Dismantling the anti-politics machine in aid: political mētis and its limits

ABSTRACT

In a recent article in NPE, Rajesh Venugopal (2022. Can the anti-politics machine be dismantled? New political economy, 1–15) concluded that the anti-politics machine was still in operation. He argued that development planners held a cognitive divide between the realm of political dynamics – an unknowable terra incognita – and the realm of operational technical knowledge. This article revisits and expands that argument. It takes a particular kind of adaptive project as an analytical entry point, arguing that their focus on political practice reimagines the ontology of politics as a kind of expert mētis which is situated, relational and emergent. Such projects hold out hope that the anti-politics machine can be dismantled by displacing the cognitive frames of the development planner from the centre stage and emphasising political practice during implementation. However, shifting attention to implementation reveals other elements of the anti-politics machine’s operation. Drawing on interviews with policy advocates, the article shows that the anti-politics machine does not simply work through the cognition of the planner: it also acts through bureaucratic resistance to political practice during project implementation, produced through operational, accountability and financing processes that shackle practice, particularly in spaces with heterogeneous interests and values.

KEYWORDS:

• Development
• technocracy
• mētis
• politics
• implementation

Can the anti-politics machine be dismantled?

In a recent article, Rajesh Venugopal asked ‘can the anti-politics machine be dismantled (Venugopal 2022)?’ His answer was no: for the development planner, political and social dynamics constitute a domain of uncertainty and risk which is cognitively unknowable. Analytical tools which do make politics legible simply bring a portion of that political realm into the planners’ technical realm of knowledge, redrawing the boundary of what is known or cognitively graspable. What once fell in the political realm is stripped of its political quality through the automatic operation of a combination of bureaucratic, epistemic and techno-ethical processes. The divide is cognitive in nature and produced by these deeply embedded institutional processes. Efforts to make development political have only made schemes to politicise aid technical.

Venugopal’s question and response are timely: James Ferguson’s argument that development is an anti-politics machine is now over thirty years old (Ferguson 1996 [1990]). Since its publication, political economy analyses have become mainstream tools and governance programmes have sought to become more political and more contextually responsive.

This article revisits Venugopal’s argument. It does so by focusing on a particular group of adaptive management projects that set political practices as central to politically aware aid, emphasising implementation over ex ante planning. The article does not deny the depoliticising processes identified by Venugopal in relation to planning, but rather explores the operation of these depoliticising processes in relation to politically-aware implementation. In 2017, I was part of an evaluation team for one such project funded by what was then the Department for International Development (DFID),¹ the Zambia Accountability Programme (ZAP: see Lodge and Paxton 2017, DFID 2018) – and one component in particular, the inclusive growth component. The project team consisted of a small team of expatriate lobbying experts who partnered with and supported two well-regarded Zambian think-tanks and a range of civil society organisations in Lusaka (churches, business associations and unions). Together, they sought to influence economic policy and agricultural policy reform. The expatriate team helped refine their partners lobbying and communications skills. There can be little doubt that this project component was explicitly political in its practices and purposes: it worked to mobilise and support an advocacy coalition to call for policy change with significant structural, governance and social implications in relation to control of grain reserves and Zambia’s economic and fiscal policy. To be clear, I do not suggest that this project did not also require expertise (it demanded considerable public relations know-how and project management skills) and I do not make any remark in relation to its legitimacy. The point is rather that ZAP was self-consciously political and seems to hold out the possibility that the anti-politics machine can be dismantled. Nor is it alone: as Venugopal acknowledges, it was part of a wider set of ‘adaptive’ programming adopting ‘politically smart, locally led development’ practices (Booth and Unsworth 2014, Guthiel 2019, Wild 2021).

Adaptive, political practice projects like ZAP (see e.g. Faustino and Fabella 2011, Booth and Unsworth 2014, Faustino and Booth 2014, Pett 2020) prompt a theoretical reframing of politics/technics which foregrounds the role of practice during project implementation, rather than just the cognitive state of the planner and their institutional environment. Following James C. Scott (1998, p. 309), I argue that ‘political knowledge’ of this kind falls within the cognitive horizons of the expert as a matter of mētis (situated, practical know-how) rather than techne, universal technical knowledge (see also Kumar 2021, ‘improvisation’; Mowles et al. 2008). Political mētis signifies the meetings over coffee or dinner, the formulation of arguments, the positioning and lobbying, the attendance of ceremonies and relation-building through which politics is prosecuted. Programmes like ZAP prompt an ontological reimagination of politics as an object of expert practice.

The article suggests that projects like ZAP constitute a ‘third generation’ of project which legitimise political programming through a managerial framing rooted in discourses of complexity and ‘adaptive management’. This third generation of political programming seems well suited to defuse the depoliticising processes identified by Venugopal, insofar as the bureaucratically palatable managerial frame creates space for political practice during implementation, separate from the cognitive limits of planners. Drawing on seventeen interviews with leading policy advocates and practitioners, the article suggests that this potential has not been realised: operational, accountability and financial processes constrain political practice during implementation. Analysis of third generation projects reveals that the limits on political work are not simply cognitive and do not simply affect planners – they mask an underlying and ongoing squeamishness and discomfort in acknowledging the compromises and engagements of practice during implementation.

