Supermax Administration and
the Eighth Amendment:
Deference, Discretion, and
Double Bunking, 1986–2010
Keramet Reiter*
This Article explores the constitutionality of supermax prisons,
focusing on one of the earliest and largest supermaxes in the United States,
California’s Pelican Bay State Prison, and one of the first court cases to
consider supermax constitutionality, Madrid v. Gomez. Although
international human rights bodies have condemned indeterminate periods of
solitary confinement in the harsh conditions of American supermax prisons
as torture, no American court has found that the harsh conditions or long
durations of confinement in these institutions violate the Eighth Amendment
prohibition against cruel and unusual punishment. This Article examines
why. Part I suggests the simplest answer: courts evaluating supermax
confinement have simply deferred to prison administrators’ assertions that
the institutions are constitutional. Introducing the cases that have examined
supermax prison conditions, with a specific focus on Madrid, this part
analyzes the role of judicial deference in Madrid and other, similar decisions.
Parts II and III present evidence of two key, overlooked mechanisms of this
judicial deference. Part II analyzes historical evidence from archives and oral
history interviews to demonstrate how prison administrators in California
and Arizona worked to design supermax institutions maximally free of
public oversight, demonstrating that deference to prison officials is not only
initiated by courts, but also actively cultivated by supermax designers and
administrators. A second mechanism of judicial deference to supermax
administrators is a presumption that no empirical evidence exists with which
to evaluate prison administrators’ claims. But this presumption is
unwarranted. Part III provides an example of the kinds of empirical
* Keramet A. Reiter, Assistant Professor, Department of Criminology, Law & Society, and School of
Law, University of California, Irvine; Ph.D. in Jurisprudence and Social Policy, University of California,
Berkeley 2012; J.D., University of California, Berkeley School of Law 2009; M.A., John Jay College of
Criminal Justice, City University of New York 2006; B.A., Harvard University 2003. Thanks especially
to the participants in the 2013 Michigan Human Rights Workshop, who provided helpful, critical
comments on a prior draft of this piece.

89

90

UC IRVINE LAW REVIEW

[Vol. 5:89

evaluations of supermax administrators’ claims that are possible, even in
the context of limited evidence and broad administrative discretion. Part IV
suggests two judicial and two nonjudicial avenues to check the broad
discretion in the design and operation of supermax prisons; this discretion
has produced a lack of public, and especially judicial, oversight of potentially
egregious constitutional violations. Supermaxes represent the most extreme
end of mass incarceration in the United States. Looking closely at the
motivations behind supermaxes, their ungovernability, and their potential
for reform therefore has implications for mass incarceration more broadly. If
we could grapple with limiting the supermax, perhaps this would give us
clues as to how we might grapple with limiting other aspects of the supersized
American criminal justice system.
Introduction ...................................................................................................................... 90
I. Madrid and the Constitutionality of Supermax Prisons ......................................... 94
A. Madrid v. Gomez ................................................................................................ 96
1. Assessing the Constitutionality of Supermax Isolation .................... 98
2. Assessing the Constitutionality of Supermax Double Bunking .... 101
B. Deference in Prison Conditions Cases Generally.................................... 103
C. Deference in Supermax Cases After Madrid v. Gomez ............................. 108
1. Comer v. Stewart ....................................................................................... 109
2. Wilkinson v. Austin ................................................................................. 111
II. Deference by Oversight Avoidance ....................................................................... 112
A. Discretion in Design .................................................................................... 113
B. Discretion in Operation ............................................................................... 117
III. Deference by Empirical Assumption ................................................................... 121
A. Double Bunking ............................................................................................ 122
B. Initial Indicators of Violence ...................................................................... 125
C. In Search of a Relationship Between Supermaxes, Overcrowding,
and Violence ............................................................................................... 127
IV. Less Deference, More Visibility, Better Incentives .......................................... 133
Conclusion ........................................................................................................................ 138
INTRODUCTION
“Supermax” prisoners in the United States live for months (and often years)
at a time in windowless, poured-concrete boxes. Each “box,” or cell, includes
roughly eighty square feet of space—about the size of a handicap bathroom stall,
or a parking space. Prisoners spend at least twenty-two hours of every day in these
cells, under fluorescent lights that never turn off. They only leave their cells four or
five times per week, for showers or for brief, solitary exercise periods in “dog
runs”—concrete pens with roofs at least partially open to natural light. A guard in
a central booth controls each prisoner’s automated, steel cell door. Phone calls are
not permitted, unless a prisoner is to be notified of a death in his immediate family.
The rare human interaction—with a doctor, a lawyer, or an in-person family visit—

2015]

SUPERMAX ADMINISTRATION

91

takes place from within a cage on wheels or behind bulletproof glass. Usually,
prisoners live in total solitary confinement in these cells. Sometimes, however,
prison systems have more supermax prisoners than supermax prison cells. In these
cases, prisoners in supermax cells are double bunked, spending twenty-two or more
hours of every day in that eighty-square-foot cell with a cellmate.
Correctional administrators, not judges, assign prisoners to supermaxes, based
on in-prison behavioral assessments. These are prisons within prisons, where the
deprivation conditions are extreme. Supermaxes are a modern phenomenon,
growing out of and expanding with America’s turn toward harsher sentencing
policies and mass incarceration in the 1980s.1 Arizona opened the first supermax in
1986, and California quickly copied the Arizona prototype, opening two more
supermaxes in 1988 and 1989.2 The federal prison system opened the United States
Penitentiary, Administrative Maximum (nicknamed AdMax, or ADX) in Florence,
Colorado, in 1994.3 (The ADX designers had looked to other state supermax
facilities as a model for ADX, and designed an institution structurally similar to the
California and Arizona supermaxes.)4 By the late 1990s, nearly every state had a
supermax prison.5 Each facility costs between $8 and $230 million to build.6 Today,
scholars estimate that between twenty thousand and eighty thousand prisoners
across the United States are held in supermax prisons, under conditions of longterm solitary confinement (or sometimes conditions of long-term segregation, with
one other cellmate, as the case may be).7
This massive investment in building the infrastructure to maintain thousands
of prisoners in long-term isolation is surprising in light of both U.S. prison history
and U.S. prison law prior to the 1980s. The earliest U.S. prisons, which opened in
Pennsylvania and New York between 1790 and 1829, maintained prisoners in
1. See, e.g., Franklin E. Zimring, The Scale of Imprisonment in the United States: Twentieth Century
Patterns and Twenty-First Century Prospects, 100 J. CRIM. L. & CRIMINOLOGY 1225, 1230–33 (2010).
2. MONA LYNCH, SUNBELT JUSTICE: ARIZONA AND THE TRANSFORMATION OF A MERICAN
PUNISHMENT 5, 135–38 (2010); Keramet A. Reiter, Parole, Snitch, or Die: California’s Supermax Prisons and
Prisoners, 1997–2007, 14 PUNISHMENT & SOC’Y 530, 532, 535 (2012).
3. Ed Pilkington, ADX Florence Supermax Prison: The Alcatraz of the Rockies, GUARDIAN (Apr. 10,
2012, 2:00 PM), http://www.theguardian.com/world/2012/apr/10/abu-hamza-isolationsupermax-prison.
4. Interview with Anonymous, Architect, Fed. Bureau of Prisons (Feb. 2, 2011) (on file with
author).
5. NAT’L INST. OF CORR., U.S. DEP’T OF JUSTICE, SUPERMAX HOUSING: A SURVEY OF
CURRENT PRACTICE 4–6 (1997), available at http://static.nicic.gov/Library/013722.pdf; Alexandra
Naday et al., The Elusive Data on Supermax Confinement, 88 PRISON J. 69, 75 (2008).
6. DEL NORTE CNTY. BD. OF SUPERVISORS, COMPREHENSIVE ECONOMIC DEVELOPMENT
STRATEGY: DEL NORTE COUNTY, CALIFORNIA 2006–2008, at 10 (2006), available at http://www
.jirwinconsulting.com/CEDS050506.pdf; Mark Roberts, Butner Supermax Prison ‘Locks Down’ Inmates,
WRAL.COM (Dec. 1, 1998), http://www.wral.com/news/local/story/131674/; The “Supermax” Prison
in Butner Will House the Worst of the Worst, WRAL.COM (May 13, 1998), http://www.wral.com/news/
local/story/131663/.
7. Naday et al., supra note 5 at 77; Jean Casella & James Ridgeway, How Many Prisoners Are in
Solitary Confinement in the United States?, SOLITARYWATCH (Feb. 1, 2012), http://solitarywatch.com/
2012/02/01/how-many-prisoners-are-in-solitary-confinement-in-the-united-states/.

92

UC IRVINE LAW REVIEW

[Vol. 5:89

solitary confinement.8 Within eighteen months of opening in 1821, New York’s
Auburn State Prison abandoned the practice of solitary confinement because so
many prisoners lost their minds.9 Pennsylvania was slower to abandon the solitary
confinement system. In 1842, novelist and social critic Charles Dickens condemned
the ongoing use of solitary confinement in Pennsylvania’s Eastern State Penitentiary
as “worse than any torture of the body.”10 And by the mid-nineteenth century,
hundreds of deaths and cases of insanity had been documented and attributed to
the use of long-term solitary confinement in New York and Pennsylvania.11 In light
of this evidence, many jurisdictions took steps to limit the duration of solitary
confinement, or to eliminate the practice entirely.12 At the turn of the twentieth
century, the Supreme Court conclusively condemned the practice, albeit in dicta. In
1890, in In re Medley, the Supreme Court devoted more than a page (of a short,
fifteen-page opinion) to describing in no uncertain terms the severity and futility of
solitary confinement as a punishment: “A considerable number of the prisoners fell,
after even a short confinement, into a semi-fatuous condition . . . and others became
violently insane; others still, committed suicide.”13 In re Medley marked the end of a
century of on-again-off-again uses of extended solitary confinement in the early
United States penitentiaries, until the 1980s.
Even in its modern form (well-lit, technologically advanced, hyperhygienic),
long-term isolation in supermax prisons like California’s Pelican Bay and the Federal
Bureau of Prisons’ ADX, has been correlated with hallucinations and symptoms of
post-traumatic stress disorder, increased suicide risk, exacerbations of existing
mental health problems, and higher recidivism rates.14 And international human
8. See LAWRENCE M. FRIEDMAN, A HISTORY OF A MERICAN LAW 219–20 (3d ed. 2005);
ADAM JAY H IRSCH, THE RISE OF THE PENITENTIARY: PRISONS AND PUNISHMENT IN EARLY
AMERICA 59–60 (1992); M ICHAEL IGNATIEFF, A JUST MEASURE OF PAIN: THE PENITENTIARY IN
THE INDUSTRIAL REVOLUTION, 1750–1850, at 194–97 (1978); ORLANDO F. LEWIS, THE
DEVELOPMENT OF AMERICAN PRISONS AND PRISON CUSTOMS, 1776–1845, at 118–19 (Patterson
Smith 1967) (1922); DAVID J. ROTHMAN, THE DISCOVERY OF THE ASYLUM: SOCIAL ORDER AND
DISORDER IN THE NEW REPUBLIC 82–83, 94–96 (1971).
9. Peter Scharff Smith, The Effects of Solitary Confinement on Prison Inmates: A Brief History and Review
of the Literature, 34 CRIME & JUST. 441, 457 (2006); see also REBECCA M. MCLENNAN, THE CRISIS OF
IMPRISONMENT: PROTEST, POLITICS, AND THE MAKING OF THE A MERICAN PENAL STATE, 1776–
1941, at 57 (2008).
10. CHARLES DICKENS, Philadelphia and Its Solitary Prison, in A MERICAN NOTES AND PICTURES
FROM ITALY 97, 99 (Oxford Univ. Press 1957) (1842), available at http://www.onlineliterature.com/dickens/americannotes/8/.
11. Craig Haney & Mona Lynch, Regulating Prisons of the Future: A Psychological Analysis of Supermax
and Solitary Confinement, 23 N.Y.U. REV. L. & SOC. CHANGE 477, 484 (1997).
12. Id. at 484–86.
13. In re Medley, 134 U.S. 160, 168 (1890) (holding that Colorado’s new policy of keeping deathsentenced prisoners in solitary confinement constituted a significant additional punishment beyond the
sentence of death; holding that the solitary confinement policy violated the ex post facto clause as to
Medley, who had committed his crime before the solitary confinement policy became law; and
overturning Medley’s death sentence on this basis).
14. See, e.g., Haney & Lynch, supra note 11, at 506, 509, 521, 524, 530; Reiter, supra note 2, at
533, 552–56; Kevin Johnson, Inmate Suicides Linked to Solitary, USA TODAY (Dec. 27, 2006, 10:59 PM),
http://usatoday30.usatoday.com/news/nation/2006-12-27-inmate-suicides_x.htm; Mary Beth

2015]

SUPERMAX ADMINISTRATION

93

rights bodies have condemned modern supermaxes as places of torture that violate
international human rights laws.15 Yet no U.S. court has held that supermax prison
conditions, like those at Pelican Bay, violate the Eighth Amendment prohibition
against cruel and unusual punishment. This Article examines why not.
Specifically, this Article addresses two questions about the legality of
supermaxes. First, why have U.S. courts repeatedly upheld the constitutionality of
supermaxes? Second, should the constitutionality of supermaxes be reconsidered,
and if so, how?
California’s Pelican Bay State Prison is the focal point of the analysis in this
Article. California’s prison system is second in scale only to Texas’s.16 Pelican Bay
was one of the first and largest supermaxes built in the United States.17 And Pelican
Bay was the site of one of the first challenges to the constitutionality of
supermaximum security confinement, in the federal lawsuit Madrid v. Gomez.18 Many
courts have followed Judge Thelton Henderson’s reasoning in the Madrid case in
subsequent challenges to supermaxes located in other districts.19
Part I introduces the cases that have examined supermax prison conditions,
focusing especially on Madrid v. Gomez, which assessed the constitutionality of the
conditions at California’s Pelican Bay State Prison. In Madrid and other similar cases,
U.S. federal courts have held that supermaxes are necessary tools of safety and
security, based on the assertions of prison administrators that this is the case. In
fact, prison-law scholars have clearly established (and criticized) a broader pattern
of federal court deference to prison administrators’ claims, as discussed in Part I.
Generally, courts justify this deference because (1) prison administrators need
discretion to manage the difficult populations of people in prison, and (2) prison
administrators have expertise in their own management needs.
However, the supermax phenomenon reveals two key, overlooked
mechanisms of this deference. First, the deference is not only initiated by courts,
but also actively cultivated by supermax designers and administrators. The
correctional administrators who designed and ran the first supermaxes sought to
Pfeiffer, Prison Suicide Rates Rise; Solitary Confinement Adds to Risk, POUGHKEEPSIE J. (Apr. 11, 2011),
http://www.poughkeepsiejournal.com/article/20101017/NEWS01/112030002/SPECIAL-REPORT
-Prison-suicide-rates-rise-solitary-confinement-adds-risk; Peter Schworm, Before Suicide, Inmate Pleaded for
End to Isolation, BOS. GLOBE (Mar. 13, 2012), http://www.bostonglobe.com/metro/2012/03/12/
suspect-double-murder-takes-his-own-life-after-being-held-isolation/7osEIHfBDQp0VcFetpw0jI/
story.html.
15. See, e.g., A MNESTY INT’L, USA: THE EDGE OF ENDURANCE: PRISON CONDITIONS IN
CALIFORNIA’S SECURITY HOUSING UNITS 9–10, 28, 52 (2012), available at http://www.amnestyusa
.org/sites/default/files/edgeofendurancecaliforniareport.pdf; Solitary Confinement Should Be Banned in
Most Cases, UN Expert Says, UN NEWS CENTRE (Oct. 18, 2011), https://www.un.org/apps/news/
story.asp?NewsID=40097.
16. E. ANN CARSON & WILLIAM J. SABOL, U.S. DEP’T OF JUSTICE, NCJ 239808, PRISONERS
IN 2011, at 3 (2012).
17. See Reiter, supra note 2, at 531.
18. See Madrid v. Gomez, 889 F. Supp. 1146, 1155 (N.D. Cal. 1995).
19. See, e.g., Wilkinson v. Austin, 545 U.S. 209, 213 (2005) (discussed infra); Comer v. Stewart,
230 F. Supp. 2d 1016, 1023, 1033–34, 1071 (D. Ariz. 2002).

94

UC IRVINE LAW REVIEW

[Vol. 5:89

avoid all forms of public oversight—from legislators, media, and courts—of the
facilities. Continued judicial deference to supermax administrators further
exaggerates this institutionalized lack of oversight. Part II analyzes historical
evidence from archives and oral history interviews to demonstrate how prison
administrators in California and Arizona worked to design supermax institutions
maximally free of public oversight, and, over the next twenty-five years, continued
to operate supermaxes largely free of public oversight.
A second mechanism of judicial deference to supermax administrators is a
presumption that no empirical evidence exists with which to evaluate prison
administrators’ claims. But this presumption is unwarranted. In fact, prison
administrators’ claims about what management tools they require and how these
tools operate in practice are empirically testable. Part III provides an example of the
kinds of empirical evaluations of supermax administrators’ claims that are possible,
even in the context of limited evidence and broad administrative discretion. By
analyzing empirical data about supermax operation over the last twenty years, this
part seeks to demonstrate the importance of empirical data to any legal analysis of
the constitutionality of an institution of punishment.
Part IV suggests two judicial and two nonjudicial avenues to check the broad
discretion in the design and operation of supermax prisons, which has produced a
lack of public, and especially judicial, oversight of potentially egregious
constitutional violations. The judicial avenues seek to limit the degree of deference
given to correctional administrators imposing the most extreme possible in-prison
punishments, by expanding categorical restrictions excluding certain groups from
supermax prisons, and also by expanding the procedural protections preceding
supermax assignment. The nonjudicial avenues include increasing the visibility of
supermaxes by requiring more regular and more detailed production of data and by
facilitating regular, independent monitoring of supermaxes.
I.

M ADRID AND THE CONSTITUTIONALITY OF SUPERMAX PRISONS

The technologically advanced, extremely harsh conditions of the first
supermax prisons in Arizona and California quickly attracted judicial scrutiny. A
federal court in California led the inquiry. Judge Henderson, Chief Judge of the
Northern District Court of California from 1990 through 1997, remembered when
he and a few other judges on the court first heard about the conditions at Pelican
Bay: “[W]e were looking and our mouths were open, and we said [the judges to each
other]: ‘You can’t do that. That’s unconstitutional.’”20 Within three years of Pelican
Bay’s opening, Judge Henderson was overseeing a three-month long trial to evaluate
the constitutionality of the conditions at the prison. In January of 1995, he issued a
137-page opinion in the case of Madrid v. Gomez.21 For the next fifteen years, he

20. Interview with Thelton Henderson, Chief Judge, Dist. Court for the N. Dist. of Cal., in S.F.,
Cal. (May 24, 2011) (on file with author).
21. Madrid, 899 F. Supp. 1146.

