Intersectional coproduction and infrastructural violence: experiences from Pakistan


 The delivery of projects for the coproduction of services raises multiple questions about how different structural barriers prevent and hinder the participation of various sectors of the population. Intersectionality theory provides a critical lens to examine the delivery of such coproduction projects to refine any strategies to include vulnerable perspectives or perspectives that get silenced by existing hierarchies. This paper presents an intersectionality-led analysis of the delivery of a project to improve public safety in Pakistan. The project mapped existing concerns about urban violence of different groups of the population. The project used a multilayered approach to facilitate the engagement of excluded views, both in the constitution of the research team and in the involvement of communities. An intersectionality framework is applied to analyse the deployment of the project in terms of design, innovation, planning, and signification. The analysis shows that there are limitations to how far coproduction exercises can challenge existing social structural barriers.


Introduction
The Sustainable Development Goals emphasize equity and its interactions with sustainability (Leach et al., 2018). The Sustainable Development Goal 11 (SDG11), advances the notion of inclusive cities, and the motto of 'leave no one behind' has also reached the New Urban Agenda (López-Franco Intersectionality emerges from a feminist analysis of the multiple forms of oppression affecting marginalized groups in society. The concept initially engaged with the intersection between gender-discrimination and racism in the US legal system (Crenshaw, 1991). Since then, intersectionality has served as a point of reference to relect upon the multiple ways in which overlapping systems of power dictate people's life chances and engagement with the state (Kumar, 2017). The concept of intersectionality as emerging from overlapping systems of opresion is a powerful critique of frameworks to understand diversity and inequality that rely on a systematic list of multiple forms of identity and their intersections. Intersectionality, instead, turns the focus to daily experiences of discrimination in speciic contexts (Sultana, 2009;Truelove, 2011). Forms of collaborative governance may be recognized as a tool to work with and claim intersectional concerns in urban service provision, but their operation may also lead to further inequalities (Beebeejaun, 2017). Intersectional approaches depend on existing knowledge systems, creating an apparent dilemma because intersectionality critiques emerge from the various structures of power and knowledge that they purport to criticize (Collins, 2015). Salem (2016) argues that intersectionality is a travelling theory that needs to be grounded in context and experience since the neoliberal academic context has facilitated it travelling to the 'mainstream', co-opting it and often erasing its 'radical traits'. Intersectionality relates to visible and invisible forms of violence and the institutionalization of social arrangements that systematically oppress individuals and are internalized and accepted by the powerful and the powerless (Castán Broto, 2013;Scott-Villiers and Oosterom, 2016). This paper focuses on applying an intersectionality lens to a case of an action-research project to help the coproduction of safety in urban spaces in Pakistan. The research question is: how did the project address intersectionality concerns, and with what results? We apply the framework developed by Castán Broto and Neves Alves (2018) to raise critical questions about the delivery of service coproduction projects. The framework distinguishes four types of service coproduction. Each type raises different questions about intersectionality: Context-speciic technological innovations and coproduced design; Coproduction of institutional innovation and systems of provision within a collective space, such as a neighbourhood; Coproduction of planning processes, rules, and regulations that may establish new frameworks for existing systems of service provision; and Coproduction of new systems of signiication, new principles of practice, or even a change of paradigm, such as, for example the recognition of informal processes as part of the urban condition. Ta b l e 1 Critical questions for analysis (after Castán Broto and Neves Alves, 2018)  Each type raises different critical questions, which are summarized in Table 1.
This paper seeks to tackle the practitioner dilemma explained above, through an intersectionality lens. We analyse an example of a coproduction project that combined embedded ethnographic and participatory research methods to develop the irst-ever database on gender, urban violence, and access to infrastructure in Pakistan. The research was funded by the International Development Research Center Canada under the 'Safe and Inclusive Cities Grant' and led by the Institute of Business Administration in Karachi and King's College London. The project responded to a common critique that the lack of information about people's experiences hinders decisionmaking about public services in urban environments (Castán Broto et al., 2015a). The research project aimed at creating an opportunity to engage with working-class people over three years (2013)(2014)(2015)(2016) about how they experienced city life mapping different inequalities, violence, and needs, while creating mechanisms for communities to coproduce a safer city (Anwar et al., 2016).