The scope of the article’s argument is the broader emergence of ‘third generation’ programmes in the years following the financial crash of 2007. Whilst the article takes the UK as its primary site, the theoretical contributions in relation to operation of the depoliticising processes and the possibility of transcending them extend beyond the UK into other traditional donors. Further, these programmes emerge within a wider community of practice centred on networks like Doing Development Differently (Wild et al. 2014, Wild 2021), revealing sites across bilateral donors, multilateral donors and UN agencies. To take one prominent example, in USAID these have taken the form of an emphasis on organisational learning through a significant ‘Collaborating, Learning and Adapting’ initiative (interview I, USAID official) which has become central to USAID’s working practices.

The article is structured as follows. The second section theorises the anti-politics machine. The third section makes the case for a third generation of project, following the emergence of projects like ZAP. The fourth section then outlines the promise of third generation projects in dismantling the anti-politics machine. The fifth section reviews empirical accounts which foreground the institutional blockages to political practice during implementation. The sixth section offers a summary and closing points.

Theorising the anti-politics machine

Taking inspiration from ZAP, this section extends the theoretical apparatus through which Venugopal treats the anti-politics machine to explore depoliticisation in the context of development practice, particularly during implementation.

Depoliticisation in the authorising arena

The development industry must legitimise spending through the identification and justification of technical projects. Venugopal disaggregates the processes by which these processes depoliticise aid into three (Venugopal 2022, p. 4): the first process insists that development spending must for ethical reasons be justified technically, contrasting the clean altruism of technical arguments with the venal and interested domain of politics; the second process operates through the legal-bureaucratic demand for predictability, where the complexities of social dynamics are reduced to ‘the rational and systematic implementation of plans’ (ibid); and the third is international development’s disciplinary reliance on economics and its epistemological orientation. Each of these processes is discursive in nature, insofar as it concerns the form and language of development representations. Together they produce the cognitive divide between an unknowable political realm on the one hand and an arid technical domain of operational knowledge on the other. The arena in which they work may be described as the ‘authorising arena’ (Honig 2018).

Before addressing the operation of the three depoliticising processes, it is worth sketching the techno-politics of project design within the authorising arena. Critical accounts suggest that aid bureaucracies represent societies in such a way as to foreground problems to which the development industry affords a solution, necessarily framing the country as a particular kind of legible object – Egypt is represented as a thin ribbon of rich farmland, surrounded by desert and placed under increasing demographic pressure (Mitchell 2002); Lesotho is a rural backwater, disconnected from labour markets (Ferguson 1996 [1990]). Critiques of these representations have argued that they are interested, disenfranchising (Escobar 1995), ignorant (Hobart 1993) or just plain wrong (Ferguson 1996 [1990]). Nevertheless, problems are framed to prompt a repertoire of possible ‘travelling’ models (Behrends et al. 2014), each of which have been established as legitimate in the field on the basis of these (mis)representations (Whitty 2019). In aid, a travelling model is a deterritorialised, corporate or epistemic product which is applied, suitably adjusted to account for particularities of any given environment (Murray Li 2007, pp. 230–69; 2011, Behrends et al. 2014, p. 2, Mosse 2011). Travelling models are therefore the vehicles by which the ethical, bureaucratic and epistemic processes are composed and legitimised. They are top-down and technical. Each model is legitimised through categories cemented in global policy (e.g. orphans and vulnerable children Green 2007, p. 140) and maintained through a scaffold of quantitative practices of data collection and use (see e.g. Rottenburg and Merry 2015, Adams 2016).

Adaptively managed, political practice projects like ZAP are applications of one a kind of technical model, albeit one where (unusually) the project details are only partially determined up front. How do adaptively managed, political practice programming relate to Venugopal’s three depoliticising processes? Each process may be taken in turn. The first is an ethical reluctance or squeamishness to engage with messy political processes. It may be disregarded in relation to this kind of programming, which is ipso facto designed explicitly to engage in political practices or mētis (I discuss this further below). The second and third are, respectively, the operation of bureaucratic/legal processes and the epistemic commitments of economics. Their operation must be treated with care. As Venugopal notes, economics is only one of a number of epistemic communities within the aid landscape: the UK development apparatus has no fewer than thirteen disciplinary cadres, most of which generate their own travelling models.² As the article suggests below, adaptive, political practice projects draw on epistemic references in political science, user-experience design, and agile or adaptive project delivery (Pett 2020). Economics is not a primary reference point. Indeed, insofar as economics focuses on transactional framings of results-based management, it is the antithesis of this kind of programming.