2015]

SUPERMAX ADMINISTRATION

95

would oversee day-to-day policies and practices at the institution, reading regular
monitoring reports from prisoners’ rights attorneys who visited the institution at
least annually.22
Judge Henderson has a reputation as one of the most liberal federal district
court judges in the United States. He was one of the first African American lawyers
to work for the Civil Rights Division in the U.S. Department of Justice, where he
investigated the 1963 white supremacist bombing of the 16th Street Baptist Church
in Montgomery, Alabama.23 Since his early days as a civil rights attorney, Henderson
has presided over a number of controversial cases, including a case upholding
environmental protections for dolphins; a case overturning a murder conviction of
an alleged Black Panther member; a case in which he attempted to block
enforcement of California’s Proposition 209, which prohibited public entities from
engaging in affirmative action; and, most recently, a class action prison conditions
case in California in which he issued a population reduction order affecting more
than thirty thousand state prisoners.24 If any judge was predisposed to find the
conditions of confinement in a supermax like Pelican Bay unconstitutional, it was
Judge Henderson.
However, Judge Henderson never found that Pelican Bay’s harsh conditions
of confinement, or the practice of imposing indeterminately long solitary
confinement on some prisoners, were per se unconstitutional. In March of 2011,
Judge Henderson closed the Madrid case.25 That summer, dozens of Pelican Bay
prisoners coordinated a three-week-long, statewide prisoner hunger strike,
protesting their indefinite solitary confinement.26 At that point, more than five
hundred prisoners at Pelican Bay had been in solitary confinement for more than
ten years.27 And, with the Madrid case closed, they had little hope that the courts
would provide any potential avenue for reform. The international human rights
community, however, was paying closer attention in 2011 than it had been in the

22. See Interview with Thelton Henderson, supra note 20.
23. SOUL OF JUSTICE: THELTON HENDERSON’S A MERICAN JOURNEY (Ginzberg Video
Productions 2006).
24. Id.; see, e.g., Order to Reduce Prison Population, Coleman/Plata v. Brown, Nos. S-90-0520
LKK JFM P, C01-1351 TEH (E.D. & N.D. Cal. 2010) (Three-Judge Court); Earth Island Inst. v. Evans,
No. C03-0007TEH, 2004 WL 1774221 (N.D. Cal. Aug. 9, 2004) (redefining what constitutes dolphinsafe tuna); Coal. for Econ. Equity v. Wilson, 946 F. Supp. 1480 (N.D. Cal. 1996) (issuing a preliminary
injunction against enforcement of Proposition 209), vacated, 122 F.3d 692 (9th Cir. 1997); Spain v.
Rushen, 543 F. Supp. 757 (N.D. Cal. 1982) (issuing a writ of habeas corpus, functionally overturning
Johnny Spain’s in-prison murder conviction), aff’d, 701 F.2d 186 (9th Cir. 1983), vacated, 464 U.S. 114
(1983).
25. Judge Closes Prison Abuse Case After 20 Years, KQED NEWS (Mar. 22, 2011), http://
blogs.kqed.org/newsfix/2011/03/22/judge-closes-prison-abuse-case-after-20-years/.
26. Sal Rodriguez, One Year Anniversary of Pelican Bay Hunger Strike Against Solitary Confinement,
SOLITARY WATCH (July 3, 2012), http://solitarywatch.com/2012/07/03/one-year-anniversary-ofpelican-bay-hunger-strike-against-solitary-confinement/; see also Keramet Reiter, The Pelican Bay Hunger
Strike: Resistance within the Structural Constraints of a U.S. Supermax Prison, 113 S. ATLANTIC Q., 579 (2014).
27. Julie Small, Under Scrutiny, Pelican Bay Prison Officials Say They Target Only Gang Leaders (KPCC
Southern California Public Radio, Aug. 23, 2011).

96

UC IRVINE LAW REVIEW

[Vol. 5:89

late 1980s and early 1990s; Juan Mendez, the UN Special Rapporteur on Torture
said in August of 2011 that more than fifteen days in solitary confinement, in a
prison like Pelican Bay, could constitute torture in violation of international human
rights covenants.28
UN officials’ recent assertions that the conditions at Pelican Bay might
constitute torture, Judge Henderson’s initial instinct about California’s supermax
(“that’s unconstitutional”), and the 1890 Supreme Court decision dismissing solitary
confinement as futile, all inspire a central question about the Madrid decision and
others like it: by what rationale have U.S. courts repeatedly upheld the
constitutionality of supermaxes? This section first examines the rationale applied in
Madrid, then examines the rationales applied in subsequent, similar cases, and finally
contextualizes the supermax cases in the framework of a pattern of deference in
U.S. prisoners’ rights cases decided in the late twentieth and early twenty-first
centuries.
A. Madrid v. Gomez
In 1995, Judge Henderson released a substantial opinion evaluating the
constitutionality of multiple aspects of the conditions of confinement and the
operational policies at Pelican Bay State Prison. The Madrid decision documented a
number of constitutional violations at Pelican Bay State Prison, including the
practice of placing mentally ill prisoners in long-term solitary confinement with
minimal access to psychiatric treatment, and a correctional “code of silence” that
worked to systematically conceal allegations of excessive uses of force.29 The Madrid
court ordered the California Department of Corrections to work with plaintiffs’
lawyers to design a remedial plan to bring the institution into compliance with basic
constitutional norms prohibiting cruel and unusual punishment. The court also
appointed a special master to regularly monitor the institution’s progress.30
However, the court did not find that the basic conditions of confinement in the
Pelican Bay Security Housing Unit (SHU) were, in and of themselves,
unconstitutional. Specifically, the court considered (and found constitutional) two
common aspects of confinement in the Pelican Bay SHU, which will be discussed
in detail here: (1) the constitutionality of the basic, restrictive conditions of
confinement; and (2) the constitutionality of double bunking some prisoners under
these conditions.
For both of these challenges to the constitutionality of the conditions within
the Pelican Bay SHU, the Madrid court applied a then newly minted Supreme Court
rule articulated in Farmer v. Brennan,31 a 1994 prisoners’ rights case. In Farmer, the
28. Special Rapporteur of the Human Rights Council, Interim Report of the Special Rapporteur of the
Human Rights Council on Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment, U.N. Doc.
A/66/268 (Aug. 5, 2011).
29. Madrid v. Gomez, 889 F. Supp. 1156, 1259–60 (N.D. Cal. 1995).
30. Id. at 1282–83.
31. Farmer v. Brennan, 511 U.S. 825 (1994).

2015]

SUPERMAX ADMINISTRATION

97

Supreme Court established a two-prong rule for evaluating the constitutionality of
a condition or conditions of confinement: (1) the prisoner must claim a serious
actual deprivation, or a serious risk of harm, a claim that requires an objective
evaluation of seriousness; and (2) the prison officials inflicting the deprivation or
imposing the risk of harm must have done so unnecessarily and wantonly (i.e., the
official must have known of the risk of harm), a claim that requires a subjective
evaluation of the state of mind of prison officials.32 A prisoner’s initial burden of
proof under the Farmer test is high: he must both establish the existence of a
significant harm and also prove that the harm was intended, or at least that the
prison official knew of the likelihood of the harm.33 I discuss the Madrid court’s
application of Farmer to each aspect of confinement.

32. Id. at 834.
33. Id. at 837, 842 (defining deliberate indifference and describing the plaintiff’s burden of
proof). Interestingly, two of the concurrences in Farmer revolved around the issue of the reasonableness
of this second prong, and questioned the requirement that a prison official’s subjective state of mind
be established. Id. at 852 (Blackmun, J., concurring); id. at 858 (Stevens, J., concurring). Prisoners face
a particularly high burden in attempting to establish that a prison condition violates the Constitution.
Many of them are representing themselves; ninety-three percent of prisoner petitions are filed pro se
or without legal representation. ADMIN. OFFICE OF THE U.S COURTS, JUDICIAL BUSINESS OF THE
UNITED STATES COURTS: 2011 ANNUAL REPORT OF THE DIRECTOR 189
tbl.C-13, available at http://www.uscourts.gov/uscourts/Statistics/JudicialBusiness/2011/
JudicialBusiness2011.pdf. General population prisoners are in legally vulnerable positions stripped of
many of the rights of citizenship. They are also likely to be uneducated and to have limited access to a
lawyer or legal advice. In all but two states, prisoners are not allowed to vote. In most other states, even
people who have been released from prison are disenfranchised for some period of time, and sometimes
permanently. See Criminal Disenfranchisement Laws Across the United States, BRENNAN CENTER, https://
www.brennancenter.org/sites/default/files/analysis/RTV%20Map%2010%2016%2013.pdf.
According to the Prison Policy Initiative, nineteen percent of the U.S. prison population is completely
illiterate, and forty percent are “functionally illiterate,” which means someone would be “unable to write
a letter explaining a billing error.” Peter Wagner, Section I: Crime & Punishment in the U.S., PRISON POL’Y
INITIATIVE, http://www.prisonpolicy.org/prisonindex/rootsofcrime.html (last visited Oct. 18, 2014).
Moreover, prisoners seeking to bring lawsuits challenging the conditions of their confinement face not
just the social barrier of low literacy but legal barriers as well. For example, the Prison Litigation Reform
Act is national legislation passed in 1996 that limits the ability of prisoners to bring civil rights claims.
See Prison Litigation Reform Act (PLRA): Myths and Facts, SAVE COALITION, http://www
.savecoalition.org/myths.html (last visited Oct. 18, 2014) (noting that between 1981 and 1995, the rate
of prisoner filings of civil rights cases decreased from 29 per 1000 prisoners to 25 per 1000 prisoners,
while the prison population itself doubled); see also Margo Schlanger, Civil Rights Injunctions Over Time: A
Case Study of Jail and Prison Court Orders, 81 N.Y.U. L. REV. 550, 566 (2006). A supermax prisoner’s
vulnerability is further exaggerated by his placement in extreme isolation. Supermax prisoners usually
only have the ability to request legal research materials if they have a pending court deadline. Under
Lewis v. Casey, 518 U.S. 343, 351 (1996), the prisoner has the burden of proving that he was not able to
pursue a legal claim because of inadequate library access, so most prisons prioritize access for prisoners
who may be facing court filing deadlines. See also CTR. FOR CONSTITUTIONAL RIGHTS & NAT’L
LAWYERS GUILD, JAILHOUSE L AWYER’S H ANDBOOK 43–47 (Rachel Meeropol & Ian Head eds., 5th
ed. 2010). Since supermax prisoners have no contact with other prisoners, they usually cannot seek out
advice or help from so-called jailhouse lawyers. They are also not permitted to make phone calls, absent
an emergency. See Reiter, supra note 2, at 531; Reiter, supra note 26, at 580, 585.

98
1.

UC IRVINE LAW REVIEW

[Vol. 5:89

Assessing the Constitutionality of Supermax Isolation

The Madrid court applied the Farmer analysis to prisoners’ claims that the
conditions in the Pelican Bay SHU constituted cruel and unusual punishment. First,
the court established facts about the conditions in the supermax units. Second, it
evaluated whether there was a serious risk of harm to prisoners in these units based
on these conditions (the objective prong of Farmer). Ultimately, the court found that
there was no serious risk of harm to prisoners (the objective prong of Farmer), so
this obviated the need to evaluate the “state of mind” of the prison officials who
designed and operated these units (the subjective prong of Farmer). At each point
of analysis—factual description, evaluating risk of harm, and considering whether
to evaluate state of mind—the court incorporated a deferential consideration of
whether prison officials had articulated a legitimate penological justification for the
potentially unconstitutional restrictive conditions at issue.
First, the Madrid court described the conditions in the SHU in stark terms:
“[T]he conditions . . . may press the outer bounds of what most humans can
psychologically tolerate . . . .”34 The court described these “outer bounds” of human
tolerance in vivid detail: “The overall effect of the [supermax] is one of stark sterility
and unremitting monotony. Inmates can spend years without ever seeing any aspect
of the outside world except for a small patch of sky.”35 But even within these kinds
of descriptions, the court did not describe the conditions as decisively unreasonable:
“[T]he totality of the SHU conditions may be harsher than necessary to
accommodate the needs of the institution with respect to these populations.”36 In
describing the institution, the court weighed its harsh characteristics against what is
necessary for institutional safety and security. The court described the SHU as not
just a potentially intolerable place of stark and sterile conditions, but also as a place
that housed “some of the most anti-social and violence-prone prisoners in the
system.”37 Prison administrators, the court noted, have “the paramount
responsibility” and “remarkably difficult undertaking” of managing these
prisoners.38 The court, then, connected the description of the conditions in the SHU
to the necessity for these conditions as tools “to maintain the safety and security of
both staff and inmates.”39
This initial description of the Pelican Bay SHU, highlighting the violent nature
of the prisoners there and the importance of maintaining institutional safety and
security, did not incorporate the perspectives of prisoners and their lawyers (who
argued that some prisoners in the SHU were not dangerous, but mentally ill, and
that other prisoners did not deserve to be there), nor the first-hand observations of
Judge Henderson (who visited the Pelican Bay SHU numerous times over the
34.
35.
36.
37.
38.
39.

Madrid, 899 F. Supp. at 1267.
Id. at 1229, 1267.
Id. at 1263 (emphasis added).
Id. at 1160.
Id. at 1159.
Id. at 1159–60 (citations omitted).

2015]

SUPERMAX ADMINISTRATION

99

course of the Madrid litigation40). Rather, the initial description precisely reflected
the claims of correctional administrators about who was housed in the Pelican Bay
SHU and why.
Interviews I conducted with the California prison administrators who
designed the Pelican Bay SHU reveal the close alignment between how the Madrid
court described the institution and how correctional administrators described it. For
instance, Carl Larson, a prison administrator who oversaw the design and
construction of the Pelican Bay SHU in the 1980s, explained that revolutionary
groups in the prisons, like the Black Panthers and the Black Guerilla Family, incited
violence, which necessitated institutionalizing a long-term lock-up unit: “We needed
a SHU unit . . . we were having a hell of a time . . . we had eleven staff members
murdered in 1971.”41 Other administrators asserted that the SHU had indeed
accomplished its safety-and-security goals. Craig Brown, the director of California’s
Youth and Adult Correctional Authority (YACA), the agency that approved the
building of Pelican Bay, argued: “We had a severe violence problem, and it got a lot
better [after Pelican Bay opened] . . . less officers are killed at the hands of inmates
than in the ‘70s . . . I think the SHU did what it was supposed to do . . . I actually
think gangs got better.”42 The Madrid court’s descriptions of the SHU—as harsh
but perhaps necessary—reflected exactly these correctional explanations and
justifications for the SHU.43 And, as I will show in the next part, the same
administrators who articulated these justifications for the SHU in the 1990s and
2000s had designed the institution in the 1980s.44
Once the Madrid court established descriptive facts about the Pelican Bay
SHU—a stark place of restrictive confinement, but one necessary for isolating
dangerous prisoners in order to maintain institutional safety and security—the court
then evaluated the constitutionality of supermax conditions.45 The court
acknowledged in vivid terms the severity of the conditions in the Pelican Bay SHU,
but found that the risk of harm to most prisoners housed in these conditions did
not rise to the level of a truly serious deprivation or risk.46
In analyzing the question of the seriousness of the harm to prisoners housed
in SHU conditions (i.e., the objective prong of the Farmer analysis), the Madrid court
deferred again to prison administrators’ claims about the necessity of the SHU for
maintaining safety and security: “The severe restrictions on social interaction further
defendants’ legitimate interest in precluding opportunities for disruptive or gang

40. See Interview with Thelton Henderson, supra note 20.
41. Interview with Carl Larson, former Dir. of Fin., Cal. Dep’t of Corr., in Sacramento, Cal.
(Feb. 22, 2010) (on file with author).
42. Interview with Craig Brown, former Undersec’y, Youth & Adult Corr. Auth., Cal., in
Sacramento, Cal. (Jan. 22, 2010) (on file with author).
43. Madrid v. Gomez, 889 F. Supp. 1146, 1263 (N.D. Cal. 1995).
44. See Interview with Craig Brown, supra note 42.
45. See Madrid, 899 F. Supp. at 1263–64.
46. Id. at 1267.

100

UC IRVINE LAW REVIEW

[Vol. 5:89

related activity and assaults on other inmates or staff.”47 As the Madrid court
explained, an established penological interest in certain conditions of confinement
must be considered, along with the relative harm of that condition of confinement,
in a careful weighing of the purpose of the restriction against the potential harm of
the restriction.48 The Madrid court’s final conclusion about whether the restrictive
SHU conditions rose to the level of a constitutional violation (satisfying the first
prong of the Farmer analysis) was that prison officials had adequately justified these
conditions, given “the wide-ranging deference they are owed . . . [d]efendants are
thus entitled to design and operate the SHU consistent with the penal philosophy
of their choosing, absent constitutional violations.”49 The court added the phrase
“absent constitutional violations,” but in evaluating the SHU conditions, the court
found no violations precisely because the defendants had defended an adequate
“penal philosophy.”50 In sum, the court weighed the severity of the conditions
against the justifications for the conditions, and found that the justifications
adequately rationalized the severity.
Because the court found that the need for the SHU conditions justified their
seriously restrictive nature, the court did not reach the second prong of the Farmer
analysis—the subjective question of whether the prison officials operating the
SHUs knew the conditions might be generally harmful. Rather than incorporating
the principle of deference to prison officials into the second, subjective prong of
the Farmer test, which was explicitly designed to be deferential to what prison
officials knew or should have known, the Madrid court incorporated the principle
of deference to prison officials into its findings of fact, as well as into the objective
prong of the Farmer test.51 In this way, the court expanded the Farmer principle of
judicial deference.
The Madrid court essentially found that few, if any, conditions of confinement
(short of sedating prisoners against their will52) would actually be so severe as to
outweigh prison officials’ justifications for supermax conditions, as tools of safety
and security required to control the most dangerous prisoners.53 This expanded
prison officials’ discretion to define nearly any harsh condition of confinement as
necessary for safety and security and, therefore, virtually exempt from constitutional
review.
This also added an additional burden of proof for prisoner plaintiffs. In
addition to being required to establish a conditions-based harm, and to establish
that prison officials knew or should have known about the harm, post-Madrid
prisoner plaintiffs now also have to establish that the conditions-based harm is not

47.
48.
49.
50.
51.
52.
53.

