The Challenge of Just Federal Sentencing for “Boat Defendants” Free

Abstract

In the U.S. federal criminal justice system, sentencing guidelines for drug-related offenses are pegged to drug type and quantity. This research explores sentencing for a group of federal drug offenders who are found with unusually large quantities of powder cocaine (typically >150 kg): poor, unarmed foreign nationals detained at sea and charged under the Maritime Drug Law Enforcement Act. As an unusual type of drug courier, this boat defendant population offers an instructive window into federal sentencing because the defendants are so “similarly situated,” yet they are prosecuted across multiple federal courts and have been sentenced by hundreds of judges. Using a defendant-level dataset built from publicly available data, the study explores sentencing outcomes across circuits, districts, courthouses, judges, and codefendants before and after the First Step Act (FSA) took effect in December 2018, which granted judges increased discretion to impose sentences below mandatory minimums. Results show that average imposed sentences were markedly lower than the estimated guideline minimums, suggesting that courts are recognizing drug weight to be a poor indicator of culpability, particularly post-FSA. The study also shows stark variation in imposed sentences within and across federal districts and among judges within a common court, which points to the unpredictability of sentencing outcomes despite a highly homogeneous offender population. The authors call for a critical reevaluation of federal drug sentencing guidelines to promote fairness and consistency and call for further research on judicial discretion in driving sentencing disparities.

I. Introduction

War on Drugs is the common name given to long-running policy efforts in the United States seeking to combat drug production, trafficking, sale, and use through widespread enforcement of criminal prohibitions. The adoption and application of severe sentencing regimes in drug prohibition enforcement have been especially controversial, prompting oft-expressed concerns about racial disparities, excessive incarceration and other harsh punishment practices, and undue and lasting consequences for affected communities.¹ The federal system’s approach to sentencing drug offenses has been subject to extended criticisms, in part because the type and quantity of the drug involved often drive sentencing outcomes, rather than factors more closely connected to the offenders’ motives and culpability.²

Federal sentences for drug crimes were developed in a domestic context based largely on the supposition that drug volumes proxied for seriousness of the offense, resulting in mandatory minimum sentencing terms and guideline sentencing ranges pegged to drug volumes. Many critics have detailed the problems with this sentencing approach, including those who point to the particular problems associated with the application of stiff quantity-based penalties to drug couriers. Couriers are low-level actors who are typically paid by drug trafficking organizations to transport large volumes of drugs via land, sea, or air, in exchange for a modest payday.³

This article describes research focused on a specific population of drug courier: the hundreds of men who are intercepted every year by the U.S. Coast Guard (USCG) on cocaine-laden boats and brought to the United States for prosecution under the Maritime Drug Law Enforcement Act (MDLEA).⁴ Typically caught with hundreds of kilos of high-purity cocaine, they face serious federal penalties, including a mandatory minimum statutory sentence of ten years in prison.

Our interest in this population is part of a broader project on the extraterritorial reach of U.S. law. Here, however, our goal was to explore the passage of these so-called boat defendants through the U.S. criminal justice system. We were particularly curious about how federal courts deal with a type of individual that, arguably, they were never intended to process: noncitizen, nonresident defendants detained far from U.S. shores while moving volumes of narcotics that far exceed the upper bounds contemplated by sentencing guidelines.⁵ Further, as mostly non-English-speaking indigent mariners plucked from the ocean and taken to U.S. courts, they are almost literally fish out of water, wholly unfamiliar with U.S. laws and legal procedures and thus susceptible to various forms of coercion and ill treatment.⁶

In the face of such unusual legal subjects, we wondered what courts consider to be a just sentence. In the process of exploring this question, we draw attention to a particularly vulnerable subgroup within the category of noncitizen drug offenders, an offender population that has received relatively little attention from sentencing scholars.⁷ In so doing, we believe we help shine a distinctive light on various aspects of federal drug policy and federal sentencing practices.

As a defendant-level analysis of federal drug sentencing, our approach is distinct. Rather than begin with the sentencing outcome from known data, we have instead built our own dataset to “follow the defendant” from the initial at-sea encounter with law enforcement through incarceration and eventual deportation from the United States. The approach offers important insights into many of the variables that are often opaque to sentencing analysts, including details of the offense, aspects of the defendant’s entry into the system, and post-sentencing dynamics. In addition, our sample of men detained at sea with cocaine ensures a population of defendants who are considered “similarly situated,” that is, with very little variation in many case and individual characteristics (table 1).⁸ Holding these factors constant allows us to focus more analytical attention on other variables’ impacts on sentencing, including how codefendants are rewarded (or not) for substantial assistance, the influence of different court norms, and the sentencing strategies of presiding judges.

This research builds on previous work on boat defendants by journalists, human rights organizations, and legal scholars. Collectively, they have drawn attention to different aspects of the men’s treatment by U.S. law enforcement and U.S. courts, including (a) human rights violations associated with due process delays; (b) a circuit court “split” around nexus requirements for MDLEA prosecution; (c) cross-court variation in the application of the “safety valve” sentencing provision; and (d) challenges to the constitutionality of U.S. jurisdiction over smuggling vessels without nationality.⁹ To date, however, there has been no systematic assessment of these defendants as a distinct offender population within the federal system and no large-scale analysis of their sentencing.

This analysis should be considered preliminary and exploratory. The next section outlines our methods and then describes the men’s passage through the criminal justice system, including several features peculiar to this population. Subsequent sections describe sentencing patterns and explore some of the factors underlying the sentencing variation we found. We conclude with a discussion of the study’s implications for scholars and practitioners.

II. Methods

Our analysis draws on our Extraterritorial Detainee Database, a repository of all boat cases since the advent of MDLEA in 1986, which we constructed using available dockets and case documents from the Public Access to Court Records (PACER) Website, and data accessed through the Bureau of Prison’s (BOP) “Inmate Search” tool. The companion article in this issue of Federal Sentencing Reporter describes our method in detail.¹⁰

We focus here on (a) defendants detained at sea (b) with cocaine (c) between FY 2014 and FY 2020¹¹ (d) for whom a sentence is known.¹² The final sample size of 2,144 boat offenders sentenced to imprisonment should be considered incomplete, as we likely missed defendants indicted under other lead charges and others whose sentences had not yet been imposed during the data collection phase. In addition, we could not find complete information on every defendant due to sealed records or missing information, so sample sizes (denoted by n) may vary depending on the variable under scrutiny.

