Content uploaded by Robert Koulish
Author content
All content in this area was uploaded by Robert Koulish on Jul 29, 2015
Content may be subject to copyright.
A preview of the PDF is not available
The Immigration Detention Risk Assessment
Abstract
In early 2013, U.S. Immigration and Customs Enforcement (“ICE”) deployed nationwide a new automated risk assessment tool to help determine whether to detain or release noncitizens pending their deportation proceedings. Adapted from similar evidence-based criminal justice reforms that have reduced pretrial detention, ICE’s initiative now represents the largest pre-hearing risk assessment experiment in U.S. history — potentially impacting over 400,000 individuals per year. However, to date little information has been released regarding the risk assessment algorithm, processes, and outcomes.
This article provides the first comprehensive examination of ICE’s risk assessment initiative, based on public access to ICE methodology and outcomes as a result of Freedom of Information Act requests. This article argues that immigration risk assessment in its current form will not reduce current over-detention trends. The unique aspects of immigration enforcement, laws, and the impacted population will likely frustrate accurate calibration of the risk tool, and effective implementation of even a calibrated tool — in turn frustrating constructive impact of ICE’s risk assessment initiative on over-detention. Consequently, the immigration risk assessment may only add a scientific veneer to enforcement that remains institutionally predisposed towards detention and control.
Additionally, this article argues that even if more accurate, evidence-based immigration detention were achieved under a future risk assessment regime, it would nonetheless likely be accompanied by several disadvantages. Particularly, risk assessment could facilitate a transition from mass detention to mass supervision of an even wider net of supervisees, by justifying lesser deprivations of liberty such as electronic supervision.
... In addition, they had also received ICE guidance and training manuals from the American Immigration Council via an FOI request. ICE had deployed the automated RCA tool to assist in detention decisions (Noferi & Koulish, 2015). Noferi and Koulish found that the RCA detailed summaries which were sent to detention facilities had five sections: overview regarding detention or release, special vulnerabilities, mandatory detention assessment, public safety risk assessment, and flight risk assessment. ...
... The RCA was also found to collect non-citizens' basic personal information including gender, age, as well as country of citizenship and origin. Noferi and Koulish note that a more transparent risk assessment system that would also respect individual rights could be implemented by immigration enforcement authorities, further advising that public disclosure, government oversight, audit trails, and individual access to data could provide more due process in decisions relating to the detainees (Noferi & Koulish, 2015). Kerwin et al. (2015) had also received information from ICE on detainee population details via FOI requests that were submitted by reporters from the Boston Globe and Associated Press. ...
... Larsen and Piché acknowledge that KIHC was indeed built for specific profiles of inhabitants, whether the policies and features were recognized as "worthy acts of cultural accommodation or as extensions of systemic racism" (Larsen & Piché, 2009). Similarly, Noferi & Koulish (2015) found racial profiling to be present with the ICE RCA tool in the US. Noferi and Koulish provide an example from their FOI responses of RCA cases, with one case being of a single, 26-year-old Salvadorian male who had no criminal history or immigration history. ...
In this report, we take stock of English-language scholarly publications appearing during the past two decades in which scholars and researchers used FOI requests to generate data for their studies on prison and jail systems. We review these studies looking at trends in frequency over time, where they come from, what other data sources they used along with FOI, and what novel contributions FOI requests have allowed researchers to make to existing prison and jail scholarship.
... 13 For most other noncitizens, risk is determined on a case-by-case basis using an automated risk assessment tool that assigns a public safety and flight risk score based on DHS enforcement priorities. However, the algorithm has been criticized as being unduly weighted in favor of detention (Noferi and Koulish 2014;Koulish 2016). ...
... Ultimately, a holistic and human rights-based approach to reforming immigration detention is a necessary part of accomplishing this goal. See, e.g., Noferi and Koulish (2014); Koulish (2016). 5 See, e.g., Legomsky (1999); Kimball (2009);Stefanelli (2020). ...
Every year, thousands of people are detained in United States immigration detention centers. Built to prison specifications and often run by private companies, these detention centers have long been criticized by academics and advocacy groups. Problems such as overcrowding and lack of access to basic healthcare and legal representation have plagued individuals in detention centers for years. These failings have been illuminated by the COVID-19 pandemic, which has disproportionately impacted detained migrants. Against a human rights backdrop, this article will examine how the U.S. immigration detention system has proven even more problematic in the context of the pandemic and offer insights to help avoid similar outcomes in the future.
... Much of the research on migration control and technology tends to focus on sophisticated systems that involve algorithms or artificial intelligence (AI), and for good reason. Automated decision making is already incorporated into Canada's asylum process [16] and in Immigration and Customs Enforcement's immigrant detention determinations [15,35,36]. Biometric technologies have become ingrained in the identification and risk assessment of people crossing international borders [8,9,37,38]. ...
