Rachel Reyes has waited 16 months for any acknowledgment that her son’s death matters. On March 15, 2025, 23-year-old U.S. citizen Ruben Ray Martinez was shot by an ICE agent on South Padre Island, Texas, after a night celebrating his birthday. The Department of Homeland Security took 11 months to publicly disclose the agency’s involvement — and did so only after the watchdog group American Oversight filed a public-records request. Body-camera footage from state and local officers, reviewed by NPR, “did not clearly show whether the vehicle struck an agent,” contradicting DHS’s account that Martinez had “intentionally run over” an officer. A Cameron County grand jury declined to bring criminal charges against the federal agent in late February 2026. No admission of fault has been made.

Martinez’s killing is not an isolated incident. At least 10 people have died in encounters with immigration-enforcement officers since the current administration took office, according to an Associated Press tally. The Trace, a nonprofit journalism site focused on gun violence, counted 29 instances in which ICE agents fired at civilians during that period, at least seven of them fatal. The ACLU documented dozens of incidents across eight states since January 2025 in which immigration officers rammed vehicles, smashed windows, and struck people with cars. The pattern’s persistence — not the individual tragedies — is the analytical subject.

Two independent root causes produce the same outcome: no accountability.

The first is a policy structure that prioritizes enforcement volume over encounter safety. Following the July 2026 fatal shootings of Lorenzo Salgado Araujo in Houston and Joan Sebastián Durán Guerrero in Maine days later, DHS planned a halt to nonurgent vehicle stops. Within roughly one day, Trump posted on Truth Social: “We CANNOT give up one of I.C.E.’s most important and effective Crime Fighting tools, THE TRAFFIC STOP! Once we do, we are playing right into the criminal’s hands.” White House Press Secretary Karoline Leavitt reinforced the reversal, calling vehicle stops “a necessary tool that ICE agents need in order to continue their deportation campaign.” The safety measure was countermanded at the political principal level within 24 hours. This incentive hierarchy — enforcement throughput above reduction of lethal encounters — manifests in every downstream failure: no mandatory timely disclosure requirement, no standardized encounter protocol, no incentive to implement body cameras, and direct presidential countermanding of safety pauses.

The second root cause is the legal insulation of federal agents from prosecution. The article states that federal agents have “broad legal protections” while performing their duties, making prosecution “difficult and relatively rare.” This chain is structurally independent of the management-incentive chain — even a DHS leadership committed to reducing fatalities would face the same prosecutorial barriers. Removing it would require statutory change, not agency policy change.

A third, convergent root cause is structural information asymmetry: the investigating agency controls both the evidence and the timeline of its release, with no independent verification mechanism to compel disclosure. The 11-month delay in acknowledging Martinez’s death, the withheld body-camera footage in Minnesota (where Hennepin County Attorney Mary Moriarty reported that her office had finally received evidence the Justice Department had previously withheld), and the absence of agent-worn body cameras are convergent expressions of this root. DHS pledged to equip all immigration-enforcement agents with body cameras; the pledge remains unfulfilled. ICE agents involved in both the Houston and Maine fatal shootings were not wearing cameras. Attorney Charles Stam, representing Reyes, articulated the structural ceiling: “If a U.S. citizen killed by an ICE agent cannot get a serious federal investigation or an admission of fault, imagine the odds facing families whose immigration status is precarious, who lack legal representation, or whose loved one’s death did not happen to be caught on camera.”

The conflict’s stakeholder map reveals systemic role gaps.

The conflict pits families of people killed by immigration-enforcement officers against the federal enforcement apparatus — ICE, Homeland Security Investigations, and DHS leadership — over accountability for use of force. Families seek acknowledgment, transparency, and consequences; the enforcement apparatus asserts defensive justification for each killing and continues the operational practices (vehicle stops, absence of body cameras) that produce them.

