PUBLIC LAW AND LEGAL THEORY RESEARCH PAPER SERIES
PAPER NO. 590

FEBRUARY 2018

WRONG TURN ON THE EX POST FACTO CLAUSE
PAUL D. REINGOLD & KIMBERLY THOMAS
CALIFORNIA LAW REVIEW, VOL. 106, 2018

THE SOCIAL SCIENCE RESEARCH NETWORK ELECTRONIC PAPER COLLECTION:
HTTP://SSRN.COM/ABSTRACT=3126467

Electronic copy available at: https://ssrn.com/abstract=3126467

Wrong Turn on the Ex Post Facto Clause

Paul D. Reingold* and Kimberly Thomas**

ABSTRACT

*

University of Michigan Law School, Clinical Professor of Law. For this article we took

inspiration for the title (as well as style cues) from David Shapiro’s pungent “comment” Wrong Turns:
The Eleventh Amendment and the Pennhurst Case, 98 HARV. L. REV. 61 (1984).
**

University of Michigan Law School, Clinical Professor of Law. We thank Michigan Law

School student Amanda Blau for excellent research assistance, and the Law School for its research
support of clinical faculty.

Electronic copy available at: https://ssrn.com/abstract=3126467

The Ex Post Facto Clause bars any increase in punishment after
the commission of a crime. But deciding what constitutes an increase
in punishment can be tricky. At the front end of a criminal case,
where new or amended criminal laws might lengthen prisoners’
sentences if applied retroactively, courts have routinely struck down
such changes under the Ex Post Facto Clause. At the back end,
however, where new or amended parole laws or policies might
lengthen prisoners’ sentences in exactly the same way if applied
retroactively, courts have used a different standard and upheld the
changes under the Ex Post Facto Clause. Because the harm is identical and lies at the core of what the Ex Post Facto Clause is
supposed to protect against, we think the asymmetry is mistaken.
Parole is an integral part of punishment: it determines how
much time people will serve on their sentences. Until the 21st
century, black-letter law forbade even modest parole changes that
were adverse to prisoners. If a change in the parole regime might
lead to longer sentences, then courts insisted that the change be
applied prospectively only. Over the last two decades, relying on
language in two US Supreme Court parole cases decided in 1995 and
2000, the lower courts have shifted parole–ex post facto doctrine by
180 degrees. Prisoners can no longer prevail even when the change
in the state parole regime is almost certain to lead to significantly
longer sentences.

Electronic copy available at: https://ssrn.com/abstract=3126467

In the context of parole, the courts have repudiated past doctrine and strayed far from the purposes of the Ex Post Facto Clause.
In this article, we review the history, show how the current case law
is misguided and illogical, and put forward a new framework that
would restore the Ex Post Facto Clause to its rightful place.

Abstract ........................................................................................................... 101
Introduction .................................................................................................... 103
I. The Ex Post Facto Clause and Parole: Early Protection Against Changes
that Might Increase Punishment .......................................................... 110
II. Late 20th Century Supreme Court Doctrine .............................................. 117
A. A Modest Change of Course ........................................................ 117
B. An Opening for Opponents of Parole Release ............................. 121
C. Garner v. Jones: A Wrong Turn Initiated .................................... 126
III. Post-Garner: Wrong Turn Completed ...................................................... 130
A. Garner’s Reading by the Courts of Appeal .................................. 130
B. Coda on the Ex Post Facto Clause at Sentencing......................... 137
IV. Identifying the Wrong Turn and Getting Back on the Right Track .......... 141
A. Distinguishing Two Categories of Ex Post Facto Claims ............ 141
B. The Red Herring of Discretionary Decision-Making .................. 146
C. A Sustained Look at Category 2 “Possible” Ex Post Facto
Cases ............................................................................................ 151
1. Burdens of Proof and Persuasion ........................................... 153
2. How Individualized Must an Ex Post Facto Showing Be? .... 159
D. Group 2 “Possible” Cases: Putting It Together............................ 160
E. A New (Old) Approach to Ex Post Facto Doctrine ..................... 163
Conclusion ...................................................................................................... 165

INTRODUCTION
The Ex Post Facto Clause says, “No State shall . . . pass any . . . ex post
facto Law . . . .”1 Although the Latin phrase “ex post facto” literally encompasses any law passed “after the fact,” by 1800 the US Supreme Court had
recognized, in Calder v. Bull, that the constitutional prohibition on ex post

1.

U.S. Const., art. I, § 10, cl. 1.

facto laws applies only to penal statutes.2 In Calder, Justice Chase described
the reach of the Ex Post Facto Clause as follows:
1st. Every law that makes an action done before the passing of the law,
and which was innocent when done, criminal; and punishes such
action. 2d. Every law that aggravates a crime, or makes it greater than
it was, when committed. 3d. Every law that changes the punishment,
and inflicts a greater punishment, than the law annexed to the crime,
when committed. 4th. Every law that alters the legal rules of evidence,
and receives less, or different, testimony, than the law required at the
time of the commission of the offence, in order to convict the
offender.3
Justice Chase’s four categories in Calder were originally viewed as exclusive:
if a change of law did not fit within those four categories, then it was not
covered by the Ex Post Facto Clause.4 For much of the nineteenth century, the

2.

3 U.S. 386, 390–392 (1798); but see Evan C. Zoldan, The Civil Ex Post Facto Clause,

2015 WISC. L. REV. 727 (arguing that the historical doctrine is misplaced and that the clause should
apply to civil as well as criminal laws); see also Thomas Jefferson, Letter to Isaac McPherson, August
13, 1813 (“The sentiment that ex post facto laws are against natural right is so strong in the United
States, that few, if any, of the State constitutions have failed to proscribe them. The federal constitution
indeed interdicts them in criminal cases only; but they are equally unjust in civil as in criminal cases
and the omission of a caution which would have been right, does not justify the doing what is
wrong.”).
3.

Calder, 3 U.S. at 390.

4.

Justice Chase himself may have taken a broader view, noting that “All these, and similar

laws, are manifestly unjust and oppressive.” Calder, 3 Dall. at 391. See also Zoldan, supra note 3, at
743-749 (citing historical material in support of the broader view).

Ex Post Facto Clause played a fairly narrow role: it was primarily invoked to
prevent new punishments from being imposed retroactively for past criminal
conduct.5
In the late 1800s, however, the precise contours of the Ex Post Facto
Clause became less clear as the US Supreme Court struggled to apply the
Calder categories consistently. In that epoch the Court expanded the reach of
the Ex Post Facto Clause to bar not just substantive changes to criminal laws,
but also some arguably procedural changes that affected significant rights or
seriously disadvantaged criminal defendants.
Two cases exemplify the Court’s more expansive interpretation. First, in
1883, in Kring v. Missouri,6 the Court held that “any law passed after the
commission of an offence which . . . ‘in relation to that offence, or its conse-

5.

See e.g., Fletcher v. Peck, 10 U.S. (6 Cranch) 87, 138-39 (1810) (invalidating retroactive

law that forfeited title and permitted state seizure of estates for past criminal acts); Cummins v.
Missouri, 71 U.S. 277 (1866) (invalidating state constitutional provision that barred people from
holding public office or practicing their professions absent taking an oath stating that they had not
supported the rebellion); Wayne Logan, Democratic Despotism and Constitutional Constraint: an
Empirical Analysis of Ex Post Facto Claims in State Courts, 12 WM. & MARY BILL OF RTS. J. 439
(2004).
6.

107 U.S. 221 (1883). Kring involved a plea to second-degree murder that was overturned

on appeal, resulting in a conviction for first-degree murder (and a death sentence) on remand. The law
in effect when the defendant committed his crime and pled guilty treated his plea as an acquittal of the
higher charge. But a new state constitution, applied retroactively, abrogated that law. The state court
held that the “change is a change, not in crimes, but in criminal procedure, and such changes are not ex
post facto.“ Id. at 224 (citation omitted). The U.S. Supreme Court reversed, holding (5-4) that the
amendment could not be applied retroactively. The label “crime” or “criminal procedure” was of no
moment: what mattered was the change in circumstances to the defendant’s detriment. Id. at 228–229.

quences, alters the situation of a party to his disadvantage,’ is an ex post facto
law.”7 Second, in 1898, in Thompson v. Utah,8 the Court held that retroactive
procedural statutes can violate the Ex Post Facto Clause unless they “leave
untouched all the substantial protections with which existing law surrounds the
person accused of crime.”9 In Thompson the Court struck down a Utah law that
retroactively reduced the size of criminal juries from twelve to eight persons,
because the change deprived the defendant of “a substantial right involved in
his liberty.”10

7.

Id. at 235 (citation omitted). On the other hand, the very next year the Court held that

permitting a felon to testify against an accused was not an ex post facto violation even though felons
were forbidden from testifying in criminal cases when the defendant committed his crime; the change
was viewed as merely procedural. See Hopt v. Utah, 110 U.S. 574 (1884).
8.

170 U.S. 343 (1898).

9.

Id. at 352.

10.

Id. at 351-52 (holding that the change violates the Ex Post Facto Clause because it

“materially impairs the right of the accused”). Later in the same term, in Thompson v. Missouri, 171
U.S. 380 (1898), the Court distinguished Kring, saying that the right at issue in Kring was “a
substantial one—indeed, it constituted a complete defense against the charge of murder in the first
degree—that could not be taken from the accused by subsequent legislation,” and therefore was “not
simply a change in procedure.” Id. at 383-84. In Thompson, handwriting samples were admitted
against the accused, resulting in his conviction; on appeal the court held that the admission of the
samples was error and reversed on that basis. In the meantime, the state amended its laws to allow the
admission of handwriting samples, which were then used at the trial on remand to convict the
defendant again. The Court found no ex post facto violation because the evidentiary change did not
“affect the substantial rights of one put on trial for a crime” nor did it “require ‘less proof, in amount or
degree,’ then was required at the time of the commission of the crime. . . .” Id. at 387.

It took almost another hundred years before the Court’s more expansive
interpretation of the Ex Post Facto Clause was put to rest.11 In the 1990 case
Collins v. Youngblood,12 the Court reversed Kring and Thompson, holding that
those decisions went beyond Justice Chase’s definitions in Calder. The Court
rejected the Kring and Thompson rationale (that the defendant need only be

11.

Other cases skirted the issue but without resolving it. For example, in Beazell v. Ohio, 269

U.S. 167 (1925), at the time of the crime (embezzlement) joint defendants were entitled by state law to
separate trials. By the time of the trial, joint defendants were no longer entitled to separate trials
(except in capital cases). Id. at 169. The Court found no ex post facto violation, stating that “ . . . it is
now well settled that statutory changes in the mode of trial or the rules of evidence, which do not
deprive the accused of a defense and which operate only in a limited and unsubstantial manner to his
disadvantage, are not prohibited.” Id. at 170. The Beazel Court cited with approval Kring and
Thompson, as well as Hopt and Thompson v. Missouri, even though those two sets of cases are not
easy to reconcile. Id. at 171.
Likewise, in Dobbert v. Florida, 432 U.S. 282 (1977), the Court held that a retroactive
change in state law did not violate the Ex Post Facto Clause. In Dobbert, at the time of the crime
(capital murder), state law forbade the imposition of the death penalty if a majority of the jury
recommended mercy. By the time of trial, the law had been amended to make the jury’s decision a
recommendation that was not binding on the court. The trial court rejected the jury’s 10-2
recommendation for mercy and imposed the death penalty. The Court said it is “well settled . . . that
‘[t]he inhibition upon the passage of ex post facto laws does not give a criminal a right to be tried, in
all respects, by the law in force when the crime charged was committed.’ Gibson v. Mississippi, 162
U.S. 565 (1896). ‘[T)he constitutional provision was intended to secure substantial personal rights
against arbitrary and oppressive legislation, see Malloy v. South Carolina, 237 U.S. 180, 183 (1915),
and not to limit the legislative control of remedies and modes of procedure which do not affect matters
of substance.’ Beazell v. Ohio, supra, at 171.” Dobbert, 432 U.S. at 293. The Court described the
change as “procedural” with no change in “the quantity or degree of proof necessary to establish [the
defendant’s] guilt.” Id. at 294 (quoting Hopt v. Utah, 110 U.S. at 589-590).
12.

497 U.S. 37 (1990).

“substantially disadvantaged”) because “the prohibition which may not be
evaded is the one defined by the Calder categories,”13 and Calder says nothing
about “disadvantaging” defendants. Accordingly, retroactive changes do not
run afoul of the Ex Post Facto Clause unless they make innocent conduct criminal (Calder category 1), aggravate the crime (category 2), increase the punishment (category 3), or change the type or quantum of proof required for a conviction (category 4).14 Post-Collins, a significant “disadvantage” to the defendant is not enough unless the change also fits within one of the four Calder
categories.15

13.

Id. at 46 (internal citations omitted); see also Beazell, 269 U.S. at 169-70 (1925).

14.

Collins, 497 U.S. at 46. In Collins, the Court upheld the reformation of an improper

conviction pursuant to a law that was passed after the defendant committed his crime but was applied
retroactively to him. Absent the new statute, the error would have entitled the defendant to a new trial.
The Supreme Court said that the statute did not “punish as a crime an act previously committed, which
was innocent when done; nor make more burdensome the punishment for a crime, after its
commission; nor deprive one charged with crime of any defense available according to law at the time
when the act was committed. Its application to respondent therefore is not prohibited by the Ex Post
Facto Clause . . . .” Id. at 52. For a modern case addressing the sufficiency of the evidence prong of
the Ex Post Facto Clause, see Carmell v. Texas, 529 U.S. 513 (2000) (holding that a statutory
amendment changing the corroborating evidence requirement for convictions of sexual offenses
violates the Ex Post Facto Clause when applied retroactively); see also Danielle Kitson, It’s An Ex
Post Fact: Supreme Court Misapplies the Ex Post Factor Clause to Criminal Procedure Statues, 91 J.
CRIM. L. & CRIM. 429 (2001) (discussing Carmell).
15.