As the emerging literature on migration studies has demonstrated, migrants who are seeking asylum around the world are increasingly finding that the process is mediated by a variety of new technologies. While the process of digitizing various aspects of migrant protection may promise improvements, new technologies also risk limiting access to asylum for migrants who are unable to overcome these new digital barriers to entry. This article explores the digitization of asylum by examining the context and consequences of the U.S. government’s deployment of a smartphone app called CBP One in early 2023 which suddenly became one of the main pathways for migrants to seek asylum along the U.S.–Mexico border. In doing so, this article makes two contributions to the literature on the digitization of asylum. First, the article shows how CBP One, which was not initially designed for asylum seekers, morphed into a tool that took center stage in border enforcement statecraft during a period of exceptional migration policies. Second, this article examines the range of what have been referred to as “glitches” with CBP One, to demonstrate how the app created new digital barriers to asylum. Rather than accepting glitches as mere accidents, this article argues that these glitches are the result of a political decision to force already vulnerable migrants to rely upon experimental technologies that hinder rather than facilitate their asylum-seeking process.
... Inquiry into algorithmic enforcement technologies have further drawn attention to the way in which familial relationships-as pieces of data that are analyzed and mobilized to produce risk profiles-are key to shaping uneven practices of border enforcement and policing. Family relationships are key to risk profiling (Amoore, 2006;Evans and Koulish, 2020;Nofferi and Koulish, 2014) and tactics of premediation (De Goede, 2008) that underscore algorithmic approaches to border security, which combine the biometric attention to the individual body and its embodied characteristics with the imagined impacts of networks, connections, and associations such as family relationships that could impact national security. For instance, Wilson and Weber (2008: 133) detail how risk profiling relies much more on aggregate information and associations than any risk factor specifically connected to the embodied individual migrant, such that "the association of illegalized border crossers with biometric technologies fuses a disparate array of amorphous dangers-terrorism, organized crime and uncontrolled migration-as a single unified threat to national sovereignty." ...
Scholars have increasingly focused on the role of the family within border enforcement practices. In this paper, we build on and extend these research efforts to propose a research agenda driven by a new understanding of the relationship between families and immigration enforcement. Drawing on examinations of emerging enforcement strategies, including family separation and public information campaigns, we suggest that the family as a social unit and set of relationships is increasingly targeted within the regulation of transnational migration, what we term “relational enforcement.” Greater attention to relational enforcement tactics, processes, and impacts helps to frame geographies of border enforcement.
... (2008). Noferi and Koulish (2014) were first to bring immigration risk into the crimmigration literature. ...
Objective
This article measures Immigration and Customs Enforcement (ICE) officers’ punitive biases in immigration enforcement, comparing rates of dissent with low‐ and high‐risk classification scores. It provides evidence that contextual political factors alter dissent with risk classification assessment (RCA) scores and describes how such dissent informs more punitive versions of ICE's RCA scoring rules.
Methods
We implement a statistical decomposition technique that compares the gap between low‐ and high‐risk dissent rates in risk assessments. Historical and qualitative evidence describes the effect of punitive biases in algorithmic editing.
Results
Results of this investigation show that larger gaps in dissent rates between low‐ and high‐risk classifications result in more punitive scoring updates in later versions of the RCA algorithm.
Conclusion
Our article provides evidence that risk assessment tools implemented to reduce biases in immigration enforcement were updated to accommodate the punitive bias of ICE officers and supervisors.
... In January 2013, DHS launched the RCA-the largest risk assessment tool in the country-as part of its expanding detention regime (Nofferi and Koulish, 2014). The RCA would, in theory, measure a migrant's flight risk and risk to public safety in order to determine whether he or she should be detained by ICE. 9 The RCA is an automated risk tool designed to help determine whether to detain or release non-citizens pending their deportation proceedings. ...
This Report demonstrates that the Immigration and Customs Enforcement (ICE) violates legal requirements to provide immigrants with an individualized custody determination. Trump’s enforcement policies brought a surge of low-risk immigrants into ICE custody. The detention risk tool was supposed to train officers and strongly discourage them from detaining low-risk immigrants who posed no harm to society and were not a flight risk. Data received pursuant to FOIA show the opposite result. ICE has failed to perform the individualized assessment and restrict its use of civil detention to only those whose high levels of dangerousness and risk of flight justify their incarceration. The data show that officers have manipulated the risk tool by subjecting low-risk immigrants to blanket detention, which has come to define the no-release Trump immigration policy in the New York City area.
