Risk Assessment Tools: An Algorithmic Solution to the Due Process Problem in Immigration

Anu Chugh^[1]*

Abstract

Risk assessment tools have become instrumental in bail reform; the tools increase access to due process by diminishing bias and standardizing decision-making. States across the country have adopted these mechanisms to assist in decision-making in the criminal adjudication process, such as in bond determinations and sentencing proceedings. This Note examines the need for greater due process protections for immigrant detainees during bond determinations and argues that risk assessment tools can serve to address this need. By comparing the Public Safety Assessment, a risk assessment tool used in criminal adjudication, to the Risk Classification Assessment, the risk assessment tool currently used by Immigration and Customs Enforcement, this Note finds that a reformed, independent risk assessment tool administered by Immigration Judges would provide immigrant detainees with increased access to due process.

 

Table of Contents

Introduction 274

I. The Due Process Problem in Immigration 275

A. Supreme Court Jurisprudence on Due Process for Detained Undocumented Immigrants 275

B. Due Process in Practice: The Involvement of Immigration and Customs Enforcement and the Quasi-Judicial Nature of Immigration Courts 279

II. Procedural Due Process: Risk Assessment Tools at Work 283

A. Risk Assessment Tools in the Criminal Context: How Procedural Due Process is Afforded 284

1. What Factors Are Considered by the PSA? 285

2. How Has the PSA Affected Judicial Decisions? 285

3. How Has the PSA Improved Due Process Provisions? 288

B. Risk Assessments in the Immigration Context 291

1. How Does the RCA Work? 291

2. How Does the RCA Differ from the PSA? 292

C. Why Has the RCA Failed to Provide Due Process in Its Current Form? 296

1. The RCA Does Not Satisfy Current Legal Standards of Due Process 296

2. The RCA Has Failed to Reduce the Number of People Detained Because of Their Inability to Post Bond 301

3. The RCA Is Fundamentally Unfair 305

III. Reform of the Risk Classification Assessment 306

A. Changing the Decisionmaker 306

B. Changing the Factors 308

C. Changing the Outcomes 310

Conclusion 311

 

Introduction

Risk assessment tools are an innovation at the forefront of bail reform. Several local and state governments, like in Texas, New Jersey, and New Mexico, have adopted various pretrial risk assessment tools in their criminal adjudication processes, and states including Nebraska, Wyoming, and New Hampshire are considering their use.^[2] In 2013, Immigration and Customs Enforcement (“ICE”) implemented its own risk assessment tool called the Risk Classification Assessment (“RCA”).^[3] Upon the initial introduction of ICE’s tool, a very small percentage of immigrants were actually released from detention, and under the Trump administration, that number dwindled to almost zero.^[4] Scholars have analyzed ICE’s tool at length and roundly criticized its mechanics and use in immigration detention decisions.^[5]

Despite the significant attention the RCA has received, the relationship between risk assessment tools (“RATs”) and due process rights in the immigration context has largely escaped scrutiny. This Note explores that relationship and argues that RATs can improve due process protections for noncitizen detainees. Part I includes a review on relevant due process case law and explains the tenuous administration of due process in the immigration context. Part II examines how RATs, like the Public Safety Assessment, have increased due process protections by diminishing bias and standardizing decision-making. Comparing the Public Safety Assessment to the RCA, and with discussion on the differences between state criminal and federal immigration systems, this Note concludes that RATs in immigration can increase due process protections for immigrant detainees. Part III argues that a risk assessment tool in immigration will have a positive effect on due process if calibrated and used properly. Specifically, this Note proposes that first, a risk assessment tool in the immigration context should be changed substantially to reflect the factors which clearly affect an immigrant’s propensity to return to court and second, immigration judges—rather than ICE officials—must be the primary evaluators of an immigrant detainee’s eligibility for release.

I. The Due Process Problem in Immigration

On an average day, ICE conducts over five hundred removals of individuals who are illegally present in the United States.^[6] These removals are the result of ICE’s Enforcement and Removal Operations group, which targets individuals who have violated U.S. immigration laws and manages the immigration enforcement process, including the stages of identification, arrest, bond determinations, and detention.^[7] While the U.S. Constitution affords due process rights to “persons” generally, the Supreme Court has held that an American citizen’s rights to due process are not equally due to noncitizens in the immigration context.^[8] This section of the Note delves into recent Supreme Court jurisprudence regarding the extent to which noncitizens are afforded due process during immigration hearings. As the law stands today, noncitizens subject to immigration hearings face legislative restraints to their due process rights, even though such restraints are not constitutionally permissible as applied to the due process rights of American citizens.^[9]

A. Supreme Court Jurisprudence on Due Process for Detained Undocumented Immigrants

Today, 8 U.S.C. § 1226(a), a main section of the Immigration and Nationality Act (“INA”), provides for the non-mandatory detention of immigrants facing deportation hearings.^[10] This provision allows the government to either detain a noncitizen pending a removal proceeding, release the individual on a “bond of at least $1,500 with security,” or release them on conditional parole.^[11] Of those detained by ICE during the 2020 fiscal year, 60,994 individuals were subject to non-mandatory detention.^[12] Though there are other provisions of the INA that pertain to the detention of undocumented immigrants, § 1226(a) is the only one that allows for the discretionary release of such detainees.^[13] This Note focuses on due process as it pertains to detainees under § 1226(a).

Though the courts have long held that undocumented immigrants are guaranteed due process by the Fifth Amendment,^[14] the Supreme Court has delivered little guidance on the extent to which due process requires bond hearings for undocumented immigrants facing deportation.^[15] In recent cases, the Court has deferred to lower courts or executive agencies to decide how long an undocumented immigrant may be held and how often an undocumented detainee must receive a court hearing, if at all. In 2001, the Court ruled on a challenge to 8 U.S.C. § 1231(a)(6), which allows the federal government to detain an undocumented immigrant beyond the standard ninety-day removal period.^[16] In Zadvydas v. Davis, Zadvydas was held beyond the ninety-day period because he was stateless and no country would grant him citizenship to allow for his removal from the United States.^[17] The majority ruled that “if removal is not reasonably foreseeable, the court should hold continued detention unreasonable and no longer authorized by statute.”^[18] The Court also held that after a six-month period of detention, once an undocumented immigrant makes a showing that there is “no significant likelihood of removal in the reasonably foreseeable future,” the government must provide evidence to rebut that showing.^[19] In sum, the Court determined that where an undocumented immigrant is held for six months, a hearing must be held to determine the validity of continuing detention.^[20]

Two years later, in Demore v. Kim, the Supreme Court upheld the mandatory detention provision in § 1226(c) against a facial challenge raised in a habeas petition.^[21] Hyung Jyon Kim, a Legal Permanent Resident with South Korean nationality, raised the claim when immigration authorities detained him after his conviction of burglary and petty theft, arguing that his detention “violated due process because the INS had made no determination that he posed either a danger to society or a flight risk.”^[22] Though the Court acknowledged that “the Fifth Amendment entitles undocumented immigrants to due process of law in deportation hearings,”^[23] it held that “Congress may make rules as to undocumented immigrants that would be unacceptable if applied to citizens” because of its broad powers over naturalization and immigration.^[24] Ultimately, the Court held that because “the Due Process Clause does not require [Congress] to employ the least burdensome means to accomplish its goal,” mandatory detention without an individualized hearing under § 1226(c) was permissible under the Constitution.^[25]

Justice Kennedy’s controlling concurrence in Kim provided the fifth vote required to find that Kim’s detention was not a violation of due process, but it did not rule out such violations in other cases.^[26] Justice Kennedy asserted that due process could require “an individualized determination as to [the detainee’s] risk of flight and dangerousness if the continued detention became unreasonable or unjustified.”^[27] He also stated that if a detainee’s deportation proceedings were unreasonably delayed, “it could become necessary” to determine whether the detention was truly aimed at facilitating deportation and protecting against flight risk or dangerousness.^[28] With these provisions, Justice Kennedy’s concurrence would require the courts to use a due process analysis to require individualized bond hearings for immigrant detainees in at least some cases, but there is little guidance as to how often these hearings might be conducted and what must be considered when determining lawful detention.

Most recently, in 2018, the Supreme Court ruled on a challenge to § 1226(a).^[29] In Jennings v. Rodriguez, the Ninth Circuit found that the government must provide undocumented detainees with a hearing every six months in order to re-evaluate their detention.^[30] On certiorari, the Supreme Court applied a strictly textual reading to the statute, holding that nothing in § 1226(a) requires the government to provide a hearing to evaluate an undocumented immigrant’s detention every six months.^[31] Avoiding the due process question, the Court’s reasoning for denying periodic hearings was limited to the plain text meaning of the statute. The statute itself, according to the Court, only requires an “initial bond hearing established by existing regulations.”^[32] It is not specified when this hearing must occur, who must conduct it, or what must be considered during the evaluation of detention.^[33] Therefore, the six-month standard was deemed arbitrary and overruled.^[34] The Court expressly left open the question of what due process the Constitution requires of extended detention in the immigration realm.^[35]

Though it has been established that undocumented immigrants are entitled to due process, the Court has not expressed a clear standard that ICE must meet in order to detain individuals without violating their constitutional rights. In other words, the Court has not articulated whether ICE must provide undocumented detainees with bond hearings, notice of the evidence against them, or legal representation—all of which are commonly regarded as due process requirements in non-immigration contexts.^[36] Furthermore, the Court has held that Congress may legislatively impose procedures on undocumented immigrants’ detention that the Due Process Clause would otherwise bar against American citizens.^[37] The Court’s refusal to limit Congress’s curtailment of due process rights for immigrants makes it clear that ICE may constitutionally detain undocumented immigrants without affording them any semblance of what an American citizen may consider to be due process.

B. Due Process in Practice: The Involvement of Immigration and Customs Enforcement and the Quasi-Judicial Nature of Immigration Courts

Three main decisionmakers exist in the immigration detention context: district directors, immigration judges, and the Board of Immigration Appeals. District directors,^[38] otherwise known as ICE officials, are the actors directly tasked with determining whether an undocumented detainee may be released on bond.^[39] The other decisionmakers, immigration judges and the Board of Immigration Appeals, are secondary actors, and their roles are discussed later in this Section. Per 8 C.F.R. § 1236.1(c)(8), an ICE official may determine whether a detainee is eligible for release based on his own discretion.^[40] An ICE official’s decisions are limited only by the “existing regulations” noted in Jennings, which establish two requirements for a § 1226(a) detainee to be released in a bond determination: first, “the alien must demonstrate to the satisfaction of the officer that such release would not pose a danger to property or persons,”^[41] and second, it must be shown “that the alien is likely to appear for any future proceeding.”^[42] In sum and substance, § 1236.1(c)(8) states that ICE officials have full control over bond hearings, and the administration of due process is within their purview.^[43]

Under its broad mandate of administering bond determinations, ICE may adopt various programs and tools to assist in its decision-making and in the enforcement of those decisions.^[44] In 2013, ICE officials began using the Risk Classification Assessment (“RCA”), an algorithmic risk assessment tool, to evaluate a detainee’s eligibility for release based on factors that measure her risk of flight and possible danger to public safety. Though the RCA was implemented as a reform measure,^[45] it quickly became another tool through which ICE could justify detaining almost every single undocumented immigrant it arrests. By 2015, changes to the RCA’s scoring rubric suspended release for undocumented immigrants who were categorized as risks to public safety based on “recency of entry, recency of removal order, and abuse of a visa or waiver program,” rather than on genuine danger considerations.^[46] After the 2016 election, ICE officials were instructed to “treat all immigration violations alike,” thereby nullifying the difference between individuals who were low, medium, and high risk and condemning them all to detention without bond.^[47] It is important to note that in addition to these policy changes, the RCA was also created with a built-in objection feature, allowing officers and supervisors to depart from the RCA’s recommendation and implement a different outcome.^[48] Though the RCA remains part of an ICE official’s toolbox in conducting bond hearings, it does not currently outweigh, or even inform, the discretion allotted to ICE in § 1226(a).

A secondary decisionmaker in the immigration context is the immigration judge (“IJ”). IJs do not derive their authority from Article III of the U.S. Constitution. Instead, as dictated by Congress, IJs are appointed and overseen by the Department of Justice (“DOJ”), thus making them agents of the executive branch.^[49] As Article I judges, IJs preside over administrative courts, where they “examine and determine various matters, arising between the government and others, which from their nature do not require judicial determination.”^[50] However, the matter at hand, the detention of noncitizens, requires the enforcement of due process; IJs are susceptible to review by federal judges and therefore are more likely than ICE officials to be arbiters of procedural due process.^[51] About 460 IJs are managed by the Office of the Chief Immigration Judge, a part of the DOJ that “establishes operating policies and oversees policy implementation for the immigration courts.”^[52] Apart from executive oversight, IJs are further limited by statute. 8 C.F.R. § 1236.1(d) states that the authority of an IJ to make custody and bond decisions is reserved for the appeal of decisions made by a district director.^[53] Therefore, IJs are only limited to reviewing decisions rather than making their own determinations. Though IJs are statutorily empowered to “exercise their independent judgment and discretion and may take any action consistent with their authorities,” they are also bound to carry out the duties delegated to them by the Attorney General.^[54] IJs, though judges in name, are thus enforcers of executive policy.