Venugopal suggests that other technical fields are subordinated to economics and must be framed and justified in economics terms. While it is stretching the economic logics’ limits to argue that outcomes such as reductions in maternal mortality or infant mortality, or education outcomes, or democratic process improvements, are primarily justified through instrumental economic rationality, yet he is right to underline the crucial importance of economics. A senior civil servant suggested that they act as ‘umpires’ within the UK aid apparatus (interview K, former DFID Aid Effectiveness Director). That is, they are positioned at crucial gateway points, reviewing business cases or assessing claims to value for money and evidence. In this, they are not primarily drawing on the disciplinary apparatus of economics but rather acting as administrators and guardians of the bureaucratic and managerial processes (Venugopal’s second process; see the account of Mosse (2011) for the interaction between anthropologists and economists in the World Bank). Within the UK then, economic epistemic processes and bureaucratic / managerial processes are therefore thoroughly blurred within the authorising arena, such that in many cases the processes may scarcely be distinguished. Adaptive, political practice projects sit in awkward relation to each, as I elaborate below.

Depoliticisation in the implementation arena

A project is never simply implemented or delivered, any more than a technical model is taken directly off the shelf for implementation without adjustment (Latour 1987, Mosse 2005, Lewis and Mosse 2006b, Behrends et al. 2014). It requires the expert work of translation from the abstract forms of a travelling model into the specific context, institutional languages and meanings of a particular project’s actors (Mosse 2004, p. 647; drawing on Actor Network Theory). Once started, projects undergo a process of constant negotiation and translation as meanings are contested, activities negotiated and actors enrolled. This process lasts the life of the project. Tania Li presents an account of a conservation project in Indonesia which sought to reform the conduct of the people of sixty-seven villages in relation to a neighbouring national park, where the attempt to depoliticise is itself a project or attempt that never concludes (Murray Li 2007). She shows how villagers’ contestation and resistance eventually ‘punctured’ attempts to determine an exclusively technical discourse around a conservation and development project (Murray Li 2007), presenting the technical account with a challenge it could not contain. Nor is resistance the only mode. Gal et al. (2015) show how the process of translation involves a transformation of project ideas, to produce new assemblages and registers – for example, where locally prevalent forms of care merged with the project’s rights-based ideas. This is part of the ordinary practice of project of project performance and implementation.

The processes of negotiation and translation happen in an ‘implementation arena’, which concerns the actors and agents involved in project implementation. The implementation and authorising arenas are not strictly separate, nor finally defined: the implementation arena is porous, lacks clear boundaries and is characterised by constant criss-crossing within its bounds (Long 2001, p. 32, Mosse and Lewis 2006, pp. 10–1). What, then, does depoliticisation mean in the implementation arena? In his ethnographic account of a technical assistance programme, Richard Rottenburg suggests that aid discussions are depoliticised – or articulated through a shared technical language or metacode – at the insistence of both the aid agencies and their developing country interlocutors. Development cooperation occurs under conditions of heterogeneity ‘of not only interests but also basic orientational knowledge’, with significant effects:

The tricky combination of setting conditions and offering assistance, of hegemony and equality, and of the superiority of one party over the other results in a perfidious form of camouflaged disciplining. (Rottenburg 2009, p. 193)

Under these conditions, resorting to the metacode is necessary to enable cooperation and to obscure the subordination of local knowledges and orientations (Rottenburg 2009, pp. 190–1). The technical language comprises the language through which this disciplining process happens, enabling all parties to work together whilst managing the discomforts caused by the tensions entailed in the aid relationship. Ordinary aid relations are depoliticised simply in order to implement the project.

Work in both the authorising arena and the implementation arena is framed by technical discourses which are abstract, based on universal policy categories and travelling models. These serve to obscure the political practices of compromise and translation necessitated by implementation. David Mosse argues that projects are failed when their support in the authorising arena fails – project teams must therefore maintain a coherent narrative that retains legitimacy in the authorising arena whilst accommodating the translations and vicissitudes of project implementation (Mosse 2005, p. 158).

In theory, then, the anti-politics machine operates doubly: it flattens and simplifies the social and political practices needed to legitimise a technical project model; and it reciprocally flattens and simplifies the representations of the project’s implementation processes. For traditional forms of development programming, the translation ‘chain’ by which accounts are provided obscures the dynamics that occur within the project arena; they serve to hide the formulations of unofficial scripts whose existence is necessary to accommodate the tensions that exist in the official project plans (Rottenburg 2009). As Scott observes ‘the formal order is always and to some considerable degree parasitic on informal processes, which the formal scheme does not recognise, without which it could not exist, and which it alone cannot create or maintain’ (Scott 1998, p. 310; see also Wynne 1988).

For the team charged with project implementation, politics is not (just) a matter of cognition but of lived, situated practice. Whilst such practice does not address explicitly or necessarily even analyse the institutions by which resources are distributed or people governed, it is politics: it concerns the contestation of meaning within an implementation arena comprised of actors with different value systems – whether contestation over a specific policy, as is usual for third generation political projects (e.g. procurement rules of the food reserve agency, as in the case of ZAP) or even just the project resources itself – constitutes political work.

