Negotiating ‘Hearts and Minds’: conflict, infrastructure, and community support in Colombia

ABSTRACT Research has shown that the counterinsurgent proposition of ‘winning Hearts and Minds’ is more complex than building a road. This paper examines how project workers in three infrastructure projects in Colombia sought community support not for military intelligence or to improve government-community relations, but to intervene with armed groups on the project’s behalf. The findings highlight the role of community institutions in negotiating between two actors – rather than being ‘won over’ by either. This paper also indicates the limitations of community agency in the face of changing local orders, questioning the local empowerment of goods delivery in conflict areas.


Introduction
In an office in Bogota, I sit across the table from a former Minister of Transport.He explains that the construction of infrastructure -roads, electrical grids, dams, bridges, pipes -provokes attacks by armed groups against both communities and construction companies, requiring the military 'to manage this public disorder'. 1This may be due to the supposedly counterinsurgent nature of these projects: Zeiderman observes that since the mid-20 th century, public works in rural and isolated territories in Colombia 'were conceived as antidotes to insurgency' against groups such as the Revolutionary Armed Forces of Colombia (FARC) or National Liberation Army (ELN), winning the 'Hearts and Minds' of civilians away from armed groups and towards the state. 2NTACT Clara Voyvodic claravoyvodic@gmail.com University of Bristol This article has been corrected with minor changes.These changes do not impact the academic content of the article.
However, closer examination of public goods delivery and 'Hearts and Minds' reveals cracks in the efficacy and legitimacy of the strategy.Extensive literature highlights how such programmes are often counterproductive, 3 presume the idea of an overly simplistic legitimacy marketplace' where local communities allocate their loyalty to the highest bidder, 4 and may be prescriptive and imperialistic in its approach. 5So if its counterinsurgent credentials are so dubious, what -if any -is the value of civilian support in the construction of infrastructure in conflict areas?
I argue that project workers -engineers, managers, contractors, and public officials -perceive a strategic value in gaining civilian support during the construction of the infrastructure project.Rather than try to win civilian loyalty away from armed groups, project workers rely on their relationship to navigate complex security environments so that civilians will intervene with armed groups on the project's behalf.However, the ability of civilians to exert influence is affected by fluctuations in the local order.
I examine three cases of infrastructure in Colombia: the construction of tertiary roads and schools in the municipality of Leiva (Nariño), of a secondary road in the municipality of Barbacoas (Nariño), and of a hydroelectric dam in the municipality of Chaparral (Tolima).Despite variation in type of infrastructure and location, project workers in all three cases viewed the community's relationship to the armed group as a source of security.Project workers appealed to the Communal Action Councils (Juntas de Accion Comunal -JAC), indigenous councils (cabildos) or afro-Colombian community councils (consejos comunitarios) 6 to negotiate with armed actors, including the FARC, ELN, and neoparamilitary groups.
This capacity to negotiate was affected by changes in the local order between the armed group and the military.Where the state showed little interest in securing the project at all, armed groups had fewer incentives to adhere to community needs.When the state elected to remove the armed group out by force, the resulting confrontations robbed civilians of direct influence.However, these cases indicate that when the state and armed group reached a 'tacit coexistence', 7 civilian institutions took a central role in negotiating the shared presence of project workers and the armed group in the territory.
To demonstrate the important -but limited -role of civilian institutions in infrastructure construction, I first review the research on counterinsurgency, public goods delivery and aid, and local orders.I discuss the data collection process, including case selection, methods, and ethical considerations.I then present the findings, tracing the construction of the three projects and the strategies used by project workers in 'winning' community support.Finally I discuss how these cases inform our understanding of infrastructure and counterinsurgency, further questions, and the importance of putting local agency and it's limitations at the heart of public goods delivery in conflict areas.