This direct subordination to the Attorney General raises concerns over the independence and impartial nature that judges are generally expected to maintain. It also poses issues of executive interference in matters that are considered to be judicial. For example, in 2018, Judge Morely of the Philadelphia Immigration Court administratively ended proceedings because he believed the defendant in the case, an unaccompanied minor, was not receiving notices about his court dates, which in turn created a due process problem.^[55] On appeal from this decision, the Board of Immigration Appeals, the third decisionmaker also known as “Appellate Immigration Judges,”^[56] remanded the case, ordering Judge Morely to schedule another hearing and ordering the defendant removed in absentia if he failed to show.^[57] The new hearing was scheduled twelve days later; instead of administratively closing the case at this hearing, Judge Morely issued a continuance on the basis that twelve days was not enough notice to afford due process.^[58] The DOJ removed Judge Morely from the case and assigned a new IJ to hear the case.^[59] Attorney General Sessions then certified the case for his own review and categorically denied IJs the authority “to suspend indefinitely immigration proceedings by administrative closure.”^[60] This decision was met with opposition from IJs. The Vice President of the Immigration Judges’ union, Judge Amiena Khan, said that the removal of Judge Morely was “another transparent way, surprisingly transparent in this instance, for the agency to come in and re-create the ideology of this whole process more towards a law enforcement ideology.”^[61]

The three decisionmakers in the immigration context are subject to direct interference from the executive branch. ICE officials must comply with the policies issued by their superiors. IJs, though independent in theory, are easily overruled by their superiors as well, and the Board of Immigration Appeals also serves under the Attorney General of the United States.^[62] The quasi-judicial nature of IJs and the way in which ICE officials make bond determinations demonstrate a lack of due process for immigrants subject to deportation. Though immigration detention has been intentionally placed within the purview of the executive, the judiciary has found that undocumented immigrants are entitled to certain constitutional rights, such as due process.^[63] In light of this reality, tools to safeguard due process are imperative. The RCA, though a risk assessment tool on its face, is not an independent source of decision-making information that ICE officials can properly rely on. Therefore, ICE must alter and implement the RCA in a way where due process is not easily curtailed.

II. Procedural Due Process: Risk Assessment Tools at Work

Due process is a right afforded to all American citizens and to all those charged with criminal offenses, but less process is extended to immigrant detainees. Criminal defendants are guaranteed due process in two forms by the Bill of Rights: the Fifth, and Fourteenth, Amendment guarantee substantive due process,^[64] while the Sixth Amendment establishes procedural due process.^[65] These amendments include the right to a speedy trial, the right to a jury, the right to counsel, and the right to confront opposing witnesses.^[66] Recently, some courts have found that defendants are also entitled to procedural due process rights when bail is set.^[67] These rights include “a hearing at which the arrestee has an opportunity to be heard and to present evidence,” as well as “an impartial decisionmaker” and “timely proceedings.”^[68] As a reform measure, risk assessment tools (“RATs”) have been used in the criminal context for pretrial detention and sentencing determinations, aiming to reduce the number of people held because they cannot afford to pay bail and to “divert lower-risk defendants to reduced or alternative sentences.”^[69] This Section discusses how RATs have increased procedural due process for criminal defendants and how they have helped decrease the number of defendants detained before trial, which sharply differs from the RCA in its current form. This Section concludes with the argument that the RCA must be changed in order to provide immigrant detainees with a degree of procedural due process that is not currently afforded to them.

A. Risk Assessment Tools in the Criminal Context: How Procedural Due Process is Afforded

In the criminal justice system, RATs are currently in use at the local, state, and federal levels.^[70] Each of the fifty states has implemented a risk assessment program to assist in decision-making for some part of the criminal adjudication process, whether it is for pretrial release, probation or parole supervision, prison programming, or presentencing.^[71] The specific tools used by states or localities vary by jurisdiction; most use one of several independently designed algorithms, while others have created their own programs.^[72] One of the most widely used RATs is the Public Safety Assessment (“PSA”), developed by the Laura and John Arnold Foundation.^[73] The Foundation, now known as Arnold Ventures, designed the PSA with the following goals in mind: to use jail “only when absolutely necessary,” to enhance public safety, and to limit “the costs of incarceration on families and communities.”^[74] Over two hundred jurisdictions across the country implemented this tool for use in pretrial release decisions.^[75] As discussed below, it has improved individuals’ access to procedural due process by improving the information available to judges and ensuring that fewer people are incarcerated simply because of their inability to post bond.

1. What Factors Are Considered by the PSA?

The PSA uses nine factors to assign two risk scores to defendants. Both scores fall on a scale from one to six, with the first score predicting the likelihood of a defendant committing a new crime while on pretrial release and the second score assessing the likelihood of that defendant failing to appear for a future court date.^[76] The factors considered are: age at the time of arrest, whether the current offense was violent, if pending charges existed at the time of arrest, whether the defendant had prior misdemeanor or felony convictions, whether the defendant had prior violent convictions, prior failures to appear (weighted differently if it occurred within the last two years), and whether the defendant had ever been incarcerated.^[77] According to Arnold Ventures, these factors help “jurisdictions shift away from a decision-making process based on a defendant’s financial resources to one that prioritizes a defendant’s risk of pretrial failure.”^[78]

2. How Has the PSA Affected Judicial Decisions?

The PSA has altered judicial decision-making by standardizing information about defendants and reducing judges’ discretion. To illustrate this point, consider the bail-setting process in Harris County, Texas before and after the County implemented the PSA in 2017. In recent years, Harris County has been the subject of bail reform litigation, and several changes to the Harris County criminal justice system have occurred over the last several years.^[79] Prior to 2017, Harris County judges would consider a report prepared by Pretrial Services^[80] that included information regarding employment, monthly expenses and income, family members, and prior arrests.^[81] Judges would also consider a risk score based on factors such as whether the defendant owned a vehicle, had a home telephone, and the equivalent of a high school diploma, as well as active parole or probation status.^[82] Based on all of this information, Pretrial Services would provide a recommended amount of money bail to the judge for her approval.^[83] Following the setting of a bail amount, the defendant would either agree to pay or remain in jail because of inability to put up funds.^[84] In this context, a judge had a great deal of discretion; depending on what demographics were provided, she was able to give more weight to certain facts (such as type of employment or place of residence) and decide whether a defendant could be released on personal bond, no matter how low the bail recommendation might be.^[85]

The problems with this system were that first, the “assessment was focused on ties to the community, which [are] not necessarily predicative of risk for failure to appear, or risk of committing another crime;”^[86] second, there were informal rules about which groups of people might be denied bond based on certain demographics;^[87] and third, the Pretrial Services reports often did not include the same information about each defendant.^[88] After the PSA was implemented in 2017, Harris County Pretrial Services “stopped using its interview-based assessment process” and focused on gathering the information required by the new assessment tool.^[89] This not only standardized the information provided to a judge but also reduced the amount of time required to interview a defendant after his arrest.^[90] Now, judges in Harris County assign one of three outcomes to a defendant: release without bail (on the promise that the defendant will return for his court dates), release on cash bond (where applicable),^[91] or pretrial detention.^[92] Though judges may still exercise their discretion on occasion, pretrial detention in Harris County may now only be ordered if the PSA renders a flag for new violent criminal activity or generates a score of six for either the risk of new criminal activity or failure to appear.^[93] Detention is no longer ordered simply because of a lack of funds, job, home, or landline.^[94] Informal rules, such as requiring homeless defendants to provide a residential address in exchange for pretrial release, can no longer exist.^[95] In addition,

[t]hough the PC [probable cause] court judge can still refuse to set a PR [personal recognizance] bond despite the PSA’s recommendation, the fact that the PSA recommends PR bonds in a large number of cases means that many more people are getting released on PR bonds than ever before. The end result is fewer poor people are sitting in jail just because they are poor.^[96]

Likewise, in jurisdictions where the PSA has been adopted, like in Cook County, Illinois, and the states of New Jersey and Arizona, the tool’s nine standardized factors limit judicial discretion by invalidating informal rules and improper measures of risk.^[97]

3. How Has the PSA Improved Due Process Provisions?

First and foremost, for bail proceedings at the state level, the PSA ensures that defendants are judged using a standardized set of factors, all of which have been selected specifically to assess a defendant’s propensity to reoffend or fail to appear in court.^[98] Other than age, the PSA does not rely on “race, gender, ethnic background, income, substance abuse, mental health, employment status, marital status, or any demographic or personal information.”^[99] The PSA eliminates the fragmented data that jurisdictions like Harris County have relied upon in the past to make bail determinations because it relies on less information than traditional systems.^[100] All of this information can also be accessed from sources other than the defendant.^[101] Secondly, the PSA eliminates a judge’s reliance on recommendations made by auxiliary groups, like Harris County Pretrial Services.^[102] Although this group still functions as an information-gatherer for the bail determination process, it no longer provides its own risk assessment recommendations to the judge for approval.^[103] Instead, Pretrial Services generates risk scores directly through the PSA, and plugs them into whichever bail or release schedules are provided by the state.^[104] The judge then issues a determination based on those facts and her intuition.^[105] Finally, the PSA reduces the amount of time it takes for a jurisdiction to issue a bail determination, in turn reducing the amount of time an arrestee may be detained before seeing a judge.^[106] Ultimately, in terms of procedural due process, the PSA standardizes the evidence going before a judge, thereby eliminating discrepancies in information, increasing a judge’s impartiality by relying on risk-related data rather than demographic data (such as race, income, etc.), and increasing the timeliness of a determination.

Critics of the PSA argue that the tool does not remove bias from the judge’s ultimate decision^[107] and that it may not satisfy the requirement of an individualized hearing prescribed in the Supreme Court’s holding in Stack v. Boyle.^[108] In particular, Sandra Mayson and other critics argue that RATs propagate racial inequality through the nature of prediction itself,^[109] are a better predictor of arrest than crime,^[110] and contribute to the lower threshold of risk required to justify detention for defendants with pending charges.^[111] Though it is true that inherent bias continues to pervade the criminal justice system through the use of RATs, tools like the PSA move the system toward the regulation, modernization, and standardization of some judicial decision-making. In addition, the PSA’s developers have announced that “[a]ll studies to date show the PSA does not exacerbate racial disparities.”^[112] Although the PSA issues risk scores based on algorithms rather than a holistic analysis of an arrestee’s human nature, it does serve the purpose of an individualized hearing.^[113] Like interview-based assessment schemes, the PSA measures the odds of an individual reoffending or failing to appear in court.^[114] Judges are able to consider a research-based analysis of a particular defendant’s likelihood to reoffend or fail to appear;^[115] the PSA’s risk-based predictions focus on an individual’s history and compare it to a studied aggregation of criminal histories, failures to appear, and offenses while released on bail.^[116] In other words, the PSA determines an individual’s risk score by comparing their nine factors to those of 1.25 million other individuals who have made their way through the criminal system in multiple jurisdictions.^[117]

Furthermore, the PSA was developed by researchers who gathered the “most comprehensive dataset of pretrial cases ever assembled in the United States with the goal of developing a universal risk assessment.”^[118] Hundreds of risk factors were considered when developing the tool, and ultimately nine were found to be the most predictive of new crime, new violence, and failure to appear.^[119] To validate the PSA, researchers selected 500,000 cases from different jurisdictions where the outcome was already known, input the relevant information into the tool, and documented the similarities between the actual outcome and the PSA’s outcomes. According to Arnold Ventures, the PSA was successfully validated and “the results confirmed that the assessment does not over-classify non-whites’ risk levels.”^[120] Ultimately, though this risk score is determined by a mathematical analysis, it is individualized because it considers specific details about a person’s history, is an objective prediction of that person’s risk level, and is still subject to a judge’s ultimate review and interaction with the defendant.

B. Risk Assessments in the Immigration Context

The federal government employs the RCA tool to decide whether immigrant detainees are eligible for release but, as discussed earlier, the actual number of detainees who are released from ICE custody is extremely low.^[121] In addition, the current use of the RCA propagates the denial of due process that is deeply embedded in the American immigration system. Therefore, the RCA in its current form is ineffective. It should be reformed to not only focus on risk-oriented factors, like the PSA, but the overall administration of the tool must be shifted into the hands of the IJ.