Adaptive management as a third generation of ‘political’ projects

The first generation of ‘political’ projects emerged in the late 1980s and 1990s (with very limited examples before 1980, see Andrews 2013, pp. 4–5). Aid provision had shifted to poverty reduction as its central logic. Failures to trigger development processes through economic capital investments or structural policy reform had opened up state functions and institutions as areas for intervention (Craig and Porter 2006, pp. 64–75, Carothers and de Gramont 2013). The political projects were narrow: they concentrated on reform to the functions of specific state agencies, initially under the broad heading of Good Government (ODA 1993) and then ‘Good Governance’: they included decentralisation, anti-corruption, human rights, public financial management, the judiciary and at times democratic processes (Leftwich 1994, Brautigam and Knack 2004, Carothers and de Gramont 2013). Early political analyses were ‘very formulaic’ (interview M, former DFID governance advisor). They were political in the sense that the targets were the public rather than the private sector. Otherwise, they were largely blind to the specificities of the local situation or were driven by highly-interested analyses (Ferguson 1996 [1990], Leftwich 1994). Like technical assistance programmes in other domains, their reforms were based on universally applicable models, often directly transplanted from Western institutions (Andrews 2013, pp. 7–11).

As UK aid committed to poverty reduction, the second generation of political projects widened the lens of analysis and intervention abruptly to address ‘governance’ more broadly. It was no longer enough to concentrate reforms on state agencies: the state had to be situated within the political and social institutions, its relations, its elite settlements, its limitations. Attention turned to how aid might support or undermine governance reforms (Brautigam and Knack 2004, Rajan and Subramanian 2007). The sector instituted ‘demand-side’ projects seeking to develop social accountability and thence improved service delivery (Carothers and de Gramont 2013, pp. 136–9). Participatory processes became mainstreamed, aiming to bring citizens in; donors committed to shared ownership of aid strategies through Poverty Reduction Strategy Papers (Craig and Porter 2006, Booth 2012, 2003). As the occupations of Afghanistan and Iraq coincided with ‘bottom billion’ analyses focusing attention on institutional and conflict-drivers of poverty (Collier 2008), UK aid pivoted to work in fragile states, intensifying the focus on political institutions and elite settlements, leading to the operationalisation of state-building processes (Whaites 2008, Di John and Putzel 2009, Carothers and de Gramont 2013).

The abrupt expansion of scope in programming, the work in fragile environments and the increased use of state systems for aid delivery triggered demands for improved analytical capability. Amongst others, these took the form of Political Economy Analyses (PEAs) which were designed ‘to enhance their understanding of the economic, political and social processes that drive or block policy reform’ (Moore and Unsworth 2006, Hyden 2008, Unsworth 2009, Copestake and Williams 2014, p. 134, Yanguas and Hulme 2015). In the new UK department, DFID, governance officials with the support of the Secretary of State developed ‘Drivers of Change’ as a particular form of PEA to help the department ‘build more effective partnerships’ at the macro level (interview M, former DFID Governance Advisor). It also provided analytical tools to inform governance programming (Hudson and Leftwich 2014). PEAs allowed aid agencies to generate representations of developing countries’ institutions sufficient to identify starting points for intervention, their feasibility and focus for intervention (Copestake and Williams 2014, p. 135).

The first and second generation of ‘political’ projects were political in their goals (Carothers and de Gramont 2013, pp. 10–1). That is, the outcome or goal that they seek constitute the rules by which the day-to-day of politics is carried out, ‘small-p politics’ as opposed to ‘Big-P’ Politics (interview I, former DFID governance advisor). Both first and second generations are therefore heavily associated with ‘governance’ as a category legitimising technical models. Yet whilst they were political in their goals, they were frequently critiqued by institutional theorists (Pritchett and Woolcock 2004, Andrews 2013) and critical scholars alike (Murray Li 2007, 2011) for being too rote, too model-based, and not ready enough to encourage local commitment or contestation. To use Carothers and de Gramont’s terms, their means were technical in the sense that the projects follow technical models which provide replica solutions to externally identified problems.

The third generation of projects marked a significant shift away from both the first and the second generation. Their third generation employed political means rather than ends. It had its own epistemic lineages. Complexity theory constitutes one strand. Framing development problems as complex (that is, as being subject to messy social dynamics that make predictability impossible) promotes management approaches that are non-linear, drawing on systems theory (Ramalingam et al. 2009, Ramalingam 2013). The Thinking and Working Politically strand moves beyond the emphasis on political analysis of prior generations with a growing focus on working politically – that is, on political practices loosely based on political science (Faustino and Fabella 2011, Carothers and de Gramont 2013, Dasandi et al. 2019). A third prominent strand highlights the lack of political commitment and persistent mimicry behaviours in traditional technical assistance work. It emphasises local commitment to local problems and neo-institutional theory and has been framed into a specific approach known as ‘Problem-Driven Iterative Adaptation’ or PDIA (Andrews 2013, Andrews et al. 2017).