III. Boat Defendants’ Path into and Through the Federal Criminal Justice System

A. Observed/Reported Crime

Between FY 2014 and FY 2020 U.S. military and intelligence assets tracked thousands of “primary maritime drug movements”: vessels smuggling high-purity powder cocaine and sometimes other drugs (cannabis, heroin) northward out of South America.¹³ The vessels ranged from narco subs to luxury sailboats, but the most common are twenty- to thirty-foot open fishing boats with outboard motors and crews of three or four men (mean crew size, 3.4; range, 1–14). Crews comprised seasoned fishermen and sometimes mechanics and hired hands. The men’s average age is 42 (median, 41; range, 19–77; n = 2,126). All but one were male.

The men are not kingpins. Most were hired to crew the vessels in exchange for between $5,000 and $50,000. Sentencing memos indicate that they are often motivated to participate in smuggling ventures out of economic desperation. For example, after the 2016 earthquake in Ecuador, which devastated fish stocks and coastal livelihoods and left fishermen struggling for alternative income sources, there was a surge in Ecuadorian fisherman detained with cocaine at sea.¹⁴ The growing numbers of Venezuelans interdicted at sea between FY 2014 and FY 2020 also appear to reflect that country’s increasingly dire economic situation.¹⁵ In Colombia and Mexico, mariners also described being forced to join smuggling crews under duress.¹⁶

B. Entry into the System

1. Detention at Sea and Drug Quantity

Of the thousands of vessels that moved drugs northward out of South America by sea every year, the USCG, the principal United States agency tasked with maritime drug law enforcement, reported intercepting less than 10 percent.¹⁷ All U.S.-led maritime drug interdictions in the Western Hemisphere follow a standard protocol coordinated out of the Joint Interagency Task Force–South based in Key West, Florida.¹⁸ Government auditors’ reports show that interdictions conducted between FY 2014 and FY 2020 resulted in at-sea detention of 3,638 mariners as suspected drug smugglers, an average of 455 per year.¹⁹ Three-quarters of the men were detained in the eastern Pacific, the rest in the Caribbean, following the interception of their vessels by USCG cutters and others by USCG law enforcement detachments aboard U.S. Navy ships.²⁰ Smuggler boat crews are very rarely armed—we found fewer than five cases in which guns were encountered during an at-sea interdiction. Suspected smuggling vessels were usually interdicted alone, although in some cases boats were caught as one was refueling another, or as drugs were being transferred between boats at sea. In those instances, the crew of all vessels involved were detained together and typically processed as codefendants in a single case.

Of the men detained at sea, about 14 percent appear to have been transported to other countries for prosecution; the remainder were destined for the United States. Of those, however, roughly 344 (9.5%) appear to have been released without prosecution before arriving in the United States.²¹ Of the remaining 2,770 men, 2,312 (83.5%) were found on vessels with powder cocaine. The average cocaine load was very large: 854 kilograms (median, 722 kg; range, 1–7,683 kg), and 90 percent of vessels held at least 150 kilograms. Load volumes were typically higher on vessels intercepted in the eastern Pacific (mean, 892 kg) than in the Caribbean (mean, 720 kg). Crews sometimes jettisoned their drug load during the interdiction; if no cocaine bales were recovered but cocaine residue was subsequently found on the vessel, the men were typically charged with less than two kilograms. Sixty-seven men were interdicted on boats that were also smuggling other types of drugs.

2. Role in the Offense and Nationalities

When interdicted by law enforcement, crews sometimes identified the vessel’s “master” or “captain”; we infer from court documents that the first individual listed on the indictment (27% of sample) is likely to be the person authorities determined to be “in charge” and therefore to play a leadership or supervisory role in the offense.

Other roles include so-called load guards, men flown from a cocaine-destination country to a cocaine-exporting country to accompany a shipment north. In the eastern Pacific load guards are often from Guatemala and Mexico; in the Caribbean men from the Dominican Republic often play this role. This is one reason that crews (and thus codefendants) are of mixed nationality. It is also why in some of our analyses specific nationalities can proxy for the load guard role. We were unable to consistently track other roles, such as the “minor role” that some courts recognize for crew members who appear to have little knowledge of or involvement in the offense.

Less than 1 percent of the men in the sample appeared to be U.S. citizens;²² 98 percent are from majority-Spanish-speaking countries and can be assumed to be Hispanophone.²³ Most are from Ecuador and Colombia, followed by the Dominican Republic, Venezuela, Mexico, and Central American nations; the remainder are from an additional eighteen countries. Not surprisingly, Ecuadorians and Colombians tend to work the eastern Pacific smuggling routes, while Colombians, Venezuelans, and Dominicans work the Caribbean routes (table 2).

We found evidence that 21 men in the sample (1%) had a prior U.S. felony conviction, usually from a prior smuggling attempt. This is likely an underestimate since it was difficult to track criminal history for all defendants.

3. Prearrest Detention, Arrest, and Intercourt Transfer

Following an interdiction, the suspected smugglers were detained for some time on the open decks of USCG cutters for transfer to the United States.²⁴ Their paths into the United States were rarely direct. Many were transferred between USCG cutters and were routed through Panama City and Guantánamo Bay. The 99 percent that are not U.S. citizens were subsequently paroled into the United States and arrested upon landing on U.S. soil, on a date that usually coincided with that of the case filing. However, about 13 percent (n = 307) were transferred from the jurisdiction in which they first arrived to another jurisdiction for prosecution. Such intercourt transfers were rare before 2017 (<5%/year) but became more common afterward, following revision to the relevant venue statute.²⁵

Excluding men transferred between jurisdictions (which caused a lag between arrest and case filing), an average of 19 days elapsed between the men being detained at sea and the date of their case filing (n = 1,851; median, 16; range, 0–77). About one in five (19%) appear to have spent more than 30 days in prearrest detention.

There is surprisingly little correlation among the number of days taken to deliver the men to the United States, the location of the ocean interdiction, and the location of the court in which they are prosecuted. For example, only 74 of the 1,285 men picked up in the eastern Pacific were indicted in California—the rest were taken to courts on the Gulf and East Coasts. Diplomacy can matter more than distance. For example, it took 77 days for the crew of a Cameroonian-flagged vessel to be transported from the Caribbean interdiction site to eventual arrest in Tampa, in part because Cameroon took over two months to waive its jurisdiction over the vessel.²⁶

C. Prosecution

The men were charged under one or more of the Title 46 statutes; the vast majority were also charged with 21 U.S.C. § 960(b)(1) and other counts. We are unaware of any cases in which charges were dropped prior to adjudication. We found no instances in which the defendants were released on bail; all appear to have been detained pretrial.

1. Distribution by Court

Boat prosecutions in this time period were strongly concentrated in the Eleventh Circuit, followed by the First Circuit (fig.1). Because of the uneven distribution of cases, we focus our analysis on the four circuits and the five districts within them that prosecuted at least forty boat defendants between FY 2014 and FY 2020: Florida’s Middle and Southern Districts, California’s Southern District, New York’s Southern District, and Puerto Rico (table 3).

