Introduction
Me: ‘So it’s like “the Council Conclusions say that if there is refoulement there should be consequences”, which raises the question “is this refoulement?”, to which the answer is “we don’t know”?’ Them: ‘Yes, the principle of non-refoulement should be upheld.’ Me: ‘Yes, but whether that happens, we don’t know. And that’s that?’ Them: ‘Yes.’Footnote 1
The question of how states and international organizations address tensions between norms and interests is central to international studies. Such tensions culminate in the realm of migration, where normative principles relating to protection that are enshrined in international law often clash with geopolitical containment interests resulting in violent deterrence.Footnote 2 Faced with such tensions, governance actors often resort to inaction and ambivalence.Footnote 3 But while we know why certain international actors may utilize such strategic ambiguity, there is a lack of empirical evidence and theorization on how they do so.Footnote 4 Addressing this knowledge gap is important because it helps us to better comprehend how international law can be upheld or eroded, how opaque negotiations and considerations shape foreign policy, and how knowledge production enables or precludes political agency.
An emerging body of literature turns to the notion of epistemic politics, the exercise of power through producing or obstructing knowledge, to investigate how practices of (un-)knowing generate and legitimize inaction and ambivalence.Footnote 5 Building on and contributing to this literature, this paper explores EUrope’sFootnote 6 engagement with contested regional refugee return to Syria-under-Assad and its response to instances of potential refoulement. It does so through the concept of agnopolitics, the strategic use of ignorance. Agnopolitics is leveraged as an analytical instrument to unpack the migration diplomacy processes through which international actors seek to navigate contradicting normative positions and geopolitical interests – here between non-refoulement and refugee containment.Footnote 7
Hosting over 1.5 million and 3.5 million Syrian refugees, respectively the highest number of refugees worldwide per capita and in total, Lebanon and Türkiye are key partners in EUrope’s containment of Syrian refugees.Footnote 8 Aid linked to the refugee presence is a crucial source of income for Lebanon.Footnote 9 For Türkiye, too, EUropean support for hosting Syrian refugees has been a key financial and political asset since the 2016 ‘EU–Turkey Statement’.Footnote 10 Such interdependencies suggest significant EUropean influence. Despite EUrope’s insistence that conditions for safe return to Syria were not in place,Footnote 11 however, Lebanese and Turkish authorities were becoming increasingly assertive in coercing such returns even prior to the demise of the Assad regime.Footnote 12 In response, international actors routinely reiterated refoulement as a red line to their Lebanese and Turkish counterparts. Beyond their generic insistence that returns should be safe and voluntary, however, EUropean representatives were often publicly silent in the face of claims that this red line was being crossed.Footnote 13
This may be because in many cases containment interests override humanitarian and protection considerations in EUrope’s external migration policy.Footnote 14 In Lebanon and Türkiye, EUropean engagement prioritizes migration and border management to prevent onward movement to Europe.Footnote 15 I thus assume that many EUropean actors have an incentive to disregard claims on refoulement. If they were to acknowledge that their self-proclaimed red line is being crossed, they would have to act on this to maintain their diplomatic and political credibility.Footnote 16 Such public ‘naming and shaming’ or withholding funding could then jeopardize their relations with crucial containment ‘partners’.Footnote 17
A important question, then, is how these donors, including the EU and its member states, respond to claims that refoulement is taking place. This paper shows that such responses include ignoring and discrediting these claims. I identify and theorize three specific agnopolitical mechanisms through which this happens: ontological agnopolitics, where actors construct unknowability of what precisely constitutes refoulement, for instance claiming that it does not apply to deporting irregularized refugees; epistemological agnopolitics, where actors construct unknowability regarding the empirical identification of refoulement, for example when they discredit refugees’ deportation testimonies as anecdotal and biased; and material agnopolitics, where funding conditionalities and partnerships perpetuate or reproduce unknowing, as when minimal access and monitoring is accepted as an inevitable contextual contingency.
My case study of the agnopolitics in (return) migration diplomacy thereby contributes to three academic literatures. First, it extends our understanding of deportation by demonstrating how the legal complexities or contextual complications in proving refoulement are politically extended and exacerbated.Footnote 18 Second, it develops literature on migration diplomacy by foregrounding that migration diplomacy does not merely entail partnership, coordination, and negotiation, but also evasion, denial, and distancing.Footnote 19 And, third, it enriches the broader field of international studies by extending insights on the epistemic politics of (migration) diplomacy with the concepts of ontological, epistemological, and material agnopolitics.Footnote 20
Theory: migration diplomacy, (non-)refoulement, and agnopolitics
‘They know well. […] But it’s their excuse not to intervene. But it’s not a real excuse, it’s a political excuse. […] It’s impossible for them not to know.’Footnote 21
The concept of migration diplomacy focuses on the nexus between migration and foreign policy.Footnote 22 It is concerned with what Tsourapas has summarized as ‘the use of diplomatic tools, processes and procedures to manage cross-border population mobility, including both the strategic use of migration flows as a means to obtain other aims, and the use of diplomatic methods to achieve goals related to migration’.Footnote 23 The question of return and the related principle of non-refoulement are pivotal contentions in such dynamics.Footnote 24
There are three so-called ‘durable solutions’ for displacement: integration, resettlement, and return. Return is increasingly favoured by host and donor states.Footnote 25 If conditions of safety and voluntariness are not met, however, this ideal solution to displacement risks violating a central principle of international law: non-refoulement. The core of the non-refoulement norm is ‘the idea that a person should not be sent to a country where she may face persecution or a serious human rights violation’.Footnote 26 This norm originates in the 1951 Refugee Convention, but has since expanded to human rights treaties.Footnote 27 As such, non-refoulement is widely recognized as customary international law.Footnote 28 This means that non-refoulement is ‘the most fundamental of all obligations owed to refugees’,Footnote 29 also for countries that have not signed the Refugee Convention, such as Lebanon, or that maintain geographical limitations to its application, like Türkiye.
Legally, different conventions and bodies of law can have different interpretations of the scope and applicability of the principle. For the purpose of this paper, refoulement focuses on direct forced return, not including ‘constructive’ or ‘disguised’ refoulement (where return is coerced through destitution and undue incentivization) or ‘indirect’ or ‘chain’ refoulement (where states transfer people to other states that engage in refoulement).Footnote 30 Beyond the fact that Lebanon and Türkiye are bound to the non-refoulement principle through its customary law status, both countries are also parties to the Convention Against Torture, which explicitly prohibits refoulement.Footnote 31 My understanding of refoulement therefore focuses on the risk of ‘torture or state-sponsored indignity’, rather than general ‘ill treatment’ or ‘deprivation’.Footnote 32
To avoid refoulement, returning states have an obligation to provide returnees with the chance to prove ‘danger of experiencing torture or other pertinent harm’ and to have such claims independently assessed.Footnote 33 Returning states have a ‘duty of inquiry’Footnote 34 and an obligation to ‘provide effective remedies’ not to expel ‘arbitrarily’.Footnote 35 This becomes even more pertinent in cases where the country of return is infamous for its routine use of torture, as was the case with Syria under Assad.Footnote 36 For the case at hand, then, the fine line between the perfect durable solution and the violation of international law hinges on demonstrating that return was either voluntary or that there had been an individual and objective assessment that established that the person forcibly returned was not at risk of torture.
