“This Is Shared Work:” Negotiating Boundaries in a Social Service Intermediary Organization

This article discusses the results fromour fieldwork at a social service intermediary organizationworking to reform criminal justice institutions in a large city in the American South. Our findings focus on organizational staff’s relationships with information and communication technologies (ICTs), both in the course of their daily work of delivering care work to vulnerable participants, as well as the project’s broader political goals to reduce recidivism and repair community relationships with local police. The group needed to distinguish and negotiate the various—and often competing—needs and commitments of the civic actors involved. As on-site researchers, we were asked to design and deploy digital tools to support the organization in exchange for conducting research on organizational uses of technology. This work draws from our time with the group to ask: how might community-based researchers revisit and realign our research methods to better respond to the changing needs and practices of a research site? Our observations identified three recurring technological concerns expressed by staff that pointed to competing agendas and needs within the organization, specifically across different levels of scale: operational, proximal, and temporal. We then discuss these patterns around broader organizational concerns to reflect on how they impacted our own research methods and commitments. Finally, we reflect on the limitations of participatory methods in issue-oriented organizations that do progressive work across multiple scales and agendas.

Growing trends in the public sector point to the work of public, private, and municipal actors become more entangled to combine increasingly scarce resources to issues of public concern (Dombrowski et al., 2013;Erete et al., 2016). As boundaries blur within public sector work, scholarship shows this introduces additional challenges as communities must contend with competing agendas and efforts (DiSalvo, Lukens, Lodato, Jenkins, & Kim, 2014;Voida, Dombrowski, Hayes, & Mazmanian, 2014), funding sources and requirements (Goecks et al., 2008), and differing expectations around labor and management (Harmon, Bopp, & Voida, 2017;Voida et al., 2014). When information and communication technologies (ICTs) are deployed to support this intermediary work, they are woven into the intricate fabric of human and non-human actors that all participate in and negotiate the various politics, ethics, and power struggles involved in community efforts around different issues.
As community-based researchers, we too are part of this fabric and our research efforts necessarily about the socio-technical networks we purport to study (Lodato & DiSalvo, 2018). This article focuses its analysis on community-based research efforts that use more interventionist design methods, such as participatory design (Björgvinsson, Ehn, & Hillgren, 2010) or action research (Hayes, 2011). Our interests lie in the boundaries between researcher and fieldsite and how these boundaries are established and negotiated through the design and deployment of ICTs. Specifically, we ask how community-based researchers might revisit and realign our research methods to better respond to the changing needs and practices of a research site? Our role as inhouse technical support and organizational consultants provided unique insight into how the staff enacted their various responsibilities and the myriad ways that digital tools were expected to support their work. In order to push for social change in the complex and interwoven criminal justice ecosystem of an urban city in the American South, the organization had to necessarily bridge the agendas, ideologies, and needs of involved actors, including vulnerable participants, local residents, police officers, social service agencies, legal actors, and both incoming and outgoing political representatives. We hope to contribute to growing conversations in the academy around how and when to intervene through research, especially when working in complex public contexts.

Intermediary Work: From Grassroots to Institutions
We draw from and build on existing bodies of work that study intermediary organizations in public sector work. More traditional intermediary work in these environ-ments connect disparate resources or organizations to build greater capacity for action (Asad & Le Dantec, 2015;Parker et al., 2012). More recent research in this space focuses on various strategies and common practices used across organizations (Hansen, Koepfler, Jaeger, Bertot, & Viselli, 2014;Lodato & DiSalvo, 2018), which often include myriad actors, from city and municipal employees to volunteer activists and institutional actors. These works reveal common technological challenges across the political spectrum of public work, such as outdated computers or incomplete datasets. Scholarship points to the ways in which intermediary work does not strictly address technological concerns, but looks at how ICTs mediate and participate in larger and more complex social issues, from economic security (Vyas & Dillahunt, 2017) to cultural identities (Wyche & Grinter, 2012) to urban crime (Erete et al., 2016). A common challenge specific to intermediary organizations is the ways in which competing and often contradictory values and agendas must co-exist and the work must be malleable enough to address these complexities (Le Dantec & Edwards, 2008;Stoll et al., 2010). As Voida et al. (2014) argue, it is difficult enough to put into values into practice when everyone in an organization shares those same commitments, much less when a project spans different categories of civic actors, each of whom have their own set of values, commitments, and responsibilities, which the organizational staff must contend with through their work.
Recent work has paid closer attention to the ways in which non-human actors also participate in and facilitate intermediary work with sensors, processors, and data exerting as much agency as human actors to address social issues (Forlano, 2016;Harmon et al., 2017). Here, we use "intermediary" in a different sense to focus on role of ICTs in public sector work. Literature discusses how sensors and data are influential actors in citizen sensing projects (Erete et al., 2016) and social media platforms operate in tandem with neighbors and residents to negotiate shared concerns around local, small-scale community engagement (Asad & Le Dantec, 2015;Hansen et al., 2014). Another important perspective to consider is self-reflexive: our participation as researchers is also a kind of intermediary work as we exert influence over research sites regardless of our methodology (Holmer, DiSalvo, Sengers, & Lodato, 2015;Khovanskaya, Sengers, Mazmanian, & Darrah, 2017).