The centrality of community support
The road to winning civilian support requires more than the road itself.In Afghanistan, Kilcullen argues that counterinsurgent success emerges when people use the 'process of the road construction . . .as a framework around which to organise a full spectrum strategy . . . to separate insurgents from the people', rather than just rely on providing the road itself. 8Colombian policies in the 1950s and 1960s sought to empower local organisations such as the JACs as a bulwark against FARC advancement. 9Developed alongside Plan Lazo, this approach put locally-geared public goods delivery at the heart of its counterinsurgent strategy, 10 including infrastructure provision. 11JACs, along with the later creation of consejos comunitarios and cabildos, brought marginalised communities into the political system: a rationale best characterised as 'decentralize to pacify'. 12As Rempe observes, this approach was intended to help 'security forces overcome 'the traditional suspicion of the military held by the people in violent regions', improving intelligence and support for internal security operations"'. 13Kalyvas terms this support defection, where one actor seeks to 'maximize the support they receive from that population and minimize the support that rival groups receive from the same population' through a variety of instruments, not all violent. 14ronically, as Gutierrez argues, this attempt to win over communities in Colombia did not go as planned, as civilian institutions often became so close to the FARC that 'even the state' knew about the relationship. 15This raises questions about the efficacy of this locally-geared Hearts and Minds approach.For example, in neighbouring Peru the Fujimori government created the FONCODES programme to empower local communities through infrastructure construction to improve relations with the state and prevent the further 'insertion' of insurgent groups. 16However, Palmer observes that FONCODES only resulted in temporary empowerment during the project's construction and only superficial improvement in local-government relations. 17Sexton and Zurcher's recent examination of locally-geared infrastructure projects in Afghanistan also demonstrates that these projects had little effect on the presence of armed groups, and that robust consultation with local communities only prevented negative attitudes towards the state, rather than improving government-community relations. 18herefore if the objectives of 'winning Hearts and Minds' appear largely unsuccessful, why would project workers continue to appeal to local communities during infrastructure construction?To answer this, I turn to the literature on humanitarian work in conflict areas.Civilians may choose a myriad of ways to react to aid projects, either by actively opposing and threatening the project, tolerating it, or promoting and acting on the project's behalf. 19Breslawski examines the third option, where aid workers access difficult territories in conflict areas by 'winning over' the community so that they intervene with armed groups on behalf of the project. 20The strategy appears anecdotally in aid work done in Somalia, Afghanistan, and Syria, 21 yet this negotiated dynamic is still underexplored.As Breslawski argues, 'community acceptance has received scant attention in the academic world, despite the fact that community acceptance has emerged as one of the strongest factors in determining humanitarian security and access'. 22any project workers in Colombia view the local communities as more important to their security in conflict areas than the military, even if their governmental roles are more charged than 'neutral' humanitarian organisations. 23A senior official in Bogota explained that 'it is not recommended [to enter with the military] . . .Because when they leave, things will be worse.And the [armed] groups will then understand the project as a military objective'. 24An engineer working in another municipality of Nariño also argued that 'the army did not provide security, the community did'. 25This approach often put project workers at odds with other branches of government, as the Colombian government since Uribe in the 2000s has prioritised military protection of infrastructure construction. 26herefore project workers often build infrastructure in the midst of rapidly changing local orders between the military and armed groups.Drawing from strategies of 'community acceptance' in the humanitarian literature, but recognizing the intersection of goods delivery and counterinsurgency, I examine how community support becomes most valuable when the local order resembles what Staniland refers to as a 'tacit coexistence'.This dynamic occurs when neither the state nor armed groups is confident in securing the territory militarily. 27Therefore, the role of civilian institutions becomes crucial in negotiating project workers' safety. 28 explore the role of community support in infrastructure construction during conflict, I took a grounded approach in a two-phase data collection in Colombia. 29Colombia has a comparatively low-level international intervention in its statebuilding processes. 30This means that project workers are usually directly associated with the state, such as the three engineers I first interviewed at a government agency in Bogota.I used these interviews to identify how project workers valued community support.I then conducted indepth fieldwork on different infrastructure projects in Nariño with 25 interviews with local leaders, project workers, and international observers in early 2019.

Case selection and research design
Bearing in mind the 'ethical imperative of research' in conflict zones, 31 I paid particular attention to the relative position, perspective, and power of the participants.In my own positionality as a native non-Colombian Spanish speaker, project workers and community leaders saw me as a neutral actor.However, as a white researcher from a privileged university in the global North, I had an ability to leave when situations were complex that not all of my participants possessed.Indeed, security concerns prevented me from visiting Leiva at the time, and I had to rely on interviews with project workers in the city of Pasto.Therefore I took every precaution to ensure that the data was collected with the consent and complete anonymity of all participants to ensure their contribution did not generate any problems during or after the fieldwork process. 32ariño has been targeted by the state for infrastructure construction and seen a growing presence of multiple armed groups since the early 1990s, with the region largely ignored by both the central state and armed actors prior to that decade.Military efforts to dislodge the guerrillas and coca production were introduced in the north-east Cordillera region, including the municipality of Leiva by the late 1990s. 33The central municipality of Barbacoas faced this military crackdown later in the mid-2000s as cultivation shifted from the neighbouring department of Putumayo. 34I selected infrastructure projects in these two municipalities to unpack the community support strategy used by project workers despite military presence.