The surrounding community is structured in concentric rings. The intimate ring includes Reyes, attorney Stam, and passengers in Salgado Araujo’s van who disputed ICE’s account to Houston Public Media. They are engaged in direct legal and political advocacy but lack institutional power to compel a federal response. The mid ring includes the Cameron County grand jury (which declined to charge), the Hennepin County Attorney’s office, Texas Department of Public Safety officers whose body cameras recorded the Martinez scene, and local law enforcement at scene-management sites. These actors have investigative and documentary capacity but no enforcement authority over federal actors. The outer ring includes civil-society organizations (American Oversight, the ACLU), national media (NPR, The Trace, AP, The New York Times, where Reyes published an op-ed), and political actors whose statements set enforcement policy (Trump, Leavitt, DHS Secretary Markwayne Mullin, who took office in March 2026). Additionally, faith institutions, immigrant-rights mutual-aid networks, and diaspora organizations in south Texas, Minnesota, and Maine are plausible but unnamed actors — in immigrant-adjacent communities, churches, mosques, and mutual-aid networks routinely serve as first-contact points for families in crisis.

The third-side role analysis identifies which functions are active, which are compromised, and which are unfilled:

The witness role is active — NPR, The Trace, the AP, and American Oversight make the pattern visible and impose consequences through documentation. The equalizer role is partially filled by attorneys, watchdog litigators, and investigative journalists, addressing power asymmetry through legal representation, public-records enforcement, and media exposure. The arbiter role is nominally filled by grand juries and county attorneys but degrades under evidentiary withholding — the Cameron County grand jury’s decision not to charge, despite footage that did not clearly show whether the vehicle struck an agent, illustrates failure by evidence-control dynamics.

The peacekeeper role was briefly activated when DHS planned to halt nonurgent vehicle stops after the July 2026 Houston and Maine shootings, then overridden within roughly one day by Trump. The institutional actor best positioned to execute the role was overridden by political authority. The referee role is compromised — federal use-of-force policies exist but are not externally enforced; agents operate without body cameras despite the pledge; evidence reaches local prosecutors only under pressure. The provider role is needed but unfilled — no institution offers families of the killed a structured path short of litigation for basic information, case-status updates, or grief services. The mediator role is not filled by any named actor — DHS’s 11-month delay in acknowledging involvement signals institutional unwillingness to engage. The healer role is unfilled institutionally; Reyes’s New York Times op-ed is self-administered witness, not institutional healing. The bridge-builder role is aspirational — DHS’s unfulfilled body-camera pledge was a bridge-building gesture; as of July 2026, no bridge exists.

Escalation signals are active and accelerating.

Rhetoric hardening: Trump’s Truth Social post reframes operational restraint as capitulation — “We are playing right into the criminal’s hands” — and Leavitt endorses vehicle stops as “necessary.” Channel breakdown: DOJ withheld evidence from a county attorney in Minnesota; DHS was silent for 11 months on Martinez’s death until American Oversight’s FOIA request; passengers in Salgado Araujo’s van disputed ICE’s account. Public symbolic markers: the ACLU’s eight-state documentation, the 29-incident Trace count, and the July 2026 cluster of two killings within days. The temporal pattern — 16 months since Martinez’s death, 11-month DHS delay, the planned traffic-stop pause immediately reversed, ACLU documentation covering a full year of incidents — signals chronic unresolved conflict, eroding trust and hardening positions.

Proposed interventions address specific role gaps, each with recognized limits.

A federally funded but independently administered family-assistance office, modeled on DOJ victim-assistance programs, would address the provider gap by offering families of those killed in federal enforcement encounters a single point of contact for information, grief services, and case-status updates. Congress could authorize it via oversight legislation; absent legislative action, DHS could create it under existing victim-services authority. The expected effect on conflict trajectory is moderate — it reduces information asymmetry but does not change use-of-force policy.

Mandatory automatic referral of every federal-agent fatal shooting to an independent special prosecutor would address the arbiter gap by removing reliance on local grand juries dependent on federal cooperation for evidence. The Attorney General could implement it via policy; absent executive willingness, state attorneys general (as Moriarty is attempting in Minnesota) could assert concurrent jurisdiction where state law permits. The expected effect is moderate — prosecuting federal agents remains legally difficult regardless of referral mechanism, but an independent referral would increase transparency and reduce the risk of evidence suppression.

