemptions and the overall processing of the request rather than on the retrieving the information themselves. Most importantly, arbiters will weigh in on the public interest tests performed. As a third party they will assess arguments for and against disclosure and can even force authorities to reveal information which they may have previously considered not to be in the public interest because it was potentially embarrassing to the authority itself. Many times refusal decisions are not reversed and complaints not upheld following requesters’ appeals to the ico or tribunal. However, each time a request or response is placed in a new forum of judgment the chance of disclosure is renewed. Sometimes there are even resounding successes; for example, the Tribunal recently ordered the Home Office to disclose the number of emergency travel documents issued for Somalia and Eritrea and in its judgment chastised the department for its established ‘pattern of conduct’ which was ‘within neither the spirit nor the letter of foia’ as well the ico for accepting the Home Office’s ‘bare assertions’ as to why the information was exempt (Lenegan 2024). The bounds of secrecy which govern freedom of information are thus contingent, malleable and defined through the interaction of requesters, authorities and arbiters negotiating specific cases rather than definitively fixed in advance.