The project followed previous efforts at addressing poor people's experiences, such as the inluential Orangi Pilot Project (OPP) that led a longterm engagement in community-led sanitation (Hasan, 2010). OPP adopted an integrated perspective, focusing on how the delivery of urban services (housing, water and sanitation, transport) helped construct safe environments that relected the experiences and aspirations of vulnerable residents. The research team worked in twelve neighbourhoods, across two urban areas -the mega-city of Karachi and the smaller, twinned cities of Rawalpindi and Islamabad, home to Pakistan's administrative capital. According to the latest census in 2017, Karachi has a population of sixteen million, Rawalpindi ten million, and Islamabad two million. The census is disputed because it reveals new population balances -and questionable demographic numbers -impacting political power in the country. Since 1998, Karachi, Rawalpindi, and Islamabad have experienced a growth rate of at least 1.5 percent (up to 2.5 percent in Islamabad). Population growth has not been accompanied by a parallel improvement of governance and services. Pakistan has the highest rate of urbanization in South Asia, with an annual urbanization rate of 3.06 percent and 36.38 percent of the population living in urban areas -although experts assert the urban population is underestimated. Pakistan's urbanization is mostly unplanned and happens without adequate infrastructure (Arif and Hamid, 2009;UNDP, 2019).
Looking at this case through an intersectionality lens, it helps to explain how the project uncovered situated forms of social discrimination related to basic infrastructure and service provision. At the same time, intersectionality enables the recognition of infrastructural violence that otherwise would remain invisible. Intersectionality theory supports the development of meaningful processes of participation that address structural forms of oppression directly. The following section begins with a discussion of types of infrastructural violence and how they relate to intersectionality concerns.

Dimensions of infrastructural violence in urban Pakistan
Urban Pakistan, and in particular Karachi, has been the subject of multiscalar political discourses since the Global War on Terror (GWOT, see Mustafa et al., 2018). Pakistan's urbanization and urban governance have been deeply affected by geopolitics, and since the Soviet invasion of Afghanistan in the 1970s and subsequent US-led invasion in 2001, approximately 2.7-3 million residents (registered and unregistered) identify as Afghan (Alimia, 2014). While the majority were displaced to Pakistan's northern areas, a signiicant number settled in the urban Sindh and Punjab -notably Karachi, Islamabad, and Rawalpindi, where the project was conducted. Following the GWOT and these population movements, Pakistan experienced a rise in 'militancy' as well as broader urban law and order challenges, peaking around 2013-2014. International actors debated the causes of violence and militancy without engaging urban residents at all and concluded that the state allowed urban centres to be 'safe havens' for terrorists (see, Felbab-Brown, 2018 1 ). Pakistan's national elite concluded that Pakistan was suffering in a proxy war at the hands of the United States and its allies and that their tactics, through drone strikes, were incentivizing citizens to join violent political groups. As a result of multiple 'security' discourses, which framed Pakistan's urban areas -especially the megacity of Karachi -as 'unwieldy' urbanizing threats and 'pressure cookers,' the state intervened through paramilitary operations, notably in Karachi in 2013, to crack down on terrorism and political violence in the city (International Crisis Group, 2017).
Terrorism has been a major concern for Pakistan since the GWOT, with urban areas as the usual targets. Attacks, usually claimed by the Pakistan Taliban (TTP) or its splinter groups initially targetted government infrastructure, citing grievances over drone strikes, and enforced disappearances. Over time attacks evolved to focus on religious minorities, deemed by these groups to be defacing Islam. Rawalpindi, Islamabad, and Karachi have been frequently targeted, e.g. in 2012, the TTP bombed a Shia procession in Rawalpindi, killing twenty-three people (Guardian, 22 November 2012); in 2015 the Jundallah, a TTP splinter group, attacked a bus in Karachi killing forty-six Ismaili Shia.
Identity-based violence is not only limited to terrorism. Minorities are increasingly victims of false blasphemy accusations, a crime punishable by death (Rahman, 2012). Despite a dearth of academic research -due to the security risks of pursuing such work -media reports document state and non-state violence towards minorities accused of blasphemy (Hashim, 2018). The Lahore Supreme Court's 2018 acquittal of Asia Bibi, a Christian woman, wrongly convicted of blasphemy in 2010, led to mass protests, blocked motorways, and vandalized property, paralyzing some cities for days. Pakistan's minorities live in fear of false blasphemy charges, compounding the forms of infrastructural violence already visible in the urban environment (Kermani, 2018).
