The Forgotten Jurisprudence of Parole and State Constitutional Doctrines of Vagueness

The majority of carceral sentences in the United States include the possibility of discretionary release on parole. Most such sentences, however, are unconstitutionally vague. Their unconstitutionality has gone unnoticed because contemporary scholarship and litigation about vague laws have focused on the U.S. Constitution in lieu of state constitutions. This Article unearths historic state court decisions holding that sentences that end through the discretionary judgment of a parole board are “void for uncertainty.” Although state void for uncertainty doctrines share some similarity with the federal vagueness doctrine, they are far more demanding as applied to criminal punishment. By urging revival of the void for uncertainty doctrine, this Article outlines a novel path for state constitutional litigation and proposes how state legislatures can reform parole statutes to put them on sound constitutional footing.

Introduction

As federal courts have narrowed protections for criminal defendants under the U.S. Constitution, some courts have recognized more robust protections under state constitutions in areas including search and seizure,1See, e.g., Jeffrey S. Sutton, 51 Imperfect Solutions: States and the Making of American Constitutional Law 82 (2018) (citing fifteen state constitutions under which chasing a suspect is sufficient to constitute a seizure, which requires reasonable suspicion); Erwin Chemerinsky, Presumed Guilty: How the Supreme Court Empowered the Police and Subverted Civil Rights 290–91 (2021) (first citing State v. Goss, 834 A.2d 316 (N.H. 2003) (discussing how the state constitution requires warrants for police searches of trash); then citing State v. Sullivan, 74 S.W.3d 315 (Ark. 2002) (discussing how the state constitution prohibits pretextual stops); and then citing State v. Farris, 849 N.E.2d 985 (Ohio 2006) (discussing how the state constitution prohibits admissibility of physical fruits of police questioning without proper Miranda warnings)). excessive punishment,2See, e.g., William W. Berry III, Cruel State Punishments, 98 N.C. L. Rev. 1201, 1242 (2020). and imposition of the death penalty.3See, e.g., Recent Case, State Constitutional Law—Capital Punishment—Washington State Supreme Court Declares Death Penalty Unconstitutional in Washington.—State v. Gregory, 427 P.3d 621 (Wash. 2018)., 132 Harv. L. Rev. 1764, 1771 (2019). The trend is part of the phenomenon that scholars call “judicial federalism” in which states interpret their own constitutions independently from analogous provisions in the federal Constitution.4See Lawrence Friedman, The Once and Future Constitutional Law: On the Law of American State Constitutions, 74 Alb. L. Rev. 1671, 1672 (2011). Enthusiasm for judicial federalism spiked in 1977 when Justice Brennan sounded a “call to state courts to step into the breach” in the wake of U.S. Supreme Court cases that rolled back protections for individual rights.5William J. Brennan, Jr., State Constitutions and the Protection of Individual Rights, 90 Harv. L. Rev. 489, 503 (1977); see also Hans A. Linde, First Things First: Rediscovering the States’ Bills of Rights, 9 U. Balt. L. Rev. 379, 383–84 (1980). State constitutional litigation has since resulted in extended protections in the realms of abortion rights,6See Ctr. for Reprod. Rts., State Constitutions and Abortion Rights: Building Protections for Reproductive Autonomy (2022), https://reproductiverights.org/‌wp-content/‌uploads/‌2022/‌07/‌State-Constitutions-Report-July-2022.pdf [https://perma.cc/‌Z9VV-5HFG]. marriage equality,7See Friedman, supra note 4, at 1678–79 (citing cases from Massachusetts, Iowa, and Connecticut that struck down prohibitions on same-sex marriage under the respective state constitutions). voting rights,8See Valerie L. Snow, Comment, State Constitutions and Progressive Crimmigration Reform, 23 U. Pa. J.L. & Soc. Change 251, 252 & n.18 (2020) (citing League of Women Voters v. Commonwealth, 178 A.3d 737, 737 (Pa. 2018)). and equitable school funding.9See John Dinan, State Constitutional Amendments and American Constitutionalism, 41 Okla. City U. L. Rev. 27, 41 (2016). Nevertheless, the majority of constitutional litigation and scholarship remains focused on the U.S. Constitution, and most state constitutional provisions have been interpreted in “lockstep” with the U.S. Supreme Court’s interpretation of analogous federal constitutional provisions.10See James A. Gardner, The Failed Discourse of State Constitutionalism, 90 Mich. L. Rev. 761, 788 (1992).

This Article applies state constitutional law to a new realm of criminal law: challenges to vague statutes. Contemporary scholarship about the vagueness doctrine has focused almost exclusively on the U.S. Constitution.11See Michael J. Zydney Mannheimer, Vagueness as Impossibility, 98 Tex. L. Rev. 1049, 1055 (2020) (analyzing Supreme Court jurisprudence on vagueness without mention of state constitutional law); Robert Batey, Vagueness and the Construction of Criminal Statutes—Balancing Acts, 5 Va. J. Soc. Pol’y & L. 1, 13–16 (1997) (same); Ryan McCarl, Incoherent and Indefensible: An Interdisciplinary Critique of the Supreme Court’s “Void-for-Vagueness” Doctrine, 42 Hastings Const. L.Q. 73, 73–76 (2014) (same); Note, The Void-for-Vagueness Doctrine in the Supreme Court, 109 U. Pa. L. Rev. 67, 67–73 (1960) (same); John F. Decker, Addressing Vagueness, Ambiguity, and Other Uncertainty in American Criminal Laws, 80 Denv. U. L. Rev. 241, 242–44 (2002) (same). Although supporters of “judicial federalism” have called for independent state constitutional doctrines in many areas of law, none appear to have applied that call to the doctrine of vagueness.12See, e.g., Chemerinsky, supra note 1, at 291–95; Sutton, supra note 1, at 177; Brennan, supra note 5, at 491, 500–01; Berry, supra note 2, at 1205–06, Snow, supra note 8, at 261–62. Yet, two features of the vagueness doctrine make it particularly apt for state constitutional analysis. First, the vagueness doctrine differs from many other constitutional protections because it has roots in the principle of separation of powers.13See Sessions v. Dimaya, 138 S. Ct. 1204, 1224–28 (2018) (Gorsuch, J., concurring). Most state constitutions have more robust doctrines of separation of powers than the U.S. Constitution, and therefore one would expect state constitutional doctrines of vagueness to be more robust than the federal analog.14See Jim Rossi, Institutional Design and the Lingering Legacy of Antifederalist Separation of Powers Ideals in the States, 52 Vand. L. Rev. 1167, 1190–91 (1999). Second, for over a century, state courts applied a far-reaching vagueness doctrine to criminal punishment, and the U.S. Supreme Court has largely ignored that precedent in developing its jurisprudence of vagueness.

On a practical level, state constitutional development of the vagueness doctrine has potential to impact hundreds of thousands of carceral sentences. This Article unearths a history of how state courts applied the vagueness doctrine to carceral sentences that included the possibility of discretionary release through a parole board (henceforth referred to as “indeterminate sentences”). The federal government and sixteen states prospectively abolished indeterminate sentences in the late twentieth century, but thirty-four states continue to rely heavily on indeterminate sentencing.15See Kevin R. Reitz, Edward E. Rhine, Allegra Lukac & Melanie Griffith, Robina Inst. of Crim. L. & Crim. Just., American Prison-Release Systems: Indeterminacy in Sentencing and the Control of Prison Population Size 27 (2022), https://robinainstitute.umn.edu/‌sites/‌robinainstitute.umn.edu/‌files/‌2022-05/‌american_prison-release_systems.pdf [https://perma.cc/‌EB7V-XBXP]. An estimated two-thirds of all carceral sentences in the United States are indeterminate sentences.16Kevin R. Reitz & Edward E. Rhine, Parole Release and Supervision: Critical Drivers of American Prison Policy, 3 Ann. Rev. Criminology 281, 283 (2020).

Most parole boards exercise unfettered discretion17See Thomas & Reingold, supra note 47, at 218. in deciding whether to grant parole, and their discretion has been regularly criticized as arbitrary,18See, e.g., Kristen Bell, A Stone of Hope: Legal and Empirical Analysis of California Juvenile Lifer Parole Decisions, 54 Harv. C.R.-C.L. L. Rev. 455, 507–08 (2019). racially biased,19See, e.g., Michael Winerip, Michael Schwirtz & Robert Gebeloff, For Blacks Facing Parole in New York State, Signs of a Broken System, N.Y. Times (Dec. 4, 2016), https://www.nytimes.com/‌2016/‌12/‌04/‌nyregion/‌new-york-prisons-inmates-parole-race.html (last visited Apr. 7, 2023) (reporting on a 2016 investigation of New York parole hearings and finding that one in four white men were released at their initial parole hearing, whereas one in six Black or Hispanic men were released at their initial parole hearing). and dominated by political influence.20See, e.g., Reitz & Rhine, supra note 16, at 286. Very few procedural protections apply at parole hearings, and some parole boards are highly unprofessional.21See, e.g., Mae C. Quinn, Constitutionally Incapable: Parole Boards as Sentencing Courts, 72 SMU L. Rev. 565, 594–95 (2019) (describing New Hampshire parole board’s use of “profanity, foul language, and mockery of mental health issues”). A 2017 investigation in Missouri, for example, found that parole board staff were playing a game in which they scored points by getting a parole candidate to repeat nonsensical words during the parole hearing, such as “platypus” or lyrics to “Folsom City Blues.”22See id. at 596.

Even in states where the parole board is more professionalized and basic procedural protections are in place, the sheer breadth of parole-release discretion makes it impossible to know when, if ever, an indeterminate sentence will end. For example, in 1988, Mr. William Palmer was sentenced to life with the possibility of parole for a kidnapping conviction when he was seventeen years old.23See In re Palmer, 479 P.3d 782, 786 (Cal. 2021). He was eligible for parole after eight years but ultimately served over thirty years because the California Board of Parole Hearings denied him parole ten times.24Id. at 785; In re Palmer, 245 Cal. Rptr. 3d 708, 712 (Ct. App. 2019), rev’d, 479 P.3d 782. In the meantime, the Board released other people at their initial parole hearings, many of whom had been convicted of murder as adults and had less impressive records of rehabilitation in prison.25See Application to File Brief of Amici Curiae and Proposed Brief of Amici Curiae Law Professors in Criminal, Administrative, and Constitutional Law in Support of Petitioner William M. Palmer at 31–32, In re Palmer, 479 P.3d 782 (No. S252145) (comparing cases of individuals granted parole prior to Mr. Palmer despite more misconduct in prison and more aggravated offenses than Mr. Palmer). The disparate length of Mr. Palmer’s sentence is not an aberration but a structural feature of parole-eligible sentences: they are explicitly indeterminate.26See Bell, supra note 18, at 459, 507 (analyzing 426 parole-release hearings in California and describing how time served on indeterminate sentences varies by multiple decades among individuals convicted of similar crimes).

Although indeterminate sentences are currently taken for granted as an established feature of the carceral landscape, they were subject to a firestorm of constitutional litigation when they were introduced in the late 1800s.27See Edward Lindsey, Historical Sketch of the Indeterminate Sentence and Parole System, 16 J. Am. Inst. Crim. L. & Criminology 9, 40–41 (1925). Many state courts that legally analyzed indeterminate sentences saw them as a deep threat to the basic principle that crimes and punishments should be defined by clear rules, not left to individual discretion. State courts in Michigan, Tennessee, Texas, Utah, Oklahoma, and Vermont struck early indeterminate sentencing statutes.28See People v. Cummings, 50 N.W. 310, 314 (Mich. 1891); Fite v. State ex rel. Snider, 88 S.W. 941, 944 (Tenn. 1905); Ex parte Marshall, 161 S.W. 112, 113–14 (Tex. Crim. App. 1913); State ex rel. Bishop v. State Bd. of Corr., 52 P. 1090, 1092 (Utah 1898); Ex parte Ridley, 106 P. 549, 550–51 (Okla. Crim. App. 1910); In re Conditional Discharge of Convicts, 51 A. 10, 14–15 (Vt. 1901). State courts highlighted the contrast between indeterminate sentencing statutes and earlier credit-based statutes that set forth rules for reducing sentences by a specific number of days for each month of time served without prison misconduct. Legislatures included no such rules in indeterminate sentencing statutes, but instead left the period of incarceration to be determined by the discretion of the parole board. Courts saw a constitutional problem with delegating such broad discretionary power over punishment to a nonjudicial agency. They reasoned that such broad discretionary authority to end a sentence was an improper abdication of the legislature’s duty to make law about punishment and rendered indeterminate sentences “void for uncertainty.”29See infra notes 84–87 and accompanying text.