Id. at 1263.
Id. at 1262.
Id. at 1262–63.
See id.
See id. at 1260–61.
Id. at 1263.
See id.

2015]

SUPERMAX ADMINISTRATION

101

justifiable as a tool of safety and security.54 In sum, by deferring to prison officials’
claims not just in the second, subjective prong of the Farmer analysis, but also in the
first, objective prong of the Farmer analysis, the Madrid court altered the power
allocation between courts, prisoners, and prison officials, thereby expanding prison
administrators’ discretionary control over supermax confinement.
2.

Assessing the Constitutionality of Supermax Double Bunking

In addition to deferring to prison officials’ claims about the need to house
some prisoners in the restrictive conditions of the Pelican Bay SHU, the Madrid
court also deferred to prison officials on another safety and security point: the
permissibility of double bunking some prisoners, two to a cell, in the same restrictive
supermax conditions. As with the prisoners’ claims challenging the harshness of the
conditions of supermax confinement, the Madrid court first outlined the facts about
double bunking in the SHU, and then applied the objective and subjective prongs
of the Farmer analysis to assess the prisoners’ claims about the unconstitutionality
of the conditions.55 Again, at each step of the analysis, the Madrid court weighed
descriptive facts about prison conditions and prisoners’ claims of constitutional
violations against prison officials’ justifications for the necessity of the conditions
at issue.56
First, the Madrid court described the practice of double bunking.57 The court
explicitly noted that as many as two-thirds of the Pelican Bay SHU prisoners had
been double bunked in the prison’s first few years of operation.58 The court
explained that all SHU prisoners were segregated from the general prison
population, but “[t]he degree of segregation and restrictions may vary . . . depending
on a variety of factors, including penal philosophy and the underlying reason for the
inmate’s segregation.”59 As in the factual analysis about the general conditions in
the SHU, in the factual analysis of the condition of double bunking, the Madrid court
addressed the “penal philosophy” justifying those conditions, weighing harsh
conditions against institutional necessity.60 Prison administrators’ justifications—
specifically the penal philosophies they promoted, and the reasons they provided
for segregation—were integral to the factual description of the condition being
evaluated.
Next, the Madrid court evaluated the seriousness of the injuries that had
resulted from double bunking. The court tallied the hundreds of cell fights that had
taken place between 1989 and 1993 at Pelican Bay and noted that “[m]any of these
cell fights have resulted in serious injuries to the victimized inmate,” including

54.
55.
56.
57.
58.
59.
60.

See id.
See id. at 1269.
See id. at 1238, 1269.
See id. at 1237–38.
Id. at 1227.
Id. at 1228.
Id. at 1237.

102

UC IRVINE LAW REVIEW

[Vol. 5:89

fractured ribs, paralysis, and brain damage.61 But, as with the question of whether
the restrictive conditions of supermax isolation were sufficiently serious to establish
a constitutional violation, the court weighed the serious injuries resulting from
double bunking against the claims of prison officials, who explained that cell fights
are unavoidable.62 The court noted: “[C]ell fights are an inevitable fact of prison life,
particularly in maximum security prisons and security housing units where inmates
are more likely to have violent histories or tendencies.”63 In alleging that the practice
of double bunking in the Pelican Bay SHU violated the Constitution, the prisoners
and their lawyers argued that cell fights could be prevented, with better housing
decisions and less double bunking.64 In assessing this claim, the court cited the
number and frequency of cell fights over three years, and the testimony of prison
officials about cell fights as “frequent” but “not as frequent as one might expect,”
and concluded that cell fights were “inevitable.”65 Rather than assessing the merits
of the plaintiffs’ claim about the potential to prevent cell fights, the court accepted
the prison administrators’ assertions that cell fights could not be prevented, which
was, in turn, based on a second assertion that prisoners indeed had violent histories
and tendencies. In the end, given the “inevitability” of fights between prisoners with
“violent histories or tendencies,” the court found that the objective prong of the
Farmer analysis—the seriousness of the alleged constitutional violation—was not
satisfied.66 Again, a penological claim made by prison administrators, in this case
about the inevitability of the harm prisoners claimed to experience, essentially
mitigated the court’s assessment of the seriousness of a condition of confinement
(double bunking).
In the double-bunking analysis, the court did not stop with the first, objective
prong of the Farmer test, but also considered the second, subjective prong. The
Madrid court found that there could be no Eighth Amendment liability for prison
administrators without proof that these officials actually knew they were housing
repetitively assaultive prisoners with bunkmates.67 If the prison officials did not
have actual knowledge of how dangerous some of the double-bunked prisoners
were, then the subjective, state-of-mind prong of the Farmer analysis was not
satisfied.
This finding, that the practice of double bunking in the Pelican Bay SHU failed
the subjective prong of the Farmer test, indirectly contradicts the court’s earlier
analysis that the isolation conditions in the Pelican Bay SHU failed the objective
prong of the Farmer test. In that analysis, the court deferentially accepted prison
officials’ claims that the most dangerous prisoners from throughout the California
department of corrections were housed in the SHU, without proof of how and why
61.
62.
63.
64.
65.
66.
67.

Id. at 1239.
Id. at 1269.
Id.
Id.
Id.
Id.
Id.

2015]

SUPERMAX ADMINISTRATION

103

each individual prisoner was dangerous.68 The court simply acknowledged that (1)
some prisoners were “unsuited for intermingling” with others, and (2) correctional
administrators had an “unenviable task” in managing these prisoners.69 At the same
time, the court declined to hold these same prison officials liable for Eighth
Amendment violations stemming from housing these same dangerous prisoners
together, two to a cell, because there was no proof that these administrators knew
exactly how dangerous these prisoners were.70 How could prison administrators have
inherent knowledge of dangerousness, without any knowledge of degree of
dangerousness? The court does not reconcile the inconsistency.
In analyzing the constitutionality of both the conditions of confinement in the
Pelican Bay SHU and the practice of double bunking some prisoners there, the
Madrid court demonstrated great deference to the claims of prison officials about
the necessity for the supermax institution. By incorporating deference to
correctional administrators into each step of analysis, the Madrid court expanded
correctional administrators’ discretion over supermaxes, allowing them to define
and influence the constitutional analysis applied to supermaxes, thereby shifting the
balance of power away from courts and prisoner plaintiffs and toward prison
officials.
Importantly, the analysis in the Madrid case is representative of (1) how courts
have analyzed prison conditions generally in the late twentieth and early twenty-first
century, and (2) how courts have analyzed the constitutionality of supermax
confinement specifically since 1995. I will address each of these trends in turn.
B. Deference in Prison Conditions Cases Generally
In the early 1990s, Judge Henderson, who oversaw the Madrid case for almost
twenty years, seemed especially predisposed to find that the basic conditions at the
Pelican Bay SHU did violate the Constitution. His first instinct was that the
conditions violated the Constitution, and he actually initiated the lawsuit against the
prison by encouraging lawyers to visit Pelican Bay and document potential
constitutional violations there.71 So why, in the end, was he so deferential to the
prison officials’ claims about the justifications for the institution?
The simplest answer is that the Madrid decision was part of a much larger trend
in federal court litigation across the United States: persistent judicial deference to
the claims and assertions of prison administrators in the context of prison
conditions challenges in and out of supermaxes.72 Prison law scholar Sharon
68. Id. at 1223.
69. Id. at 1259, 1262.
70. Id. at 1237–39.
71. See Interview with Thelton Henderson, supra note 20.
72. See Sharon Dolovich, Cruelty, Prison Conditions and the Eighth Amendment, 84 N.Y.U. L. REV.
881, 961 n.306 (2009) [hereinafter Dolovich, Eighth Amendment]; Sharon Dolovich, Forms of Deference in
Prison Law, 24 FED. SENT’G. REP. 245 (2012) [hereinafter Dolovich, Forms of Deference]; Richard H.
Fallon, Judicially Manageable Standards and Constitutional Meaning, 119 HARV. L. REV. 1274 (2006); Judith
Resnik, Detention, the War on Terror, and the Federal Courts, 110 COLUM. L. REV. 579 (2010); Giovanna

104

UC IRVINE LAW REVIEW

[Vol. 5:89

Dolovich has described a typology of three forms of deference in prisoners’ rights
cases: (1) “doctrine-constructing,” (2) procedural rule revising, and (3) “situationreframing.”73 Each form is present in the Madrid case.
First, the Madrid court applied Farmer, a judicially constructed doctrine that
incorporates deference to prison officials through the subjective prong of analysis,
which requires a prisoner plaintiff to prove not just that he or she was harmed, but
that the prison official intended the harm.74 On its face, the deferential Farmer test
is reasonable. It gives prison officials the benefit of the doubt, implicitly
acknowledging the challenges prison officials face in managing prisoners, and it
seeks to hold officials accountable only for things they can actually control.75
Moreover, the deference toward prison officials codified in the Farmer test maintains
a delicate balance of power between the legal experts in federal courts and the
security experts who run state prison systems.76 However, the Farmer test has been
applied in a manner and context that has exaggerated deference toward prison
officials. As I suggested in my analysis of Madrid above, and as Dolovich argues in
categorizing forms of deference, the doctrine-constructing deference of Farmer has
been expanded through procedural rule revising and situational reframing.
Second, the Madrid court “presume[d] the correctness” of the assertions of
prison officials, when these officials claimed prisoners were dangerous, restrictive
conditions were necessary, or cell fights were inevitable.77 This is what Dolovich
calls procedural rule revising, through which the trier of fact weighs evidence with
deference to one particular party over another (as opposed to weighing evidence
objectively).78 In Jones v. North Carolina Prisoners’ Labor Union, the Supreme Court
mandated exactly this kind of deference.79 In Jones, the Supreme Court upheld
limitations on speech and association imposed by the North Carolina prison system
on members of a prison-based labor union. The Court rejected the prisoners’
argument that the union had never actually “interfered with the prison’s
operations,” finding that the prisoners had failed to prove, with “substantial
Shay, Ad Law Incarcerated, 14 BERKELEY J. CRIM. L. 329, 339–44 (2010) (describing the
institutionalization of deference to prison officials in federal court cases through the 1980s and 1990s);
see also M ALCOLM M. FEELEY & EDWARD L. RUBIN, JUDICIAL POLICY MAKING AND THE MODERN
STATE: HOW THE COURTS REFORMED AMERICA’S PRISONS (Alfred Blumstein & David Farrington
eds., 1999) (discussing generally the history of prisoners’ rights litigation in the United States); Raja
Raghunath, A Promise the Nation Cannot Keep: What Prevents the Application of the Thirteenth Amendment in
Prison?, 18 WM. & MARY BILL RTS. J. 395 (2009) (describing patterns of judicial deference to prison
officials); Mikel-Meredith Weidman, Comment, The Culture of Judicial Deference and the Problem of Supermax
Prisons, 51 UCLA L. REV. 1505 (2004) (discussing how courts navigate deference principles with narrow
holdings, restricting the placement of mentally ill prisoners in supermaxes).
73. Dolovich, Forms of Deference, supra note 72, at 246.
74. Id.
75. Farmer v. Brennan, 511 U.S. 825, 837–38 (1994).
76. See Turner v. Safley, 482 U.S. 78, 84–85 (1987) (explaining judicial deference in the prison
context in an earlier prison conditions case).
77. Dolovich, Forms of Deference, supra note 72, at 247.
78. Id.
79. Id. at 246 (citing Jones v. N.C. Prisoners’ Labor Union, Inc., 433 U.S. 119, 121 (1977)).

2015]

SUPERMAX ADMINISTRATION

105

evidence” that the prison officials’ response was “exaggerated.”80 Unless prisoner
plaintiffs satisfied this high standard of proof, the Court presumed the
reasonableness of the prison officials’ policy.81
Third, situation reframing occurs in the way the Madrid court presented the
facts about supermax isolation conditions and supermax double-bunking
conditions, incorporating the claims of prison officials about why certain conditions
are necessary into the factual descriptions of the conditions. Dolovich describes this
as “recast[ing] relevant facts in ways that deny or disregard the lived experiences of
prisoners,” as the Madrid court did when it described the conditions in the SHU as
stark and sterile, but only potentially at the outer bounds of psychological
tolerability, in spite of evidence that many prisoners had actually found the
conditions psychologically intolerable.82
Other scholars have argued that judicial deference to institutional officials and
administrators is endemic not just to prison law but also to the law of detention
generally, including pretrial custody, immigrant, and terrorist detentions contexts.83
But the existence of widespread federal judicial deference to prison officials, as
evidenced in Madrid and discussed by scholars like Dolovich and Judith Resnik,
cannot alone explain the outcome in Madrid, upholding the basic constitutionality
of the Pelican Bay SHU. After all, the Madrid court found some aspects of the
policies and practices governing SHU confinement unconstitutional.84 The court
found that excessive use of force in the prison, as well as the practice of placing
mentally ill prisoners in long-term isolation, were blatantly unconstitutional and
required immediate remedy; as to these conditions, the Madrid court was not at all
deferential to prison officials.85
In other words, acknowledging the persistence of federal judicial deference to
prison officials is just the first step in understanding why the Madrid court deferred
to officials’ claims about some but not other conditions of confinement in the
Pelican Bay SHU. The next step is examining which factors facilitate deference in
the prisoners’ rights context. Resnik has argued that concerns with the limits of
federalism, concerns with the dangerousness of detainees, and considerations for

80. Id. (citing Jones, 433 U.S. at 127–28).
81. Id.
82. Id. at 248.
83. See Resnik, supra note 72, at 589. Judith Resnik argues that this deference is so pervasive
among federal courts and legislators that other mechanisms of civil rights enforcement must be
explored and implemented—perhaps through elevating the enforcement roles that might be played by
state court judges and nonjudicial decision makers (including “intelligence officials, police officers,
immigration authorities, lawyers, and decision makers operating at the lowest tiers of the system that
apprehend and confine such persons”). Id. at 685. Parts III and IV of this Article will argue that, in the
prison context, nonjudicial decision makers have had the significant “independent decisional authority”
Resnik advocates, but they have failed to treat disputants equally, and they have not been “protected
and monitored through public observation,” critical characteristics of nonjudicial decision makers,
according to Resnik. See infra Parts III, IV; see also Resnik, supra note 72, at 685.
84. Madrid v. Gomez, 889 F. Supp. 1146, 1280 (N.D. Cal. 1995).
85. Id. at 1279–80.

106

UC IRVINE LAW REVIEW

[Vol. 5:89

the expertise of detaining officials have all encouraged deference.86 On the other
hand, Dolovich has argued that sometimes there is a limit to the amount of
deference courts are willing to engage where “multiple forms of deference” have
already been applied in lower court rulings.87 The Madrid case suggests another
factor that affects judicial deference: whether the conditions at issue are immediately
and obviously life-threatening.88
Specifically, where the Madrid court identified egregious prisoner abuses, with
actual or potentially life-threatening consequences, the court also held that those
conditions and practices violated the Eighth Amendment.89 In other words, the
more obviously life-threatening the abuse, the less the court deferred to prison
officials. The Madrid court identified multiple instances in which excessive force was
used, such as the incident in which a prisoner’s skin burned off after officers forced
him to take a scalding bath, or the cuffing of prisoners into fetal positions for as
long as twenty-four hours, or the breaking of a prisoner’s jaw and a subsequent
refusal of medical treatment for seven days.90 In these instances, the Madrid court
applied the deferential Farmer test to analyze the fact patterns, but the court found
that prison officials had the requisite subjective intent to harm prisoners.91 Similarly,
in these same instances, the Madrid court refrained from applying any of the other
forms of deference Dolovich identified and which were discussed in the previous
subsections on supermax isolation conditions and supermax double-bunking
conditions. For example, the court did not engage in procedural rule revising or
presuming the correctness of prison officials’ assertions, nor did it reframe the fact
patterns to disregard the lived experiences of prisoners.92
This pattern of judicial avoidance of deference in the context of especially
abusive, physically life-threatening situations extends back to the earliest prisoners’
rights cases brought in the 1960s and 1970s.93 In these early cases, when abusive,
physically life-threatening situations were consistently present, courts found prison
conditions to be unconstitutional.94 For instance, in one of the earliest class action
prison conditions cases, Hutto v. Finney, the Arkansas district court, and later the
Supreme Court, noted that prisoners in Arkansas prisons were crowded into small,
dark isolation cells where infectious diseases were rampant and food provisions
inadequate; other prisoners received electrical shocks to sensitive areas of their
bodies as punishment.95 Likewise, in the rare federal cases subsequent to Madrid in

86. Resnik, supra note 72, at 652, 663.
87. Dolovich, Forms of Deference, supra note 72, at 257 n.127.
88. Madrid, 899 F. Supp. at 1257–58.
89. Id. at 1146–1283.
90. Id. at 1163, 1167–70.
91. Id. at 1248–52.
92. Id. at 1146–1283.
93. See generally FEELEY & RUBIN, supra note 72 (discussing early prisoners’ rights litigation in
the United States and the kinds of physical abuses documented in those cases).
94. Id.
95. Hutto v. Finney, 437 U.S. 678, 683 n.5 (1978).

2015]

SUPERMAX ADMINISTRATION

107

which courts have withheld deference to prison officials, abusive and physically lifethreatening situations have also been present.96 For instance, in Brown v. Plata, the
district court found “on average, an inmate in one of California’s prisons needlessly
dies every six to seven days due to constitutional deficiencies in the . . . medical
delivery system.”97 In this case, deference played a minimal role in the lower court’s
conclusion that the conditions were unconstitutional and violated the Eighth
Amendment, and the Supreme Court upheld the finding of unconstitutionality.98
Dolovich argues that perhaps Plata represents “an implicit upper limit to how much
deference to defendant prison officials the Court is willing to allow,”99 but I am
suggesting that this upper limit is only reached in the context of physically lifethreatening conditions, as existed in Hutto, Plata, and at least to a certain extent,
Madrid.
This analysis is not based on a comprehensive sample of every prison
conditions case that has been litigated in the United States between the 1960s and
2014. Indeed, such a study could provide an important test of the theory suggested
here. Rather, I am suggesting that the factors contributing to deference in prison
conditions cases need to be better understood, and that one possible explanation
turns on the severity (and physicality) of the conditions at issue. Because the
prisoners challenged such a wide array of conditions in Madrid, the case provides a
particularly useful starting point for analyzing when, why, and how courts defer to
prison administrators in assessing the constitutionality of prison conditions. But
understanding the legal context of prison conditions cases provides only preliminary
insights about the factors shaping judicial deference. As I will discuss in subsequent
sections, the historical and empirical contexts of Madrid provide further insights into
two additional mechanisms of judicial deference: how the Pelican Bay SHU was
initially designed, and how it has operated over the last twenty-five years (as
discussed in Parts II and III of this Article).
In sum, this subsection has argued that Madrid is representative of how courts
have analyzed prison conditions in the late twentieth and early twenty-first
centuries—generally concluding that prison conditions are constitutional, based on
deference to the claims of prison officials, unless there is evidence of egregious and
physically endangering abuse. The next subsection will argue that Madrid is also
representative of how courts have analyzed the constitutionality of supermax
confinement since 1995.