Indeed, 'infrastructural violence', as summarized by Rodgers and O'Neill (2012), explains how structural exclusions from infrastructure reproduce 'lows' and 'circuits' of power and materiality that feed into and reinforce each other. The violence associated can be active (where articulations of infrastructure are designed to be violent) or passive (where unintended consequences are violent). Overall, structural exclusion becomes possible due to the infrastructure (or lack of), and it also impacts how residents interact with each other.
Pakistan's minorities are often excluded from decent work opportunities and access to infrastructure and services to which they have constitutional rights. Structural exclusion is a form of terror itself (Mustafa, 2005). Many Ahmadi Muslims, for example, fear violence if they self-identify in public (Sayeed, 2017). The persecution of Ahmadis is institutionalized by the state, who deems them to be non-Muslims and expects them to keep their beliefs private or else be labelled apostates (Rahman, 2016). The structural marginalization was illustrated in 2018 when the Prime Minister appointed Ahmadi advisor to the National Economic Council, Princeton Professor Ataf Mian, was pushed out of post due to threats of repercussion by religious groups (Chaudry, 2018). Exclusion from work is frequent among the Downloaded from https://academic.oup.com/cdj/article-abstract/55/1/83/5678528 by guest on 29 May 2020 Intersectional coproduction and infrastructural violence 89 working classes and katchi abadis (unplanned settlements) where other forms of marginalization overlap.
Marginalized ethnic minorities include the Pashtuns, hailing from northern Pakistan and Afghanistan, and the Burmese-Rohingya and Bengali, hailing from Myanmar and East Pakistan (now Bangladesh). Anwar (2013) illustrates how the latter group has been excluded from accessing the national identity card (CNIC), which becomes the basis for citizenship claims on the state in terms of access to basic services, healthcare, education, and employment. Consequently, they have become segregated into 'enclaves' and forced into 'illegality' to access livelihoods and infrastructure. The Pashtuns have become a scapegoat by government and elites for terrorism and the hardening of gender norms since the GWOT (Human Rights Watch, 2014). This scapegoating has transformed into extraordinary levels of surveillance and police abuse. One media report pictures a traders' association notice in Lahore, demanding that all Pashtun traders submit their CNIC and business information to the police station to prove their status (Yousaf, 2017). In Karachi, young Pashtun males were targets of the paramilitary-led (rangers) 'operation clean up' in 2013.
Pashtuns have also become targets of anti-encroachment evictions. Despite the presence of around 52 Katchi Abadis in Islamabad, a Pashtun one in Islamabad's I-11/4 sector -externally branded as 'Afghan Abadi' (although most residents were Pakistani Pashtuns) -became a key target for forced eviction by the Capital Development Authority in 2015. Qualitative research revealed that this was the second or third internal forced displacement for many residents. They had support from civil society and political parties, who pursued a legal case to the Supreme Court (Akhtar et al. vs. Federation of Pakistan). The process was not fast enough to prevent eviction, and at the time of writing, the case is yet to be resolved. In the meantime, forced evictions of the working classes extended into Karachi through the development of large infrastructure projects like the Karachi Circular Railway (Anwar et al., 2018). Public-led efforts questioning the legality of elite developments (Bhatti, 2018) are minor in comparison.
The overlap of gender with ethnic or religious identity exposes individuals to different vulnerabilities with regard to basic services and protection. For example, a working-class Pashtun woman does not experience the city in the same ways as her male counterparts (Anwar et al., 2016;Mustafa et al., 2019). While Pashtun males are exposed to racial proiling and state aggression, Pashtun females face restricted mobilities due to cultural traditions that place the woman's role irmly within the home (Khan and Samina, 2016;Mustafa et al., 2019). For many, the stress of lacking household water or electricity cannot be overcome by venturing into public space. Resource scarcity becomes embodied by women in conservative homes, and it can put them at risk of domestic violence when males return (Anwar et al., 2019). There is an age dimension: women deemed over 'childbearing' age tend to face less restricted mobilities. Younger men are expected to secure household resources. Youthful men also tend to be the targets of arrests and disappearances.