1. How Does the RCA Work?

Information about the factors used in the RCA comes directly from the Department of Homeland Security (“DHS”), which published details about the tool in April 2012, just before the RCA’s 2013 rollout.^[122] DHS made it clear that the RCA does “not change the information that ICE collects about aliens.”^[123] The information gathered by ICE officials for input into the RCA includes: “disability or status as a crime victim; substance abuse history; immigration history and case status; ties to the community, including the length of time at current address, the number of family members residing with the undocumented immigrant, and property ownership; and authorization to work or enrollment in school.”^[124] This data produces two measurements—one for public safety risk and the other for flight risk—each weighed on a scale of low, low/medium, high/medium, or high.^[125] The RCA then combines the two measurements to generate one of four recommendations: “release, supervisor to determine, detain-eligible for bond, or detain in the custody of DHS (i.e., no bond set).”^[126] Ultimately, ICE officials retain authority over the final custody decision and are able to override any recommendations provided by the RCA.^[127]

2. How Does the RCA Differ from the PSA?

At face value, the RCA seems to function much like the PSA. However, several stark differences exist between the two. First, the RCA is administered by an ICE official, not an IJ.^[128] As discussed in Part I, ICE officials retain full authority over initial bond determinations.^[129] They do not conduct formal hearings, as IJs do, and therefore do not provide detained immigrants with a chance to present their case against detention or the setting of bond.^[130] Whereas Pretrial Services gathers all required information and has PSA recommendations reviewed by a judge, ICE officials only collect information they deem to be relevant and make decisions themselves based on the RCA.^[131] In other words, ICE officials are both the information-gatherers and the decision-makers, allowing them to influence the RCA directly by exercising their discretion over which details in a detainee’s life are relevant enough to be considered by the tool. Because the RCA determines a detainee’s public safety risk using a scoring rubric that relies “principally on static criminal history pulled from other databases, combined with a few additional inputs determined by the ICE officer,” an ICE official’s determination of which facts are important becomes critical to the RCA’s output.^[132] The facts within the ICE official’s discretion generally fall into the “special vulnerabilities” category, which includes a determination of whether a detainee has serious physical or mental illness, is pregnant/nursing, or has been a “victim of harm.”^[133] If a detainee has open warrants in the United States, it is also up to the ICE official to decide how violent the crimes alleged in those warrants might have been.^[134] The DHS has not precisely defined the meaning of “violent” as applied to these situations.^[135] Ultimately, the decision-maker using the RCA has a great deal more influence over the tool’s recommendation than the decision-maker using the PSA.

Second, unlike the PSA, ICE and DHS easily manipulate and change the RCA.^[136] From October 2012 to October 2015, DHS altered the risk levels produced by the RCA several times, allowing for “factors that once generated a medium public safety risk [to] later generate[] a high public safety risk.”^[137] In 2014, DHS changed the RCA’s scoring rubrics for public safety and flight risks, increasing the likelihood of recommendations to detain individuals with a criminal history and simultaneously making it easier to release individuals with community ties and no criminal history.^[138] In order to do so, many “property crimes, immigration crimes, and business crimes generated lower public safety scores,” as they were reduced from moderate to low-level offenses, whereas crimes such as domestic violence, most drug possessions, and prostitution were increased from low to moderate-level crimes.^[139] Essentially, because of these changes, “a low public safety risk assessment was impossible with any criminal history unless a sole offense was rated as low severity or was moderate severity and was at least five years old.”^[140] Scores for certain community ties were changed in 2014 as well, increasing a flight risk determination for an individual who had “previously fled or avoided removal” or violated conditions of supervision in a prior immigration case.^[141] A flight risk score was lowered if an individual lived with immediate family members, had lived at a stable address for six months or more, or owned property in the community.^[142] Together, these changes to the public safety and flight risk scores reflected the PSA’s arrangement: those with low public safety risks and low/medium flight risks were recommended for release, those with low public safety risks and high flight risks were recommended for a supervisor determination of release or detention, and those with high public safety scores and medium/high flight risks scored received a detention/no bond recommendation.^[143]

Manipulation of the RCA’s scoring rubric did not end in 2014. In 2015, DHS made changes to “align the RCA with the new prosecutorial priorities.”^[144] The most significant changes made at this time were to the flight risk algorithm.^[145] The algorithm was designed to no longer take an individual’s immigration history into account; “date of entry was the prime determinant of flight risk level,” with those entering after January 1, 2014 receiving an almost immediate detention recommendation.^[146] Factors such as fleeing removal or violating conditions of release, as well as having pending benefits applications or work authorizations, are no longer considered.^[147] These changes heavily reflected the directives issued by the White House at the time, and “made a few new politically-based factors determinative of [an immigrant’s] risk level.”^[148] In 2017, in response to the Trump administration’s directive to detain undocumented immigrants, ICE changed the RCA by removing the release option altogether.^[149] Though the RCA may still recommend bond in certain cases, the ultimate outcome is that undocumented immigrants captured by ICE are detained without the possibility of release.^[150] Though the implementation of the PSA may vary based on jurisdictional definitions of each risk level, the RCA is especially vulnerable to arbitrary manipulation that is arguably unrelated to actual risk. The RCA, therefore, is different from the PSA because it is easily manipulated and changed by ICE, which responds to the political directives of the executive branch.

Finally, the RCA differs from the PSA because it is not an independently created or verified risk assessment. The PSA was created by Arnold Ventures, an independent philanthropic organization.^[151] The organization spent five years researching and testing its risk assessment tool, partnering with select jurisdictions after its development and providing the tool for free in the aim to lower pretrial jail populations.^[152] In addition, independent researchers continue to validate the PSA “to maximize its accuracy and minimize its impact on racial disparities.”^[153] The RCA, on the other hand, was developed by DHS itself. Urged by the 2009 Schriro Report,^[154] ICE developed the RCA and implemented it in 2013.^[155] There are government reports on the RCA, such as the one issued by the Office of the Inspector General (“OIG”) in February 2015,^[156] as well as independent studies conducted by scholars in the field, such as Robert Koulish, Mark Noferi, and Kate Evans.^[157] The issue, however, is that independent researchers are hindered by the government’s concealment of ICE data.^[158] In their studies, Koulish, Noferi, and Evans had to conduct “multiple rounds of litigation and negotiation” to receive information consistent with their requests under the Freedom of Information Act.^[159] To the public, and even for these researchers, the methodology behind the RCA remains hidden to a certain degree.^[160] The RCA, unlike the PSA, is not a transparent, accessible tool, which obstructs independent verification and long-term improvement.

C. Why Has the RCA Failed to Provide Due Process in Its Current Form?

Although scholars such as Koulish, Noferi, and Evans have provided in-depth reviews of the RCA and how it works, few have acknowledged how the RCA currently impacts a noncitizen’s due process rights and, if improved, how it might impact due process in the future.^[161] It is important to understand how the RCA and similar tools might help noncitizen detainees, especially in the realm of due process. The introduction of risk assessment tools into ICE’s detainment process raises questions of effectiveness, fairness, and what the ultimate goal in using such tools may be. Though risk assessment tools used in the immigration context generally cannot improve a detainee’s access to legal counsel or ensure that a detainee’s case is handled in a speedy manner, they can help an immigrant detainee present evidence to the decisionmaker and can increase the impartiality of a decisionmaker’s ultimate holding.^[162] RATs can provide such benefits, and others, so long as they are properly designed and implemented.^[163] Though the RCA was created to “help serve as a technological palliative to the problem of draconian over-detention,”^[164] it has failed to improve the procedural component of the detention process nor the over-detention itself.

1. The RCA Does Not Satisfy Current Legal Standards of Due Process

In cases like Demore v. Kim and Jennings v. Rodriguez, the Supreme Court has affirmed vague due process requirements for immigrant detainees, such as initial bond hearings^[165] and individualized proceedings.^[166] Lower courts, however, have gone a step further and clearly articulated some due process requirements, including the consideration of a noncitizen’s financial circumstances when setting bond, deliberation of alternative conditions of release,^[167] and “individualized bond hearings before a neutral decisionmaker.”^[168] Though the RCA facilitates these requirements in theory, it fails to do so in practice.

In Hernandez v. Sessions, the Ninth Circuit considered a class action suit brought by immigrants detained under 8 U.S.C. 1226(a).^[169] The class, represented by named plaintiffs Hernandez and Matias, sought relief against the government’s failure to require the consideration of a detainee’s financial circumstances and alternative conditions of release at bond determinations.^[170] Applying Zadvydas, the court held that the Due Process Clause of the Fifth Amendment protects noncitizens from a deprivation of liberty without due process of law.^[171] The Ninth Circuit explained that “due process requires adequate procedural protections to ensure that the government’s asserted justification for physical confinement outweighs the individual’s constitutionally protected interest in avoiding physical restraint.”^[172] In addition, citing the Fifth Circuit’s holding in Pugh v. Rainwater, the Ninth Circuit affirmed that it is impermissible for the government to detain an indigent person for his inability to post bond if the indigent’s appearance at future proceedings could be reasonably guaranteed by an alternative form of release.^[173] Therefore, because ICE officials found that Hernandez and Matias were neither dangerous nor a significant flight risk, the government’s failure to consider their financial circumstances or alternative modes of supervised release constituted a violation of the plaintiffs’ due process rights under the Fifth Amendment.^[174]

Furthermore, in Padilla v. ICE, the Ninth Circuit discussed noncitizens’ due process rights to an “individualized bond hearings before a neutral decisionmaker.”^[175] This case pertained to immigrants detained under 8 U.S.C. § 1225(b), which provides for the expedited removal of noncitizens arriving at a port of entry or who are unlawfully residing in the United States and cannot prove that they have been in the United States for more than two years.^[176] The Padilla holding applies a fortiori to those detained under § 1226(a) because of its extensive discussion of adequate due process for immigrant detainees.^[177] Citing Zadvydas again, the Ninth Circuit affirmed noncitizens’ due process rights to “avoid physical restraint” unless detention “bears [a] reasonable relation to [its] purpose.”^[178] The court, discussing the Supreme Court’s holding in United States v. Salerno, noted that civil confinement may only be justified if there is a “full-blown adversary hearing” where there must be factual findings, specific explanations, or deliberation over presented evidence.^[179] The Ninth Circuit found that those arrested by ICE are entitled to a bond hearing when they are subjected to detention, whether such detention is mandatory or not, and a “full-blown” hearing must be conducted by an “impartial adjudicator.”^[180] Importantly, the Ninth Circuit made note of the government’s motives in detaining noncitizens, finding that “[d]ue to political and community pressure, the INS . . . has every incentive to continue to detain” and an impartial adjudicator is therefore necessary.^[181]

The RCA, then, must facilitate the procedural guarantees that ICE must provide to noncitizens in order to preserve their due process rights. This tool should help an impartial adjudicator—who must consider factual findings or specific explanations as well as financial circumstances and alternatives to detention—determine whether a noncitizen poses a great enough danger to public safety or a great enough flight risk to justify his civil detention by ICE.

However, the way in which the RCA is currently used does not satisfy these due process requirements. First, the RCA is administered by an ICE official, who works for the DHS, which is led by a political appointee who answers to the White House.^[182] The ICE official’s administration of the RCA does not guarantee impartiality.^[183] In fact, as the Ninth Circuit noted, it is likely that the official will be incentivized to detain noncitizens instead of release them.^[184] Second, the “hearings” conducted by ICE officials are not adequate per Padilla and Salerno.^[185] Though ICE officials may “ask each alien as many as 178 RCA questions during intake processing,” it is not clear how ICE verifies any factual assertions made by an immigrant detainee.^[186] Furthermore, in a report conducted regarding the RCA’s efficacy, the OIG found that “the RCA was not intended to predict factors such as . . . a detainee’s ability to pay bond.”^[187] Though the RCA was initially designed to offer release or community supervision recommendations, these alternatives to detention were eliminated by the Trump administration.^[188] Therefore, the RCA does not provide ICE officials with proper alternatives to ICE detention. As a result of these failures, immigrant detainees are denied adequate due process in the RCA’s current form.

