Collaboration by Mandate: How Shared Data Infrastructure Shapes Coordination and Control in U.S. Homelessness Services

This study argues that while federally mandated shared data systems like HMIS facilitate coordination and accountability in U.S. homelessness services, they simultaneously reinforce power asymmetries by shifting resource-constrained providers toward compliance roles and concentrating decision-making authority.

The Gist

Here is an explanation of the paper using simple language and creative analogies.

The Big Picture: A Forced Team-Up with a Shared Notebook

Imagine a group of different non-profit organizations (like soup kitchens, shelters, and job training centers) trying to help people experiencing homelessness. They are all working in the same city, but they don't always talk to each other.

To fix this, the government says, "You must work together." But they don't just say "talk more." They give everyone a mandatory shared digital notebook called HMIS (Homeless Management Information System).

The goal is great: If a person visits three different shelters, the system should know that, so the shelters can coordinate care instead of giving them the same advice three times.

However, this paper argues that while this shared notebook helps everyone see the same picture, it also acts like a strict referee that changes who holds the power.

---

The Two Faces of the Shared Notebook

The authors found that this system has two very different sides, like a coin:

1. The Good Side: The "Team Huddle"

When everyone uses the same notebook, it helps the team coordinate.

• The Analogy: Imagine a sports team where everyone wears the same jersey and uses the same playbook. If a player gets injured, the coach knows exactly what happened because the data is shared.
• In Reality: If a client moves from a shelter to a housing program, the new provider can see their history. This stops them from having to ask the same painful questions over and over. It helps the whole network see the "big picture" of how many people are homeless and where they need help.

2. The Bad Side: The "Strict Referee"

The paper argues that the rules for writing in this notebook are so strict that they create a power imbalance.

• The Analogy: Imagine a group of artists trying to paint a mural together. The government says, "You must all use the same specific shade of blue, and you can only paint in a 2x2 inch square."
  • The Big Artists (large agencies with money and tech experts) can figure out how to paint amazing pictures within those tiny squares. They control the narrative.
  • The Small Artists (smaller, underfunded shelters) are so busy just trying to fit their paint in the square that they can't paint anything creative. They just do what they are told to avoid getting in trouble.
• In Reality: The big agencies have the money to hire data experts who can analyze the numbers and tell the story. The small agencies just spend all their time filling out forms to prove they followed the rules. They lose their voice.

---

The Three Big Problems Found in the Study

The researchers interviewed experts and found three main ways this system gets messy:

1. The "Compliance Trap"

• The Metaphor: It's like a student who is so worried about following the teacher's formatting rules for an essay that they forget to actually write a good story.
• The Reality: Small shelters spend so much time making sure their data looks perfect for the government (to get paid) that they don't have time to use that data to actually improve their services. They become "data entry clerks" instead of "caregivers."

2. The "Trust Gap"

• The Metaphor: Imagine a patient is afraid to tell their doctor about a secret illness because they think the doctor might tell the police.
• The Reality: Many people experiencing homelessness are afraid that if they put their data in the system, the police or jail might see it. So, they hide information or refuse to sign up. This makes the data incomplete, like a map with huge holes in it.

3. The "Storytelling Mismatch"

• The Metaphor: The government looks at the data and sees a "failure" because a number is low. The local workers look at the same number and see a "success" because they know the context (e.g., "We didn't get people housed because there are no apartments available, not because we failed").
• The Reality: The people who collect the data (the small shelters) often can't explain why the numbers look the way they do. The people who read the data (the government) don't understand the messy reality on the ground. This leads to bad decisions.

---

The Conclusion: How to Fix the Team-Up

The paper suggests that we can't just throw more technology at the problem. We need to change how we use the "notebook."

• Don't just check boxes: Instead of using the system only to punish people for bad data, use it to help everyone learn.
• Teach the small artists: Give the smaller shelters the tools and training to understand the data, not just fill it out.
• Listen to the story: When looking at the numbers, we need to listen to the local workers who know the real story behind the stats.

In short: The shared data system is a powerful tool that can help everyone work together, but right now, it's designed in a way that makes the big players powerful and the small players just follow orders. To truly help people, we need to redesign the system so everyone has a voice in the conversation.

Technical Summary

Here is a detailed technical summary of the paper "Collaboration by Mandate: How Shared Data Infrastructure Shapes Coordination and Control in U.S. Homelessness Services."

1. Problem Statement

The paper addresses the complex dynamics of mandated collaboration in fragmented social policy domains, specifically U.S. homelessness services. While governments often mandate that organizations coordinate through laws and funding requirements, they increasingly rely on shared data infrastructures (like the Homeless Management Information System, or HMIS) to enforce this coordination.

The core problem is the tension between coordination and control:

• The Promise: Shared data systems are designed to enable interoperability, support coordinated care, and ensure accountability.
• The Reality: These systems often embed governance rules that centralize authority, reproduce power asymmetries, and shift participants from strategic collaborators to compliance-focused actors.
• The Gap: Prior research has focused on frontline data collection or algorithmic fairness, but there is limited understanding of how shared data infrastructures structure data work (collection, storage, analysis, use) and decision authority across an entire service network.