The debate about what can be shoehorned into the Calder categories has continued into the

twenty-first century. In Stogner v. California, 539 U.S. 607 (2003), the defendant was convicted of
crimes committed several decades earlier, subject to a then-existing 3-year statute of limitations. But a
retroactive amendment permitted such charges to be brought within a year of their being reported to
state authorities. The California courts found no ex post facto violation. The US Supreme Court

With this history as a backdrop, we examine the Ex Post Facto Clause
through the lens of parole. In Part I, we trace the robust ex post facto protections against increases in punishment via delayed or deferred parole, which
were firmly established by the Supreme Court and entrenched in the lower
courts by the end of the twentieth century. In Part II, we describe the Court’s
subtle but important shift to a less protective ex post facto regime in two parole
cases just before and at the millennium. In Part III, we show how lower federal
courts have taken the Court’s modest shift and turned the Ex Post Facto Clause
on its head vis-à-vis parole, all but eliminating parole from the clause’s coverage.

reversed (5-4), with acrimonious opinions on both sides. Justice Breyer, writing for the majority,
invoked Justice Chase for the proposition that “the Clause protects liberty by preventing governments
from enacting statutes with ‘manifestly unjust and oppressive’ retroactive effects.” Stogner, 539 U.S.
at 611 (citing Calder v. Bull, 3 Dall. 386, 391 (1798)). Breyer concluded that “The second [Calder]
category—including any ‘law that aggravates a crime, or makes it greater than it was, when committed,’—describes California’s statute as long as those words are understood as Justice Chase understood
them—i.e., as referring to a statute that ‘inflict[s] punishments, where the party was not, by law, liable
to any punishment.” Id. at 613. Justice Kennedy, writing for the dissent, strongly disagreed. He said
that the words of the second Calder category “do not permit the Court’s holding, but indeed foreclose
it. A law which does not alter the definition of the crime but only revives prosecution does not make
the crime ‘greater than it was, when committed.’ Until today, a plea in bar has not been thought to
form any part of the definition of the offense.” Id. at 633. For academic discussion of Stogner, see,
e.g., Ashran Jen, Stogner v. California: A Collision Between the Ex Post Facto Clause and
California’s Interest in Protecting Child Sex Abuse Victims, 94 AMER. J. OF CRIM. L. &
CRIMINOLOGY 723 (2004); Joan Comparet-Cassani, Extending the Statute of Limitations in Child
Molestation Cases Does Not Violate the Ex Post Facto Clause of Stogner, 5 WHITTIER J. CHILD &
FAM. ADVOC. 303 (2006).

In Part IV, we diagnose the causes of this constitutional “wrong turn” and
disentangle the doctrinal morass that ex post facto law has become in the
context of parole. We propose a way forward that makes sense of and harmonizes the US Supreme Court’s doctrine, the constitutional history, and the
underlying purposes of the Ex Post Facto Clause. Specifically, we distinguish
two categories of ex post facto claims that have been raised but not treated
separately in the cases—namely obvious or per se violations versus those that
require some factual development—and we set out a framework for analyzing
these distinct categories of ex post facto claims. In addition to keying off the
history and purpose of the clause, we keep an eye on the practical needs of the
people affected by the Court’s doctrine (prisoners), who have few resources to
litigate these cases and no political influence to promote doctrinal change on
their own.

I.
THE EX POST FACTO CLAUSE AND PAROLE: EARLY PROTECTION AGAINST
CHANGES THAT MIGHT INCREASE PUNISHMENT
Modern penal codes enacted starting in the early-to-mid-twentieth century
raised new ex post facto questions. In many states and in the federal system,
indeterminate sentencing replaced flat sentencing. Prisons started offering
school, job, and mental health programs that created incentives for prisoners to
work toward their own rehabilitation, with the hope of early release. Parole
boards proliferated, with the goal of making informed, professional, and
consistent decisions about early release. As these new penal models were introduced and flourished, the US Supreme Court had to decide if the Ex Post Facto
Clause applied to the many and various new features of these models—features

like indeterminate sentencing, “good time” and other sentence credits, as well
as parole itself.16 Any or all of these features could affect how much time a
person would serve on a given sentence. Parole was the paradigmatic example,
because if statutory, regulatory, or policy changes made parole harder to get
(than it had been when people committed their crimes), and if the changes
applied retroactively, then by definition some people might serve more time in
prison than they would have served but for the retroactive changes.
Before the Court examined these new penal models, its landmark ex post
facto case on changes in sentencing had been Lindsey v. Washington, decided
in 1937.17 In Lindsey, the Court found an ex post facto violation where an
amended statute changed a criminal penalty from “not more than fifteen years,”
with a judge-set minimum of six months to five years, to a flat fifteen years,
but giving the parole board authority to determine the actual length of imprisonment after the prisoner had served six months. Even though prisoners could
have served up to fifteen years under the old statute, the Court held that the
retroactive application of the new statute violated the Ex Post Facto Clause
because the “standard of punishment adopted by the new statute is more
onerous than that of the old.”18 Whether the prisoners would have received or

16.

For an interesting discussion of lex mitior, a doctrine which bars imposing the greater,

original punishment if the punishment is later decreased—the inverse of ex post facto protection—see
Peter Westen, Lex Mitior: Converse of Ex Post Facto and Window into Criminal Desert, 18 NEW
CRIM. L. REV. 167 (2015) (discussing lex mitior and the implications for purposes of punishment and
“desert”).
17.

301 U.S. 397 (1937).

18.

Id. at 400-01.

served shorter sentences under the former statute was immaterial, given that the
amendment created the potential for at least some prisoners to serve longer
sentences under the new regime.
Lindsey quoted text from Kring and Thompson—namely that “[i]t is
plainly to the substantial disadvantage of petitioners to be deprived of all
opportunity to receive a sentence which would give them freedom from custody and control prior to the expiration of the fifteen-year term.”19

This

language was later disavowed in Collins, yet the Collins Court did not reverse
Lindsey (as it had Kring and Thompson) because it conceded that some Lindsey
prisoners might wind up serving longer sentences than they could have served
when they committed their crimes, and thus their ex post facto claim satisfied
the third Calder category of “increase in the punishment.”20
In 1980—again before the Court had narrowed the reach of the Ex Post
Facto Clause in Collins21—the Court finally had to address a feature of a backend modern penal code, as opposed to a direct front-end sentencing issue, as in
Lindsey. In Weaver v. Graham,22 the Court barred the state from retroactively

19.

Id. at 401-02.

20.

Id.

21.

See 497 U.S. at 45 (holding that changes that “disadvantage” the defendant but were other-

wise unrelated to the crime or the punishment or the type or quantum of evidence required as set forth
in the Calder categories were not sufficient to make out an ex post facto claim); see supra notes 12-15
and accompanying text.
22.

450 U.S. 24 (1981). Weaver was a unanimous decision; the concurring Justices did not

disagree with the majority’s analysis of the Ex Post Facto Clause, but took issue only with whether the
new statute in fact operated retrospectively, or whether its benefits (which for some prisoners could

reducing the “gain time” that prisoners could earn to hasten their parole-eligibility date. The Court made clear that retroactive changes that could lengthen
the time served on a sentence fell within the rubric of “punishment” and thus
also fell within the ambit of the Ex Post Facto Clause.
The Weaver Court recognized that retroactive changes to such early
release provisions are no different from retroactive changes to initial sentencing
provisions (like those that were at issue in Lindsey): they implicate the Ex Post
Facto Clause because they are “one determinant” of how long a person will
serve, and because the person’s “effective sentence is altered once this determinant is changed.”23 Suspending or withdrawing such provisions constitutes
an increase in punishment because a “prisoner’s eligibility for reduced imprisonment is a significant factor entering into both the defendant’s decision to plea
bargain and the judge’s calculation of the sentence to be imposed.”24 The Court
also noted that relief under the Ex Post Facto Clause is based not on the individual’s right to less punishment, but rather on the values the Ex Post Facto
Clause was designed to protect—namely to ensure fair notice of the punishment at the time when the crime is committed, and to restrain “arbitrary and
potentially vindictive legislation”25 against the politically weakest members of
society (like criminals and prisoners).

permit or speed up parole eligibility in new ways) might outweigh its harms enough to excuse what
would otherwise be an ex post facto violation. See id. at 36-39.
23.

Id. at 32.

24.

Id. (internal citations omitted).

25.

Id. at 29.

Weaver is notable because it applied the Ex Post Facto Clause where the
increase in punishment took the form of a delay in parole eligibility, despite the
uncertainty as to whether or not the individual prisoner would in fact have been
paroled.26 The Court said the inquiry is a facial one: it “looks to the challenged
provision, and not to any special circumstances that may mitigate its effect on
the particular individual.”27 Weaver, like Lindsey before it, viewed the question
as a pure question of law to be decided by the Court.28 If the statute under
review might result in some prisoners serving longer sentences than they could
have served when they committed their crimes, then retroactive application is
barred by the Ex Post Facto Clause. It made no difference that Lindsey was a
front-end sentencing case while Weaver was a back-end parole case because
the interest and the harm were identical in both cases.
Weaver, like Lindsey, sent a strong signal to the lower courts that almost
any retroactive change that could delay prisoners’ release date renders their
sentence “more onerous” and thus is prohibited by the Ex Post Facto Clause.29
Based on Lindsey and Weaver, and despite the narrowing of the Ex Post Facto

26.

Id. at 30. “Evaluating whether a right has vested is important for claims under the

Contracts or Due Process Clauses, which solely protect pre-existing entitlements. The presence or
absence of an affirmative, enforceable right is not relevant, however, to the ex post facto prohibition,
which forbids the imposition of punishment more severe than the punishment assigned by law when
the act to be punished occurred.” Id. at 29-30 (citations omitted).
27.

Id. at 33 (citations omitted). But see Dobbert v. Florida, 432 U.S. 282, 300 (1977) (sug-

gesting that a defendant in a criminal case cannot bring an ex post facto claim “where the change has
had no effect on the defendant in the proceedings of which he complains”).
28.

Lindsey, 301 U.S. 397 (1937); Weaver, 450 U.S. 24 (1981).

29.

Weaver, 450 U.S at 36.

Clause that occurred in 1990 in Collins, from the late 1970s to the mid-1990s,
numerous US circuit courts of appeal and state supreme courts struck down
retroactive parole changes with such effects. These included not just changes in
the substantive standard to obtain parole, but also arguably procedural changes
or mixed changes that might be described as procedural or substantive, but
which could still delay parole consideration or a prisoner’s release. Examples
included less frequent parole review,30 loss of good time or other credits,31
increases in the minimum time to be served,32 and hurdles making it harder for

30.

See e.g., Rodriguez v. U.S. Parole Commission, 594 F. 2d 170 (7th Cir. 1979) (reduced

opportunity for parole review under federal rules violates Ex Post Facto Clause); Watson v. Estelle,
859 F.2d 105, vacated on other grounds, 886 F.2d 1093 (9th Cir. 1989) (same, for extending parole
review from every year to every three years); Akins v. Snow, 922 F.2d 1558 (11th Cir. 1991) (same,
for changing annual parole review to review every eight years, following a parole denial); Roller v.
Cavanaugh, 984 F.2d 120 (4th Cir. 1993) (same, changing parole review from annual to biennial
review); see also Tiller v. Klincar, 561 N.E.2d 576 (Ill. 1990) (changing parole review from every year
to every three years “constitutes ex post facto punishment . . . [because] the new law ‘constricts the
inmate’s opportunity to earn early release, and thereby makes more onerous the punishment for crimes
committed before its enactment,’” id. at 580 (citing Weaver, 450 U.S. at 35-36); Griffin v. State, 433
S.E.2d 862 (S.C. 1993) (same, for extending parole review from every year to every two years).
31.

See e.g., Arnold v. Cody, 951 F.2d 280 (10th Cir. 1991) (emergency overcrowding credit

statute amendment, which made it more difficult for prisoners who had been denied parole to obtain
release, imposes eligibility requirement which had not existed under earlier statute and thus violates
the Ex Post Facto Clause); Greenfield v. Scafati, 277 F. Supp. 644 (Mass. 1967) summarily aff’d, 390
U.S. 713 (1968) (affirming judgment of three-judge court that found an ex post facto violation in a
statute that eliminated gain time for the first six months following parole revocation as applied to an
inmate whose crime occurred before the law’s enactment).
32.

Devine v. New Mexico Dept. of Corrections, 866 F.2d 339 (10th Cir. 1989) (unfore-

seeable judicial enlargement of a state criminal statute (raising the minimum term of defendant’s

prisoners to petition for review or leniency.33 And as in Lindsey and Weaver,
the courts treated these issues as pure questions of law, to be decided by the
court. By 1995, it was black-letter law that any significant change in the parole
process that (a) applied retroactively, and (b) might delay a prisoner’s release,
was close to a per se violation of the Ex Post Facto Clause.34

sentence from ten years to thirty years), applied retroactively, operates precisely like an ex post facto
law; “If a state legislature is barred by the Ex Post Facto Clause from passing such a law, . . . a State
Supreme Court is barred by the Due Process Clause from achieving precisely the same result by
judicial construction,” id. at 342, citing Bouie v. City of Columbia, 378 U.S. 347, 353-54 (1964); see
also Marks v. U.S., 430 U.S. 188, 192 (1977) (same)).
33.

State v. Reynolds, 642 A.2d 1368 (N.H. 1994) (changing the period for filing a petition for

suspension of sentence from every two years to every four years violates the Ex Post Facto Clause).
34.

As to ex post facto sentencing cases, the U.S. Supreme Court has stayed the course that it

set in Lindsey back in 1937. In Miller v. Florida, 482 U.S. 423 (1987), Florida had amended its
sentencing guidelines between the time Miller had committed his crime and his sentence. The
sentencing court applied the amended guidelines, raising Miller’s presumptive sentence; he was then
sentenced to the new top end of the range. The Florida Supreme Court found no ex post facto
violation, but the U.S. Supreme unanimously reversed. Although it used the pre-Collins ex post facto
test in its analysis, the Court found that the parole guidelines “directly and adversely” affected the
sentence the defendant received, making “more onerous the punishment for crimes committed before
[their] enactment,” citing Weaver, 540 U.S. at 36. Miller, 482 U.S. at 435. Miller, too, therefore
survived Collins, because the change fit snugly within the third Calder category. See California Dept.
of Corr. v. Morales, 514 U.S. 499, 506 n 3 (1995), discussed next in Part II.A.