The ultimate arbiters of determining whether a particular case of return constitutes refoulement are legal authorities. My analysis, however, is not a legal one. It focuses on political actors’ contestations of whether something should be considered refoulement. For the Syrian refugees central to this case, monitoring and thereby safeguarding either the voluntary nature of returns or the due process of deportations falls under the mandate of the UN’s High Commissioner for Refugees (UNHCR),Footnote 37 which has access to these populations through its registration, service delivery, and protection programmes. In Lebanon, in the absence of a functioning asylum system, it was the UNHCR that registered (and later ‘recorded’) Syrians as refugees and established a (partial) modus operandi with the government to observe the voluntariness of the return programme run by the General Security Office (GSO) and the due process of deportations.Footnote 38 In Türkiye, the state has largely handled the administration of Syrians under the temporary protection regime and the UNHCR has mostly focused on service provision and capacity-building, but the UNHCR has also observed some protection functions, including monitoring returns.Footnote 39
Beyond legal authority and monitoring mandates, whether or not conditions of voluntariness or due process are met and the extent to which they can be observed or assessed are also political questions and thus subject to migration diplomacy. The interests of host, origin, and donor states determine both the feasibility of monitoring refugee returns and the political will to speak out on or sanction instances of forced return without due process.Footnote 40 The question of knowledge production on contested returns that may constitute refoulement thus becomes central to return migration diplomacy. In trying to understand how relevant actors navigate tensions between norms and interests that centre on return and refoulement, therefore, I draw on another emerging subfield of migration studies, which has introduced the concept of epistemic politics to better understand the governance of displacement.
Originating in the field of critical security studies, epistemic politics refers to the exercise of power through the production of knowledge as well as ignorance.Footnote 41 A wide range of migration scholars have explored this ‘other side of knowledge’, investigating how uncertainty, ambivalence, and ignorance generated by particular practices of (un-)knowing serve political interests, in ways ranging from resolving tensions and maximizing leeway and flexibility to avoiding accountability, outsourcing responsibility, and legitimizing (in)difference and (in)action.Footnote 42
Within the epistemic politics literature, I am specifically inspired by work on ignorance studies and McGoey’sFootnote 43 concept of strategic ignorance, which is concerned with ‘any actions which mobilize, manufacture or exploit unknowns in a wider environment to avoid liability’.Footnote 44 Building on feminist and postcolonial thinkingFootnote 45 and concepts such as ‘structural amnesia’, ‘non-thinking’, and ‘states of denial’, this work approaches ignorance claims not as cognitive or intellectual deficits but as socially constructed and politically productive practices.Footnote 46 The notion of strategic ignorance assumes that ignorance can be a way of exercising power, and so is not merely an object but rather a technique of governance .Footnote 47 In McGoey’s words: ‘elite power functions through strategic ignorance – through the ability to select which voices to acknowledge and which to dismiss’.Footnote 48 Studying such processes of agnopoliticsFootnote 49 helps to understand how claiming ignorance is instrumental in pursuing geopolitical interests and aligning these with normative claims.Footnote 50
My analysis builds on and extends Davies, Isakjee, and Dhesi’s concept of agnopolitics, developed to explore the legitimation of violent inaction involved in migration control.Footnote 51 Davies, Isakjee, and Dhesi present agnopolitics as an adaptation of work on agnotology and define it as ‘an expression of power’ in the form of ‘maintaining ignorance of a situation for political ends’ in a way that allows actors to ‘be mobilized into inaction’ or ‘a calculatedly limited form of involvement’.Footnote 52 Agnopolitics, as such, explicitly encompasses the intertwined claiming and (re-)production of ignoranceFootnote 53 – as also used in work outside the realm of migration studies, for instance in Gilroy’s analysis of the ‘machinations of denialism’, where he sees ‘agno-political power’ as the systemic political denial of ‘troubling information’ to maintain a status quo of ‘epistemological disorientation’.Footnote 54 ‘Ignorance’ here relates to a broader spectrum of not-knowing, where various forms of not knowing enough, not knowing for sure, and knowing but not understanding – such as doubt, uncertainty, opacity, and ambivalence – are strategically presented as ignorance claims to legitimize inaction and silence.Footnote 55
I understand the claiming and the generation of ignorance as inherently interlinked. Since the power to determine what is credible knowledge is unequally distributed, claiming ignorance also (re-)produces, institutionalizes, reifies, and imposes ignorance on other actors. The plausible deniability legitimizing silence and inaction then consists of defensive dimensions (claiming ignorance, i.e. pretending not to know) and offensive dimensions (imposing ignorance, i.e. preventing others from knowing).Footnote 56 I extend this framework on defensive and offensive dimensions of agnopolitics by identifying three specific mechanisms through which agnopolitics operates.
In terms of claiming ignorance (defensive agnopolitics), I distinguish between ontological and epistemological mechanisms. In EUropean ignorance claims there is, on the one hand, an assumption that refoulement is theoretically unknowable. Interlocutors question the definition or applicability of refoulement and the related epistemic mandate to ‘call it’.Footnote 57 I label this ontological agnopolitics, as it pertains to the nature of refoulement. On the other hand, there is a suggestion that refoulement is empirically unknowable. Interlocutors maintain that even if we could agree on what does and does not constitute refoulement there would be no way to test specific cases against such definitions due to contextual impediments. They then discredit the epistemic authority and testimonial competence of actors who do produce partial knowledge in such settings.Footnote 58 I call this epistemological agnopolitics as it regards the nature of knowing about refoulement. In terms of extending, maintaining, or imposing ignorance (offensive agnopolitics), I identify specific material mechanisms relating to the political economy of un/knowing that centre around funding conditionalities and partnerships. Taken together, these three mechanisms enable an agnopolitical reading of migration diplomacy that helps us see how EUropean ignorance is partly professed – since there are particular actors who do know – and partly maintained – since some actors arguably could know.
The ignorance claims central to my analysis often follow from the inherent organizational complexities and the extreme staff turnover in international organizations and are further exacerbated by the convoluted nature of the overlaps between these organizations.Footnote 59 Many interlocutors, moreover, understood ignorance claims as a form of ‘cognitive dissonance’Footnote 60 or the fallout of ‘misinformation’Footnote 61 produced about refugees by authorities. These readings, however, risk obscuring the political convenience of accepting such ‘misinformation’. My agnopolitical analysis thus complements these understandings by highlighting the political functionality of ignorance.
