DHS workforce expansion and 'soft feet' tactics coincide with record ICE detention volume

Summary

The Department of Homeland Security’s workforce expansion to more than 22,000 officers and ICE’s signing of more than 2,500 287(g) agreements coincided with a tactical shift to “soft feet” detentions completing in roughly 90 seconds, a sequencing The Guardian’s September 28, 2026 report documents following the Minneapolis killings and which Trump’s February “softer touch” remark and Mullin’s March “not in the lead story every single day” confirmation statement document as a stated priority of reduced news-cycle salience.
Trump’s and Mullin’s on-record statements function as smoking-gun evidence and the 287(g) expansion as hoop-test evidence for the political-response reading under Bennett-Checkel analysis, applied as cited authority, while operational-learning and capacity-endogenous hypotheses lack predictors for the Minneapolis timing tie.
Lakoff’s frame-analysis framework, applied as cited authority, identifies the administration’s “self-deportation” and “softer touch” lexical activations competing with the article’s “invisibilizing of immigration enforcement” and “soft feet” characterizations, while Entman’s four-function decomposition, applied as cited authority, populates problem definition around record detention volume, causal interpretation around 287(g) partnerships and compressed-time detentions, and moral evaluation around named cases including the deportation of US-citizen Brian José Morales García.
The Stanley and Herman-Chomsky propaganda-model framework, applied as cited authority, registers source-mix imbalance in the substrate — two immigration attorneys and named affected parties supplying critical commentary against a single DHS spokesperson (Chandler Rebel) who declined to discuss tactics — with administrative silence versus attorney commentary registering as a substantive finding under the framework.
Ellul’s integration dimension, applied as cited authority, anchors the cumulative normalization of detention not in attorney-quote commentary alone but in documented cases of detentions during routine activities — trash collection, school-bus drop-off, gas-pumping — and in the deportation of Morales García despite his stated documentary proof of citizenship.
The administration’s counterframe characterizes 287(g) partnerships as lawful delegation and “tremendous success” per Rebel’s stated comments, with operational effectiveness as the implicit metric, against which the article documents a “wide gulf” between detentions and deportees and cases in which detained individuals held legal status, judicial protection orders, or US citizenship.

The Department of Homeland Security, under Secretary Markwayne Mullin, more than doubled its workforce to more than 22,000 officers and staff in 2026, backed by $75 billion in congressional funding, while Immigration and Customs Enforcement signed more than 2,500 agreements under 287(g) — the federal program that authorizes state and local law enforcement to enforce immigration law — as of September 15, The Guardian reported on September 28, 2026. Within that operational expansion, the report records a tactical shift from high-visibility enforcement operations that witnesses described as a “siege” or an “occupation” to detentions that immigration attorney Willis described as occurring “very quietly, with very soft feet” and completing in roughly 90 seconds from agents’ arrival to removal. The shift follows the killing of two US citizens during an ICE operation in Minneapolis last year, after which President Donald Trump told reporters in February that “I learned that maybe we could use a little bit of a softer touch,” and which Mullin named as a six-month news-cycle objective at his March Senate confirmation hearing: “My goal in six months is that we’re not in the lead story every single day.” The substrate treats the Minneapolis sequencing, the on-record statements, and the parallel 287(g) expansion as the central evidence for the timing and structure of the shift.

What the substrate documents — the process trace

The substrate documents a temporally ordered sequence that anchors the analytical claim. The killing of two US citizens during an ICE operation in Minneapolis preceded Trump’s February “softer touch” remark, preceded Trump’s selection of Mullin as Secretary of Homeland Security, preceded Mullin’s March Senate confirmation hearing, and preceded the reported record detention levels. The DHS workforce expansion, the 287(g) signings, and the tactical tempo shift all fall within the same year. The newspaper headline-characterizes the result: “ICE is now detaining more immigrants than ever before.”

Four causal hypotheses are consistent with the substrate. H1 holds that the shift constitutes a deliberate political response to the Minneapolis killings, intended to reduce visibility while maintaining arrest volume. H2 holds that the shift reflects operational learning, in which what works at scale is quiet and fast. H3 holds that the shift constitutes strategic concealment in response to political opposition generated by high-profile operations. H4 holds that the shift is endogenous to expanded capacity, in which more agents and more 287(g) partners necessarily produce more distributed, less conspicuous enforcement. A refined version of H2, in which headline-volume calibration emerges as an outcome of scaling up rather than as a separately intended driver, remains consistent with the evidence: scale produces both distributed operations and lower per-incident news salience, so the secretary’s headline-volume metric could be endogenous to operational learning at scale rather than its cause.

Diagnostic evidence — Bennett-Checkel tests

Two pieces of evidence discriminate among these hypotheses. Trump’s on-record February remark — “I learned that maybe we could use a little bit of a softer touch” — attests to intent and is consistent with H1 and H3. Mullin’s on-record objective — “My goal in six months is that we’re not in the lead story every single day” — attests to news-cycle management as a measured variable and is similarly consistent with H1 and H3. Under the Bennett-Checkel hoop and smoking-gun tests, applied as cited authority, the administration’s on-record statements function as smoking-gun evidence for the political-response reading; few alternative hypotheses predict that the secretary’s success metric would be expressed in news-cycle terms.