To demonstrate the broader scope of this strategy beyond Nariño, I also examine the construction of the River Amoyá hydroelectric dam in the department of Tolima as a shadow case.Dams are usually outside the scope of 'Hearts and Minds' cases or humanitarian work as they are often heavily opposed by local communities, particularly in Colombia. 35The similarities across these cases demonstrates that community agency is unlikely to depend solely on the type of infrastructure project as much as project workers' strategy and the conditions of the local order.
Path through the mountains: schools and roads in Leiva (1995-2015) 36 In 1995, the departmental government of Nariño set out to build tertiary roads, schools, and other local infrastructure projects in the mountainous municipality of Leiva.Aside from treacherous conditions and poor existing roads, the engineers and project managers sent from Pasto also had to contend with the FARC's 29 th Front, established several years earlier.The FARC's arrival to the communities had been marked by declarations of revolutionary intent and criticisms of the state for failing to invest in the region. 37et the arrival in 1995 of engineers and project managers undermined the FARC's position.The state entered to build highly-desired roads in the mountainous municipality.In response the FARC appeared willing to allow the project, even co-opting elements of it by attending all the meetings with project engineers and workers, establishing when and where project workers could move, and even organising community labour to assist. 38his arrangement quickly fell apart as the military began a concerted incursion into the territory for the first time in the late 90s, and by 1999 were joined by the paramilitary forces of the Southern Liberation Block (Bloque Libertadores del Sur). 39The paramilitary group eventually demobilised in 2005, but many combatants regrouped under the neoparamilitary banner New Generation Self-Defence Organisation (OANG).Between 2006 and 2008 different groups wrestled over control of the drug trade, leading to multiple clashes, violence against civilians accused of acting as informants, and a dizzying arrays of alliances and enmities amongst the FARC, smaller factions of the ELN, and OANG. 40Fighting continued until 2008 and the infrastructure projects in Leiva were left to languish for almost eight years. 41As one of the head engineers noted, 'it was an impressive violence and we could not work.We tried to coordinate with the local organisations to defend ourselves for the project.But it was six or seven years of that'. 42y 2008 the combination of violence and community protests in Leiva led to a reassessment of territorial strategy by the Colombian state. 43ommunities, tired of the violence, coca crop fumigation, and lack of infrastructure investment, took to setting up blockades on the Panamericana road connecting the capital of Nariño with the rest of the country in 2008, grinding the economy of the department to a stop.After negotiations with local JACs and mayors, the government launched in 2009 a development initiative called Si Se Puede (Yes we can) to encourage rural development, including reinstating the local infrastructure projects. 44These new efforts sent project workers back to Leiva.
This time, instead of entering a relatively peaceful territory under FARC control, the project workers had to navigate a tense coexistence between FARC and state forces.Informal boundaries emerged where military forces stayed largely in the municipal centre of Leiva, while the FARC remained present in the corregimientos -sub-municipal administrative units organised around a JAC -of Nariño, Villa, and el Palmar.One official from the Department of Infrastructure of Nariño sent to check on the Si Se Puede project observed in 2012: There was a [state] military checkpoint on the road.They said, if you go past here, we are no longer responsible for your security.I then had to explain [to the guerrillas] that I was there for the educational project so that they let us in. 45he military could not guarantee the project workers' safety: instead, they had to rely on community support for the project. 46This approach also applied to other groups in the municipality.By 2008 the OANG was actively clashing with government forces. 47The engineer noted that by the time they went back, the OANG had begun to build a: Similar relationship with community councils as these councils had with the guerrillas . . .some went with the guerrillas and some with the paramilitaries . . .[We went in] always with the support of the local leaders, like the presidents of JACs.We would also work with the mayor's office.Sometimes we could not travel or move around certain times.We would communicate with local leaders and they would advise us.With other commanders it was more difficult.But there was no trouble working for these communities.The armed groups would not oppose these projects because they would also benefit, and it was a good relationship with the community. 48e JAC were considered integral to 'guaranteeing the entry into the territory' for the technical teams in carrying out the projects, 49 which indicated even on official documents how project workers had to navigate the ongoing presence of the FARC or the OANG in the territory to gain 'entry'. 50his strategy by project workers -and the community's ability to mediate with the FARC and OANG -provided communities with the desired schools and roads. 51However, the armed groups also used the community to dictate how the project was built.The head engineer noted that whenever there 'were some misunderstandings [with the FARC]' about their intentions in the territory, they were informed by members of the community that they 'had to leave'. 