Context and methods: return assemblages in Lebanon and Türkiye
Me: ‘I hear this a lot, right? “Refoulement is a red line.” But then how do we know and who determines when refoulement takes place and what happens if it does?’ Them: ‘Uuuuuh … that is a good question.’Footnote 62
My analysis focuses on EUropean engagement with forced return to Syria before the fall of the Assad regime in two contexts: Lebanon and Türkiye. These contexts differ significantly. Türkiye is a centralized and strong state with an established legal and policy framework for asylum and residency.Footnote 63 Lebanon is a fragmented and hybrid political order and lacks a formal institutional framework for governing the Syrian refugee presence.Footnote 64 Türkiye wields significant geopolitical clout in its relations with EUropean partners, especially since the 2016 EU–Turkey Statement, and is considered a powerful partner able to engage in significant ‘blackmailing’.Footnote 65 Lebanon is regarded as a more dependent refugee rentier state, reliant on ‘back scratching’.Footnote 66 Yet, following Şahin-Mencütek, as containment partners in EUropean migration diplomacy related to Syrian refugees, they can credibly compared.Footnote 67
Initially, both Lebanon and Türkiye welcomed Syrian refugees. But they also insisted that refugees’ stay would be temporary and return was the preferred durable solution. Both countries, however, have taken a ‘return turn’. In Lebanon, competing political actors have reached a de facto consensus that return is possible, necessary, and urgent, culminating in a ‘safe return’ agenda that overruled voluntariness and disregarded realities on the ground in Syria.Footnote 68 In Türkiye, Syrian refugee return has became extremely politicized, with opposition parties increasingly calling for immediate and complete, and if necessary forced, repatriation of refugees.Footnote 69 The government has maintained its formal insistence on voluntary return but simultaneously significantly increased pressure to return.Footnote 70 Thus, rhetorical and physical pressure on refugees to ‘self-deport’ to Syria has been increasing since 2019 in both countries.Footnote 71
Self-deportation has been complemented with forced deportation. In Lebanon, this happened in a context of irregularization that generated deportability.Footnote 72 Entry and residency regulations introduced in 2014 practically barred refugees from maintaining legal residency, and a 2015 government ban on refugee registration by UNHCR undercut access to formal refugee status. Syrian people in Lebanon could be forcibly returned on the grounds of such illegal entry or presence, which is referred to as ‘administrative deportation’, or on the ground of committing criminal offences, so-called ‘judicial deportations’.Footnote 73 Both such deportations were expediated by a 2019 decision of the GSO stipulating that all Syrians who entered Lebanon irregularly after April 2019 could be deported without a trial.Footnote 74 In addition to individual cases, Lebanon has seen concerted campaigns to summarily deport Syrian refugees collectively through checkpoints and raids.Footnote 75 The most infamous of those was a deportation peak in spring 2023. In a period of around three weeks, thousands of people were arrested and deported to Syria by the Lebanese army.Footnote 76
In Türkiye, Syrian refugees are safeguarded from deportation by temporary protection status. Since 2022, however, such status became more limited and conditional.Footnote 77 Here, too, irregularization produced deportability, providing a pretext for arrest and detainment often followed by deportation.Footnote 78 In 2022, authorities started implementing ‘sweeping operations’, resulting in the ‘mass arrests and deportations of Syrians’.Footnote 79 In 2023, ‘mobile migration points’ were introduced to apprehend irregular migrants and detain them in removal centres prior to deportation, a campaign that also targeted Syrian refugees who lacked verifiable address registration, had been found outside the province where they were registered, or were considered a threat to public order or national security, an accusation often made on an arbitrary basis.Footnote 80 Once in such removal centres, sometimes people were reportedly blackmailed into signing, or physically forced to sign, voluntary return forms, which they often could not read.Footnote 81
Because these returns were not systematically monitored by independent observers it is impossible to determine precisely how many people have been forcibly deported.Footnote 82 Based on insights from Syrian border gates, Lighthouse Reports suggests that half of the returns presented as voluntary – estimated to be around 600,000 for the 2017–2023 periodFootnote 83 – may in reality have been forced.Footnote 84 For Lebanon, authorities told Amnesty International that in 2019 and 2020 alone GSO deported 6,865 people under the 2019 decision.Footnote 85 UNHCR approximates that in 2023 the army forcibly returned 13,700 people from Lebanon to Syria in ‘deportation or pushback operations’.Footnote 86
Because of a lack of systematic monitoring, different actors make different claims about the nature and extent of deportations.Footnote 87 In line with the ‘complex and polycentric’ quality of the geopolitics of return,Footnote 88 I discuss these different claims by considering the actors involved as assemblages.Footnote 89 Assemblages here indicate networks of practice that ‘forge connections’ between distinct actors who have variegated interests, understandings, and capabilities but can contextually align on specific issues.Footnote 90 The assemblages presented below are thus not formal or unitary actors with direct agency or intent, but rather ‘fields’ of loosely interacting individuals and groups that are characterized by ‘heterogeneity, relationality, and flux’.Footnote 91
To juxtapose competing knowledge and ignorance claims, my analysis focuses on three such assemblages. First is the ‘protection assemblage’, which captures the variety of refugee-led organizations (RLOs) and local, national, and international NGOs that work with and for Syrian refugees in Lebanon and Türkiye. These include bar associations and the humanitarian, development, and rights organizations that take on a ‘major share’ of refugee governance in Lebanon and Türkiye.Footnote 92 Most of them have a protection mandate and aim to contribute to preventing violations of refugee rights, including refoulement.
Second is the ‘deportation assemblage’, which designates the broad and multiscalar array of relevant Lebanese and Turkish authorities, including politicians and government officials. In Lebanon, while return policy is officially the prerogative of an inter-ministerial committee presided over by the Prime Minister’s office, deportation is largely overseen by a range of security agencies, mainly the GSO and the army. These can act rather autonomously, often extralegally, and frequently do so in tandem with political actors who operate parallel to but as proxies for the official state.Footnote 93 In Türkiye, migration is centrally managed by the Presidency of Migration Management (PMM) under the Ministry of the Interior, which operates through provincial directorates of migration management.Footnote 94 Deportation is enacted by them through the gendarmerie, police, and land guard commands.Footnote 95
Third is the ‘EUropean assemblage’, referring to EUropean diplomats and donors. Here, I follow Stierl’s conceptualization (see footnote 6) to do justice to the assembled nature of ‘many Europes’.Footnote 96 I ‘studied up’ from representativesFootnote 97 associated with EU and member state delegations in Lebanon and Türkiye (embassies, Directorate General (DG) country offices) to Brussels- and capital-based institutions (DG offices, parliaments, foreign ministries). ‘EUrope’ encompasses actors with different mandates and priorities, from humanitarian to development to security. My analysis concerns the dominant epistemic logics that prevailed within this assemblage in relation to forced return. I am aware that for many EUropean stakeholders the issue was not so much lack of knowledge about deportations but lack of options to work against potential refoulement in ways that did not create further backlash for refugees. My objective is neither to promote neocolonial assumptions that ‘normative Europe’ should keep its ‘partners’ in check,Footnote 98 nor to suggest that the dilemmas of migration diplomacy have clear-cut solutions or are solely about knowledge production. Instead, my research unpacks how such dilemmas emerge and persist in a context where agnopolitical migration diplomacy helps sustain containment.
Like every analytical construct, the above assemblages are artificial. The ‘fields’ that each assemblage represents empirically overlap, as with EUropean humanitarian actors who have a protection mandate but operate under EUrope’s broader geopolitical externalization agenda. This overlap is particularly evident in the case of UNHCR, which has a protection mission but is structurally incentivized to accommodate host state and donor interests.