96. See, e.g., Brown v. Plata, 131 S. Ct. 1910, 1927 (2011).
97. Id. (internal quotation marks omitted).
98. See id. at 1928–29 (discussing the role that deference should play in the determination of
constitutionality of prison conditions).
99. Dolovich, Forms of Deference, supra note 72, at 252 n.127.

108

UC IRVINE LAW REVIEW

[Vol. 5:89

C. Deference in Supermax Cases After Madrid v. Gomez
Perhaps not surprisingly, given the extremity of the conditions in supermaxes,
prisoners and their legal advocates have challenged the institutions in most states in
which they have been built. Prisoners allege that administrative assignment to
supermaxes violates their constitutional right to due process under the Fourteenth
Amendment to the U.S. Constitution, and that continued detention in extreme
isolation violates their constitutional right to be free from cruel and unusual
punishment under the Eighth Amendment.100 Courts have considered these
challenges carefully, often ordering modifications to specific policies and
procedures governing supermax confinement, such as the procedures by which
prisoners may be assigned to supermaxes,101 including prohibiting the assignment
of mentally ill prisoners to supermaxes, and requiring additional training for
correctional officers working in supermaxes.102 However, no U.S. court has declared
the imposition of long-term solitary confinement in supermax institutions to be
unconstitutional as a violation of the Eighth Amendment prohibition against cruel
and unusual punishment.
Instead, courts have followed the pattern Madrid established. Two
representative federal court cases serve as examples of replications of the Madrid
analysis. First, I discuss the only case in Arizona in which a federal court specifically
addressed the conditions of confinement in that state’s supermax and explicitly
provided legal justifications for the extremity of the supermax conditions. Arizona
is an important case study because its supermax was the first modern supermax in
the United States, and the one on which Pelican Bay was modeled. 103 In Comer v.
100. Prisoners and their legal advocates have brought legal challenges questioning both the
ability of correctional administrators to accurately and fairly identify the most dangerous prisoners in
the prison system, and the overall methods by which correctional administrators assign prisoners to
supermaxes, alleging due process violations under the Fifth Amendment to the U.S. Constitution (as
incorporated through the Fourteenth Amendment). See, e.g., Wilkinson v. Austin, 545 U.S. 209, 225
(2005) (finding that inmates have a protected liberty interest in avoiding assignment to supermax prison
facilities); Lira v. Herrera, 427 F.3d 1164 (9th Cir. 2005), reconsidered by Lira v. Cate, No. C 00-0905 SI,
2009 U.S. Dist. LEXIS 91292, at *7–8 (N.D. Cal. Sept. 30, 2009) (upholding prisoner challenge to rules
governing placement in California’s SHUs); Charles F.A. Carbone, California Changes Policies for Prison
Gangs and Security Housing Units, PRISON LEGAL NEWS, Sept. 2004, at 33 (describing Settlement
Agreement, Castillo v. Alameida, No. C-94-2847-MJJ-JCS (N.D. Cal. Sept. 21, 2004)). Prisoners have
also challenged supermaxes as violations of their right to be free from cruel and unusual punishment.
Specifically, prisoners have alleged that their duration of confinement in supermaxes constitutes cruel
and unusual punishment, Silverstein v. Fed. Bureau of Prisons, 704 F. Supp. 2d 1077 (D. Colo. 2010),
that the conditions of total solitary confinement constitute cruel and unusual punishment, Madrid v.
Gomez, 889 F. Supp 1146 (N.D. Cal. 1995), and that the operation of the institutions constitutes cruel
and unusual punishment, id.
101. See, e.g., Lira, 2009 U.S. Dist. LEXIS 91292, at *7–8 (finding that Lira was improperly
“validated” as a gang member and unjustly spent eight years in the SHU at Pelican Bay).
102. See, e.g., Madrid, 889 F. Supp. at 1202 (finding that correctional officers mistreated prisoners
due to a lack of adequate medical training and supervision).
103. See Keramet Reiter, The Most Restrictive Alternative: The Origins, Functions, Control,
and Ethical Implications of the Supermax Prison, 1976–2010, at 55–81 (Spring 2012) (unpublished
Ph.D. dissertation, University of California, Berkeley) (on file with author).

2015]

SUPERMAX ADMINISTRATION

109

Stewart, an Arizona district court assessed the constitutionality of the conditions in
Arizona’s Security Management Unit (a supermax) in the context of evaluating a
death-sentenced prisoner’s competency to decide to relinquish his appeals.104
Second, I discuss the one Supreme Court case that has considered the
constitutionality of long-term solitary confinement: Wilkinson v. Austin.105 The
Eighth Amendment questions are secondary in Wilkinson, as in Comer. In the context
of evaluating the due process required before a prisoner can be assigned legally to a
supermax institution, the Wilkinson Court analyzed legal justifications for the
supermax conditions at issue in the case.106
1.

Comer v. Stewart

When a federal district court considered the constitutionality of long-term
solitary confinement in Arizona’s supermax, the Security Management Unit (SMU),
the court held that the interests of prison administrators in protecting staff from
prisoners and prisoners from each other outweighed any interest prisoners had in
less restrictive conditions of confinement.107 In Comer v. Stewart, Arizona prisoner
Robert Comer sought to waive his right to appeal his death sentence.108 But Comer’s
lawyers argued that the conditions in SMU I, the Arizona supermax where Comer
was being held, were so extremely harsh that the conditions had caused Comer to
withdraw his appeals and seek to facilitate his own execution.109 The Comer court
held that these claims about the coercive impacts of the SMU conditions were
unfounded and ultimately permitted Comer to waive his right to appeal his death
sentence.110
Throughout the Comer opinion, the Arizona district court detailed how
restrictive Comer’s conditions of confinement were in the SMU, and how these
restrictive conditions responded directly to how extremely dangerous Comer was.111
The court detailed the harsh reduction in privileges Comer experienced in the SMU,
including the modification of his cell “to remove the desk and stool and to reinforce
the bunk to make it more difficult for Mr. Comer to fashion weapons.”112 The court
further elaborated on the necessity of these conditions, explaining just how
dangerous Comer was, citing his history of disciplinary infractions, “including
making weapons and assaulting other inmates and staff,” and even referencing
media comparisons between Comer and the fictional horror movie character

104. See Comer v. Stewart, 230 F. Supp. 2d 1016 (D. Ariz. 2002).
105. See Wilkinson v. Austin, 545 U.S. 209 (2005).
106. Id.
107. See Comer, 230 F. Supp. 2d at 1021–22, 1034–35 (describing and accepting justifications for
restrictive conditions of confinement).
108. Id. at 1018.
109. Id. at 1026–27.
110. Id. at 1071–72.
111. See id. at 1021–22, 1034–35.
112. Id. at 1023, 1033–34, 1071.

110

UC IRVINE LAW REVIEW

[Vol. 5:89

Hannibal Lecter.113 As in the Madrid case, the Comer court deferred to the
correctional administrators’ assertions about the extreme dangerousness of certain
prisoners, and accepted this dangerousness as a justification for total isolation in
long-term solitary confinement, under conditions the Ninth Circuit had already held
to be constitutionally questionable.114
However, in contrast to the Madrid case, which involved a class of prisoners
and referenced categories of dangerous prisoners without specifically evaluating
many correctional administrators’ individual assessments of dangerousness, the
Comer case specifically assessed one individual prisoner’s dangerousness.115 And the
assessment was convincing; Comer was accused of multiple well-documented
assaults of staff and other prisoners.116 The restrictions in Comer’s housing
conditions were explicitly tied to the kinds of risks he posed as an individual.117
The Comer case is important not just for the individual assessment it provides,
though. First, the Comer case might be used to justify the restrictive conditions of
supermax confinement more generally for any Arizona prisoner who might
challenge the constitutionality of the restrictive conditions. Once the safety-andsecurity conditions have been upheld in one archetypal case, like that of Madrid or
Richard Comer, one archetypally dangerous individual is implicitly allowed to stand
for hundreds of other prisoners who have not received a similarly individualized
assessment of dangerousness. Second, the Comer case reinforced yet again the power
of the rhetorical justifications articulated by the original supermax designers to
explain the institutions they built. Indeed, the comparison the Comer court made
between Robert Comer and the popular trope of the dangerous Hannibal Lecter
echoed the descriptions of dangerous prisoners articulated by the correctional
administrators and architects who designed the first supermaxes in Arizona and
California in the 1980s. As one architect who worked closely with correctional
administrators on the Pelican Bay design project said, “The facility was designed to
house the worst of the worst, the Hannibal Lecters of the world. And in the state
of California, you find more Hannibal Lecters.”118 In the Comer case, as in the Madrid
case, correctional administrators and prison designers shaped the court’s
interpretations of the constitutionality of conditions with their virtually
unchallenged descriptions of the necessity of the supermax.

113. Id. at 1022, 1033–34.
114. Comer v. Stewart, 215 F.3d 910, 916 (9th Cir. 2000) (describing similar conditions to
Comer’s that had previously been held unconstitutional), remanded to 230 F. Supp. 2d 1016 (D. Ariz.
2002).
115. Compare Comer, 230 F. Supp. 2d at 1020, with Madrid v. Gomez, 889 F. Supp. 1146 (N.D.
Cal. 1995).
116. See Comer, 230 F. Supp. 2d at 1020, 1033–34 (referencing Comer’s assaults on staff and
other inmates).
117. Id. at 1034 (noting that Comer’s cell was modified “to make it more difficult for Mr. Comer
to fashion weapons”).
118. Telephone Interview with Anonymous, Pelican Bay Justice Architect (Aug. 4, 2010) (on
file with author).

2015]
2.

SUPERMAX ADMINISTRATION

111

Wilkinson v. Austin

While many federal district courts, in addition to those in Arizona and
California, have considered the constitutionality of individual state supermaxes, only
one case has reached the Supreme Court.119 In 2005, the U.S. Supreme Court
considered the constitutionality of a modern state supermax for the first (and only)
time in Wilkinson v. Austin.120 As the Madrid court had done ten years earlier, the
Wilkinson Court described in great detail the “severe” and “harsh” conditions in the
Ohio State Penitentiary (OSP), a supermax:
Incarceration at OSP is synonymous with extreme isolation. . . . OSP cells
have solid metal doors with metal strips along their sides and bottoms
which prevent conversation or communication with other inmates. All
meals are taken alone in the inmate’s cell instead of in a common eating
area. Opportunities for visitation are rare and in all events are conducted
through glass walls. It is fair to say OSP inmates are deprived of almost any
environmental or sensory stimuli and of almost all human contact.121
However, the Court also noted that the “severity of the conditions” in the
Ohio supermax was justified by the existence of prisoners so dangerous that they
required total isolation from all other prisoners and correctional staff.122 Specifically,
the Wilkinson Court stated: “OSP’s harsh conditions may well be necessary and
appropriate in light of the danger that high-risk inmates pose both to prison officials
and to other prisoners.”123 In other words, the Court accepted an explanation
underlying the necessity of the supermax much like the justifications articulated in
both Comer and Madrid: the Court agreed with prison officials’ assertions that the
institutions were necessary (a) to completely isolate high-risk prisoners from staff
and from each other124 and (b) to manage risk or to maintain institutional safety and
security.125
Ultimately, the Wilkinson Court reviewed the procedures by which a prisoner
could be assigned to OSP and upheld the requirements for minimal procedural due

119. See Wilkinson v. Austin, 545 U.S. 209 (2005).
120. Id.
121. Id. at 214.
122. Id.
123. Id. at 224. In Wilkinson, the Court focused on the procedural policies necessary to meet
minimum constitutional due process requirements in assigning individual prisoners to the Ohio State
Penitentiary (OSP). The discussion of the harsh conditions at OSP and their justification as necessary
safety and security measures constituted the extent of the (implicit) Eighth Amendment analysis in the
Wilkinson case. I have argued elsewhere that, in the Eighth Amendment context, legal analyses of
procedural rights have largely replaced legal analyses of whether a punishment is actually cruel and
unusual, in both the context of the death penalty and the context of the supermax. Keramet Ann Reiter,
The Most Restrictive Alternative: A Litigation History of Solitary Confinement in U.S. Prisons, 1960–2006, 57
STUD. L. POL. & SOC’Y 71, 117–18 (2012).
124. See Wilkinson, 545 U.S. at 229 (finding that “[p]rolonged confinement in Supermax may be
the State’s only option for the control of some inmates”).
125. See id. at 227 (establishing the role that “the brutal reality of prison gangs” plays in the
State’s interest).

112

UC IRVINE LAW REVIEW

[Vol. 5:89

process imposed by lower courts.126 The Wilkinson Court, however, did not evaluate
whether OSP conditions violated the Eighth Amendment prohibition against cruel
and unusual punishment. The question at the heart of the case was whether OSP
prisoners have a liberty interest in avoiding placement in supermaxes (according to
the Court, they do), not whether the OSP supermax conditions violate the Eighth
Amendment prohibition on cruel and unusual punishment.127 In fact, by the time
the Court heard arguments in Wilkinson, numerous lower federal courts, including
the Madrid court, had already decided that supermax conditions did not violate the
Eighth Amendment.128
In both Comer and Wilkinson, courts evaluated some aspect of the
constitutionality of long-term solitary confinement in modern supermax
institutions.129 And in each case, the courts deferred to prison administrators as
experts who needed discretion to manage difficult prison populations. When prison
administrators said that some prisoners were dangerous and would injure other
prisoners or correctional staff if they were allowed to live in a general prison
population, courts accepted these assertions. As discussed throughout this part, this
deference to prison officials’ claims follows a long tradition of deference in prison
conditions cases, especially absent evidence of egregious, physically endangering
abuse.
But the question remains as to exactly how this deference operates—do prison
officials facilitate deference, and do judicial decision makers resist deference? The
next two parts will address two critical mechanisms that have both facilitated and
exaggerated judicial deference to prison officials, especially in the California
supermax context. Part II documents how prison officials have actively avoided
judicial (and legislative and public) oversight, and Part III documents how courts
have avoided demanding and analyzing empirical evidence in the supermax context.
II. DEFERENCE BY OVERSIGHT AVOIDANCE
Part I discussed how deference to prison officials is pervasive in both general
prison conditions litigation and in supermax conditions litigation. However, the
supermax phenomenon reveals two key, overlooked mechanisms of this deference.
This part discusses the first mechanism: judicial deference is not simply initiated by
courts, it is also actively cultivated by supermax designers and administrators. This
part first analyzes historical evidence from archives and oral history interviews to
document how prison officials (rather than legislators or voters or judges) in
Arizona and California designed the first supermax institutions to avoid legislative,

126. Id. at 224–29.
127. See id. at 229 (“[C]laims alleging violation of the Eighth Amendment’s prohibition of cruel
and unusual punishments were resolved, or withdrawn, by settlement in an early phase of this case.”).
128. See, e.g., Madrid v. Gomez, 889 F. Supp. 1146, 1280 (N.D. Cal. 1995).
129. Wilkinson, 545 U.S. at 224–28 (evaluating the constitutionality of processes used to identify
and place inmates into supermax units); Comer v. Stewart, 230 F. Supp. 2d 1016, 1021–22, 1034–35
(D. Ariz. 2002).

2015]

SUPERMAX ADMINISTRATION

113

public, and judicial oversight. This part next analyzes supermax operational policies,
especially in California, to document how prison officials maintain broad discretion
in the day-to-day operation of the supermaxes they designed in the 1980s.
Continued judicial deference to supermax administrators, then, further exaggerates
the institutionalized lack of oversight and broad discretion prison officials have had
in designing and operating supermaxes.
A. Discretion in Design
Correctional administrators, in collaboration with architects, designed the first
supermaxes in Arizona and California.130 In both states, there was little public or
legislative oversight of the design process.131 Arizona opened its first supermax, the
SMU, in 1987.132 Lynch describes how “there was . . . no mention in any department
materials, government papers, or press accounts that this unit was anything more
than a maximum-security prison.”133 After the SMU opened, however, correctional
administrators and correctional architects alike quickly took notice of “its avantgarde nature,”134 flocking to Arizona to see the newly efficient and newly harsh
creation.
Similarly, in California, few people were aware of Pelican Bay State Prison’s
unusual design and harsh conditions of confinement until after it opened.135 Judge
Henderson recalled first learning of the 1056-bed SHU at Pelican Bay when he
started receiving letters from prisoners complaining about the harsh conditions
there.136 Steve Fama, one of the plaintiffs’ lawyers in the Madrid case, said he first
learned of the Pelican Bay SHU when Judge Henderson asked him to go visit the
prison to evaluate the constitutionality of the conditions there.137
California prison officials involved in the initial legislative approval process for
the Pelican Bay SHU explained that they had virtually total control over the siting,
building, and design of the institution.138 Craig Brown, who was the Undersecretary
of the Youth and Adult Correctional Authority (YACA), the executive agency that
oversaw the California Department of Corrections in the 1980s, described the state
legislature’s role in prison construction projects in that decade: “You’re not going
to find much in the record; it was all negotiated [off the record], and we [YACA]
pretty much had our way with the legislature.”139 California legislators, in fact, paid

130. See Reiter, supra note 103, at 84–105.
131. See Keramet Reiter, The Origins of and Need to Control Supermax Prisons, 5 CAL. J. POL. &
POL’Y, 146, 149–50 (2013).
132. LYNCH, supra note 2, at 138.
133. Id. at 136.
134. Id. at 137.
135. Reiter, supra note 123, at 108.
136. Id. at 166.
137. Interview with Steve Fama, Att’y & Co-Counsel in Madrid v. Gomez, Prison Law Office, in
Berkeley, Cal. (Oct. 13, 2010) (on file with author); Interview with Thelton Henderson, supra note 20.
138. Reiter, supra note 131, at 149–50.
139. Interview with Craig Brown, supra note 42.

