The Predicates of Military Detention at Guantánamo: The Role of Individual Acts and Affiliations

• Author: Aziz Z. Huq
• Date: 01 December 2016
• DOI: http://doi.org/10.1111/jels.12126
• Published date: 01 December 2016

The Predicates of Military Detention at

Guant

anamo: The Role of Individual Acts

and Affiliations

Aziz Z. Huq*

The military detentions at Guant

anamo have provoked intense controversy. Judges,

legislators, and military officials disagree sharply about the scope of detention authority. To

date, little is known about the granular determinants of decisions to transfer detainees

from custody at the Cuban base or to prolong their detention. This article exploits a novel

data set of classified documents disclosed by the Wikileaks organization to explore two

questions now dividing officials, legislators, and judges. First, what are the detainee actions

and affiliations that predict longer detentions? Second, what is the relationship between a

determination that a detainee committed certain acts and the further inference of

affiliation with a proscribed terrorist organization? These inquiries isolate both continuities

and divergences from the law “on the books.”

I. INTRODUCTION

Since the first 156 military detainees were transferred to the Guantanamo Bay Naval

Base in January 2002, legislators, federal agencies, and federal judges have disagreed

sharply over the legality, wisdom, and even nomenclature of their detentions (Bravin

2013; Frackt 2012). Disagreement partly stems from the absence of a clear legislative

framework. The statutory authority initially invoked for their detention—the 2001

Authorization for the Use of Military Force, or AUMF (Pub. L. 107-40, codified at 115

Stat. 224 [2001])—did not define directly the scope of lawful detention. Such authority

has instead been inferred from its purpose and context (Hamdi v. Rumsfeld 542 U.S. 507

[2004]), leaving wide latitude for intradepartmental and interbranch disagreement.

There is also uncertainty about what explains the observed heterogeneity in detention

duration. The detentions at Guant

anamo display significant internal variance. Some

*Frank and Bernice J. Greenberg Professor of Law, University of Chicago Law School, 1111 E. 60th St., Chicago,

IL 60636; email: huq@uchicago.edu. The Frank Cicero Jr. Fund provided support for this research. Ruixi Mao,

Charles Zhang, Dan Marcin, and Morgen Miller provided extensive help preparing and analyzing data for this

article. Thanks to Joey Burton for facilitating their work. Several research assistants did invaluable work. I am

especially grateful to Steve Donohue; Melissa Wu, Caitlin Foley, Sam Geloso. Paxton Williams, and Cecilia Wang

also did terrific work. Workshop participants at King’s College London, the University of Chicago Law School,

and the 2015 Conference on Empirical Legal Studies provided insightful comments. All errors remain my own.

567

Journal of Empirical Legal Studies

Volume 13, Issue 4, 567–602, December 2016

individuals were held for only a few hundred days; others have been in custody for

more than 5,000 days. The mean duration

1

of all detainees is 2,180 days; its standard

deviation is 1,525 days. Little is known about the roots of this heterogeneity.

This article exploits a novel archive of classified data, initially disclosed by the

Wikileaks organization, to investigate the relationship between individual-level factual

predicates and the duration of individuals’ detention at Guant

anamo. Specifically, it

takes up two questions. The first focuses simply on the empirical grounds of prolonged

detention: What specific factual findings about an individual’s past actions or affiliations

predict longer custody? The second concerns ongoing disputes as to whether the finding

that a detainee committed certain acts can support the inference that he is affiliated

with a proscribed terrorist organization: Given information about what a detainee has

done, that is, can an inference be drawn about whether he belongs to an organization

covered by the AUMF? By focusing on the intensive margin of detention policy (i.e.,

how long a person is detained) rather than the extensive margin (i.e., the decision to

detain in the first instance), the article diverges from earlier scholarship that conceptu-

alizes detention as a dichotomous variable (e.g., Chesney 2011; Hafetz 2012; Goodman

2009).