One important question to be asked is whether risk assessment tools, including the PSA and the RCA, impair due process. While no federal cases challenging the validity of the RCA on due process grounds have been brought, some have been brought on the state level. In the criminal context, the Supreme Court of Wisconsin ruled that risk assessments do not infringe on a criminal defendant’s due process rights.^[189] In State v. Loomis, the Supreme Court of Wisconsin heard a challenge to an algorithmic risk assessment tool called COMPAS, which was used at the defendant’s sentencing.^[190] Loomis challenged the use of the tool on three grounds:

(1) it violates a defendant’s right to be sentenced based upon accurate information, in part because the proprietary nature of COMPAS prevents him from assessing its accuracy; (2) it violates a defendant’s right to an individualized sentence; and (3) it improperly uses gendered assessments in sentencing.^[191]

The Court held that the use of a RAT at sentencing did not violate the defendant’s right to be sentenced based on accurate information^[192] or his right to an individualized sentence,^[193] nor did the assessment’s consideration of his gender affect his due process rights.^[194] In addition, in Holland v. Rosen, the Third Circuit held that the PSA does not violate procedural due process as used in New Jersey.^[195] Holland, the defendant in a criminal proceeding, alleged that the PSA violated his Fourteenth Amendment right to procedural due process by enabling the state to impose home detention and electronic monitoring on defendants “without having the option to impose monetary bail together with or in place of these non-monetary conditions.”^[196] After considering the lower priority of money bail to non-monetary conditions and the availability of modifications to bail conditions in the event of changed circumstances, the Third Circuit found that “the extensive safeguards provided by the Reform Act are not made inadequate by its subordination of monetary bail,” thereby establishing that the PSA does not violate procedural due process.^[197]

On the other hand, in Nevada, the Court of Appeals found that a prison inmate’s due process rights were violated when the parole board failed to inform him of the new RAT they used or its findings when revoking his parole.^[198] The Nevada Court of Appeals held that “[d]ue process requires disclosure of the evidence used against a person in a parole revocation hearing; therefore, Ramirez was a person ‘entitled to receive such information.’”^[199] This challenge amplified the government’s responsibility to inform a defendant about the use of RATs in certain circumstances but did not affect the use or design of the tool itself. Other cases have been brought to challenge the use of RATs, but the respective courts have refused to rule on the due process question because the appellee failed to preserve the due process question at the district level.^[200] While these cases consider the effects of risk assessments to a certain degree, they do not discuss the potential benefits in detail. As a result, no existing case law disqualifies the RCA, modified from its current form, from being a positive source of change to the availability of due process rights for immigrant detainees.

2. The RCA Has Failed to Reduce the Number of People Detained Because of Their Inability to Post Bond

One of the goals in adopting a risk assessment tool is to reduce the number of people detained because of their inability to post bond.^[201] For states like New Jersey and New Mexico, adopting the PSA was a criminal justice reform aimed at decreasing the number of people detained because of their financial circumstances, in turn promoting fairness and equality in the outcome of pretrial hearings.^[202] As a result of the implementation of the PSA, there was a substantial decrease in the number of people incarcerated pretrial in New Jersey, as well as a reduction in the amount of time that defendants were detained before their initial pretrial release.^[203] In New Mexico, the same effects were seen as a result of the state’s adoption of the PSA.^[204] This reduction facilitates due process not only because it ensures that defendants are not held simply because of their inability to pay for their release, but also because it increases the timeliness of a bail hearing and allows people to collect evidence on their own behalf more freely, possibly with the help of an attorney. Though the RCA could also produce lower populations of detained noncitizens, thereby reducing the time these individuals remain incarcerated and improving their ability to represent themselves in a deportation proceeding, its current design does not allow for these reforms.

New Jersey’s use of the PSA has proved successful in reducing the number of people detained because of their inability to post bond. In New Jersey, “one of the driving factors in the decline in New Jersey’s pretrial jail population is that, [using the PSA], defendants spent half as much time in jail from the time of commitment to when they are initially released pretrial.”^[205] Compared to data from its money bail system in 2014, New Jersey’s implementation of the PSA reduced the population of its jails by a few thousand defendants per day in 2017.^[206] This means that there were over 750,000 fewer beds occupied by those awaiting trial over the course of the full year, in comparison to jail populations in 2014.^[207] Because of the PSA, New Jersey’s courts were able to release defendants on their own recognizance or under supervised conditions more quickly than before, reducing the amount of time a defendant was detained pretrial by 40%.^[208] According to Arnold Ventures and the New Jersey court system, “overall, New Jersey’s pretrial jail population had declined 43.9% from the end of 2015 to the end of 2018.”^[209]

In New Mexico, early studies of the PSA’s effects show similar results. In a study of Bernalillo County’s pretrial outcomes after the PSA was implemented in June 2017, the Institute for Social Research at the University of New Mexico found that the percentage of individuals detained before trial decreased from 566, the majority of whom “had a cash bail or bond ordered, but did not post it,” to 344, all of whom were detained because a judge granted the government’s motion for preventative detention or because of the PSA’s recommendations.^[210] In addition, the PSA’s recommendations for detention decreased from 47.7% to 36.9% between December 21, 2017 and June 30, 2020.^[211] On the whole, Bernalillo County’s implementation of the PSA has substantially decreased the number of defendants who are kept in jail simply because of their inability to pay bail.^[212] As other counties in New Mexico adopt the PSA, future studies will likely show the same outcomes.

The RCA, on the other hand, does not produce such outcomes, thereby failing to ensure that government detention is justified and that detained noncitizens are not simply imprisoned because they cannot post bail. According to ICE’s own Enforcement and Removal report for the 2019 fiscal year, “there were 226,400 removals with detention involved, a slight increase from 209,928 such removals in FY 2018.”^[213] This equates to about 85% of removals by ICE’s Enforcement and Removal Operations (“ERO”) in 2019.^[214] ICE’s explanation for this rate of detention is as follows:

Because ERO’s limited detention beds account for only a small fraction of those who are amenable to removal, the agency’s detention resources are primarily focused on individuals who represent a threat to public safety, for whom detention is mandatory by law, or who may be a flight risk.^[215]

By this logic, at least 85% of noncitizens who have been arrested by ICE are detained because they may be a flight risk or a threat to public safety.

However, because the RCA has been altered to no longer offer a release recommendation, there is no clear set of factors that designates a noncitizen as a flight risk or as a threat. It is also unclear whether the remaining 15% of arrestees are released because they are able to post bail, their RCA recommendation is reversed by an ICE official, an immigration judge orders their release, or for some other reason. While there is no consistent information regarding ICE’s detention populations, “the average daily population of detained immigrants increased from approximately 7,000 in 1994, to 19,000 in 2001, and to over 50,000 in 2019.”^[216] In 2019, the average daily population in ICE detention was 50,165 and 510,854 people were detained for the year.^[217] Compared to the 2018 fiscal year, ICE’s average daily population in detention facilities increased by 19%.^[218]

Generally, RATs have been adopted as measures of bail reform.^[219] In states like Texas, New Jersey, and New Mexico, RATs have been used to reduce the number of people behind bars before their cases are resolved,^[220] directly amending the due process violations of detaining those who cannot pay for release and those who are found to be flight risks or threats based on partial or irrelevant information. The RCA, too, was adopted in response to the Schriro Report, which recommended that “[t]he ideal system should create the capacity to detain and to supervise aliens consistent with assessed risk” as one of its seven areas of focus for reform.^[221] Therefore, the RCA should also function as a tool of bail reform and amend the due process violations ICE commits against its many detainees. Instead, however, it has been altered in such a way to increase the number of people incarcerated before their deportation hearings or removal without proper justification.^[222]

States have adopted a more rigorous statistical risk assessment program for pretrial bail determinations because judges had been overinflating their sense of the risk of pretrial violence, sending people to jail on this basis.^[223] State adoption of RATs has led to significant reductions in detention.^[224] ICE’s use of the RCA, on the other hand, does not reduce the agency’s mass detention of noncitizens. Though the OIG has not specified such a goal, merely stating that ICE adopted the RCA “to assist its release and custody classification decisions” and “to promote consistency and transparency in detention-related decision making and to better align field office decisions with ICE policies and priorities,”^[225] the Schriro report emphasized that “ICE should ascertain each participating alien’s need for supervision” by developing a risk assessment tool that determined an appropriate supervision strategy based on factors such as propensity for violence.^[226] Because the RCA was designed with need-based supervision and detention in mind, it should have led to the reduction of detention populations as did state RATs. Instead, the RCA has been subject to arbitrary manipulation and therefore is not a valid component of a noncitizen’s due process.^[227]

3. The RCA Is Fundamentally Unfair

The RCA fails to rectify due process violations because it does not adequately justify the civil detention of noncitizens, does not provide ICE arrestees with the proper opportunity to make factual showings about their circumstances, is not evaluated by a neutral decisionmaker, and is too easily manipulated by political actors. As previously discussed, RATs have been adopted as reforms in bail and pretrial detention.^[228] They are intended to help decisionmakers make fairer determinations, as well as to ensure that detainees are assessed based on accurate, relevant information.^[229] In 2015, even before the Trump administration dismantled the release option, the OIG announced that “the tool is time consuming, resource intensive, and not effective in determining which aliens to release or under what conditions.”^[230] The RCA not only fails to meet due process requirements but also does not meet the expectations that ICE established when it implemented the tool. Because the tool is ineffective as designed, easily manipulated, and improperly utilized, it is fundamentally unfair. To be a proper source of due process protections, it is imperative that the RCA be reformed as discussed below.

III. Reform of the Risk Classification Assessment

Having assessed the functionality and implications of RATs in the state criminal and federal immigration contexts, this Part focuses on potential reforms for the RCA to improve and guarantee procedural due process protections for immigrant detainees in the long term. Three possible avenues of change are presented, each of which would improve the RCA on its own, but which together would completely overhaul the current design of the tool and prevent manipulation from political forces. These changes include the delegation of RCA administration to IJs, the alteration of the factors considered for the RCA’s outcome, and the addition of various RCA outcomes.

A. Changing the Decisionmaker

In order to promote fairness to the greatest extent, IJs must be the main government entity to use the tool. Currently, ICE officials are the administrators of the RCA, and the tool’s outcomes are not even shared with the IJ during appeals of bond determinations.^[231] As discussed previously, ICE officials answer directly to those who are politically appointed to conduct the business of the DHS and are not impartial decisionmakers.^[232] ICE’s use of the RCA has been heavily influenced by political forces, as reflected in the many changes the tool has undergone in recent years.^[233] Therefore, less partial decisionmakers must be employed.

Though IJs are also appointed by the executive branch, there are two distinct differences between their decision-making and that of an ICE official. First, IJs must ensure that due process is satisfied.^[234] Pursuant to guidance set out by the Ninth Circuit, “[i]mmigration proceedings, although not subject to the full range of constitutional protections, must conform to the Fifth Amendment’s requirement of due process.”^[235] IJs preside over these “immigration proceedings”^[236] and must fulfill requirements including “factual findings, . . . specific explanations, [or] deliberation” over presented evidence^[237] in such proceedings. Therefore, though IJs are given their powers through Article I of the Constitution rather than Article III, they must still abide by the constitutional provisions that set out due process rights.^[238] In fact, IJs are subject to review by federal courts, which sets out a path through which IJ determinations can be overruled for violations of due process.^[239] ICE officials, on the other hand, are not subject to the same level of review. Though these officials must abide by due process requirements in theory, their decisions are not directly reviewable by an independent judiciary and enforcement of due process in this context is difficult.^[240] Because IJs are constitutionally required to provide such protections to those undergoing immigration proceedings and their decisions can be reviewed by Article III judges, IJs are better positioned to make fair determinations about release or detention.

Second, these judges do not report directly to ICE supervisors.^[241] Though IJs are answerable to the Attorney General, they are not under the constant purview of direct policymakers.^[242] They are not subject to any regular policy changes made within ICE itself, and any major changes to an IJ’s authority must be made by the Attorney General through the adjudication of a case.^[243] IJs have the discretion to change decisions made by ICE officials, but currently, “bond hearings do not provide any meaningful review of the initial decision to detain, but rather serve only to review the bond amount set.”^[244] If IJs were given the task of administering the RCA, they would have full access to the facts surrounding an immigrant’s presence in the United States, the factors influencing flight risk and public safety risk scores, as well as the immigrant’s financial circumstances. Because they are less influenced by ICE policy and bound by due process requirements, IJs would be better suited to use the RCA and determine an appropriate decision regarding release or detention. This solution may also involve congressional action through the plenary power to limit executive control over IJs. In the alternative, ICE officials should be heavily monitored in their use of these tools, and strict guidance must be put in place.