2. Venue Specialization

Tampa (MD-Fla.) prosecuted the majority of boat defendants (55% of total, 69% of those prosecuted in the Eleventh Circuit). This concentration appears to be related to Tampa’s central role in an Organized Crime Drug Enforcement Task Force, known as Operation Panama Express (PANEX), begun in the 1990s.²⁷ Under PANEX, Tampa’s federal prosecutors work in coordination with other agencies²⁸ and with Colombian counterparts to “degrade” transnational drug trafficking networks, especially those based in Colombia. It therefore seems likely that the boat defendants remanded to Tampa were those that the DOJ assessed to be potentially useful to ongoing investigations into Colombia-based transnational criminal groups.²⁹

The Tampa courthouse has therefore become the country’s boat case specialist, where boat cases can dominate the court’s criminal drug caseload.³⁰ The court has reportedly developed a deep bench of experienced public defenders, empaneled Criminal Justice Act (CJA) attorneys,³¹ judges, prosecutors, and Spanish court translators, all of whom are well versed in the particularities of the defendants and the offense. Perhaps for this reason, Tampa prosecutors tend to skip the complaint hearing that is common in other courts.³² These factors contribute to efficiencies in defendant intake and processing: Tampa-bound detainees spent significantly less time in prearrest detention than those destined for other courts, and their case lengths were about three months shorter than in other courts.³³

Some sentencing researchers in other contexts have correlated court efficiency with shorter sentence lengths.³⁴ The Tampa federal court, however, has a general reputation for being “tough” on felony offenders.³⁵ Tampa’s relatively punitive approach may result from specific court cultures. In the case of boat defendants, for example, both Tampa prosecutors and defense attorneys appear confident that sentences are determined by well-understood norms, governed by a “pre-programmed, fixed game with a set of rules.”³⁶ Defense attorneys report, for example, that judges routinely deny minor role adjustments for boat defendants, which can contribute to higher recommended guideline ranges and, ultimately, higher sentences.³⁷ Another court-specific consideration is that, given Tampa’s central role in PANEX, prosecutors may be particularly incentivized to induce cooperation by rewarding only cooperators with recommended sentencing benefits³⁸ and by advocating for more punitive sentences for noncooperators.³⁹

D. Adjudication

In our sample of men found at sea with powder cocaine between FY 2014 and FY 2020 and prosecuted in the United States, all but two were declared indigent and publicly defended. For typical cases, one man is represented by a federal public defender (23%, 532 of 2,312) while his codefendants are represented by attorneys appointed under the CJA (77%, n = 1,780). (A fairly standard arrangement in drug cases with multiple defendants, to avoid any potential conflicts of interest that could arise from representing multiple codefendants.)

We found plea agreements for 84 percent of the men in the sample (n = 1,942). This is an underestimate since it does not include defendants who had not yet pleaded guilty during our data collection phase (ca. 2021–23), nor does it include those who pled guilty but did not sign an agreement. At least fifteen defendants had all charges dismissed, although most of these were later reindicted. (Our sample includes only those re-indicted individuals who received a sentence of imprisonment by or before FY 2023.) At least eleven defendants were acquitted after a trial.

1. Cooperation and Its Risks

For prosecutors, boat defendants are potentially vital sources of information about the criminal networks that hired them. Further, the men can embody the jurisdictional link that allows the United States, via conspiracy charges, to extradite “higher-up” members of drug trafficking networks who reside in other countries—targets that might be difficult to prosecute otherwise.⁴⁰

Once detained, boat defendants had multiple opportunities to provide the cooperation, described in sentencing parlance as “substantial assistance,” that prosecutors seek and reward at sentencing, starting with their first debrief upon arriving on U.S. soil through formal proffer interviews with prosecutors, and then again both prior to and after sentencing. Dockets did not allow us to consistently track the presence of formal motions filed by prosecutors to indicate “substantial assistance to authorities.”⁴¹ However, we know that about 30 percent of defendants in our sample (n = 582 of 1,918; data not available for all defendants) pleaded guilty days (if not weeks) before their codefendants. Timing is often an important marker of cooperation because the codefendant who is “first in” is typically most likely to get credit from prosecutors for offering substantial assistance and therefore most likely to receive a formal motion advocating for a sentence reduction below the guidelines and below any applicable statutory mandatory minimums.⁴²

We therefore used “first in” to proxy for pre-sentencing cooperation in our analysis, though we acknowledge this is an imperfect proxy for who did and did not receive formal credit from prosecutors for providing substantial assistance.⁴³ In addition, even in cases in which defendants did not receive such a motion from prosecutors, sentencing judges may still be influenced at sentencing if the defendant or his attorney highlights that he was willing to cooperate.

Boat defendants who offer information about transnational drug trafficking organizations face a variety of risks—risks that are recognized and often invoked by defense attorneys, prosecutors, and judges alike. Those risks include the risk of being a target of violence or extortion in county jails (pretrial)⁴⁴ and in federal prisons (post-sentencing), where boat defendants and those they implicate can be held in the same facility. It also means risks to their families at home during their incarceration⁴⁵ and risks when they return home after their sentences.⁴⁶

2. Time in Adjudication

The median time that defendants spent in U.S. jail pre-sentencing (their individual “case length”) was 6.4 months (mean, 8.8 months), which represents the time between initial case opening at arrest through the court’s imposition of sentence. Case lengths increased dramatically after 2019, coincident with the COVID pandemic. Case lengths were significantly shorter in Tampa than in other districts.⁴⁷

3. Sympathy for Boat Defendants

Defense attorneys described using sentencing memos to engage judges’ sympathy toward their clients, usually by emphasizing their low education levels, childhoods characterized by poverty and violence, ongoing responsibility for family members, lack of economic opportunity, trauma from experience of civil war or criminal violence, and economic or physical coercion and exposure to violence by criminal actors. Defense attorneys report that judges at sentencing can be influenced by defendants’ personal stories.⁴⁸ However, we were unable to measure the effect of such information on sentencing outcomes.

E. Sentencing and Sanctions

Of the 2,313 men found with cocaine, 2,167 men were convicted. Of these, 2,144 received a prison sentence (at least 5 were sentenced to “time served”; the sentences of 18 others are unknown). The 2,144 men sentenced to incarceration represent 93% of the 2,312 we estimated to have been detained at sea with cocaine and brought to the United States for prosecution under MDLEA for maritime cocaine trafficking.⁴⁹ We did not track how many appealed, though many of the plea agreements we reviewed included waiver provisions that limited or eliminated the grounds on which defendants could do so.