Interventionist Work: On Friction and Research
The researcher-as-intermediary is not a new concern to community-based research: more interventionist methods have actively incorporated this positioning into its modes of inquiry, such as participatory design (Björgvinsson et al., 2010;Holmer et al., 2015) action research (Hayes, 2011), and more design-based methods like service design (Stickdorn, 2011) and design ethnography (Khovanskaya et al., 2017). These different traditions call for varying degrees of intervention with a re-search community and aim for different outcomes from the work. Recent literature suggests a move towards deeper and more involved interventions. Some call for more rigorous interrogations of oppressive sociopolitical institutions that touch our fieldsites (Dimond et al., 2013;Le Dantec & Edwards, 2008) while others ask us to incorporate more socially just and ethical approaches to our research (Dombrowski, Harmon, & Fox, 2016;Fox et al., 2016). Korn and Voida (2015) present a framework to look at the spectrum of interventions possible through research and join other researchers in active calls for facilitating more "friction," or agonism, through our various methods, be they ethnographic, design-based, or quantitative (Korn & Voida, 2015). It is also important to interrogate the specific site of friction: in some cases, the more ethical choice may be to use research to maintain existing practices to resist friction, like in instances where said friction hinders just or ethical outcomes (e.g., policies that makes it more difficult for participants to receive social services) (Håkansson & Sengers, 2014;Khovanskaya et al., 2017).
Across these different scales and strategies, it is imperative to be introspective and interrogate what forms of power and oppression exist within our research sites and how our participation may impact these power dynamics, intentionally or otherwise . Below, we detail the history of the fieldsite to highlight the ways in which its work is both intermediary and interventionist, and brings to the surface different motivations, agendas, and concerns.

Site
The lead author conducted ethnographic fieldwork with a criminal justice organization in a major city in the American South whose goal was to 'divert' vulnerable populations to social services instead of arresting them. These populations were vulnerable people who are homeless and/or facing extreme poverty, struggling with substance abuse, and/or experiencing mental health issues. As a result, they couldn't be reached by more traditional social service programming, and were thus more susceptible to repeated interactions with law enforcement. The organization was created to address these systemic breakdowns and was the result of multiple years of grassroots campaigns and policy work. These campaigns were originally led by a sex workers advocacy group who worked with local police over years to establish non-punitive ordinances in areas known for high sex worker activity. The campaign eventually grew to include other survival/quality of life crimes, e.g., loitering, panhandling, or possession of forbidden substances, and eventually led to the formal creation of the organization.
The organization itself is an intermediary between a wealth of affected actors, including police officers, attorneys, judges, community-based organizations, activists, advocates, policymakers, social workers, and the organization's program participants, who were often in extreme poverty and dealing with mental health and/or substance abuse concerns. The organization's main work focused on 'diversions,' which are the mechanisms by which trained police officers choose to offer social services to a participant instead of arresting them. Once participants have consented to enter the program, they directly interface with the organization's social workers who connect participants to social service providers who have been vetted to be respectful of participant's life experiences, including trauma-informed social services, non-punitive program requirements (i.e., re-arrests or relapses will not put participants at risk of 'getting kicked out' the program), and working with trans-inclusive and non-religious partners. The organization also mediates relationships between the participant and various program partners (e.g., social service providers, legal advocates, participants' attorneys).
During our 8-month tenure with the organization, they were operating as a pilot program, which meant that there was immense pressure-both within the organization and from program partners-to show the efficacy of the program and its goals in order to secure additional funding and establish the pilot as official city programming. Given their focus on criminal justice reform, part of the challenge of the work was to capture successes in ways that were legible to bureaucrats and policymakers unfamiliar with the complexities of social work, much less as the unique demands of social work in this region of the country. While the complex sociopolitical context of the American South is beyond the scope of the article, we wish to acknowledge the fraught regional politics that have not historically supported social welfare and educational programs, leading to heavy racial segregation in the city, historical and institutionalized anti-Blackness (e.g., redlining), and the criminalization of Black communities. As a result, there are fraught relationships between many neighborhoods and criminal justice actors (e.g., police officers), as well as the state (e.g., service providers), which was particularly true of the geographic region where the pilot was being tested. The program formally launched (i.e., started diversions) in October 2017 and is open for diversions 4 days a week (typically late-to-overnight shifts when many quality of life crimes occur). As of the time of writing, there are 70 participants in the program.