The 287(g) expansion functions as a hoop test under the same framework. If H1 or H3 is true, distributing enforcement to local law enforcement partners is a foreseeable implementation. The substrate passes that hoop: “The administration is able to essentially hide a multiplied ICE force not only in unmarked cars, and not only behind masks, but also within local law enforcement through the 287(g) program,” Willis said. H4 cannot be eliminated: expanded capacity is a documented cause, and the Memphis-style mass operations and “mega-master” mass hearings show that the tactical repertoire has expanded rather than uniformly shifted. H4 does not predict the Minneapolis timing, however, and the smoking-gun statements make H1 and H3 the dominant explanations, with H2 and H4 as contributing rather than primary causes.

The “soft feet” framing coexists in the same period with concentrated mass operations. The El Corralón bar sweep in Memphis on August 29, in which more than a dozen agents entered at 1 a.m., shut off the music, instructed everyone to gather on the dance floor, and asked one by one about legal status, resulted in 120 people detained. “Mega-master” hearings — mass immigration-court hearings scheduled at the last minute so that many people cannot attend — produced more than 13,000 in absentia removal orders among the 53,808 removal orders in absentia issued in June, a 55% increase, according to researchers Joseph Gunther and Brandon Marrow at bklg.org. The substrate supports a “soft feet alongside mass operations” reading rather than a sole-modality reading.

Frame audit — competing metaphors and Entman decomposition

Lakoff’s frame-analysis framework, applied as cited authority, distinguishes the administration’s lexical activation from the article’s. The administration frames the enforcement direction under the rubric of “self-deportation,” with Rebel stating that DHS is “working diligently to ensure that aliens in our country illegally can no longer fly unless it is out of our country to self-deport.” Trump framed the tactical adjustment as “a softer touch.” The article frames the partnership expansion as Willis characterized it — the “invisibilizing of immigration enforcement” — and characterizes the enforcement overall as an “expanded dragnet fueled by police partnerships” and as a “mass-deportation agenda.” Lakoff’s vocabulary, applied as cited authority, treats these as competing metaphors for the same operational conduct.

The article’s structure foregrounds individual cases — Morales García, Garcés Pérez, “Y,” the Cuban man in Miami, the Ecuadorian father in Danbury — in Iyengar’s sense of episodic rather than thematic framing, applied as cited authority, while embedding those cases within aggregate figures. Linguistically, terms such as “removal orders in absentia,” “voluntary deportation,” and “third-country deportation” carry what the substrate reports as institutional euphemism; Lincoln-Goldfinch states on the record that “Almost every single detained parent that I talk to has been pressured into signing a voluntary deportation,” recasting the voluntariness frame.

Entman’s four-function decomposition, applied as cited authority, populates as follows. Problem definition: the asserted expansion of detention volume and reach, anchored in the headline characterization that “ICE is now detaining more immigrants than ever before” and qualified in the passage documenting “a wide gulf between the number of people ICE is detaining each month and the number of people who are deported.” Causal interpretation: 287(g) partnerships multiplied the enforcement footprint while compressed-time “soft feet” tactics shortened the time-window of each detention — a combination Willis described in quoted commentary: “The detentions are happening in a flash — we’re talking, like, 90 seconds from the time ICE shows up to that person being driven away.” Moral evaluation: the article’s critique is anchored in named cases (Morales García as a US citizen deported despite documentary proof of citizenship; “Y” placed on a Venezuela-bound flight despite a judicial protection order; Garcés Pérez shot by an ICE agent with a bullet lodged near his spine; the Guatemalan man killed in a fiery crash in Grand Rapids while fleeing federal immigration agents; the 120 detained at El Corralón) and in Lincoln-Goldfinch’s on-record characterization that “We’re seeing a lot of racial profiling.” Treatment recommendation: implicit, oriented toward procedural accountability rather than formally stated.

Propaganda-model source mix

The Stanley and Herman-Chomsky propaganda-model framework, applied as cited authority, registers source-mix asymmetry. The substrate’s documented source mix: two immigration attorneys (Willis, Lincoln-Goldfinch) whose clients are uniformly affected; one DHS spokesperson (Rebel); named victims and witnesses (Garcés Pérez, Morales García, “Y,” the Cuban man in Miami, the Ecuadorian father in Danbury, the Guatemalan man killed in Grand Rapids). Absent from the substrate: ICE field-agent perspectives, named 287(g) local law enforcement partner statements, and DHS internal voices beyond Rebel. Rebel’s quoted statement is restricted to “working diligently to ensure that aliens in our country illegally can no longer fly,” and when asked about tactics, Rebel stated that “DHS will not disclose or discuss law enforcement operations, strategy or tactics.” Rebel separately stated that “We have had tremendous success when local law enforcement works together with us.” Administrative silence versus attorney commentary registers, under the propaganda-model frame, as a substantive finding rather than merely a description of who answered the phone. The article’s own editorial characterization (“expanded dragnet,” “invisibilizing,” “mass-deportation agenda”) is editorial rather than sourced commentary; treating it as such respects sourcing discipline.