In general, two features can be identified which unite these strands. The first feature is that such projects prioritise political practice or mētis – informed by PEAs and such analyses, but crucially not limited to that analysis. Political practice entails forging relationships with and knowledge of coalitions and interest groups, their dynamics and the play of power (Eyben 2006, 2010, Whitty 2019). It is typically characterised as involving an entrepreneurial, emergent and improvisational stance, probing the space available for mobilisation, alignment or movement (Eyben 2006, Faustino and Fabella 2011, Booth and Unsworth 2014, Kumar 2021). Most policy accounts highlight the importance of local leadership and locally salient problems (Booth and Unsworth 2014, Wild et al. 2015, Andrews et al. 2017) to circumvent the development sector’s long difficulties with ‘ownership’ and ‘political will’, which lead to – at best – projects which demand extensive translation (Lewis and Mosse 2006a), or – at worst – empty mimicry (Andrews 2013), resistance (Murray Li 2007) or obfuscation (Rottenburg 2009).

For example, in the case of ZAP’s ‘inclusive growth’ strand the project team partnered with local think tanks working on economic and agricultural issues and supported their established research programmes and purposes, whilst developing their advocacy functions; further, they convened and facilitated like-minded Zambian civil society organisations to advocate for specific policy changes and form common platforms; and they commissioned research to support their policy arguments. For example, if the government announced a change in the procurement regulations around grain reserves which advantaged certain large landowners, they would build on their ongoing work with the agricultural think tank; they would help convene a group of civil society organisations and farmers unions with whom they already had links; and they would support the think tank and civil society to produce a shared policy communique addressing the change. The project inputs were primarily ongoing central lobbying expertise and the ability to make resources available quickly. Rather than a project document specifying activities up-front, the project document outlined a broad theory of political change underpinning the project, and enabled a flexible interpretation of that combined with ex ante evaluations. The tale is one of incremental relational work to build influence and collaborative responses to unfolding events.

The second feature is technical/managerial. Third generation projects have increasingly been designed in order to defuse the depoliticising processes identified by Venugopal, and articulated elsewhere as a kind of counter-bureaucracy (Natsios 2010, Valters and Whitty 2017). They have instituted experimental programming approaches involving multiple exploratory approaches; frequent learning, the willingness to emphasise what works; tools that emphasise searching and innovating over following plans; and an implementation culture of learning from failure and exploration. In the case of ZAP, for example, the theory of change was rather high-level, focusing on general principles of good lobbying work (develop evidence base and arguments; build networks, relations and coalitions); workplans adapted quickly to policy opportunities; output and outcome targets were sufficiently generic to permit good policy reform work.

Unlike the first and second generations, therefore, the third generation is not ‘political’ because its goal is reform of the institutions governing political life. It is political because its projects adopt lobbying practices and associated managerial arrangements whose aim is the change of policies, regulations and practices that govern ordinary economic and social life generally. These processes celebrate, acknowledge and formalise the commonplace project management compromises and translations that characterised development implementation. The focus on political practice subverts aid’s depoliticising processes, which are typically treated as discursive and representational: it makes the case that politics is about a kind of situated and relational practice, acknowledging and bringing to the surface the kinds of work commonplace and inevitable in both authorising and implementation arenas. It reimagines the ontology of development politics and the cognitive divide.

As such, the emergence of third generation programming may be read as a response to managerialist reforms within DFID which had been increasingly entrenched since 2007 and particularly since the new Conservative-led administration in 2010 (Valters and Whitty 2017). The emerging ‘third generation’ programmes offered a management alternative to these rigid bureaucratic frameworks, framed by results-based management and evidence-based policy programming that increasingly characterised development policy. Their heyday may be taken as being from 2010 to 2020. During this time, several communities of practice emerged, notably the Doing Development Differently, Thinking and Working Politically and Problem-Driven Iterative Adaptation (Guthiel 2019). For example, the Doing Development Differently community (DDD, see: Wild et al. 2014) acted as the most inclusive of the communities and collected case studies from forty-five different aid projects, each adopting some aspect of adaptive programming.³ At its height, DDD claimed over four hundred signatories from sixty countries (a mix of practitioners, policy-workers and donor officials from the World Bank and other regional development banks, from UN agencies and from several bilateral donors).⁴ Thereafter, cuts to the aid budget and the absorption of DFID into a new Foreign, Commonwealth and Development Office (FCDO) reduced the space dramatically within the UK at least (interview J, former DFID governance advisor and consultant), although as noted it retains significance within an international community of practice.

Third generation political projects: the hope for dismantling the anti-politics machine?

The discomfort was people in the governance sector feeling that people in the aid sector were having to lead a double life where they talked one language to their bosses and then did something totally different on the ground. (Interview A, INGO policy advocate)

What I was good at was explaining sausages being made by our Philippine partners. And it looks super messy, and ugly. And then I turn it into a hot dog that the donors recognise. (Interview B, INGO practitioner)

In this section, I analyse the operation of the anti-politics machine on the specific features of third-generation political projects, suggesting how these projects’ particular traits hold out hope for dismantling the anti-politics machine cognitive operation by focusing beyond the authorising arena, to cover processes of implementation. I draw on a total of seventeen interviews with policy advocates and practitioners (twelve interviews were conducted in 2022; supplemented by five interviews conducted in 2013–14 during my PhD). Interviewees were selected primarily from members and contributors to the Doing Development Differently policy community working in the UK, which acted as a common platform for promoting adaptive and political practice programming, incorporating proponents of several strands of policy movements which together constitute the community.