The ACLU could extend its documentation into a formal legal challenge arguing that the pattern of force without accountability constitutes a due-process violation, forcing a federal court to assume the referee role the executive has vacated. The expected effect is uncertain — it depends on whether a federal court accepts jurisdiction; it could produce a binding ruling on use-of-force procedures.

American Oversight could push for a congressional hearing that includes Reyes and families of Durán Guerrero and Salgado Araujo, creating a provisional mediator role where DHS must answer directly to affected parties. The expected effect is moderate — it produces a public record and political cost, not accountability.

Community organizations in the eight ACLU-documented states could coordinate a national coalition of local legal defense groups and immigrant advocacy organizations that monitors enforcement patterns, issues public alerts, and provides on-the-ground de-escalation support during traffic stops — addressing the peacekeeper gap at the community level. The expected effect is localized; it cannot prevent federal-level policy continuation but may reduce harm at the community level.

The structural constraints on intervention are severe and well-documented.

The enforcement apparatus is politically directed toward escalation. The actor with the most direct peacekeeping authority — DHS Secretary Mullin — was overridden within a day of attempting operational restraint. Trump’s public position — that traffic stops must continue without qualification — signals that resolution-cluster interventions require either a change in political leadership or countervailing institutional power (courts, Congress) the community does not currently command. Federal agents have broad legal protections while performing their duties, making prosecution “difficult and relatively rare.” Mediation presupposes willingness to engage; current administration posture precludes. Equalization requires leverage generated by parties themselves; the third side can facilitate but cannot substitute.

Sub-state alternatives within constraint exist: community-based restorative-justice programs at the county level could partially fill healer and provider roles without requiring federal cooperation. Faith-community mediation between families and local ICE field offices could serve a bridge-building function at point of contact. Diaspora-led monitoring networks could provide witness-infrastructure redundancy independent of institutional journalism cycles. These address emotional and informational needs without reaching the operational and political level where use-of-force policy is set. Families and communities are agents constructing informal prevention and healing capacity within constraint, not only objects of third-side analysis.

Confidence assessment and unknowns.

Confidence in the dominant causal chain — the management-incentive structure producing lethal force without accountability — is moderate. The causal links from management incentive through to individual encounter outcomes are well-supported by the documented sequence: proposed halt, presidential reversal, continued fatalities. The convergence across multiple analytical categories is supported by independently verified data points (AP tally, The Trace dataset, ACLU report). Confidence is not high because the substrate does not provide complete data on all 29 shooting incidents, the training-adequacy branch contains an unresolved gap regarding whether DHS protocols exist that agents failed to follow, and the relative independent causal weight of the legal-immunity chain versus the management-incentive chain cannot be quantified.

Key unknowns include: whether DHS’s planned traffic-stop pause had institutional support beyond the brief window before Trump’s reversal (clarifying whether DHS leadership is a potential peacekeeper-bearer or a captured one); whether the body-camera deployment pledge is delayed by procurement, policy disagreement, or deliberate stonewalling; whether the Cameron County grand jury received the same evidentiary package NPR obtained from the Texas Department of Public Safety (determining whether arbiter failure was by design or by evidence suppression); whether a federal court would accept jurisdiction over a pattern-of-force question; whether any of the 29 ICE shooting incidents has produced a body-camera video contradicting DHS’s account as the Texas DPS footage did for Martinez; and whether a community peacekeeper network can scale across eight states without centralized coordination.

The Swiss Cheese Model applies.

Each layer — policy restraint, body-camera documentation, standardized encounter protocol, timely disclosure, independent investigation, prosecutorial accountability — has holes aligned in the same direction. The management-incentive chain and the legal-immunity chain are structurally independent but produce the same outcome: no accountability. Fixing one without the other would not resolve the pattern. Reyes’s words capture the human weight of that structural alignment: “It shouldn’t be our new normal. They’re destroying more families with all this chaos and it’s not right.”

Analytical techniques used in this piece

This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.

Root-Cause Analysis
Traces a symptom back along its causal chain to the conditions that actually generated it.
Stakeholder Mapping
Charts the parties to a situation — their interests, power, and alignments.
The Third Side
Takes the vantage of the surrounding community that has a stake in resolving a conflict (Ury).