Methods
The lead author conducted fieldwork with this organization for approximately eight months, from August 2017 to March 2018, spending an average of 6-20 hours a week in the office. She documented her observations through extensive field notes. Together with the second author, we analyzed our notes based on the fundamentals of qualitative data analysis (Miles & Huberman, 1994;Stoecker, 2012;Van Maanen, 2011), using induc-tive codes to analyze notes and iterate on our codes to organize our findings below.
The lead author's observations were based on her time with staff at the organizational office. At the start of her fieldwork, the organization had 6 full-time staff members: the executive director, the social work manager, two social workers, an operations manager, and an administrative assistant. The group experienced some flux, including firing two employees, hiring two part-time social workers, replacing a full-time social worker, and replacing the operations manager. Given the pressures of the pilot phase, the lead author completed tasks to support the operational needs of the organization, especially during staff shortages, which mainly included administrative tasks (e.g., answering phones, taking notes, ordering office supplies), logistics (e.g., directing partners to the office, coordinating deliveries) and supporting individual staff members with technological problems or emotional support.
It is important to make clear that we did not conduct any fieldwork with the field site's participants and only had minimal interaction with them as they occasionally visited the office to meet with the care staff. These interactions were limited to checking in (e.g., greeting them, showing them the wait area, letting care staff know they were present). Additionally, we did not study the diversions themselves, but rather the norms, expectations and practices of the organizational staff, as well as their daily information practices. These data practices varied, ranging from hybrid handwritten and typed forms and documents to using various digital channels to contact participants and partners to using social media to coordinate trainings and events. Staff were expected to complete a number of tasks daily, including documenting their progress, updating procedural and institutional protocols, and generating organizational reports, like meeting minutes, professional updates, and social work case notes. These tasks spanned a variety of digital tools, including but not limited to shared drives on the local area network, project management platforms, office suite software, and apps to access different databases and resources.
The lead author was given access to some of these data via the organizational shared drive, though permissions were set such that she could not access private or personal information (e.g., hiring documents, participant medical files). Broadly, she had access to the organization's meetings, events, and daily work practices, as well as direct access to the staff members themselves while they were in the office. The lead author was primarily consulted on technological questions and concerns (e.g., installing the correct drivers), she was also invited to participate in broader organizational decision-making processes and took on non-technological responsibilities. She attended staff and partner meetings, created internal documentation (e.g., inventory, communications plans), and developed digital and non-digital prototypes to deploy on-site. Some of the prototypes were longer term projects, such as the participant/partner database described in Section 4.2, which would be regularly used by social work staff and occasionally by external partners. Other prototypes were more focused on management or organizational practices, such as designing forms and templates.
The labor described above was offered at the onset of the fieldwork and accepted with the full consent of the staff (the lead author also explained her role as an embedded researcher to each new staff member as they joined the organization before asking their consent to observe and assist them). When we were initially invited to the organization to design and implement various digital tools, both the lead author and the staff members agreed that the researcher would also help support organizational stability and growth on a day-to-day basis as needed. These efforts were extended in the spirit of action research, an approach to community-based research in the discipline of human-computer interaction that "satisfies both the need for scientific rigor and promotion of sustainable social change" (Hayes, 2011). Practicing action research is to prioritize the needs of community partners, which in this case asked the lead author to deploy her skills with project management and IT to provide ad hoc administrative support and technology training as needed, which also provided the opportunity for her to engage in the research inquiries and interventions described above, which she did alongside the staff, rather than dictated or prescribed to them.

Findings
Below we share the lead author's observations on interactions between organizational staff and various digital artifacts in the office, both in the course of their daily work and also as situated within the broader goals of the pilot project. We outline three categories where expressions of conventional technological concerns (e.g., usability, security, privacy) reflected broader tensions in the group. These tensions spanned three levels of scale: operational scale, referring to concerns that were more urgent for the day-to-day operations of the group; proximal scale, concerns impacting the core staff members working in the office vs. other, more distant stakeholders; and temporal scale, negotiating actions to be taken in the more immediate present vs. in the future.

