Two ICE-involved fatal shootings within 48 hours in July 2026 share the same operational signature. On July 7, Lorenzo Salgado Araujo — a 52-year-old Mexican immigrant who had lived in the United States for 35 years, a business owner and father of three who had immigrated at 17 — was shot dead in his car by an ICE agent in Houston while on his way to work. His three co-workers survived and were taken into ICE custody. On July 8, Joan Sebastián Durán Guerrero — a 25-year-old Colombian immigrant holding a work permit, working two jobs as a food delivery driver and custodian at a veterinary office to support his wife and three-year-old daughter — was fatally shot by ICE in Biddeford, Maine, after agents rammed into his car and approached his driver’s side window. His daughter watched, still in her pajamas.

In both cases, authorities confirmed that ICE agents were seeking other individuals at the time of the shooting. Homeland Security Secretary Markwayne Mullin told Sen. Angus King of Maine that ICE was looking for someone else in the Durán case — a statement that, on its face, establishes Durán Guerrero was not the agency’s intended target. (The column misspells the surname as “Mullen”; web verification confirms Mullin was sworn in as the 9th DHS Secretary on March 24, 2026.)

The “looking for someone else” formulation is a correlational pattern marker, not a causal explanation. The gap between “aiming for someone else” and “killing the person you encounter” is where the question of institutional violence lives. The agency’s subsequent claims about what happened in each case have been independently contradicted. ICE stated that Salgado Araujo had been trying to run agents down with his car; passengers in the vehicle and available footage contradicted that account, according to reporting from CNN, Truthout, Democracy Now, and Houston Public Media. In Minneapolis, U.S. citizen Renée Good was fatally shot by an ICE officer on January 7, 2026; video quickly disproved the agency’s account, and a New York Times analysis found the agent “was not in the path of the victim’s SUV when he fired three shots at close range.” A second U.S. citizen, Alex Pretti, was fatally shot by CBP officers in Minneapolis on January 24 during anti-ICE protests.

The convergence of disputed official accounts across three incidents — Minneapolis, Houston, and Biddeford — separated by geography, personnel, and time points to a Methods failure in how use-of-force justifications are constructed after the fact, not merely an operational one before it. The pattern does not establish discriminatory intent in any individual shooting. It does establish that the agency’s institutional account of its own lethal incidents has been independently contradicted in multiple cases, and that the verification step distinguishing target from non-target does not occur before lethal force is applied.

Root cause #1: verification absent at the field level

The decision chain in both July incidents followed the same sequence. Agents arrived at a location without confirmed target identity, encountered someone other than their intended subject, and discharged lethal force. No protocol in the operational chain requires agents to verify that the person before them matches the intelligence product before using lethal force. The decision to fire is the decision; identification happens after the fact, in a post-hoc justification that witnesses and video have now repeatedly contradicted.

The Houston–Biddeford 48-hour interval is consistent with operational tempo outpacing the intelligence-gathering and verification pipeline. ICE made over 10,000 arrests in five days in early July 2026, with immigration arrests averaging approximately 1,400 per day during the first 12 days of the month, according to agency data reported by the New York Times and other outlets. Agency leaders ordered top ICE officials to focus more officer effort on apprehension. The throughput pressure is real, and it directly compresses the verification window: agents are deployed before their intelligence products are refined enough to distinguish targets from non-targets in the field.

Descending one level: why are intelligence products insufficiently specific? The political Environment — addressed below — sets volume targets that compress the verification pipeline, meaning intelligence analysts face the same throughput demands as field agents. The Environment–Methods convergence is the mechanism by which policy pressure translates into the operational conditions that produce fatal errors. The chain is inferential — the temporal correlation between intensified enforcement and increased fatal encounters is asserted by the columnist, and the mechanism (volume pressure → aggressive under-verified operations) is plausible but not independently verified by encounter-to-force ratios or internal ICE directives, neither of which appears in the source material.

Individual agent misconduct — poorly trained or reckless officers making isolated decisions — is the alternative explanation. The geography-separated, personnel-separated, time-separated convergence across three incidents sharing the same operational signature argues against treating each as anomalous. The structural account — that absent verification protocols and volume pressure together produce the conditions for these shootings — is more parsimonious across the pattern.