B. Changing the Factors

The RCA should be changed substantially to reflect the factors which substantially affect an immigrant’s propensity to return to court. In the RCA’s current state, “immigration authorities use factors copied from pretrial risk determination in the criminal justice system . . . without adoption of important considerations that have developed alongside these factors in the criminal context.”^[245] In her study of immigration detention, Professor Denise Gilman compared the factors used in detention hearings in both the criminal system and the immigration context.^[246] Looking at the factors laid out by the Bail Reform Act of 1984, which dictates many of the facts considered in typical bail hearings, and those found in the information which the Board of Immigration Appeals has used in bond re-determinations, Professor Gilman noted that the systems rely on much of the same information.^[247] In fact, ten of the thirteen factors she identified are used in both systems.^[248] These include length of residence and family ties in the community/the United States, history of criminal or immigration-related conduct, employment history in the community/in the United States, and the individual’s character, among others.^[249] She found that, these factors are used to determine the possibility of release and what support programs or requirements might be imposed after release, but “in the immigration system, the factors are used to set bond amounts, which in turn limits to varying degrees the possibility for eventual release even where release has been deemed viable enough for a bond to be set.”^[250]

This comparison shows how closely the immigration and criminal adjudication systems rely on the same information, despite their different objectives. The objectives of the immigration system, arguably, are to determine a noncitizen’s immigration status, to evaluate new or pending applications for legal status, and to remove the most serious offenders of U.S. immigration law.^[251] The immigration system is not designed to punish offenders;^[252] immigration detention, therefore, should not be used to remove people from society while their status is decided. Detention should only be used to prevent those with the highest risks of flight from absconding. Therefore, it is important that the RCA reflects those factual findings that most affect the immigration system’s objective of “[s]ecuring our nation’s borders and safeguarding the integrity of our immigration system.”^[253]

As of October 2016, the factors used to assess a noncitizen’s threat to public safety have been recalibrated “to better reflect actual threats to public safety,” with most traffic offenses no longer generating risk points and the scores for drug distribution, weapons offenses, and gang affiliation being increased to the “highest” level.^[254] These positive changes mean that the RCA does not detain noncitizens for nonthreatening violations alone; in a due process context, this means that ICE arrestees are not subjected to detention based on actions that have no bearing on the government’s interest in protecting the public. However, for the RCA to become a better measurement of a noncitizen’s flight risk, which in turn would produce more accurate recommendations for release, support programs, requirements, or detention, the way in which flight risk is calculated must be changed. As of October 2016, the RCA does not consider facts such as work authorization, enrollment in a school program, valid government identification, or the current status of valid immigration filings or proceedings.^[255] In terms of due process, the RCA would be better equipped to issue fair recommendations if it relied on factors that were more accurately predictive of risk than on conditions such as the existence of final removal orders or prior abuse of a visa program.^[256] These factors, though indicative of a noncitizen’s legal immigration status, do not properly measure whether the individual is likely to remain in the community or present themselves at a future hearing.^[257] By using factors such as legal representation,^[258] work authorization, or enrollment in a school or training program, the RCA will be better suited to gauge whether a noncitizen is inclined to leave a community for risk of deportation. Finally, ICE does not factor in the likelihood of success on the merits of a noncitizen’s legal status appeal.^[259] It is likely “that migrants will appear at hearings and remain law abiding if this behavior will result in a grant of permission to remain in the United States.”^[260]

Improving the RCA’s ability to accurately predict an individual’s flight risk requires a substantial change in the information used to calculate this disposition. Detaining an individual as they wait for their immigration status to be determined or for their deportation hearing to be held does not satisfy any particular government interest unless they pose a substantial flight risk, and it is clear that the RCA’s current factual findings do not satisfy the due process requirements in place to protect from arbitrary civil detention. Finally, in order to prevent the constant changes made to the RCA, the new factors and their point value, along with a clear explanation of why each factor is used, should be designed by an independent third party. As in the case of the PSA, a third party would be able to design the tool in a way that best serves the purposes of both the government and the noncitizen. The RCA is too malleable as it is, and political actors will only continue to alter the tool if allowed. In the alternative, the RCA’s factors and explanations should be codified into law in order to prevent quick, partisan changes.^[261]

C. Changing the Outcomes

The RCA must also be tied to various outcomes, rather than simply “detain” or “release.” Like in New Jersey and New Mexico, these outcomes can include release on one’s own recognizance, release on regular check-ins with government officials, release on monitoring, release on money bond not to exceed a certain reasonable amount, or where necessary, detention.^[262] According to the Congressional Research Service, ICE employs a program called the Intensive Supervision Appearance Program (“ISAP”) to facilitate ICE’s supervision over those noncitizens who have been released from detention.^[263] Currently, the program is in its third iteration.^[264] This supervision includes “face-to-face and telephonic meetings, unannounced home visits, scheduled office visits, and court and meeting alerts.”^[265] However, these “alternatives to detention” are only used for those individuals who have not been considered a flight risk.^[266] In its description of which individuals are enrolled in ISAP, the Congressional Research Service disclosed that 90% of the 87,384 enrollees did not have a criminal history, and 95–99% of all enrollees in the previous iteration of ISAP appeared at their various court hearings.^[267]

The RCA currently does not provide recommendations for release under supervision.^[268] Though ICE already has functional and successful alternatives to detention in place, the RCA does not recommend any noncitizens for ISAP, which reduces the number of individuals eligible for release.^[269] Like the PSA, the RCA should be changed to incorporate alternatives to detention, providing more noncitizens with the opportunity to be released under supervision even if they pose a higher flight risk. From a due process standpoint, this will reduce the number of people detained in ICE facilities because of a lack of financial resources, a lack of accurate factual findings, and a failure to reasonably meet the proper standard of government justification for detention.

Conclusion

Ultimately, the RCA could be a great tool of reform. However, because of the way it is currently designed and administered, the tool fails to accomplish the goals set out by reformers. The RCA does not ensure that noncitizens arrested by ICE are given a proper opportunity to present facts in their own favor, does not reduce the overall number of people detained by ICE, and uses malleable factors to determine risk scores. The tool is not independently designed or monitored, and there is no clear reasoning behind the factors it uses. The RCA holds great promise, but it can only achieve reform goals if it is overhauled. If major changes are made to the tool, immigrant detainees would be afforded greater due process, would be given a fairer chance in immigration proceedings, and would have a greater probability of being released back into their communities. It is imperative that the immigration detention system be reformed, and improving the Risk Assessment Tool is essential to that reform.

 

* J.D. Candidate 2022, Columbia Law School; B.A. 2019, Rutgers University. ↑

1.         Where Are Risk Assessments Being Used?, Mapping Pretrial Just., https://pretrialrisk.com/national-landscape/where-are-prai-being-used/ [https://per
   ma.cc/G9JQ-3Q88]. ↑

2.   U.S. Dep’t of Homeland Sec. Off. of Inspector Gen., OIG-15-22, U.S. Immigration and Customs Enforcement’s Alternatives to Detention (Revised) 4–5 (Feb. 4, 2015) ↑

3.   Kate Evans & Robert Koulish, Manipulating Risk: Immigration Detention Through Automation, 24 Lewis & Clark L. Rev. 789, 830–32 (2020). ↑

4.       Id.; see John Koepke, Danger Ahead: Risk Assessment and the Future of Bail Reform, 93 Wash. L. Rev. 1725 (2018) (noting that predictive models might systematically overestimate risk and may be susceptible to ill-defined concepts of dangerousness); see also Mark Noferi, The Immigration Detention Risk Assessment, 29 Geo. Immigr. L.J. 45 (2014) (analyzing the RCA’s methodology and outcomes). ↑

5.   Keeping America Safe, Immigr. & Customs Enf’t (2021), https://www.ice.gov/ [https://perma.cc/US7R-ZDKQ]. ↑

6.         Enforcement and Removal Operations, Immigr. & Customs Enf’t (2021), https://www.ice.gov/about-ice/ero [https://perma.cc/SB5W-E72H]. ↑

7.   Demore v. Kim, 538 U.S. 510, 522 (2003); see Carrie Rosenbaum, Immigration Law’s Due Process Deficit and the Persistence of Plenary Power, 28 Berkeley La Raza L.J. 118, 126 (2018) (explaining that immigration policy and the plenary power doctrine have enabled Congress to deny constitutional protections to non-citizens). ↑

8.   Demore, 538 U.S. at 522. ↑

9.   8 U.S.C. §1226(a). ↑

10.      Id. ↑

11.         FY 2020 Detention Statistics, Immigr. & Customs Enf’t (2021), https://www.ice.gov/detain/detention-management [https://perma.cc/VLF5-8RVG]. Pursuant to § 1226(c), in 2020, at least 98,375 noncitizens were subject to mandatory detention because of their status as convicted criminals. ↑

12.   8 U.S.C. §1226(a); 8 U.S.C. §1225(b) (mandatory detention for immigrants who have not been admitted or paroled to the United States); see 8 U.S.C. §1226(c) (mandatory detention for immigrants who have committed certain criminal offenses or is deportable for reasons articulated in 8 U.S.C. §1227(a)(2)(A)(i)); 8 U.S.C. §1231(a)(2) (mandatory detention for immigrants who are not admissible 8 U.S.C. §1182(a)(2) or 1182(a)(3)(B) or deportable under 8 U.S.C. §1127(a)(2) or 1227(a)(4)(B)). ↑

13.   Plyler v. Doe, 457 U.S. 202 (1982). ↑

14.   “Due process” in any context under the Fifth or Fourteenth Amendments is amorphous at best, as exemplified in the three-factor balancing test case in Mathews v. Eldridge and applied with uncertainty in Turner v. Rogers. Mathews, 424 U.S. 319, 335 (1976); Turner, 564 U.S. 431 (2011). The three-factor test includes balancing 1) the private interest at stake, 2) the erroneous deprivation of this interest through the procedures used, and 3) the government’s interest. Mathews, 424 U.S. at 321. In Turner, the Supreme Court cited the Mathews test to find that Turner’s due process rights were violated when he was neither provided counsel nor the benefit of alternative procedures in a civil contempt proceeding that resulted in his incarceration. Still, the Court noted that “the Due Process Clause does not automatically require the provision of counsel at civil contempt proceedings to an indigent individual who is subject to a child support order, even if that individual faces incarceration (for up to a year).” Turner, 564 U.S. at 447–49. ↑

15.   8 U.S.C. §1231(a)(6). Though this statutory provision is different from 8 U.S.C. §1226(a), it also addresses non-mandatory detention and is subject to due process limitations. ↑

16.   Zadvydas v. Davis, 533 U.S. 678 (2001). ↑

17.   Id. at 699–700. ↑

18.   Id. at 701. ↑

19.   Id. ↑

20.   Demore v. Kim, 538 U.S. 510 (2003). ↑

21.   Id. at 514. ↑

22.   Id. at 523. ↑

23.   Id. at 522. ↑

24.   Id. at 528. ↑

25.   Id. at 532–33. ↑

26.   Id. at 532. ↑

27.   Id. ↑

28.   Jennings v. Rodriguez, 138 S. Ct. 830 (2018). ↑

29.   Id. ↑

30.   Id. ↑

31.   Id. at 847. ↑

32.   Id. at 847–48. ↑

33.   Id. at 851. ↑

34.   Id. ↑

35.   Gretchen Frazee, What Constitutional Rights Do Undocumented Immigrants Have?, PBS (Jun. 25, 2018), https://www.pbs.org/newshour/politics/what-constitution
    al-rights-do-undocumented-immigrants-have [https://perma.cc/6LC3-EK74]. ↑

36.   Demore v. Kim, 538 U.S. 510, 522 (2003). ↑

37.   8 C.F.R. § 215.1. ↑

38.   8 C.F.R. § 1236.1(c)(6); 8 C.F.R. § 1236.1(c)(8). ICE officials are overseen by a hierarchy of leaders, all of whom are led by the Secretary of Homeland Security, who is a political appointee. At the time of this writing, the Secretary of Homeland Security is Alejandro Mayorkas. Mayorkas was appointed by President Biden and confirmed by the Senate in February 2021. See Press Release, Dept. of Homeland Sec., Alejandro Mayorkas Sworn in as Secretary of Homeland Security (Feb. 2, 2021), https://www.dhs.
    gov/news/2021/02/02/alejandro-mayorkas-sworn-secretary-homeland-security [https://perma.cc/LE23-HBQY]. ↑

39.   8 C.F.R. § 1236.1(c)(8). ↑

40.   Id. ↑

41.   Id. ↑

42.      In the immigration context, the administration of due process refers to the procedure through which ICE decides to detain or release an arrested noncitizen. Statutorily, the district director who decides whether to release an individual may consider any circumstances that he or she deems relevant. In this process, due process is not guaranteed because district directors can consider as many as or few factors in their determination as they deem relevant, thereby failing to ensure fair and just determinations. See 8 C.F.R. § 1236.1(c)(6)(ii). ↑