52Another project worker observed that they could work in the territory as long as they were 'prudential' and depended on the relationship between the guerrilla and the community. 53The fact that the JACs were considered heavily infiltrated by armed groups 54 was leveraged by project workers as a way to ensure their security, and by the FARC as an effective conduit for relaying their conditions of entry and to tell the engineers when they had to leave. 55hanges in the project could affect this mediated agreement.The head engineer, who continued to work on occasional projects in Leiva after Si Se Puede concluded, lamented that new project managers and government officials were jeopardizing the progress made: 'New [project managers] bring external contractors where there are issues with corruption.There is more tension between the presidents of the JAC and the mayors, but it is saveable.We just have to have a reunion every month'. 56e engineer viewed the relationship to the JAC presidents as crucial to continued safe entry into the territory, even if less experienced colleagues did not.This reflects not only how the project workers' security strategy can fortify civilian leadership with local authorities such as the mayor, but also the complicated and informal nature of such a strategy.
Connections in the jungle: the Barbacoas-Junín road (2012-2018) 57 Project workers deployed a similar strategy of community support in the construction of a secondary road in Barbacoas.An unpaved path was the only main connection between the municipality to the rest of the department, whose population largely depended on a network of river systems for mobility.The path was also the focal point for the presence of multiple armed actors that sought to control its scant 55 kilometres.Neoparamilitary groups were concentrated at the initial entry point in the village of Junín, with the Rastrojos established since 2006 and then slowly pushed out by the Clan del Golfo in 2014. 58Military operations in the early 2000s in the area had been concentrated on the road and rarely penetrated the jungle.As a result, the ELN, with the Martyrs of Barbacoas were concentrated from the mid-point stop at Buenavista while the FARC's Daniel Aldana Front -which controlled the territory north of Barbacoas -also maintained a strong presence on the northern part of the road and still patrolled further south.
Promises by the Nariño regional government to pave and improve the road to viable conditions failed to materialise, often succumbing to lack of political will and corruption. 59Efforts were also met by extortive demands by the Rastrojos, as well as the ELN and FARC, despite support for the project by the local communities. 60As one local afro-Colombian leader noted, 'the FARC was charging vacuna [extortion] because they did not care about having the road well paved . . .so the "vacunas" were very present'. 61his dynamic changed when a group of local women mobilised in 2010, leading to national and international coverage. 62The national government eventually took action and by 2012 construction officially recommended with a battalion of military engineers and a protective military unit deployed in early 2013 63 against the wishes of the community. 64As one indigenous Awa leader complained, 'they did not warn us that a military battalion would arrive . . .there were no protocols of relation with the community . . .we were not okay with this permanent presence of the armed forces'. 65espite some initial clashes it became evident that this military strategy was not sufficient to dislodge armed groups from the territory.On a visit to Barbacoas near the end of 2013, a UN official noted that the military and the FARC has established an uneasy truce, even appearing to agree to control different sections of the road, with the army telling him that they could not ensure his safety after a certain point, much like the army officials in Leiva.Instead, the UN official noted that 'having a relationship with a social leader would help'. 66A later report by the Office of the United Nations High Commissioner for Human Rights also indicated that the military presence on the Junín-Barbacoas road did not devolve into clashes with armed groups, but had rather resulted in a 'tense calm', where the army controlled the road to the south near Junín, while the FARC and ELN continued to have presence in the north. 67he regional government began to prioritise the community as their entry point into the territory.In 2014 the Nariño government created a Special Unit for the Junín-Barbacoas road.The Unit was intended to put the road management in civilian hands, who did not see the military as a source of security: '[the military battalion of engineers] has affected our project because there is always tension when the army arrives.It ends in shootings . . .'. 68 The social officer understood that his role was to win community support to secure the project from armed groups: 'Here there has to be a social bridge.The engineers came and [the armed groups] thought they were here to do an investigation.They would threaten them.They know me and I have more access.You have to ask permission of the people.They would then ask for permission from the armed group that I could be there'. 69her efforts by the Unit to win back community support including censuring one of the contracting companies and removed it from the project for mismanaging their construction and presenting falsified documentation on its completion after public outcry in Barbacoas. 