2. Methodology

The study employs a qualitative case study approach focusing on the FL-507 Central Florida Continuum of Care (CoC), a federally mandated network covering Orange, Osceola, and Seminole Counties.

• Data Collection: Six semi-structured, one-hour interviews were conducted with stakeholders managing and using HMIS in the region (August–September 2025).
  • Participants: Four staff from the coordinating lead agency and two from participating service providers.
  • Sampling: Snowball sampling starting with HMIS lead contacts.
• Analytical Framework: The researchers utilized a data value chain protocol to trace data movement through four stages:
  1. Collection: Recording/gathering data.
  2. Storage: Maintaining and securing records.
  3. Analysis: Processing data into indicators or models.
  4. Usage: Applying outputs to decisions and accountability.
• Analysis Technique: Inductive thematic analysis using open coding, double-coding by two researchers, and bottom-up affinity diagramming.

3. Key Results

The findings reveal that while standardization facilitates interoperability, it creates significant friction in collaborative data practices due to resource disparities and power imbalances.

A. Data Collection

• Compliance vs. Utility: Data collection is driven primarily by funding and reporting obligations (HUD standards) rather than operational needs. Providers often collect minimal data to satisfy compliance, missing nuances needed for equity assessment.
• Trust Deficits: Clients often withhold data due to fears of surveillance (e.g., by law enforcement), complicating collection.
• Capacity Gaps: Smaller providers lack the resources to invest in digital infrastructure, prioritizing direct service over data documentation. This leads to uneven participation.
• Legal Asymmetry: Larger partners (e.g., hospitals) possess significant "legal capital" to dictate data-sharing terms, whereas CoCs often lack the legal capacity to negotiate equitable agreements.

B. Data Storage

• Centralization vs. Customization: While HMIS centralizes storage, providers often maintain parallel local systems (including paper-based systems) to accommodate clients with low digital literacy.
• Data Integrity Risks: Attempts to integrate non-standardized local data into HMIS often result in duplication, missing rates, and confusion, forcing the CoC to restrict customization to preserve data integrity.

C. Data Analysis

• Compliance-Driven: Analysis is predominantly used for federal reporting and accountability rather than system-level learning.
• Interpretive Authority: Meaningful analysis requires both technical skills and deep contextual knowledge. Larger, established agencies possess the capacity to interpret data strategically, while smaller organizations rely on off-the-shelf reports or experiential judgment.
• Divergent Narratives: Without shared interpretive frameworks, the same metrics can support conflicting narratives (e.g., low acceptance rates viewed as failure vs. evidence of high demand requiring more resources).

D. Data Usage

• Communication Challenges: Coordinating data communication is difficult due to differing assumptions between providers and funders regarding what constitutes "success."
• Storytelling as Mitigation: To counter the limitations of raw data (e.g., missing identification status), participants rely on qualitative storytelling and hyper-local data to build trust and inform policy.
• Power Concentration: The ability to frame data narratives is concentrated among those with analytical capacity, often marginalizing smaller providers.

4. Key Contributions

The paper makes three primary contributions to the fields of Human-Computer Interaction (HCI), public-interest technology, and civic data infrastructure:

1. Qualitative Deep Dive: It provides an in-depth examination of HMIS, illustrating how mandated data sharing simultaneously enables coordination/accountability while reproducing power asymmetries in decision-making.
2. Beyond Collection: It expands the scope of data research beyond collection to include collective analysis and the transformation of data into community narratives, highlighting how frictions (like compliance shifts) concentrate authority.
3. Contextualizing HCI: It situates homelessness-related HCI research within broader policy and organizational contexts, demonstrating how data governance and infrastructure jointly shape data production, interpretation, and fairness.

5. Significance and Implications

The study challenges the assumption that data mandates inherently foster collaboration. Instead, it argues that shared data infrastructures institutionalize governance logic that defines whose knowledge counts.

• Reframing Governance: The authors propose shifting from a compliance regime to a learning infrastructure. This involves connecting standardization with participatory interpretation.
• Design Recommendations:
  • Capacity Building: Expand analytical capacity among smaller providers.
  • Layered Standards: Implement a stable reporting core with sanctioned local extensions to allow for experimentation.
  • Shared Interpretation: Pair metric definitions with documentation of underlying assumptions and separate data quality checks from narrative interpretation.
  • Feedback Loops: Design mechanisms to circulate insights across governance levels, ensuring that data informs policy rather than just policing it.

Conclusion: Effective data-driven collaboration requires more than just shared technology; it demands equitable distribution of analytical capacity, authority, and the ability to interpret data within its specific social context. Without these, mandated data systems risk reinforcing existing inequities and disempowering the very organizations they aim to coordinate.