II.
LATE-TWENTIETH-CENTURY SUPREME COURT DOCTRINE

A. Morales: A Modest Change of Course
In 1995, in California Dept. of Corrections v. Morales,35 the US Supreme
Court slightly tightened the parole ex post facto standard. In Morales,
California had reduced the frequency of parole review for murderers who
committed a second murder. California law still required an annual paper
review, as well as an individualized finding that the delay in parole review
would not harm the inmate’s chances for parole.36 On those limited, unusual
facts, the Court held that the less frequent parole review was not a per se
violation of the Ex Post Facto Clause. Rather, the test was whether, all things

35.

514 U.S. 499 (1995).

36.

On these facts, the Ninth Circuit had held: “By increasing the interval between parole

hearings, the state has denied Morales opportunities for parole that existed under prior law, thereby
making the punishment for his crime greater than it was under the law in effect at the time his crime
was committed. Logic dictates that because a prisoner cannot be paroled without first having a parole
hearing, a parole hearing is a requirement for parole eligibility. Akins v. Snow, 922 F.2d 1558, 1562
(11th Cir.), cert. denied, 501 U.S. 1260 (1991). Accordingly, any retrospective law making parole
hearings less accessible would effectively increase the sentence and violate the ex post facto clause.
We base this conclusion on the Supreme Court’s observation that the denial of parole is a part of a
defendant’s punishment. Warden v. Marrero, 417 U.S. 653, 662 (1974). The [Supreme] Court went on
to note that ‘a repealer of parole eligibility previously available to imprisoned offenders would clearly
present the serious question under the ex post facto clause . . .’ ” Morales v. Cal. Dept. of Corrections,
16 F.3d 1001, 1004 (9th Cir. 1994) (parallel cites omitted).

considered, the change “produces a sufficient risk of increasing the measure of
punishment attached to the covered crimes.”37
Morales was a narrow, practical decision. For obvious reasons, prisoners
who are incarcerated for murder and who then commit a second murder are
highly unlikely to be paroled, especially in the early years after the second
homicide. So changing their in-person review from every year to every three
years (and conducting an annual paper review instead, and making an individualized finding that the delayed review will not increase their risk of an erroneous parole denial, and preserving the board’s ability to shorten the period if
warranted), is hardly a high-risk venture. The Morales Court rejected the prisoner’s argument that the legal standard should be “any conceivable risk of
affecting a prisoner’s punishment.”38 The Court said that such an amorphous
standard would necessarily include anything that “might create some speculative, attenuated risk of affecting a prisoner’s term of confinement,”39 even
such petty changes as reduced access to the prison law library, or a slightly
shorter parole hearing, or the replacement of an old parole board member by
someone new.40 The Court distinguished such “attenuated” changes from the
statute before it, which applied only to a very small number prisoners, who
were provided with several layers of protection, and who were highly unlikely
to be paroled in any event. Morales was thus only the smallest step away from
the universally accepted legal regime described above—that the Ex Post Facto

37.

Morales, 514 U.S. at 509.

38.

Id. at 508.

39.

Id. at 508–509.

40.

Id.

Clause prohibits retroactive changes that might increase the time some prisoners would serve.41
Indeed, just two years later, in 1997, in Lynce v. Mathis,42 the Court again
held that the Ex Post Facto Clause applies to increased punishment in the form
of the retroactive loss of opportunities for parole, citing with approval
Weaver’s admonition that a “prisoner’s eligibility for reduced imprisonment is
41.

Writing for the Court in Morales, Justice Thomas suggested that courts should be realistic

about how parole boards use their limited resources. If multiple murderers are extremely unlikely to be
paroled, and if the state screens such cases carefully to ensure that the modest increased interval
between parole hearings will not delay cases that deserve to be heard, then no prisoners will be harmed
and the board can concentrate on cases in which a real parole decision needs to be made. As the Court
noted, “the evident focus of the California amendment was merely ‘to relieve the [Board] from the
costly and time-consuming responsibility of scheduling parole hearings’ for prisoners who have no
reasonable chance of being released,” citing In re Jackson, 39 Cal.3d 464, 473; 703 P.2d 100, 106
(1985) (quoting legislative history). Morales, 514 U.S. at 507.
In dissent, Justice Stevens belittled the cost/burden rationale, noting that murderers who
commit a second murder are a tiny class, of fiscal/resource insignificance. He argued that the state was
more forthcoming in its own briefs, inviting “the Court to ‘reexamine’ its ex post facto jurisprudence
‘[i]n view of the national trend towards the implementation of harsher penalties and conditions of confinement for offenders and inmates.’ Brief for Petitioners 11 (footnote omitted).” Id. at 521. “The
danger of legislative overreaching against which the Ex Post Facto Clause protects is particularly
acute when the target of the legislation is a narrow group as unpopular (to put it mildly) as multiple
murderers. There is obviously little legislative hay to be made in cultivating the multiple murderer
vote. For a statute such as [the California amendment], therefore, the concerns that animate the Ex Post
Facto Clause demand enhanced, and not (as the majority seems to believe) reduced, judicial scrutiny.”
Id. at 522. Justice Stevens was prescient in believing that acceptance of the state’s invitation would
lead to longer sentences for all, retrospectively as well as prospectively. See infra notes 47 through 51
and accompanying text.
42.

519 U.S. 433 (1997).

a significant factor entering into both the defendant’s decision to plea bargain
and the judge’s calculation of the sentence to be imposed.”43 In Lynce, state
law authorized the department of corrections to award early release credits to
prisoners when the prison population exceeded pre-set levels. A later statute
canceled the credits for some offenders after the credits had been awarded, and
in some cases after the prisoners had been released.
The state tried to distinguish Weaver on the grounds that it had involved
credits earned by the prisoner, rather than credits provided by the state to
alleviate overcrowding. The Court dismissed that distinction: “[I]n Weaver, we
relied not on the subjective motivation of the legislature in enacting the . . .
credits, but rather on whether objectively the new statute ‘lengthen[ed] the
period that someone in petitioner’s position must spend in prison.’”44
In Lynce the Court made clear that the Ex Post Facto Clause forbids
changes that retroactively reduce a prisoner’s opportunity for parole. This is so
because the reduced opportunity includes the risk of a longer sentence for some
prisoners, and because the defendant relies on the parole system in place when
entering (or rejecting) a plea, and the judge relies on the parole system in place
when determining what the sentence will be.45 The Court credited what
43.

Lynce, 519 U.S. at 445-46, citing Weaver, 450 U.S. at 32. Lynce came seven years after

Collins, and thus necessarily the Court was applying a legal standard consistent with Calder and
Collins. See id. at 441, n 13 (explaining that the law in question falls within the four Calder
categories).
44.

Id. at 442 (emphasis added). Here, too, the prisoner-plaintiff did not have to prove that he

would serve more time, but only that someone in the same position could suffer that fate under the
amendment as written.
45.

Weaver, 450 U.S. at 32; Lynce, 519 U.S. at 445-46.

prosecutors and criminal defense lawyers know in their bones— that the parole
regime in place when a person commits a crime influences not just what will
happen at the back end of the sentence, many years down the road, but also
what happens at the front end, namely whether the defendant will plead guilty
or go to trial, and what sentence the judge will impose.
B. An Opening for Opponents of Parole Release
But sometimes all it takes is a few words in an opinion to trigger a seismic
shift. As noted above, before Morales (and even in Morales itself) the Supreme
Court had decided its Ex Post Facto Clause cases as pure questions of law.
Courts could read the new or amended statute, figure out if any prisoners to
whom it applied might serve longer sentences as a result of the law’s retroactive application, and, if the answer was yes, enjoin its retrospective use. Even
Morales can be viewed as a straightforward application of this black-letter law.
The Court simply said, on the unique facts of the California statute, that the risk
of any prisoners ever serving a longer sentence (than they would have served
but for the deferred parole review) was so remote as not to present an
actionable ex post facto claim.
The problem with Morales was not what it did, but what it insinuated
with a few words that Justice Thomas wrote and the tone in which he wrote
them. First, in disparaging the prisoner’s claim that the Ex Post Facto Clause
forbids a change that has “any conceivable risk of affecting a prisoner’s punishment,” Thomas said that such an approach “would require that we invalidate
any of a number of minor (and perhaps inevitable) mechanical changes that
might produce some remote risk of impact on a prisoner’s expected term of
confinement. . . . [T]he judiciary would be charged under the Ex Post Facto

Clause with the micromanagement of an endless array of legislative adjustments . . . .”46
Second, in responding to a comment in Justice Stevens’ dissent, Justice
Thomas dropped a short footnote about the plaintiffs’ burden of persuasion in
these cases. Justice Stevens had criticized the majority for saying that the
prisoner’s claim of increased punishment was “speculative” because Stevens
believed that the amended California law would “inevitably delay the grant of
parole in some cases.”47 Stevens accused the majority of speculating about the
accuracy of the board’s predictions, the suitability of an entire class of
prisoners for parole in the future, and the length of time that would actually
elapse between hearings (despite the board’s ability, in theory, to intercede
early in exceptional cases). Justice Stevens further argued that “[t]o engage in
such pure speculation while condemning respondent’s assertion of increased
punishment as ‘speculative’ seems to me not only unpersuasive, but actually
perverse.”48
In responding to this accusation, Justice Thomas said that Stevens’
suggestion that the speculation “should run in the other direction” (to favor the
prisoner) “effectively shifts to the State the burden of persuasion.”49 “Although
46.

Morales, 514 U.S. at 508.

47.

Id. at 525. Stevens was concerned not so much with the prisoners’ delayed review in the

early years. Rather, he feared that as time went by and prisoners approached the date when they would
get serious board consideration, the less-favorable schedule would surely result in delayed parole for
some people. In our view his concern was justified and prescient, given the holdings of later cases (as
will be shown below).
48.

Id. at 526.

49.

Id. at 510, n 6 (internal quotation marks omitted).

we have held that a party asserting an ex post facto claim need not carry the
burden of showing that he would have been sentenced to a lesser term [under
the prior statutory scheme], . . . we have never suggested that the challenging
party may escape the ultimate burden of establishing that the measure of
punishment itself has changed.”50
Well, yes and no. Up to and including Morales, the Court’s exclusive
focus had been on whether a retroactive law created a sufficient risk that some
prisoners might serve more time than they would have served but for the
change. This was treated as a question of a law that the Court had decided
without much input from the parties—and, if anything, the Court had given the
benefit of the doubt to prisoners. As Justice Stevens noted, “In light of the
importance that the Framers placed on the Ex Post Facto Clause, we have
always enforced the prohibition against the retroactive enhancement
scrupulously.”51
Put another way, although the Court did not approach these cases from a
burden-of-proof or burden-of-persuasion perspective, based on the Court’s
decisions for decades, in practice the prisoners’ burden had been feather-like.
All they had to show was a sufficient risk that, either at sentencing or at parole,
some prisoners might serve more time as a result of the change in law. That
was so because relief under the Ex Post Facto Clause, as the Court unanimously held in Weaver, is based not on an individual’s right to less punishment, but on the values the Clause was designed to protect—namely, to ensure

50.

Id. (citation omitted).

51.

Id. at 516.

fair notice of the punishment at the time when the crime is committed, and to
restrain “arbitrary and potentially vindictive legislation,” especially against
unpopular groups.52
Not long after Morales was decided in 1995, however, the circuit courts
of appeal were already reading the decision broadly, perhaps reflecting the
“get-tough-on-crime” mentality that was sweeping the country and that Justice
Stevens had alluded to in his Morales dissent.
For example, in Shabazz v. Gabry,53 the Sixth Circuit reviewed a facial
attack on a law that reduced the frequency of in-person parole interviews for
parolable lifers from the fourth year of incarceration and every two years
thereafter to the tenth year and every five years thereafter. The change came on
the heels of the election of a new governor and legislative majority who had
run in part on a “law and order” platform.
The Michigan law was implemented with none of the protections
provided in Morales. All 800+ qualified parolable lifers got blanket notices that
their next scheduled review would be deferred by three years (to implement a

52.

Weaver, 450 U.S. at 29.

53.

123 F.3d 909 (6th Cir. 1997). We focus on the Sixth Circuit here because it is our home

circuit. We know its cases best, but we also think its cases are representative. (Full disclosure: the
authors’ clinical law program was counsel of record in some of these cases.) Examples from other
circuits include Hamm v. Latessa, 72 F.3d 947 (1st Cir. 1995) (statutory amendment excluding
inmates with “from-and-after” sentences from parole board hearings did not violate the Ex Post Facto
Clause); Roller v. Gunn, 107 F.3d 227 (4th Cir. 1997) (same, for amendment reducing frequency of
parole reconsideration hearings); Ellis v. Norris, 232 F.3d 619 (8th Cir. 2000) (repeal of statute that
gave prison officials discretion to award additional good-time credits did not violate the Ex Post Facto
Clause).

new five-year review schedule), no matter how long the prisoners had served or
how close the board’s vote to deny parole had been at their last previous
review. Thereafter, they would be reviewed for parole only every five years.
The change did not include any individualized assessment of the prisoners.
This was a classic “retroactive change” that before Morales unquestionably
would have been struck down under the Ex Post Facto Clause.
But the Sixth Circuit held, “The Morales test requires a showing of sufficient risk of increased punishment, not merely ‘some ambiguous sort of disadvantage’ suffered by an inmate.”54 The court found that the plaintiffs had not
proven that the postponement of review (in and of itself) necessarily produced
a “sufficient risk of increasing the measure of punishment attached to the
covered crimes.”55

54.

Id. at 914 (internal citation and quotation marks omitted). The Sixth Circuit’s phrasing

drew on the Supreme Court’s rejection in Collins of the Kring and Thompson line of cases, which held
generally that a mere “disadvantage” was insufficient to support an ex post facto claim, thus restoring
the exclusivity of the four Calder categories. Id. at 912. In doing so, however, the Sixth Circuit ignored
the import of the Court’s parole ex post facto cases—namely, that Weaver had survived Collins
despite relying on the “disadvantaged” rationale, and that Lynce had been decided in the prisoners’
favor (unanimously) after Collins. Both decisions were based not on a hypothetical or attenuated
disadvantage, but on the Court having found that some prisoners might well wind up serving longer
sentences—which is all the Court had ever required in the context of sentencing or parole.
55.