Public opinion in the early twentieth century overwhelmingly favored indeterminate sentencing.30See Gray Cavender & Michael C. Musheno, The Adoption and Implementation of Determinate-Based Sanctioning Policies: A Critical Perspective, 17 Ga. L. Rev. 425, 439 (1983). A public referendum in Michigan, for example, abrogated the Michigan Supreme Court’s holding that indeterminate sentences were unconstitutional.31See Marvin Zalman, The Rise and Fall of the Indeterminate Sentence, 24 Wayne L. Rev. 45, 60 (1977); see also Cummings, 50 N.W. 310. Coinciding with strong public opinion in favor of indeterminate sentencing, most state courts developed a clever jurisprudence of parole that effectively insulated indeterminate sentences from constitutional challenge.

The central principle of that jurisprudence is a legal fiction that this Article calls the “separate-parole principle”: an indeterminate sentence is defined under law as the maximum term regardless of any possibility of parole. Courts deemed release on parole as equivalent to a change in prison conditions that brought about no legal change to the underlying sentence. By placing parole apart from the legal definition of indeterminate sentences, the uncertainty and broad discretion that accompanied parole was also placed apart from indeterminate sentences.32See Ex parte Lee, 171 P. 958, 959 (Cal. 1918) (“It has uniformly been held that the indeterminate sentence is in legal effect a sentence for the maximum term. It is on this basis that such sentences have been held to be certain and definite, and therefore not void for uncertainty.”).

In a line of cases beginning in the 1980s, the U.S. Supreme Court contravened the separate-parole principle in favor of what this Article calls the “integrated-parole principle”: the presence (or absence) of the possibility of parole is an integral part of the legal definition of the sentence. The Court has repeatedly stated that the possibility of parole is significant enough to transform the constitutionality of a sentence: a sentence that is unconstitutionally excessive can be cured by adding the possibility of parole.33See Rummel v. Estelle, 445 U.S. 263, 265–66 (1980); Solem v. Helm, 463 U.S. 277, 300–03 (1983); Ewing v. California, 538 U.S. 11, 30–31 (2003); Graham v. Florida, 560 U.S. 48, 82 (2010); Miller v. Alabama, 567 U.S. 460, 479–80 (2012); Montgomery v. Louisiana, 577 U.S. 190, 212 (2016); see also infra Section II.A. In an example of what state constitutional scholars call the “lockstep” approach, almost all state supreme courts have followed the U.S. Supreme Court in adopting the integrated-parole principle.34Cf. Gardner, supra note 10, at 788 (describing how most states interpret their state constitutional provisions in “lockstep” with the U.S. Supreme Court); see infra Section II.B (discussing how nearly all states have followed the U.S. Supreme Court in adopting the integrated-parole principle in state proportionality analyses). The move has been largely unwitting; state supreme courts generally have not set forth an explicit rationale for moving to the integrated-parole principle nor cited their prior precedents that had rejected it.

Nevertheless, this Article argues that the move to the integrated-parole principle is a step forward because it discards a formal legal fiction in favor of a more realistic and functional understanding of indeterminate sentencing. The problem, however, is that adoption of the integrated-parole principle leaves indeterminate sentences open to the forceful state constitutional arguments brought against them over a century ago. Insofar as parole is an integral part of the sentence, it leaves the sentence itself up to nonjudicial discretion and renders it void for uncertainty.

Contemporary scholars have recognized that parole boards are now exercising sentencing power, and they have proposed several theories under which “sentencing by parole board” could violate the U.S. Constitution. David Ball, for example, argues that because parole boards make sentencing decisions that rely on facts about the crime without the presence of a jury, they violate the Sixth Amendment right to a jury trial.35See W. David Ball, Heinous, Atrocious, and Cruel: Apprendi, Indeterminate Sentencing, and the Meaning of Punishment, 109 Colum. L. Rev. 893, 902–03, 932–33 (2009). Alexandra Harrington and others argue that inadequate judicial review and procedural protections at parole hearings violate the Eighth Amendment and the Due Process Clause when applied to individuals serving life sentences for juvenile offenses.36See, e.g., Alexandra Harrington, The Constitutionalization of Parole: Fulfilling the Promise of Meaningful Review, 106 Cornell L. Rev. 1173, 1204–05, 1220 (2021); Sarah French Russell & Tracy L. Denholtz, Procedures for Proportionate Sentences: The Next Wave of Eighth Amendment Noncapital Litigation, 48 Conn. L. Rev. 1121, 1131–32 (2016). Mae C. Quinn argues that there is a penumbral right to sentencing by a judge or jury, and parole boards are constitutionally incapable of respecting that right.37See Quinn, supra note 21, at 569–70. This Article supports these arguments but suggests that they have not yet gone far enough in diagnosing the problem with parole boards exercising sentencing power. Insofar as sentences end by way of a discretionary judgment about open-ended criteria, they violate the state constitutional doctrines of vagueness. Just as clear rules are needed to define a conviction that starts a sentence, so too clear rules are needed to mark the end of a sentence.

Despite the strength of state constitutional arguments against indeterminate sentencing, this Article does not call for abolition of indeterminate sentencing. Instead, the Article recommends that state legislatures should “legalize” parole-release decision-making. The Article does not advocate for a specific model statute but argues for a jurisprudential principle by which state courts can assess whether revised parole statutes are void for uncertainty. The principle is that parole statutes must, at a minimum, be specific enough to provide fair notice about what a person needs to do (or not do) during the prison term to earn an enforceable right to release on parole. As seen in the Appendix to this Article, almost all existing parole statutes give parole boards discretion to grant release based on widely open-ended criteria, such as “the welfare of society.”38See infra Appendix, which cites statutes in Alabama, Colorado, Connecticut, Georgia, Maryland, Massachusetts, and New York. A handful of existing statutes do, however, set forth specific rules for parole release in regard to at least some subset of people serving indeterminate sentences.39See Reitz, Rhine, Lukac & Griffith, supra note 15, at 49–51. The Article highlights these statutes as a helpful starting point for state legislatures going forward.

This Article does not claim that compliance with a state’s constitutional doctrine of vagueness is the only, or even the most, pressing reason to “legalize” parole. Several policy reasons stand in favor of adopting parole statutes that provide fair notice about requirements to earn an enforceable right to parole. For example, empirical research shows that rules restricting discretion have the potential to reduce the impact of racial bias and other biases on decision-making.40See, e.g., Erik J. Girvan, Wise Restraints?: Learning Legal Rules, Not Standards, Reduces the Effects of Stereotypes in Legal Decision-Making, 22 Psych. Pub. Pol’y & L. 31, 33 (2016). In other work, I have argued that respect for the agency and dignity of incarcerated people calls for rules about what is required to earn early release on parole.41See Kristen Bell, Toward a Normative Theory of Parole Grounded in Agency, 31 Phil. Issues 24, 35 (2021). This Article does not include a discussion of such policy rationales, but instead focuses on legal theory. The central claim of the Article is that, in addition to whatever policy reasons there are to reform parole statutes, state constitutional law itself provides a reason for such reform.

This Article is divided as follows: Part I describes the constitutional litigation against indeterminate sentences that began in the late nineteenth century. That litigation led some courts to strike indeterminate sentences as unconstitutional and others to insulate indeterminate sentences from constitutional threat by adopting what this Article refers to as the “separate-parole principle.” Part II describes how the U.S. Supreme Court abandoned the separate-parole principle in favor of the integrated-parole principle, and how state supreme courts largely followed the Court in doing so under their state constitutions. Part III argues that in light of the abandonment of the separate-parole principle, indeterminate sentences are void for uncertainty under most state constitutions that lack amendments explicitly allowing indeterminate sentencing. This Part responds to two strong objections to the argument: first, that the argument fails to duly recognize differences in legal contexts; and second, that the argument logically extends to the erroneous claim that judicial discretion in sentencing is unconstitutional. Part IV contrasts the historic state vagueness doctrine with the contemporary federal vagueness doctrine. Part V describes the Article’s proposal for a path forward, recommending that legislatures “legalize” parole-release statutes.

A note on terminology and context before proceeding: For purposes of this Article, the term “indeterminate sentence” refers to any carceral punishment that includes an opportunity for an administrative agency (e.g., parole board) to grant conditional liberty (e.g., release on parole) at its discretion. A person granted release on parole is given conditions they must follow, and if a parole officer decides that the individual has violated conditions, the parole board can return the person to prison. A “determinate sentence” refers to any carceral punishment that does not allow for the possibility of discretionary release through a parole board. A determinate sentence may include a period of conditional liberty in the community that is often called “supervised release” and resembles being on parole.42See Fiona Doherty, Indeterminate Sentencing Returns: The Invention of Supervised Release, 88 N.Y.U. L. Rev. 958, 997 (2013). What makes an indeterminate sentence different from a determinate sentence is not whether it is followed by a period of conditional liberty. Rather, the distinct feature of the indeterminate sentence is that early release from prison is possible through the discretionary decision of a parole board.

Notably, some contemporary commentators use the term “indeterminacy” more broadly to describe sentences that allow for early release though any mechanism at the back-end of a sentence.43See Reitz, Rhine, Lukac & Griffith, supra note 15, at 10. The most common back-end release mechanisms are discretionary parole systems and credit systems.44Id. at 14. This Article uses the phrase “credit system” to refer to a statutory scheme under which specified units of time are deducted from the total sentence for each program a person completes and/or for each unit of time that a person serves without disciplinary infractions.45Id. Clemency provides another form of back-end release mechanism. The Article uses the term “clemency” to refer to the exclusive constitutional power of a state governor or the U.S. President to grant pardons and commutations.46Id.

As the next Part describes, parole-release systems were designed to share features with both credit systems and clemency. Parole release is akin to clemency in that both operate by using wide discretion. But unlike clemency, and more akin to credit systems, parole systems are creatures of statute that are operated by an administrative agency with some degree of independence from the executive office.

I. The Historic State Court Jurisprudence of Parole

Prior to the late nineteenth century, all carceral punishments were determinate sentences.47See Kimberly Thomas & Paul Reingold, From Grace to Grids: Rethinking Due Process Protection for Parole, 107 J. Crim. L. & Criminology 213, 216 (2017). The invention of indeterminate sentencing has its roots in nineteenth-century prison reformers who argued that the aim of punishment should be rehabilitation rather than retribution.48See id. at 216–17. Officials operating prisons in Australia, Ireland, and New York pioneered a new approach to punishment that created a system of rewards for good behavior in prison.49See Doherty, supra note 42, at 964–84. The system of rewards gradually increased the degree of liberty that a person had in prison, and the ultimate reward was early release into the community.50See id. at 978, 980. These models were met with increasingly glowing reports in the late nineteenth and early twentieth centuries; consensus developed that traditional sentences defined by “mere lapse of time” should be replaced with sentences measured by “satisfactory proof of reformation.”51See id. at 980, which quotes the Declaration of Principles that were adopted at an international prison conference in 1870.

Legislatures began passing statutes that created credit systems in the mid-1800s, followed by statutes creating discretionary parole-release systems in the late 1800s.52See Lindsey, supra note 27, at 10. From 1877 to 1922, the federal government and all but four states adopted some form of indeterminate sentencing laws that authorized release on parole after some minimum period of imprisonment.53See id. at 21, 69. By 1939, every state had some form of parole system.54See 4 U.S. Dep’t of Just., The Attorney General’s Survey of Release Procedures: Parole, at VII (1939). Proponents of indeterminate sentencing argued that discretion was essential to parole-release decisions because, in theory, these sentences were tailored to each individual’s unique progress toward rehabilitation.55See Thomas & Reingold, supra note 47, at 218–19; Zalman, supra note 31, at 76; Michele Pifferi, Individualization of Punishment and the Rule of Law: Reshaping Legality in the United States and Europe Between the 19th and the 20th Century, 52 Am. J. Legal Hist. 325, 337–38 (2012); Sanford H. Kadish, Legal Norm and Discretion in the Police and Sentencing Processes, 75 Harv. L. Rev. 904, 926 (1962). From the perspective of state courts, however, parole-release discretion threatened to make indeterminate sentencing statutes unconstitutional.