Finding durable solutions to displacement is central to UNHCR’s mandate. Under pressure from donor and host states, UNHCR has increasingly prioritized repatriation as a solution, creating tensions in relation to protection – including protection from refoulement .Footnote 99 In Lebanon and Türkiye, too, UNHCR is dependent on host states, without whose permission it cannot operate.Footnote 100 This has resulted in state pressure to encourage return. Janmyr notes that: ‘Lebanon’s recent policies aiming to decrease the number of Syrians in the country, by reducing access to territory and encouraging return to Syria, have heavily affected UNHCR’s own ability to execute its international protection mandate.’Footnote 101 At the same time, UNHCR is also vulnerable to pressure from its donors, including EUropean countries prioritizing containment. As the majority of ‘activities by UNHCR are funded almost exclusively from voluntary national contributions’, whereby governments are ‘narrowly pursuing their own interests or ideas by extensively earmarking funds’, donors crucially influence the direction of UNHCR operations. Footnote 102
Spanning these assemblages, my analysis is based on over 140 in-depth interviews with Lebanese, Turkish, and international representatives of intergovernmental organizations, humanitarian and development NGOs, and rights organizations; Lebanese politicians, state officials, and civil servants; and experts associated with European actors, based in Lebanon, Türkiye, and Europe. These interviews were conducted between March 2023 and May 2025 in person – during four visits to Beirut and Ankara of several weeks each – and online. Interviews focused on the period before the fall of the Assad regime and, depending on the interlocutor’s preference, were either recorded and transcribed verbatim or documented through extensive note-taking. Interviews were complemented by field notes and informal conversations with practitioners and experts. The majority of transcripts and notes were coded through NVivo.
To explore the strategic nature of ignorance, agnopolitical analysis interrogates something that is presented as a passive state (to be ignorant) as a potentially active move (to ignore). It is through alternatives to or resistance against claims of not-knowing that passive states of ignorance can be read as active modes of ignoring.Footnote 103 My analysis concerns aggregated actors who are hybrid in nature and operate within broader and overlapping assemblages. This means that rather than aspiring to prove direct ‘intent’ – a ‘pyrrhic challenge’Footnote 104 that is often dismissed as ‘conspirational’Footnote 105 – I foreground and trace proclaimed interests rather than explicit intentions, which ‘allows for a more comprehensive understanding of both the implicit and explicit, the agential as well as structural, forms of strategy’.Footnote 106
Below, I therefore present the contending knowledge and ignorance claims on deportation and refoulement of the three assemblages identified. In analysing some of EUrope’s positioning as strategic ignorance, I draw on the emic descriptions of interlocutors from the protection assemblage. The paper highlights and theorizes these interlocutors’ understandings of the agnopolitics they are up against in protecting refugees from refoulement. In asserting what actors in the EUropean assemblage do or could know, I worked on the assumption that if interlocutors from the protection assemblage and/or myself were able to access particular information or knowledge, these much more resourceful actors would, or could, have such access too.
Analysis: contesting potential refoulement to Syria
Non-refoulement is non-refoulement. This is non-derogable. So this is our bread and butter. […] non-refoulement is absolute. In principle. Sounds lovely on a convention, sounds lovely when we talk about this amongst ourselves. In practice, refoulement, it’s… you know, it’s whether or not the government or the authorities choose to apply it.Footnote 107
Protection assemblages and allegations of refoulement
When individuals are forcibly returned to a country that is broadly considered unsafe, most protection experts agree that deportation breaches the principle of non-refoulement if no individual risk assessments have been conducted through a ‘due process’.Footnote 108 This in essence consists of three stages. First, an individual indicates a fear of returning; second, these concerns are objectively considered; and, third, a deportation decision is made that is conditional on these considerations. In both Lebanon and Türkiye, interlocutors explained, this process is only partially in place and is hardly ever systematically observed.
Working closely with RLOs and local and national rights organizations, international organizations, investigative journalists, and academic researchers have claimed for years that Türkiye engaged in refoulement of Syrian refugees because many returns were not voluntary and due process was not observed.Footnote 109 Journalists and rights organizations have documented the use of physical coercion to sign voluntary return forms in removal centres where people were regularly barred from access to legal help.Footnote 110 Nevertheless, the individual risks were in most cases not assessed because the claim was that these returns were voluntary and people were returned to ‘safe zones’ in Syria that were under Turkish control.Footnote 111
In Lebanon, due process was undermined by the limited capacity of the Lebanese judiciary.Footnote 112 Since 2019, therefore, UNHCR has engaged with the GSO to investigate whether refugees to be deported are at risk. According to interlocutors, however, UNHCR was not informed of all cases and did not have sufficient access and time to properly assess cases. Some civil society protection experts reported that UNHCR appeals to the general prosecutor to halt deportations in cases that they had assessed as potential refoulement, were often disregarded – either by the prosecutor or, if the prosecutor did order deportations to be halted, by the GSO, which proceeded with the deportation anyway. This resulted in documented individual cases whereby Syrian people who indicated a fear of returning had been deported to Syria without due process or intervention by the judiciary or UNHCR.Footnote 113 Forced deportation to Syria without due process also occurred through collective summary deportations such as those that took place in spring 2023, when people were sometimes allegedly handed over directly to the Syrian army.Footnote 114 Some of them were reported to have been arrested, tortured, or forcibly di sappeared once back in Syria.Footnote 115
For the protection assemblage, then, refoulement was a recurring reality.Footnote 116 A human rights expert from an international organization concluded: ‘Non-refoulement was meant to be, like, the red line, right … And it stopped being a red line a long time ago. So, yeah, the thing was done’.Footnote 117
The deportation assemblage and denials of refoulement
Lebanese and Turkish authorities unanimously refuted contentions of refoulement. They paid routine lip service to the principle of non-refoulement publicly and in their engagement with donors and rights organizations.Footnote 118
Even in the face of findings by the protection assemblage, Türkiye maintained that all returns to Syria were safe, voluntary, and dignified, denying any claims to the contrary.Footnote 119 Turkish authorities stressed that ‘Turkey neither has a right, capability, a system, [or] a practice’ of deporting people to Syria.Footnote 120 This means that claiming voluntariness for all returns has become ‘essential’ for Türkiye and has been institutionalized through voluntary return forms.Footnote 121 To ‘increase the government’s credibility’, Turkish authorities have reportedly asked Syrian border gates to register any return as voluntary.Footnote 122
Lebanese authorities similarly systematically highlight voluntary returns. They showcase GSO’s return programme as proof of Lebanon’s commitment to non-refoulement.Footnote 123 An international protection expert explained how a ministerial representative had vehemently told them: ‘we do not do refoulement; look, there are 1.5 million people here, we’re not forcing them to return, we have facilitated return’.Footnote 124 When pressed about forced returns, Lebanese authorities’ often deny deportations take place. They also apparently uphold the impression of voluntariness through forcing people to sign voluntary return forms.Footnote 125 When Lebanese authorities acknowledge that forced returns take place, they insist on the legality of these deportations under the 2019 decision, which they consider to have legalized any forced return of the designated category of Syrian people in Lebanon.