Id. The court noted that “no reliable statistical analysis was available . . . because the

statute had been in effect for too short a period.” Id. at 914. The Sixth Circuit said that the district court
had erred in relying on “anecdotal observations and personal speculation to conclude that the amendments may present sufficient risk of increased punishment.” Id. at 914-15 (emphasis added). Based on
the limited data available (from 1993 to 1995), the court of appeals found that the prisoners had failed
to prove that the delay in parole hearings would inevitably lead to delayed paroles.

C. Garner v. Jones: A Wrong Turn Initiated
The Sixth Circuit’s interpretation of Morales got a big boost from the
Supreme Court in 2000 in Garner v. Jones.56 In Garner, Georgia had reduced
the frequency of parole review from every third year to every eighth year for
lifers who had previously been denied parole. In finding an ex post facto
violation, the Eleventh Circuit had distinguished the Georgia rule from the
California law upheld in Morales because:
[T]he set of inmates affected by the retroactive change [namely all
prisoners serving life sentences] is ‘bound to be far more sizeable than
the set [of murderers who commit a new murder] . . . at issue in
Morales’ . . . . The Georgia law sweeps within its coverage . . . ‘many
inmates who can expect at some point to be paroled,’ and thus ‘seems
certain to ensure that some number of inmates will find the length of
their incarceration extended in violation of the Ex Post Facto
Clause . . . .’”57

56.

529 U.S. 244 (2000). For another case in the same term that was more amenable to an ex

post facto claim and that reinvigorated the Calder analysis, see Carmell v. Texas, 529 U.S. 513 (2000).
That case was decided a few months after Garner, and, in finding an ex post facto violation, it
characterized Collins as defining the fourth category in Calder, instead of eliminating it, as some had
suggested. Carmell overturned a conviction obtained on the testimony of a child victim alone, which
was not allowed under the law in effect when the crime was committed. The Court said that the
amendment was “unquestionably a law ‘that alters the legal rules of evidence, and receives less, or
different, testimony, than the law required at the time of the commission of the offence, in order to
convict the offender,’” citing Calder’s fourth category of forbidden retroactive changes. Carmell, 529
U.S. at 530.
57.

Garner, 529 U.S. at 249, citing Jones v. Garner, 164 F.3d 589, 594-95 (11th Cir. 1999).

The Supreme Court nevertheless reversed. It reiterated that retroactive
parole changes fall within the ex post facto prohibition if they create “a
sufficient risk of increasing the measure of punishment attached to the covered
crimes.”58 Although the Garner Court acknowledged that the requisite risk
could be inherent in the text authorizing the change, the Court held that less
frequent review alone did not make the risk self-evident, “and it ha[d] not
otherwise been demonstrated on the record.”59
One would think (as the Eleventh Circuit had found, and as Justice Souter
argued in dissent in Garner)60 that if nearly all parole review for a large class
of prisoners is postponed from every three years to every eight years, then
surely some prisoners would serve longer sentences than they would have
served under the previous, more generous review schedule. This would be
especially true the longer the new rule remained in place. Over time, more and
more prisoners would approach the date when the board might well view them
as good candidates for parole, yet most would still be reviewed only every
eight years instead of every three years. Statistically, it seems all but certain

58.

Id. at 250; see also Robert Renjel, Casenote: Garner v. Jones: Restricting Prisoners’ Ex

Post Facto Challenges to Changes in Parole Systems, 52 MERCER L. REV. 761, 772-74 (2001)
(discussing Garner and interpreting it to mean that the prisoner must show a “sufficient risk of
increased punishment” and noting that the Court did not address the evidence required to make this
showing).
59.

Id. at 251.

60.

Id. at 261.

that some prisoners would serve more time than they would have served but for
the retroactive, deferred review.61
But in Garner, writing for the majority, Justice Kennedy played down the
specifics of the California law that controlled the outcome in Morales and
played up the language that counseled a hands-off approach to all parolerelated discretionary decisions. Justice Kennedy also paid homage to the
specter of “micromanagement” and said that “[s]tates must have due flexibility
in formulating parole procedures and addressing problems associated with
confinement and release.”62 The Court relied heavily on two important
qualifications in the Georgia parole law. One gave the board discretion to
shorten the eight-year period for worthy candidates, and the other permitted

61.

Justice Souter emphasized this very point in his dissent. Id. at 260-61. (“Before the board

changed its reconsideration Rule, a prisoner would receive a second consideration for parole by year
10, whereas now the second consideration must occur only by year 15; those who would receive a
third consideration at year 13 will now have no certain consideration until year 23, and so on. . . . If a
prisoner who would have been paroled on his fourth consideration in year 16 under the old Rule has to
wait until his third consideration in year 23 under the new Rule, his punishment has been increased
regardless of the average.”) We think he is right. In any other context it is hard to imagine that a
similar blanket delay of discretionary decision making—where a decision is necessary in order to get
the sought-after benefit—would be assumed to have a benign effect.
62.

Id. at 252. Avoiding micromanagement and granting flexibility are appropriate where

retroactive changes create little or no risk of longer sentences. But if the purpose of the Ex Post Facto
Clause is to prevent states from increasing after the fact the amount of time that prisoners serve, then
the focus should be not on the level of intrusion exercised by the courts—which might well be a valid
concern in prison conditions cases—but rather on the effect of the change in the law. See Part IV,
below. As the Court itself noted, “The presence of discretion does not displace the protections of the
Ex Post Facto Clause.” Id. at 253.

expedited review if new information warranted it. (There was no record of
whether the board in fact did either.)
Garner also went well beyond Morales with regard to the prisoner’s
burden. Without citation, the Garner Court said, “In the case before us,
respondent must show that as applied to his own sentence the law created a
significant risk of increasing his punishment.”63 To the contrary, as noted
above, in order to make out an ex post facto violation in the past, the Court had
only required a finding that some prisoners might serve longer sentences due to
the retroactive application of the new or changed law.64
In Garner the Court was also unclear about whether proof as to others
would suffice. It said that the board’s “policy statements, along with [its] actual
practices, provide important instruction as to how the Board interprets its
enabling statute and regulations, and therefore whether, as a matter of fact, the
amendment . . . created a significant risk.”65 Accordingly, “[w]hen the rule
does not by its own terms show a significant risk, the [inmate] must demonstrate, by evidence drawn from the rule’s practical implementation by the
agency charged with exercising discretion, that its retroactive application will

63.

Id. at 255.

64.

The one possible exception was the case of criminal appeals, see Dobbert, 432 U.S. 282,

300-01 (holding that there can be no ex post facto violation where the change in law “had no effect on
the defendant in the proceedings of which he complains.”). But Garner was a Section 1983 action,
which challenged the retroactive parole laws and policies directly. It is hard to see why it would be
viewed as an “as applied” challenge as opposed to a facial challenge, given the legal standard used by
the Court in Weaver, Lynce, and Morales.
65.

Garner, 529 U.S. 244, 256 (2000).

result in a longer period of incarceration than under the earlier rule.”66 The
Court remanded the case to give the prisoner the chance to make a factual
record.67

III.
POST-GARNER: WRONG TURN COMPLETED
A. Garner’s Reading by the Courts of Appeal
A number of circuit cases show how far the pendulum swung after Garner. In Dyer v. Bowlen,68 a state parole board denied parole based on the new
(stricter) substantive standard that was in effect at the time of the prisoner’s
parole hearing rather than on the old (more lenient) standard that was in effect
when the prisoner had committed his crimes more than twenty years before.69
The prisoner was denied relief in the state courts, and he then filed a federal
habeas petition, which he lost. Again, one would think that this would be an
easy case for an ex post facto violation because the substantive standard for
parole had changed to the prisoner’s disadvantage. But, citing Garner, the

66.

Id. at 255 (emphasis added). In this sentence, the Court conspicuously omitted for whom

the longer period of incarceration must be demonstrated—the prisoner bringing the claim or any
prisoners subject to the same regime.
67.

Id. at 256-57.

68.

465 F.3d 280 (6th Cir. 2006).

69.

The two changes highlighted by the appellate court were that the new standard placed an

importance on the seriousness of the offense unrelated to the offender’s rehabilitative efforts, which
had not been part of the earlier standard, and that the new standard provided that if conditions were
met, the board “may” grant parole, whereas the earlier standard had stated that the board “shall” grant
parole. Id. at 282-83.

Sixth Circuit held that the only way to be sure if the change amounted to an ex
post facto violation was to vacate and remand the case for additional fact-finding.70 The court said:
Intuitively, the retroactive application of new parole statutes . . . might
effectuate a sufficient risk of increased punishment, but the ultimate
result depends upon how the parole board actually exercises its discretion. . . . [T]he Supreme Court has made clear that in order for us to
conduct the necessary ex post facto inquiry, we must determine
whether [the prisoner] has produced specific evidence of a sufficient
risk of increased punishment.71
The court conceded that under the Garner standard the plaintiff need not show
that he “actually received a more serious punishment,” but only that he
suffered the requisite risk.72
A third illustrative Sixth Circuit case is Foster v. Booker.73 In Foster, the
plaintiffs were parolable lifers who historically had been paroled at roughly the

70.

A dissenting judge in Dyer agreed that the case should be vacated but said that remand for

fact-finding was unnecessary. In the dissent’s view, the change violated the Ex Post Facto Clause on
its face. The dissent would have remanded for the board to “make its determination under substantive
criteria no more onerous than those applicable at the time of the . . . crime.” Id. at 295.
71.

Id. at 286.

72.

Id. at 288; cf. Richardson v. Penn. Bd. of Probation and Parole, 423 F.3d 282, 292 (3rd

Cir. 2005) (holding that an ex post facto violation was not established because the inmate failed to
prove that the new law created a significant risk of increasing his punishment).
73.

595 F.3d 353 (6th Cir. 2010). Note that in the trial court, the pleadings were titled Foster-

Bey v. Rubitschun, but Westlaw reported the district court’s decision as Bey v. Rubitschun, 2007 WL

same time as prisoners who had committed similar crimes but were given long
indeterminate sentences.74 As noted above,75 after Michigan elected a conservative new governor and legislature, the new administration amended the
parole laws to abolish the existing parole board and to give the governor
authority to appoint a new board.76 The new board quickly adopted a “life

7705668 (E.D. Mich. Oct. 23, 2007), which reverted to Foster v. Booker on appeal, by which time a
new “official capacity” state defendant had been substituted into the case.
74.

Id. at 360. The parole board sought parity because in Michigan most serious felonies are

punishable by “life or any term of years.” Mich. Comp. L. § 791.234. Due to an anomaly in the state’s
parole law and practice, parole eligibility could actually be attained sooner under the lifer law than on
a long indeterminate sentence. As a result, for decades many defendants requested, and many judges
imposed, a life sentence in order to give well-behaved prisoners the chance for an earlier parole. The
board therefore viewed lifers and long indeterminate prisoners alike. See Bey v. Rubitschun, No. 0571318, 2007 WL 7705668, at *4, *13-14 (E.D. Mich. Oct. 23, 2007).
75.

See the discussion of Shabazz v. Gabry, 123 F.3d 909 (6th Cir. 1997), supra at 126-

27 and note __.
76.

Foster, 595 F.3d at 363. The plaintiffs in Foster challenged the cumulative effect of a

series of statutory and policy changes that the plaintiffs said created the requisite risk of delayed
release as applied retroactively to the class. The changes included not just delaying parole review for
lifers (which was challenged in Shabazz) but also included eliminating mandatory in-person interviews
after the first review and substituting paper (file) reviews; increasing the size of the parole board from
seven to ten members (but still requiring a majority vote for lifer paroles); taking the parole board out
of civil service and thus eliminating board tenure and substituting four-year terms; firing the existing,
non-partisan board and replacing it with the governor’s political appointees (who came mostly from
law enforcement and prosecutors’ offices); requiring that a majority of the board members have no
past connection to the department of corrections (when in the past nearly all board members had come
from the DOC); and eliminating prisoners’ right to appeal an adverse parole decision to court while
granting such a right to the prosecutor and the victim. See Bey v. Rubitschun. 2007 WL 7705668, at
*4, (district court summary of the statutory changes). As to delayed parole review, the district court

means life” policy (even though the ostensible substantive standard for parole
had not changed), with the result that release rates for parolable lifers—which
had already declined sharply due to parole-board resource issues and a mushrooming prison population—fell to microscopic levels compared to long-term
historical averages.77 The board also stopped treating parolable lifers and long
indeterminate prisoners the same in making its parole decisions.78
The Sixth Circuit nevertheless reversed the district court’s grant of
summary judgment in the prisoners’ favor. The court held that any risk of more
onerous punishment could be attributed to the changes in the board’s exercise
of its discretion. Citing Garner, the court stated, “[T]he most that can be said
here is that, based on experience, the new Board’s discretion was informed and
then exercised in a way that made it more difficult for plaintiffs to secure
release on parole.”79 Ultimately, the Foster court required that in order for
prisoner-plaintiffs to prevail, they must prove that any adverse changes could

found that, in practice, the board almost never reduced the longer period of review. Some board
members were unaware they could even do so. Id. at *17-18.
77.

Id. at 19-23 (district court’s factual findings).

78.

Id. at 11 (“Representatives of the Parole Board from the years leading up to 1992 con-

sistently testified that nonmandatory lifers were treated the same as prisoners serving long indeterminate sentences . . . for purposes of parole, whereas now the board aligns non-mandatory lifers
with mandatory lifers.”); id. at 13-14.
79.