43.   8 C.F.R. § 1236.1. ↑

44.         Robert Koulish & Katherine Evans, Injustice and the Disappearance of Discretionary Detention Under Trump: Detaining Low Risk Immigrants Without Bond 4 (iLCSS, Working Paper No. 5, 2020). ↑

45.   Robert Koulish & Ernesto Calvo, The Human Factor: Algorithms, Dissenters, and Detention in Immigration Enforcement 11 (iLCSS, Working Paper No. 1, 2019). Though the argument may be that the violation of one administrative law (here, illegally crossing the border) would likely lead to the willful violation of another (an administered court hearing date), the emphasis on recency of entry and removal orders was based on enforcement priorities rather than actual flight risk. The focus on such violations were a result of prosecutorial priorities rather than actual measures of risk. Id. ↑

46.   Koulish & Evans, supra note 44, at 5; see Exec. Order No. 13,767, 82 Fed. Reg. 8,793 (Jan. 25, 2017) (ordering the end of “catch and release,” thereby resulting in detention for all undocumented immigrants who are apprehended by ICE); Exec. Order No. 13,768, 82 Fed. Reg. 8,799 (Jan. 25, 2017) (ordering the strict enforcement of immigration detention laws). ↑

47.   Koulish & Evans, supra note 44, at 4. ↑

48.   Unlike Article III judges, immigration judges are appointed by the Attorney General (“AG”) and governed by statute. Though immigration judges “shall exercise their independent judgment and discretion,” they are ultimately acting as the AG’s delegates in presiding over immigration cases. 8 C.F.R. § 1003.10(b). Article III judges, on the other hand, are constitutionally vested with strong independence and are not directly subject to any other branch of government. Vicki Jackson, Packages of Judicial Independence: The Selection and Tenure of Article III Judges, 95 Geo. L.J. 965, 1022 (2007); see also 8 C.F.R. § 1003.10 (dictating the appointment, powers and duties, and governing standards for immigration judges). ↑

49.   Crowell v. Benson, 285 U.S. 22, 50 (1932). ↑

50.   Background on Judicial Review of Immigration Decisions, Am. Immigr. Council (June 1, 2013), https://www.americanimmigrationcouncil.org/research/background-judicial-review-immigration-decisions [https://perma.cc/2J3V-5RFV]. ↑

51.        Office of the Chief Immigration Judge, U.S. Dept. of Just. (Dec. 7, 2020), https://www.justice.gov/eoir/office-of-the-chief-immigration-judge [https://perma.cc/
    DM3J-H5BC]. ↑

52.   8 C.F.R. § 1236.1(d) (2022). ↑

53.   8 C.F.R. § 1003.10(b) (2022). ↑

54.   Aleksandar Cuic, The Trump Administration and Immigration Judges: Decreased Judicial Independence or Increased Efficiency?, 51 Case W. Res. J. Int’l L. 155, 161 (2019). ↑

55.   8 C.F.R. § 1003.1(a)(1) (2022). ↑

56.   Cuic, supra note 54, at 161. ↑

57.   Id. For context, administrative closures allow judges to “temporarily close cases and take them off their active docket” because of limited resources or jurisdictional/procedural issues that prolong a case. The Life and Death of Administrative Closure, TRAC Immigr. (Sept. 10, 2020), https://trac.syr.edu/immigration/reports/
    623/ [https://perma.cc/S9C4-T7PK]. ↑

58.   Cuic, supra note 54, at 161. ↑

59.   Matter of Castro-Tum, 27 I. & N. Dec. 271, 271 (A.G. 2018). ↑

60.   Antonio Olivo, Immigration Judges, Worried Trump Is Seeking to Cut Them Out, Fight Back, Wash. Post (Aug. 9, 2018), https://www.washingtonpost.com/local/social-issues/immigrationjudges-worried-trump-is-seeking-to-cut-themoutfightback/2018/
    08/09 (on file with the Columbia Human Rights Law Review); see Cuic, supra note 54, at 161–62 (discussing the question of IJ independence). ↑

61.   8 C.F.R. § 1003.1(a)(1) (2022). ↑

62.   Reno v. Flores, 507 U.S. 292, 306 (1993) (“It is well established that the Fifth Amendment entitles aliens to due process of law in deportation proceedings.”). ↑

63.   U.S. Const. amend. V.; see also United States v. Salerno, 481 U.S. 739 (1987) (explaining that the Due Process Clause of the Fifth Amendment provides “substantive due process” protections). ↑

64.   U.S. Const. amend. VI. ↑

65.   Id. Regarding bail, precedent set in Stack v. Boyle establishes that “[s]ince the function of bail is limited [to serving as an assurance of the defendant’s presence in court], the fixing of bail for any individual defendant must be based upon standards relevant to the purpose of assuring the presence of that defendant.” 342 U.S. 1, 4 (1951). This follows from the Eighth Amendment, which prohibits excessive bail. U.S. Const. amend. VIII. ↑

66.   ODonnell v. Harris Cnty., 892 F.3d 147, 157(5th Cir. 2018). ↑

67.   Id. at 159. ↑

68.   Rick Jones, From the President: The Siren Song of Objectivity: Risk Assessment Tools and Racial Disparity, NACDL (Apr. 2018), https://www.nacdl.org/article/april2018
    -fromthepresidentthesirensongof [https://perma.cc/2V24-UVWZ]; Brandon Garrett & John Monahan, Assessing Risk: The Use of Risk Assessment in Sentencing, Judicature (2019), https://judicature.duke.edu/articles/assessing-risk-the-use-of-risk-assessment-in-sentencing/ [https://perma.cc/V7QC-7GGF] (discussing increased judicial use of RATs). ↑

69.   See The First Step Act’s Risk Assessment Tool, Urb. Inst. (Apr. 30, 2021), https://apps.urban.org/features/risk-assessment/ [https://perma.cc/AG8Q-2W73] (describing a RAT used at the federal level); A.I. in the Criminal Justice System, EPIC, https://epic.org/algorithmic-transparency/crim-justice/ [https://perma.cc/9FHE-3X9S] (explaining that RAT algorithms have come into greater use at both the federal and state levels); Rhys Dipshan et al., The Most Widely Used Risk Assessment Tool in Each U.S. State, LAW.COM (July 13, 2020), https://www.law.com/legaltechnews/2020/07/13/the-most-widely-used-risk-assessment-tool-in-each-u-s-state/ [https://perma.cc/K46C-4HF3] (highlighting tools used by local criminal justice systems). ↑

70.   Dipshan et al., supra note 69. ↑

71.   Where Are Risk Assessments Being Used?, supra note 1. Many states and counties have adopted the VPRAI-R and COMPAS, two risk assessment tools that were independently developed (as in, they were created by third-party organizations, rather than a state’s government itself). The VPRAI-R was developed by Lumonisity, Inc., led by Marie VanNostrand. COMPAS was developed by Northpointe, Inc., now known as Equivant. See Common Pretrial Risk Assessments, Mapping Pretrial Just. https://pre
    trialrisk.com/the-basics/common-prai/ [https://perma.cc/T3Z5-G562] (describing the origin of several widely used RATs); see also Risk Assessment Factsheet, Stan. L. Sch. Pol’y Lab (June 19, 2019), https://www-cdn.law.stanford.edu/wpcontent/uploads/2019/06/
    VPRAI-Factsheet-FINAL-6-20.pdf [https://perma.cc/A9LZ-XRH5] (discussing the background of the VPRAI-R). Other states have had tools specially designed for their jurisdiction. See How Many Jurisdictions Use Each Tool?, Mapping Pretrial Just., https://pretrialrisk.com/national-landscape/how-many-jurisdictions-use-each-tool/ [https://perma.cc/FE2M-VQKG] (describing 11 states with state-developed tools). ↑

72.   How Many Jurisdictions Use Each Tool?, supra note 71. At least 5 states and 59 counties use the PSA, which impacts about 56.3 million people. Comparatively, no other RAT is used in as many jurisdictions, making the PSA one of the most widely used RATs. Id. ↑

73.   Public Safety Assessment FAQs (“PSA 101”), Arnold Ventures (Mar. 18, 2019) [hereinafter Public Safety Assessment], https://craftmediabucket.s3.amazonaws.com/
    uploads/Public-Safety-Assessment-101_190319_140124.pdf [https://perma.cc/6Z3A-TNLY]. ↑

74.   Beth Schwartzapfel, Can Racist Algorithms Be Fixed?, Marshall Project (July 1, 2019), https://www.themarshallproject.org/2019/07/01/can-racist-algorithms-be-fixed [https://perma.cc/J867-RBQC]. ↑

75.   About the Public Safety Assessment: How the PSA Works, Advancing Pretrial Pol’y & Rsch (2022). https://advancingpretrial.org/psa/factors/ [https://perma.cc/
    9R8T-ASE7]. The PSA assigns a score for the likelihood of a New Criminal Arrest and Failure to Appear, which are then matched on a matrix to recommend release. See Release Conditions Matrix, Advancing Pretrial Pol’y & Rsch. (2021), https://advancing
    pretrial.org/psa/factors/release-conditions-matrix/ [https://perma.cc/7F6L-HHMD]. ↑

76.   About the Public Safety Assessment, supra note 75. ↑

77.   Public Safety Assessment FAQs (“PSA 101”), supra note 73. ↑

78.   ODonnell v. Harris Cnty., 892 F.3d 147 (5th Cir. 2018); Keri Blakinger, The Beto Effect: Transforming Houston’s Criminal Justice System, Marshall Project (Feb. 25, 2020), https://www.themarshallproject.org/2020/02/25/the-beto-effect-transforming-houston-s-criminal-justice-system [https://perma.cc/J52A-ZSNA]. ↑

79.   Pretrial Services investigates and compiles information on people who are charged with a crime by Harris County. “This information assists the judiciary with release and detention decisions . . . . The department structures all activities to address the purposes of bail: to assure appearance in court and minimize danger to the community.” See Frequently Asked Questions, Harris Cnty. – Pretrial Services, https://pretrial.harriscountytx.gov/Pages/FAQs.aspx [https://perma.cc/CP2X-JKU9] (“The report assists the judiciary with release and detention decisions . . . . The department structures all activities to address the purposes of bail: to assure appearance in court and minimize danger to the community.”). ↑

80.   Plaintiffs’ Exhibit 8, Preliminary Injunction Hearing, ODonnell v. Harris Cnty, Civil No. 16-1414 (S.D. Tex. Mar. 2017). ↑

81.   Id. ↑

82.   This recommendation is also based on a bail schedule, which was used until the implementation of the PSA in 2017. See Misdemeanor Bail Schedule for the Harris County Criminal Courts at Law, Harris Cnty. Crim. Cts. at Law (Sept. 6, 2012), http://www.ccl.hctx.net/criminal/Misdemeanor%20Bail%20Schedule.pdf [https://perma.cc/F3KV-LBVC]. ↑

83.   How Does the Process of Bail Work in Harris County?, Aslett L. Firm PLLC (2021), https://aslettlaw.com/criminal-defense/process-bail-work-harris-county [https://perma.cc/7PMT-Q66F]. ↑

84.   Lise Olsen, Videotapes Reveal Flaws in Harris County Bail Bond Hearings, Hous. Chron. (Nov. 29, 2016), https://www.houstonchronicle.com/news/houston-texas/houston/article/Videotapes-reveal-flaws-in-Harris-County-bail-10642138.php [https://perma.cc/YR4H-AR8E]. ↑

85.   Robert Arnold, Harris County Implements Sweeping Change to Bail System, Click2houston.com (July 31, 2017), https://www.click2houston.com/news/
    2017/07/31/harris-county-implements-sweeping-change-to-bail-system/ [https://perma.cc/RCN7-89Q8]. ↑

86.   Michael Hardy, In Fight Over Bail’s Fairness, a Sheriff Joins the Critics, N.Y. Times (Mar. 9, 2017), https://www.nytimes.com/2017/03/09/us/houston-bail-reform-sheriff-gonzalez.html (on file with the Columbia Human Rights Law Review). ↑

87.   Andrew Kragie, Changes Coming to Harris County Bail System; Head of Pretrial Services Gives Details, Hous. Chron. (July 5, 2017), https://www.chron.com/
    news/houston-texas/houston/article/harris-county-bail-system-reform-pretrial-11267694.php [https://perma.cc/9ZNW-KWGT]. ↑

88.   “We’re Very Thankful”: COVID-19 Reinforces Houston Pretrial Reform Efforts, Arnold Ventures (July 31, 2020), https://www.arnoldventures.org/stories/were-very-thankful-covid-19-reinforces-houston-pretrial-reform-efforts/ [https://perma.cc/RBU2-86L5]. ↑