The transgender community has recently gained some state recognition and access to services. Traditionally, the Khawaja Sira (transgender) community was associated with piety, but recently they became some of Pakistan's most vulnerable people. Driven towards sex work as their main livelihood, the Khawaja Sira community has faced gender-based violence and social exclusion (FDI/NAZ, 2017). However, due to civil society activism (including some religious groups), a law was introduced to recognize a third gender that allows people to self-identify on oficial documents. Activism has had positive impacts on public discourse and political participation, but the Khawaja Sira community remains fearful of repercussions of self-identifying on documentation (Nisar, 2018). Violence against the Khawaja Sira continues unabated (Khan, 2018).
In urban Pakistan, infrastructure has reproduced the spectrum of violence, from the denial of basic services to target killings of those attempting to remediate the situation, such as the 2013 assassination of Perween Rahman, director of the OPP. OPP had established itself as a reputable organization for catalyzing infrastructure development in poor communities via a community-led approach. Such an approach involved mobilizing groups by lane (street) and incentivizing them to map their areas, raise funds for and then implement infrastructures. Knowledge coproduction underscored OPP's success: mapping proved useful for the local government, who struggled to grasp Orangi's rapid unplanned development. OPP's constituents became an invaluable resource, incentivizing the government to engage with them over decades (Hasan, 2006). The case shows that urban violence may affect disadvantaged communities, but also the professionals that mediate the coproduction process. For that reason, intersectionality questions are relevant to understand all the stages of the coproduction process, including the formation of action research teams.

Methodology
We used an intersectionality framework to review a project that uses coproduction to tackle active and passive forms of infrastructural violence. Our objective was to deliver a critical analysis of a coproduction project, evaluating the extent to which it had incorporated intersectionality concerns and with what impacts. The original coproduction project examined was a threeyear multidisciplinary research project on gender and urban violence in What kinds of violence have been experienced by the inhabitants of speciied localities? How often? And by whom? How is violence deined and experienced by these inhabitants? How is private and public violence linked within these localities?
The project focused on how residents experienced 'the urban' in the broadest sense, considering access to infrastructure and services, environmental, economic, social, and political issues. The objective was to examine the project critically within its social context, in line with previous research on the deployment of coproduction projects (e.g. Green 2017). In this paper, we separate the original research project from the critical analysis conducted for this paper.
The research project started with the lived experience of residents, who revealed their perceptions and experiences of urban violence through openended interviews, focus groups, and participant photography. A key inding was that the stereotyping of certain groups of men -ethnic and refugee in particular -as 'conservative', violent, or even terrorists, compounded their experiences of state and infrastructural violence. They faced the regular destruction of their livelihoods, 'encounter killings' by the paramilitary and police, and exclusion from basic services (see Mustafa et al., 2019). The stress of enduring exclusion from infrastructure and basic services resulted in hostile encounters between the state and the community. Public forms of violence bled into the home and affected domestic relationships and women's security.
The project documented forms of gendered violence, such as the restriction of women's mobilities and their bodily integrity. Due to expectations to uphold family honour through their 'purity', in most neighbourhoods, women had access restricted in public spaces, also limiting access to educational, work, and governance institutions. This limited women's quality of life and aspirations, having a knock-on effect on household livelihood security. Livelihood insecurity and exclusion from basic services contributed to incidents of domestic violence (Anwar et al., 2016;Mustafa et al., 2019).
The exclusions which had the most pronounced impact on gender relations and the security of women and men were: economic opportunities, access to water and sanitation, and transport. Residents of the neighbourhoods were excluded from state provision to different degrees, due to weak governance, poor quality services, and the label of 'illegality'. Informal providers have become critical to the functioning of daily life. However, in some cases -particularly in terms of water and land -such informal networks have become embedded in violent groups in collusion with the state. The price of water, for example, can be extremely high, leaving residents with no choice but to invest a signiicant amount of income on water, or live with the physical and psychological strain of water scarcity (see Anwar et al., 2019).
Building upon a critical relection by the authors of this project facilitated through critical analysis in conversation with one of the co-authors of this paper (who was not involved in the original research), the analysis focused on examining the four types of coproduction and how they were deployed. Using the guide presented in Table 1, we examined the research process from an intersectionality perspective. Three themes were used to structure the narrative analysis: Designing coproduction processes to avoid invisibilizing multiple perspectives on what services matter and how they matter.
Develop procedures that open up the decision-making process to all groups and actively address groups that could be excluded, for explicit or implicit reasons. Address the possible manifestations of symbolic violence in terms of which interventions are justiied and how they reproduce existing forms of oppression and exclusion.