During the process of advocating for this kind of third generation programme, ‘the language of adaptive management became the thing that landed within some donor organisations … ’ (interview D, think tank official). Emphasis on the managerial element to the exclusion of the practice is therefore in part a tactical reaction to the institutional forces of the donor agencies, as key actors in the authorising arena.

We talked about adaptive management rather than thinking and working politically more, because you can’t talk about politics or the World Bank are not allowed to suddenly it’s a slightly more acceptable technical language … (Interview E, think-tank official)

To gain policy traction you need a recipe; you need a brand; you need an idea; you need something which is ‘toolkitable’ (interview A, INGO policy advocate). The adaptive managerial principles of iteration, flexibility and dynamic learning in quick feedback loops affords the ‘toolkitable’ model which holds traction within the authorising arena – which as outlined above, comprises a set of technical models organised broadly by categories of outcome. The depoliticising processes identified by Venugopal are therefore in operation: ‘social realities and development problems must be completely decomposed and reordered to fit within a managerial iron cage’ (Venugopal 2022, p. 4). Yet within the management framing of adaptive management, the much-critiqued ‘planner’ (Easterly 2006) is displaced from the cage in favour of the ‘searcher’ whose moment comes during implementation. In Easterly’s terms, the planner creates the management environment for the searchers – or implementers – to do their job.

The depoliticising emphasis on managerial processes is reinforced through the discursive emphasis on the use of ‘complexity’. The term complexity has come to stand not for a formal mathematical approach to ‘agent-based modelling social network analysis’ but rather an applied epistemology whose practical insight was that uncertainty should be taken seriously as a matter of management (interview C, think-tank official and practitioner; D, think-tank official). Political contestation and messy social dynamics are reduced to a depoliticised ‘complex’ or ‘adaptive system’, using a metaphor from the physical sciences. The solution – so the argument goes – is to be found in a technical managerial response. Inevitably, the technical language is exclusionary:

A really big moment for me was an exchange with some Nigerian colleagues […] where we did a presentation on adaptive management and one of them said. ‘Oh, Okay. So what you do is you come on, you look at what we’re doing anyway, and have been doing for years. And then you formulate it and turn it into a new idea. Then we had to learn the language. And it’s like it’s been taken away from us, and then suddenly it’s something we have to do to get funding or to get attention. (Interview A, INGO policy advocate)

But as an experienced practitioner from the Philippines observed, adaptive management is a performative dressing up of the ordinary political practice of advocacy in a managerial language: ‘I don’t even know the language … I had to look it up on Wikipedia, right? So I said, “Yes, that’s what we’re doing!” I’ll appropriate whatever language you want’ (interview B, INGO practitioner).

Ordinarily, the anti-politics machine works doubly, first by flattening the social and political dynamics in the authorising arena to legitimise a project model. Second, reciprocally, it flattens out in project representations the existence of any translations necessitated by project implementation, leaving the centre blind to the tensions intrinsic in implementation. For third generation programming, in contrast, the activities, outcomes, and detailed budgets are left undetermined in advance; the processes of ongoing negotiation and translation are openly acknowledged within this structure, permitting the relational practices to emerge and evolve, and creating the space to acknowledge in accounts the emergence of these new compromises. The translations are therefore brought to the surface and made not just speakable but, in principle, celebrated.

Whilst the framing of third generation political projects leads with a management model that is ostensibly technical and depoliticising, that management model is designed precisely to create the space for political practice. Political practice is situated, relational and improvisational where ‘actually often you don’t really know how to make change happen: you have to figure it out as you go’ (interview D, think-tank official). Following James C. Scott (1998, p. 309), that practice can be understood as mētis or situated, practical know-how rather than techne, universal technical knowledge (see also Kumar 2021). Scott’s analogy is to the knowledge of the pilot rather than the navigator: the situated, practical knowledge of the locality by workers in the development sector (and, since it highlights the knowledge of individuals in the sector itself, does not necessarily prioritise realisation of local values in any emancipatory sense). It is technical, in the sense of a craft, but political in its application. In the case of ZAP, for example, the workplan, outcome structures and financial systems were iterative and intended to allow experimentation. They permitted the project to react to the priorities of its partner think-tanks and civil society organisations, whilst co-developing campaigns – tying research into policy messaging and convening – in reaction to the policy announcements of the government. As the project produced, their relations expanded, deepened and evolved (Lodge and Paxton 2017, DFID 2018). Third generation projects specifically acknowledge the importance of these processes of translation and purposively create a space for the processes and attendant practices.