Actors in Lebanese and Turkish deportation assemblages thus largely deny forced returns and contend that even deportations without due process need not be considered cases of refoulement because due process obligations can be overruled by security considerations, which would give security services the discretionary power to sideline the judiciary.Footnote 126
The European assemblage and ignorance on refoulement
Officially, the EU would not support (politically or financially) any organized return to Syria as long as UNHCR’s 2018 thresholds for return are not met.Footnote 127 It routinely reiterated refoulement as a ‘red line’ – both publicly, as in for instance the chair declarations of the annual Brussels Conferences for the Future of Syria and the Region,Footnote 128 and in diplomatic engagement. Indeed, interlocutors indicated that this ‘red line’ was so undisputed for the EU that member state delegations would not even have to consult their headquarters on sticking to it.Footnote 129 At the same time, EUrope did not specify explicit sanctions for violating non-refoulement.Footnote 130
In this context, on the one hand, behind the scenes there was concern with deportations. With regard to Lebanon, interlocutors indicated that, in private conversations, donor representatives and UNHCR acknowledged that some deportations probably constituted refoulement. One knowledgeable interlocutor told me: ‘If it walks like a duck, looks like a duck, and quacks like a duck …’Footnote 131 Regarding Türkiye, too, refoulement claims by the protection assemblage were widely discussed within the EUropean assemblage.Footnote 132 Interlocutors told me that in such discussions, the ‘voluntary’ nature of returns was always ‘between quotation marks’.Footnote 133 At the same time, on the other hand, diplomats and donors publicly avoided use of the term refoulement. A EUropean expert said: ‘For me, this is refoulement. I wasn’t always allowed to say it like that … because it’s the red line, internationally speaking. You have to be really careful with the terminology you are using.’Footnote 134
Where public engagement with the question of deportations taking place without due process does take place, donor representatives, including EU officials, often emulated the framing of returns used by Lebanese and Turkish authorities, reflecting the disproven but institutionalized assumption of ‘states’ good faith at borders.’Footnote 135 In Türkiye, when asked about deportation and refoulement, interlocutors often steered the conversation away from Syria and, in line with the government narrative that distinguishes between Syrian ‘refugees’ and other ‘migrants’, stressed that such concerns were more pressing with regard to Afghans. In Lebanon, like the Lebanese authorities, EUropean actors seemed disproportionately interested in the GSO’s facilitated return programme, which served as a distraction from more problematic return practices.Footnote 136
EUrope also follows government narratives when it comes to matters of scale and systematicity. EUropean diplomats in Lebanon considered that there were ‘only a handful’ of such deportations and that these numbers were ‘negligible’,Footnote 137 or ‘not significant’Footnote 138 enough to raise concerns of refoulement. One of them reflected: ‘It was acceptable because it was in limited numbers’.Footnote 139 EUropean diplomats in Ankara similarly worried about coerced return, but insisted on its small scale. One interlocutor seemed to voice a broader sentiment when they stated, ‘I don’t want to be cynical, but there were four million refugees here and then if there are two hundred, four hundred cases … who cares?’Footnote 140 Another EUropean interlocutor found that we cannot ‘afford the illusion’ that ‘one case of refoulement would mean crossing a red line’.Footnote 141 This is telling because, as a rights defender noted, individual cases never seem to ‘accumulate’.Footnote 142 It would also explain why in Lebanon the 2023 deportation peak did lead to diplomatic intervention. The response to the very public and collective deportations carried out by the army highlights the absence of such a response to the individual deportations carried out by the GSO, which though greater in number are less visible and continued both concurrently and afterwards, apperently without much comment.Footnote 143
For refoulement to be politically significant for the EUropean assemblage, it apparently needs to be structural. Protection actors went to great lengths to establish ‘clarity about a trend of incidents, because a trend shows it’s policy’.Footnote 144 They indicated that in Türkiye, coerced return ‘became more systematized’ over time and by 2023 the tenability of the ‘small scale rhetoric’ had become ‘questionable’.Footnote 145 Actors from the EUropean assemblage, however, regularly referred to cases of refoulement as ‘incidents’ and ‘operational aberrations’ and often maintained they could not be regarded as ‘policy’ if there was no ‘strategy’ behind them.Footnote 146 As a EUropean refugee expert, who was based in Beirut, put it: ‘We really believe that it’s not structural policy; if we got the feeling that it is, that would change matters.’Footnote 147 A Western diplomat in Ankara told me: ‘I mean, I don’t think anyone would deny refoulement has happened. […] But then the question is: is it systemic? And my answer is: no, it has happened, but it’s not a policy.’Footnote 148 In the end, it seems EUrope’s red line was not so much refoulement, but systematic refoulement.
When faced with the protection assemblage’s allegations of refoulement and the deportation assemblage’s denials, the EUropean assemblage thus responded by acknowledging concerns on forced deportation but simultaneously steering these concerns away from the notion of refoulement and emulating state authorities’ narratives that highlighted voluntary return and the small-scale and exceptional nature of deportations. Reflecting broader silence on rights violations by Lebanese and Turkish partners,Footnote 149 donors and diplomats are ‘turning a blind eye on forced returns’.Footnote 150 As a human rights lawyer summarized it: ‘there is no appetite to … to push on this’.Footnote 151 This was confirmed by a Europe-based refugee policy specialist, who acknowledged that there is a tendency to prioritize ‘convenience’, explaining: ‘if they don’t raise it too much there [in regional host countries], maybe we needn’t talk about it too much here’.Footnote 152 A Turkish refugee expert described EUropean non-engagement with rights violations in similar terms, noting ‘whatever happens in Turkey stays in Turkey’.Footnote 153 An international human rights expert summarized their experience with EUropean counterparts working on Türkiye as follows: ‘There is a, kind of, wall of silence, essentially, around this problem. Or, if not a wall of silence, then a willingness to minimize the problem or to say “well, yes, there are cases” and not to understand that it’s sort of quite a systematic problem.’Footnote 154
Crucially, actors in the EUropean assemblage frequently legitimized this passivity by claiming ignorance about return, suggesting they did not know enough about forced returns to speak out or intervene. This is remarkable, because diplomats saw gathering information and informing their capital as the core of their job, with embassies functioning as ‘information brokers’.Footnote 155 Yet they routinely claimed ignorance. Diplomats’ assumptions that numbers of deportations – and hence potential refoulement – were minimal and incidental often paradoxically went hand in hand with the admission that they did not and could not know the scale or systematicity of deportations. A diplomat cited by Ingleby et al. similarly said ‘it is hard to know’ about abuses, presenting precisely this ignorance as a reason for ‘not being that confrontational’ about the matter.Footnote 156
In my own conversations, interlocutors associated with the EUropean assemblage similarly reported that they ‘don’t have data, […] don’t know’,Footnote 157 ‘miss a lot’,Footnote 158 simply ‘can’t verify’Footnote 159 or ‘honestly, have little insights into’Footnote 160 what is going on regarding deportations – that they were ‘struggling’ to know.Footnote 161 Importantly, such references to ignorance were often followed by deference to state authorities’ claims: EUropean narratives seemed to fall in line with government narratives not despite the claims that the interlocutors did not know, but because of these claims. They motivated silence and inaction through supposed not-knowing. To further unpack how EUropean actors deflect the knowledge claims on refoulement presented by the protection assemblage, applying the theoretical framework introduced above, the next section analyses EUropean ignorance claims as forms of ontological, epistemological, and material agnopolitics.