We constructed a narrative of how the project unfolded using the questions in Table 1 as a guide. The analysis illustrates the extent to which a research project can challenge structural inequalities in society and support the coproduction of safer urban spaces.

Coproducing security in Karachi: an account of an action-research project
Designing coproduction processes to visibilize multiple perspectives on urban services The research was designed by three Pakistani academics, one British academic, and eleven research assistants (RAs) based in Karachi. The sampling strategy targeted 2462 residents across twelve neighbourhoods, seven in Karachi, and ive in Rawalpindi and Islamabad. The sample consisted of ifty percent women and men, with one transgender participant. The sampling technique changed from random to a snowball process, due to the constraints of starting research in neighbourhoods where the team had Downloaded from https://academic.oup.com/cdj/article-abstract/55/1/83/5678528 by guest on 29 May 2020 Intersectional coproduction and infrastructural violence 93 not developed familiarity. RAs received training on the risks associated with snowball sampling.
The neighbourhoods comprised different municipal, ethnic, and geographical characteristics. Vulnerabilities were studied through a comparative analysis of neighbourhoods. In Karachi, the project studied seven neighbourhoods in three towns (Orangi Town, Bin Qasim Town, and Jamshed Town), that straddle the city's centre and periphery, including a mix of working-class settlements and Katchi Abadis. In Rawalpindi and Islamabad, the project studied ive, including two minority Katchi Abadis in Islamabad and three older working-class settlements in Rawalpindi. The methods included a baseline survey, vulnerabilities and capacities analysis, repeated qualitative interviews, focus group discussions, and participatory photography.
The project team was designed to be as representative as possible. The RAs were selected from similar socio-economic and ethnic groups to the residents of the twelve neighbourhoods. The RAs played a central communication role because they were the 'face' of the project, each engaging with the neighbourhoods for three-four days per week throughout the project. They were ive males and six females, aged between twenty-two and fortyive, representing the working class, the lower middle, and middle classes of Pashtun, Punjabi, Sindhi, Bengali, and Mohajir ethnicities.
Bringing the RAs into the conceptualization of the research brought people's views into the design directly because RAs could convey -to some extent -participants' lived experiences. The team debated what would be a viable, embedded, research approach that was safe (possibly even 'empowering') for all. The team recognized RAs' disparate views due to different lived experiences and identities. The team welcomed diversity because it is essential to uncover different perspectives and experiences and to be able to create trust with participants from different social groups. Rather than changing the RAs' views, the team adopted two mutually constitutive approaches: (i) an intensive training course on research ethics and safeguarding, as per an ethical protocol approved by King's College Ethics Committee and (ii) an ongoing training process identifying and managing positionality and subjectivities. The team emphasized the importance of participants speaking for themselves, not guided by the RAs, but rather that the RAs could create and maintain safe spaces for participants to share their views.

Opening up the decision-making process and avoid exclusions
The project started with an inception workshop in Pakistan, where the whole team discussed the research design. Nearer the end of the workshop week, the team designed the survey questions. The RAs challenged the team'sand each other's -assumptions, regarding how to appropriately frame questions. Debates ensued regarding issues such as how to ask a Pashtun male about his ownership of a weapon without making him uncomfortable; or how to ask a young woman about her experiences of violence without raising any risks. Discussions made evident to the co-investigators that the formulation of survey questions by those who lacked lived experiences of the neighbourhoods was problematic.
RAs raised important debates around gendered agency and mobility. Female RAs in Karachi came from different backgrounds: one was working class, two lower middle, and one middle class. Two were Mohajir, born in the city, and one Sindhi, having migrated from a rural area. Two had attended university, and the other had basic education. The age range was twenty-two-to-thirty-four. One had transgressed social norms as a working woman; the others had not. They had different perspectives on how they should engage with communities. For example, the working woman felt comfortable engaging with women and men of different ethnic backgrounds, while the others felt initial fear and reluctance about it being appropriate and safe. We realized that the university-approved ethical and risk assessment procedures were not necessarily designed to address the full range of emotions experienced by RAs. The co-investigators adopted a negotiated approach to deal with ethical or risk assessment dilemmas.
One intersectional dynamic to overcome was related to age and gender hierarchies. Regardless of a speciic ethnic group, Pakistanis typically default to elders as those with more signiicant insights. There is also a gender dynamic, where older males' perspectives are privileged over older females' ones. RAs were affected by these subjectivities. For example, the project had two Principal Investigators -one woman and one man -of a similar age. Some RAs would turn to the male for 'the inal say' on how to proceed. However, the PI was also aware of his positionality, and he would intervene and ask the RAs to relect on these notions. The female PI would do the same, and their united front challenged RAs to become aware of and shift their social normativities.