1. Sentencing Before and After the First Step Act of 2018

With the 1994 Crime Bill, Congress created what has been called the “safety valve” provision, allowing federal courts to sentence certain low-level drug offenders below an otherwise applicable statutory mandatory minimum sentence as long as the defendants meet several criteria.⁵⁰ Although most boat defendants generally met those criteria, prior to December 2018 the men were ineligible for safety valve consideration because MDLEA offenses were not included under the original provision. As a result, judges inclined to sentence a boat defendant below an applicable mandatory minimum triggered by drug quantities would be statutorily unable to do so unless a prosecutor filed a formal cooperation motion attesting to “substantial assistance” provided.

In December 2018 Congress enacted major federal sentencing reform known as the First Step Act (FSA), which includes provisions expressly extending the safety valve provision to cover offenders charged under MDLEA and other offenses. This statute gave judges new authority to impose sentences below mandatory minimum terms for boat defendants who met the safety valve criteria.

Of the 2,144 men in our sample who were sentenced to time in prison, 1,377 (63%) were sentenced in the five years before December 2018, and 792 (37%) were sentenced in the five years after (including our data collection phase, ca. 2021–23). This breakdown makes it possible to observe the potential impact of the FSA on sentencing outcomes.

F. Corrections

1. Distribution Among Federal Prisons

Men in our sample served time in at least 89 federal prisons in 33 different states. About 30 percent of the sample (n = 642) were no longer in BOP custody at the time of our data entry; we do not know where they were incarcerated. Of the 1,502 we did track to a facility, the largest number were held in the (privately run, now closed) McRae Correctional Facility in Georgia (n = 244, 16.2%), followed by the (privately run, now repurposed) North Lake Correctional Facility in Michigan (n = 97, 4.5%), the Federal Corrections Institution Fort Dix in New Jersey (n = 90; 6%), and Yazoo City Medium in Mississippi (n = 64, 4.3%). These data reflect conditions at the time of our data entry (2021–23), a period during which the men were moved among multiple prisons, due in part to the pandemic and to the Biden administration’s closure of privately run prisons. It therefore does not represent current patterns of incarceration for boat offenders.

2. Access to Prison Programs

As a prison population, boat offenders are unusual. They are foreign nationals who were paroled into the United States and therefore have an “A” number indicating that they are in the country legally and are “documented.” They are therefore distinct from “undocumented criminal aliens” and thus not the target of special federal rules for that population.⁵¹ At the same time, their records all include orders for immediate deportation upon completion of their sentences, which may be perceived as evidence of undocumented status. These dual characteristics appear to create confusion around whether or not boat offenders are eligible to earn sentence-reducing credits while incarcerated.⁵²

3. Rule 35(b)

Motions under rule 35(b) are filed by prosecutors to recognize, post-sentencing, substantial assistance that an offender provides, allowing offenders to reduce their sentences after initial imposition.⁵³ Incarcerated offenders often try to generate useful information while in prison and can be encouraged to do so by judges at sentencing.⁵⁴ We were unable to systematically track the use of rule 35 because dockets did not always show these motions, and even when they do, they may be filed years after an offender is incarcerated.

4. Prisoner Transfers

The International Prisoner Transfer Unit in the Department of Justice Criminal Division maintains bilateral treaties that allow prisoners to be returned to their home country prior to completing their sentences. The home country then assumes responsibility for administering the transferred sentence.⁵⁵ Countries covered by criminal transfer treaties relevant to this research include all Central American countries, Mexico, Ecuador, and Venezuela.⁵⁶ Ecuador has in the past been particularly active in repatriating boat offenders.⁵⁷ Colombia does not have a prisoner transfer treaty with the United States. In our sample, 131 men were released from prison after serving less than 50 percent of their sentences, of whom three-quarters were Ecuadorian citizens, likely repatriated under a prisoner transfer treaty.⁵⁸

G. Deportation and Parole

Men who complete their sentences in the United States are transferred from BOP custody to Immigration and Customs Enforcement, where according to court documents they stay in custody for about three months before being deported home. The majority of men in our sample had also been sentenced to five years of supervised release, but as deportees, these conditions are meaningful only if they reoffend in a U.S. jurisdiction within five years. We found a few men who were caught again at sea within five years of completing a former MDLEA sentence and so faced a sentencing enhancement over and above that imposed on other repeat offenders.

IV. Sentence Lengths: Patterns

A. Guideline Recommended Minimum Sentence Length

Given the near absence of guns or observed violence, the severity of the offense in these MDLEA cases is calibrated under federal sentencing law primarily by the drug quantity involved. Among the men sentenced to prison time, 98 percent were found with more than 5 kilograms of powder cocaine and therefore were subject to the federal statutory mandatory minimum sentence for this quantity, which is 10 years of imprisonment (120 months). In addition to a minimum set by statute, drug quantities also serve as the principal factor in determining applicable recommended (advisory) sentencing ranges under the Federal Sentencing Guidelines. Notably, 91 percent of the defendants were found with cocaine quantities of 150 or more kilograms, and the vast majority were found with 450 or more kilograms (table 4). Such large drug quantities produce very high base offense levels under the Federal Sentencing Guidelines, as detailed in table 4.

A defendant’s criminal history is the other key factor that determines advisory sentencing ranges under the Federal Sentencing Guidelines. The majority of boat defendants in our sample were Criminal History Category I, with little or no criminal history in the U.S. Even as first offenders, however, being found with more than 150 kilograms of cocaine is generally associated with guideline ranges starting at 188 months (about 15 years) and increasing up to 293 months, or over 24 years (table 4).

Based on these applicable guideline provisions, we calculated each individual’s estimated recommended guideline minimum sentence using their known criminal histories and the reported cocaine quantities with which they were detained.⁵⁹ The result was an average guideline minimum of 216 months, or 18 years (range, 51–262 months; n = 2,132). For 75 percent of our sample, we calculated a 235-month guideline minimum.

Two caveats are important here for understanding how we estimated applicable guideline ranges. First, the guideline minimums that we calculated are based on the quantities of cocaine reported in court documents as being present on the smuggling vessel. This may not, however, be the volume a defendant was charged with trafficking (we did not systematically record the drug quantity cited in the indictment). For example, in Tampa and in Manhattan, it appears that prosecutors often charge boat defendants with trafficking “more than 5 kg” of cocaine, irrespective of the actual quantities in the boat.⁶⁰ That quantity is enough to qualify for the federal mandatory minimum of ten years but is well below the guideline range suggested by the actual volume of cocaine present. In contrast, the San Diego and Miami courthouses appear to indict based on the actual volumes of cocaine found, resulting in base offense levels that range from 32 to 38.⁶¹ Second, we were not able to systematically assess other facts and factors (e.g., “role in the offense”) that can sometimes change the final offense level used by judges at sentencing.