A. Early Indeterminate Sentencing Statutes Held Unconstitutional

In the 1886 case of People v. Moore, the Michigan Supreme Court became the first court to strike a statute that resembled a modern indeterminate sentencing statute.56See People v. Moore, 29 N.W. 80, 81, 84 (Mich. 1886). The statute did not reference anything called “parole” but instead referred to “conditional pardons.”57See id. at 81. Traditionally, a conditional pardon occurred when the governor granted a person’s release from prison prior to the end of their sentence and retained authority to return them to prison if conditions of the pardon were violated.58See, e.g., id. The Michigan statute at issue in Moore altered this traditional process by giving a prison board (rather than the governor) the authority to return a person to prison based on an alleged violation of a condition.59Id.

The Michigan Supreme Court struck that statute as unconstitutional on the ground that it violated due process of law.60Id. at 82–83. The court reasoned that unless the governor personally revokes the pardon, the pardoned individual is a free citizen who cannot be arrested without a warrant, nor detained without bail, nor punished without a conviction.61See id. at 81–82. The strong language of the opinion, written one year after the enactment of the Thirteenth Amendment, stated that allowing prison staff to revoke conditional pardons was tantamount to allowing a form of partial slavery.62Id. at 81. See generally U.S. Const. amend. XIII. The court described a person who was released on a conditional pardon as “half free and half slave[,] . . . let out with a rope around his body, as it were, with one end in the hands of the warden, to be hauled back at the caprice of that officer.”63Moore, 29 N.W. at 81.

Following Moore, the Michigan legislature passed a statute that allowed the prison board to grant and revoke release from prison in the form of “parole” rather than a conditional pardon from the governor.64See People v. Cummings, 50 N.W. 310, 310–11 (Mich. 1891). In 1891, the Michigan Supreme Court struck this statute as unconstitutional in People v. Cummings.65See id. at 314. The court cited the language in Moore, and described a person who was incarcerated and awaiting release on parole as a “slave of the prison board.”66Id. at 312–13. The problem was that a convicted person’s liberty was granted or revoked not on the basis of rules but by the whim of the board.67Id. at 312. The court emphasized that the convicted person had no enforceable rights against the board; the board’s “discretion is not reviewable by the courts; it is arbitrary.”68Id.

The court’s reasoning is rooted in what scholars call the “principle of legality”: there can be no crime without law, and no punishment without law—nullum crimen sine lege, nulla poena sine lege.69See generally Pifferi, supra note 55, at 326–27. Often referred to as the “first principle” of criminal law, it requires crimes and punishments to be defined by established, clear rules.70See Joshua Dressler & Stephen P. Garvey, Criminal Law: Cases and Materials 91 (8th ed. 2019). A crime must be defined in terms of specific acts, for example, “intentional infliction of great bodily harm,” rather than indefinite categories, such as “injurious to public morals.”71See generally id. at 105. So too a punishment must be specifically defined, for example, “two years imprisonment,” rather than left open-ended, such as “punishment for however long the common good requires.”72See, e.g., People ex rel. Hinckley v. Pirfenbrink, 96 Ill. 68, 70–71 (1879) (holding that it was illegal to imprison a person for contempt if the contempt order did not state what the person needed to do to be released or how long he would be detained). The principle that punishment must be specific did not prohibit judges or juries from exercising discretion in selecting a sentence in a given case.73See Apprendi v. New Jersey, 530 U.S. 466, 481 (2000). The legislature could fix a specific punishment for a crime or fix a broad range of punishment and give a court discretion to impose a sentence within that range.74See, e.g., id. at 519–20 (Thomas, J., concurring). Either way, the principle of legality required that the punishment imposed had to itself be clearly defined.75See Pirfenbrink, 96 Ill. at 70 (“All judgments must be specific and certain. They must determine the rights recovered or the penalties imposed.”); People v. Combs, 26 N.E.2d 668, 669 (Ill. App. Ct. 1940) (“It is a well settled rule in criminal law that a sentence must be certain, definite and consistent in all its terms and should not be ambiguous.”).

A punishment that is defined to end at the discretion of a parole board lacks clear definition: it is indeterminate. As the Michigan Supreme Court explained in Cummings, a person who had been convicted and sentenced to an indeterminate term had no way of knowing how long the punishment in prison would actually be.76People v. Cummings, 50 N.W. 310, 312 (Mich. 1891). If the sentence was fifteen years with the possibility of parole after one year, the sentence could be one year or fifteen years.77Id. The sentence was not defined by rules but left to the prerogative of the prison board.78Id. The court reasoned that if the legislature had intended to embed entirely discretionary decisions into the very definition of a sentence, it was effectively abdicating its obligation to make law about punishment.79See id.

The court recognized that there was an alternative way to interpret the indeterminate sentencing statute.80See id. at 311. Perhaps the statute did not embed the discretionary possibility of parole into the definition of the sentence but instead authorized the board to release people prior to the end of a sentence.81Id. at 312–13. If so, however, the court reasoned that the board would be exercising the pardon power; there was no other way to release a person prior to the end of a legally valid sentence.82Id. at 311. The statute would, therefore, be delegating to the board the pardon power and would be unconstitutional because the pardon power belonged exclusively to the governor.83Id. at 311–12. Thus, the court struck the indeterminate sentencing statute as unconstitutional because it either (a) defined the sentence to include the possibility of parole and therefore constituted an abdication of the legislature’s duty to make law that set punishment, or (b) did not change the definition of the sentence but authorized early release in a way that delegated clemency power to an entity other than the governor.84See id. at 311.

In so holding, the court distinguished between release on parole and release through a credit system.85See, e.g., id. at 314–15. The Michigan Supreme Court followed the reasoning of the Massachusetts Supreme Judicial Court, which had held that the possibility of reduced prison time through accrual of credits was built into the definition of a sentence.86Id. (citing In re Op. of the Justices, 79 Mass. (13 Gray) 618 (1859)). Accrual of credits reduces the sentence itself; release upon accrual of credits is, therefore, not release before the end of the sentence (not “early release”), and so it did not infringe on the pardon power.87See id. This reasoning was subsequently repeated by many state supreme courts.88See, e.g., Fite v. State ex rel. Snider, 88 S.W. 941, 943 (Tenn. 1905) (“[A]n act of the Legislature specifically defining credits for the good conduct . . . becomes a part of the sentence, and inheres into the punishment assessed.”); In re Wadleigh, 23 P. 190, 191 (Cal. 1890) (stating that a provision for credits “enters into and becomes a part of the judgment of the court below”); Ex parte Ridley, 106 P. 549, 555 (Okla. Crim. App. 1910) (“[A]n act of the Legislature specifically defining credits for good behavior . . . becomes a part of the sentence and inheres into the punishment assessed . . . .”). The same reasoning, however, was not applied to parole because parole was wholly discretionary rather than governed by clear rules. If the possibility of release on parole was part of the legal definition of the sentence at the outset, the sentence would be void for uncertainty.

The Tennessee Supreme Court likewise drew a stark distinction between credit statutes that set forth specific formulas for time reduction and a statute that provided for early release on a discretionary basis. The court struck an 1891 statute that allowed the board of workhouse commissioners to “deduct, for good conduct, a portion of the time for which any person has been sentenced.”89Fite, 88 S.W. at 941. The court identified that the constitutional infirmity of the statute was that the legislature did not specify a fixed scale for credit reductions but instead left reductions to the discretion of the agency.90Id. at 944. The court concluded that the statute was “clearly void as a delegation of legislative authority.”91Id. It reasoned that the commissioners had no authority to release people based on discretionary judgments; only the governor could grant discretionary early release through the pardon power.92Id. The court upheld statutes that provided for clear rules in the definition of good-time credits, explaining that these clearly defined credit reductions “inhere[] into” the punishment itself.93Id. at 943. The court later upheld an indeterminate sentencing statute that allowed for discretionary release on parole, but that statute required the governor’s involvement in approving each grant of parole.94See Woods v. State, 169 S.W. 558, 561–62 (Tenn. 1914).

The highest criminal court in Texas struck an indeterminate sentencing statute as void for uncertainty in 1913.95See Ex parte Marshall, 161 S.W. 112, 114 (Tex. Crim. App. 1913). The statute at issue required that, upon conviction, a judge was to impose a statutory minimum fine of $500 and a statutory maximum term of five years in prison.96Id. The person was eligible for release on parole after having served the minimum and prior to having served the entire period.97Id. The court reasoned that the sentencing law rendered the punishment too indefinite: “one cannot understand what punishment is to be meted out to persons found guilty of the offenses.”98Id. Interestingly, the court did not strike the parole-eligible sentencing statute on explicitly constitutional grounds.99See id. It relied on a foundational provision of the state penal code, which expressed what would now be recognized as the void-for-vagueness doctrine:

Whenever it appears that a provision of the penal law is so indefinitely framed or of such doubtful construction that it cannot be understood, either from the language in which it is expressed, or from some other written law of the state, such penal law shall be regarded as wholly inoperative.100Id. (quoting Tex. Penal Code art. 6 (1856)).

The highest courts of Michigan, Tennessee, and Texas saw the same basic problem: a statute that defined a sentence as including the possibility of discretionary release was in direct conflict with the principle that sentences had to be defined by clear rules. The Michigan court’s opinion was frequently cited for its reasoning in this regard, particularly in states that subsequently struck their parole statutes on the ground that the statute infringed on the pardon power.101See, e.g., Ex parte Ridley, 106 P. 549, 551 (Okla. Crim. App. 1910); In re Conditional Discharge of Convicts, 51 A. 10, 13–14 (Vt. 1901); Fite v. State ex rel. Snider, 88 S.W. 941, 942 (Tenn. 1905). Nevertheless, a legalized parole system did develop in Michigan. Public support for parole-eligible sentencing was strong, and eleven years after the Michigan Supreme Court struck the parole statute, voters ratified a referendum to amend the constitution to allow for indeterminate sentences.102See Zalman, supra note 31, at 60. Several other states also adopted constitutional provisions to explicitly allow for the creation of a parole board with authority to grant discretionary release on parole.103See, e.g., Selsor v. Trammell, No. 01-CV-721, 2013 WL 6535700, at *6 (N.D. Okla. Nov. 22, 2013) (citing Okla. Const. art. 6, § 10); Ex parte Prout, 86 P. 275, 275 (Idaho 1906) (citing Idaho Const. art. 4, § 7); Cardisco v. Davis, 64 P.2d 216, 218 (Utah 1937) (citing Utah Const. art. 7, § 12).

A handful of states avoided the concerns articulated by the Michigan court because the governor directly exercised the power to grant, deny, and revoke parole.104See Fehl v. Martin, 64 P.2d 631, 632 (Or. 1937) (“[P]arole is a conditional pardon.”); Sheldon L. Messinger, John E. Berecochea, David Rauma & Richard A. Berk, The Foundations of Parole in California, 19 Law & Soc’y Rev. 69, 100 (1985); Fuller v. State, 26 So. 146, 147 (Ala. 1899); Woodward v. Murdock, 24 N.E. 1047, 1048 (Ind. 1890). Advisory boards were devised to help the governor by providing recommendations about granting and revoking parole, but the governor was ultimately the sole authority. In these states, “parole” was a conditional pardon. The governor’s pardon came with conditions, and if a person violated a condition, the governor could revoke the pardon.105See, e.g., Flavell’s Case, 8 Watts & Serg. 197, 199 (Pa. 1844) (“[T]he governor may annex to a pardon any condition whether precedent or subsequent not forbidden by law, and it lies upon the grantee to perform the condition. If he does not, in case of a condition precedent, the pardon does not take effect; in case of a condition subsequent, such as this before us, the pardon becomes null; and if the condition is not performed, the original sentence remains in full vigour . . . .”). Because the governor himself controlled the system, there was no threat that the pardon power would be infringed.106See Lindsey, supra note 27, at 38. Operating a system in this way had limitations, including making a great deal of work for the governor and regularly embroiling the governor in the politics of punishment.107See, e.g., Messinger, Berecochea, Rauma & Berk, supra note 104, at 100–02 (discussing how in California, governors moved for the establishment of parole boards to relieve themselves of the work, and political risk, entailed in reviewing growing numbers of clemency petitions). Most states that operated their parole systems through the governor’s executive pardon power later transitioned to systems in which the governor was not directly responsible for parole decisions.108See, e.g., 12 Timothy P. Wile, Pennsylvania Law of Probation & Parole § 1:4 (3d ed. 2021). Pennsylvania prisoners were granted parole through conditional pardons from the governor until the legislature enacted the Parole Law of 1941, which created the parole board and gave it exclusive authority to grant parole. Id.