To these ends, this study exploits a novel source of data. In April 2011, the

Wikileaks organization disclosed a set of 765 classified “detainee assessments” (Savage

et al. 2011). These individual assessments, which remain classified, were produced

between 2002 and 2009 by intelligence analysts within the military joint task force (JTF)

responsible for counterterrorism detention operations at Guant

anamo. The assessments

vary widely in length, detail, and date of production. Almost all contain a narrative of

the reasons that the government credited for detaining an individual Guant

anamo

detainee. These narratives pick out certain acts and affiliations of the individual detainee

as relevant to the security risk posed by the detainee and to his intelligence value.

Further, most of the 765 assessments contain ordinal assessments of the individual’s risk

value and intelligence value on a four-tiered scale, as well a recommendation either to

release or to detain the given individual. This archive furnishes the best publicly avail-

able evidence of the government’s internal reasoning process and conclusions in respect

to specific detainees.

To complement the Wikileaks archive, this study employs a separate archive of

data, compiled on an ongoing basis by the New York Times (NYT), that quantifies the

duration of detention. By associating the Wikileaks data with the NYT timing data, it is

possible to relate individual-level factual predicates identified by the government to

individual-level detention durations.

The article’s three main findings can be summarized as follows. First, some (but

not all) acts and affiliations have independent value as predictors of both the risk values

assigned to detainees and also the length of individuals’ detention, but the exact range

of predictors varies. For example, participation in active hostilities and affiliation with a

radical mosque, while identified in government documents as risk factors, do not

1

This is calculated as of December 20, 2015, when the study terminated.

568 Huq

predict assignment of a higher risk rating. Second, not all acts and affiliations are equally

predictive of riskiness and prolonged detention. The best predictor of a higher risk

assignment is the only variable related to behavior during detention: the making of

explicit threats while in custody. In contrast, explicit threats and affiliation with a radical

mosque had little or no effect on the duration of detention. The best predictors of lon-

ger detention are affiliation with the Taliban or al Qaeda. Some acts, such as attendance

at a terrorist training camp, participation in hostilities, or residence at a terrorist guest-

house, also predict longer incarcerations, but to a lesser extent.

Third, some courts have suggested that evidence of affiliation with a proscribed

organization is a necessary predicate of lawful detention. However, they have then

inferred membership in the Taliban or al Qaeda from certain acts (e.g., attendance at a

training camp or residence at a guesthouse). I demonstrate that Al Qaeda membership

is tightly correlated to several variables, including attendance at a training camp and res-

idence at a terrorist guesthouse. Membership in the Taliban, by contrast, is not positive-

ly correlated with any behavioral predicate of detention.

The study proceeds as follows. Section II summarizes debates about the predicates

of military detention. Section III details the data sets used in this study, in particular the

Wikileaks files. Section IV describes the manner in which data were extracted from these

archives. Section V considers whether the data contained in detainees’ assessments

should be viewed as evidence of the reasons for detention, or solely as post hoc justifica-

tions for transfer decisions. Section VI presents descriptive summary statistics concerning

population-level patterns. Section VII then uses those data in the study’s central econo-

metric analyses of the relationships between individual acts and affiliations on the one

hand, to risk value, detention duration, and detention/release decisions on the other.

Finally, Section VIII contextualizes the findings of this study within wider legal and poli-

cy debates about military detention.

II. THE ROLE OF INDIVIDUAL ACTS AND AFFILIATIONS IN

DETENTION POLICY

A. The Unsettled Law of Detention

Terrorism-related detention operations at Guant

anamo initially proceeded under the

September 16, 2001 AUMF (Pub. L. 107-40, codified at 115 Stat. 224 [2001]). This envis-

aged the “use” of “all necessary and appropriate force against those nations, organiza-

tions, or persons [the president] determines planned, authorized, committed, or aided

the terrorist attacks [of 9/11].” The 2001 AUMF does not expressly refer to detention.

In 2006, Congress enacted the Military Commissions Act (MCA). Defining the epony-

mous tribunals’ scope, this Act defines an “unlawful enemy combatant” as a person

“engaged in hostilities or who has purposefully and materially supported hostilities

against the United States or its co-belligerents who is not a lawful enemy combatant

(including a person who is part of the Taliban, al Qaeda, or associated forces)” (10

U.S.C. § 948a(I)(i) [2006]). In 2011, Congress clarified the 2001 AUMF by providing

expressly that military detention authority encompassed any person “who was a part of

569

Predicates of Military Detention at Guantanamo