... In January 2013, DHS launched the RCA-the largest risk assessment tool in the country-as part of its expanding detention regime (Nofferi and Koulish, 2014). The RCA would, in theory, measure a migrant's flight risk and risk to public safety in order to determine whether he or she should be detained by ICE. 9 The RCA is an automated risk tool designed to help determine whether to detain or release non-citizens pending their deportation proceedings. ...
This Report demonstrates that the Immigration and Customs Enforcement (ICE) violates legal requirements to provide immigrants with an individualized custody determination. Trump's enforcement policies brought a surge of low-risk immigrants into ICE custody. The detention risk tool was supposed to train officers and strongly discourage them from detaining low-risk immigrants who posed no harm to society and were not a flight risk. Data received pursuant to FOIA show the opposite result. ICE has failed to perform the individualized assessment and restrict its use of civil detention to only those whose high levels of dangerousness and risk of flight justify their incarceration. The data show that officers have manipulated the risk tool by subjecting low-risk immigrants to blanket detention, which has come to define the no-release Trump immigration policy in the New York City area.
Through an analysis of interviews with Southern California attorneys, supplemented by archival materials, this article contributes to the literature on gangs, critical criminology, and Gothic tropes by examining how the ambiguous nature of gang profiling allows state actors to target racialized others in various legal and administrative venues with little evidence and few procedural protections. I conceptualize gang phantasmagoria as the constant, amorphous, unpredictable, and haunting threat of racialized gang allegations and argue that the dynamic shapes the work of legal practitioners and constitutes a state mechanism of racial terror. Specifically, first I argue that government officials deploy the specter of gangs to both portray asylum seekers as monstrous threats and justify restrictions in asylum eligibility. I then illustrate how the potential for gang phantasmagoria to upend asylum applications and trigger the deportation of their clients elicits constant low-grade anxiety for attorneys. Consequently, attorneys are forced to adopt more cautious approaches to legal work in a way that indirectly facilitates the social control of young Latinx immigrants.
Canada Border Services Agency (CBSA) officers use the National Risk Assessment for Detention (NRAD) process to evaluate the ‘riskiness’ of immigration detainees. The NRAD’s key tool is a 2-page document laying out ‘risk factors’ with corresponding points that add up to scores of ‘dangerousness’ allegedly posed by non-citizens. CBSA officers then recommend detention in either a provincial prison or a lower security ‘immigration holding centre’. In a national context of no legislated upper time limits on detention periods, and where telephonic and other access to incarceration sites is impeded, the NRAD form’s outcome portends serious, long-term consequences.
This article examines why deportation and imprisonment for immigration offenses rose under presidential administrations that claimed to favor more “humane” approaches to immigration enforcement. I examine the politics of enforcement discretion on the US-Mexico border during the administrations of Bill Clinton (1993–2001) and Barack Obama (2009–17). Drawing on historical and ethnographic research, I argue that the Clinton and Obama administrations took a punitive humanitarian approach to enforcement discretion aimed at punishing “illegal immigration” at the border while protecting “legal immigrants” with long-standing ties to the United States from deportation. The findings show that such an approach extended crime control to US-Mexico border enforcement. This blend of humanitarian and punitive approaches systematized criminal enforcement priorities and expanded the discretion of border agents to deport and imprison. Just as other scholars have shown how liberal reform contributed to the rise of the carceral state, this article shows how immigration policies that blended humanitarianism and security punished the very people such policies were designed to protect.
Twelve years after the September 11th attacks, countering domestic terrorism remains a top priority for federal law enforcement agencies. Using a variety of reactive and preventive tactics, law enforcement seeks to prevent terrorism before it occurs. Towards that end, community policing developed in the 1990s to combat violent crime in inner city communities is being adopted as a means of collaborating with Muslim communities and local police to combat 'Islamist homegrown terrorism.' Developed in response to paramilitary policing models, community policing is built upon the notion that effective policing requires mutual trust and relationships among local law enforcement and the communities they serve. Thus, traditional community policing is premised on their convergence of interests.While community policing in counterterrorism appears facially sound, this Article proffers that this endeavor is fraught with peril – both for collective civil liberties interests and local police’s interests in preserving relationships of trust. Accordingly, community policing exacerbates, rather than resolves, the underlying subordination of Muslims post-9/11 manifested in preventive counterterrorism policies, notwithstanding the increase of homegrown terrorism threats from non-Muslim groups. The Article asserts three critiques of community policing in counterterrorism: it is more akin to counterradicalization taken from military counterinsurgency strategy than the partnership-based traditional community policing model; to the collective detriment of communities it divides them into 'Good Muslims' willing to cooperate with law enforcement on the federal government’s terms and 'Bad Muslims' who demand a meaningful quid quo pro that ensures protection of Muslim communities’ civil rights and liberties; and it deputizes Muslim leaders to gather and share seemingly innocuous information about their communities that may be used adversely to their collective interests as part of the predominantly prosecution-driven counterterrorism regime. As such, CCP as currently envisioned betrays its rhetoric of empowerment and mutual trust, and is just another weapon in the federal government’s toolkit that perpetuates the 'terrorist other' stereotype. Unless systemic reforms are made to federal preventive counterterrorism strategies, community policing is likely to aggravate existing civil liberties violations and impair otherwise good relations between Muslim communities and local police. Thus, a serious rethinking of proposals to implement community policing in counterterrorism is warranted.