The relative positions of the research team and their afiliations had a deinitive inluence on group dynamics. The RAs found the context intimidating and saw the co-investigators as seniors to 'respect' and not challenge. The onus was on the most experienced to break down hierarchies and disavow such perspectives. The articulation of a lat structure, in oral and written reporting, encouraged new members to participate but without displacing the hierarchical power structure. Humorous examples were used to challenge the hierarchical image. Small steps created an atmosphere where every opinion could be expressed. Such a day-to-day attitude to the running of the project helped to build relationships of trust over time and mitigated power differentials. Downloaded from https://academic.oup.com/cdj/article-abstract/55/1/83/5678528 by guest on 29 May 2020 Intersectional coproduction and infrastructural violence 95

Examine the manifestations of symbolic violence
The project generated knowledge on gender and urban violence to inform policy discussions. Most participants were pleased to be involved. For many, it was the irst time they had been able to express their views about these issues without distrust or fear of reprisal. The most challenging neighbourhood to enter was a more afluent one, in Rawalpindi. Its public spaces were extremely gendered: young males controlled the streets. Male participants were suspicious at irst, questioning the project's concerns about gender issues, and the presence of women researchers roaming the streets. Male RAs' empathy with these concerns helped to create a space to overcome them. Fundamentally, these attitudes were driven by masculinities, which had been threatened over time due to changes in the community from in-migration of different ethnic groups and engagement with globalized digital media. Through this process, these young men became engaged in detailed, repeated interviews and participatory photography. The success of this process relied entirely on having male RAs within the team who these young men perceived as legitimate interlocutors.
Mitigating power differentials takes time, and not all the embedded inequalities can be challenged through a single project. Even when the researchers are familiar with the context, community relations are embedded in multilayered structures and dynamics that are inaccessible to the team. The objective is to create a similar intimacy to that already experienced by the research team so that the coproduction process leads to teamwork across different settings, for example, through the explicit establishment of a shared goal (Castán Broto et al., 2015b).
Certain groups of people tended to be overlooked. For example, following the compromises made to obtain ethical approval, we did not engage with people under eighteen years old. We witnessed instances of gendered violence against men, women, and transgender people under eighteen, but could not engage them directly in the research. Furthermore, the Khawaja Sira community is particularly vulnerable and excluded, but aside for one participant, they were mostly inaccessible to the team.
Moreover, researchers considered it dangerous, from a safeguarding perspective, to engage with speciic groups and draw attention to them through the research. The Ahmadis, the lesbian, gay, and bisexual community are identity categories with the mark of illegality. People living with disabilities were also inaccessible without speciic methods to target them.
This experience raised questions around the institutional power dynamics that inluence the construction of spaces of participation (Cornwall and Coehlo, 2007). The project team actively considered what kind of space was created. The team created a safe space for the RAs, and RAs used this to construct safe spaces for participants to express their perspectives. However, like in other projects, participation became a means to create boundaries, because: [the difference between] 'normative expectations (of participation) and empirical realities presents a number of challenges for the projects of democratization and development. It becomes evident that the participation of the poorer and more marginalized is far from straightforward, and that a number of preconditions exist for entry. Much depends on who enters these spaces, on whose terms and with what 'epistemic authority' 2 .
The project focused on building inclusive institutions to transform existing systems of governance -rather than delivering material outputs or technologies. Following the four decades' experience of the OPP, which operated in similar areas, and following previous efforts at community mobilization, a strong focus was put onto community-led water and sanitation interventions that could reduce quotidian forms of infrastructural violence. OPP lessons were transferred from an RA who worked with them as a community mobilizer.
Water has a particular political valence in Pakistan. Community development projects tend to centre around water. There is a compelling narrative about Pakistan's water 'running dry' and citizens' resolve to participate in water governance. Formal and informal networks intervene in water management, responding, and contributing to what has become a patronagebased service. The more power one has -through their gender, ethnicity, access to cash, social networks, or location -the easier it is to navigate the water economy. Marginalized urban women are particularly excludedespecially those deemed within the age bracket of sexual activity. On the one hand, they must ensure household needs are met, while on the other they deal with restrained mobility. When there is insuficient water at home, they must negotiate access without compromising their 'femininity' by leaving. One participant left her home to obtain cash to buy water from a tanker. Her husband became suspicious. When she returned he inlicted violence upon her.