114

UC IRVINE LAW REVIEW

[Vol. 5:89

minimal attention to what kind of prison was built in California’s Del Norte County,
the site chosen for Pelican Bay.140 Correctional administrators recalled telling
legislators that the “Spartan” but “not Draconian” facility was necessary and
receiving few queries in response.141 Ruth Wilson Gilmore also described how,
throughout the 1980s, the California legislature increasingly ceded authority over
prison siting, designing, and financing to the California Department of
Corrections.142
In Arizona, the legislature also ceded authority to the state department of
corrections. The Arizona legislature passed a law in 1985 exempting correctional
administrators from the usual administrative agency requirements for public
hearings and legislative reporting preceding any significant rule and policy changes
(under the Administrative Procedure Act).143 This effectively expanded correctional
administrators’ control over penal policies by exempting them from the kind of
legislative and public oversight to which most other state agencies were subject. 144
Given the discretion correctional administrators had in designing the Arizona SMU
and the Pelican Bay SHU, the initial public invisibility of these institutions, as
described by Judge Henderson and attorney Fama is, perhaps, not surprising.
But the invisibility of these institutions was not accident or serendipity. Rather,
correctional administrators designed supermaxes like California’s Pelican Bay SHU
explicitly to avoid attracting attention, especially judicial attention.145
Correspondence between California’s Undersecretary of Corrections, Rodney
Blonien, and state senator and chair of the Joint Legislative Committee on Prison
Construction and Operations, Robert Presley, reveals one motivation behind the
building of Pelican Bay: avoiding further litigation over conditions of confinement
in existing maximum security prison in the state.146 Specifically, in a May 1986 letter
to Presley, Blonien requested a $250 million legislative allocation to build a new
maximum security prison in California; he argued that building a new prison was
the only way to appease the judges in both the Toussaint and Wilson cases, which had
condemned the decrepit, deteriorating conditions at the state’s existing maximum
security prison, San Quentin, and had ordered the state to make improvements. 147
That same year, the legislature authorized a new high-security prison in Del Norte
County, which would become Pelican Bay.148
140. Reiter, supra note 131, at 149–50.
141. Second Interview with Carl Larson, former Dir. of Fin., Cal. Dep’t of Corr., in Sacramento,
Cal. (Apr. 14, 2010) (on file with author).
142. RUTH WILSON GILMORE, GOLDEN GULAG: PRISONS, SURPLUS, CRISIS, AND
OPPOSITION IN GLOBALIZING CALIFORNIA 93–94 (Earl Lewis et. al. eds., 2007).
143. LYNCH, supra note 2, at 140.
144. Id. at 140; see also Shay, supra note 72, at 344–61 (describing the systematic exemption of
state correctional systems from state administrative procedures acts).
145. Letter from Rodney J. Blonien, Undersec’y, Cal. Youth & Adult Corr. Agency, to Robert
Presley, Chairman, Joint Comm. on Prison Constr. & Operations (May 30, 1986) (on file with author).
146. Id.
147. See id.
148. Reiter, supra note 131, at 149–50.

2015]

SUPERMAX ADMINISTRATION

115

The Toussaint case explicitly challenged the conditions in long-term lock-down
units at San Quentin, where prisoners were in semipermanent segregation. 149
Attorney Steve Fama, who represented the plaintiffs in the Toussaint case (and who
would later represent plaintiffs in the Madrid case), explained that in the 1980s, the
California Department of Corrections figured out how to avoid litigation over
unconstitutional prison conditions by simply building new prisons.150 Specifically,
Fama said that he thought a Ninth Circuit opinion issued in the Toussaint case might
have helped to pave the way for the idea of the supermax at Pelican Bay. 151 Fama
said: “At a particular point there, the Department opened New Folsom [later
renamed California State Prison-Sacramento], and the Ninth Circuit held that the
[Toussaint] order did not apply, and this gave the Department the idea of a way out
of the consent decree.”152 Not only was the Pelican Bay SHU designed, in part, to
avoid litigation over long-term lockdown conditions in older, more decrepit
facilities, but it was designed to itself resist litigation, carefully built to meet
minimum standards for space, clean air, light, and daily exercise that had been
established in Toussaint and other similar cases.153
Arizona correctional administrators took a much more confrontational stance
against prison litigation than California administrators. The Arizona SMU’s
predecessor, CB-6, came under judicial scrutiny in the early 1980s. The Arizona
Department of Corrections Director, Sam Lewis, resisted the litigation and the
settlement, going so far as to ban ACLU lawyers from visiting their clients in CB6.154 Over the next decade, Arizona correctional officials and legislators collaborated
to institutionalize their resistance to prisoners’ rights litigation, drafting and
successfully advocating for the passage of a federal law limiting prisoner litigation
nationwide (the Prison Litigation Reform Act, passed in 1996).155 Arizona’s SMU
supermax opened in 1986, in the midst of this institutionalization of a general
correctional culture of resistance to judicial oversight of the prisons.156 Across the
United States, high-security and isolation-style prisons attracted judicial scrutiny in
the 1970s and 1980s.157 And, as in Arizona and California, many of these states built
new supermaxes designed specifically to comply with those minimum standards

149. Toussaint v. McCarthy, 597 F. Supp. 1388 (N.D. Cal. 1984), aff’d in relevant part, rev’d in part,
801 F.2d 1080 (9th Cir. 1986).
150. Interview with Steve Fama, supra note 137.
151. Id.
152. Id. (referring to Rowland v. U.S. Dist. Court for N. Dist. of Cal., 849 F.2d 380 (9th Cir.
1988)).
153. See Reiter, supra note 123, at 106–10 (discussing the relationship between prison conditions
litigation in the 1970s and 1980s and the physical structure of prisons that were built during the latetwentieth century prison-building boom across the United States).
154. LYNCH, supra note 2 at 180–82 (discussing Black v. Ricketts, a settled matter in which the
ACLU challenged the confinement conditions of Arizona’s high security prison units).
155. Id. at 190–91.
156. Id. at 135–38.
157. Reiter, supra note 123, at 104.

116

UC IRVINE LAW REVIEW

[Vol. 5:89

courts had described in earlier orders, consent decrees, and settlements designed to
remedy unconstitutional prison conditions.158
The initial invisibility—to state legislators, to the general public, and to
courts—of the first supermaxes in Arizona and California is one indication of the
administrative discretion inherent in the design of the facilities. The fact that many
supermaxes were built to the precise minimum standards of confinement described
in earlier court cases, with the explicit purpose of avoiding further prison conditions
litigation, provides evidence of the broad administrative discretion underlying the
design of supermax facilities.159
This administrative discretion in the design of the first supermax facilities, in
turn, has two important implications for understanding the mechanisms of judicial
deference to prison administrators. First, the earliest supermax designers in Arizona
and California actively cultivated the judicial deference they later received. Prison
officials built new prison facilities in order to avoid the persistent judicial scrutiny
that had plagued older facilities, and they built the facilities to the precise
specifications of preexisting minimum standards for constitutionally acceptable
conditions of confinement, designed to satisfy subsequent judicial scrutiny.160
Second, when supermaxes like the Pelican Bay SHU did face judicial scrutiny,
prisoner plaintiffs lacked any evidence with which to counter the claims of prison
officials about the necessity and usefulness of the SHU.161 There was no legislative
record about different possible intents or justifications for the Pelican Bay SHU, no
public debate over whether the institution was necessary, no already-collected
evidence about long-term rates of violence in the prisons and the short-term impact
of the SHU.162 Because of this absence of a public record about how and why the
SHU was designed, there was practically no evidence at the disposal of the Madrid
court or the prisoners’ lawyers to use to establish alternative justifications for the
Pelican Bay SHU, or to develop counterarguments to levy against prison officials’
claims of safety and security necessity. Instead, there was a court record, in cases
like Toussaint, establishing unconstitutional conditions of confinement in
overcrowded, deteriorating nineteenth-century California prisons.163 The hyperhygienic, well-lit Pelican Bay SHU cells, made of easy-to-clean poured concrete,
seemed, exactly as prison officials had intended, to largely remedy the concerns
earlier courts had with unconstitutional conditions in prisons like San Quentin.
Even when some Pelican Bay SHU prisoners were housed two to a cell,164 the
slightly crowded SHU conditions still seemed better than much more overcrowded
conditions in much older cells at prisons like San Quentin in the early 1980s.

158.
159.
160.
161.
162.
163.
164.

Id. at 104–06.
Id. at 106–10.
Id.
Id. at 110–12.
Id. at 94–97.
Id. at 167.
Id. at 170.

2015]

SUPERMAX ADMINISTRATION

117

The day-to-day operation of supermaxes provides yet another example of the
continued role of administrative discretion in shaping supermax policies and
procedures over the last twenty years, and the way this discretion, in turn, facilitates
judicial deference. Specifically, correctional administrators determine how, when,
and for how long prisoners are assigned to supermaxes, again with little oversight
from legislators, the public, or courts.165
B. Discretion in Operation
The administrative rules governing California’s supermax units, or SHUs,
provide myriad examples of the day-to-day operational discretion correctional
administrators have in overseeing supermaxes. First, correctional administrators,
not judges or juries, assign prisoners to supermaxes based on in-prison behavioral
assessments. Second, these administrators interpret the prison rules, deciding
whether a given prisoner’s transgression merits supermax confinement.166 For
instance, officers have many options in responding to prisoners who violate rules—
they can ignore the prisoner, or choose a range of charges to levy against the
prisoner from an informal prison rule violation to a formal felony charge.167 Third,
correctional administrators define and categorize evidence that indicates gang
membership.168 In California, three pieces of evidence—such as tattoos of symbols
associated with gangs, documentation of associations with other gang members
based on a prisoners’ correspondence or on observations of in-prison
conversations, or literature associated with gangs—are required for gang
“validation.”169 Gang validation, in turn, may, at the discretion of correctional
administrators, result in indefinite placement in a California SHU.170 Prisoners who
have been validated as gang members and assigned to indeterminate SHU terms say
the only way to complete an indeterminate SHU term is to “parole, snitch, or die.”171
“Parole” refers to the fact that prisoners might be released directly from the SHU
upon the expiration of their criminal sentences.172 Because assignment to the SHU
is an administrative process, it does not affect the prisoner’s maximum criminal
sentence.173 “Snitch” refers to the prisoner’s option to prove he is no longer active
in a gang by agreeing to identify gang members and describe gang activity to the
165. Id. at 167.
166. Reiter, supra note 2, at 541–43 (discussing discretionary supermax policies); Reiter, supra
note 131, at 152.
167. See, e.g., CAL. CODE REGS., tit. 15, § 3341.5(c)(9) (West, Westlaw through Reg. 2014, No.
44, Oct. 31, 2014); see also Reiter, supra note 2, at 542–43 (discussing discretionary supermax policies);
Reiter, supra note 131, at 152.
168. Reiter, supra note 2, at 542.
169. Id.
170. CAL. CODE REGS., tit. 15, §§ 3000, 3341.5, 3378(c)(4). For further discussion of
administrative discretion in the supermax context, see also Reiter, supra note 2, at 543.
171. Reiter, supra note 2, at 536.
172. Id. (stating that most prison sentences in California include mandatory three-year parole
terms, so “release” is synonymous with “parole”).
173. Id. at 531.

118

UC IRVINE LAW REVIEW

[Vol. 5:89

prison system’s internal gang investigation unit.174 Beyond the obvious, “die”
suggests the hopelessness and helplessness an indeterminate SHU term can inspire.
Across the United States, administrative decisions to discipline prisoners with
SHU terms, or to validate prisoners as gang members and assign them to the SHU,
include few procedural protections for prisoners. In Wilkinson, the Supreme Court
held that prisoners must be told why they are being assigned to a supermax and
must have some opportunity to rebut the evidence against them.175 But they are not
guaranteed a hearing, a lawyer, the right to call witnesses, or any other traditional
criminal procedural protections.176 And in California, as in Arizona, administrative
rules applying to only a single prison, like Pelican Bay State Prison, are not subject
to the basic procedural protections of the state’s Administrative Procedure Act,
which usually requires, for instance, public notice and comment periods for new
administrative rules.177
This administrative discretion over who gets sent to supermaxes and why
facilitates judicial deference in much the way administrative discretion over the
initial design of supermaxes facilitated judicial deference. Just as prison officials
designed the first supermaxes, they also designed the policies governing the
facilities.178 And just as with the administrative process underwriting the first
supermaxes, administrative discretion over day-to-day operational policies
contributes to a lack of evidence with which to counter prison officials’ claims.
The Madrid case provides a perfect example of the challenge a judge faces in
evaluating prison officials’ claims when those claims are asserted in the context of
policies over which prison officials exercise significant discretion. As discussed in
Part I, the Madrid court deferred to prison officials’ claims that the Pelican Bay SHU
housed “some of the most anti-social and violence-prone prisoners in the system,”
and this deference to prison officials’ claims is unsurprising in light of the pattern
of judicial deference in prison conditions.179 But a better understanding of the
discretion prison officials exercise over the determinations of antisociality and
violent predispositions suggests that the Madrid court’s judicial deference was not a
legal decision but a practical necessity. Because little evidence is collected through
the cursory disciplinary and gang validation processes underlying SHU
confinement, and even less of this evidence is made available to prisoners, the
prisoner plaintiffs in Madrid had nothing to work with in attempting to counter
prison officials’ claims about their inherent dangerousness.180 Importantly, where
the prisoner plaintiffs in Madrid were able to counter claims about inherent
dangerousness with evidence that they had been brutally subdued by prison officials

174.
175.
176.
177.
178.
179.
180.

Id. at 542.
Wilkinson v. Austin, 545 U.S. 209, 229 (2005).
Id. at 228.
Shay, supra note 72, at 377–78.
Reiter, supra note 123, at 96–97.
Madrid v. Gomez, 889 F. Supp 1146, 1160 (N.D. Cal. 1995).
Id.

2015]

SUPERMAX ADMINISTRATION

119

(e.g., skin scalded off or jaw broken and wired shut), they overcame the presumption
of judicial deference and proved that some of the conditions of their confinement
were, in fact, unconstitutional.181 Where there was no evidence of this kind of
egregious physical abuse—as in the placement of the vast majority of prisoners in
isolation conditions, or in double-bunked isolation conditions—the Madrid court
had minimal evidence with which to evaluate (or counter) the underlying claims of
prison administrators that the prisoners were dangerous.182
Over the last twenty-five years of Pelican Bay’s operation, prisoner plaintiffs
have sought to collect better evidence about the SHU assignment and gang
validation process in California, and have brought multiple individual lawsuits to
limit the discretion inherent in the process.183 Between 2011 and 2013, prisoners in
the Pelican Bay SHU organized three prison-wide hunger strikes to protest their
conditions of confinement, especially the gang validation and indeterminate SHU
term policies.184 In total, thirty thousand prisoners participated in the third hunger
strike, which lasted more than three weeks.185 Following the first hunger strike, in
2011, in response to media requests about durations of confinement in the Pelican
Bay SHU, the California Department of Corrections and Rehabilitation (CDCR)
released data indicating that there were more than 500 prisoners who had been in
isolation in the Pelican Bay SHU for more than ten years.186 Less than one year later,
the Center for Constitutional Rights filed a lawsuit, Ashker v. Brown, on behalf of
these prisoners, alleging that being in solitary confinement for more than ten years
violates the Eighth Amendment prohibition against cruel and unusual
punishment.187 In June of 2014, the federal district court in Northern California
certified a class of ten prisoners, each of whom had been in isolation for ten years
or more, in a case challenging these long durations of confinement.188
These recent events demonstrate the value of basic empirical data to prisoner
plaintiffs seeking to challenge their conditions of confinement and overcome the
presumption of judicial deference to the claims of prison officials. Whereas the
Madrid prisoner plaintiffs failed to win their argument that conditions in the Pelican
Bay SHU were unconstitutional, the certification of a new class of prisoners in the
Ashker suit has reopened the question, in light of the information that has been
gathered over the last twenty-five years about who has been assigned to the SHU,
181. Id.
182. Id.
183. See, e.g., Lira v. Cate, No. C-00-0905 SI, 2009 U.S. Dist. LEXIS 91292, at *48 (N.D. Cal.,
Sept. 30, 2009) (successfully challenging the evidence on which a gang validation was based); Carbone,
supra note 100.
184. See Reiter, supra note 26, at 579–81.
185. Id. at 581.
186. Small, supra note 27.
187. Ashker v. Brown: Synopsis, CENTER FOR CONST. RTS., http://ccrjustice.org/pelican-bay
(last visited Oct. 18, 2014).
188. Order Granting in Part Motion for Class Certification; Denying Motion to Intervene at 21,
Ashker v. Gov. of Cal., No. C-09-5796-CW (N.D. Cal. June 2, 2014), available at http://ccrjustice.org/
files/6.2.14%20Order%20Granting%20Class%20Cert.pdf.

120

UC IRVINE LAW REVIEW

[Vol. 5:89

on what basis, and for how long.189 Much of this information has been obtained
piecemeal, through individual litigation and journalist and researcher requests for
data.190 Each lawsuit and each request for more information constrains the
discretion prison officials have over the SHU assignment process—requiring the
officials to specify more precisely how the process works and to track more
accurately who is subject to SHU conditions. For instance, in October of 2013, the
CDCR released details of a revised policy for identifying gang members and placing
them in the SHU based on a narrower range of evidence than had previously been
authorized (“[u]nsubstantiated confidential information from a single source” is no
longer considered acceptable evidence of gang membership).191 CDCR also began
systematically reviewing all of the case files of validated gang members in the SHU;
as of October 2013, it had completed 528 reviews and found that 343 prisoners
were eligible for less restrictive conditions of confinement.192 This new evidence,
gathered and published by CDCR for the first time, calls into question whether
every prisoner in the SHU is as dangerous as previously alleged—and whether the
discretion and associated judicial deference accorded to prison officials is
justified.193
In this part, I have described how the same prison officials who defined the
SHU for the Madrid court also first designed and then operated the SHU with little
public or legislative oversight. Evidence of the litigation-avoidant intentions of the
Pelican Bay designers suggests that prison officials institutionalized resistance to
judicial oversight in the supermax design. And evidence about the discretionary
procedures by which prison officials assign prisoners to supermaxes suggests that
these officials facilitated ongoing deference by failing to collect or produce evidence
to defend their assertions.
In sum, existing research about the discretion inherent in the design and
operation of supermax prisons—from the initial invisibility of the institutions, to
the intentions of their designers to avoid legislative and judicial scrutiny, to the
vague rules governing the day-to-day operation of the institutions—provides critical
background to understanding the Madrid court’s finding that the basic conditions of
confinement at the Pelican Bay SHU were constitutional. Judicial deference in
prison conditions must be understood not just as a common legal presumption, but
in the context of the wide-ranging discretion prison officials have to design and
impose punishment, absent judicial review. Within this institutional context of
administrative discretion, federal court deference to prison officials exaggerates
existing administrative discretion. Courts cede the power of independent review to
189.
190.
191.