70The Awa also benefited from locally hired labour -much like project managers in Leiva had tried to do 71 that in turn incentivised them to support the project with the armed groups: The contractors came in and even used local labour . . .We then made it clear to everyone that we all would benefit from such a road.The armed groups understood our message in the territory.Yes, there were attacks against the military but that was just between the groups. 72ongside hiring local indigenous labour, government officials also promised to provide smaller public goods and services that the Awa had requested.One official viewed this as a major reason for why the project did not face 'any more significant inconveniences', 73 including FARC efforts to mine the road.The governor of the Awa cabildo intervened: 'we all protested that that was not alright.We are very interested that the road project advance for our benefit.The patients were not reaching the hospitals in leaving for Junín.It was a very high priority.The FARC stopped mining'. 74The consejo comunitario of Nueva Esperanza also interceded with efforts by armed groups to extort the project workers.When the Clan del Golfo began extorting the Junín-Barbacoas contractors in 2014, a local leader explained: okay well that is alright . . .there is a strong pressure from the communities who say we want this [infrastructure] here.That makes them be aware and have a conscience and then they ask for less money because if not the project is just abandoned there.There is not opposition from the guerrillas, not even from paramilitaries, because the communities come out'. 75r project workers, strengthening a positive relationship with the community meant its officials and project managers could rely on the community to secure their presence by interceding on their behalf with armed groups.If the community were to provide project workers with military intelligence it would jeopardise civilians own security -which the state was unable to guarantee even with the military -and the security of the project.As the official for the Junín-Barbacoas project explained, the project construction was always aware that it was sharing space with armed groups despite the presence of the Colombian military: There were unofficial rules.The workers are from there and they tell us when to go in.We can only go in until 10pm.The local leaders themselves tell us these rules.They will accompany us and only for half an hour, and then for example we could not work anymore.The technicians were also from the area and they would anticipate the risks.Today yes, tomorrow no, only go up until that point, don't get too far from us. 76local inhabitant of Barbacoas even explained that the FARC agreed not to interfere with the road project but required that the workers refrain from using a camera; 'a very simple rule' to keep the peace.77 It ensured that the community could have a road, the project workers could secure their presence and continue working on the project, and the FARC reduced the chances of intelligence being put in the hands of Colombian military.Armed groups, like the FARC in Leiva, used the community as an intermediary to impose guidelines on project workers that prevented the state from gaining a military advantage.
In the heartland of the FARC: Amoyá hydroelectic (2008-2013) 78 In 1999 the utilities company Generadora Union was awarded the rights to build a hydroelectric dam on the River Amoyá in the municipality of Chaparral located in the south-central department of Tolima.The proposed construction was particularly ambitious given the presence of the 21 st Front of the FARC.The mountainous region was the oldest stronghold of the group, where in 1964 a small group of rebels fought the Colombian army in the battle of Marquetalia, the mythologised origin story for the leftist guerrillas. 79he Colombian government's initial foray into the territory was to send aid money to local families (Plan Familias Guardabosques -PFGB) in 2003, purportedly for coca substitution but also to assess viability for a hydroelectric project. 80Unlike the 29 th Front in Leiva, the FARC in Chaparral prohibited local villagers from accepting governmental aid, seeing it as a threat to their territorial control. 81Faced with the looming risk of FARC backlash, Generadora Union tried to deploy non-military strategies to safely enter the territory in 2005.Like the project managers and officials in Leiva and Barbacoas, Generadora Union considered the relationship between the FARC's 21 st Front and the local community in Amoyá an asset to securing the project.An environmental impact report in 2003 by Generadora Union and its partner ISAGEN (another utilities company) observed that the close relations between the FARC and the local community presented an opportunity to mitigate risk: 'The presence of the armed group generated a certain level of "security and protection" in the region, as there were no longer any fights, thefts, or family abuse.Although this situation could be considered a threat to the Project, we could also think that the coordination and co-management that are consistent with the principles of [local] participation and agreement between all the different actors involved in the processes in the region, could generate the adequate conditions to ensure that the project is carried out in the best possible terms.It is important to note and highlight the need to carry out a work of concertation and sensitization of the Project that involves the various social actors of the region, the ignorance of any of the parties may be misunderstood and generate an inappropriate environment for the development of the Project [emphasis own]'. 