Foster v. Booker, 595 F.3d. at 364, citing Garner, 529 U.S. at 253. But as noted above,

Garner also held that the mere fact that a board decision is discretionary “does not displace the
protections of the Ex Post Facto Clause,” because there is always the “danger that legislatures might
disfavor certain persons after the fact . . . even in the parole context.” 529 U.S. at 253.

not be accounted for by how the board exercises its discretion—a standard that
is nearly impossible to meet in any discretionary setting.80
An even more extreme example is the Fourth Circuit’s holding in Burnette v. Fahey.81 In Burnette, Virginia had ended indeterminate sentencing and
parole (both prospectively) in 1995. From that date forward, the parole board
only dealt with prisoners who had been sentenced under the prior parole
regime. But after 1995 the board also changed its internal parole policies and
practices. It ceased using risk-assessment instruments; it stopped interviewing
prisoners and instead farmed out parole review to non-board parole examiners;

80.

Foster contrasts sharply with Mickens-Thomas v. Vaughn, 321 F.3d 374 (3rd Cir. 2003),

one of the rare post-Garner cases in which the prisoner prevailed. In Mickens-Thomas, the
Pennsylvania parole board had changed its parole criteria (based on a statutory amendment) to make
“concern for public safety” the overriding factor for parole. Id. at 380. The predictable result was that
violent offenders got far less sympathetic review, and the number of paroles plummeted. Id. The Third
Circuit conducted a thorough review of all the evidence, including statements by the board before and
after the change, and statistical data comparing release rates before and after the change.
The court concluded that the board “mistakenly construed [the statutory change] to signify
a substantive change in its parole function.” Id. at 391. The court found that the change—as applied
retroactively in a habeas case—violated the Ex Post Facto Clause, and it remanded to the board with
instructions to reconsider the prisoner’s case under the former standard. The prisoner had served over
forty years, and his original mandatory life sentence had been commuted by the governor (based on
the recommendation of an earlier board). When on remand the board again denied parole, the Third
Circuit found that the board had failed to comply with the court’s mandate and instead had again used
the very factor that violated the Ex Post Facto Clause. The court issued a writ of habeas corpus
granting the prisoner his unconditional release. Mickens-Thomas v. Vaughn, 355 F.3d 294 (3rd Cir.
2004).
81.

687 F.3d 171 (4th Cir. 2012).

and it mostly stopped meeting as a board and instead voted electronically. 82
Before the changes, the board had relied on fourteen factors listed in its policy
manual in making the parole decisions. After the changes, the board relied on
“the serious nature and circumstances of the crime”—something the prisoner
cannot change no matter how long the prisoner serves—to deny parole in 45
percent of the cases (95 percent in geriatric cases),83 even when the other
factors in the board’s manual or in the statute favored release. Parole rates for
violent felonies dropped precipitously—initially from over 40 percent to below
20 percent, and then (from 2002 to 2008) to around 3 percent.84 In other words,
these prisoners were some thirteen or fourteen times less likely to be granted
parole than before the changes. Prisoners with violent felonies typically served
about 38 percent of their sentence before the changes, but afterward, of the
much smaller group who were paroled, many had served 85 percent of their
sentence.85
On these facts, the district court granted the state’s motion to dismiss
based on the pleadings alone, without discovery. The Fourth Circuit affirmed.
Citing the Sixth Circuit’s decision in Foster, the Fourth Circuit said that the
prisoners had failed to prove that the parole rates could not have occurred just
by changes in the way the board exercised its discretion:
[T]he de facto abolition of discretionary parole . . . is at the crux of the
Inmates’ complaint. . . . [I]t is implausible based on the facts alleged
82.

Id. at 176.

83.

Id.

84. Id. at 176-77.
85.

Id.

that the Board has adopted any such policy. The factual allegations
suggest that the Board has become harsher with respect to violent
offenses, but they do not indicate that the Board has implemented a de
facto prohibition of parole for persons convicted of [violent] offenses.
In the absence of such facts, we cannot reasonably infer that the Board
is failing to exercise its discretion as required by state law.86
This is an exceedingly odd statement of an ex post facto claim. Prisoners
do not have to prove that the parole board has “de facto abolished discretionary
parole,” but only have to demonstrate a “sufficient risk” of increased punishment.87 Compared to cases like Weaver and Lynce—where the plaintiffs
prevailed on their ex post facto claims because some prisoners might plausibly
spend more time in prison than before the changes took effect—the claim in the
Virginia case seems like the easiest of calls (in the prisoners’ favor). The plaintiffs showed that their chances for parole were reduced by around 93 percent
(from a 40+ percent parole rate to a 3 percent parole rate in the years before
and after the 1995 changes, respectively).88 As Garner made clear, the issue in
ex post facto cases is not whether the board is exercising discretion, but how it
is exercising its discretion, in the real world, on the ground.89 Burnette is
especially striking because the prisoner-plaintiffs were never given the chance
to conduct discovery or demonstrate how the board was exercising its discretion.

86.

Id. at 185, citing Garner, 529 U.S. 244, 256 (2000).

87.

Burnette, 687 F.3d at 185 (4th Cir. 2012); Garner v. Jones, 529 U.S. 244, 250 (2000).

88. See Burnett, at 176-77.
89.

Garner, 529 U.S. at 256.

B. Coda on the Ex Post Facto Clause at Sentencing
Peugh v. U.S., a 2013 front-end sentencing case that relies on Weaver,
Morales, and Garner, deserves special attention.90 In Peugh, the US Supreme
Court had to decide if changes to the federal sentencing guidelines violated the
Ex Post Facto Clause.91 When Mr. Peugh committed his crime, his guideline
range was thirty to thirty-seven months.92 By the time he was sentenced,
however, changes in the guideline scoring system had raised his range to
seventy to eighty-seven months.93 He argued that the change was a textbook ex
post facto violation. The government argued that the guidelines were not
“laws” covered by the Ex Post Facto Clause,94 because after U.S. v. Booker95

90. 569 U.S. 530, 133 S. Ct. 2072 (2013); see also Morales, 514 U.S. 499, 508 (1995)
(suggesting parallel application of the Ex Post Facto Clause to sentencing and parole when the
Court stated that it was concerned “with the micromanagement of an endless array of legislative
adjustments to parole and sentencing procedures”).
91.

Id.; see also Andrew C. Adams, One-Book, Two Sentences: Ex Post Facto Considerations

of the One-Book Rule after United States v. Kumar, 39 AM. J. CRIM. L. 231, 245 (2012) (describing
circuit split in application of U.S. Sentencing Guidelines at that time).
92. Peugh, 133 S. Ct. at 2078.
93. Id. at 2079.
94.

This argument resurfaced a debate as to whether the Ex Post Facto Clause applies only to

statutes (“any ex post facto . . . Law”) or also applies to other regulations, rules, or policies that might
or might not be binding upon the parole board or other state authorities. Before Garner, most courts
had extended the Ex Post Facto Clause to formal rules and regulations that have the force and effect of
law, and that are binding on the state officials who administer them. See, e.g., Barna v. Travis, 239
F.3d 169 (2nd Cir. 2001) (holding that the Ex Post Facto Clause does not apply to guidelines that do
not create mandatory rules for release but are promulgated simply to guide the parole board in the
exercise of its discretion); Himes v. Thompson, 336 F.3d 848, 854 (9th Cir. 2003) (holding that more

the guidelines were only advisory: they could not “control” the defendant’s
sentence, which remained at the discretion of the trial court. The district court
and the Seventh Circuit agreed and found no constitutional violation.
The Supreme Court reversed (5-4), noting that:
Each of the parties can point to prior decisions of this Court that lend
support to its view. On the one hand, we have never accepted the
proposition that a law must increase the maximum sentence for which
a defendant is eligible in order to violate the Ex Post Facto Clause.
See, e.g., Lindsey v. Washington (1937). Moreover, the fact that the
sentencing authority exercises some measure of discretion will also not
defeat an ex post facto claim. See Garner v. Jones (2000). On the other

onerous parole regulation violated the Ex Post Facto Clause and that the term “laws” includes “‘every
form in which the legislative power . . . is exerted,’ including ‘a regulation or order’”) (internal citation
omitted). After Garner, some courts extended the Ex Post Facto Clause to informal rules or practices
that influence or control decision making in practice. See, e.g., Michael v. Ghee, 498 F.3d 372, 383
(6th Cir. 2007) (noting that, for ex post facto purposes, the issue is not whether the parole guideline is
a law but whether the challenged “guidelines present a significant risk of increasing the plaintiff’s
amount of time actually served”); Fletcher v. Dist. of Columbia, 391 F.3d 251 (D.C. Cir. 2004) (noting
that the Supreme Court in Garner has “foreclosed our categorical distinction between a measure with
the force of law and ‘guidelines [that] are merely policy statements,’” holding that either can be the
source of an ex post facto violation) (internal citation omitted); Mickens-Thomas, 355 F.3d at 384
(finding that parole policies can be the source of an ex post facto violation). Both Garner and Peugh
lean heavily toward the more liberal view, though that view may have its limits. See, e.g., Peugh, 133
S. Ct. at 2091 (Thomas, J., dissenting) (“It is difficult to see how an advisory Guideline, designed to
lead courts to impose sentences more in line with fixed statutory objectives, could ever constitute an ex
post facto violation.”).
95.

543 U.S. 220 (2005).

hand, we have made it clear that mere speculation or conjecture that a
change in law will retrospectively increase the punishment for a crime
will not suffice to establish a violation of the Ex Post Facto Clause.
See California Dept. of Corrections v. Morales (1995). The touchstone
of this Court’s inquiry is whether a given change in law presents a
“‘sufficient risk of increasing the measure of punishment attached to
the covered crimes.’” Garner, quoting Morales. The question when a
change in law creates such a risk is “a matter of degree”; the test
cannot be reduced to a “single formula.”96
The Court held that because “[t]he federal system adopts procedural measures
intended to make the [g]uidelines the lodestone of sentencing . . . a retrospective increase in the [g]uidelines range . . . creates a sufficient risk of a higher
sentence to constitute an ex post facto violation.”97
The Peugh majority obviously viewed its opinion as a straightforward
application of the general ex post facto standard set forth in Lindsey, Morales,
and Garner. But unlike in the back-end parole cases post-Garner, to prevail
Mr. Peugh had to prove only that the new procedures and scoring rules created
a “sufficient risk” of a higher sentence.98 In short, while Peugh strongly
supports our view that there is a unitary ex post facto standard that applies to
all cases in theory, Peugh also highlights how sentencing cases and parole
cases are treated differently in reality. In Peugh, the judge’s discretion in apply-

96.

Peugh, 133 S. Ct. at 2081-82 (internal citations omitted or shortened).

97.

Id. at 2084.

98.

The proof was easy in his case because he was in fact sentenced to the top of the new

range. Id. at 2079.

ing the guidelines was not determinative, in sharp contrast to the Garner-based
cases, like Foster and Burnette, where the board’s discretion was fatal to the
plaintiffs’ claim. If the Supreme Court applied the Ex Post Facto Clause to
parole in the same way that it applied the clause to sentencing (in Peugh), then
it is hard to see how Garner (and consequently cases like Foster and Burnette)
would not come out the other way.
To sum up, over the last twenty-plus years—from Morales in 1995 to
today—the legal standard to be applied in ex post facto parole cases turned 180
degrees from what it had been for the fifty-plus years before Morales. It went
from a regime in which any retroactive change that might harm prisoners’
opportunity for parole was treated as close to a per se ex post facto violation, to
a regime in which almost no retroactive change in a parole statute, regulation,
or policy can ever rise to the level of an ex post facto violation. The burden of
persuasion also switched from a near presumption that any retroactive change
that might delay parole consideration violates the Ex Post Facto Clause, to a
near presumption that anything having to do with parole is effectively unchallengeable because it involves the exercise of discretion. Under current ex post
facto doctrine, in practice the prisoner must show that other prisoners’ (or even
the individual plaintiff’s) delayed release is all but certain, and must prove that
changes in how the board exercises its discretion could not account for the
delayed consideration or release.99

99.

For example, in Foster the prisoners produced evidence showing (1) that the changes were

proposed and implemented to make current violent felons serve longer prison terms, (2) that in
practice parolable lifers were being evaluated under a harsher substantive standard than before, and (3)
that parolable lifers were released at record low levels in the decade after the change. Bey, 2007 WL

IV.
IDENTIFYING THE WRONG TURN AND GETTING BACK ON THE RIGHT TRACK
In this section, we diagnose how the wrong turn occurred, and we propose
a fix that we think is consistent not just with Supreme Court case law, but also
with the history and purpose of the Ex Post Facto Clause, and with the
practicalities of litigation brought by prisoner-plaintiffs who lack the resources
of other litigants.
Several errors have given rise to the wrong turn. First, courts have glossed
over the fact that some ex post facto claims are easy or obvious violations on
the face of the changed statute, regulation, rule, or policy, or in its operation,
while other ex post facto claims require additional proofs to determine if the
claim has merit. Second, courts (in applying Morales and Garner) have misanalyzed the role and relevance of “discretionary” decision making. Third, based
on the sparse language of Garner, lower courts have imposed burdens in ex
post facto cases that are inconsistent with the purpose of the clause, are
contrary to the Court’s own analysis in Weaver, Morales, and Lynce, as well as
in Peugh, and are impossible to meet given the realities facing prisoner-plaintiffs in the courts.

A. Distinguishing Two Categories of Ex Post Facto Claims
Having already fingered Garner as the primary source of the wrong turn
on the Ex Post Facto Clause, we think a closer look at Garner is warranted.
First, we note that Garner left one crucial aspect of previous ex post facto law
7705668, at *10-15, *19-23 (factual findings of the district court). But that still wasn’t enough for the
prisoners to prevail in the Sixth Circuit.

undisturbed. Before Garner, the Court had typically treated ex post facto
claims as presenting questions of law that could be resolved by looking at the
language or the operational effect of the statute, regulation, rule, or policy that
was the source of the alleged change. Weaver had made the clearest statement
of this approach:
Whether a retrospective state criminal statute ameliorates or worsens
conditions imposed by its predecessor is a federal question. The
inquiry looks to the challenged provision, and not to any special
circumstances that may mitigate its effect on the particular
individual.100
Garner did not change or reject this analysis. In fact, Garner’s language
reinforces that a category of ex post facto claims exists where the courts should
need to look only to the law, regulation, rule, or policy that is being challenged
in order to determine whether, as a matter of law, there is a “sufficient risk of
increasing the measure of punishment attached to the covered crimes.”101 We
agree that where the risk of a longer sentence is apparent, or is predictable with
reasonable certainty, then the plaintiff should win under the Ex Post Facto
Clause as a matter of law. We will refer to this first group as “per se” ex post
facto claims because nothing more is required for the trial court to determine
that the risk of increased punishment is sufficiently high.102

100.