Discussion: ontological, epistemological, and material agnopolitics
Unvoluntary returns are also covered under voluntary returns. You know it, I know it, all of us know it. […] I don’t think that they don’t know it. You know, there’s a kind of a blind eye quality to that. […] Nothing is hidden or covered from the European Union. […] There is a lot of data if you wish to have it.Footnote 162
In the above, I established why EUropean donors and diplomats may have had an interest in falling in line with Turkish and Lebanese denial of refoulement: this allowed them to uphold their normative commitment to international refugee law without compromising the partnerships that allowed them to pursue geopolitical containment interests. For a diplomat, this was a matter of ‘sweeping the dust under the carpet’, of ‘not wanting to collapse the tower of cards’.Footnote 163 To be able to do this, EUropean actors had to disregard knowledge presented by the protection assemblage.
Actors from all three assemblages, the previous section showed, acknowledged (at least off the record) that instances of forced return to Syria occured and that these often did not meet the due process standards for guaranteeing the safety of returnees. For actors from the protection assemblage, this constituted refoulement. Actors from the deportation assemblage denied this. Actors from the EUropean assemblage formally maintained they could not know for sure whether this was the case, claiming ignorance of the nature of refoulement as well as the nature of their own red line. In the below, I pursue an agnotological analysis to dissect how such unknowing works.
Ontological agnopolitics: from ‘red line’ to ‘grey area’
Ontological strategies of agnopolitics relate to definitions and applicability and the related power to define. In terms of definitions, interlocutors noted that actors from the EUropean assemblage often suggested that there was no universal definition of refoulement, calling it a ‘grey area’.Footnote 164 This was related to doubts about the universal applicability of the non-refoulement principle. While, as my theory section acknowledged, there are indeed varying legal interpretations of the scope of the non-refoulement principle, the ignorance claims maintained here were routinely dismissed by experts of the protection assemblage, also in their engagement with EUropean partners. There was, for instance, the regular assertion that protection from refoulement is only beholden to refugees, meaning people with formal refugee status (provided by either a state or UNHCR), and that when such status is absent or partial or people are otherwise irregularized, non-refoulement principles do not apply. This is incorrect, as diplomats were regularly told by protection experts in various briefings.
Another discussion diplomats engaged in when asked about refoulement is that on safety. In line with the position of Lebanese authorities, they sometimes suggested that a deportation is only refoulement if and when the people who are deported are actually ‘mistreated’ (arrested, detained, killed, or disappeared).Footnote 165 But refoulement constitutes the risk of harm not the actual infliction of harm and, as elaborated on in the theory section, hinges on due process guarantees to assess such risks (even in the case of forced returns to so-called ‘safe countries’ or ‘safe zones’). Interlocutors from the protection assemblage also noted this ‘lack of kind of, I think, legal knowledge and understanding of the principle of refoulement’ among their EUropean counterparts and found it conspicuous in light of the many instances where legal experts have briefed donors on refoulement concerns and their applicability in relation to Syrian refugees.Footnote 166
In addition to generating doubt about the scope and applicability of the non-refoulement principle, a second dimension of the ontological agnopolitics of refoulement regards the contention of who decides when deportation is or is not refoulement. Mostly, EUrope appears to follow the lead of UNHCR in assessing whether deportations may constitute refoulement. A humanitarian expert working for a major donor told me that when they raised concerns on refoulement with their manager, the reply was: ‘do we have it on paper from the UN that there have been refoulements?’ Considering UNHCR’s mandate, this seems logical. But, considering the complicated position of UNHCR outlined above, when EUrope defers to UNHCR’s assessment in terms of the voluntariness and safety of a return – and thus the assessment of potential refoulement – this can be considered a form of epistemic instrumentalization of UNHCR’s limited capacity, independence, and access. Interlocutors familiar with UNHCR explained that often UNHCR cannot call a deportation refoulement even if it considers it to be so, and says so in private to partners. A representative of an international human rights organization summarized it as follows: ‘they know exactly what’s going on, […] but they’re very mute on everything’.Footnote 167
This is because UNHCR sometimes faces pressure by Lebanese and Turkish authorities.Footnote 168 But also, interlocutors explained, because it has to consider the sensitivities of its donors – donors represented by the very interlocutors who say they follow the lead of UNHCR.Footnote 169 Thus, UNHCR, in line with its general ‘depoliticization’ and ‘technocratic distancing’,Footnote 170 likely faces similar incentives as its donors to look away.Footnote 171 An independent Turkish consultant concluded that this resulted in a situation where ‘when I talk with UNHCR, I feel that I’m talking with someone from PMM’.Footnote 172 Donors are well aware of this ‘delicate’Footnote 173 situation – it is ‘almost just accepted that this kind of thing is happening’Footnote 174 – but nevertheless continuously reiterate that UNHCR is the sole authority able to proclaim refoulement.
Actors associated with the EUropean assemblage thus created ambiguity about the applicability of refoulement, regularly relativizing and complicating it, and often outsourced its determination to actors politically disincentivized from declaring it. As forms of defensive ignorance that enable claims of not-knowing, such ontological agnopolitics allowed some EUropean actors to maintain that their regularly declared red line of refoulement was, upon further scrutiny, hard if not impossible to draw.
Epistemological agnopolitics: from ‘red line’ to ‘black box’
If ontological agnopolitics regards the proclaimed impossibility of even drawing a red line, epistemological agnopolitics concerns a proclaimed impossibility of knowing when such a red line is crossed. EUropean donors and diplomats often legitimized their silence and inaction by suggesting that in the case of deportations from Türkiye and Lebanon to Syria, there was simply no way to produce the kind of information necessary to establish refoulement. The contention here was that since structural access to the border and to the people to be deported was absent or elusive, one could not know about deportations taking place and the extent to which due process was or was not observed or coercion applied. Actors who claimed otherwise were in many instances considered unreliable by these interviewees.