89.   Id. ↑

90.      It is important to note that while cash bond is still technically an option for judges to use, the implementation of the PSA was intended to reduce the number of people detained for their inability to pay bond. Public Safety Assessment FAQs (“PSA 101”), supra note 73. In New Jersey, money bail is not an option on the bail schedule provided for judges to use. See N.J. Cts., Pretrial Release Recommendation Decision Making Framework 4 (Mar. 2018), https://njcourts.gov/courts/assets/criminal/decmakframwork.pdf?c=qrn [https://perma.cc/T2D9-CTDP] (highlighting that money bail is not an option for judges to use in New Jersey). ↑

91.   How Does the Process of Bail Work in Harris County?, supra note 83; see Bunin et al., NADCL Found. Crim. Just., The Harris County, Texas Bail Manual 8–11 (Sept. 2018), https://www.nacdl.org/getattachment/e11b870d-0647-4468-a9aa-408204004
    509/the-harris-county-bail-manual.pdf [https://perma.cc/3UV7-EZRV] (showing Harris County’s bond schedule). ↑

92.   Bunin, supra note 91, at 3–10. Though there is no literature on judicial overrides of the PSA in Harris County, judges can choose to override the PSA’s determination and impose a pretrial outcome (release on recognizance, release on bond, detention) of their own volition. See Alex Albright, If You Give a Judge A Risk Score: Evidence from Kentucky Bail Decisions, Olin Ctr. L., Econ., & Bus. Fellows’ Discussion Paper Series 85, at 85 (Sept. 3, 2019) (describing how judges have the ability to override recommendations). ↑

93.   See Bunin et al., supra note 91, at 4 (explaining that the PSA examines risk based on failure to appear, new criminal activity, and new violent criminal activity). “Defendants who do not threaten public safety and are likely to appear for scheduled court dates should not remain in jail simply because they cannot afford bail.” Id. ↑

94.   Schwartzapfel, supra note 74. ↑

95.   How Does the Process of Bail Work in Harris County?, supra note 83. ↑

96.   Where Are Risk Assessments Being Used?, supra note 1. ↑

97.   Public Safety Assessment FAQs (“PSA 101”), supra note 73, at 2. ↑

98.   Id. ↑

99.   Id. at 3–4. ↑

100.   Oftentimes, prosecutors’ offices and public records can provide a defendant’s prior convictions and age. See Chloe Anderson et al., Evaluation of Pretrial Justice System Reforms That Use the Public Safety Assessment, MDRC Ctr. Crim. Just. Rsch. at 7, n.9 (Nov. 2019) (“At this stage, the PSA uses information from the defendant’s in-state criminal history—which is available from state databases once fingerprints are taken—to calculate a risk score.”). ↑

101.   What Happens in the First 24 Hours After My Arrest in Texas?, Aslett L. Firm PLLC (2021), https://aslettlaw.com/criminal-defense/what-happens-in-the-first-24-hours-after-my-arrest-in-texas/ [https://perma.cc/HY4J-P635]. ↑

102.   Id. ↑

103.   Bunin et al., supra note 91, at 8–11. ↑

104.   See Note, Judicial Discretion in Granting Bail, 27 St. John’s L. Rev. 56, 56-57 (1952) (describing the judge’s discretion and the “judicial hunch” in granting bail in criminal prosecutions); Milton Heumann et al., Going with Your Gut: “Hunches” and “Hunching” in Judicial Decision-Making, 55 Crim. L. Bull. 1, 10 (2019) (examining judicial decision-making and “hunches,” including in setting bail). ↑

105.    Public Safety Assessment FAQs (“PSA 101”), supra note 73. ↑

106.   Megan Stevenson, Assessing Risk Assessment in Action, 103 Minn. L. Rev. 303, 327-333(2018). ↑

107.   Brandon Buskey & Andrea Woods, Making Sense of Pretrial Risk Assessments, The Champion (June 2018), https://www.nacdl.org/Article/June2018-Making
     SenseofPretrialRiskAsses [https://perma.cc/G9SW-623D]. ↑

108.   Sandra G. Mayson, Bias In, Bias Out, 128 Yale L.J. 2218, 2224 (2019). ↑

109.   Id. at 2252. ↑

110.   Sandra G. Mayson, Dangerous Defendants, 127 Yale L.J. 490, 535 (2018). Mayson argues at various points throughout this article that the data used to determine risk levels is inherently biased because defendants who do not receive adequate support services (thereby possibly failing to appear in court or being re-arrested) are more likely to be deemed “high-risk.” This outcome raises the question of what data RATs should be using, considering a failure to appear may be less associated with actual risk and more related to a failure in the state to support a defendant. Id. at 541–45. ↑

111.   About the Public Safety Assessment, Advancing Pretrial Pol’y & Rsch., https://advancingpretrial.org/psa/about/ [https://advancingpretrial.org/psa/about/]. ↑

112.   This is not to say that risk assessments are perfect. Risk assessments simply provide a more streamlined, less biased approach to pretrial hearings than prior programs, like in Harris County, TX. For discussions of the limitations of risk assessment tools, see John L. Koepke, Danger Ahead: Risk Assessment and the Future of Bail Reform, 93 Wash. L. Rev. 1725, 1725 (2018) (“[M]any are questioning the extent to which pretrial risk assessment instruments actually serve reform goals.”). See also Matt Henry, Risk Assessment: Explained, The Appeal (Dec. 14, 2019) (presenting the challenges and risks associated with pretrial risk assessments). ↑

113.   How the PSA Works, supra note 75. ↑

114.   About the Public Safety Assessment, supra note 111. ↑

115.   How the PSA Works, supra note 75. ↑

116.   About the Public Safety Assessment, supra note 111. In developing the PSA, Arnold Ventures studied 1.25 million pretrial records across various jurisdictions. These records were used to create the risk scores, which in turn predict an individual’s propensity to fail to appear or to commit a crime post-release. ↑

117.   Developing a National Model for Pretrial Risk Assessment, Laura & John Arnold Found. at 3–4 (Nov. 2013), https://craftmediabucket.s3.amazonaws.com/
     uploads/PDFs/LJAF-research-summary_PSA-Court_4_1.pdf [https://perma.cc/FH35-FBZ3]. ↑

118.   Id. ↑

119.   Id. at 5. ↑

120.   See discussion infra Part II.C.2. ↑

121.         Mary Ellen Callahan, U.S. Dept. of Homeland Sec., Privacy Impact Assessment Update for the Enforcement Integrated Database (EID) at 4 (Apr. 6, 2012), https://www.dhs.gov/xlibrary/assets/privacy/privacy_piaupdate_EID_april2012.pdf [https://perma.cc/4XLC-X6YS]. ↑

122.   Id. ↑

123.   Id. ↑

124.   Id. ↑

125.   Evans & Koulish, supra note 3, at 804–05. ↑

126.   Id. at 820–21. ↑

127.   Id. at 802. ↑

128.   See supra Part I. ↑

129.   Evans & Koulish, supra note 3, at 802–04; see also Dept. of Just., Immigration Court Practice Manual 130 (Dec. 2016), https://www.justice.gov/eoir/page/file/
     1284746/download [https://perma.cc/6UVT-TZSV] (describing the kinds of proceedings that IJs oversee). ↑

130.   Id.; see Frequently Asked Questions, supra note 79 (explaining that Harris County Pretrial Services compiles information relevant for judges to make bail determinations). ↑

131.   Evans & Koulish, supra note 3, at 804. ↑

132.   Id. at 806–07. ↑

133.   Id. at 811–12. ↑

134.   Id. at 812. The PSA relies on the definition of violence as per the statutes of the various states in which it is used. For a list of statutory offenses that are considered violent in New Jersey, see N.J. Cts., Public Safety Assessment New Jersey Risk Factor Definitions – January 2019, at 5–7 (Jan. 2019), https://njcourts.gov/courts/assets/
     criminal/psariskfactor.pdf [https://perma.cc/QJD5-MKVG]. ↑

135.   Evans & Koulish, supra note 3, at 804–32. ↑

136.   Id. at 805. ↑

137.   Id. at 824. ↑

138.   Id. at 821. ↑

139.   Id. at 822. ↑

140.   Id. at 822–3. ↑

141.   Id. ↑

142.   Id. ↑

143.   Id. at 826. ↑

144.   Id. at 828. ↑

145.   Id. at 830. ↑

146.   Id. at 828–29. ↑

147.   Id. at 830. ↑

148.   Mica Rosenberg, Trump’s Catch-and-Detain Policy Snares Many Who Have Long Called U.S. Home, Reuters (June 20, 2018), https://www.reuters.com/
     investigates/special-report/usa-immigration-court/ [https://perma.cc/JSM4-DNGV]. ↑

149.        Daniel Oberhaus, ICE Modified Its “Risk Assessment” Software So It Automatically Recommends Detention, Vice (June 26, 2018), https://www.vice.com/
     en/article/evk3kw/ice-modified-its-risk-assessment-software-so-it-automatically-recommends-detention [https://perma.cc/J5FF-8TJY]. ↑

150.   About, Arnold Ventures (2021), https://www.arnoldventures.org/about [https://perma.cc/R2MC-D7V3]. ↑

151.   About the Public Safety Assessment, supra note 111. ↑

152.   Id. ↑

153.         Dr. Dora Schriro, U.S. Dept of Homeland Sec., Immigration Detention Overview and Recommendations 3, 17 (Oct. 6, 2009). This report was produced for the Secretary of Homeland Security under the Obama Administration and “provides a comprehensive review and evaluation of the U.S. Immigration and Custom Enforcement (ICE) system of Immigration Detention.” Nina Bernstein, Report Critical of Scope of Immigration Detention, N.Y. Times (Oct. 6, 2009), https://www.nytimes.com/interactive/
     projects/documents/immigration-detention-overview-and-recommendations#p= (on file with the Columbia Human Rights Law Review) (providing an overview of the most concerning issues exposed by the Schriro Report). ↑

154.   Alternatives to Detention, supra note 2, at 4. ↑

155.   Id. ↑

156.   Evans & Koulish, supra note 3; Noferi, supra note 4 (conducting an in-depth study of the RCA). ↑

157.   Evans & Koulish, supra note 3, at 796–800. ↑

158.   Id. ↑

159.   Id. ↑

160.   Jayashri Srikantiah, Reconsidering Money Bail in Immigration Detention, 52 U.C. Davis L. Rev. 521, 539-540 (2018). ↑

161.   ODonnell v. Harris Cnty., 892 F.3d 147, 159 (5th Cir. 2018). ↑

162.   Srikantiah, supra note 160, at 539, 544. ↑

163.   Noferi, supra note 4, at 4 (“It would improve the uniformity of detention decisions, effectively identify dangerous individuals, and increase ICE’s successful use of alternatives to detention (ATD). With RCA, ICE could limit detention for those identified with a ‘propensity for violence,’ which the Report said was distinct from that of the criminally incarcerated.”). ↑

164.   Jennings v. Rodriguez, 138 S. Ct. 830, 847 (2018). ↑

165.   Demore v. Kim, 538 U.S. 510, 515 (2003). ↑

166.   Hernandez v. Sessions, 872 F.3d 976, 990–91 (9th Cir. 2017) (“Detention of an indigent ‘for inability to post money bail’ is impermissible if the individual’s ‘appearance at trial could reasonably be assured by one of the alternate forms of release.’” (quoting Pugh v. Rainwater, 572 F.2d 1053, 1058 (5th Cir. 1978) (en banc))). ↑

167.   Padilla v. Immigr. & Customs Enf’t, 953 F.3d 1134, 1142–43 (9th Cir. 2020); see also Stack v. Boyle, 342 U.S. 1, 1 (1951) (holding that bail hearings must be individualized and take the defendant’s financial limitations into account). ↑

168.   Hernandez, 872 F.3d at 967. ↑

169.   Id. at 982. ↑

170.   Id. at 990 (citing Zadvydas v. Davis, 533 U.S. 678, 690, 693 (2001)). ↑

171.   Id. (citing Singh v. Holder, 638 F.3d 1196, 1203 (9th Cir. 2011)). ↑

172.   Id. (citing Pugh v. Rainwater, 572 F.2d 1053, 1058 (5th Cir. 1978)). ↑

173.        Id. at 990–91 “Setting a bond amount without considering financial circumstances or alternative conditions of release undermines the connection between the bond and the legitimate purpose of ensuring the noncitizen’s presence at future hearings. There is simply no way for the government to know whether a lower bond or an alternative condition would adequately serve those purposes when it fails to consider those matters.” Id. at 991. ↑

174.   Padilla v. Immigr. & Customs Enf’t, 953 F.3d 1134, 1143 (9th Cir. 2020). ↑

175.   Id. at 1139 (“Section 1225(b) provides for ‘expedited removal’ of ‘arriving’ noncitizens at ports-of-entry and inadmissible noncitizens apprehended within the United States who cannot prove that they have been in the United States for more than two years.”). ↑