Most states created parole boards that granted and revoked release on parole with some independence from the governor.109See Lindsey, supra note 27, at 49. Rather than strike these parole systems as unconstitutional and/or pass a constitutional amendment, most state courts adopted an intricate jurisprudence of parole that insulated indeterminate sentences from constitutional challenge.

B. The Emergent Separate-Parole Principle

The Supreme Court of Illinois was among the first courts to announce this jurisprudence when it upheld an indeterminate sentencing statute in the 1894 case of People ex rel. Bradley v. Illinois State Reformatory.110See People ex rel. Bradley v. Ill. State Reformatory, 36 N.E. 76, 78 (Ill. 1894); see also State ex rel. Att’y Gen. v. Peters, 4 N.E. 81 (Ohio 1885). The litigant in Bradley argued that, because the indeterminate sentence imposed on him was not for a specified time, both the sentence in his case and the statute that authorized it were void for uncertainty.111Bradley, 36 N.E. at 78. The court acknowledged the long-standing common law principle that “all judgments must be specific and certain, and must determine the rights recovered or the penalties imposed.”112Id. (quoting People ex rel. Hinckley v. Pirfenbrink, 96 Ill. 68, 70 (1879)). Nonetheless, the court held that an indeterminate sentence was not uncertain because the legal definition of the sentence was the maximum term, regardless of any possibility of early release on parole.113Id. The Massachusetts Supreme Judicial Court used similar reasoning, explaining that an indeterminate sentence of six years with the possibility of parole after three years was defined under law as a term of six years.114Oliver v. Oliver, 48 N.E. 843, 843 (Mass. 1897).

By defining the punishment as a clear point in time—the maximum period allowed by statute—courts effectively neutralized the threat that indeterminate sentences were void for uncertainty or an unconstitutional delegation of legislative authority.115See James M. Kerr, The Indeterminate-Sentence Law Unconstitutional, 55 Am. L. Rev. 722 (1921). They deemed the possibility of parole, with all the uncertainty that accompanied it, as separate from the sentence itself. Just as the possibility of a pardon did not make a sentence void for uncertainty, neither did the possibility of parole. At least sixteen courts across the country took this approach, including Illinois, New Jersey, Indiana, Iowa, New York, Kansas, Massachusetts, Tennessee, and Pennsylvania.116See People v. Joyce, 92 N.E. 607, 613 (Ill. 1910); Ex parte Marlow, 68 A. 171, 173 (N.J. 1907); Skelton v. State, 49 N.E. 901, 903 (Ind. 1898); State v. Perkins, 120 N.W. 62, 64 (Iowa 1909); People ex rel. Clark v. Warden of Sing Sing Prison, 78 N.Y.S. 907, 908–09 (Sup. Ct. 1902); State v. Tyree, 78 P. 525, 527 (Kan. 1904); Oliver, 48 N.E. at 844; Woods v. State, 169 S.W. 558, 562 (Tenn. 1914); Commonwealth v. Kalck, 87 A. 61, 63 (Pa. 1913). In 1918, the California Supreme Court stated that “[i]t has uniformly been held that the indeterminate sentence is in legal effect a sentence for the maximum term. It is on this basis that such sentences have been held to be certain and definite, and therefore not void for uncertainty.”117Ex parte Lee, 171 P. 958, 959 (Cal. 1918).

The view that an indeterminate sentence was equivalent to the maximum term is what this Article refers to as the “separate-parole principle.” The principle was invoked in several ways to uphold the constitutionality of indeterminate sentences. In addition to insulating indeterminate sentences from the argument that they were void for uncertainty, the separate-parole principle was also used to explain why indeterminate sentences did not infringe on the judiciary’s power to impose sentences.118See Lindsey, supra note 27, at 46 n.47 (collecting cases holding that parole statutes do not vest judicial power in the parole board). Litigants argued that insofar as the parole board has the power to determine the length of time people actually serve in prison, the board impinges on the judiciary’s core powers to set and impose sentences.119See, e.g., Woods, 169 S.W. at 558. Courts rejected these arguments by reasoning that because the “sentence” is deemed to be the maximum term imposed by law, the parole board does not have power over the length of the “sentence.”120See, e.g., Ex parte Lee, 171 P. at 959. The legislature sets the “sentence,” the judiciary imposes the “sentence,” and the parole board administers the “sentence.”121See id.

The problem then was how to distinguish parole release from release through clemency. Given that the legal definition of the sentence was the maximum term, by what authority could a parole board release a person prior to the end of that term? Traditionally, use of the clemency power was the only way to grant release to a person before the end of the sentence.122See Kerr, supra note 115, at 732. State constitutions gave clemency power to the governor and prohibited delegation of that power outside the office of the governor.123See id.

To distinguish the parole power from the clemency power, courts maintained that the distinguishing feature of clemency is that it extinguishes a conviction or otherwise exempts an individual from punishment (or some part of the punishment) under law.124See, e.g., United States v. Wilson, 32 U.S. 150, 160 (1833) (noting that the meaning of the pardon power flows from English tradition under which a pardon was an “act of grace” that “exempt[ed] the individual . . . from the punishment”). To explain how conditional commutations could exempt an individual from punishment but nevertheless allow reimprisonment if conditions were violated, courts turned to principles of contract law.125See, e.g., Ex parte Prout, 86 P. 275, 277 (Idaho 1906); Wilson, 32 U.S. at 161 (“A pardon is a deed, to the validity of which, delivery is essential, and delivery is not complete, without acceptance.”). A conditional commutation is a gift of sentence reduction that the governor offers to an imprisoned individual. If the imprisoned individual accepts the gift, the individual binds herself to the conditions and is granted liberty.126Wilson, 32 U.S. at 161; In re Conditional Discharge of Convicts, 51 A. 10, 12 (Vt. 1901) (explaining that when a prisoner accepts a pardon that includes conditions, “the prisoner voluntarily submits himself thereto by accepting the pardon, and he is bound thereby”). While at liberty, the person does not remain “sentenced” or “punished.” If the individual violates the conditions, the governor is empowered to cease extending the gift of reprieve and return the individual if any time remains on the initial sentence.127In re Conditional Discharge of Convicts, 51 A. at 12–13.

Courts reasoned that, unlike granting clemency, granting release on parole did not extinguish a sentence or any part of it.128See Lindsey, supra note 27, at 46–47. When a person was released on parole, the person was not exempt from punishment but rather continued to serve the very same sentence in a different geographic location.129See, e.g., State v. Duff, 122 N.W. 829, 830 (Iowa 1909); State ex rel. Att’y Gen. v. Peters, 734 N.E. 81, 87 (Ohio 1885). In its 1885 opinion upholding an indeterminate sentencing statute, the Ohio Supreme Court explained that the term “penitentiary” did not narrowly refer to some geographic place of confinement but to anywhere that the legislature designated for keeping criminals under the custody of prison officials.130Peters, 734 N.E. at 86. The court cited prior statutes that had defined county jails as extending to any quarry, road, or other place where “convicts may be advantageously employed without the walls of the prison.”131Id. (quoting Act of Feb. 11, 1832, § 12, 30 Ohio Local Law 294). The court held that because the legislature defined being on parole as remaining in custody, a parolee in the community remained in the “penitentiary.”132See id. at 86–88; see also George v. Lillard, 51 S.W. 793, 794 (Ky. 1899). A pardon exempts a person from the punishment or some part of it, but a paroled prisoner “remains under the sentence to which he has been condemned . . . . So, strictly speaking, it cannot be said there has been a change of punishment to a less severe one.” Id. Release from prison to parole was akin to an administrative matter like prison transfer rather than any type of reduction in punishment; as such, courts reasoned that it did not infringe on the pardon power.133See Kerr, supra note 115, at 733–34 (explaining this argument with case citations and then critiquing it).

Many other state supreme courts followed the Ohio Supreme Court’s reasoning in holding that the power to parole did not infringe on the pardon power because parolees were continuing to serve their sentences.134See, e.g., Commonwealth ex rel. Banks v. Cain, 28 A.2d 897, 901 (Pa. 1942) (“The parolee . . . serves the remainder of his sentence by having his liberty restrained in a manner analogous to that employed in the ‘trusty’ or ‘honor’ system of prison discipline.”). Courts that embraced this view of parole as serving the sentence in a different type of “penitentiary” were clear that parole boards lacked authority to discharge people from parole prior to the end of the full legal sentence.135See Lindsey, supra note 27, at 47 & n.48. Statutory provisions that allowed for early discharge from parole without approval from the governor or a court were generally struck as unconstitutional.136See id.; see also Bd. of Prison Comm’rs v. De Moss, 163 S.W. 183, 188 (Ky. 1914). Discharge from parole prior to the maximum term would mean actual reduction in the sentence imposed by law, and such reductions could be achieved only through exercise of the pardon power or through judicial resentencing.137Compare De Moss, 163 S.W. at 188 (striking a provision that allowed for early discharge from parole), with George v. People, 47 N.E. 741, 746 (Ill. 1897) (upholding a provision that allowed for early discharge from parole because it required approval from the governor).

Although accepted by many courts, the view of parole as penal custody in the community was also met with powerful criticism. The Michigan Supreme Court highlighted that the position entails the consequence that a person who is out on parole for over ten years is no different in the eyes of the law than a “trusty’ . . . who is allowed at times to be outside at work or to run errands.”138See People v. Cummings, 50 N.W. 310, 315 (Mich. 1891). This consequence seemed absurd to the Michigan Court; it evoked an image of parolees as slaves who could be hauled back to prison as if wardens held ropes around their waists.139 Id. at 313. The Supreme Court of Utah also rejected the view that parole release was a change in prison conditions. It reasoned that parole release reduced the sentence, and the indeterminate sentencing law therefore must be void because sentences could be reduced only through exercise of the pardon power.140See State ex rel. Bishop v. State Bd. of Corr., 52 P. 1090, 1092 (Utah 1898). A year later, in 1899, the Utah Constitution was adopted, and included a specific provision that created a board of pardons and gave it power to grant commutations.141See Cardisco v. Davis, 64 P.2d 216, 217–18 (Utah 1937) (citing Utah Const. art. 7, § 12).

The Vermont Supreme Court also criticized the view that parole release was simply a matter of “penal administration.”142In re Conditional Discharge of Convicts, 51 A. 10, 14 (Vt. 1901). The court called this understanding of parole a “guise” that failed to recognize the reality that release on parole was not functionally any different than release on a conditional commutation.143Id. The practical reality was the same regardless of whether a person was released on a conditional pardon or on parole: the person was released at someone’s discretion prior to serving the full period of imprisonment, the release was subject to conditions, and the person could be returned to prison upon violation of the conditions. The Vermont court concluded that conditional pardon and parole differed in name only; because the pardon power was exclusive to the governor, so too the parole power was exclusive to the governor.144See id. at 14–15. In a 1921 American Law Review article, James Kerr put forward the same argument, concluding that indeterminate sentences are unconstitutional, and no amount of “phraseology” by the courts could save them.145Kerr, supra note 115, at 732–33.