Operational Scale
At the start of our fieldwork, the executive director asked us to create digital systems to support operational work within the organization. These data were to be used to support both participant needs of the smaller scale, dayto-day social work, as well as the larger scale work of policy change and criminal justice reform. While some tasks could be completed by organizational staff alone, others needed input from partners and thus required more coordination, like contacting service providers to check the availability of resources or checking various court dates with different attorneys. We worked with staff to learn about their work practices and to arrive at consensus on which digital artifacts would best serve the group's needs, which revealed competing agendas for how these operational concerns should be addressed. Specifically, there were tensions between improving the existing, ad hoc assemblage of myriad digital artifacts or creating two customized digital infrastructures to streamline both internal and external tasks. Given the urgent nature of the work, these technical questions became a way to negotiate how to prioritize what kind of work got done: more administrative tasks or more diversions.
The existing digital tools were cobbled together by staff members based on their existing skillsets and familiarity with digital tools, though by their own admission was risky as participant information was being shared across multiple channels, such as email, text, and verbally over the phone. These practices were more reactive as they were typically addressing the urgent needs of participants. Alternatively, while the new digital systems would be more secure, they would be costly in other ways, such as costing money for additional resources. The new systems would require additional labor from staff for systems training and maintenance, whereas they were currently focusing their already constrained efforts on diversions. One instance of these operational tensions took place during a staff meeting to prepare for an upcoming partners meeting. Here, a social worker asked the team for advance for how to communicate the nuances of a roleplay scenario: SW: I don't want to go up there and say the wrong thing. Should I say (it this way)? Executive Director (ED): Well… SW: What about (a different way)? (silence) ED: This is shared work, we're not going to let you fail. SW: Okay, but what should I say?
When consulting the social worker after the meeting, she expressed frustration at the executive director being "not helpful" when she needed advice. She identified a disconnect between her ask for effective communication during a meeting and her receiving what she thought were irrelevant reminders of the organization's core principles of cooperation and consensus. Here we see how the tension between the material needs of the day-today and the 'big picture' can be malleable and contested in light of different operational priorities and practices.

Proximal Scale
The executive director eventually solicited the lead author to customize an existing product, Salesforce, as an all-encompassing system to coordinate both internal and external tasks. From this, we observed another set of tensions along the scale of proximity: while the decision to use Salesforce was justified as a compromise for the concerns described above, it introduced some ambiguity around who the tool was supposed to serve. Combining both internal and external tasks in a single artifact was meant to better facilitate staff members continuing their daily work with minimal training and IT setup. However, the artifact was also meant to be accessible by partners, such as attorneys and services providers, which introduced concerns about who the system should be built for-internal staff or external stakeholders.
In one meeting, team members discussed which specific technical features should be incorporated into the system. Specifically, the social work manager asked for text field input for one section of the new system so social workers could qualitatively describe the participants' goals in detail. She explained that she had seen participants in other organizations receive "overly prescriptive care" and she wanted to keep the focus on the participants' self-determined progress and self-reported experiences. She emphasized the importance of recording the quality of each participant's interaction with a social service provider, saying that how participants felt about delivered services were crucial to their wellbeing, as well. She told us about transphobic agencies she had previously worked with who delivered services to cis-gender participants, but trans participants reported completely different, traumatic and harmful experiences. By contrast, the executive director wanted to deploy predetermined sets of checkboxes and dropdown menus to facilitate the speed of data entry and to make data easier to interpret for external stakeholders. Here, the dissent over data entry formats reveals differing expectations for who would be the primary users for the digital system.
We observed how these technical discussions echoed this same proximal tension in other, nontechnical conversation: for example, in staff meetings, care workers in the group often described their approach as using "radical love" to "heal" people, stating reminders to "(diversion) calls (from participants) change everything." In the same meeting, the director described organizational success in very different terms, stating that "if they (the police department) feel like this is their program, we've done our job." Given the intermediary nature of the organization, we observed staff members often acknowledge the complex and nuance motivations driving the project, but it was these technological conversations that made material the different organizational missions and commitments.