B. Actual Sentence Lengths

Actual prison sentence lengths imposed by judges in boat cases averaged 108.5 months (median, 108; range, 15–360, n = 2,144), dramatically lower than the advisory guideline minimum average of 216 months. The difference between the minimum guideline and actual sentences was –108 months, or 9 years (n = 2,132; median, –115 months; range, +159 to - 217 months) and was normally distributed (fig. 2).

As figure 2 indicates, 94 percent of defendants in our sample were sentenced below the minimum guideline for the quantity of cocaine found at the time of the offense (<0 on x-axis), 1 percent at the guideline (0 on x-axis), and 5 percent above it (>0 on x-axis). In fact, only 2.5 percent of defendants with an estimated base offense level of 38 were sentenced within the corresponding guideline range, and only 0.6 percent above it (data not shown).

Overall, sentence lengths showed very little relationship to the drug quantities with which these offenders were detained (fig.3). This reality could in part reflect how prosecutors pursued charges and sentences, as defendants were more likely to be charged with “more than 5 kg” regardless of the actual quantity of cocaine found on the smuggling vessel; that is, the precise drug quantity did not clearly drive prosecutorial decision making. It also likely reflects the broader views of sentencing judges that the precise drug quantities involved in these offenses are poor proxies for defendant culpability and the advancement of relevant sentencing goals.

1. Changes over Time

The FSA had a clear impact on sentencing. Before December 2018, the average sentence for boat defendants in our sample was 115 months (median, 120 months), and 61 percent were sentenced at or above the mandatory minimum of ten years of imprisonment. After the passage of the FSA, the average sentence length dropped by 17.4 months to an average of 97.6 months (median, 94.5), leaving only 30% sentenced at or above the mandatory minimum of ten years.⁶² The difference in sentences before/after the FSA is highly statistically significant.⁶³ Moreover, the FSA correlated not just with lower sentences overall but also with a contraction in the range of sentences imposed, with fewer outliers than in years prior to the FSA (fig. 4).

2. Variation by Circuit

The aggregate sentencing picture depicted in figure 4 belies considerable variation in the sentences imposed on boat defendants across federal circuits and districts. For example, there was a stark difference in median sentence lengths imposed in the Ninth versus Eleventh Circuits: 72 versus 120 months, respectively, a difference of 4 years (table 5). The circuit with the lowest median sentences overall was the Second Circuit, in which judges imposed median sentences of 54 months (n = 44) (table 5). Thus, it is reasonable to wonder if divergent sentencing cultures may also have an impact on outcomes.

3. Variation by District

Variation in district-level sentencing was marked in our data. For example, despite comparable guideline minimums for all years,⁶⁴ the median sentence in Florida’s Middle and Southern Districts was 120 months but only 90 in the District of Puerto Rico and 72 in the Southern District of California (table 5). Those interdistrict differences were more pronounced before and less pronounced after the FSA became law in December 2018, when boat defendants became eligible for safety valve relief.⁶⁵ In Florida’s Southern District, for example, after the FSA, median sentences fell by 36 months (from 120 to 84 months). In Puerto Rico median sentences fell by 21 months (from 108 to 87 months), with 55 percent of defendants sentenced below the 10-year mandatory minimum prior to the FSA and 78 percent sentenced below that minimum after 2018.

4. Sentencing Variation Within Districts

The three courthouses in the Southern District of Florida that consistently heard boat cases offer a chance to explore within-district sentencing variation. We found that mean sentences imposed in Key West were consistently higher than those imposed in Miami or in Fort Lauderdale, both before and after the FSA, with sentences in Key West averaging 20 months more than those in Miami. Nevertheless, Key West sentencings appear to have responded more dramatically to the FSA, dropping average sentences by 34 months, or almost 3 years (table 5).

5. Venue Experience

Tampa (MD-Fla.) saw 55 percent of all boat defendants and processed them efficiently. That court is also associated with some of the highest average (117.5) sentences in the country (table 5). The relatively “tough” sentences for boat defendants in Tampa align with both the court’s punitive reputation and what might be expected given the court’s central role in PANEX, in which prosecutors and judges may pursue and impose sentences that are known to be severe enough for most to incentivize cooperation (see above). This hypothesis appears to be supported by the fact that the Tampa court also had a particularly high level of sentencing variation (average sentence: 117.5 months; median 120; standard deviation 37.4; range 24–360 months). That variation existed both before and after the FSA and points to the fact that aggregate sentencing data obscure the routine imposition of sentences that are both markedly lower (possibly for cooperators) and markedly higher (for non-cooperators) than the mean or median measures suggest.

6. Within-Court Sentencing Variation among Judges

We found marked differences in the sentence lengths imposed by judges within a common court, suggesting idiosyncratic responses to existing guidelines and different applications of sentencing discretion. Looking at the period following the FSA, we found that judges appeared to have applied this discretionary power differently. In Miami (SD-Fla.), for example, the median sentence for all judges was 101 months before the FSA and 84 months after (table 5). However, within that court, presiding Judge W⁶⁶ stands as an outlier who imposed significantly higher sentences (by 26 months) after the FSA.⁶⁷

7. Judges in Tampa

The twelve Tampa judges in our sample have the most experience with boat defendants. As such, they offer a rare opportunity to explore judicial discretion in sentencing within a single courthouse and across a population of similarly situated defendants. Here, we focus on three judges: presiding Judges X, Y, and Z (table 6). As of 2023, they had been on the bench for 30, 31, and 15 years, respectively. These judges were chosen because their imposed sentences tended to vary most from the court average. For example, while the court overall sentenced 65 percent of boat defendants to the mandatory minimum of 120 months or less (all years), Judge X only sentenced 55 percent at or below 120 months, and Judge Y sentenced only 40 percent below that minimum. Judge Z, however, skewed in the opposite direction, sentencing 87 percent of all defendants below the mandatory minimum (table 6).

These three judges sentenced 494 boat defendants to prison time, or about 40 percent of all Tampa defendants in our sample. Before the FSA, both Judges X and Y imposed average sentences of 137.3 and 128.3 months, which corresponded to an average of 89 and 96 months lower than the guideline minimum. These judgments were similar to the court average, which was 124.5 months, or 95.1 months lower than the guideline minimum. Judge Z, in contrast, imposed a much lower average sentence of 112.5 months, or an average of 104 months below the guideline minimum.