After 9/11, a series of revitalised strategies and techniques were promoted that were intended to govern the highly uncertain conditions and threats that risk techniques seemed no longer able to prevent. Precaution, preparedness and speculative pre-emption have been particularly prominent. But all establish a negative and defensive framework of ‘freedom from’. In the past decade, programs of resilience, and particularly resiliency training, have been developed with the aim of creating subjects able to thrive and prosper under conditions of extreme uncertainty. Reflecting many of the assumptions and goals of neo-liberal politics, resiliency has already emerged as a principal technology for military and business, and may be the answer to the neoliberal dream of a society of extreme entrepreneurs.
Big Data is increasingly mined to rank and rate individuals. Predictive algorithms assess whether we are good credit risks, desirable employees, reliable tenants, valuable customers-or deadbeats, shirkers, menaces, and "wastes of time." Crucial opportunities are on the line, including the ability to obtain loans, work, housing, and insurance. Though automated scoring is pervasive and consequential, it is also opaque and lacking oversight. In one area where regulation does prevail-credit-the law focuses on credit history, not the derivation of scores from data. Procedural regularity is essential for those stigmatized by "artificially intelligent" scoring systems. The American due process tradition should inform basic safeguards. Regulators should be able to test scoring systems to ensure their fairness and accuracy. Individuals should be granted meaningful opportunities to challenge adverse decisions based on scores miscategorizing them. Without such protections in place, systems could launder biased and arbitrary data into powerfully stigmatizing scores.
The Article addresses itself to immigration law governing the admission and expulsion of aliens, exploring the gap between the "plenary power" doctrine-the notion that Congress and the executive branch have broad and often exclusive authority over immigration decisions-and the actual practice of many federal courts in the immigration field. Federal courts, the author argues, often apply two distinct sets of constitutional norms in immigration cases, one set drawn from immigration law proper and applied directly to constitutional cases, and a second, "phantom norm" set, borrowed from mainstream public law and applied in the interpretation of immigration statutes. The result is an immigration law regime in which the plenary power doctrine has been undermined, but not reversed, through statutory interpretation. The Article explores the reasons for the application of "phantom norms" to immigration statutes and the problems created by that practice, ultimately recommending a direct reassessment of plenary power as a constitutional doctrine. In so doing, the Article serves not only as an examination of the interpretive dynamics peculiar to immigration law but also as an exemplary study of a general problem of legal process-the relationship between statutory interpretation and constitutional law.
In the analysis of the budgetary game between politicians and bureaucrats, the emphasis has been on the politicians' uncertainties about bureaucratic performance and the ensuing advantages enjoyed by nonelected officials. Such a perspective neglects the uncertainties that bureaucrats must confront. This paper models these dual uncertainties--superiors about subordinates, subordinates about superiors--where each party uses its informational advantage strategically. Model 1 assumes a legislature that is sufficiently organized to create a monitoring system and examines the bureau's responses to parametric variations in that system. Model 2 assumes a fully strategic legislature, one able to create an optimal monitoring and incentive structure. The two models have sharply contrasting implications.
The convenors of this symposium have asked us to think about due process. Much of the due process literature in the United States considers how individuals can defend themselves against the power of the state and focuses upon equipping individuals for their confrontations with the state. Some commentators see the purpose of procedural safeguards as enhancing accuracy—that a correct result is achieved when state and individual clash. Others look beyond this goal and articulate aspirations for due process in addition to generating acceptable outcomes. Thoughtful commentators such as Jerry Mashaw and Frank Michelman have spoken about aspects of due process that advance individual dignity and efficacy.
In this essay, I suggest a related but somewhat different set of issues for inclusion in our understanding of due process. For me, fairness to the individual and the generation of acceptable outcomes does not exhaust the content of due process theory. These comments are a preliminary effort to articulate another dimension of that theory. I seek to analyze the effects of and upon a third group, perhaps best described as the "public," in the ongoing process of dispute resolution. I am interested both in the role of the public in the process, and the function of the process for the public.