Temporal Scale
The third tension we observed was concerned with temporality: decisions about different digital structures revealed difficulties in balancing more immediate concerns with needs that would impact the organization in the future, particularly beyond the pilot phase when it would need to scale up to become a more robust program. These tensions are not mutually exclusive: there were concerns about temporality as mentioned in Section 4.1 as more administrative work was not directly addressing immediate concerns in the organization, i.e., prioritizing data maintenance instead of participant health and wellbeing. These temporal concerns also came up in discussions of specific technical needs for Salesforce: again, conversations around a single data field pointed to differences in what work was more important. In this example, a meeting was held to identify what health-related data should be captured for the Salesforce database to be shared with external partners. On a technical level, staff wanted to balance a generalizable data set so they could better anticipate if certain resources or kinds of care were needed versus collected data that would be unique to each participant's specific goals. As one staff member put it, one challenge with data collection would be to "be creative with how we track certain progress goals, like showering everyday." This was a particularly important conversation to have as external partners needed to have access to certain kinds of participant data (e.g., demographics), but making accessible certain kinds of personal data might result in vulnerable participants facing discrimination, e.g., based on their sexuality or health status. In this meeting, there was contention over the inclusion of a checkbox to connote a participant's HIV status: by having this information available in the database, staff and service providers might be able to offer more appropriate care, but staff members were concerned for participants' dignity, at best having deeply personal data stored on corporate servers, and at worst, what were to happen to participants if there were to be a data breach. Staff discussed other kinds of data as being helpful in future work-for example, tracking instances of rearrests-but in this instance, the technical conversation reflected deeper concerns about present decisions that might open the door to different scales of impact in the future, specifically introducing more traumatic or irreparable kinds of harm to folks already experiencing multiple kinds of trauma.

Discussion
By the end of our fieldwork with the organization, the lead author ended up designing and building the Salesforce system solicited in Section 4.2, but she also adjusted her research practices to better reflect and respond to the tensions described in the above section. Ultimately, researchers and the organization mutually agreed to conclude our fieldwork as a service design collaboration rather than an action research project. As such, researchers provided prototypes, recommendations, best practices to the organization rather than engage in more collective and in-depth design processes to catalyze more impactful change within the organization. This decision was largely made because of concerns around temporal scale-fieldwork was taking longer than agreed upon but researchers had academic commit-ments that had to take priority, like teaching. Below, we reflect on operational and proximal tensions common to community-based research-that is, what kind of work can we commit to and to whom are we responsible?and discuss our responses to the different organizational needs we observed in light of our research commitments and priorities. We focus specifically on the kinds of interventions we tried to deploy as researchers, as well as the friction that was produced across different relationships in the research project.

From Social to Technical Prototyping
The organizational contradictions were helpful cues for us as researchers as they eventually became the boundaries our research collaborators drew around where we should and should not intervene on-site. The above contradictions can be seen as a kind of social prototyping (6) as the organization itself needed to figure out what their priorities were, in all their complexities. There is an extent to which the inconsistencies above are expected artifacts of a new organization still working through growing pains. Indeed, with early interventions from the lead researcher, the team iterated through kinds of practices to prototype within the organization what best worked for them, such as working through conflict as an entire team and in-person during staff meetings rather than through online communications.
By the end of the fieldwork, however, we decided as researchers to prioritize the completion of the technical artifact over building organizational capacity, and thus focused our efforts on different concerns around technical requirements. We did not feel like a focus on the technological was completely eschewing organizational concerns as the digital platforms served as shared artifacts between the site and the researchers through which broader concerns could be articulated and negotiated. Nascent organizational practices, concerns, and power dynamics were made material through discussions of the digital tool itself. As such, the lead author shifted her tactics: instead of facilitating dialogue and debate across team members and organizational practices, she facilitated conversations to focus on shared understandings (and misunderstandings) of the artifact. The digital tool was a way to articulate both technological needs and to give form to kinds of work that should or should not be done.
As mentioned above, we ultimately prioritized concerns of the temporal scale: because funding for the project was conditional, we had to adjust our research to the fact that there was more at stake than participants' wellbeing. These temporal concerns also directly impacted operational concerns: if efforts focused too much on the short-term (e.g., administrative tasks), the organization would not be able to sustain the diversion work to make claims about the longer-term goals (e.g., policy change, criminal justice reform), thus putting at risk the possibility of the project existing in the future.
It was through the lens of these operational concerns that the lead researcher was able to articulate changing boundaries to staff members: she would be able to make more lasting impact by taking a step back from administrative work and focusing on technological implementation as her time at the site was limited and there was more organizational capacity to complete administrative tasks as there was for deploying digital systems. It was important to hear and acknowledge each staff member's priorities-be it policy or care deliverybecause even though we could not address each of them through research, they were ways for us to identify how our work and impact at the fieldsite could best support the broader organizational work.