Compare id., with Madrid v. Gomez, 889 F. Supp. 1146, 1283 (N.D. Cal. 1995).
See, e.g., id. (using plaintiff depositions as evidence of current SHU practices).
CAL. DEP’T OF CORR. & REHAB., FACT SHEET: SECURITY THREAT GROUP
PREVENTION, I DENTIFICATION AND M ANAGEMENT STRATEGY (2013), available at http://
www.cdcr.ca.gov/stg/docs/Fact-Sheet-STG-reforms.pdf.
192. Id.
193. See id. at 2 (stating that a case-by-case examination is taking place and more than half of
the inmates reviewed have been released to general population).

2015]

SUPERMAX ADMINISTRATION

121

prison officials when they do not carefully evaluate those officials’ claims about why
certain restrictive conditions of confinement are necessary. And prisoners
themselves become increasingly vulnerable to abuses, as an unreasonable burden of
proof shifts onto their shoulders.
However, as discussed in this subsection, evaluating prison officials’ claims
about the necessity of supermax confinement is difficult in light of policies and
procedures that discourage collecting basic evidence about why and how prisoners
are assigned to supermaxes. But even where such evidence has not been collected
and published, as with supermaxes in California, the presumption that the evidence
does not exist, or could not be collected, is unwarranted. In fact, the recent increase
in evidence available about the Pelican Bay SHU following the hunger strikes and
the Ashker lawsuit suggests that the evidence is there, it just needs to be sought out
and evaluated. In the next section, I discuss exactly what kind of evidence is
available regarding the need for supermax confinement and how this evidence might
be assessed. While this section has focused on discretion in design and operation as
one mechanism underlying and facilitating judicial deference, the next section
focuses on the presumption that no empirical evidence exists with which to evaluate
prison administrators’ claims as a second mechanism of judicial deference.
III. DEFERENCE BY EMPIRICAL ASSUMPTION
This part examines the kinds of data the CDCR has collected about its
supermax institutions, and evaluates whether evidence supports the legal claims
correctional administrators make about the safety and security purpose and
violence-prevention outcomes of the supermax. This part highlights the absence of
data, in correctional records and in response to information queries, about the
operation and effectiveness of supermax prisons, especially in California. Just as
supermaxes have continued to be arenas of broad correctional discretion, in spite
of many legal challenges to their constitutionality, so have they also continued to be
largely invisible. Indeed, invisibility and discretion go hand-in-hand in ways that this
part will explore. And, as discussed in the prior part, invisibility and discretion are
critical mechanisms that facilitate and underwrite the presumption of judicial
deference in prison conditions cases.
This part is divided into three sections. The first presents and analyzes data
about double bunking in the California supermaxes between 1989 and 2010, and
the second two sections present and analyze qualitative data about incidents and
quantitative data about rates of violence in the California supermaxes. Both the
double bunking and violence data contradict the articulated purposes of the
supermax as institutions that maintain safety and security and reduce high levels of
institution and department-wide violence by controlling extremely dangerous
prisoners. These contradictions, in turn, reveal the perverse incentives—to build
additional, harsher, and less visible supermax cells—inherent in the multifaceted
discretion of supermax incarceration. In the context of the widespread judicial
deference to prison officials discussed in Part I, this evidence is all the more

122

UC IRVINE LAW REVIEW

[Vol. 5:89

important. Judicial deference to prison officials designing and operating facilities at
the outer boundaries of constitutionally acceptable conditions potentially blurs
those very boundaries. Where courts presume that evidence with which to evaluate
prison officials’ claims is not available, they not only exaggerate existing official
discretion, they also facilitate the continued invisibility and excessive harshness of
supermax incarceration.
A. Double Bunking
The Madrid court described the central purpose of the Pelican Bay SHU as
containment of “the worst of the worst” prisoners—the gang affiliates and the
“Hannibal Lecters”—in the state system.194 The court also noted that California
regulations “provide that housing ‘shall be in single cells (when possible) in security
housing.’”195 However, as noted by the Madrid court, some prisoners, in spite of
their dangerousness and the single-cell function of the SHU, were double bunked
in the SHU in the first years it was open.196 This section follows up on this factoid
from the Madrid case, examining the practice of double bunking in California SHUs
over time, and suggesting how the ongoing practice of double bunking might be
interpreted. This section argues that the practice of double bunking, which seems
to contradict the underlying supermax theory of safety and security achieved
through total isolation, provides critical insights into how supermaxes have
developed and operated. Examining the practice of double bunking reveals the
underlying incentives, rooted in the context of mass incarceration, driving supermax
growth and sustaining the practice of long-term isolation, even in the face of myriad
legal challenges and growing public condemnation. In the context of mass
incarceration, the practice of double bunking in the Pelican Bay SHU appears to be
not a practical or theoretical contradiction, but a natural corollary of the same
incentives that produced supermaxes in the first place.
Table 1 and Figure 1 in Appendix A examine double-bunking practices over
the last twenty years in California’s two main supermaxes: the Pelican Bay SHU and
the Corcoran SHU, both opened in the late 1980s.197 The Pelican Bay SHU has a
“design capacity” for 1056 single-occupancy cells, and the Corcoran SHU has a
design capacity of 1024 single-occupancy cells.198 However, both SHUs have
consistently housed significantly more prisoners than their “design capacity” would
indicate. On February 17, 2010, 1118 prisoners were housed in the Pelican Bay SHU
(eleven percent of prisoners had a cellmate), and 1439 prisoners were housed in the
Corcoran SHU (fifty-eight percent of prisoners had a cellmate).199 Table 1 and
194. Madrid v. Gomez, 899 F. Supp. 1146, 1155 (N.D. Cal. 1995).
195. Id. at 1237.
196. Id.
197. Reiter, supra note 2, at 530.
198. Id. at 524.
199. E-mail from James S. Derby, Assoc. Dir., Div. of Planning, Acquisition & Design, Cal.
Dep’t of Corr. & Rehab., to author (July 22, 2011, 5:12 PM PDT) (on file with author). I calculated the
percentage of prisoners who are double bunked by subtracting the design capacity of the Corcoran

2015]

SUPERMAX ADMINISTRATION

123

Figure 1, in fact, demonstrate that double bunking happens often in California’s
SHUs. In any given year, at least one-quarter and often as many as two-thirds of the
prisoners in the Pelican Bay and Corcoran SHUs have been double bunked.200 Rates
of double bunking, in fact, have varied drastically over time. In some years, like 1993
and 1994, as many as two-thirds of all supermax prisoners were double bunked.201
In other years, like 2000 and 2001, only one-quarter of all supermax prisoners were
double bunked.202
While rates of double bunking have varied greatly, the proportion of supermax
prisoners relative to the overall prison population has remained relatively constant,
at between 1.6% and 2.3% of the overall prison population.203 See the final column
of Table 1, in Appendix A, for these numbers. In fact, the raw number of supermax
prisoners has increased steadily, with the increases in the raw numbers of the overall
prison population.204 This explains how the rate of supermax use has remained
relatively constant.
In addition to double bunking supermax prisoners to keep up with increasing
populations, the CDCR has also opened additional supermax units, or SHUs.205 In
1995, the CDCR opened a second SHU at Corcoran.206 And in 2000, the CDCR
opened an overflow SHU at the Central California Institute at Tehachapi.207 The
fact that double-bunking rates in California SHUs have fluctuated in the California
prison system, while the overall proportion of prisoners confined in SHUs has
remained relatively constant, suggests that prison population increases generally,
and overcrowding in particular, are important factors affecting rates of supermax
double bunking.
In other words, patterns of supermax housing and double bunking are closely
intertwined with patterns of mass incarceration in the United States. Indeed,
supermaxes have been built as part of expansions in state prison populations and
state prisons, and they have become overcrowded right along with state prisons.
Both Arizona and California built their supermaxes as part of statewide investments
in prisons; California built twenty-three prisons between 1984 and the early 2000s,
and Arizona spent hundreds of millions of dollars on prison infrastructure over the

SHU (1024) from the total population (1439). This number (415) represents the “extra” prisoners being
housed in the SHU; every “extra” prisoner is necessarily housed in a cell with one other prisoner, so
the number of double bunked prisoners is twice the number of “extra” prisoners (830). The percentage
of prisoners double celled is therefore 830 divided by the total SHU population (1439): 0.576, or fiftyeight percent.
200. Note that these two figures first appeared in Reiter, supra note 2, at 544, 546.
201. Id. at 546.
202. Id.
203. Id.
204. Id. at 245.
205. Id.
206. Id.
207. Reiter, supra note 2, at 543–45; E-mail from Carl Larson, former Dir. of Fin., Cal. Dep’t
of Corr. & Rehab., to author ( July 20, 2011) (on file with author).

124

UC IRVINE LAW REVIEW

[Vol. 5:89

same years.208 Carl Larson, who was instrumental in designing and building the SHU
at Pelican Bay, explained that “temporary overcrowding” of SHUs was a problem
before the first SHUs even opened at Pelican Bay and Corcoran:
[Although] the Corcoran SHU and all of the other new prisons designed
and constructed since 1982 were designed for one inmate per cell . . . based
on our inmate population projections at the time, which proved to be
accurate, to accommodate “temporary overcrowding” we added a second
bunk and a second locker to each cell and increased the utility
infrastructure.209
Other states copied the California model; between 1990 and the early 2000s,
nearly every state built a new, freestanding supermax facility or retrofitted an
existing prison unit to impose supermax conditions, and many states double-bunk
prisoners in these facilities. In addition to California, Colorado, Connecticut,
Massachusetts, New Jersey, North Carolina, and Pennsylvania double bunk some
supermax prisoners.210
Although the Madrid court acknowledged the “ever-increasing population
pressures” facing prison officials administering supermaxes, the court did not
acknowledge how these pressures created perverse incentives to expand supermax
use, make supermax confinement maximally uncomfortable, and maintain the
invisibility of the institutions. First, with an ever-increasing prison population,
prison officials needed more and better methods of control, including an everincreasing number of supermax cells.211 Second, confinement in these supermax
cells, as the Madrid court acknowledged in discussing the frequent cell fights in
crowded cells, prioritized institution-wide safety and staff safety over the safety of
individual prisoners. For instance, that court noted, and accepted as “inevitable,”
that there were at least 683 cell fights in the Pelican Bay SHU between its opening
in December of 1989 and the production of assault data pursuant to the Madrid
litigation in January of 1993.212 Prison officials operating supermaxes—especially in
a state where hundreds of extremely dangerous cell fights annually are deemed
constitutionally acceptable—have as much of an incentive to make supermax
conditions maximally uncomfortable as minimally safe. After all, the more
uncomfortable a supermax cell is, the more potential it has to function as a tool of
control, deterring bad behavior by individual prisoners who want to earn their way
out of the supermax, and deterring bad behavior by all other prisoners who fear
earning their way into the supermax. Or so the theory goes. If the supermax is
simply a tool for control, the more uncomfortable the better, and double bunking
some supermax prisoners operates as just another form of discomfort. (The

208. GILMORE, supra note 142, at 93–94; LYNCH, supra note 2.
209. E-mail from Carl Larson, supra note 207.
210. Various postings to stopsolitary@mail.lawhelp.org (Aug. 9, 2012) (on file with author)
(responses by lawyers and activists to Taylor Pendergrass’s query whether a state double cells).
211. See Madrid v. Gomez, 889 F. Supp. 1146, 1237 (1995).
212. Id. at 1238 & n.180.

2015]

SUPERMAX ADMINISTRATION

125

evidence about violence in supermaxes in the subsequent sections reinforces this
point about the incentives to create, rather than eliminate, discomforts.)
Finally, double bunking facilitates a third perverse incentive to keep
supermaxes invisible. As discussed in the second part, California’s supermax
designers explicitly built the institutions to avoid the attention of courts, and as the
Madrid case reiterated, the less judicial attention the better, at least from the
perspective of prison officials struggling to manage challenging prison
populations.213 Double bunking, then, is a means to facilitate the continued
invisibility of supermaxes; double bunking blurs the line between maximum security
prisons and supermaxes, creating a literal grey area. The practice of double bunking
within supermaxes makes the institutions appear to be a kind of imprisonment on
the same spectrum with other harsh conditions of confinement—a series of
extended thirty-day periods in isolation, for instance, or a different degree of
decreased access to privileges, like out-of-cell time, communal meals, access to
books, radios, and televisions. As long as supermaxes appear to be simply a different
degree of imprisonment, and one that is clean, well lit, and leaves few physical scars,
they are more difficult to define as a new or unprecedentedly hard form of
punishment. In other words, specific unconstitutional practices are harder to
identify, describe, and challenge legally when supermaxes appear to be just a new
(and variable) combination of existing tools of control.
B. Initial Indicators of Violence
Much as the goal of total isolation in the California supermaxes was
compromised from the first day the institutions opened, complete with the extra
bunk bed added at the last minute, so was the goal of total safety and security. The
myriad safety and security compromises at the Pelican Bay SHU, from dangerous
prisoners being double bunked together, to prison officials using excessive force to
abuse prisoners, were discussed in the preceding sections. Like double bunking,
violence initiated by prison officials is not unique to the Pelican Bay SHU; similar
incidences took place at the Corcoran SHU, California’s other main supermax, in
the early years of its operation. And many supermaxes across the United States
experienced similar incidents of excessive use of force and extreme violence
initiated by prison officials, especially in the first few months and years after a state
supermax opened, as documented in anecdotes, journalistic accounts, and
litigation.214 These incidents provide another indicator of the second perverse
incentive of supermaxes: impose the harshest possible conditions.
213. Id. at 1263.
214. See, e.g., United States v. LaVallee, 439 F.3d 670 (10th Cir. 2006); Jones-El v. Berge, 374
F.3d 541 (7th Cir. 2004) (describing guards beating up prisoners when Wisconsin’s supermax first
opened); United States v. Verbickas, 75 F. App’x 705 (10th Cir. 2003) (detailing gruesome abuses of
prisoners by correctional officers at the federal supermax, for which officers were sentenced to threeplus years in prison); Affidavit of Chase Riveland ¶15, Osterback v. Moore, No. 97-2806-CIV-HUCK
(S.D. Fla. 1997) (describing excessive use of pepper spray in Florida’s supermaxes); JAMIE FELLNER &
JOANNE M ARINER, HUMAN RIGHTS WATCH, COLD STORAGE: SUPER-M AXIMUM SECURITY

126

UC IRVINE LAW REVIEW

[Vol. 5:89

Local news reports and court records from the early 1990s document that
between 1989 and 1994, five prisoners in the Corcoran SHU were killed, allegedly
as a result of the actions of prison guards.215 Guards injured an additional forty
prisoners from the Corcoran SHU in these years.216 Most of these injuries, and all
five deaths, stemmed from “gladiator fights,” staged by prison officers.217 Officers
would choose two prisoners at a time, who were known to be from enemy gangs,
to release into small, outdoor exercise yards attached to the Corcoran SHU.218 The
rival gang members would fight each other and when the fights got heated, prison
officers shot at the fighting prisoners.219 On five different occasions between 1989
and 1994, prison officers shot and killed a prisoner during these gladiator fights.220
Prosecutors charged eight correctional officers with civil rights violations for staging
the gladiator fights and facilitating the five prisoner deaths.221 Although all eight
officers were ultimately acquitted, surveillance cameras documented the causes of
each of the five prisoner deaths—a shot fired by an officer during a fight between
rival gang members.222
Then, in 1998, twelve more Corcoran correctional officers were indicted in
two separate cases and charged with civil rights violations for setting up and inciting
fights and rapes in the Corcoran SHU.223 The charges were widely reported in local,
national, and even international papers.224 The correctional officers’ union
supported their indicted members, both financially, by paying the legal costs of the
members’ defense, and publicly, through a media campaign to educate the public
about Corcoran, described as the place “Where Hell Begins.”225 In its public
relations campaign, the union argued that guards at Corcoran “maintain a thin blue
line” against thousands of the “state’s most violent criminals.”226 The comments of

CONFINEMENT IN INDIANA 28, 50 (1997), available at http://www.hrw.org/reports/1997/usind/
(discussing how supervisors encouraged guards to use unnecessary force against prisoners when
Indiana’s supermax first opened, and discussing how guards kept supermax temperatures at forty
degrees Fahrenheit for extended periods); Andy Davis, State Settles Pepper-Spray Suits: Ex-inmate at Varner
Supermax Unit to get $4,000 for ‘05 Cases, ARK. DEMOCRAT-GAZETTE, Feb. 17, 2011, at 1B (describing
instances of abusive uses of pepper spray following the opening of the Arkansas supermax); Press
Release, American Civil Liberties Union, ACLU Sues CT Corrections Chief Over Abuse of Prisoners
Housed at Notorious Virginia “Supermax” (Feb. 7, 2001), available at www.clearinghouse.net/chDocs/
public/PC-CT-0001-0002.pdf (last visited Oct. 18, 2014) (describing the violent deaths of two
Connecticut prisoners housed under contract in Virginia’s supermax).
215. Matthew Heller, Prison of the Damned, INDEPENDENT, Jan. 28, 2001, at 10.
216. Id.
217. Id.
218. Id.
219. Id. at 13–14.
220. Id.
221. Id.
222. Id. at 13.
223. Id. at 15.
224. See, e.g., Robert B. Gunnison, 8 Prison Guards at Corcoran Indicted: Civil Rights Charges in Slaying
of Inmate, S.F. CHRONICLE, Feb. 27, 1998, at A1; Heller, supra note 215.
225. Heller, supra note 215, at 15.
226. Id.