The language of complexity and the need for engagement with its problems has become mainstream. It has been embedded at the heart of DFID/FCDO’s project guidelines. For example, the SMART Rules opened with the framing: ‘In a complex and fragile world, we need to ensure excellent management of DFID programmes. Delivering results and addressing the underlying causes of poverty and conflict requires programmes that can adapt to and influence the local context’ (my emphasis: DFID 2014). Nor are they limited to governance, the usual ‘home’ for political programming: indeed, governance may not be the ‘natural’ starting point (interview F, independent policy advocate). As noted above, first and second generation political programmes were traditionally defined by their goals, not their methods, whilst third generation adaptive projects identify no particular kind of purpose, but constitute a management orientation which can be applied to any effort of institutional change. Conversely, traditional governance projects’ linear technical models are an ill fit to the adaptive and flexible programming. As a consequence, governance had not initially ‘methodologically embraced the full implications of adaptive management’ (interview F, independent policy advocate): the model-based approach afford an ontological understanding of politics as an analytical object – a ‘checklist’ focus – which were antipathetic to complexity or adaptive management (interview G, policy advocate). Even the adoption of Political Economy Analysis in the second generation was often ‘static’ (interview F, independent policy advocate) in the sense of a one-off manual for the rules and mechanics of the game but lacking guidance in how practically to play it. More recently, the implications of adaptive management have been picked up in other institutional change programmes, such as health systems and market-strengthening programmes.

Yet these models remain relatively marginal. Even before the budget for UK aid was slashed and the space for development programming narrowed following the merger, the amount of resourcing for genuinely adaptive programmes were – in the words of one practitioner – ‘absolutely miniscule’. Indeed, projects using these processes seriously are often designed to stay unnoticed:

Little enclaves below the radar, people can still stay below the radar for budgets not too big and they don’t come to the notice of, you know, big fish. (Interview A, INGO policy advocate)

Programming focusing on the improvisations and unexpected opportunities of situated, political practice remains marginal in comparison to the mechanical implementation of technical models.

A false hope? Operational, accountability and financing processes shackling implementation

Why have these projects remained marginal in UK development, despite a central place in the primary project guidelines and viable project models? Interviewees highlighted several inhibitory operational, accountability and financing processes which shackle implementation. One advocate who worked for a consultancy noted the difficulty in recruiting team leaders who adopt adaptive approaches (interview C, D). The sector’s team leaders are trained to deliver linear programming. An NGO manager seeking to drive adaptive practices in her organisation talked of the need to mainstream adaptivity across their project management trainings, which were oriented towards traditional ways of working (interview G). Others stressed the importance of shifting procurement processes either by using relational contracting or persuading procurement to use single tender actions (interview A, C, D). This requires changes in guidance and the formal approval of models within procurement systems set up otherwise. Others yet identified the need to innovate within accountability, budgeting and reporting systems to permit flexibility in budgeting and funding approval (interview E). It is no use being flexible and adaptive in reacting to emerging political opportunities, if the finance processes demand a three week turnaround to approve non-budgeted activities (interviews A, C). Yet others stressed the importance of monitoring, evaluation and learning systems adapted to flexibility rather than audit (interviews, A, B, E, G).

Advocates for political practice emphasised how it was necessary to find pragmatic, organisationally-oriented workarounds for standard organisational practices in financial forecasting and management, evaluation, monitoring, human resources (with a small literature of practitioner-oriented how-to notes, e.g. Booth 2011b, Booth and Unsworth 2014, Wild et al. 2014, Wild et al. 2015, Derbyshire and Donovan 2016, Sharp et al. 2019, Sharp 2021).

These, however, are workarounds: interviewees stressed that they work against the grain of operational, accountability and financing processes, all of which were designed for linear programming: for pre-designed activities, budgeted and agreed beforehand, to ensure control by the donor. Adaptivity is not incentivised in the normal run of work. As a consequence, the interviewees stressed scepticism that adaptive programming and its commitment to flexibility is in fact being taken seriously:

These very seasoned and sort of sceptical contractors basically […] said, ‘We don’t believe you. So the only way [we will be successfully adaptive] is if you give us a hotline. So when some lower ranking Monitoring, Evaluation and Learning person says you’ve got to stick to the indicators, we can just go over their heads and say, they’re blocking us’ […] They thought, yeah, all this nice hand-wavy stuff coming from the big bosses was not going to filter down to the people who work nine to five and then just want to go home. (Interview A, INGO policy advocate)

People would point a finger at the donor and say, ‘but the donor won’t let me I’d love to adapt, but the donor won’t let me’ […] And then we started realising that actually, there are quite a lot of internal constraints that that stop us from behaving in the way that we say we want to like incentives and behaviours, right? (interview G, INGO and consultancy practitioner)

For adaptive management to work requires more than an acceptance of its validity but also engagement ‘not just with the senior people, but with the parts of the organisation you would never normally talk to: so it’s the HR, compliance, finance, procurement […] unless you get those people they will frustrate you’ (interview A, INGO policy advocate). Multiple organisational processes tie the implementation arena inescapably to the planner-based framing of the authorising arena.

The first difficulty – that of the bureaucratic processes’ reach into the implementation arena – reveals a second and deeper difficulty. In principle, third generation political projects allow the project team the space to identify and work with actors with whom there is an alignment of interests, institutional assumptions and orientational knowledge. In the case of ZAP, for example, the project team worked closely with think tanks and civil society organisations who shared similar progressive, liberal ideologies and opportunistically with organisations with whom they could align on particular campaigns (Booth 2011a, pp. s20–s22, Carden 2009, Faustino and Fabella, 2011).