82wever, this strategy was complicated since the dam project itself was not desired by local civilians.The officials and project managers of Generadora approached the local JACs to offer them additional concessions in road constructions and small schools that would help develop the local villages in exchange for serving as part of the security protocol in the project's construction and avoiding the army.The project workers viewed the community as a necessary conduit to reaching the FARC's approval for the dam, since 'by obligation, [the community] had to go consult the guerrillas'. 83et initial efforts to establish a good working relationship with the JAC struggled to gain traction due to opposition by the FARC, who had begun to 'jump at shadows' since the implementation of the PFGB, accusing civilians of passing information to the military if they cooperated with the project. 84eneradora Union's community support strategy was finally completely disrupted by a change in management when ISAGEN took full control in 2006.Unlike Generadora Union, ISAGEN insisted on accompaniment by the military, 85 despite project workers within the same company preferring to avoid it. 86his lead to active combat between the FARC and the state between 2006 and 2008, with direct clashes rising from 85 events per year in 2005 to nearly 180 in 2007, 87 as well as increasing attacks by FARC and military against civilians. 88The FARC were aggressive against the project in this period, considering it a 'trap for the Army to consolidate territorial control', and continuing to victimise civilians as potential informants (sapos) if they appeared to support the project. 89etween 2008 and 2010 the intensity of the battles began to wane and both actors established informal boundaries, with the FARC in mountains of Las Hermosas while the military established a base near the village of Chaparral in the valley. 90The engineers and builders still needed assurances for their safety as their work took them sometimes up into the mountains.ISAGEN's social team had been trying to work around the difficulties generated by the military presence in the area by convincing civilians that the presence of project workers was not designed to communicate intelligence to the army, even actively avoiding military accompaniment where possible. 91owever, local communities were now hostile to the project, worried about flooding and angry about the increased military presence and fighting in the territory, leading protests to emerge under the banner ASOHERMOSAS between 2007 and 2009. 92By 2010 ISAGEN refocused efforts to win back the community, when the director of ISAGEN agreed to use a Transparency Table -established in 2007 but virtually abandoned since -as a platform to improve communication between the company, the military, and the local community. 93The Table brought civilian leaders from JACs and ASOHERMOSAS to voice their grievances.ISAGEN also brought military officials to the Table to make promises and assurances to civilian leaders that they would investigate cases of victimisation and respect local communities in the territory. 94ith a tense co-existence established in the local order, and ISAGEN attempting to win over the community through the Transparency Table, the FARC became more wary of losing community support; 'the guerrillas did not want to antagonise the community, so they cautiously accepted the implementation of the project'. 95One local leader even explained that the shift in FARC attitude towards the dam was a reflection of the relationship between the FARC and the community: 'if [the community] wanted the company to come in, the guerrillas would accommodate that.But if they wanted to oppose [the dam], the comrades would be there as well'. 96Much like the leaders in Barbacoas, civilians explicitly referenced their own agency in affecting the behaviour of the armed group.This appeared to be true to an extent.With community support, project workers were able to enter the more remote parts of the mountain as long as -as one ex-FARC combatant explained -the military stayed 'in the valley'. 97This was considered a success by ISAGEN leadership, who viewed improved relation with civilians through the Transparency Table as key to the 'continuation of the construction at the most critical moments of the project'. 98

Infrastructure and the limitations of local agency
Political actors often frame infrastructure construction in conflict areas as proof of the state's successful delivery of goods in the face of armed opposition, 99 despite the messy realities that project workers navigate on the ground.President Santos declared in 2013 that the River Amoyá dam in Chaparral had been 'built through fire and blood' and 'against the will of the FARC'. 100In Barbacoas, Major Sánchez Peralta of the Colombian army assured that engineers and project workers 'depended on our trained troops and special forces, including anti-explosive unites to counter criminal actions and keep not just our engineers safe during their work but also our civilian population, the primary motive for our work'. 101In Leiva, the language was less charged but still positioned as a political win; the project heralded, according to the Governor of Nariño, a new step towards peace and cultures of legality. 102State actors positioned the constructions as military victories and statebuilding successes, 103 despite the fact that the projects had ultimately been constructed with the tacit -if begrudging -approval of armed groups.