2015]

SUPERMAX ADMINISTRATION

127

the union in defense of their members reiterated the second perverse incentive of
supermax incarceration—toward maximally harsh conditions. Not only do
maximally harsh conditions serve as a potential deterrent, they are also selfreinforcing. The harsher the conditions, the worse the prisoners experiencing those
conditions will be perceived to be. The prison guards who orchestrated the
Corcoran gladiator fights and were charged with civil rights violations based on
incontrovertible evidence turned the viciousness of the fights back against the
prisoners, alleging that anyone acting the way the prisoners had been induced to act
was inherently violent.227
C. In Search of a Relationship Between Supermaxes, Overcrowding, and Violence
Because correctional administrators have claimed—both in interviews
explaining the need for the first supermaxes, and in court cases challenging the
conditions in these prisons—that supermaxes are necessary for safety and security,
a natural corollary question to the one about whether supermaxes have been used
for total isolation is: have supermaxes had a positive impact on institutional safety
and security? And, has double bunking in supermaxes compromised any positive
impact supermaxes might have had? This section evaluates the available data
collected by the CDCR over the last twenty years about various rates of violence.
Because California correctional officials do not collect sufficient data to answer
these questions, this section will also outline what data would be needed in order to
conclusively answer these questions.
Based on the claims of correctional administrators about the safety and
security value of total isolation incarceration, two hypotheses about the structure
and value of supermax incarceration should be considered: (1) The use of supermax
incarceration decreases violence throughout a state prison system by isolating the
most dangerous, violence-prone prisoners. (2) Double bunking the most dangerous
prisoners in conditions designed for total isolation increases rates of violence. This
section will evaluate the available data about supermax housing practices and
incarceration based on twenty years of supermax incarceration in California. The
short story: the available data can neither confirm nor reject the hypotheses
suggested above. This section will conclude by suggesting what data might be
collected in an effort to evaluate these hypotheses.
First, why is there no reliable statistical data with which to conclusively
evaluate the two hypotheses proposed above? There are two problems with violence
data collected and published by the CDCR. The biggest problem is a collection
problem: the CDCR collects and releases data about violent incidents—including
homicides, suicides, suicide attempts, prisoner-on-staff assaults, and prisoner-onprisoner assaults—at the institutional rather than unit level.228 Therefore, violent
227. See Keramet Reiter, The Supermax Prison: A Blunt Means of Control, or a Subtle Form of
Violence?, 17 RADICAL PHIL. REV. 457, 461 (2014), for a further discussion of the ways in which
supermaxes encourage and produce violence.
228. See, e.g., COMPSTAT DAI STATISTICAL REPORT – 13 MONTH (Sep. 12, 2014), available at

128

UC IRVINE LAW REVIEW

[Vol. 5:89

incident data from Corcoran and Pelican Bay includes violent incidents that took
place throughout the institution, both within the SHUs (these are the supermax
units, which make up roughly one-quarter to one-third of the total population at
each), and within the general population units.229 Prisoners in the general population
units spend many more hours per day out of their cells (congregating with each
other and participating in group activities) than prisoners in the SHUs, who spend
only an hour or two out of their cells every few days in a solitary exercise yard. 230
The second problem with violent incident data is an interpretive problem. Either
(a) very small numbers of incidents are involved, as in the case of homicides and
suicides; or (b) incident reporting is subject to correctional administrators’
discretion about how to characterize and whether to report nonfatal incidents of
violence.
In spite of these two serious problems with the California violent incident
data, this subsection presents two different analyses of the data. First, the data is
charted descriptively, over time, in order to investigate whether there are any
obvious or consistent trends in the data. Second, a differences-in-differences
regression model is presented in order to explore whether there are any consistent
trends in the data. California’s two supermaxes have now been open for just over
twenty years.231 The fact that there are two institutions’ worth of statistics to
compare and two decades’ worth of multiple violence measures provides some hope
that some trends over time might be visible, such as long-term increases or
decreases in multiple measures of violence. If there were consistent trends, these
might compensate for the two main flaws in the data as collected. Unfortunately,
no such trends are visible. Each method of analysis will be discussed in turn below.
The results of the descriptive graphs appear in Appendix B (Figures 2 through
16); the graphs are labeled and grouped together based on the kind of violence
measure each describes. Five measures of violence are included: homicides, suicides,
suicide attempts, prisoner-on-prisoner assaults, and prisoner-on-staff assaults. For
each measure of violence, there are three graphs: (1) comparing the rate of a given
kind of violence at Corcoran, Pelican Bay, and throughout the California prison
system (labeled as DOC, for department of corrections, on the graphs); (2)
comparing the rates of double bunking in the SHU at Corcoran to the rates of
violence throughout that institution; (3) comparing the rates of double bunking in
the SHU at Pelican Bay to the rates of violence throughout that institution.
Correctional administrators’ claims about the safety and security value of
supermaxes would be supported by either (1) a consistent decrease in overall rates
of violence, over the last twenty years, in institutions with supermaxes, and
throughout the California prison system; or (2) evidence of a consistent relationship
http://www.cdcr.ca.gov/COMPSTAT/docs/DAI/2014_07/2014_07_DAI%20High%20Security.pdf.
Note that violent incidents are reported by institution, not by units, within the institution.
229. Id.
230. Heller, supra note 215.
231. LYNCH, supra note 2, at 2.

2015]

SUPERMAX ADMINISTRATION

129

between fluctuating rates of double bunking in the supermaxes and fluctuating rates
of violence throughout the department of corrections. Unfortunately, the
descriptive trends in the graphs in Appendix B fail to tell one obvious story about
either overall rates of violence in the California prisons, or about the relationship
between rates of violence and double bunking in California.
In looking at Figures 2, 5, 8, 11, and 14, which each compare a single measure
of violence (homicides, suicides, suicide attempts, prisoner-on-prisoner assaults,
and prisoner-on-correctional officer assaults) at Corcoran, Pelican Bay, and
throughout the department of corrections since 1989, when the first SHUs opened
at Corcoran and Pelican Bay, one fact does stand out.232 Rates of violence at
Corcoran and Pelican Bay, whatever the particular measure, are consistently higher
than the rates of violence throughout the prison system.233 But again, the violence
data do not reveal whether these violent incidents occurred in the supermax units
at these prisons, in the general population units, or in both. Moreover, there is no
way to tell based on these data whether the violence rates at Corcoran and Pelican
Bay would have been higher or lower if these prisons had not had supermax units.
The data allows for no conclusions about whether supermaxes have either
exacerbated or reduced violence in the California prison system.
Figures 3, 4, 6, 7, 9, 10, 12, 13, 15, and 16 each compare rates of double
bunking in the supermax units at Corcoran and Pelican Bay, respectively, to rates of
violence in each unit. Again, the story is one of absences. For some measures of
violence in some institutions, violence seems to increase as double-bunking rates
increase. For instance, Figures 14 and 26 show this kind of pattern. In Figure 4: Rate
of Violent Death Compared to Proportion of Prisoners Double-Bunked, Pelican Bay State Prison,
1989–2006, between 1994 and 1998, the peak years of double bunking in the Pelican
Bay SHU, homicide rates at the institution also peaked. Similarly, in Figure 16: Rate
of Prisoner-on-Prisoner Assaults, Compared to Proportion of Double-Bunked Prisoners, at
Pelican Bay, 1989–2006, in those same years of SHU double bunking at Pelican Bay,
prisoner-on-prisoner assaults also peaked. But for other measures of violence,
violence actually seems to decrease as double-bunking rates increase. See Figure 12:
Rate of Prisoner-on-Staff Assaults Compared to Proportion of Double-Bunked Prisoners,
Corcoran, 1989–2006 for an example of this; as rates of double bunking in the
Corcoran SHU increased, rates of prisoner-on-staff assaults actually decreased.
However, even the apparently contradictory trends seen in these graphs could
be capturing changes in violence rates due to violent incidents that took place
outside of the supermax units, elsewhere in the prison institutions. Modeling these
apparent relationships with a bivariate regression, for instance, is futile, because
there is no way to directly evaluate the relationship between double bunking in the
supermax units and rates of violence in the supermax units, since the only data
available here actually consists of institution-wide violence rates.
232. See infra Appendix B, Figure 2; Appendix B, Figure 5; Appendix B, Figure 8; Appendix B,
Figure 11; Appendix B, Figure 14; see also LYNCH, supra, note 2, at 2.
233. See infra Appendix B, Figure 2.

130

UC IRVINE LAW REVIEW

[Vol. 5:89

One possibly creative statistical model through which the relationship between
supermax housing practices and violent incidents might be teased out is a
differences-in-differences estimation. This model allows a comparison between
variables across panels of data, if the panels of data are drawn from the same
broader context. Differences-in-differences estimations, then, might be used to
compare employment outcomes in different firms operating within the same
broader state policy context.234 In the instant case, violence outcomes in different
prisons operating in the same broader policy context of one state corrections
department are compared using a differences-in-differences model. Between
Corcoran and Pelican Bay, a differences-in-differences model estimates the
relationship between the independent variable of supermax housing and the
dependent variable of violence outcomes. In theory, such a panel-based,
differences-in-differences model might isolate the effect of changes in (1) rates of
prisoners housed in supermaxes, and in (2) rates of double bunking among these
prisoners on institutional violence rates. Indeed, this model provides a creative way
to explore whether there is a clear relationship between supermax housing practices
and rates of violence in California. However, even with the two panels of data drawn
from Corcoran and Pelican Bay, there are a limited number of observations; for
some measures, as few as nine years of data are available, and no more than
seventeen years of data are available for any measure. (Unfortunately, the data are
reported annually, not monthly.) Moreover, for many measures, like homicides and
suicides, there are very few instances of violence in any given year.235 Given these
data limitations, the inconclusive results reported in Table 2 are unsurprising.
Specifically, the model presented in Table 2: SHUs, Double Bunking, and Violence
in California Prisons – Correlation Models & Results, in Appendix B, estimates the effect
of (1) the differences in the proportion of all prisoners housed in the SHU between
each institution (Model A), and (2) the differences in the double-bunking rates between
each institution (Model B) on the differences in violence rates between each
institution, over a decade. Model A tests the first hypothesis presented in this
section: that the use of supermax housing, isolating some prisoners, will produce
overall reductions in violence throughout institutions. Model B tests the second
hypothesis presented in this section: that double bunking dangerous supermax
prisoners will produce increases in violence in these institutions.
These tables show that the directions of the relationships between supermax
housing practices and violence rates are inconsistent; sometimes the coefficient in
the regression is positive and sometimes it is negative, suggesting that sometimes
violence rates increase with increased rates of supermax housing use or with

234. See generally Stephen G. Donald & Kevin Lang, Inference with Difference-in-Differences and Other
Panel Data, 89 REV. ECON. & STAT. 221 (2007) (discussing differences-in-the-differences modeling).
235. See infra Appendix B, Figure 9. Note that the regression results presented in Figure 9 are
based on the natural logs of the housing rates and violence rates. In some years, however, there were
no suicides or no homicides at a particular institution; in this case, rather than dropping the variable
(because there is no natural log of 0), a zero was used.

2015]

SUPERMAX ADMINISTRATION

131

increased rates of double bunking, but sometimes violence rates decrease with
changes in housing practices. Moreover, very few of these relationships are
significant: the standard errors are large, so the apparent relationships could well be
due to chance and may not predict future relationships with much accuracy.
Model A evaluates the hypothesis that increased use of supermax housing
might lead to decreases in institutional levels of violence. The results reported in
Table 2 suggest that this hypothesized relationship might be true for three of the
five measures of violence: homicides, attempted suicides, and prisoner-on-prisoner
assaults. For all three of these measures, the regression coefficient is negative; as
supermax housing rates increase, homicide rates decrease, attempted suicide rates
decrease, and prisoner-on-prisoner assault rates also decrease. The regression
coefficient remains negative even when a year variable is included in the regression
model (columns labeled (2)), to control for the effects of time (and, by extension,
other variables that might be closely correlated with time). However, none of these
results are significant; the standard errors are relatively large, so the possibility that
the apparent relationships are due to chance cannot be ruled out.
Moreover, the results for two other measures of violence—suicides and
prisoner-on-staff assaults—further complicate the findings. For each of these two
measures of violence, the apparent relationship reverses when the control variable
for the year is included in the model. So, suicides appear to be negatively related to
supermax housing for the basic differences-in-differences regression estimation (as
supermax housing use increases, suicides go down), but positively related when a
year variable is included in the model (as supermax housing use increases, suicides
increase).236 The exact reverse is true for prisoner-on-staff assaults. However, the
relationship between prisoner-on-staff assaults and supermax housing in the basic
differences-in-differences regression estimation, without a control variable for the
year included, is the only significant relationship in Model A. Thus, the positive
relationship between increased supermax housing and increased prisoner-on-staff
assaults is one of the more robust relationships in this model. This relationship,
however, contradicts the first hypothesis presented in this section, suggesting that
an increased use of supermax housing might actually exacerbate rates of violence in
the prison. However, the significant relationship disappears (and the apparent
relationship also reverses, becoming negative) when the control variable for year is
included in the model. The fact that controlling for more variables in the model
reverses the apparent relationships is another indication that the potential positive

236. In fact, of all the measures of violence, completed suicides might be the one that could be
expected to increase with increased uses of isolation; mental health professionals have argued that
isolation is a risk factor for suicide. Don Thompson, Record California Inmate Suicides Are Double National
Rate, FREE REPUBLIC (Jan. 2, 2006, 3:38 PM), http://www.freerepublic.com/focus/f-news/1550815/
posts (noting that seventy percent of the forty-four prisoners who committed suicide in California in
2005 were in solitary confinement); Fact Sheet: Psychological Effects of Solitary Confinement, SOLITARY
WATCH, http://solitarywatch.files.wordpress.com/2011/06/fact-sheet-psychological-effects-final
.pdf (last visited Feb. 3, 2012).

132

UC IRVINE LAW REVIEW

[Vol. 5:89

relationship between supermax housing and prisoner-on-staff violence is weak and
not likely predictive.
In sum, for three measures of violence, Model A suggests that increased rates
of supermax housing use are associated with decreased homicide rates, suicide
attempts, and prisoner-on-prisoner assaults, but the standard errors for these
associations are large, so the significance of these findings is limited. Moreover, for
two measures of violence, suicides and prisoner-on guard assaults, the relationships
between increases in rates of supermax housing are more ambiguous.
Model B evaluates the hypothesis that increased rates of double bunking in
supermaxes might lead to increased rates of institutional violence. In fact, this model
reveals that, for three measures of violence, increased rates of double bunking are
negatively correlated with rates of homicide, prisoner-on-staff assault, and prisoneron-prisoner assault (i.e., as double-bunking rates increase, those three violence rates
decrease). For the assault statistics, these relationships have low standard errors;
they are significant and unlikely to be due to chance. For suicides and attempted
suicides, which admittedly constitute a different kind of violence, increased rates of
double bunking are correlated with increased rates of self-inflicted violence. The
standard errors are relatively large; only the relationship between double bunking
and suicides, without controlling for year, produces a significant p-value.
In sum, for three measures of violence, Model B suggests a possible negative
relationship between double bunking and violence. As double-bunking rates
increase, violence rates decrease. This finding might suggest that either (a) the
prisoners in the supermaxes are not so dangerous as alleged because double bunking
them does not produce aggravations in rates of violence, or (b) the right prisoners
are being double bunked together, so possible violent altercations are being avoided
by safely bunking known friends together. Either way, there is an absence of
evidence to support the claim that total isolation of certain individually dangerous
prisoners is strongly associated with reductions in institutional levels of violence
across a variety of measures of violence.
The statistical data in California is conclusively inconclusive as to the
effectiveness of supermaxes in general or of double bunking in particular. As with
the descriptive graphs discussed above, the differences-in-differences estimation
fails to reveal a clear relationship between supermax housing practices and violence
rates at Corcoran and Pelican Bay. However, the statistics are useful for what they
do not show. They fail to either confirm or reject the claims of correctional
administrators about the relationship between the restrictive conditions of
supermax prisons and the safety and security of prison systems with supermaxes.
Correctional administrators, then, have produced little-to-no evidence of whether
the supermax institution has succeeded or failed at its stated safety and security
purposes, for more than twenty years now.
Indeed, very few analyses of the relationship between supermaxes and
violence, in any state, exist. The few studies that do exist suggest that perhaps

2015]

SUPERMAX ADMINISTRATION

133

supermaxes actually exacerbate violence between prisoners and staff.237 For
example, Peter Kratcoski found some evidence that higher security prisons often
have higher levels of violence, documenting that seventy-one percent of assaults on
staff occurred in a maximum-security unit that housed less than ten percent of the
facility’s prisoners.238 And Chad Briggs, Jody Sundt, and Thomas Castellano found
some evidence of a correlation between supermaxes and prisoner-on-staff
assaults.239
Correctional administrators have collected virtually no data with which to
examine the justification for supermaxes: violence control and reduction. These data
(or the absence thereof, as the case may be) reiterate the third perverse incentive of
the supermax—the incentive to maximum invisibility. Given the discretion
correctional administrators had both in designing and operating supermaxes and the
deference courts have shown to correctional administrators’ claims about the need
for and purposes of supermaxes, correctional administrators have no incentives to
produce more or better data about supermax operation. Indeed, they have
incentives to produce less data so that their claims can remain straightforward and
simple.
When courts defer to prison officials’ claims about prison conditions, the
deference implicitly assumes that data with which to evaluate these claims is not
available, is impossible to collect, or is too complicated to analyze. But even working
with extremely limited existing data, this part has documented many findings
relevant to understanding whether supermaxes are constitutional, including: the
relationship between prison populations and supermax double bunking, which
suggests that overcrowding, as much as security concerns, drives double-bunking
decisions; the qualitative evidence of violence in supermaxes; and the absence of
evidence that supermaxes curb violence. Deference to prison officials often
depends on an assumption of the absence of empirical data; this section has sought
to demonstrate that that assumption is unwarranted.
IV. LESS DEFERENCE, MORE VISIBILITY, BETTER INCENTIVES
The previous parts documented how courts have repeatedly deferred to
correctional administrators’ claims about the necessity for supermax prisons as tools
of institutional safety and security. Empirical evidence, however, demonstrates that
supermaxes were built and operated in direct contradiction to these correctional
administrators’ claims. For instance, although correctional administrators described
supermaxes as institutions of total isolation, designed to protect staff and prisoners
alike from “the worst of the worst,” or the “most dangerous” prisoners in the

237. See generally Chad S. Briggs et al., The Effect of Supermaximum Security Prisons on Aggregate Levels
of Institutional Violence, 41 CRIMINOLOGY 1341 (2003); Peter C. Kratcoski, The Implications of Research
Explaining Prison Violence and Disruption, 52 FED. PROBATION 27 (1988).
238. Kratcoski, supra note 237, at 28.
239. Briggs et al., supra note 237, at 1365.

134

UC IRVINE LAW REVIEW

[Vol. 5:89

system, supermaxes have regularly housed two prisoners to a cell.240 In fact,
correctional administrators failed to clearly articulate the many purposes a supermax
would serve, consistently operated the institutions in contradiction to the purposes
they have articulated, and failed to collect and analyze evidence that supports their
claims to expertise and to being skilled promoters of institutional safety and security.
These failures suggest the perverse incentives underlying supermax incarceration—
incentives to maximize the number of supermax cells in operation and to maximize
the harshness of the conditions there while also minimizing the visibility of these
institutions. The pattern of deference federal courts have shown to correctional
administrators in evaluating the constitutionality of supermaxes has only
exaggerated these perverse incentives. Judicial deference, along with administrative
discretion and the relative invisibility of supermaxes, has essentially left supermax
prisoners at the mercy of inconsistently disorganized and inexperienced (at least in
terms of analyzing and applying empirical evidence) administrators. This section will
consider two ways that prison visibility might be increased, while simultaneously
limiting judicial deference to prison administrators and two ways that the degree of
judicial deference deployed in Madrid and similar cases might be directly limited or
discouraged.
First, better data about our prisons, and especially about supermaxes, could be
collected; paying empirical attention to American imprisonment is something that
criminal law practitioners and scholars should (in fact, must) do in order to make
determinations about how the law does and should work. Better data collection can
be facilitated at the state and federal legislative level through legislation requiring
systematic collection of information about who is housed at what level of prison
security, for how long, and with what prevalence of violence, for instance. And
better data collection can be facilitated in courts if both lawyers and judges demand
and evaluate empirical evidence to back up the expert claims of prison officials
about things like the purpose and effectiveness of supermaxes.
The example of California’s supermaxes is instructive. Evidence of the
practice of double bunking in California supermaxes potentially contradicts the
stated rationale of supermaxes as necessary for safety and security. Specifically,
California’s use of supermax double bunking suggests that perhaps California
supermaxes have been used not just as a tool of safety and security, but as a tool for
managing overcrowding. When the Madrid court deferred to correctional
administrators’ penological justifications for both (1) the restrictive conditions of
supermax confinement, and (2) the differently restrictive conditions of double
bunking in supermax confinement, the court did not consider that the actual
justification for the conditions could have been overcrowding pressures rather than
safety-and-security concerns.241 If rates of overall prison overcrowding are, in fact,
determining the degree of restrictive conditions of supermax prisons, then this

240.
241.