However, such projects risk confronting the development sector with an abiding discomfort, rooted in the differences in actors’ orientation, systems of knowledge and value in the implementation arena:

We don’t talk to people, right? We don’t talk to people in other silos, we don’t talk to constituents, because [the aid apparatus will] think we’ve been corrupt. Or they’ll they’ll say we did favouritism … . And we found that essentially, the aid, people in the aid industry are not in that space, they are observers of that space. (Interview H, PDIA policy advocate)

In traditional ‘technical’ or ‘top-down’ development processes, the problems and values are stated within the project’s authorising arena, translated from the political and technical discourses in London and Washington D.C. into the context. Within second generation governance programming, for example, the development discourse-defined model-based approach is antagonistic to the search for locally-driven problems, since their analytical frames are imposed from outside and they do not acknowledge the politics which happens in the act of engagement and discussion.

A leading proponent of PDIA which insists on local ownership of the problem statement has suggested that progress with aid agencies was limited due to ‘hegemonic management mechanisms’ (interview H, advocate for PDIA). By finessing depoliticising processes in the authorising arena, third generation political programming in principle creates the space for relational and improvisational political practice in the implementation arena. Yet it is stymied by the compromises needed to meet local priorities. The sheer extent of the transformation required in the managerial constraints and the celebration of compromise in the field renders such projects marginal and ‘under the radar’. The challenge is not (only) cognitive; the insistence on technical legitimacy is grounded in tensions encountered during implementation, which are in turn produced because the sector’s authorising arenas have never been institutionally comfortable to undertake the political compromise and flexibility by which emancipation might be possible.

Conclusions

Can the anti-politics machine be dismantled? Venugopal points to the operation of discursive processes of depoliticisation producing a cognitive divide between the political and the technical: the political world can only ever be represented as stripped down and reduced to technical approaches. This article has focused on third generation political programmes and used them as an entry point to explore the operation of the anti-politics machine further. In particular, it has taken the reimagination of politics entailed by these third-generation projects, whereby ‘politics’ is not discursive or representational but a matter of situated practice or – to use James C Scott’s term – mētis. I have suggested that this reimagination celebrates, formalises and foregrounds the normally-obscured processes of translational and negotiation that happen in project implementation; in the case of third-generation programming, these commonplace practices are turned towards specific areas of policy change. The article suggests they entail an ontological transformation of ‘politics’ as an object, displacing the focus from the technical discourses in the authorising arena and towards the arena of practice.

Third generation programmes create a technical model resting on depoliticised language of ‘complexity’ and ‘adaptive systems’ which are framed to meet the discursive and depoliticised expectations of the authorising arena; yet which are also designed specifically to create the space for relational, political practice through managerial commitments to adaptivity and flexibility. The managerial commitments suggest a form of technical assurance, that this kind of political practice is legitimate. Rather than the complete masking of earlier efforts to be political, political practices are made visible, albeit clothed in a technical language of complexity and adaptive systems. This kind of project thereby hold out the possibility of subverting or circumventing the operation of the anti-politics machine as described by Venugopal.

However, that possibility has not – for the most part – eventuated. Despite commitments in project guidance, adaptive programming has remained relatively marginal within UK aid. The foregoing analysis suggests that this is because of ongoing squeamishness of political practice, reinforced by a suspicion of such practices’ need for flexibility and room for manoeuvre in project planning. Bureaucratic controls continue to exercise force in implementation; the squeamishness for engaging with politics may be a response to the heterogeneity of interests and assumptions. That suspicion is expressed through the multiple processes binding implementation to a project’s authorising arena. Whilst negotiation over project implementation is commonplace, the ordinary operations of the anti-politics machine hide these processes from view in project representations that cannot acknowledge political divergences within the implementing arena. Third generation projects are not proof against its operations.

If third generation projects offer a way where aid can be political, perhaps the question they raise is whether aid should be political. The squeamishness about politics is revealing. In normal development programming, the work of negotiation and mobilisation and the possibility of political divergence are typically obscured and denied. In contrast, the third generation’s search for political allies and commitment to lobby are overt. In the case of ZAP the DFID-funded project team openly sought to partner with think-tanks and civil society with shared views to convene a political advocacy community arguing – amongst other things – for fiscal propriety and an appropriate policy on debt and expenditure. The ordinary technical figleafs for the prescriptions of the international liberal order are discarded; the politics is open. They force the question: how political should the sector get?

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Brendan S. Whitty

Brendan Whitty is a Lecturer at the University of St Andrews. He researches the techno-politics of the international development sector. He is interested in the intersection of the international development sector and processes of global capitalism. He has worked in the development sector since 2004.

Notes

1 In 2020 DFID was integrated with the Foreign and Commonwealth Office (FCO) as part of a new Foreign, Commonwealth and Development Office (FCDO).

2 See the technical competency frameworks: https://www.gov.uk/government/publications/dfid-technical-competency-frameworks.

3 See the project website for case studies: http://web.archive.org/web/20180705031921/http://doingdevelopmentdifferently.com/.

4 http://web.archive.org/web/20180705031921/http://doingdevelopmentdifferently.com/.