Ellul’s integration dimension

Ellul’s integration dimension, applied as cited authority, concerns the cumulative normalization of detention as an ambient feature of community life. The substrate anchors this reading not in attorney-quote commentary alone but in the cases it documents: a Cuban man in Miami detained moments after dropping a garbage bag into the bin; an Ecuadorian father in Danbury, Connecticut arrested just after seeing his six-year-old son onto the school bus; people detained while pumping gas, on their way to work, or on their way home. The reach extends further to Morales García’s deportation — a US citizen born in Denver who possessed documentary proof of citizenship, riding with his boss to a job as an air-conditioning installation technician, stopped by Gillespie County sheriff’s deputies and Texas Department of Public Safety officers working with ICE under 287(g), who told the officers of his citizenship and stated his birth certificate and social security card were at home, deported to Mexico anyway, and returned only after his lawyers submitted documents proving citizenship. The reach into documented-citizen cases sharpens the integration reading relative to attorney-quote commentary alone.

The substrate also reports that “On one day, a client might go to an ICE check-in and leave without incident. The next day, a different client with the same immigration status might be detained.” That unpredictability, per Willis, “makes it very hard to counsel people” and is “very hard to prepare people for what they can expect.” Federal circuit courts are split on whether immigrants who came to the US without lawful admission, including many asylum seekers, can be detained indefinitely without access to bond hearings; many detained parents accept voluntary deportation rather than wait months or years for their cases to process.

The administration’s counterframe

Per Rebel’s quoted statements, the administration frames local-law-enforcement partnerships as operational effectiveness — “tremendous success” — and tactical non-disclosure as standard practice. Under this counterframe, the 287(g) agreements function as lawful delegation of federal authority, the unpredictability Willis cites reflects case-by-case enforcement discretion, and the temporal compression of detentions reflects operational efficiency rather than a strategy of reduced visibility. The headline detention statistic can be characterized under the operational-effectiveness framing; the workforce doubling can be characterized as capacity-building; and the 287(g) expansion can be characterized as lawful delegation of federal authority. The implicit premise required for the article’s structural juxtaposition — Rebel’s “tremendous success” framing set against the article’s critical framing — to function persuasively is that detention volume is a reliable proxy for enforcement success; the article itself documents a “wide gulf” between detentions and deportees and reports cases in which detained individuals held legal status, judicial protection orders, or US citizenship, which complicates that proxy.

Boundaries of documentation

The substrate documents timing, scale, named statements, and individual cases. It does not document — and the journalistic evidentiary record typically cannot surface — interagency deliberative documents or internal DHS memoranda that could directly link the tactical change to the Minneapolis-killings trigger event. Doubly-decisive evidence categories absent from the substrate, in the Bennett-Checkel framework, would include: a leaked operational planning document naming the tactical change; a time-series analysis correlating arrest timing with news-cycle salience; statements from named 287(g) local law enforcement partners describing shifts in referral volume or composition; and congressional testimony recording the rationale for 287(g) expansion pace. Any of those would function as doubly-decisive — both necessary (the political-response reading would predict their existence in some form) and sufficient (their content would confirm the political-response reading against the refined-learning alternative). Their combined absence does not disconfirm the political-response hypothesis, because the administration has refused to discuss tactics on the record, but it marks the boundary of what the available evidence can establish.

Substrate figures used throughout — $75 billion in congressional funding, more than 22,000 DHS officers and staff, more than 2,500 287(g) agreements as of September 15, 53,808 in absentia removal orders in June, the 55% increase, the 90-second figure, more than 13,000 in absentia orders from mass hearings, 120 detained at El Corralón, and approximately 23,000 third-country removals — are properly attributed to The Guardian as the reporting source. The Minneapolis incident, the Austin shooting of Garcés Pérez, the Grand Rapids fatal crash, the Memphis nightclub raid, and the Phoenix “Y” case are all substrate-attributed to The Guardian’s reporting. Independent web verification did not retrieve confirming sources for these figures within the available search results; they remain substrate-attributed. Substrate items not surfaced in the analytical work — TPS scope figures (350,000 Haitians and Syrians; approximately 1.3 million immigrants from TPS-designated countries; approximately 170,000 Salvadorans awaiting decision), the expected revocation of business and tourism visas for up to 200,000 foreigners, the public-charge rule allowing officials to block lawful permanent residency based on use of public benefits, and the federal-circuit split on indefinite detention without bond hearings — were judged to fall outside the tactical-shift frame’s selectivity scope under news-floor discipline.

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.

Frame Audit: Surfaces the frame an argument adopts and what that framing quietly includes or excludes.
Process Tracing: Reconstructs the step-by-step causal pathway of a specific historical event.
Propaganda Audit: Reads a message for propaganda technique — loaded framing, manufactured consensus, and demonization.
Principal–Agent Problem: An agent acting for a principal has its own interests, which can quietly diverge.