Weaver, 450 U.S. at 33 (citations omitted).

101.

Garner, 529 U.S. at 250.

102.

The same would be true for a losing case, where it is facially obvious or easily predictable

that the alleged risk is too low, and the plaintiff should lose as a matter of law.

As to cases where the risk is not apparent or predictable from the text or
operation of the statute, regulation, rule, or policy, Garner says that
the [plaintiff must] . . . demonstrate, by evidence drawn from the rule’s
practical implementation by the agency charged with exercising
discretion, that its retroactive application will result in a longer period
of incarceration than under the earlier rule.103
Thus, while recognizing the familiar category of historical amendments or
changes of the sort presented in Lindsey, Weaver, Morales, and Lynce—all
classic “per se” cases—Garner also implicitly admits of a second category of
cases where the effect of the changes cannot be readily discerned or predicted
by the text or overt operation of the changed law or policy. In these cases,
factual development is needed to determine whether or not there is an ex post
facto violation.104 We will refer to this second group as “possible” ex post facto
claims.
Though we see Garner as allowing for both kinds of ex post facto cases—
“per se” and “possible” cases—we disagree, as noted above, with the Court’s
holding that the facts of Garner fall into what we are calling the second
“possible” group. Let’s look again at the change: review of potential parolees
was deferred from every three years to every eight years; the change covered
all parolable lifers who had been denied parole at least once before; and the

103.
104

Id. at 255.
Id. (“When the rule does not by its own terms show a significant risk, the respondent

must demonstrate, by evidence drawn from the rule’s practical implementation by the agency
charged with exercising discretion, that its retroactive application will result in a longer period of
incarceration than under the earlier rule.”)

new review schedule had no time limit (meaning that prisoners whose
likelihood of parole increased with the passing years would still be reviewed
only every eight years forever into the future). The change was applied
wholesale against a large class of prisoners and lacked the extra procedural
protections guaranteed by the California statute in Morales (which included not
just an annual paper review, but also a particularized finding that the two-year
delay would not harm the individual prisoner). As in the Sixth Circuit’s Foster
case, the delay was not tied to how close the board’s vote had been in the
previous review, or how much time the prisoner had served.105 We think that
when legislators or prison authorities change the normative rules of parole
sufficiently to result in the likely delayed release of some prisoners over time,
that alone should be sufficient to meet the traditional legal standard under the
Ex Post Facto Clause as a matter of law, in line with Weaver, Morales, and
Lynce. We view Garner as squarely such a “per se” case.
By implicitly holding that deferral of parole review for five years for all
lifers was not a “per se” case, the Court blurred the bright line that had been set
in Weaver, Lynce, and Morales. As a result, even core “per se” cases (like Dyer
v. Bowlen, where the parole board had applied a new harsher substantive parole
standard retroactively, yet the Sixth Circuit still thought it had to “get more
facts” and remanded the case for discovery) are being viewed by the lower
courts as “possible” cases, contrary to Weaver, Lynce, and Morales (and
contrary to the rationale of Peugh). Yet in each of those four cases the same
105.

As noted in the Foster case, the delayed review became so routine over time that some

board members were unaware that the review period could be shortened. See supra notes 73 through
79 and accompanying text.

argument could have been made; namely, that you cannot know for sure what
the effect will be until you see it played out. The takeaway of Weaver, Lynce,
Morales, and Peugh, however, is that where the risk of increased punishment is
sufficiently clear on the face of the change, that is enough to make out an ex
post facto claim, and nothing more is required.
Moreover, to do otherwise—as Garner (perhaps inadvertently) has
encouraged the lower courts to do—thwarts the purpose of the Ex Post Facto
Clause. The Framers viewed the Clause as a bulwark against vindictive legislatures106 and ex post facto laws as “contrary to the first principles of the social
compact.”107 From the early nineteenth century, the Supreme Court recognized
that the Ex Post Facto Clause protects people from legislatures (or policy
makers, in modern parlance) inflamed by the “feelings of the moment” or
subject to “sudden and strong passions.”108 Indeed, in Weaver, the Court
(unanimously) noted that the clause not only ensures notice to the public of
crimes and punishments, but also serves to protect disfavored groups from such
vindictive changes, and promotes separation of powers by making legislatures
the authors of prospective criminal laws and courts the enforcers of those laws
after they are passed.109
106.

See Wayne A. Logan, The Ex Post Facto Clause and the Jurisprudence of Punishment, 35

AM. CRIM. L. REV. 1261, 1267 (1998).
107.

THE FEDERALIST NO. 44, at 287 (James Madison); see also THE FEDERALIST No. 48

(Alexander Hamilton).
108.

Fletcher v. Peck, 10 U.S. 87, 138 (Marshall, C.J.).

109.

Weaver, 450 U.S. at 28-29. See also Evan C. Zoldan, Reviving Legislative Generality, 98

MARQ. L. REV. 624, 654 (2014) (“When a legislature enacts retroactive legislation, it acts with the
knowledge of conduct that has already occurred. As a result, …retroactive legislation permits the

B. The Red Herring of Discretionary Decision Making
In Garner, the Court deferred to the parole board in part because the
decision to extend the review interval had been made by the board itself, and
therefore, like the decisions of prison officials in conditions cases, could be
characterized as “discretionary.”110 What is striking about Garner is that it
reads not like an ex post facto case at all, but rather like a prison conditions
case of the same period. In the years before the millennium, the Court had
decided several prison cases that were designed in no small part to get courts
out of the business of supervising prisons, and to reduce the federal courts’
burgeoning docket of prisoners’ rights litigation.111 These cases emphasized the
broad discretion that prison authorities needed (in order to run their institutions
safely), and at bottom said that interference by federal courts was appropriate

legislature to punish … an individual without naming him specifically but with knowledge of whom
the legislation will … harm.”).
110. Garner, 529 U.S. at 259.
111.

See e.g., Sandin v. Conner, 515 U.S. 472 (1995) (holding that discipline in segregated

confinement did not present the type of atypical, significant deprivation in which a State might
conceivably create a liberty interest). The Sandin Court said that “the involvement of federal courts in
the day-to-day management of prisons often squander[s] judicial resources with little offsetting benefit
to anyone. In so doing, it has run counter to the view expressed in several of our cases that federal
courts ought to afford appropriate deference and flexibility to state officials trying to manage a volatile
environment.” Id. at 482-83, citing Wolff v. McDonald, 418 U.S. 539, at 561–563; Hewitt v. Helms,
459 U.S. 460, at 470–471; Jones v. North Carolina Prisoners’ Labor Union, Inc., 433 U.S. 119, 125
(1977). The Court said, “Such flexibility is especially warranted in the fine-tuning of the ordinary
incidents of prison life . . . .” Sandin, 515 U.S. at 483 (1995). Congress had expressed similar
sentiments in passing the Prison Litigation Reform Act, 42 U.S.C. § 1997 et. seq. (1996) (making it
much harder for prisoners to file, or to win, civil rights cases in federal courts).

only in exceptional circumstances. As a result, prisoners could, for example, be
transferred from one prison to another,112 or they could be moved from general
population into a disciplinary setting,113 and prison libraries could be maintained and modified,114 without undue judicial interference. Garner was decided at a time when prison authorities got huge deference from the Court in
prison conditions cases. We think this blurred the lines, and obscured the key
jurisprudential differences, between cases brought under the Due Process
Clause (alleging deprivation of a constitutional liberty or property interest) or
the Eighth Amendment (alleging cruel and unusual punishment), and cases
brought under the Ex Post Facto Clause (alleging a prohibited retroactive
increase in punishment).
Despite the Garner Court’s importation of the term “discretionary,” the
Court did little to explain what the parole board’s exercise of discretion
regarding parole vis-à-vis the Ex Post Facto Clause has to do with prison
officials’ exercise of discretion regarding prison management or conditions visà-vis the Due Process Clause or the Eighth Amendment.115 We think the

112.

Meachum v. Fano, 427 U.S. 215 (1976).

113.

Sandin v. Conner, 515 U.S. 472 (1995).

114.

Lewis v. Casey, 518 U.S. 343 (1996).

115.

If legislators or executive officials were to take steps—even “discretionary” steps—to

increase prisoners’ punishment after the fact (for example, by making prisoners serve more time, or to
require hard labor where none had been required before), we think the Ex Post Facto Clause would
apply to such changes. Conditions cases, where the discretion involves things like moving prisoners
from one facility to another or changing out the volumes in the prison law library, bear only the most
attenuated connection to punishment, as Morales (correctly) makes clear. Morales, 514 U.S. at 508-11.
But just because we defer to prison officials on some issues (that get marginal protection under a

answer is very little, and that the Court misapplied any such assumed or
subconscious analogy.116
The Ex Post Facto Clause protects against changes in law or policy
(typically aimed at classes of prisoners and applied retrospectively to all such
individuals) that increase punishment, and it should apply regardless of
whether prison officials or parole boards are making “discretionary” policy
decisions or are carrying out the mandatory will of the legislature or the executive.117 The ability of the parole board to use its discretion to change its

different part of the Constitution) does not mean that we should defer to prison officials or parole
boards if the effect of their actions is to increase sentences or to impose other forms of punishment
retroactively in violation of the Ex Post Facto Clause.
116.

In his concurrence in Garner, Justice Scalia made it seem like almost any discretionary

decision of the board would be beyond judicial review—though it isn’t clear if that is because in his
view discretionary decisions cannot ever come under the Ex Post Facto Clause because they are not
laws, or because they are discretionary. Compare Garner, 529 U.S. at 257-58 with id. at 259 (Scalia,
J., concurring).
117.

Foster v. Booker presented an interesting issue in this regard. Over time, statutory changes

in sentencing laws had resulted in much longer felony sentences. The result was a mushrooming
prison population without a concomitant increase in the size of the parole board or its resources. In
response, the parole board made a “discretionary” decision to focus on short-term prisoners who could
be released quickly and easily, freeing up badly needed bed space. Consequently, long-term prisoners,
and especially parolable lifers, did not get reviewed on the schedule required by statute. See Bey v.
Rubitschun, No. 05-71318, WL 7705668, *21-23 (E.D. Mich. Oct. 23, 2007). Although prisoners had
sued (and had won declaratory and injunctive relief) under the Due Process Clause to enforce the
statutorily mandated parole review schedule, see Swearington v. Johnson, 709 F.2d 1509 (6th Cir.
1983), we think the Ex Post Facto Clause is the better claim. The parole board’s policy favored one
group of prisoners over another, with the result that people convicted of more serious crimes wound up
serving more time than they would have served had their review not been delayed. As the then chair of

administrative rules is distinct from its ability, once it is applying those rules, to
make the discretionary decision whether to grant or deny parole in a given case.
It is in the latter kind of decision-making that discretion typically gets the most
deference from reviewing courts; yet even there courts should step in if the
board is violating the Ex Post Facto Clause in a specific case.118 If we are
wrong, and if “discretion” in the broadest sense gets the board a free pass, then
it is hard to see why the Ex Post Facto Clause would prevent a board, for
example, from deciding “in its discretion” not to release prisoners until they
have served, say, 80 percent of their sentences, or for that matter to eliminate
parole altogether, even if those changes overturn decades of consistent policy
to the contrary.119 Indeed, we reiterate that Garner itself acknowledges that

the board noted, “It is fair to say that the board was overwhelmed by the numbers at some point, and
that we had to put our energy and resources into interviewing prisoners who were most likely to be
paroled. Lifer interviews got pushed back, and even when we did lifer interviews, it was more to
comply with the law, and not with an eye to moving anyone forward to parole, because we were so far
behind in our work. In the best of circumstances we kept just marginally abreast of the regular parole
cases, and no doubt . . . the lifers suffered for it.” See Plaintiffs’ Brief in Support of Cross-Motion for
Summary Judgment at 38, Foster-Bey v. Rubitschun, No. 05-cv-71318 (E.D. Mich. Nov. 1, 2006),
ECF No. 114. And of course in Foster-Bey itself the district court found that the parole board did not
just gradually get more conservative over time, but rather the Governor signed legislation that eliminated the existing parole board and created a new board for the purpose of making violent prisoners
serve longer sentences. Bey v. Rubitschun, 2008 WL 7020690, at *10-12, reversed, Foster v. Booker,
595 F.3d at 353. We think that is the very definition of an ex post facto violation.
118.

See Mickens-Thomas (I) and Mickens-Thomas (II), discussed supra at note 80.

119.

The elimination of parole is the paradigmatic example. The entire criminal justice system

is built upon the parole regime in place when the defendant commits the crime. The prosecutor relies
on it in choosing what to charge and what to offer by way of plea bargain; the defense counsel relies

“[t]he presence of discretion does not displace the protections of the Ex Post
Facto Clause.”120
Peugh also undercuts the Garner Court’s and other courts’ reliance on
“discretion.” In Peugh, the government argued that the Ex Post Facto Clause
did not apply because judges retained discretion in sentencing and were not
on it in counseling the defendant whether to plead guilty or go to trial; the defendant relies on it in
making that choice; the probation department relies on it in recommending a sentence; and the judge
relies on it imposing the sentence. If, after the fact, the board can simply (in its discretion) stop
granting paroles, so that all prisoners must serve the maximum term instead of having a fair chance at
parole upon serving the minimum, then the Ex Post Facto Clause is a worthless shell. We think the
historical legal standard is spot on: if the change of law that triggers or governs the board’s altered
exercise of its discretion creates a sufficient risk that some prisoners will serve more time than they
would have served in the past absent the change, then the Ex Post Facto Clause should prohibit the
change, period.
120.