C. Continued and Cross-Substantive Reliance on the Separate-Parole Principle

Indeterminate sentencing became the predominant form of sentencing in the twentieth century.146See Kate Stith & Steve Y. Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake Forest L. Rev. 223, 227 (1993) (describing how indeterminate sentencing became “firmly entrenched” in state and federal jurisdictions before World War I and continued through the late twentieth century). By the 1970s, parole boards exercised authority to release over 70% of incarcerated people across the country.147Thomas & Reingold, supra note 47, at 218. Most state supreme courts consistently relied on the separate-parole principle to insulate indeterminate sentences from constitutional challenge and marshaled a strict distinction between sentencing and parole for nearly one hundred years.148See supra Section I.B. Through the 1970s, state courts continued to hold that indeterminate sentences must be defined by law as the maximum term, lest they be void for uncertainty.149See, e.g., Woodward v. Murdock, 24 N.E. 1047, 1048 (Ind. 1890); Fuller v. State, 26 So. 146, 147 (Ala. 1899); Fehl v. Martin, 64 P.2d 631, 632–33 (Or. 1937); Commonwealth v. Logan, 327 N.E.2d 705, 706 (Mass. 1975); In re Lynch, 503 P.2d 921, 925–26 (Cal. 1972); State v. Deats, 489 P.2d 662, 664 (N.M. Ct. App. 1971). Decisions to grant, deny, or revoke parole were not sentencing decisions but administrative decisions about the conditions of custody.150See, e.g., Kadish, supra note 55, at 920. Courts frequently used language that signaled parole’s distance from the strictures of law.151See, e.g., id. From the late 1800s through the 1970s, parole was characterized as “the granting of a favor,”152Ughbanks v. Armstrong, 208 U.S. 481, 487–88 (1908). a chance for “executive grace,”153See, e.g., State v. Fain, 617 P.2d 720, 724–25 (Wash. 1980); Orme v. Rogers, 260 P. 199, 202 (Ariz. 1927). and a “question of grace and mercy.”154Commonwealth v. Kalck, 87 A. 61, 64 (Pa. 1913) (citing Commonwealth ex rel. Bates v. McKenty, 52 Pa. Super. 332, 339 (1912)). In this sense, parole was closely akin to clemency,155See, e.g., Murphy v. Commonwealth, 52 N.E. 505, 509–10 (Mass. 1899) (“It is as correct . . . to say that the duration of his sentence is uncertain because the governor may pardon him . . . as it is to say that it is uncertain because . . . the [parole] commissioners may release him . . . .”). the only difference being that clemency was a gubernatorial reprieve from punishment whereas parole was a transfer to a different kind of “penitentiary.”

The development of this jurisprudence was likely influenced by the strong public support for indeterminate sentencing in the early to mid-twentieth century. While the opinions upholding the constitutionality of parole included legal reasoning, they were also infused with excitement and promise about rehabilitative reforms.156Cavender & Musheno, supra note 26, at 439. As renowned criminal law scholar Sanford Kadish put it in 1962, courts shaped parole into a “prerogative of mercy” not because doing so was sound legal theory,157See Kadish, supra note 55, at 920. but rather because “the grace conception [of parole] provided a ready rationale for upholding indeterminate sentence laws against constitutional attack as void for uncertainty, and as prohibited delegations of the power to punish.”158Id.

Regardless of why courts adopted this jurisprudence of parole, state court caselaw shows remarkable consistency in adhering to it. In addition to applying the separate-parole principle in answering questions of statutory interpretation,159See, e.g., Oliver v. Oliver, 48 N.E. 843, 843–44 (Mass. 1897) (deeming an indeterminate sentence equivalent to the maximum term when interpreting a statute that provided grounds for divorce if a spouse had been sentenced to “five years or more”). courts used it in applying the ex post facto doctrine and state constitutional prohibitions against cruel and unusual punishment. The next two Sections describe this cross-substantive application of the separate-parole principle.

1. Ex Post Facto Context

The ex post facto clause of federal and state constitutions prohibits, among other things, sentencing a person under a law that increases the amount of punishment that would have been permitted at the time of the crime.160See Calder v. Bull, 3 U.S. 386, 390 (1798). A law violates the ex post facto clause if it falls into one of four categories: laws that criminalize previously innocent conduct, laws that expand the definition of a previously existing crime, laws that inflict greater punishment for the same crime, and laws that change rules of evidence for conviction. Id. Like the void for uncertainty doctrine, the ex post facto doctrine is undergirded by the principle of nullum crimen sine lege, nulla poena sine lege.161See Jerome Hall, Nulla Poena Sine Lege, 47 Yale L.J. 165, 170–71 (1937). Laws must be both certain and stable over time in order to provide fair notice about crimes and punishments and restrict officials’ discretion to expand the definitions of crimes and punishments.162See, e.g., Murphy v. Commonwealth, 52 N.E. 505, 507 (Mass. 1899) (explaining the “uncertainty” that ex post facto laws introduce, as well as “the injustice of punishing an act which was not punishable when done, or of punishing it in a different manner from that in which it was punishable when done”).

In Commonwealth v. Kalck, the Pennsylvania Supreme Court considered the argument that a judge violated the ex post facto clause when imposing an indeterminate term of twenty years with the possibility of parole after fifteen years.163Commonwealth v. Kalck, 87 A. 61, 62 (Pa. 1913). When the defendant had committed the crime (second-degree murder), the law stated that the punishment applicable to that crime allowed for the possibility of parole after no more than five years.164Id. at 62–63. After the crime and prior to sentencing, the legislature enacted a statute that gave the sentencing judge discretion to set any period of time as the minimum time served prior to parole-release eligibility.165Id. at 63. The defendant argued that the sentence violated the ex post facto clause because the punishment had increased; the judge had imposed a fifteen-year minimum prior to parole-release eligibility rather than only five years.166Id. The court held that there was no ex post facto violation based on the separate-parole principle: an indeterminate sentence was legally equivalent to the maximum term, and in this case, the maximum term remained fixed at twenty years.167Id. The court cited seven other state supreme courts for the separate-parole principle and explained that the change in the minimum period prior to eligibility for parole release was not a change in punishment but instead a change in “penal administration” or “prison discipline.”168Id. at 63–64. The court cited with approval the reasoning of a lower court in Pennsylvania, which had reasoned that in order for the indeterminate sentencing statute to be constitutional, “it necessarily follows that the maximum sentence is the only portion of the sentence which has legal validity, and that the minimum sentence is merely an administrative notice.”169Id. at 64 (citing Commonwealth ex rel. Bates v. McKenty, 52 Pa. Super. 332, 339 (1912)).

In Murphy v. Commonwealth, the Massachusetts Supreme Judicial Court considered the argument of a defendant who was sentenced to fifteen years with the possibility of discretionary release on parole after ten years.170Murphy v. Commonwealth, 52 N.E. 505, 506, 509 (Mass. 1899). The statute under which the defendant was sentenced did not allow for early release based on accrual of credits.171Id. at 507. Under the statutes that had been in effect at the time of the defendant’s crime, the maximum punishment had been the same—fifteen years—but instead of having an opportunity for parole, a person could reduce their sentence through accrual of good-time credits.172Id. at 509. If a person serving a fifteen-year sentence earned all possible credits, they would be entitled to a three-year sentence deduction and conditional release after twelve years.173Id. The court relied on the distinction between the credit system, which created a “right” to release upon accrual of credits, and the parole system, which created only the “favor” of discretionary release.174See id. Whereas parole was separate from the legal definition of the sentence given its discretionary nature, the Massachusetts Supreme Judicial Court had previously made clear that credits were integrated into the sentence: credits for good behavior were “‘intended to be an actual reduction of sentence as a right, and not as a favor,’ and ‘therefore operated upon the sentence itself.’”175Id. at 508 (quoting In re Op. of the Justices, 79 Mass. (13 Gray) 618, 620 (1859)). The court therefore held that applying the law that allowed for the possibility of parole but removed the opportunity to earn credits violated the ex post facto clause.176Id. at 510.

The Supreme Court of Kansas and the Supreme Court of California relied on similar reasoning when each respectively held that the ex post facto clause is violated by retroactive application of a law that created the possibility for discretionary release on parole but removed the opportunity to earn a reduction in sentence through accrual of credits.177See State v. Tyree, 78 P. 525, 526–27 (Kan. 1904); Ex parte Lee, 171 P. 958, 960 (Cal. 1918).

2. Cruel and/or Unusual Punishment Context

State supreme courts also relied on the separate-parole principle in analyzing whether indeterminate sentences violated clauses in state constitutions that prohibit cruel and/or unusual punishment. The phrase “cruel and/or unusual punishment” is used here because some state constitutions prohibit “cruel and unusual punishment,” whereas others prohibit “cruel or unusual punishment.”178See Berry, supra note 2, at 1206. In applying these clauses to indeterminate sentences through the 1970s, state appellate courts were explicit in reasoning that an indeterminate sentence is equivalent to the maximum term.179See Miller v. State, 49 N.E. 894, 896–97 (Ind. 1898); People ex rel. Bradley v. Ill. State Reformatory, 36 N.E. 76, 78 (Ill. 1894); People v. Joyce, 92 N.E. 607, 611–12 (Ill. 1910); Commonwealth v. Brown, 45 N.E. 1, 2 (Mass. 1896); State v. Newman, 152 S.E. 195, 197 (W. Va. 1930); Johnson v. State, 218 S.W.2d 687, 689 (Ark. 1949); Sims v. Balkcom, 136 S.E.2d 766, 768–69 (Ga. 1964); State v. Peters, 430 P.2d 382, 383–85 (N.M. 1967); Washington v. Rodriguez, 483 P.2d 309, 311 (N.M. Ct. App. 1971); State v. Freeman, 574 P.2d 950, 958 (Kan. 1978); Schmidt v. State, 584 P.2d 695, 697 (Nev. 1978); State v. Freitas, 602 P.2d 914, 921 (Haw. 1979); In re Lynch, 503 P.2d 921, 924–25 (Cal. 1972); State v. Kimbrough, 46 S.E.2d 273, 277 (S.C. 1948); Cason v. State, 23 S.W.2d 665, 667 (Tenn. 1930); State v. Fain, 617 P.2d 720, 725 (Wash. 1980).

In 1948, the Supreme Court of South Carolina struck a thirty-year sentence with the possibility of parole as unconstitutional when imposed on a defendant convicted of burglary.180See Kimbrough, 46 S.E.2d at 275–77. In doing so, the court dismissed the suggestion that the defendant’s sentence was mitigated by the likelihood that he would be released on parole prior to serving thirty years.181See id. at 277. Likewise, the Supreme Court of Alaska held in 1968 that a thirty-six-year sentence was unconstitutional for several counts of check fraud. The court explained that the fact that the defendant would be eligible for parole after five years did not mitigate the harshness of his sentence.182See Faulkner v. State, 445 P.2d 815, 816–17, 819 (Alaska 1968). The court noted that it was unknown whether the defendant would ever be granted parole, and a concurring opinion noted that the court should not abdicate its responsibility to review the constitutionality of sentences simply because a person is eligible for release on parole.183See id. at 819; id. at 823 (Rabinowitz, J., concurring).