After the FSA, that disparity widened. Judges X and Y maintained their pre-FSA practices of imposing sentences that were both on average 93.2 and 92.5 months below the guideline minimum. Judge Z, however, appeared to respond forcefully to the FSA’s passage. Post-FSA, the median sentence this judge imposed was 141.8 months below guideline (mean: 67.5 months)—much less than what they had imposed prior to the FSA (the difference was highly significant: Pr[T < t] = 0.0000, two-sample t-test with equal variances). Judge Z’s post-FSA median sentence was therefore 75 months lower than the median sentences imposed during the same time period by their colleague Judge Y, and 48 months lower than the Tampa court overall (table 6).

Some of these variations in sentencing appear to be driven by the three judges’ differential willingness to reward for cooperation.⁶⁸ For example, before the FSA, the median sentence imposed by Judges X and Y for all defendants (120 months and 135 months, respectively) was the same as for those who pleaded guilty first (“first in”). In effect, these two judges did not appear to reward for substantial assistance (to the extent that first-in defendants showed extra levels of cooperation with prosecutors). Judge Z, however, imposed significantly lower sentences on the first to plead. During this period, court records show that Judge Z was also interested in understanding how prosecutors might recommend more than one codefendant for cooperation-related sentencing reductions.⁶⁹

The FSA seemingly had distinct effects on the three judges’ treatment of those who cooperated. Judges X’s and Y’s sentencing practices changed little: they continued to impose the same sentences on cooperators that they had before the FSA (median = 120 months), despite the fact that the court median declined to 97 months. As before the FSA, Judge Z continued to reward those who were first in, but their slightly lower sentences were not significantly different from those of defendants who pleaded out after their codefendants. In effect, Judge Z’s sentencing practices post-FSA no longer showed the marked leniency toward cooperators that they had before. This suggests that the FSA allowed Judge Z to revise their sentencing calculations dramatically downward for all defendants. This washed out the benefits of cooperation, and the range in the sentences this judge imposed contracted noticeably.

V. Explaining Sentence Length Variation

The preceding section describes how sentence lengths imposed on boat offenders varied according to factors such as drug quantity, cooperation, court, and judge. To more systematically assess the relative importance of each variable, we ran a series of multivariate regressions. To partially control for the influence of drug weight on sentencing, we used data only from the 91 percent of defendants found with 150 or more kilograms of cocaine (see table 4). Our dependent variable was the imposed sentence in months for all those sentenced to prison time.

A. Expected Relationships Between Dependent and Independent Variables

Here we list the relationships we expected between the dependent variable (imposed sentence length in months) and the independent variables, which we categorize below by case-level, defendant-level, and court-level (+ and – emphasize the expected direction of the correlation):

CASE CHARACTERISTICS

• Estimated minimum guideline (table 4): Likely to be strongly and positively correlated with sentence lengths, since it is the starting point for prosecutorial and judicial sentencing decisions (+).

• Severity of the offense (kg cocaine): Higher cocaine volumes seem an obvious predictor of longer sentence lengths, although the relationship does not appear strong (see fig. 3) (+).

• “Multidrug” smuggling event: The presence of drugs in addition to cocaine on the vessel might proxy for a more sophisticated criminal enterprise and perhaps greater culpability of the crew (+).

• Detention in eastern Pacific Ocean: Boats smuggling cocaine in the eastern Pacific carry larger loads, possibly suggestive of higher levels of organization and culpability (+).

INDIVIDUAL CHARACTERISTICS

• Criminal history: Defendants with a U.S. criminal history will face higher sentences (+).

• Role in the offense: Boat masters (here proxied for by “lead” defendants) will be seen as more culpable, resulting in sentencing enhancement (+).⁷⁰

• Acceptance of responsibility: Having signed a plea agreement will be associated with lower sentences (–).

• Age and age squared: We square the age term because we anticipate a curvilinear relationship (inverted U), with younger and older men receiving less severe sentences (–) than middle-aged men (+).⁷¹

• Cooperation: Proxies for substantial assistance (as estimated by being “first in” to plead) will be associated with lower sentences, because cooperation is an important factor in allowing for sentences below the applicable mandatory minimum term and also for an additional sentence reduction, although the effect may weaken after the FSA (–).

• Country of origin: As described in the next subsection, courts may be more sympathetic to Venezuelans (–) than to Colombians (+).⁷²

• Other mitigating factors: Those who spent more than 30 days in the extreme conditions of prearrest detention⁷³ (which defense attorneys described as potential due process violations) will be associated with lower sentences (–).

• ‘Safety valve’ effect: Sentences will be significantly lower after the FSA (–).

VENUE CHARACTERISTICS

• Courthouses: Preliminary analyses described above suggest the influence of court cultures on sentencing practices at the courthouse level. We used dummy variables to control for the effect of Tampa (+), Miami (+), San Juan (+), Manhattan (–), and San Diego (–).

• Judge: Judicial discretion appears important in explaining sentencing variation, as noted above. We used dummy variables to control for the effects of four judges whose sentencing stood out in the context of their courts: Judge X (+), Judge Y (+), and Judge Z (–) in Tampa (MD-Fla), and Judge W (–) in the Miami and Key West courts of Florida’s Southern District.

B. Model Limitations

The models do not account for judges’ consideration of the risks associated with defendants’ cooperation. It seems likely that these risks and associated duress are considered by judges as “personal mitigating factors” that may compound the downward variance associated by the cooperation alone.

In addition, the models cannot control for the impact of prosecutorial and defense attorney discretionary decision making. Prosecutors in different cases may push for different sentencing terms in plea agreements based on various known and unknown factors; different defense attorneys may likewise advise their clients to accept or reject plea offers based on various known and unknown factors.

C. Regression Results

We ran four regression models (table 7): model A, on sentence lengths for the total sample (of boat defendants interdicted FYs 2014–20); model B, sentence lengths imposed before the FSA; model C, sentence lengths imposed after the FSA; and model D, sentence lengths imposed after the FSA in the Tampa courthouse of Florida’s Middle District. All models were highly significant, and each explained between one-third and one-half of the overall variation in sentence lengths (adjusted R² = 0.35–0.54). For all but a few variables (discussed below), the signs of the coefficients confirmed our expectations about the directions of the relationship between dependent and independent variables, but the significance of the relationships was in some cases surprising. Below we review the results of models A–C with respect to the importance of different “passage points” through which defendants experience the criminal justice system. We describe the Tampa-specific model D in the following subsection.