Root cause #2: accountability disabled by federal non-cooperation

The institutional review function that should process these contradictions has been structurally disabled. The causal chain operates as follows: without investigation, there is no finding; without a finding, there is no consequence; without consequence, there is no deterrent.

As of the column’s publication on July 19, 2026, no criminal charges or civil judgments had resulted from the Good or Pretti killings in Minneapolis. The FBI withheld key evidence from state investigators for months. Federal prosecutors turned over evidence to Minnesota investigators on approximately July 13, 2026 — six days before the column was published — after months of legal battles documented by the Atlantic, the Guardian, House Oversight Democrats, the Minnesota Reformer, and Politico. The Atlantic described the investigations as “moribund” due to federal non-cooperation. The Good family attorney’s characterization of the evidence turnover as “an important and meaningful step towards justice and accountability” confirms that accountability had not yet been achieved as of the column’s publication.

This is a Measurement failure operating above the level of any individual agent or any individual incident. The FBI’s months-long refusal to share evidence with state investigators is a federal-agency-level decision that disables state-level review regardless of the merits of any specific case. The temporal pattern — Minneapolis killings in January 2026, Houston and Biddeford killings in July 2026 — is consistent with the inference that absent accountability enables recurrence, though this link depends on the assumption that agents are responsive to consequence probability. The accountability gap is not merely a matter of institutional delay; the FBI’s pattern of evidence withholding in ICE-involved fatalities extends beyond the Minneapolis case, and whether the gap is isolated or systematic is itself an open question.

Political environment as contributing cause

The Trump administration’s intensified deportation campaign establishes the political Environment in which these operational failures occur. Volume pressure produces rapid, broad-sweep operations where agents arrive at locations without confirmed target identification. Donegan characterizes the agency as “a group of reckless, trigger-happy bigots tasked with carrying out the president’s mass deportation and ethnic cleansing regime” and the enforcement campaign itself as the source of the operational posture. The Houston–Biddeford 48-hour interval is consistent with operations tempo driven by policy-level enforcement priorities.

This is a contributing condition, not an independently sufficient cause. Removing the political pressure would alter operational tempo but would not by itself create the verification protocols or accountability mechanisms that are absent. The connection between policy-level enforcement directives and the specific use-of-force decisions that killed Salgado Araujo and Durán Guerrero is inferential: the temporal association with the deportation campaign does not by itself demonstrate that policy directives caused the particular decisions. Comparative use-of-force data across administrations, or internal ICE directives specifying encounter volume targets, would resolve the causal question; neither appears in the source material.

The stakeholder map: who holds power, who bears the cost

The parties in this situation occupy sharply unequal positions.

ICE institutional leadership and the Trump administration enforcement apparatus hold high power and high interest — they are the definitive stakeholders, though Donegan’s column contests the legitimacy of that position. From the administration’s frame, legitimacy is preserved and the position remains Definitive or Dominant; from the column’s frame, the position shifts toward Dangerous — power and urgency without legitimacy. Sen. Angus King holds congressional oversight power but variable engagement; his extraction of a briefing from the DHS secretary on the Durán case is the only instance of Senate-level oversight visible in the source material, and the Mullin statement directly establishes Durán Guerrero was not the agency’s target. That single point at which a Senate-level actor enters the accountability structure is the column’s nearest approximation to an oversight hook.

The dependent stakeholders — those with high stake but low power — include Salgado Araujo’s three co-workers, who witnessed the Houston shooting and were taken into ICE custody, making them unreliable witnesses by structural design: their testimony is shaped by the fact that the agency controls their detention and deportation timeline. They include Durán Guerrero’s wife and three-year-old daughter, who was present at the scene in her pajamas at approximately 7 a.m. They include the Houston neighborhood community, represented in the column by Maria Cervantes: “ICE hunts us like animals. I can’t help but think, who’s next? Am I going to get a bullet to my head for being brown?” They include the posthumous interests of Renee Good and Alex Pretti — U.S. citizens whose deaths extend the pattern beyond immigrant communities to general-enforcement-failure framing.