Micro-and Macro-Level Friction
We were able to gain insight into the boundaries around our work as researchers by observing the different perspectives on technological requirements and organizational priorities. We saw boundaries drawn across the team of who does what kinds of work: whereas the social work staff focused on participants' needs and delivering care, the ED was able to frame this on-the-ground work to external stakeholders by translating how the microlevel work moved forward the macro-level policy goals. By the time we had transitioned our research methods to more of a service design model, the challenge was to focus our efforts on how to best contribute to the organization with our remaining time and capacities. Here, our concern was around proximal concerns: on whom should we focus our support? As described above, individual staff members had differing commitments to stakeholders on the project, whether it was supporting individual participant needs or translating these efforts to external partners to influence policy or culture change.
As described above, much of our early fieldwork was spent (to mixed levels of success) designing and implementing digital tools as a way to address problems that existed outside ICTs as well, such as lack of organization and clashing communication and management styles. There was no team management software that would address the power imbalances that belied the working relationships between the executive director and staff. This is not to reduce the agency of the rest of the team: we observed staff continually contest power relations within the office and they were not unaware of the institutional complexities of their work. Yet, in spite of this, the team did not seem to substantially push back against the hierarchical model imposed by the director, which posed a challenge to us as action researchers, who were invested in and make claims to distributions of power, specifically in the workplace and for those who are underrepresented in decision-making processes.
In our observations of organizational practices, we observed how concerns across different levels of proximal scale-specifically the implicit pressure from city policymakers and the police force to lead a "success-ful" criminal justice reform pilot-posed challenges to the micro-level work of the social work team. Taking the perspective of the staff, we could interpret their micro-disagreements over field inputs and dropdown menus to prioritize the wellbeing of their participants over the perceived success of the pilot by external parties. Despite the friction that it introduced to their dayto-day work, researchers observed how staff members negotiated boundaries within their group-'picked their battles,' so to speak-so they could focus their efforts on their participants.
We took our cues from the social work team adjusted our research methods at the fieldsite. Our challenge was to maintain our commitment to socially just research  at a site that largely maintained status quo power relations. Rather than try and influence radical change within the organization-or even worse, influence change in the affiliate institutions, such as the police force-we instead did our best to align our efforts with and support the team's ongoing radical work of prioritizing their participants. Staff members took on additional administrative practices-such as maintaining shared calendars-so long as they did not interfere with their prioritizing participants ("calls change everything"). In turn, we acquiesced larger design decisions-such as building out a Salesforce database-and took seriously some of the more minute details as opportunities to better support the rest of the staff and their commitment to participant wellbeing. Over time, we learned to adapt our research methods to respect the complexities and contradictions of the site: we would not be able to co-design more collective organizational processes on the team, nor would we implement radical digital tools to challenge the larger scale, complex breakdowns of longstanding sociopolitical systems. We could, however, use text boxes in Salesforce instead of dropdown menus so that participants could share experiences with social workers and the social workers could, in turn, approach social service delivery with the dignity and respect they strive for.

Conclusion
Our fieldwork at a social service intermediary offered insight into the various challenges of doing work that is entangled in various public concerns, touches various stakeholders, and is responsible to different needs and project goals. The tensions of having to serve multiple agendas simultaneously-specifically across temporal, operational, and proximal scales-makes the staff's work even more complex and fraught. As researchers, we reflected on these organizational dynamics to try and develop strategies to build on and evolve our own work as intermediaries, particularly when our methodologies are interventionist by nature and added to the challenges our collaborators faced at the field site. By observing and respecting the various competing boundaries at the organization, we attempted to adjust our re-search practices to better support the priorities of the staff and the goals of their work. We encourage our peers to center the concerns and priorities of their research participants in their research, which will be invaluable for articulating what kinds of work are or are not appropriate for that collaboration at that time. Moreover, we strongly suggest that researchers learn to value friction, rather than avoid it: while it may be a source of discomfort at first, that friction is ultimately fruitful for participants to articulate what is important to them, for us to build more trusting and sustainable collaborative relationships, and so we can orient our research to be richer, better informed, and more impactful. We hope this work will provoke conversation and reflection across other researchers, academics, and designers who seek to use participatory methods with issue-oriented communities or organizations.