Young men from refugee populations, such as Burmese-Rohingya and Afghans, or poor young men from the same ethnic groups, are often excluded from public spaces as they are labelled criminals. Exclusion from basic services (even when they have citizenship claims to them) goes handin-hand with stereotyping and state violence (e.g. from the police). These stereotypes contribute to their marginalization in broader society.
Institutions in urban Pakistan are, for the most part, marred by patriarchal power. Young women can hardly walk into a government agency to demand their rights without facing sexual harassment, threats to their bodily integrity, or questions over their reputation. Every institution needs to make better efforts to provide safe spaces for young urban women.
Young refugee men (or those from marginalized ethnic groups) have varying capacity to participate. If they have CNICs, they are entitled to take part in participatory spaces, but they fear that their CNICs will be coniscated (a common practice in Karachi and Islamabad). Many men would rather avoid any interface with the state, even mediated by NGOs. Infrastructural violence permeates the lives of women and young men.
While the project was able to characterize experiences of infrastructural violence across different groups of the population, it was not able to address every single constraint that prevented participation. There are inherent limitations in community development processes to represent the views of every member of the community. Coproduction processes face the impossibility of representing all perspectives given institutional, practical, and safety constraints. Excluded groups will be negatively affected by coproduction. The challenge, in this case, is how to evaluate impacts on those who remain invisible despite the best efforts to make their concerns visible.

Discussion
The case shows the importance of relecting on whose uses are prioritized in the research design. The project was feasible because it started from a consultation with the communities affected. Communities were not simply questioned about their experiences and knowledge, but also, they were invited to deine critical concepts that informed the assumptions of research and its objectives. There are, however, limitations to this approach. First is the question of funding. Research teams, such as the one formed by the co-investigators, start from the development of an application for research funding, a process that depends on deining the terms of the research narrowly. Even projects that begin from local experiences and a long history of engagement with the research site, like this one, can be framed narrowly to meet funding requirements. Projects then require corrective action to incorporate the four types of coproduction outlined in Table 1 (design, innovation, planning, signiication) and examine the whole life of the project, outcomes, and aftermath. Coproduction processes may end up reproducing the same forms of infrastructural violence that the process intends to address.
The analysis also raises a question about whose interests and needs are most often overlooked. Some groups are systematically excluded from coproduction engagements, such as the Khawaja Sira. The project recognized the drivers of exclusion of this group but could not develop appropriate means to engage them. Coproduction goes hand-in-hand with political activism that denounces the inequalities that the coproduction process makes visible, even when those are not fully addressed.
Mapping baselines is useful to explore whose services are prioritized, but ultimately, services depend on political drivers. The question of access to decision-making arenas becomes fundamental. What are, particularly, the decision-making arenas in which coproduction can impact peoples' lives? The project provided a corrective to alternatives to current models of service coproduction in Pakistan, in which coproduction is used as an alternative to public-private partnership and the semi-privatization of water and sanitation (for comparison, see Farooqi, 2016).
However, coproduction happens in the context of infrastructural violence that reproduces existing forms of oppression and exclusion. An intersectionality-conscious approach challenges discourses that support infrastructural violence. For example, discourses on terrorism and militancy supported paramilitary operations that led to hundreds of enforced disappearances and abductions of young men. Whatever its results in reducing crime rates in Karachi, the operation left behind a legacy of trauma among families, permeating several generations. A coproduction environment offered the opportunity for people to express their experiences of violence and insecurity in the city without fear. Simultaneously, the project cannot transcend cultural barriers that may also result in symbolic violence, for example, through the acceptance of hierarchies of status and knowledge in the work of the RAs.
The analysis demonstrates the growing importance of the intersectionality lens in any community-oriented process, mainly when it is oriented towards service coproduction in a context of environmental violence (Grunenfelder and Schurr, 2015). Intersectionality poses a layer of responsibility for those delivering a coproduction project. An intersectionality lens is an instrument to reinforce community development processes, ensuring the representation not only of multiple voices but also of multiple identities alongside multiple ways of understanding the world. While power relations cannot be eradicated, those who have control of the project also have responsibility for managing power relations and ensuring that alternative views are heard.