Garner, 529 U.S. at 253. We concede that in parole decision-making some organic change

should be expected over time, as old parole board members leave and new ones are appointed. We see
similar cyclical swings with appellate courts, as the mood of the country or the mores of the majority
shift. But with parole boards, as with courts, typically the “shape” of these cycles will be a relatively
flat sine curve with a fairly long amplitude. As noted above, in Foster-Bey the parole board had civil
service protection and lifetime tenure for decades before statutory amendments eliminated the parole
board and replaced it with new gubernatorial appointees. See Bey v. Rubitschun. 2007 WL 7705668, at
*4, discussed at note 76. Cf. Julio A. Thompson, A Board Does Not A Bench Make: Denying QuasiJudicial Immunity to Parole Board Members in Section 1983 Damages Actions, 87 MICH. L. REV.
241, 252 (1988) (noting that, as of 1988, parole board members were typically selected by governors,
subject to legislative approval, and usually appointed for three to six years). When parole rates decline
sharply from long-term historical norms on the heels of a new administration taking power and
amending parole laws or regulations, that is exactly the kind of retroactive increase in punishment that
we think the Ex Post Facto Clause was intended to prevent. See e.g., Burnette v. Fahey, discussed
supra at 137-139, or Mickens-Thomas (I) and Mickens-Thomas (II), discussed supra at note 80.

bound by the sentencing guidelines.121 The Court rejected that argument, holding that because the guidelines served as the “lodestone” in sentencing,
changing the guidelines—even if they were discretionary—nevertheless
presented a sufficient risk that some prisoners would get or would serve longer
sentences.122
Accordingly, we think that “discretion” is a bright-red herring in ex post
facto analysis. As to both “per se” cases and as to “possible” cases, whether the
change is “discretionary” should have little or nothing to do with the Court’s
analysis under the Ex Post Facto Clause. We therefore think the increasingly
ubiquitous Garner-based notion in the lower courts—that prisoner-plaintiffs
must also prove that the change could not have occurred as a result of the
board’s “exercise of its discretion”—is dead wrong, and ought to be excised
from ex post facto analysis.
C. A Sustained Look at “Possible” Ex Post Facto Cases
We now turn to the second category of ex post facto cases—the
“possible” cases—where it is the implementation of the change of law or policy
that will determine whether or not there is a sufficient risk of prisoners serving
longer sentences. We agree that this second category makes good sense analyt-

121. Peugh, 133 S. Ct. at 2085-87.
122.

Peugh, 133 S. Ct. at 2084. We also note that legislatures exercise their “discretion” when

they choose to amend a statute, and governors exercise their “discretion” when they sign the amended
legislation into law. No one is forcing them to do these things. Yet despite the fact that they are
exercising their discretion, if the text or the effect of the change is to increase punishment after the fact,
the changes cannot be applied retroactively consistent with the Ex Post Facto Clause.

ically where changes in parole law or policy may not explicitly or obviously
increase punishment. Prisoners are prone to challenge even de minimus
changes, and this second category provides a useful mechanism to review what
Morales called the more “attenuated” cases, which may require a fact-based
decision. The category of “possible” ex post facto cases also illustrates that
what at first blush might look like a benign change can violate the Ex Post
Facto Clause if the change creates a sufficient risk that some prisoners will
serve longer sentences.
In Garner, the Court held that “[t]he requisite risk is not inherent in the
framework of [the] amended Rule . . . , and it has not otherwise been
demonstrated on the record.”123 The Court thus treated the case as a “possible”
ex post facto case and remanded it back to the trial court for factual
development.124 The Court said that the relevant inquiry would look at how the
board is implementing the change, whether the change is being used to deny
parole and lengthen terms of custody, what policies animated the change, and
how the board is actually exercising its discretion on the ground.125 These

123.

Garner, 529 U.S. at 251 (internal citation omitted).

124.

See supra Part IV.A. (discussing why we think placing Garner into the group of “possi-

ble” ex post facto claims was a mistake). On remand in Garner, the district court was to determine
whether the amended Georgia rule, in its operation, created a significant risk of increased punishment.
See Jones v. Garner, 211 F.3d 1225 (Mem) (11th Cir. 2000). But the plaintiff died before that inquiry
could be completed. The district court found that the ex post facto claim did not survive the plaintiff’s
death and dismissed his claims as moot. Order, Jones v. Garner, No. 95-cv-03012 (N.D. Ga. June 19,
2001), ECF No. 81.
125.

Garner, 529 U.S. at 255 (suggesting that “the general operation of the Georgia parole

system may produce relevant evidence and inform further analysis on the point.”)

strike us as appropriate things to do with an underdeveloped “possible” ex post
facto claim.126
But more factual development should only be required for true “possible”
cases, where the changes to the parole regime are arguably de minimus or
attenuated, yet not so de minimus or attenuated that the trial court can dismiss
the cases outright on the pleadings under Rule 12(b)(6). But Garner also
muddled the treatment of ex post facto cases (and especially “possible” ex post
facto cases) in two important ways, which we tackle next.

1. Burdens of Proof and Persuasion
First, the Garner Court sent a confusing message by the awkward way it
addressed the question of the prisoner’s burden of persuasion. Up to and
including Morales, the Supreme Court’s legal standard in ex post facto cases
had been “whether a given change in law presents a ‘sufficient risk of
increasing the measure of punishment attached to the covered crimes.’”127 In
Garner, however, the Court substituted the word “significant” for “sufficient”—almost as if the two were synonymous.128 Justice Kennedy did not do
so consistently, and he still recited the legal standard as set forth above (using

126.

As noted above, on the facts of Garner, we think this procedure is unnecessary or

misguided, and a major deviation from the Court’s jurisprudence before and after Garner (in Lindsey,
Weaver, Morales, and Peugh). See Part IV, supra.
127.

Peugh, 133 S. Ct. at 2081-82 (citing Garner quoting Morales). In Peugh, the Court

reaffirmed that the line between an ex post facto violation and a permissible change “is a matter of
degree.” Id. at 2082, quoting Morales at 509.
128.

See generally Garner, 529 U.S. 244 (2000).

the “sufficient” language).129 But reading the opinion, one cannot help but
come away with the feeling that the prisoner-plaintiff lost (what we view as)
his “per se” claim in no small part because he failed to prove a “significant”
risk that he (or others) might serve more time.130 No one on the Court seemed
to have noticed the switch, but the lower courts certainly did!131
The shift is subtle but extremely “significant.” It is also wrong. The only
time the Court had used the words “significant” or “substantial” regarding an
ex post facto claim was back in the pre-Collins day, when defendants or
prisoners could win an ex post facto claim (even if they did not fit within the
four Calder categories) by showing, variously, “a legal signification more
injurious to the accused than was attached to them by the law existing at the
time of the transaction,”132 or a change that deprived them of a “substantial
right involved in [their] liberty,”133 or that the retroactive change “substantially
alter[ed] the consequences attached to a crime already completed . . . .”134 In
this context the words “significant” and “substantial” were used to give criminal defendants or prisoners extra protections that the Calder categories did not
otherwise cover.

129. Garner, 529 U.S. at 250.
130. Id. at 251, 254.
131.

The post-Garner losing cases consistently use the word “significant.” See e.g., Burnette v.

Fahey, 687 F.3d 171, 184-85 (4th Cir. 2012); Foster v. Booker, 595 F.3d 353, 361 (6th Cir. 2010);
Dyer v. Bowlen, 465 F.3d 280, 289-90 (6th Cir. 2006); Richardson v. Penn. Bd. of Probation and
Parole, 423 F.3d 282, 293-94 (3d. Cir. 2005).
132.

Kring, 107 U.S. at 249.

133.

Thompson, 497 U.S. at 352.

134.

Weaver, 450 U.S. at 33.

But after Collins, any use of those terms would be improper, because
either the claim fits within one of the four Calder categories, or it does not. The
Court has said unequivocally that the Ex Post Facto Clause applies equally to
sentencing and to parole because both determine how long the person will
serve (and both affect the sentence imposed).135 Since sentencing and parole lie
at the heart of the third Calder category (increased punishment), requiring a
higher burden of persuasion than in Weaver, Morales, Lynce, and Peugh cannot
be right, as all four of those cases are likewise third–Calder third-category
cases. In Garner, the switch from “sufficient” risk to “significant” or
“substantial” risk occurred without citation to any authority, and none exists.
We think the only question (as to the plaintiff’s burden of persuasion) is
whether or not a “sufficient risk of increased punishment” has been shown, and
of course the plaintiff’s burden of proof (at least in a Section 1983 action) is the
same as in any other civil case: 51 percent (a preponderance of the evidence).
Nor is there any policy reason to raise the burden of persuasion from
“sufficient risk” to “significant risk” — or , for that matter, to require prisoners
to prove the negative fact that any increased punishment could not be attributable to the board’s exercise of its discretion. To the contrary, the default in ex
post facto cases should run the other way because of the nature of the harm.
The harm at stake in ex post facto cases is the worst legal harm that people can
suffer short of state-imposed death—namely, the forced loss of liberty.136 No

135.

See supra notes 34 and 54.

136.

Nor is any after-the-fact damage remedy likely to be available. A Section 1983 action for

unlawful confinement would require the prisoner-plaintiff to show that the illegality of the law in
question was “clearly established” in order to overcome the defense of qualified immunity. See, e.g.,

increased punishment can be imposed, absent notice when the crime was
committed (as to what the punishment would be). The very foundation of the
Ex Post Facto Clause is to prevent the state from illegally extending a person’s
loss of liberty after the fact. So a doctrine that effectively requires the harm to
occur, or the risk of the harm to be “significant,” before courts will say that the
Ex Post Facto Clause has been violated would be anathema to the Framers137
and is a cold comfort to prisoners. Yet that is exactly the situation today, as
Garner has been read by the lower courts.138 Prisoners have little hope of ever
meeting the current legal standard until the harm to them has already occurred.

Wilson v. Layne, 526 U.S. 603, 609 (1999) (citing Harlow v. Fitzgerald, 457 U.S. 800, 818 (1982)).
But, by definition, the law would not be “clearly established” unless or until the prisoner had won the
ex post facto case, and so the state would rarely if ever pay damages for the illegal extra imprisonment
it imposed. See, e.g., Taylor v. Reilly, 685 F.3d 1110, 1114-17 (D.C. Cir. 2012) (holding that parole
officials applying current parole regulations to prisoners would not have reason to know that doing so
would create significant risk of prolonged incarceration, which is required for prisoners’ rights under
the Ex Post Facto Clause to be clearly established).
137.

See generally THE FEDERALIST NO. 84 (Alexander Hamilton) (“The creation of crimes

after the commission of the fact, or, in other words, the subjecting of men to punishment for things
which, when they were done, were breaches of no law, and the practice of arbitrary imprisonments,
have been, in all ages, the favorite and most formidable instruments of tyranny.”)
138.

If hindsight is 20/20, then we can say with confidence that current ex post facto doctrine

reaches the wrong result nearly every time. For example, when (in 1993) parolable lifers in Michigan
first challenged the increase in the interval between their parole reviews, the prisoners lost because the
Sixth Circuit held that not enough time had elapsed for them to prove that the delays alone would
inevitably result in longer prison terms. Shabazz v. Gabry, 123 F.3d 909, 914-915 (6th Cir. 1997).
When (in 2005) they brought a global challenge to all the retroactive changes of the previous decade,
the prisoners lost because the Sixth Circuit held that they couldn’t prove that the changes were not

Moreover, the kind of proof required (to demonstrate that the risk of
delayed release is “significant” and that changes in how the board exercises its
discretion could not account for the delayed consideration or release) is a kind
of proof that prisoners are uniquely ill-equipped and ill-positioned ever to
acquire. Most prisoners’ rights cases are filed in pro per,139 and there is no
reason to think cases raising ex post facto claims are an exception to the rule.140

caused by the board’s exercise of its discretion. See Foster v. Booker, 595 F.3d 353, 361 (6th Cir.
2010).
What we know in hindsight is that parolable lifers in Michigan wound up serving vastly
longer sentences than they would have served under the regime that existed when they were sentenced,
and vastly longer sentences than they likely would have served but for the legislative, executive, and
board policy changes that were applied retroactively to them in the late-twentieth and early-twentyfirst centuries. See e.g., CAPPS, When “life” did not mean life: A Historical Analysis of Life Sentences
Imposed

in

Michigan

Since

1900

(Lansing,

Sept.

2006),

http://www.capps-mi.org/wp-

content/uploads/2013/04/When-life-did-not-mean-life-2006.pdf. Indeed, the changes were so clear and
so harsh that (in 2014) twenty-seven employees and former employees of the Michigan Department of
Corrections signed a statement decrying the plight of the state’s parolable lifers and urging reforms.
See MDOC Professionals Comment on Lifer Paroles (February 2014) at http://www.capps-mi.org/wpcontent/uploads/2014/02/rev-Michigan-Department-of-Corrections-Professionals-Comment-on-LiferParoles.pdf. The signatories included three long-serving former MDOC directors, a deputy director,
two parole board chairs (including the one who had chaired the post-1992 conservative “life means
life” board), as well as a raft of wardens, deputy wardens, and former parole board members. Id.
139.

See Margo Schlanger, Trends in Prisoner Litigation, as the PRLA Enters Adulthood, 5

U.C. Irvine L. Rev. 153, 166-67 (2015).
140.

In a study that our clinic did in the late 1980s, we looked at twelve months of pro se

prisoners’ filings in the Eastern District of Michigan. The study (using random sampling) showed that
of the 585 cases filed, 40 percent of the cases were dismissed by magistrate judges under 28 U.S.C. §
1915(d) before service of process. Another 57 percent were dismissed on motions to dismiss or on

It is one thing to say, as Justice Thomas said in Morales,141 that prisoners have
the “burden of persuasion” in “per se” cases like Lynce, Weaver, and Morales,
where in reality the Court read the statute and held as a matter of law that the
challenged amendments did (or as in Morales did not) “produce[] a sufficient
risk of increasing the measure of punishment attached to the covered
crimes.”142 It is a completely different thing to require prisoners to prove a
“significant risk”143 and to prove the negative fact that any harm they suffer
could not be the result of a change in the way the board exercises its
discretion.144
To meet these burdens under current ex post facto doctrine would require
not just lawyers to represent the plaintiffs, but also elaborate and expensive
discovery. It almost certainly would also require the services of high-end
statistical experts to exclude all other variables that might arguably be the cause
of longer prison terms. None of this is within the reach of unrepresented
prisoners. Nor can prisoners amass the sort of practical on-the-ground evidence

summary judgment, and in most of those cases the prisoners never filed another document after the
form complaint. At any point in the process, less than 7 percent of the prisoners ever had a lawyer, and
in those cases the lawyer withdrew before the end of the case about 60 percent of the time. See Hardin
v. Straub, 490 U.S. 536 (1989), Plaintiff’s Reply Brief, Appendix (on file with the authors). These
findings were consistent with more detailed studies at the time. See e.g., T. Eisenberg and S. Schwab,
The Reality of Constitutional Tort Litigation, 72 CORNELL L. REV. 641 (1987) (noting that “[o]ver the
three years studied, only seventeen prisoner constitutional tort cases were counseled . . .”).
141.