In re Lynch, decided by the Supreme Court of California in 1972, appears to be the most recent case that provides an in-depth discussion of the separate-parole principle. The question presented in Lynch was whether an indeterminate sentence imposed for the crime of indecent exposure violated Article I, then-Section 6 (now-Section 17) of the California Constitution, which prohibited cruel or unusual punishment.184See In re Lynch, 503 P.2d at 922. The maximum term of the sentence was lifetime incarceration, and parole release was available after one year.185Id. at 927. The court cited the long-established rule that the definition of an indeterminate sentence under law is the maximum term,186Id. at 924–26. and it summarized the historic reasoning that, absent this rule, indeterminate sentences would be void for uncertainty and violate principles respecting the separation of powers.187Id. at 925–26. The court emphasized that the decision to grant parole is a discretionary action by an executive agency, and there is no right to release on parole prior to the end of the maximum term.188Id. at 925. “[A] defendant’s liability is to serve the maximum term, and he is therefore entitled to know that the maximum in his case is lawful.”189Id. After deciding to treat the indeterminate sentence as lifetime incarceration for the purpose of proportionality analysis, the Lynch court held that the punishment was unconstitutionally disproportionate to the crime of second-offense indecent exposure.190Id. at 940. The court articulated three factors by which to assess the proportionality of a sentence: the gravity of the offense in relation to the severity of the punishment, the extent to which more serious crimes are punished with less severe sentences within the jurisdiction, and the extent to which other jurisdictions punish the same crime with less severe punishments. Id. at 930–33. State supreme courts in Hawaii, Kansas, Utah, and Arizona each adopted the three-factor method in reliance on Lynch. See State v. Solomon, 111 P.3d 12, 26–27 (Haw. 2005); State v. McDaniel, 612 P.2d 1231, 1241–42 (Kan. 1980); State v. Gardner, 947 P.2d 630, 639–40 (Utah 1997); State v. Mulalley, 618 P.2d 586, 590 (Ariz. 1980). The approach in Lynch was praised by scholars at the time,191See Bruce W. Gilchrist, Note, Disproportionality in Sentences of Imprisonment, 79 Colum. L. Rev. 1119, 1125 & nn.26–27, 1131 (1979) (discussing “a rational model of proportionality,” which “requires that courts measure the severity of punishment by reference to the maximum length of the offender’s sentence”); Alan M. Dershowitz, Indeterminate Confinement: Letting the Therapy Fit the Harm, 123 U. Pa. L. Rev. 297, 328–29 (1974) (praising the “thoughtful and detailed opinion” in Lynch and its commitment to assessing proportionality based on the maximum authorized term). and its reasoning that indeterminate sentences are equivalent to the maximum term was cited by supreme courts in several states, including New York,192See People v. Broadie, 332 N.E.2d 338, 341 (N.Y. 1975); see also People v. Mosley, 358 N.Y.S.2d 1004, 1007–08 (Monroe Cnty. Ct. 1974) (“The Court will not sit idly by and justify an unconstitutional deprivation of rights on the ground that an administrative agency may later come along and correct or reduce the injustice. The sentence imposed on every defendant must pass constitutional scrutiny.”). Massachusetts,193See Op. of the Justices to the House of Representatives, 393 N.E.2d 313, 319, 320 n.10 (Mass. 1979) (explaining that the lack of eligibility for parole is not relevant in determining the proportionality of a sentence). and Washington.194See State v. Fain, 617 P.2d 720, 724–25 (Wash. 1980) (deeming a life sentence with the possibility of parole as equivalent to lifetime incarceration for the proportionality analysis because the possibility of parole is a chance for grace that is not legally enforceable).

II. The U.S. Supreme Court Starts a New Approach to Parole

Despite its dominance across state courts prior to 1980, the separate-parole principle is now rarely cited and appears all but forgotten in judicial opinions and scholarship. As this Part explains, the separate-parole principle’s disappearance coincides with the U.S. Supreme Court’s rejection of it, which began in Rummel v. Estelle and Solem v. Helm and was solidified in a subsequent series of cases including Harmelin v. Michigan, Ewing v. California, Graham v. Florida, Miller v. Alabama, and Montgomery v. Louisiana.195See Rummel v. Estelle, 445 U.S. 263, 280–81 (1980); Solem v. Helm, 463 U.S. 277, 297, 300–03 (1983); Harmelin v. Michigan, 501 U.S. 957, 996 (1991); Ewing v. California, 538 U.S. 11, 22 (2003); Graham v. Florida, 560 U.S. 48, 69–71 (2010); Miller v. Alabama, 567 U.S. 460, 465 (2012); Montgomery v. Louisiana, 577 U.S. 190, 212 (2016). In these cases, the Court replaced the separate-parole principle with what this Article refers to as the “integrated-parole principle”: that the presence or absence of the possibility of parole is an integral part of how the law defines the sentence. After describing these cases, this Part explains that state supreme courts have followed the U.S. Supreme Court in adopting the integrated-parole principle in their cruel and/or unusual punishment jurisprudence. This Part then demonstrates how the Court and various state supreme courts have also relied on the integrated-parole principle in their caselaw regarding the ex post facto clause and procedural due process.

A. U.S. Supreme Court Announces the Integrated-Parole Principle

The Eighth Amendment prohibits “cruel and unusual punishments”196U.S. Const. amend. VIII. and was rarely invoked by the U.S. Supreme Court until 1972.197See Berry, supra note 2, at 1203. Since that time, the Court has relied on the Eighth Amendment to place various limits on capital punishment and has generally upheld challenges to noncapital punishments.198See id. at 1203–04. Since 1972, it has published eight opinions on whether noncapital sentences violate the Eighth Amendment.199See Rummel, 445 U.S. 263; Hutto v. Davis, 454 U.S. 370 (1982) (per curiam); Solem, 463 U.S. 277; Harmelin, 501 U.S. 957; Ewing, 538 U.S. 11; Lockyer v. Andrade, 538 U.S. 63 (2003); Graham, 560 U.S. 48; Miller, 567 U.S. 460. Prior to 1970, the Court addressed only one noncapital proportionality claim. See Weems v. United States, 217 U.S. 349 (1910). The distinction between sentences with versus without the possibility of parole has made a remarkable difference in these opinions. In the three cases in which the Court held that a noncapital sentence violated the Eighth Amendment, the sentence did not include the possibility of release on parole.200See Solem, 463 U.S. at 281–82, 303 (holding that life without possibility of parole (LWOP) is unconstitutional when imposed on a recidivist defendant for using a “no account” check); Graham, 560 U.S. at 82 (holding that LWOP is unconstitutional when imposed on a juvenile for a nonhomicide offense); Miller, 567 U.S. at 489 (holding that LWOP is unconstitutional when imposed on a juvenile as a mandatory sentence). And in the five cases in which the Court upheld noncapital punishments, three of the sentences included the possibility of release on parole.201See Rummel, 445 U.S. at 265–67 (holding that life with possibility of parole (LWPP) is not unconstitutional when imposed for forgery on a recidivist defendant); Hutto, 454 U.S. at 371, 375 (holding that a forty-year sentence is not unconstitutional for distribution and possession with intent to distribute marijuana); Harmelin, 501 U.S. at 961, 994–96 (holding that LWOP is not unconstitutional for a first-offense possession of cocaine); Ewing, 538 U.S. at 28–31 (holding that LWPP is not unconstitutional for shoplifting when imposed on a recidivist defendant); Lockyer, 538 U.S. at 70, 77 (upholding a sentence of LWPP for theft of videotapes).

Rummel is the first case in which the Court stated that the possibility of parole mitigates the harshness of a sentence for the purpose of an Eighth Amendment analysis. Mr. Rummel argued that his sentence—life with the possibility of parole (LWPP)—was cruel and unusual as applied to his crime of obtaining $120.75 by false pretenses.202Rummel, 445 U.S. at 265–66. The State of Texas sentenced him to life under a habitual offender statute; Mr. Rummel had prior convictions for passing a forged check for $28.36 and using a fraudulent credit card to obtain $80.203Id. In upholding Mr. Rummel’s sentence, the Court highlighted that he would be eligible for release on parole after approximately twelve years.204Id. at 280. The Court explained that Mr. Rummel’s sentence should not be treated as equivalent to a twelve-year sentence, nor should it be treated as equivalent to an entire lifetime in prison.205Id. at 280–81. The possibility of parole, “however slim,” distinguished his sentence from the more severe punishment of life without the possibility of parole (LWOP).206Id. at 281.

In a dissenting opinion, Justice Powell argued that “the possibility of parole should not be considered in assessing the nature of the punishment.”207Id. at 286–87 (Powell, J., dissenting). Central to Justice Powell’s reasoning was the absence of any right to release on parole.208See id. at 294. He characterized the decision to grant release on parole as “an act of executive grace.”209Id. at 293. Justice Powell’s opinion about the possibility of parole changed three years later, however, when he wrote the majority opinion in Solem.

The defendant in Solem, Mr. Helm, was very similar to Mr. Rummel in that he had prior convictions for relatively minor crimes.210See Solem v. Helm, 463 U.S. 277, 279–80 (1983) (noting that the defendant had prior convictions for third-degree burglary, obtaining money under false pretenses, grand larceny, and driving while intoxicated). Mr. Helm was sentenced to LWOP under a South Dakota habitual offender statute for writing a “no account” check in the amount of $100.211Id. at 281–82. In holding that Mr. Helm’s sentence violated the Eighth Amendment, the Court stated that its opinion was not inconsistent with Rummel.212Id. at 303 & n.32. Writing for the Court, Justice Powell explained that Rummel was “clearly distinguishable” because the sentence at issue in Rummel was LWPP, whereas the sentence at issue in Solem was LWOP.213Id. His opinion characterized LWOP as a “far more severe” punishment than LWPP.214Id. at 297.

In holding that LWOP was disproportionate to Mr. Helm’s offense, the Court applied the same three-factor analysis that the California Supreme Court had developed in Lynch and which other state supreme courts had subsequently adopted.215See supra note 190 (stating the three-factor Lynch test and citing state supreme courts that subsequently applied it); Solem, 463 U.S. at 292, 295–300 (applying the same three factors without citation to any state supreme court). The opinion, however, does not cite Lynch or any other related state supreme court opinion. It is unclear why the Court copied the three-factor test from Lynch without citation, but a speculative explanation is that Lynch was widely recognized for its forceful restatement of the historic separate-parole principle. Solem abandoned that principle with very little argument.

The Court asserted that LWPP is distinctly less severe than LWOP because release on parole is “regular,” “the normal expectation,” and “possible to predict,” whereas “a [g]overnor may commute a sentence at any time for any reason without reference to any standards.”216Solem, 463 U.S. at 300–01. The opinion cites no parole statutes or rates of release to support the claim.217The only cited support is the transcript from oral argument, see id. at 302, where Justice Stevens said that “[p]robably quite a lot” of people were released on parole in South Dakota, to which South Dakota’s attorney replied, “There have.” Transcript of Oral Argument at 8–9, Solem, 463 U.S. 277 (No. 82-492). Ironically, South Dakota’s clemency system at issue in Solem released more people in practice than the Texas parole system at issue in Rummel. The dissent in Solem indicated that approximately 50% of people sentenced to LWOP prior to 1964 were granted release through South Dakota’s commutation system.218See Solem, 463 U.S. at 316–17 (Burger, C.J., dissenting). Justice Powell’s dissent in Rummel indicated that only 21% of people were granted release through the parole system in Texas.219See Rummel v. Estelle, 445 U.S. 263, 294 (Powell, J., dissenting).

Notably, no Justice in Solem took a position on the relative harshness of LWOP to LWPP that was consistent with the position they took in Rummel.220The Justices who participated in both opinions include Rehnquist, White, Burger, Powell, Brennan, Marshall, and Stevens. Whilst in the majority, the Justices trumpeted a difference between LWOP and LWPP; however, when the same Justices dissented, they argued it was a distinction without a difference. The reasoning may have less to do with maintaining a consistent jurisprudence about parole, and more to do with changes in the composition of the Court.221See Nancy Keir, Solem v. Helm: Extending Judicial Review Under the Cruel and Unusual Punishments Clause to Require ‘Proportionality’ of Prison Sentences, 33 Cath. U. L. Rev. 479, 510 (1984) (supporting the argument that the holding in Solem was “completely at odds with Rummel”). In the three years between Rummel and Solem, Justice Stewart retired (he was in the Rummel majority), and Justice Blackmun joined (he was in the Solem majority).222See id. at 494 n.113, 503 n.179.

Regardless of its ignoble origin, the idea that the possibility of parole mitigates the harshness of a sentence has become a consistent feature in the Court’s Eighth Amendment jurisprudence. In its 1991 opinion in Harmelin v. Michigan, the Court held that LWOP for a drug conviction did not violate the Eighth Amendment.223See Harmelin v. Michigan, 501 U.S. 957, 994 (1991). Writing for the Court, Justice Scalia stated that although the difference between LWOP and LWPP may be “negligible” in some cases, LWPP is categorically less severe than LWOP.224See id. at 996.

The distinction between LWOP and LWPP was central to the reasoning of both the plurality and dissent in Ewing v. California. The judgement of the Court in Ewing was that LWPP was not grossly disproportionate to the theft of three golf clubs under California’s three strikes statute.225See Ewing v. California, 538 U.S. 11, 30 (2003). In her plurality opinion, Justice O’Connor noted that the explicit distinction between Rummel and Solem was that Mr. Rummel’s sentence included the possibility of release on parole, whereas Mr. Helm’s did not.226Id. at 22. In his dissent in Ewing, Justice Breyer likewise picked out parole eligibility as the key distinction between Rummel and Solem.227Id. at 38 (Breyer, J., dissenting). In a companion case to Ewing, the Court emphasized that the possibility of parole was the distinguishing factor that made a life sentence constitutional as applied to a third-strike offense in California.228See Lockyer v. Andrade, 538 U.S. 63, 74 (2003).