Inverting this counterinsurgent framing, project workers saw community support as crucial to the successful construction of infrastructure, rather than the other way around.In all three cases, some degree of uneasy coexistence with armed groups -pressurized by the close presence of state military forces -generated the space for civilians to take a greater role in negotiating the safety of the infrastructure construction.Therefore project workers went beyond the project itself, relying on local labour, smaller infrastructure projects, and increased transparency to win over the community.Their objective was not to separate the civilians from the armed group, but rather rely on their closeness.
This strategy reflects the fact that relationships between armed groups and local civilians can cut both ways, allowing civilians to influence armed groups as well.Arjona argues that strong local institutions are able to negotiate with armed imposition, 104 and Rubin observes that groups may even seek civilian institutions because it can better support rebel groups with information gathering and resource mobilization, even if this agency improves the bargaining ability of civilians. 105The creation of these local civilian institutions also improve collective action capacity to pressure the state in providing desired goods or services, 106 especially in generating media attention and agenda setting, 107 as shown in all three cases.However, civilian agency is still limited.ISAGEN designated the Transparency Table in Amoyá as a relationship-building platform to enable the safe construction of the project.ISAGEN continued to host discussions at the Table between 2014 and 2017, promising improvement of roadworks, the building of schools, better energy provision, and assurances of measures taken to mitigate the environmental damage caused by the dam.Yet by 2018 none of them had been implemented.With the dam finished and the FARC officially demobilised from the area, the interest by ISAGEN in continuing to use the Table had faded, leaving civilian leaders frustrated. 108hese cases also challenge the supposition that armed groups always act on behalf of the wishes of the community.Multiple actors across the three cases, from project workers to civilians themselves, echoed the expectations in humanitarian circles that armed groups acquiesce to the wishes to the people.One analyst in Bogota even argued that the FARC always allowed projects popular with communities. 109This may be the case in some circumstances, as the FARC demonstrated in Leiva before the arrival of the military.Yet local support is not always foremost in the minds of armed groups, especially in the face of military pressure. 110Where all-out confrontation emerged, the projects were often abandoned or became military targets.Yet where there was no military presence at all, armed groups could reject any governmental projects entirely -as in Chaparral -or seek to extort it beyond viability -as in Barbacoas, despite the community's wishes.
These local dynamics constrain the strategies of project workers and civilian agency.As Worral posits, armed groups not only respond to external orders (which dictate military incursions) but also how this affects their relationship with other societal agents and structures within their locality, 111 such as the relationship they have with JACs, consejos comunitarios, and cabildos.Similar to Staniland's tacit coexistence, Idler argues that pacific coexistence between armed groups and the military generates a greater incentive for opposing actors to 'strive for the social recognition of the local population in order not to be out-ruled by the other'. 112Therefore armed groups may have similar incentives to project workers in winning community support when they coexist with military forces: their willingness to listen to civil leaders is therefore less a reflection of an ideological position and more a response to the immediate security concerns in the local order.
The agency and limitations of civilians mediating between project workers and armed groups deserves further exploration.This paper has demonstrated that the logic and decision of project workers to seek community support functions as a temporary security strategy rather than a long-term statebuilding project that reflects the fluctuating and complex nature of local orders. 113uture research could build on this to better understand why and how civilians and armed groups play their part in this dynamic.Assumptions about a positive relationship between civilians and armed groups overlooks key variations in when and why armed groups are more susceptible to community pressure.The type of government leading the project may influence armed group response, just as the group's own internal order may shape its interactions with local social actors and external players. 114Indeed, this is a concern in the humanitarian sector as well, since there are cases where aid workers have endangered themselves by assuming there were linkages between communities and armed groups where there were not. 115hese cases represent only a small sample of a wider universe where governments build infrastructure in areas controlled by armed groups.However, here I have unpacked the confrontations, miscommunications, and negotiations that can underpin public goods delivery in conflict areas.I show how actors -even within government -may disagree on strategy during infrastructure construction.I explore how the fragility of a 'tacit coexistence' keeps the ground shifting underfoot and project workers must constantly renew their appeal to local communities to gain safe passage.Finally I demonstrate both the opportunity these projects provide for local agency to assert itself and the limitations generated by local orders and finite construction.Identifying community support as a pragmatic response to complex local orders rather than a blunt measure of counterinsurgency success allows researchers and policy-makers to better assess public goods delivery in contexts where armed groups, project workers, military forces, and civilian institutions must unhappily co-exist.

Figure A4 .
Figure A4.Amoyá Hermosas dam in the municipality of Chaparral.