In upholding defensive ignorance to legitimize inaction, EUropean interlocutors stressed that ‘we don’t see it, because we are not there’ and wondered ‘how to prove that it happens until you’re there yourself?’Footnote 175 This, however, ignores the possibility that others might know. In many instances, interlocutors from the EUropean assemblage indicated that it was not just that they did not know, but that they could not know, that no one could credibly know about refoulement in this ‘black box’.Footnote 176 In the words of one diplomat: ‘We don’t have the full picture. But you almost can’t have the full picture here.’Footnote 177 This, however, was only partially the case. In the situation that prevailed until the fall of the Assad regime in Syria, UNHCR, the organization that EUrope seemed to defer to in order to determine potential refoulement, did indeed only have limited knowledge on deportations. Despite its mandate,Footnote 178 UNHCR did not have structural access to the border or to places where people to be deported were held and could thus not observe all returns in either Lebanon or Türkiye.Footnote 179
EUropean actors therefore complemented the information they received from UNHCR with knowledge from the protection assemblage, which they obtained via the humanitarian and development NGOs that they funded and the various rights organizations that advocated with them. Indeed, donors proudly referred to the ‘good networks’ they hade with NGOs working with refugees, explaining that they received both solicited and unsolicited assessments, reports, and briefings from them. In line with this, interlocutors from the protection assemblage indicated that EUropean actors were well aware of forced deportations lacking due process as they constantly informed them about this publicly as well as confidentially. This happened through regular coordination bodies and international fora as well as targeted advocacy by rights groups and confidential briefings by humanitarian partners. Rights organizations funded by EUrope explained to me that their EUropean donors and partners ‘know where to find us’, concluding that ‘they know very well what is going on’.Footnote 180 A former European Commission official stated such issues were ‘constantly talked about internally’ and were ‘common concerns’, to the extent that ‘it’s not really credible that EU officials are not aware of what’s happening in the Turkish migration system’. They added: ‘They know. Everybody knows. People are closing their eyes.’Footnote 181
While they were actively approached for their knowledge, however, NGOs and international human rights watchdogs simultaneously had their ability to know questioned and were regularly discarded as ‘activist’,Footnote 182 ‘alarmist’,Footnote 183 and ‘having their own interests’.Footnote 184 Despite depending on them for their assessments and understandings, EUropean actors routinely questioned the validity of the information provided by actors in the protection assemblage, both in their conversations with me and in interactions with these protection actors that were conveyed to me. These actors’ knowledge was construed as ‘anecdotal’Footnote 185 – ‘even less than case studies’, according to one senior diplomatFootnote 186 – and thereby as in need of verification or discardable. A representative of an international human rights organization recalled:
We’ve tried endlessly to engage with the EU and with EU member state governments. […] And, you know, what we find is that… […] when we present these kind of factual findings, the answer is ‘well, we can’t verify your claims’. […] So this is the answer we typically get on all of this stuff about pushbacks, summary deportations, returns, that kind of thing. That’s the standard argument: ‘we can’t verify’. Which is very surprising to us because that’s not something they say in relation to other human rights abuses that we’ve reported. […] And here they seem to be even sort of challenging the veracity with this kind of ‘we can’t verify’. […] It’s in bad faith, essentially.Footnote 187
International NGOs and rights organizations usually relied on local RLOs for their own knowledge production. But it appears as if it was exactly the close ties to refugee communities that guaranteed RLOs’ access to the type of data that other organizations subsequently requested, which were then used to downplay their validity or significance. This resonates closely with Aradau and Perret’s documentation of the epistemic politics that place migrant testimonies ‘a priori under suspicion’ and Tazzioli’s account of migrants’ knowledge being ‘constantly discredited as untrue or misleading’.Footnote 188 This allowed EUropean actors to claim an indefinite lack of ‘accurate and objective’ knowledge in the face of evident and extensive refugee knowledge, the presentation of which ‘turns out to be pointless’.Footnote 189 Donors maintained that they were ‘concerned and closely monitoring the situation’, but required more information.Footnote 190 One diplomat mused ‘I’d love to see a report about the numbers of deportations that allegedly occur in Lebanon’, disregarding the regular publishing of reports of exactly this kind by RLOs.Footnote 191 Similarly, another international donor repeated multiple times that there were just no organizations that could provide information about the context of deportation, thereby dismissing regular RLO briefings on this context.Footnote 192
Material agnopolitics: funding return monitoring
Through forms of ontological and epistemological agnopolitics EUropean actors could deny or discredit much of the knowledge on forced return to Syria that the protection assemblage claimed. Turning our attention from the realm of defensive ignorance – claiming one’s own lack of knowledge or understanding – to that of offensive ignorance – preventing others from producing or disseminating knowledge – it is obvious that there was much information about the form and degree of potential refoulement that actors from the protection assemblage did not have either. Rather than being a contextual given beyond the influence of EUropean donors and diplomats, however, interlocutors noted that this situation might be improved through making EUropean funding conditional on access and by extending EUropean funding to those organizations that did produce the minimal knowledge on coerced return that was available. They saw the non-occurrence of such significant conditionality and RLO funding partly as indications that ignorance on refoulement was not merely partly professed but also to some extent conveniently maintained. Upholding the material dimensions of agnopolitics that help legitimize EUropean inaction in the face of potential refoulement was, for them, a matter of ‘political will’ as much as complexity and contingency.Footnote 193
As outlined earlier, EUrope was the main donor for Lebanese and Turkish hosting of Syrian refugees, indicating a significant power relation.Footnote 194 Such funding was made nominally conditional on respecting the non-refoulement principle. But it was not made conditional on access to the sites and populations that would allow the monitoring of the actual upholding (or violation) of this principle, even though this was widely and repeatedly advocated for.Footnote 195 EUropean actors tended to convey such limited access and monitoring as a fact of life, claiming that their counterparts in other states would simply not accept conditions of this kind. Even the suggestion that the rare monitoring visits to removal centres in Türkiye by donors should be unannounced was disregarded as a politically unthinkable violation of Turkish sovereignty, leading a journalist to conclude that ‘they’re not trying to see it’.Footnote 196 A EUropean diplomat seemed to second this when they concluded that EUrope was ‘apparently fine-ish’ with a lack of Turkish transparency.Footnote 197 In the Lebanese context, this deflection was acknowledged by senior diplomats, one of whom observed: ‘As much as it is complicated politically in Syria, [it] should not be complicated in Lebanon’.Footnote 198 They concluded: ‘we are in a very complicit silence, you know, when it’s much easier to turn a blind eye, instead of having the right positions’.