See supra note 46 - 48, and accompanying text.

142.

Morales, 514 U.S. at 509, 510 n.6.

143.

Garner, 529 U.S. at 250-51.

144.

Foster v. Booker, 595 F.3d 353 (6th Cir. 2010).

that would be the focus of the factual inquiry, and that invariably requires
depositions, document requests, and other in-depth discovery. If a prisoner
must produce “specific evidence” of the risk, as a practical matter his action is
usually doomed, especially if he is proceeding pro se.145 We note that even
with lawyers who have the time and resources to do elaborate and expensive
discovery, the burden has still proven to be too high.146

2. How Individualized Must an Ex Post Facto Showing Be?
Second, Garner is unclear about whether the prisoner must show that
some prisoners will serve longer sentences or must show that the prisonerplaintiff himself will serve a longer sentence. Again, as with “sufficient” versus
“significant,” Garner appears to say both.147 This, too, has confused the lower
courts, leading them to require a higher burden of persuasion or burden of
proof—for example, by making prisoners show a risk of increased confinement
with respect to the specific plaintiff raising the claim as opposed to similarly
situated potential parolees.148 This kind of showing is familiar to courts where
the plaintiff must prove individualized harm in order to win both on liability

145.

One of the ironies of the current doctrine is that ex post facto sentencing claims —where

the legal standard is easier to meet—are more likely to be brought by appointed trial or appellate
counsel, while ex post facto parole cases arise long after the prisoner has a right to counsel.
146. See e.g., Foster v. Booker, 595 F.3d 353, 361 (6th Cir. 2010) (holding that plaintiff must
show that his harm is not attributable to board’s exercise of discretion); Richardson v. Penn. Bd. of
Probation and Parole, 423 F.3d 282, 284 (3rd Cir. 2005) (holding that the parole board may be using
improper standard but plaintiff failed to show that he was individually harmed by it).
147.

See Garner, 529 U.S. at 250-54 (suggesting either or both).

148.

See e.g., Burnette, 687 F.3d at 184; Richardson, 423 F.3d at 291.

and damages (or to win injunctive relief) and may have been mistakenly
invoked as a kind of default, but it should not be used in ex post facto cases. On
this issue, even the Morales Court consistently referred to “classes” of
prisoners, “some prisoners,” “any prisoner’s actual term of confinement,” and
the like.149 The Court pointedly did not require a showing that Mr. Morales’
own punishment would increase.150 Garner is unique in suggesting such a
requirement when Weaver, Morales, Lynce, and Peugh do not.
D. “Possible” Cases: Putting It Together
Where further factual development is required (which will be the norm for
“possible” ex post facto claims), there is yet one more reason why the burden
of persuasion on prisoner-plaintiffs should be light. In nearly all of these cases,
it is the state defendants (prison staff, corrections administrators, parole board
members) or their agents who possess the information that the court needs in
order to make an informed decision. The state defendants will be the keepers of
the statistics that reveal parole rates: they will have the memos and emails that
reveal the state’s motivation in implementing the changes (to the extent that
motivation or credibility might be relevant), the parole files and notes bearing
on how the changes have affected the board’s actual decision making on the
ground, among other potentially relevant information.
In other situations of information asymmetry, courts have often imposed a
series of shifting burdens. In workplace discrimination cases, for example, the
plaintiff must make out a prima facie case showing that discrimination could

149.

Morales, 514 U.S. at 512-13, 520.

150.

Id. at 512-14.

account for the alleged harm.151 The burden then shifts to the employer to rebut
this prima facie case by articulating some legitimate, nondiscriminatory reason
for the employment action.152 If the defendants succeed, the plaintiff still gets a
chance to avoid dismissal if the plaintiff can show that the innocent explanation
is a pretext.153 A similar evidentiary progression is used in Batson154
challenges—where a criminal defendant alleges discriminatory jury strikes by
the prosecution—for the same reasons.155 A claim brought by a prisoner under
the Religious Freedom Restoration Act (RFRA)156 shares this structure as
well.157 While these areas of law may have distinctive characteristics, the
consistent parallel—and the one that matters in the context of ex post facto

151.

McDonnell Douglas Corp. v. Green 411 U.S. 792, 802-03 (1973) (laying out this

framework for Title VII workplace discrimination claims).
152.

Id. Texas Department of Community Affairs v. Burdine, 450 U.S. 248 (1981), clarified that

in the context of Title VII claims, the employer bears a “burden of production” at this stage and that
the plaintiff retains the “ultimate burden of persuading the court that she has been the victim of
intentional discrimination.” Id. at 255-56.
153.

McDonnell Douglas, 411 U.S. at 804.

154.

Batson v. Kentucky, 476 U.S. 79 (1986).

155.

Id. at 93-94.

156.

42 U.S.C. § 2000bb.

157.

Another area of law in which a similar burden-shifting structure exists is the antitrust Rule

of Reason, where the plaintiff must initially show that the restraint produces anticompetitive effects in
the relevant market, then the burden shifts to the defendant to come forward with a legitimate
precompetitive justification for the restraint, and then the burden shifts back to the plaintiff. See
Standard Oil Co. v. United States, 221 U.S. 1 (1911); Board of Trade of City of Chicago v. United
States, 246 U.S. 231 (1918); see also Daniel C. Fundakowski, The Rule of Reason: From Balancing to
Burden Shifting, 1 PERSPECTIVES IN ANTITRUST 1 (2013).

prisoner litigation—is the significant information asymmetry between the
plaintiff and the defendants. Although these burden-shifting schemes are not
without trenchant critiques158—which we do not take lightly—we nonetheless
think that, in an area that at present lacks coherent doctrine, looking to familiar
and (relatively) easy-to-implement structures may help courts make decisions
that are more consonant with the purpose of the Ex Post Facto Clause and more
consistent from case to case. Burden shifting is one such possibility, though
courts should be open to others as well.
If burden shifting were adopted for “possible” ex post facto claims, evidence that might carry the plaintiff’s initial burden could include: a prima facie
showing that the prisoner’s “eligibility for reduced imprisonment [was] a significant factor entering into both the defendant’s decision to plea bargain and the
judge’s calculation of the sentence to be imposed,”159 evidence that the purpose
of the change was to “get tough” on prisoners or otherwise to extend sentences,
or evidence from past or present DOC or parole board officials that similarlysituated prisoners seem to be serving longer sentences than in the past. We also
think that in “possible” ex post facto cases the trial court should appoint coun-

158.

See Martin J. Katz, Reclaiming McDonnell Douglas, 83 NOTRE DAME L. REV. 109, 170

(2007) (“[C]ritics argue that McDonnell Douglas stacks the deck against plaintiffs by requiring them
to disprove every reason offered by the defendant for its action . . . .”); Elizabeth C. Williamson, City
of Boerne v. Flores and the Religious Freedom Restoration Act: The Delicate Balance Between
Religious Freedom and Historic Preservation, 13 J. LAND USE & ENVTL. L. 107, 124-25, 138-39
(1997) (explaining that concerns that “RFRA’s ‘boundless’ requirement that the least restrictive means
be used when neutral laws affect religious freedom . . . will surely affect governments’ abilities to
protect religious structures”).
159.

Lynce, 519 U.S. at 445-46, citing Weaver, 450 U.S. at 32.

sel and permit discovery as early as practicable but certainly if the prisonerplaintiff meets his initial burden of persuasion as to the plausible effect of the
change.
The burden would then shift to the government to show that, in its
operation, the challenged change does not pose a “sufficient risk” of increasing
punishment, which the plaintiff could then challenge as pretextual or wrong.
This might still pose a high bar for prisoner-plaintiffs, but it would be a sea
change over their current burden of having to prove both that there is a “significant” risk of increased incarceration and that the harm they have suffered
cannot be attributable to the exercise of the board’s discretion. In our view, the
initial burden on the plaintiff should be quite light, given the extraordinary
nature of the harm, the historical legal standard applied in these cases, and the
prophylactic purpose of the Ex Post Facto Clause.

E. A New (Old) Approach to Ex Post Facto Doctrine
To summarize, the Ex Post Facto Clause is all but gutted if (1) prisoners
cannot prove their ex post facto claims until after they have already suffered
the very harm (increased punishment) that the Clause was designed to protect
against, and (2) the prisoners’ burden of proof is raised to the point that pro se
litigants (or even represented plaintiffs) can never meet it. Yet under current ex
post facto doctrine, almost no changes to parole regimes can be challenged
successfully, no matter how harmful their effect.160 The irony, of course, is that

160.

Before Morales and Garner, prisoner-plaintiffs won nearly every ex post facto case

relating to obstacles put in the way of their parole eligibility. See supra Part I. Since Morales and
Garner, it is hard to find a winning parole ex post facto case on those issues. See e.g., Wallace v.

until very recently, nearly all the statutory and policy changes regarding parole
over the past fifty years have been in the direction of harsher treatment for prisoners, as a result of political shifts from the 1960s to the 2000s, combined with
the fact that the Court has not accepted a parole ex post facto case since Garner
in 2000.161
In this section we have tried to make sense of the Court’s cases and
impose some order by clarifying two types of ex post facto claims—those in
which there is a “per se” violation that can be decided as a matter of law and
those in which there is a “possible” violation that requires fact-finding. In both
of these types of cases, where the criminal defendant or prisoner-plaintiff is
challenging a normative, structural-type of change that affects all similarly
situated prisoners, we think the Supreme Court’s long-established legal

Quarterman, 516 F.3d 351 (5th Cir. 2008) (changing parole board voting requirement from threemember panels to the entire eighteen-person board does not violate the Ex Post Facto Clause);
Richardson v. Penn. Bd. of Probation and Parole, 423 F.3d 282 (3rd Cir. 2005) (parole board may be
using improper old standard but prisoner-plaintiff failed to show that it harmed him individually);
Newman v. Beard, 617 F.3d 775 (3rd Cir. 2010) (new requirements that prisoners attend sex offender
therapy and admit guilt in therapy in order to be parolable do not violate the Ex Post Facto Clause).
161.

Only in the last few years has the pendulum begun to shift back, as the cost of mass

incarceration has spiraled upward, fueling a reaction rooted not primarily in notions of justice but in
efforts to save or to redistribute public tax dollars (and perhaps rooted also in modern research
showing that imprisonment rates and crime rates have less to do with each other than we once
thought). See e.g., Michelle Alexander, The New Jim Crow 101 (2012) (noting that “violent crime is
not responsible for mass incarceration. As numerous researchers have shown, violent crime rates have
fluctuated over the years and bear little relationship to incarceration rates—which have soared during
the past three decades regardless of whether violate crime was going up or down. Today violent crime
rates are at historically low levels, yet incarceration rates continue to climb.”).

standard remains good law and should always apply. As the Court confirmed in
Peugh, “The touchstone of [the] Court’s inquiry is whether a given change in
law presents a ‘sufficient risk of increasing the measure of punishment attached
to the covered crimes.’”162
If Garner is read narrowly, as Peugh read it, and as we think it should be
read, then most parole ex post facto cases should fit comfortably into the first
category of obvious “per se” claims. Historically, the Court has had little
trouble determining as a matter of law whether a change in a sentencing or
parole regime creates a sufficient risk of increasing some prisoners’ punishment. Morales is a good example even though it went against the prisoner: the
harm to twice-convicted murderers of slightly delayed parole review was negligible. Peugh is also a good example: there the disagreement among the Justices
was about whether the changes to the parole guidelines fell within the ambit of
the Ex Post Facto Clause, but no one doubted that some prisoners would serve
longer sentences as a result of those changes. We think Garner itself was also
good example (just gone wrong): the delayed parole review met the standard of
Weaver, Morales, and Lynce (as later applied by the Court in Peugh) and therefore should have been treated as a “per se” ex post facto violation.

CONCLUSION
Today the Ex Post Facto Clause no longer protects a powerless
disenfranchised minority (prisoners) from “arbitrary and potentially vindictive
legislation” and the passing political forces that give rise to it, as the Supreme

162.

Peugh, 133 S. Ct. at 2082 (quoting Garner, 529 U.S. at 250).

Court said the Ex Post Facto Clause must.163 To the contrary, the Court’s
modest “about face” in Morales and its abstruse opinion in Garner have
resulted in far longer sentences for some prisoners whom the legislative or
executive branches specifically targeted for harsher treatment long after the
prisoners committed their crimes. The harsher treatment has resulted in
precisely the harm that the Ex Post Facto Clause was designed to prevent. The
extra time people serve on the back end of their sentences—as the result of
delayed or denied parole—can add years to their incarceration, and cumulatively can add hundreds of millions of dollars to the costs of corrections
nationally, now with close to zero constitutional protection under the Ex Post
Facto Clause. At the same time, in its sentencing ex post facto cases (most
recently represented by Peugh), the Court has continued to apply its traditional
scrupulous ex post facto standard, holding that changes which might result in
longer sentences for some defendants cannot be applied retroactively. While
the Court’s analytical split may be inadvertent and may have been heightened
by the lower courts, it is no less illogical. The Ex Post Facto Clause remains
robust when applied to sentencing, but the clause has become toothless as
applied to parole, despite historically identical doctrine and identical harm. The
wrong turn needs to be corrected.

163.

Weaver, 450 U.S. at 29 (1981).