The Court cemented its position that the possibility of parole mitigates the harshness of a sentence in its recent trilogy of cases on juvenile sentencing: Graham v. Florida, Miller v. Alabama, and Montgomery v. Louisiana. In Graham, the Court held that LWOP was unconstitutional when imposed on a juvenile for a nonhomicide offense.229Graham v. Florida, 560 U.S. 48, 82 (2010). Justice Kennedy, writing for the majority, emphasized the harshness of LWOP and explained that the possibility of parole is what distinguished Rummel from Solem.230Id. at 69–70. He cited Solem for the claim that “the remote possibility of [clemency] does not mitigate the harshness of the sentence”231Id. at 70 (citing Solem v. Helm, 463 U.S. 277, 300–01 (1983)). but did not cite the Solem dissent, which had explained that clemency had been granted frequently at the time.232See Solem, 463 U.S. at 316–17 (Burger, C.J., dissenting). Justice Kennedy also described LWOP as egregious because it “den[ies] the defendant the right to reenter the community.”233Graham, 560 U.S. at 74. He did not cite the Rummel dissent, which had explained that LWPP also deprives the defendant of any right to reenter the community.234See Rummel v. Estelle, 445 U.S. 263, 302 (1980) (Powell, J., dissenting).

The Court subsequently held, in Miller, that mandatory LWOP sentences are unconstitutional as applied to juveniles.235Miller v. Alabama, 567 U.S. 460, 465 (2012). The distinction between LWOP and LWPP was so cemented that the Court no longer accompanied it by any discussion or citation. In the third sentence of the majority opinion, Justice Kagan asserted that LWPP is less severe than LWOP.236Id. (“State law mandated that each juvenile die in prison even if a judge or jury would have thought . . . a lesser sentence (for example, life with the possibility of parole) more appropriate.” (emphasis added)). In Montgomery, the Court held that Miller applied retroactively to cases on collateral review.237See Montgomery v. Louisiana, 577 U.S. 190, 206 (2016). At the end of the opinion, Justice Kennedy wrote that states do not have to relitigate cases in which mandatory LWOP was imposed on a juvenile.238See id. at 212. Another option is to grant the possibility of parole in these cases.239Id. (“A State may remedy a Miller violation by permitting juvenile homicide offenders to be considered for parole, rather than by resentencing them.” (citing Wyo. Stat. Ann. § 6-10-301(c) (2013) (juvenile homicide offenders eligible for parole after 25 years))). There is no discussion of parole other than an assertion that being considered for parole “ensures” that juveniles who mature will not serve the rest of their lives in prison.240Id. This suggested assurance belies the fact that there is no right to be granted parole; parole-release decisions remain a matter of executive discretion.241See Greenholtz v. Inmates of Neb. Penal & Corr. Complex, 442 U.S. 1, 15 (1979).

The fact that the Court holds the possibility of parole to be such an integral part of the sentence is difficult to explain in light of the century-old precedent that consistently rejected this idea. What might explain such a stark shift in thinking about the relationship between sentences and the possibility of parole? One answer might be that the U.S. Supreme Court has simply not had its attention focused on the void for uncertainty issue when deciding its Eighth Amendment cases. The oversight is understandable because the uncertainty concern arises partly through separation of powers principles, and the U.S. Supreme Court does not have authority over how states interpret their own separation of powers principles.242See Rossi, supra note 14, at 1218. Further, the federal government prospectively abolished indeterminate sentencing in the Sentencing Reform Act of 1984.243See Doherty, supra note 42, at 996. The absence of federal indeterminate sentences imposed after the Act was passed has left the Court with limited occasions to consider how federal separation of powers questions might arise in indeterminate sentences.

Another explanation for the shift in thinking about parole might be the path-dependency of the law. The distinction between LWOP and LWPP was most prominent in Solem, where it seems to have been invoked to distinguish Rummel without overruling it.244See supra text accompanying notes 210–16. Then, in Ewing, decided when mass incarceration was reaching its peak, the distinction provided a way to uphold a draconian sentence without overruling Solem.245See supra text accompanying notes 225–27. Next, juveniles seeking relief from LWOP pushed the distinction in Graham and Miller, and the Court used it to grant relief without overruling Rummel or Ewing.246See supra text accompanying notes 229–36. The cases did not raise any issues with the constitutionality of parole-eligible sentences themselves, and doing so was not necessarily in the interest of any of the litigants.

Regardless of why the U.S. Supreme Court may have rejected the separate-parole principle in favor of the integrated-parole principle, the fact remains that the Court has clearly done so in its Eighth Amendment jurisprudence. As the next Section explains, state supreme courts have largely followed this move in applying their own state constitutional analogs to the Eighth Amendment.

B. State Courts Adopt the Integrated-Parole Principle

An analysis completed by the author in 2020 considered how each state’s appellate courts have interpreted the possibility of parole when considering whether sentences violate cruel and/or unusual punishment clauses under state constitutions.247Kristen Bell, 51 Jurisdictions Parole Survey (2020) (on file with author). The vast majority of state courts have followed the U.S. Supreme Court in accepting the integrated-parole principle: that parole is part of the sentence and its presence can be dispositive as to whether a sentence is constitutional.248See, e.g., Well-Yates v. People, 454 P.3d 191, 198 (Colo. 2019); State v. McCleese, 215 A.3d 1154, 1166 (Conn. 2019); Martinez v. State, 442 P.3d 154, 156–57 (Okla. Crim. App. 2019); Taylor v. State, 86 N.E.3d 157, 166 (Ind. 2017); Steilman v. Michael, 407 P.3d 313, 319 (Mont. 2017); State v. Smith, 892 N.W.2d 52, 64–66 (Neb. 2017); State v. Hopkins, No. S-1-SC-35052, 2016 WL 3128776, at *6 (N.M. May 26, 2016); Hobbs v. Turner, 431 S.W.3d 283, 286–87 (Ark. 2014); State v. Null, 836 N.W.2d 41, 70–72 (Iowa 2013); Diatchenko v. Dist. Att’y, 1 N.E.3d 270, 284–85 (Mass. 2013); State v. Juarez, 837 N.W.2d 473, 480, 482 (Minn. 2013); Linde v. State, 78 A.3d 738, 743–44 (R.I. 2013); State v. Mossman, 281 P.3d 153, 163–64 (Kan. 2012); Bradshaw v. State, 671 S.E.2d 485, 490 (Ga. 2008); State v. Labrie, No. 03-S-82-86, 2004 WL 895969, at *7–9 (N.H. Super. Ct. Apr. 13, 2004); State v. Davis, 79 P.3d 64, 68–69 (Ariz. 2003); Thomas v. State, 916 S.W.2d 578, 584 (Tex. Ct. App. 1996); People v. Thompson, 633 N.E.2d 1074, 1077 (N.Y. 1994); State v. Ferguson, 519 N.W.2d 50, 54 (S.D. 1994); State v. Busler, 704 S.W.2d 310, 312 (Tenn. 1986); State v. Bishop, 717 P.2d 261, 269–71 (Utah 1986). A handful of states have some caselaw from the early 1980s that retains the separate-parole principle in regard to cruel and/or unusual punishment analysis,249See People v. Taylor, 464 N.E.2d 1059, 1062 (Ill. 1984); State v. Turner, 676 P.2d 873, 876 (Or. 1984); State v. Cooper, 304 S.E.2d 851, 858 (W. Va. 1983); State v. Kido, 654 P.2d 1351, 1359 (Haw. Ct. App. 1982); Kelly v. State, 622 P.2d 432, 438 (Alaska 1981). but after 1985 only one such case was found.250See People v. Cadena, 252 Cal. Rptr. 3d 135, 143–44 (Ct. App. 2019). The California Supreme Court ordered that the opinion be unpublished.

In many states, adoption of the integrated-parole principle has arisen in the context of state constitutional challenges to LWOP sentences imposed on people convicted of crimes while under the age of eighteen. Several state courts have held that although juvenile life sentences violate the state constitution in some (or all) cases if they exclude the possibility of parole, the sentences are constitutional if they include the possibility of parole.251See, e.g., Diatchenko, 1 N.E.3d at 276. The structure of this reasoning has not been limited to the context of juvenile sentencing. The Supreme Court of Michigan, for example, held that LWOP is unconstitutional for first-offense drug possession under its constitution, and that the constitutional problem is cured by adding the possibility of parole.252See People v. Bullock, 485 N.W.2d 866, 875–78 (Mich. 1992). In so holding, the court gave no explanation of how or why adding the possibility of parole could cure the constitutional defect in the sentence.

The state shift to the integrated-parole principle often reflects adherence to the U.S. Supreme Court’s position without independent reflection or consideration of state court precedent that contradicts it. For example, in 1994, the New York Court of Appeals held that the possibility of parole makes a sentence less severe than the maximum term for purposes of proportionality analysis under its state constitution.253See Thompson, 633 N.E.2d at 1077. The court relied on Harmelin to support that position.254See id. (citing Harmelin v. Michigan, 501 U.S. 957, 1028 (1991)). The opinion repeatedly cites a New York case from 1975, People v. Broadie, but ignores the statement in Broadie that sentences that include the possibility of parole are to be treated as equivalent to the maximum term.255See id. at 1075–81; People v. Broadie, 332 N.E.2d 338, 341 (“In considering punishments the maximums must be examined . . . .” (citing In re Lynch, 503 P.2d 921, 925–26 (Cal. 1972))).

The California Supreme Court’s 2021 case, In re Palmer, provides another example of a state court abandoning the separate-parole principle without citation to its prior caselaw.256See In re Palmer, 479 P.3d 782, 789–91 (Cal. 2021). Mr. Palmer had been given a mandatory life sentence for a kidnapping conviction when he was age seventeen. After Mr. Palmer was repeatedly denied parole and served thirty years, he argued that his punishment was grossly disproportionate to his offense. The parole board then granted him release during the pendency of the appeal.257Id. at 785–86, 791. Mr. Palmer argued that the court should still decide his proportionality claim because if his sentence was unconstitutional, he should no longer be required to serve any part of the parole term. The court of appeal agreed and held the sentence disproportionate, but the California Supreme Court reversed, holding that it did not need to decide whether his sentence was disproportionate.258Id. at 786, 791–92.

The court reasoned that the “term” was defined as the “period of actual confinement prior to release on parole,”259Id. at 792 (quoting People v. Jefferson, 980 P.2d 441, 447 (Cal. 1999)). and that parole is a “distinct phase”260Id. at 793 (quoting People v. Nuckles, 298 P.3d 867, 871 (Cal. 2013)). that follows completion of the term. The opinion cited Lynch more than ten times in describing proportionality analysis under the state constitution, but never mentioned its holding that an indeterminate term is deemed to be the maximum term.261See generally id. If the court had applied that part of the holding in Lynch, doing so would have necessitated answering whether Mr. Palmer’s indeterminate life sentence was disproportionate: under Lynch, the grant of parole did not alter the fact that the sentence under law continued until the end of his life.

C. Cross-Substantive Reliance on the Integrated-Parole Principle

This Section describes how the integrated-parole principle has replaced the separate-parole principle in areas of law beyond the context of cruel and/or unusual punishment analysis. The Section describes U.S. Supreme Court caselaw and a sampling of state supreme court cases that illustrate a cross-substantive trend of rejecting the separate-parole principle in the context of ex post facto and procedural due process analysis.

1. Ex Post Facto Context

A law violates the ex post facto clause if it retroactively inflicts “greater punishment” than would have been allowed at the time of the crime.262See Calder v. Bull, 3 U.S. 386, 390 (1798). As previously discussed, state supreme courts historically held that laws that limited the availability of parole did not fall into the category of laws that inflict “greater punishment.” Their reasoning relied on the separate-parole principle: an indeterminate sentence was defined under law as the maximum term, and aspects of when or how parole release may occur were not part of the punishment under law.263See supra Section II.B. An apparent exception is State ex rel. Woodward v. Board of Parole, 99 So. 534 (La. 1924). The Louisiana Supreme Court held that it violated the ex post facto clause to retroactively revoke the possibility of parole among people serving life sentences. Id. at 536. Notably, however, the Louisiana court’s reasoning turned on the fact that removing the possibility of parole was a substantial disadvantage rather than a change to punishment. See id. The court relied in part on Kring v. Missouri, 107 U.S. 221 (1883), in which the U.S. Supreme Court held that a law can violate the ex post facto clause if it creates a substantial disadvantage for a criminal defendant. Kring, 107 U.S. at 235; see Woodward, 99 So. at 703. In 1990, the Court overruled Kring and held that a law cannot violate the ex post facto clause unless it falls into one of the four categories set forth in Calder; a law that creates a substantial disadvantage is not included in the Calder categories. Collins v. Youngblood, 497 U.S. 37, 50 (1990); see Calder, 3 U.S. 386.