Some protection experts saw such dynamics as acquiescence with limited monitoring. They noted that if donors can incentivize their Lebanese border control partners to monitor (and prevent) departures to Cyprus at the sea border, they should be able to demand them to allow monitoring at the land border with Syria. They pointed out that EUropean actors have direct political and financial ties to the security agencies that obstruct the access needed to do better monitoring.Footnote 199 In light of this, they argued, making aid conditional on restoring UNHCR access to the border should not be taboo. Yet, in the words of a refugee protection expert, there seemed to be widespread resignation about the observation that ‘the mandate holder is not upholding their mandate’.Footnote 200
There have been instances when EUrope has made use of its leverage to stop deportation practices, for instance when Western donors in Lebanon apparently used financial and political means to pressure Lebanese authorities to stop deportations in spring 2023. Analysts have noted, however, that such interventions may stem from concerns regarding liability or complicity in violations rather than regarding the violations as such, and occurred predominantly in the case of particularly blatant deportations by agencies receiving direct European funding.Footnote 201 In Türkiye, too, EUropean responses appeared to focus more on distancing EUrope from violations than on denouncing or addressing them.Footnote 202
The material dimensions of agnopolitics go beyond the conditionality imposed on funding and extend to the partnerships for such funding. As we have seen, it was organizations in the protection assemblage, specifically RLOs, that produced much of the limited knowledge about refoulement that was available under the constraints outlined above. This in turn constituted a significant part of the knowledge base for UNHCR and EUropean donors and diplomats. Analysts from the Turkish protection assemblage said they had therefore suggested to partners from the EUropean assemblage that more funding for monitoring was the only logical response to ‘anecdotal findings’, reasoning that ‘even with the little information we have, we can identify this trend’.Footnote 203 Organizations in the Lebanese protection assemblage similarly petitioned donors to prioritize return monitoring, calling it an ‘evergreen or a recurring theme in our recommendations’.Footnote 204 The international monitoring mechanism they plead for would streamline and upscale the ad hoc and fragmented monitoring already happening under UNHCR oversight.Footnote 205
Following this, there were instances where EUrope funded RLOs to complement UNHCR’s knowledge production on return. However, structural and extensive funding for RLO-led monitoring of coerced return does not seem to have emerged. This, various protection experts claimed, was a major reason why knowledge on deportation remained fragmented and partial. RLO representatives from Türkiye reported that they ‘kind of got too much info; we don’t have the capacity to take the testimonies in a proper way’, adding that ‘some groups have a lot of testimonies and information they could be sharing, but it may be a security thing’.Footnote 206 This reflects the significance of what Canzutti and Aradau have called skewed ‘data work’, which strategically imposes the ‘burden of proof’ precisely on those actors whose epistemic authority and capacity is systematically undermined.Footnote 207
Thus, what interlocutors from the protection assemblage pointed out repeatedly is that deportation monitoring was not merely already partly happening but could be far more extensive – if never complete or infallible – if funding were to focus on different types of organizations (those willing and able to monitor rather than those formally mandated to) and different activities (not merely general protection monitoring, but documentation of deportation).Footnote 208 Thus, they claimed, the funding, mandating, and programming logics of the EUropean donors in relation to monitoring returns sometimes obstructed the production of the very knowledge that EUrope indicated was necessary to establish whether refoulement happens. Material agnopolitics ensured the continued ‘fragmented and dispersed knowledges’ that were then presented as a legitimation of silence and inaction.Footnote 209 In the words of an international protection expert: ‘It’s very difficult to monitor red lines and to know what has actually happened, or to know the certain thresholds if they don’t have, like, a system to actually monitor these kinds of actions.’Footnote 210
Reflections and contributions
‘[The issue of refugee return] encapsulates everything from how willing they [EUropean donors] even are to look at what’s happening in Lebanon in detail, because if they looked at it, then they might have to act on it.’Footnote 211
It was indeed hard for EUrope to know about deportations from Türkiye and Lebanon to Syria under Assad and therefore whether or not they constituted refoulement. There was, however, a lot that was known about this that was made ignorable through mechanisms of ontological agnopolitics (questioning the scope and undercutting the applicability of the non-refoulement principle and the mandate to determine it), epistemological agnopolitics (undermining the epistemic authority of the core actors actually producing knowledge on deportations), and material agnopolitics (upholding funding logics that concede to a status quo of limited access and monitoring). Through these mechanisms, actors in the EUropean assemblage could maintain both their claimed commitment to non-refoulement and international law and their containment partnerships with actors potentially engaging in refoulement.
These insights extend work on strategic ignorance – in migration studies and beyond – by operationalizing defensive and offensive agnopolitics through the mechanisms of ontological, epistemological, and material agnopolitics.Footnote 212 Applying this to the phenomenon of migration diplomacy enriches the field of international studies by introducing agnopolitics as a new form of power relevant to the negotiations and contestations central to transnational governance: agnopolitics adds a relevant new layer to the literature on (geo)politics. My analysis has elevated discussions on epistemic politics from intrastate to interstate scales:Footnote 213 ‘strategic ambiguity’ is not just leveraged towards people on the move, but fundamentally informs the ‘relational architectures’ that constitute containment partnerships as well;Footnote 214 it is not merely a ‘constitutive political technology of refugee governance,’ but also of migration diplomacy.Footnote 215
This sheds new light on the logics of containment and the outsourcing of protection that characterize EUrope’s external migration politics.Footnote 216 In the face of host country shifts from temporary hosting to partially coerced return, EUropean migration diplomacy ultimately might externalize deportation.Footnote 217 In so far as they prioritize containment over ‘burden-sharing’,Footnote 218 EUropean actors may replace direct (and highly visible and criticized) engagement in pushbacks and non-entrée on their own borders with indirect (and much less scrutinized or deplored) engagement with deportations potentially constituting refoulement through their ‘partners’.Footnote 219
In this way, an agnopolitical analysis can help rethink potential EUropean facilitation of the ‘fragmentation and informality’ that regional host states draw on to ‘blur responsibility over safe and dignified return’.Footnote 220 It reveals how EUrope may extend, benefit from, and appropriate such impunity. It is this ‘diffuse migration power’Footnote 221 deployed in processes of refugee return diplomacy that can help account for the broader eroding of the principle of non-refoulement that scholars have observed.Footnote 222 An agnopolitical perspective on migration diplomacy helps us explore how EUrope might ‘offload’ norm violation onto regional host statesFootnote 223 and thereby maintain its often ‘schizophrenic attitude towards international refugee law’.Footnote 224
This also offers new insights in the international politics of ‘red lines’. Traditionally, red lines are a diplomatic tool to constrain the actions of adversaries.Footnote 225 If directed at diplomatic partners, however, my analysis shows that proclaimed ‘red lines’ may operate as discursive constructs for domestic audiences more than diplomatic leverage vis-à-vis international partners.Footnote 226 In such contexts, the malleability of ‘red lines’ becomes central to their political utility. I have shown how ‘red lines’ can be ‘unknown’ through theoretical ambivalence and empirical doubt so that they can be proclaimed but never be crossed. This resonates with other work that shows that, contrary to dominant theory, the functionality of ‘red lines’ in international diplomacy might follow from ambiguity.Footnote 227 Perhaps red lines do not ‘fail’ if they do not deter, but instead serve other diplomatic ends.Footnote 228
This illustrates how agnopolitics brings a new theoretical perspective to the emerging literature that studies the elusiveness of border violence, including refoulement, as not merely a matter of capacity, but also of political will.Footnote 229 As critical interlocutors pointed out, EUrope funds and promotes the ‘smart’ border control that can monitor and prevent entry, but appears to allow the ‘dumb’ border that fails to monitor and prevent forced exit.Footnote 230 Its migration diplomacy makes legible one dimension of borders but reifies the illegibility and inaccessibility of other dimensions.Footnote 231 It is in interrogating these contradictions that the agnopolitics of migration diplomacy suggests new ways of thinking about political responsibility and accountability in outsourcing migration containment and the related deportation regimes.Footnote 232 Deconstructing ignorance claims and the sometimes active politics involved in maintaining frames of inevitable, inescapable ignorance opens up space for epistemic resistance and mobilization against violent inaction in the face of refoulement.Footnote 233
Acknowledgements
I am grateful to all interlocutors for so graciously sharing their time and insights. I am also thankful to Maja Janmyr, Annika Lindberg, Lisa Marie Borrelli, and Nadine Kheshen, who have provided very helpful input, and Melle van der Woude, who provided assistance with document collection. Several anonymous reviewers have also offered important and insightful reflections. I also extend my thanks to the agencies that have funded the research projects underlying this paper: the Dutch Research Council (project: EU-Actors and the Contested Return of Syrian Refugees from Lebanon and Turkey), the European Commission (Horizon Europe project: GAPs – Decentering the Study of Migrant Returns and Return Policies), and the Swiss Council for International Studies (project: Tracing Syrian Refugee Return Across South/North Divides).
Open access