Model A, which covers the whole sample of sentenced boat offenders found with ≥150 kilograms of powder cocaine (n = 1,492), shows that cocaine volumes, as predicted, were positively correlated with but explained little about the variation in sentence lengths (for every additional kg of cocaine, the increase in sentence imposed was 0.003 months). The presence of other drugs had no significant effect. Being detained in the eastern Pacific, however, was strongly predictive of being sentenced to 9.6 more months, all else held equal (although the significance weakened in models B and C). We are not sure why. It is possible that, compared to relatively short Caribbean routes, cocaine trafficking in the eastern Pacific, which typically takes six days and requires boat crews to rendezvous with other boats for loading/unloading and refueling, requires more capital, planning, and at-sea coordination and is therefore dominated by transnational criminal groups. As a result, defendants working those routes may be considered more culpable; it is also possible that the information those boat crews can offer are particularly valued for prosecutions of transnational criminal groups, and prosecutors may pursue higher sentences to encourage greater cooperation.

Conversely, being held for more than thirty days in prearrest detention was not predictive of shorter sentence lengths in model A (–5.0 months, p>|t| = 0.57) but strongly predictive in model C, such that those who were detained longer than thirty days pre-arrest could expect an imposed sentence of twenty-one fewer months (after the FSA; –21.0 months, p>|t| = 0.000). This finding may reflect judges’ increased discretion to acknowledge the duress of prearrest detention, especially following national media coverage of the hardships faced by boat detainees chained on USCG vessels for long periods.⁷⁴

Among defendant characteristics, being middle-aged was positively correlated with longer sentences, but age overall had no predictive strength. However, being the lead defendant—a status associated with being “master” of the smuggling vessel—was significantly predictive of longer sentences across all models.

Models A–C show that venue matters: As expected, in all three models the courthouse in which an offender is sentenced is highly predictive of the sentence length, even when all other variables are held constant. However, the sign of the coefficients was surprising. Only the Tampa courthouse was associated with higher sentences, all else held equal—a strong confirmation of this court’s punitive reputation. In contrast, the Miami, Manhattan, San Diego, and San Juan courthouses were predictive of lower sentences in models A–C. These findings possibly point to powerful “court cultures” that can overdetermine sentencing patterns, even accounting for key sentencing calculations.

As predicted, being sentenced by Judge X or Judge Y in Tampa or by Judge W in Florida’s Southern District (Judge W imposed sentences on boat defendants in both Miami and Key West during this period) was strongly predictive of higher sentences, all else held equal. After the FSA (model C), offenders who came before Judge Y in the Tampa court could expect about 20 more months than other boat defendants, all else held equal. Meanwhile, those sentenced by Judge Z in the same court could expect about 37 months less. Overall, these results suggest that court cultures are clearly not all-determining, because they can mask considerable variation among judges within a given court.

The recommended guideline minimum sentence length was, as expected, positively correlated with imposed sentences, and the variable was significant (p>|t| < 0.05 in all models). However, the magnitude of the effect was small. For example, Model C estimates that for every one-month increase in the guideline minimum, the actual sentence imposed increased by only 0.24 months. As expected, having a criminal history in the United States was associated with significantly longer sentences (44 more months before the FSA and 47 more after). Admission of guilt (via plea agreement) was particularly important before the FSA (model B), when it was associated with a strong likelihood of receiving 35 fewer months than others, all else held equal.

Timing of sentencing with respect to the FSA clearly mattered. Model A confirms that overall sentences declined by 13 months after 2018, all else held equal (p>|t| = 0.000). The FSA had a noticeable effect on the leniency offered to cooperators, as our preliminary analysis showed. That is, before the FSA, those who pleaded out first (a proxy for cooperation) were associated with 5 months’ shorter sentences, all else held equal (model B). After the FSA, the relationship reversed direction and ceased to be significant, likely because judges were more able to account for other factors in sentencing below a mandatory minimum.

D. Multivariate Model for Tampa

Model D explores sentencing variation in Tampa (MD-Fla.) alone (n = 242; table 7). Not surprisingly, the model results vary little from model C, in which Tampa accounted for 242 of the 423 observations. However, the importance of due process concerns stands out in the Tampa-only regression: Being held in detention for more than 30 days prior to the initiation of the case (usually coincident with the date of the defendant’s arrest) was significant and associated with sentences that were 37.3 months shorter than for others, all else held equal (p>|t| = 0.000). Meanwhile, particularly severe sentence enhancements were faced by repeat offenders in Tampa, equivalent to 72.2 months, or about 6 years (p>|t| = 0.000). The model also confirms the findings of Model C, showing that even when other variables are held at their means, in the context of the Tampa court Judge Z was associated with sentence lengths that were 37.2 months shorter than those imposed by other judges.

As expected, given Tampa’s concentration of PANEX cases, the reward of substantial assistance (first in) was predictive of sentences that were lower by 7.1 months, even after the FSA. Although the significance was not strong (p>|t| = 0.04), it was significant, unlike the effect in the larger sample (Model C; p>|t| = 0.933). This suggests that the Tampa court may be unusual in rewarding for substantial assistance at the sentencing stage.

VI. Conclusions and Future Research

This research is a preliminary analysis of a rich dataset describing the experience and fate of a relatively unknown and underexamined population of drug offenders within the federal criminal justice system. The men typically are poor and unarmed foreign nationals charged with two or three codefendants under Title 46, and 91 percent were found with more than 150 kilograms of cocaine on vessels subject to the jurisdiction of the United States. They were detained at sea via a standard interdiction protocol, typically had no criminal history, and once paroled into the United States were publicly defended. However, these “similarly situated” boat defendants were prosecuted and sentenced in at least 9 different circuits and 20 districts in 34 courthouses by 175 different judges, allowing for exploration of sentencing outcomes across these categories. Further, our study explores boat defendant sentencing over a period that included passage of the FSA, which provided an opportunity to explore how judges responded to an increase in their discretionary authority to sentence below a certain mandatory minimum sentencing provision for drug offenses.⁷⁵ Below, we highlight some key insights from the work and identify areas for future research.

A. Sentences Depart Dramatically from Estimated Guideline Minimums

This research was animated by the question of what courts consider a just sentence for boat defendants. According to data from the sentencing practices across a range of federal courts, the answer is apparently 108 months (9 years) in prison—the average sentence imposed on the men in our sample. This sentencing average is dramatically lower than the estimated average guideline minimum suggested by the drug quantities found when the men were interdicted at sea—a finding that points to courts’ willingness to vary substantially from lengthy drug sentences recommended by the guidelines. This finding likely reflects a collective understanding among federal judges (and likely also among federal prosecutors) that, for this type of drug courier, drug weights are poor proxies for culpability. This interpretation is strengthened by the fact that average boat sentences fell even further below guideline minimums after the FSA in December 2018, suggesting that, once given new authority to reduce sentences below mandatory minimums, the collective response of the courts was to do so.