Moira Donegan occupies a broker position: a media intermediary connecting these dependent parties to a public that currently holds power and legitimacy but lacks urgency. The coalition structure is the dependent parties anchored by the broker, seeking to activate the dormant public into a definitive stakeholder bloc that can strip the dominant stakeholders of institutional immunity.

Parties absent or marginalized in the public record are structurally significant. The unnamed individuals ICE was actually seeking in Houston and Biddeford are shielded by ongoing operations — their identities unknown, their stake-if-present high but their visibility zero. The unnamed ICE field agents involved in the specific shootings are currently non-stakeholders because anonymity plus institutional immunity equals no stake; under retroactive liability they would invert to Dependent stakeholders with urgent personal legal jeopardy and no power. Consular authorities of Mexico and Colombia have Article 36 Vienna Convention standing but have made no public statements; political constraints of bilateral relations may be suppressing engagement below what their standing would support. Local law enforcement in Houston and Biddeford absorb community-trust cost without command authority over federal operations. Immigration attorneys and advocacy organizations constitute latent infrastructure — not named in the source text but structurally necessary for wrongful-death litigation, habeas petitions for detained witnesses, and the retroactive liability claims Donegan proposes.

The post-Trump public — the “we” of Donegan’s historical analogies — is the column’s intended definitive stakeholder, currently dormant. The column’s rhetorical project is to activate that stakeholder by tying remembrance of the dead to democratic survival.

How the column’s argument is built

The column’s case is constructed through deliberate structural choices whose design is analytically significant apart from the underlying facts.

Salgado Araujo and Durán Guerrero receive detailed biographical treatment — ages, nationalities, occupations, family structure, dates, and circumstances of death. Donegan devotes a full biographical paragraph to Salgado Araujo: his age, his 35 years in the country, his occupation, his family, the circumstances of his commute to work. Good receives a single sentence naming her citizenship, the city of her death, and the manner of her killing. Pretti receives the same treatment. The disparity is structural: the column allocates narrative weight to build identification with two individuals and uses the other two as pattern markers whose presence in the list carries argumentative force through accumulation rather than through individual detail. The weighted-spoke asymmetry operates to convert individual incidents into a systemic claim.

The “looking for someone else” statement functions as the structural link between these otherwise disparate incidents. Without it, they are geographically scattered events that happened during separate enforcement operations. With it, they share an operational signature. Witness testimony contesting the agency’s account in the Salgado Araujo case and video evidence disproving the official narrative in the Good case fill the mechanism gap between the pattern and the institutional-violence conclusion the column draws. The connection is non-obvious — the column foregrounds the shootings, not the connective tissue that links them into a pattern claim.

The column’s core argumentative move connects the post-administration accountability demand to historical precedent. Donegan invokes the post-Watergate pardon and post-Civil War amnesty as structural parallels, then links them to the enforcement deaths through a single sentence that ties future political reconciliation to forgetting the dead: “moving on requires forgetting about Salgado, and Durán, and about Good and Pretti, and about everyone else that ICE has killed and will kill.” This transforms the policy argument into a historical-structural argument about how democratic institutions handle state violence — a framing that loads the accountability demand with an obligation to democratic memory rather than leaving it as a policy preference to be weighed against competing interests. The historical analogy cross-link bridges what would otherwise be separate argumentative branches: the accountability-demand branch (prosecute ICE officers, abolish the agency) and the political-transition branch (post-Trump pressure to “move on”).

What the evidence does and does not establish

The operational findings — verification gap, disabled accountability, political pressure — are concrete and well-sourced. The accountability mechanisms that should process these contradictions have been disabled through federal non-cooperation, a structural finding confirmed by the FBI’s documented months-long evidence withholding and the Atlantic’s “moribund” characterization.