176.   Id. at 1142–48. This discussion applies to those detained under § 1226(a) because it affirms “the importance of providing detained noncitizens individualized hearings before neutral decisionmakers.” Id. at 1143. Those detained under § 1226(a) are kept in non-punitive detention and are constitutionally afforded due process as discussed in Part I of this Note. Therefore, the holding of Padilla can be extended to all noncitizens detained by ICE pending the adjudication of their case or their removal, including § 1226(a) detainees. ↑

177.   Id. at 1142 (citing Zadvydas v. Davis, 533 U.S. 678, 690 (2001); Hernandez, 872 F.3d at 990). ↑

178.   Id. (citing United States v. Salerno, 481 U.S. 739, 750–51 (1987)). ↑

179.   Id. at 1143. ↑

180.   Id. at 1145 (citing St. John v. McElroy, 917 F. Supp 243, 251 (S.D.N.Y. 1996)). ↑

181.   Alejandro Mayorkas Sworn in as Secretary of Homeland Security, supra note 38. ↑

182.   Padilla, 953 F.3d at 1145. ↑

183.   Id. ↑

184.        Id. at 1142 (“The Supreme Court has held repeatedly that non-punitive detention violates the Constitution unless it is strictly limited, which typically means that the detention must be accompanied by a prompt individualized hearing before a neutral decisionmaker to ensure that the imprisonment serves the government’s legitimate goals.” (citing Salerno, 481 U.S. at 750–51)). ↑

185.   Alternatives to Detention, supra note 2, at 11. ↑

186.   Id. at 14. ↑

187.   Oberhaus, supra note 149. ↑

188.   State v. Loomis, 881 N.W.2d 749, 753 (Wis. 2016). ↑

189.   Id. at 752. ↑

190.   Id. at 757. ↑

191.   Id. at 761. The Court explained that because the COMPAS algorithm was based on Loomis’ own answers to twenty-one questions, as well as on publicly available information, Loomis “had the opportunity to verify that the questions and answers listed on the COMPAS report were accurate.” Id. ↑

192.   Id. at 771. The Court opined that the COMPAS algorithm was not the determinative factor in the Loomis’ sentencing, and therefore did not pose a threat to the validity of his sentence as an individualized determination. Id. ↑

193.   Id. at 767. The Court held that because the COMPAS’s use of the defendant’s gender was for accuracy purposes and because there was a factual basis to the lower court’s consideration of Loomis’ gender, the algorithm’s use of the information did not violate the defendant’s due process rights. Id. ↑

194.   895 F.3d 272, 297 (3d Cir. 2018). ↑

195.   Id. ↑

196.   Id. at 301. In evaluating whether the PSA, as used by New Jersey, is a violation of due process, the Third Circuit applied the Mathews test. ↑

197.   Ramirez v. State, No. 77777-COA, 2020 Nev. App. Unpub. LEXIS 319, at *1 (Nev. Ct. App. April 27, 2020). ↑

198.   Id. at *3. ↑

199.   State v. Guise, 921 N.W.2d 26 (Iowa 2018); State v. Gordon, 921 N.W.2d 19 (Iowa 2018) (appellate courts refuse to rule on the due process question raised regarding risk assessment tools because the question was not preserved). ↑

200.   The National Task Force on Fines, Fees, and Bail Practices developed a guide detailing bail reform efforts in six states. This guide details the goals that each state sought to meet by adopting risk assessment tools like the PSA. One of these goals was to make sure that defendants are not held pre-trial because they are unable to pay bond. Nat’l Task Force on Fines, Fees, and Bail Practices, Bail Reform: A Practical Guide Based on Research and Experience 5(2019). ↑

201.   Glenn A. Grant, N.J. Cts., Jan. 1 – Dec. 31 2018 Report to the Governor and the Legislature 3 (2018), https://njcourts.gov/courts/assets/criminal/2018cjrannual.
     pdf?c=taP [https://perma.cc/V5VW-8UPK]. “New Jersey has moved away from a system that relied heavily on monetary bail. Two years into its existence, CJR has begun to remove many of the inequities created by the prior approach to pretrial release.” Id. at 16. ↑

202.   Id. at 19–25. ↑

203.   Elise M. Ferguson, UNM Inst. Soc. Rsch., Bernalillo County Metropolitan Detention Center: Analysis of the Jail Population, June 30, 2020, at 8-10 (Oct. 2020), http://isr.unm.edu/reports/2020/bernalillo-county-metropolitan-detention-center-analysis-of-the-jail-population,-june-30,-2020.pdf [https://perma.cc/7ZED-NRJL]. ↑

204.   Grant, supra note 201, at 19. ↑

205.   Id. ↑

206.   Id. at 23. ↑

207.   Id. at 21. ↑

208.   Diana Dabruzzo, New Jersey Set Out to Reform Its Cash Bail System, Arnold Ventures (Nov. 14, 2019), https://www.arnoldventures.org/stories/new-jersey-set-out-to-reform-its-cash-bail-system-now-the-results-are-in/#:~:text=The%20MDRC%20
     analysis%20found%20that,people%20in%20jail%20awaiting%20trial [https://per
     ma.cc/3STJ-3AZ4]. ↑

209.  Jenna Dole, UNM Inst. Soc. Rsch., Bail Reform 7 (Oct. 2019), http://isr.unm.edu/centers/new-mexico-statistical-analysis-center/docs/bail-reform-year-1-final-report-121719.pdf [https://perma.cc/D6TS-54HN]; Ferguson, supra note 203, at 9 (showing that jail populations had decreased in part due to the implementation of the PSA). ↑

210.   Ferguson, supra note 203, at 9. ↑

211.   Id. ↑

212.   U.S. Immigr. & Customs Enf’t, U.S. Immigration and Customs Enforcement Fiscal Year 2019 Enforcement and Removal Operations Report 18 (2019), [hereinafter Fiscal Year 2019 Enforcement and Removal Operations Report] https://www.ice.gov/sites/default/files/documents/Document/2019/eroReportFY2019.pdf [https://perma.cc/9M22-FWPR]. ↑

213.   Id. ↑

214.   Id. ↑

215.        Immigration Detention 101, Detention Watch Network, https://www.de
     tentionwatchnetwork.org/issues/detention-101#:~:text=The%20average%20daily%20
     population%20of,as%20500%2C000%20immigrants%20each%20year [https://perma
     .cc/KY83-TBWB]. ↑

216.   Id. ↑

217.   Fiscal Year 2019 Enforcement and Removal Operations Report, supra note 212, at 5. ↑

218.         Why Jurisdictions Use RATs, Mapping Pretrial Injustice, https://pretrial
     risk.com/the-basics/the-case-for-rats/ [https://perma.cc/5PVH-BXS6] (last visited Oct. 13, 2020). ↑

219.   Where Are Risk Assessments Being Used?, supra note 1. ↑

220.   Schriro, supra note 153, at 18. ↑

221.   This Note argues that despite ICE’s assertion that the 85% of arrestees who were detained posed a flight risk or threat to public safety, the design of the RCA itself is highly flawed and does not properly justify detention because it does not present any alternative. ↑

222.   Chelsea Barabas et al., The Problems with Risk Assessment Tools, N.Y. Times (July 17, 2019), https://www.nytimes.com/2019/07/17/opinion/pretrial-ai.html (on file with the Columbia Human Rights Law Review) (explaining that risk assessment tools give recommendations that make future violence seem more predictable than it is). ↑

223.   Dabruzzo, supra note 207. ↑

224.   Alternatives to Detention, supra note 2 at 2, 14. ↑

225.   Schriro, supra note 153, at 20. ↑

226.   See supra Part II.B.2 and accompanying text. ↑

227.   Why Jurisdictions Use RATs, supra note 218. ↑

228.   See supra Part II. ↑

229.   Alternatives to Detention, supra note 2, at 2. ↑

230.    Evans & Koulish, supra note 3, at 795. ↑

231.    Padilla v. Immigr. & Customs Enf’t, 953 F.3d at 1145 (citing St. John v. McElroy, 917 F. Supp 243, 251 (S.D.N.Y. 1996)); see Alejandro Mayorkas Sworn in as Secretary of Homeland Security, supra note 38 (Mayorkas is a political appointee who oversees U.S. Immigration and Customs Enforcement). ↑

232.   Evans & Koulish, supra note 3, at 804–32. ↑

233.   Salgado-Diaz v. Ashcroft, 395 F.3d 1158, 1162 (9th Cir. 2005). ↑

234.   Id. ↑

235.   Id. ↑

236.   Padilla, 953 F.3d at 1145. ↑

237.   Id. ↑

238.        Fact Sheet: Immigration Courts, Nat’l Immigr. F. (Aug. 7, 2018), https://immigrationforum.org/article/fact-sheet-immigration-courts/ [https://perma.
     cc/62B8-6WSD]. ↑

239.   8 C.F.R. § 1236.1(c)(8). ↑

240.   8 C.F.R. § 1003.10. ↑

241.   Id. ↑

242.         Andrew R. Arthur, AG Certification Explained, Ctr. Immigr. Stud. (Nov. 5, 2019), https://cis.org/Arthur/AG-Certification-Explained [https://perma.cc/X4MQ-FMRW]. ↑

243.   Evans & Koulish, supra note 3, at 801. ↑

244.   Denise Gilman, To Loose the Bonds: The Deceptive Promise of Freedom from Pretrial Immigration Detention, 92 Ind. L.J. 157, 203 (2016). ↑

245.   Id. at 203–05. ↑

246.   Id. ↑

247.   Id. ↑

248.   Id. at 204–05; see also Evans & Koulish, supra note 3, at 823–34, 828–29 (discussing the factors used by the RCA). ↑

249.   Gilman, supra note 244, at 205. ↑

250.         What We Do, U.S. Citizenship & Immig. Servs., https://www.uscis.gov/about-us/mission-and-core-values/what-we-do [https://perma.cc/R3SB-H9WP]; ICE’s Mission, U.S. Immigr. & Customs Enf’t (Jan. 8, 2021), https://www.ice.gov/mission [https://perma.cc/E65E-22DA]. ↑

251.   How the United States Immigration System Works, Am. Immigr. Council (Oct. 10, 2019), https://www.americanimmigrationcouncil.org/research/how-united-states-immigration-system-works#:~:text=U.S.%20immigration%20law%20is%20complex,p
     rotecting%20refugees%2C%20and%20promoting%20diversity [https://perma.cc/8N
     NT-QAJM] (“U.S. immigration law is based on the following principles: the reunification of families, admitting immigrants with skills that are valuable to the U.S. economy, protecting refugees, and promoting diversity.”). ↑

252.   ICE’s Mission, supra note 250. ↑

253.   Evans & Koulish, supra note 3, at 827. ↑

254.   Id. at 828–29. ↑

255.   Id. ↑

256.   Id. at 833. (“By keying the flight risk assessment and the public safety risk assessment to prosecutorial priorities, the RCA ties detention recommendations to enforcement policy.”) There also lies the argument that even though violation of a visa program or illegal entry may be somewhat probative to the likelihood of court appearance, the cost of detention would outweigh the costs of a missed court appearance or abscondment. Therefore, RATs may also be calibrated to consider the cost of detention in relation to risk. See Crystal Yang, Toward an Optimal Bail System, 92 N.Y.U. L. Rev. 1399, 1483–91 (2017). ↑

257.   Gilman, supra note 244, at 222–23. ↑

258.   Id. at 221–22. ↑

259.   Id. at 222. ↑

260.   Perhaps a body like the Sentencing Commission would be effective here. The U.S. Sentencing Commission is an “independent agency in the judicial branch created by the Sentencing Reform Act of 1984” that was designed to establish sentencing policies and practices. A similar body could be established in the judicial branch to determine which factors to use in an immigration risk assessment tool, but more research would have to be done on this point. About, U.S. Sent’g Comm’n, https://www.ussc.gov/about-page [https://perma.cc/AY3Q-FBQ9]. ↑

261.   Pretrial Release Recommendation Decision Making Framework, supra note 90; N.M. Code R. § 5-401 (statue allowing for pretrial release). ↑

262.   ISAP III, which is currently operational, was preceded by ISAP II. Cong. Rsch. Serv., Immigration: Alternatives to Detention (ATD) Programs, R45804, at 7 (2019). ↑

263.   Id. at 7. ↑

264.   Id. at 2. ↑

265.   Id. ↑

266.   Id. at 8–9. ↑

267.   Evans & Koulish, supra note 3, at 804–05. ↑

268.   Oberhaus, supra note 149. ↑