B. Extreme Detention Raising Due Process Concerns Are Reflected in Sentencing Outcomes

An unusual finding, and one specific to this offender population, is the apparent willingness of courts, after the passage of the FSA in late 2018, to impose lower sentences for those defendants who were likely to have been held in an extreme detention environment for more than thirty days prior to arrest, including some time spent chained on the deck of a USCG cutter during transit to U.S. soil. This suggests that courts, on average, are acknowledging this extreme deprivation (and the due process concerns associated with it) with twenty-one fewer months in prison for those who endured it (see table 7, model C).

C. Marked Variation in Sentencing across Federal Courts

If one measure of just sentencing is uniformity in sentencing outcomes for similarly situated defendants, this study suggests that boat defendants experience the opposite: significant sentencing disparities between circuits, districts, courthouses, judges, and codefendants.

In the context of MDLEA defendants, much has been written about the Ninth versus Eleventh Circuit-court “split” on nexus requirements. The Ninth Circuit’s requirement of a demonstrable U.S. nexus for boat cases has been seen as influential in contributing to that circuit’s lower sentences, which may discourage prosecutors from bringing boat cases to the Ninth Circuit and is instead encouraging them to “venue shop” in the more legally amenable but logistically remote Eleventh Circuit.⁷⁶

Our research tells a different story. In fact, the Second Circuit is where boat offenders can expect the most lenient sentences, not the Ninth. And within the “punitive” Eleventh Circuit, it is the Key West court, not Tampa, that appears to have imposed the most punitive sentences. Overall, results point to the need for greater attention to local factors that may shape sentencing in those courts. Of particular interest is how the Second Circuit may be producing significantly lower sentences, including the possibility that it allows boat defendants to plead out to lesser, non–Title 46 charges, a theory advanced by Tampa U.S. Attorney Lee Bentley III in 2017.⁷⁷

Further, our findings indicate that it can take the U.S. Coast Guard just as long to transport men detained in the eastern Pacific to California as it does to bring them to Florida, which means that the Ninth Circuit should not, from a logistical standpoint, be considered the presumptive circuit for men detained in the Pacific. Finally—and importantly—the Eleventh Circuit is not only attractive to prosecutors because of the settled nexus issue there. It is also attractive from a processing standpoint, because its courts feature a deeper pool of expertise in boat prosecution and adjudication, especially in Tampa, given that court’s three-decade role in PANEX. In fact, the longer sentence lengths imposed on boat offenders in Tampa (relative to the national average) may have less to do with nexus requirements and more to do with sentencing practices designed to incentivize PANEX-relevant cooperation. More research is required, however, to explore these issues and their role in the concentration of boat cases in Florida’s Middle District.

D. Opening the “Black Box” of Judicial Discretion

Regardless of why the Eleventh Circuit—and Florida’s Middle District in particular—receive the most boat defendants, the Tampa courthouse appears highly efficient in processing them. That efficiency, however, does not produce uniformity in sentencing outcomes. In fact, the very different sentencing practices of Tampa’s experienced judges offer a glimpse into the role of judicial discretion for similarly situated offenders.⁷⁸ For example, we found that, controlling for all other known sentencing factors, some judges in Tampa maintained a steady, “tried-and-true” approach that did not appear to be influenced by the greater discretion allowed under the FSA, resulting in sentences that were more punitive than the court average. Further research might explore how those specific judges’ decisions are shaped by ideological, generational, or other issues. Meanwhile, other Tampa judges, including those with shorter tenure on the bench, appeared willing to question sentencing norms before the FSA and thereafter to embrace the greater sentencing discretion allowed under the expanded safety valve provision. As a result, their imposed sentence lengths fell dramatically below those of some of their court peers, while holding constant key characteristics of the offense, defendant, and adjudication.

These findings make a strong case for the need for more research into the role of judicial discretion in sentencing outcomes, with particular attention to individual judges. If the path to more just federal sentencing hinges on harmonizing judges’ decision-making practices, it is essential for scholars and policymakers—and judges themselves—to better evaluate what judges and their peers are doing and to openly debate the merits of specific sentencing approaches.

E. The Unpredictability of Sentencing Outcomes

If a measure of just sentencing is that similarly situated defendants can expect a predictable path through the criminal justice system,⁷⁹ then this study offers evidence of injustice. We show that considerable uncertainty characterizes boat defendants’ experience of the federal system. That uncertainty does not suddenly emerge at the sentencing stage but, rather, results from a series of chance events over which defendants themselves have little to no control: the chance of being detained at sea; the chance of being prosecuted in the United States; the chance of being remanded to one court and not another, to stand before one judge and not another; the chance of pleading out before crewmates; and so on. Figure 5 attempts to illustrate this “game of chance” for a subsample of particularly similarly situated men within our dataset: Colombian and Ecuadorian fishermen with no criminal history in the United States who were detained with at least 450 kilograms of cocaine in the eastern Pacific Ocean and remanded to Tampa (n = 497), and then assigned to either Judge Z (n = 86) or Judge Y (n = 78). Despite common initial characteristics, some “lucked out” in how and when they moved through the system and were ultimately sentenced to an average of 68 months’ imprisonment (just over 5.5 years). Others were dealt a “bad hand” and experienced a sequence of events that led to an average sentence of 138 months (11.5 years).

Some variations in sentencing outcomes are inevitable, as different individual judges take stock of different individual defendants. But our set of defendants are remarkably similar with respect to their offense conduct and personal backgrounds, yet despite this some dramatic sentencing variations were evident. The lesson and a pathway for needed reform may track back to the Federal Sentencing Guidelines, which were created and designed to try to reduce unwarranted sentencing disparities and yet have long been criticized for being so focused on drug quantities in determining recommended sentencing ranges in drug cases. It seems judges in all these cases view the guideline range, here dramatically increased by drug quantities, to serve as a very poor guide for a just and effective sentence. Consequently, sentencing outcomes are significantly shaped not by the Guidelines but by how much individual judges decide the deviate from them. Unsurprisingly, judges may not be consistent in how far they deviate from sentencing recommendations, though they seem to collectively find them wanting and inappropriate for serving a just outcome. The results of our data collection and analysis suggest that the U.S. Sentencing Commission, and perhaps also Congress, ought to seriously consider whether a robust and complete revision of the drug sentencing laws and guidelines is needed in all cases, and particularly in courier cases, to better achieve both just and more consistent sentencing outcomes.

Acknowledgments

This work was supported by grants from Ohio State University’s (OSU) Mershon Center for International Security Studies to OSU’s Interdiction Lab. We are grateful to the Drug Enforcement and Policy Center at OSU’s Moritz College of Law for hosting the June 2024 Webinar and to all Webinar participants, whose insights greatly improved this work.

Notes