See infra Appendix A, Table 1; Figure 1.
See Madrid v. Gomez, 889 F. Supp. 1146, 1282 (N.D. Cal. 1995).

2015]

SUPERMAX ADMINISTRATION

135

might present a constitutional problem. After all, the Supreme Court recently held
in Plata that overcrowded prison conditions were no excuse for the perpetuation of
Eighth Amendment constitutional violations and therefore had to be remedied,
even if the remedy would require the release of thousands of prisoners.242
The statistics about double-bunking rates over time in the California
supermaxes suggest that there is indeed overcrowding in California’s supermaxes.243
The decision in Plata, in turn, suggests that if the conditions of supermax
confinement are being determined not by safety concerns but by institutional
overcrowding concerns, the supermax may deserve renewed constitutional
scrutiny.244 By analogy to Plata, if overcrowded prison conditions are forcing prison
officials to either (a) assign prisoners who otherwise would be housed in less
restrictive conditions to the “stark sterility and unremitting monotony” of supermax
conditions of confinement,245 or (b) assign prisoners who otherwise would be
housed in total isolation for the safety of other prisoners or staff to have cellmates,
then there may be a constitutional problem.246 The Madrid court held that imposing
these conditions to isolate dangerous prisoners in order to promote institution-wide
safety and security was justified; the court did not hold that imposing these
conditions to relieve problems with overcrowding was justified.247
The California case puts in stark relief the enormous difference between the
theories of criminal law as taught in the academy and applied by judges, and the
creation and administration of imprisonment as a criminal punishment. In fact, this
Article documents how the creation and administration of imprisonment happens
outside of, or in the shadow of, the law. Empirical evidence—not only statistical,
but also historical—is absolutely vital both to better understanding this process of
creating and administering imprisonment and to accurately applying legal
frameworks to analyses of punishment practices. Advocates should present
evidence of supermax practices, and courts should demand it before assuming that
deference to the expertise of correctional administrators is warranted. And the
evidence should be weighed against correctional administrators’ claims about the
purposes of and necessity for certain prison practices, like supermaxes.
Another way that judicial deference to prison administrators might be limited
is through the application of categorical restrictions, excluding certain prisoners
from supermaximum confinement. Some courts, including the Madrid court, have
242. See Brown v. Plata, 131 S. Ct. 1910, 1947 (2011).
243. See infra Appendix A, Table 1; Figure 1.
244. See Plata, 131 S. Ct. at 1910.
245. Madrid, 889 F. Supp. at 1229.
246. Cf. id. at 1947. I do not mean to equate violations of standards of decency with the viability
of the safety-and-security justification for supermaxes. Under Plata, a finding that supermaxes do not
promote safety and security would be insufficient to conclude the institutions were unconstitutional. A
court would also need to find that supermaxes violated standards of decency. See id. I am simply
suggesting here that the lack of a clear purpose behind the restrictive conditions of confinement in
supermaxes might be the first step to considering whether the institutions do, in fact, violate standards
of decency.
247. See Madrid, 889 F. Supp. at 1272.

136

UC IRVINE LAW REVIEW

[Vol. 5:89

categorically excluded prisoners with preexisting mental illnesses from supermax
confinement.248 Prisoner plaintiffs could raise and courts could consider expanding
these kinds of categorical restrictions on supermax confinement. For instance, these
restrictions might be extended to exclude juveniles from supermax confinement,
much as the Supreme Court has recently determined that the death penalty and even
certain sentences of life without the possibility of parole exclude juveniles.249 Certain
offenses might also be categorically excluded from supermax confinement. Just as
the court held in Kennedy v. Louisiana that the death penalty cannot be applied to
nonhomicide offenders, and in Graham v. Florida that the sentence of life without
parole cannot be applied to nonhomicide juvenile offenders,250 federal courts
addressing challenges to supermax confinement might exclude certain kinds of inprison offenses from eligibility for long-term or indeterminate supermax
confinement. For instance, in California, more than five hundred prisoners have
spent more than ten years in supermax confinement serving indeterminate solitary
terms because of their status as gang members instead of for specific actions they
took while in prison.251 Courts could limit supermax confinement to those prisoners
who have committed a specific action that endangered prisoner safety.
In response to a spate of highly publicized critiques of supermax prisons
between 2011 and 2014,252 a number of state legislatures have proposed laws
imposing categorical restrictions on placements in supermax confinement, including
restrictions on placing juveniles, pregnant women, and the mentally ill in
supermaxes.253 These kinds of categorical restrictions have been criticized, though,
as distractions from the restrictive conditions of supermax confinement, which
some argue should be unconstitutional for anyone, vulnerable or not.254
Judicial expansions of categorical restrictions excluding certain prisoners from
supermax confinement might require a different interpretation of the purposes
248. See id. at 1267; see also, e.g., Settlement Agreement at 6, Disability Law Ctr., Inc. v. Mass.
Dep’t of Corr., 960 F. Supp. 2d 271 (D. Mass. Apr. 12, 2012) (No. 07-10463-MLW).
249. See Miller v. Alabama, 132 S. Ct. 2455 (2012) (excluding all juvenile offenders from the
mandatory punishment of life without the possibility of parole); Graham v. Florida, 130 S. Ct. 2011 (2010)
(categorically excluding all juvenile, nonhomicide offenders from the punishment of life without the
possibility of parole); Roper v. Simmons, 543 U.S. 551 (2005) (categorically excluding all juveniles from
the punishment of death); see also IAN KYSEL, GROWING UP LOCKED DOWN: YOUTH IN SOLITARY
CONFINEMENT IN JAILS AND PRISONS ACROSS THE UNITED STATES (Alison Parker et al. eds., 2012)
(documenting juvenile placement in solitary and supermax confinement and arguing that it violates
international human rights standards and should be forbidden).
250. Graham, 130 S. Ct. 2011; Kennedy v. Louisiana, 554 U.S. 407 (2008) (categorically
excluding child rapists from receiving the death penalty).
251. Small, supra note 27, at 6.
252. See, e.g., A MNESTY INT’L, supra note 15; George F. Will, The Torture of Solitude, WASH. POST,
Feb. 21, 2013, at A17; UN NEWS CENTRE, supra note 15.
253. See Ivan Moreno, Lawmakers Are Trying to Restrict Solitary Confinement, HUFFINGTON POST
(Apr. 22, 2014, 5:45 PM EDT), http://www.huffingtonpost.com/2014/04/22/solitary-confinement
_n_5194072.html (discussing efforts in a number of states to impose categorical restrictions on uses of
solitary confinement).
254. See, e.g., Weidman, supra note 72 (arguing that categorical restrictions might “inadvertently
narrow the scope of first amendment protections”).

2015]

SUPERMAX ADMINISTRATION

137

behind supermax confinement. Whereas the Madrid court treated supermax
confinement as a condition of confinement and so evaluated its constitutionality
using the Farmer standard, courts might consider treating supermax confinement as
more akin to a harsh prison sentence, like the death penalty or life without parole,
which would trigger specific protections for certain categories of people. 255 At a
procedural level, supermaxes seem more like a condition of confinement than like
a sentence to prison; correctional administrators, not judges, assign prisoners to
supermax confinement.256 But at a practical level, when prisoners spend years at a
time in segregated conditions harsh enough to trigger a liberty interest in avoiding
them,257 questions arise as to whether the supermax experience is practically, if not
also purposefully, punitive.
Third, the degree of deference deployed in Madrid might be limited by more
individualized considerations of specific prisoners’ placements in supermax
confinement. In Wilkinson v. Austin, the one Supreme Court case that has considered
the constitutionality of any aspect of solitary confinement, the Court mandated
certain minimum procedural protections, like written notice and regular review of
supermax placement, for prisoners facing supermax confinement.258 The limited
procedural protections prisoners have before being placed in a supermax are
somewhat less than the limited procedural protections undocumented immigrants
have prior to being deported, or the limited procedural protections a citizen has
prior to being preventively detained. But these protections are also somewhat
greater than the limited procedural protections a person has prior to being required
to register as a sex offender or prior to being placed on a no-fly list.259 Jennifer
Daskal, for instance, outlines a series of increased procedural protections that
should be considered for persons in the latter category. These protections, including
increased procedural requirements, and especially a robust mechanism that allows
“targets [to] rebut a presumption of continuing dangerousness,”260 could well be
applied to increase the protections available to supermax prisoners prior to
placement in supermaxes.
Because court-initiated protections are likely to face public and political
scrutiny,261 additional avenues for achieving increased visibility and decreased

255. See Madrid v. Gomez, 889 F. Supp. 1146, 1245 (N.D. Cal. 1995).
256. A MNESTY INT’L, supra note 15, at 2.
257. See Wilkinson v. Austin, 545 U.S. 209, 220 (2005).
258. Id. at 224–30.
259. See Jennifer C. Daskal, Pre-Crime Restraints: The Explosion of Targeted, Non-Custodial
Prevention, 99 CORNELL L. REV. 327 (2014).
260. Id. at 379.
261. For instance, following the Supreme Court’s decision in Plata, a three-judge panel in
California ordered the release of thousands of prisoners from the state’s overcrowded prison system.
Plata/Coleman v. Brown, Case Nos. C01-1351 TEH (N.D. Cal. 2010) and 2:90-cv-0520 LKK JFM P
(E.D. Cal. 2010), “Three-Judge Plata/Coleman v. Brown Court Order to Reduce Prison Population,”
available at http://www.cand.uscourts.gov/tehplatacase; see also Brown v. Plata, 131 S. Ct. 1910 (2011).
Governor Brown has repeatedly sought to close the case, arguing that government oversight is no

138

UC IRVINE LAW REVIEW

[Vol. 5:89

discretion vis-à-vis prisons, and especially supermaxes, should be considered. One
additional nonjudicial avenue toward these goals is indirectly facilitating prison and
supermax visibility by making the institutions more accessible to journalists, the
media, and oversight agencies. Most prisons limit journalist access to guided tours
of facilities and interviews with prisoners selected by prison officials.262 And there
are few independent or government agencies charged with monitoring prisons at
either the state or the federal level.263 Even American law students studying criminal
law rarely either visit prisons or study prison law.264 Encouraging visibility through
legislation that guarantees journalists broader access to prisons, or that facilitates
independent citizen or nonprofit monitoring of prisons, is another means to
increase the visibility and limit the often unchecked discretion in prison, and
especially supermax operation. Law schools can also facilitate prison oversight by
teaching prison law, as Dolovich has argued,265 and by facilitating legal scholar tours
and monitoring of prisons.
CONCLUSION
Supermaxes represent the outer boundaries of our criminal justice system, the
extreme end of permissible punishment. As prisons within prisons, they are literally
hidden from sight. As arenas of near total administrative discretion, they are
practically hidden from the law. At a microlevel, multiple levels of administrative
discretion have reinforced the day-to-day ungovernability of supermaxes, fortifying
the perverse incentives for expanded discretion at the genesis of the facilities.
Judicial deference to prison officials’ claims, in turn, has further expanded and
exaggerated the administrative discretion at the heart of supermax facilities. This
Article has sought to explore the mechanisms of that deference by documenting the
ways supermax designers avoided oversight in the design and operation of prison
facilities, and identifying the testable empirical assumptions underlying supermax
incarceration.
At a macrolevel, supermaxes resulted from the perverse incentives of mass
incarceration: a rapidly expanding prison system requires flexible and flexibly harsh
tools of control. Looking closely at the motivations behind supermaxes, their
ungovernability and their potential for reform, therefore, has implications for mass
incarceration more broadly. If we could grapple with limiting the supermax, perhaps
longer necessary, even though the three-judge panel has found that unconstitutional conditions
continue to exist in California’s prisons.
262. James Ridgeway, Fortresses of Solitude: Even More Rare: Journalist Access to Prison Isolation Units,
COLUMBIA JOURNALISM REV., Mar.–Apr. 2013, at 31, 32.
263. The Special Litigation Section of the Civil Rights Division of the U.S. Department of
Justice investigates prisons upon allegations of abuse or constitutional violations, but they do not
regularly visit prisons absent such allegations. In New York and Illinois, there are nonprofit
organizations (the Correctional Association and the John Howard Association), which have public
mandates to monitor conditions in state prisons, but most other states have no comparable
organizations with independent monitoring mandates.
264. Sharon Dolovich, Teaching Prison Law, 62 J. LEGAL EDUC. 218, 218, (2012).
265. Id. at 222–24.

2015]

SUPERMAX ADMINISTRATION

139

this would give us clues as to how we might grapple with limiting other aspects of
the supersized American criminal justice system.

140

UC IRVINE LAW REVIEW

[Vol. 5:89

APPENDIX A
Table 1: Rates of Double Bunking and of SHU Use, 1989–2010
Year

1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010

Corcoran
SHU
Percent
Population
Double
Bunked

0
720
764
759
785
786
1318*
1266
1369
1212
1134
1115
1221
1213
1231
1223
1220
1319
1292
1358
1382
1439

0%
58%
66%
65%
70%
70%
45%
38%
50%
31%
19%
16%
32%
31%
34%
33%
32%
45%
41%
49%
52%
58%

Pelican Bay
SHU
Percent
Population
Double
Bunked

0
1238
1176
1520
1642
1504
1470
1570
1573
1394
1251
1172
1148
1162
1215
1113
1101
1089
1100
1098
1117
1118

0%
29%
20%
61%
71%
60%
56%
65%
66%
48%
31%
20%
16%
18%
26%
10%
8%
6%
8%
8%
11%
11%

DOC
SHU
Population
as a
Percent of
Total
Prison
Population
0.0%
2.2%
2.1%
2.3%
2.2%
2.0%
2.3%
2.1%
2.1%
1.8%
1.6%
1.7%
2.0%
1.9%
1.9%
1.9%
1.9%
1.9%
1.8%
1.9%
2.0%
2.0%

* In May of 1995, the California Department of Corrections opened a second Security Housing Unit at
Corcoran State Prison. This housing unit had been planned as a SHU since the prison was built, but
was not operated as one until 1995. E-mail from Carl Larson, supra note 207. So, prior to 1995, the
design capacity of the Corcoran SHU used for calculating overcrowding in this chart was 512 singleoccupancy cells. In 1995 and thereafter, the design capacity of the Corcoran SHU used for calculating
overcrowding in this chart was 1024 single-occupancy cells. The calculation for the percentage of
double-bunked SHU prisoners is as follows: (1) Subtract the SHU Design Capacity from the SHU
population to determine how many prisoners are housed in the SHU in excess of the design capacity.
(2) Multiply the difference between the design capacity and the population by two, because every
prisoner in excess of the design capacity is, by definition, double bunked with a second prisoner. (3)
Divide the total number of double-bunked prisoners by the total population to obtain the percentage
double bunked.

2015]

SUPERMAX ADMINISTRATION
Figure 1: Percentage of Double-Bunked Prisoners, 1990–2010

141

150

UC IRVINE LAW REVIEW

[Vol. 5:89

Table 2: SHUs, Double Bunking, and Violence in
California Prisons—Correlation Models & Results266
(Observations)
A. Overall Effect of SHU
Population on Violence
Coefficient
(Standard Error)
P-Value
R-Squared
B. Overall Effect of Double
Bunking on Violence
Coefficient
(Standard Error)
P-Value
R-Squared

(Observations)
A. Overall Effect of SHU
Population on Violence
Coefficient
(Standard Error)
P-Value
R-Squared
B. Overall Effect of
Double Bunking on
Violence
Coefficient
(Standard Error)
P-Value
R-Squared

Homicide
(17)
(1)
-0.755
(0.532)
0.176
0.119

(2)
-1.145
(1.073)
0.304
0.13

-0.134
(0.316)
0.678
0.012

-0.391
(0.357)
0.293
0.133

Suicide
(9)

Attempted Suicide
(9)

(1)
-4.773
(2.154)
0.062
0.412

(2)
3.9
(3.181)
0.266
0.771

(1)
-4.773
(2.154)
0.062
0.412

(2)
3.9
(3.181)
0.266
0.771

0.882
(0.229)
0.006
0.68

0.443
(0.343)
0.244
0.776

0.882
(0.229)
0.006
0.68

0.443
(0.343)
0.244
0.776

266. Standard Errors in Parentheses; grey shading indicates significant p-values; Model (1) is a
simple regression relating the natural log of the independent variable (SHU population or double
bunking) to the natural log of the dependent variable (violence rate); Model (2) includes the year in the
regression to control for the possible effects of time. Models (A) and (B) both use differences-indifferences estimations, comparing the panel of Corcoran data to the panel of Pelican Bay data, and
evaluating the relationships, if any, between the differences in these two panels, i.e., does a comparative
change in double-bunking rates between two institutions relate to an overall change in violence rates
between two institutions?

2015]

SUPERMAX ADMINISTRATION

(Observations)
A. Overall Effect of
SHU Population on
Violence
Coefficient
(Standard Error)
P-Value
R-Squared
B. Overall Effect of
Double Bunking on
Violence
Coefficient
(Standard Error)
P-Value
R-Squared

Prisoner-on-Staff
Assaults
(17)

151

Prisoner-on-Prisoner
Assaults
(17)

(1)
0.484
(0.189)
0.022
0.305

(2)
-0.109
(0.336)
0.749
0.466

(1)
0.484
(0.189)
0.022
0.305

(2)
-0.109
(0.336)
0.749
0.466

-0.318
(0.097)
0.005
0.417

-0.199
(0.099)
0.065
0.581

-0.318
(0.097)
0.005
0.417

-0.199
(0.099)
0.065
0.581

152

UC IRVINE LAW REVIEW

[Vol. 5:89