Whether the institutional-violence inference is warranted depends on the evidentiary standard applied. The pattern — agents looking for someone else, accounts contradicted by witnesses and video, accountability mechanisms disabled — supports the claim that institutional review has failed and that operational verification is absent. It does not, without comparative enforcement data or directive-level evidence, independently establish the systemic criminality the column argues for. The operational facts generate their own momentum independent of the broader policy framework: four people killed across three cities, in each case agents were searching for someone else, official accounts contested by witnesses and video. That convergence, regardless of the column’s broader claims, identifies an operational failure and an accountability gap that the institutional evidence confirms.

The most analytically vulnerable link in the column’s argument is the move from operational failure to discriminatory policy. The pattern of operational failures does not, by itself, demonstrate that the failures stem from a discriminatory agency culture rather than from the structural conditions the operational analysis has identified. The column treats pattern as proof of policy; the operational evidence supports pattern as evidence of operational failure and accountability breakdown, which are distinct claims with different evidentiary burdens. UCR and DHS OIG data that would supply a comparative use-of-force baseline across agencies and administrations are not cited. The policy-level critique (that intensified enforcement increases encounters and thus the statistical chance of shootings) and the claim about specific agents’ culpability (that each shooting reflects discriminatory intent) are collapsed into a single argumentative move. Donegan wrote that those responsible “must be investigated, tried, fined, punished and ostracized,” a prescription that presupposes a standard of proof the column does not itself supply.

The column is not designed as a legal brief. It operates as an activation mechanism for a dormant public — a genre in which the evidentiary standard appropriate to the argument shifts from courtroom proof to rhetorical plausibility, and in which the gap between pattern and policy is a feature of the form rather than a failure of the case. But the gap is real, and Donegan does not bridge it. Advocacy journalism that seeks to prefigure post-administration legal accountability should disclose the evidentiary burden required in legal proceedings, making the gap between polemic and courtroom visible to the reader.

The two root causes are independent

Removing qualified immunity for ICE officers and creating a statutory review mechanism would address the accountability root cause. But the absent-verification root cause — no protocol requires agents to confirm target identity before using lethal force — would remain intact. The two fixes are independent: fixing accountability without fixing intelligence verification would leave the operational failure sequence unchanged. Addressing the verification gap requires inserting a gate between intelligence intake and force deployment — biometric confirmation, radio verification, or documented visual identification — at the field-supervisor level. Neither fix alone is sufficient; both are necessary.

Carry forward

The analysis identifies concrete intervention points, but several questions remain:

  • The witnesses. Salgado Araujo’s three co-workers remain in ICE custody — held by the agency whose actions they witnessed. The sharpest power imbalance in this situation is not between ICE and the broader community but between the agency and its own witnesses, who have direct knowledge but whose testimony is shaped by the fact that the agency controls their detention and deportation timeline.

  • The consular authorities. Mexico has diplomatic standing for Salgado Araujo and Colombia for Durán Guerrero under Article 36 of the Vienna Convention on Consular Relations. Neither country’s consular response appears in the source material. Whether diplomatic engagement has occurred through private channels, or whether bilateral relations with the United States have suppressed it, is unknown.

  • Local law enforcement. The Houston and Biddeford police departments operate in jurisdictions where ICE conducted fatal operations and absorb community-trust costs without command authority over federal enforcement actions. Their response — incident reports, public statements, or silence — is absent from the source material and would illuminate the local institutional dimension of the accountability gap.

  • Pre-operation verification. Could pre-operation verification protocols — confirming target identity before force is authorized — prevent the specific encounter sequence that defines these incidents? The operational chain identifies this as the intervention point, but whether such protocols exist in other federal enforcement agencies, and whether they could be imposed on ICE absent abolition, is not established by the source material.

  • Federal-state evidence sharing. What mechanisms could compel federal-agency cooperation with state investigators? The FBI’s months-long evidence withholding disabled accountability for the Minneapolis killings. The statutory landscape governing federal-state evidence sharing in this context is not addressed by the column or the available source material.

This is an analysis based on official statements, witness testimony, institutional records, and the column at issue. Facts not in dispute are stated as such; framing findings reflect the relationship between accounts, not claims about intent.

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.

Relationship Mapping
Extracts the network of ties among people, institutions, and entities.
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.