In contrast, the U.S. Supreme Court has held that a law that limits the availability of parole is capable of violating the ex post facto clause. In California Department of Corrections v. Morales, the Court considered retroactive application of a California law that extended the period between parole hearings for people convicted of multiple counts of murder.264See Cal. Dep’t of Corr. v. Morales, 514 U.S. 499, 507–09 (1995). Under the old law, this category of people was entitled to a parole hearing every year after having served a minimum period of time. Under the new law, if the parole board denied parole, it could require a person to wait three years until the next hearing.265Id. at 503. The Court held that the ex post facto clause would be violated if the defendant showed a sufficient risk that a person would serve a longer period of time in prison prior to being granted parole. The majority concluded that the evidence in the case did not prove a sufficient risk of longer imprisonment, and therefore held that the ex post facto clause had not been violated.266Id. at 509 (“The amendment creates only the most speculative and attenuated possibility of producing the prohibited effect of increasing the measure of punishment . . . .”). The Court came to a similar conclusion in Garner v. Jones, holding that a change to Georgia’s parole system could potentially violate the ex post facto clause but the evidence did not prove a significant risk of increased punishment.267See Garner v. Jones, 529 U.S. 244, 255–56 (2000); see also Paul D. Reingold & Kimberly Thomas, Wrong Turn on the Ex Post Facto Clause, 106 Calif. L. Rev. 593, 608–09 (2018).

The critical point for the purpose of this Article is not whether the Court has held that laws limiting the availability of parole actually violate the ex post facto clause. The key is that the Court’s doctrine creates potential for those laws to violate the ex post facto clause on the ground that such laws are capable of increasing legal punishment. That doctrine relies on the principle that parole is an integral part of the punishment under law. The doctrine would not make sense if the Court had retained the separate-parole principle, which insisted that the punishment under law is the maximum term irrespective of the possibility of parole.

State supreme courts have largely followed the U.S. Supreme Court in adopting the integrated-parole principle for the purpose of ex post facto analysis. Some of these courts have found ex post facto violations in retroactive changes to laws governing parole release,268See, e.g., Breton v. Comm’r of Corr., 196 A.3d 789, 802–03 (Conn. 2018); Clay v. Mass. Parole Bd., 56 N.E.3d 145, 152–53 (Mass. 2016). whereas others have found no ex post facto violations.269See, e.g., Lay v. Comm’r, Tenn. Dep’t of Corr., No. M2005-02245-COA-R3-CV, 2007 WL 2089742, at *6 (Tenn. Ct. App. July 10, 2007); In re Rosenkrantz, 59 P.3d 174, 200–01 (Cal. 2002). The common thread is that none of these courts appear to treat an indeterminate term as the maximum term for the purpose of this analysis.

2. Procedural Due Process Context

The U.S. Supreme Court has not been wholly consistent in its application of procedural due process analyses in the parole context.270See Ball, supra note 35, at 944. In 1972, the Court held in Morrissey v. Brewer that the Due Process Clause applies in decisions to revoke parole and return a parolee to prison.271See Morrissey v. Brewer, 408 U.S. 471, 485 (1972). The Court held that the minimum due process protections in parole-revocation decisions include written notice of claimed violations of parole; disclosure of evidence against the parolee; an in-person hearing before a neutral hearing body (such as a parole board); the opportunity to present witnesses and evidence; the right to confront and cross-examine witnesses; and a written statement of the evidence and reasons for revoking parole.272Id. at 489. The Court’s reasoning relied on recognizing the practical reality that the historic jurisprudence of parole had long ignored—that the condition of being on parole was substantially different from the condition of being imprisoned behind bars.273See id. at 477. In emphasizing the difference between parole and incarceration, the Court seemed to distance itself somewhat from the separate-parole principle.

The Court expressed adherence to the separate-parole principle, however, in the 1979 case of Greenholtz v. Inmates of the Nebraska Penal and Correctional Complex.274See Greenholtz v. Inmates of Neb. Penal & Corr. Complex, 442 U.S. 1, 9 (1979). The case presented the question of whether the Due Process Clause applies to decisions about whether to grant release on parole.275Id. at 3. The Court characterized parole as a favor or gift, a “mere hope” that was independent from the judicially imposed punishment.276See id. at 11. It explained that states are free to operate systems in which parole boards make release decisions on a wholly discretionary basis without any stated standards and without any procedural protections under the Due Process Clause.277See id. at 7–8. The Due Process Clause does apply, however, if a state has written its parole statute in a way that creates a presumption in favor of parole release.278See id. at 12. If so, the Due Process Clause requires some procedural protections in parole-release decisions but far less than what is required in parole-revocation proceedings. The Court reasoned that because parole-release decisions are “necessarily subjective,” those seeking release on parole are entitled only to an in-person hearing at the initial consideration and a statement of reasons for the decision.279See id. at 13–16.

Critics have argued that the reasoning in Greenholtz is inconsistent with Morrissey.280See, e.g., Ball, supra note 35, at 944. Given Morrissey’s holding that the difference between prison and parole is substantial enough to require due process protections in all parole-revocation proceedings, it seems that the difference should likewise be substantial enough to trigger due process protections in all parole-release decisions. One possible explanation of the difference is the Court’s ambiguous understanding of the legal status of parole: Morrissey is consistent with the integrated-parole principle which emerged in the 1980s, whereas Greenholtz clung to the traditional separate-parole principle.

Most states have applied their state constitutional due process clauses to parole matters in a manner that mirrors the U.S. Supreme Court. At least one state, however, has clearly rejected the separate-parole principle in analyzing procedural due process questions about parole. The Utah Supreme Court has acknowledged that, until 1991, it had relied on the idea that indeterminate sentences are equivalent to the maximum term, that parole was a continuation of custody, and that parole-release decisions were a matter of grace.281See Labrum v. Utah State Bd. of Pardons, 870 P.2d 902, 908 (Utah 1993) (citing Foote v. Utah Bd. of Pardons, 808 P.2d 734 (Utah 1991)). The court criticized that historic view as having “no basis in fact,”282Id. and highlighted the fact that the number of years a person spends in prison is not determined by a judge but “left to the unfettered discretion of the board of pardons.”283Id. It concluded that initial parole-release decisions in its indeterminate sentencing scheme were analogous to judicial sentencing decisions in determinate sentencing schemes.284Id. The court held that because “parole is sentencing,” constitutional due process protections apply to initial parole-release decisions.285Id. at 908, 911.

The Massachusetts Supreme Judicial Court has applied similar reasoning in extending heightened procedural due process protections to parole-release decisions among individuals serving indeterminate life sentences for crimes committed under the age of eighteen. In 2013, the court held that, without the opportunity for parole, life sentences imposed for juvenile convictions violate Article Twenty-Six of the Massachusetts Declaration of Rights prohibiting cruel or unusual punishment.286Diatchenko v. Dist. Att’y, 1 N.E.3d 270, 276 (Mass. 2013), superseded by statute, 2014 Mass. Legis. Serv. ch. 189 sec. 6, § 24, as recognized in Commonwealth v. Watt, 146 N.E.3d 414 (Mass. 2020). Two years later, the court held that heightened due process protections apply to parole hearings among those serving juvenile life sentences.287Diatchenko v. Dist. Att’y, 27 N.E.3d 349, 357 (Mass. 2015). The court explained that although constitutional due process protections generally do not apply to parole-release decisions in Massachusetts, the parole process among juvenile lifers is unique because the opportunity for parole is what makes their sentences constitutional.288Id. The dissenting opinion argued that the court’s holding turned the parole-release decision into a sentencing proceeding. Id. at 369–70 (Spina, J., dissenting). The majority responded to this argument in a footnote that stated some adherence to both the separate-parole principle and the integrated-parole principle. It stated that the sentence itself is defined as the maximum term with parole being a continuation of the sentence in the community. Id. at 357 n.12 (majority opinion). Parole, however, is not simply a change in prison conditions but a critical “component” of the sentence that makes the sentence constitutional. See id. Several scholars and other courts have adopted similar reasoning in regard to juvenile-lifer parole hearings.289See Harrington, supra note 36; Sarah French Russell & Tracy L. Denholtz, Procedures for Proportionate Sentences: The Next Wave of Eighth Amendment Noncapital Litigation, 48 Conn. L. Rev. 1121, 1132 (2016).

III. Indeterminate Sentences Void for Uncertainty Under Most State Constitutions

Although most jurisdictions have followed the U.S. Supreme Court in adopting the integrated-parole principle, they have not yet contended with the implications of doing so. Except for states with constitutional provisions explicitly allowing for indeterminate sentencing, state court precedent is clear that indeterminate sentences are void for uncertainty if release on parole is understood as a change in the punishment itself and release on parole is left at the discretion of a parole board. The legislature would be abdicating its duty to make law about punishment and the parole board would be infringing on the judiciary’s power to impose sentences. Most state supreme courts sidestepped this constitutional challenge by applying the separate-parole principle. Jurists insisted that the possibility of release on parole was not the possibility of a change in punishment but rather the possibility of a change in prison conditions that stood apart from the legal definition of the punishment.

Now that state courts have effectively overruled the separate-parole principle, the weight of the historic state precedent points to the conclusion that indeterminate sentences are unconstitutional in any state that leaves release decisions to the discretion of the board and lacks a constitutional amendment that allows for indeterminate sentencing.

The Appendix to this Article sets forth the substantive standard for parole-release decisions in all jurisdictions that currently impose indeterminate sentences on a substantial percentage of convicted individuals.290The Appendix does not include jurisdictions that have prospectively abolished parole systems and allow for the possibility of parole to specialized groups of individuals—those sentenced to life as juveniles, and those sentenced prior to the abolition of the parole system. For a list of these jurisdictions, see Reitz, Rhine, Lukac & Griffith, supra note 15. With few exceptions, these statutes grant the parole board wide discretion in decisions to grant release on parole. The substantive standard governing release in six states mirrors the language used in one of the first parole statutes enacted by New York in 1877: release is granted if the board finds a “reasonable probability” that a person “will live and remain at liberty without violating the law” and that “release is not incompatible with the welfare of society.”291Lindsey, supra note 27, at 22. See infra Appendix, which cites statutes from Colorado, Connecticut, Georgia, Maryland, Massachusetts, and New York, and Lindsey, supra note 27, at 21–22, which cites the Act establishing Elmira Reformatory in 1877 (providing for release if “there is a strong or reasonable probability that any prisoner will live and remain at liberty without violating the law, and that his release is not incompatible with the welfare of society”). Some statutes provide a presumption of release on parole after a minimum period of time has been served, but the parole board nevertheless retains wide discretion to deny release based on very broad factors.292See Reitz, Rhine, Lukac & Griffith, supra note 15, at 48. The California statute, for example, provides a presumption that a person who has served sufficient time to be eligible for release shall normally be granted parole, but that presumption is defeated if the board finds that “consideration of the public safety requires a more lengthy period of incarceration.”293Cal. Penal Code § 3041(b)(1) (West 2023).

Many statutes direct parole boards to create administrative laws or informal policies that provide more detail about parole-release criteria.294See Thomas & Reingold, supra note 47, at 242–43. These additional board-made rules generally do not meaningfully restrict discretion over parole-release decisions.295See id. Some of these board-made rules repeat the same substantive standard that is included in the governing statute and provide a nonexhaustive list of factors that the board should consider. The Oregon regulations, for example, provide a list of ten factors that includes participation in rehabilitation programs, misconduct in prison, “any apparent development in the inmate personality,” and whether “[t]here is a reasonable probability that the inmate will remain in the community without violating the law.”296Or. Admin. R. 255-032-0020 (2022). Other states have administrative codes that set forth more structured guidelines, but the board retains wide discretion to depart from those guidelines.297See Thomas & Reingold, supra note 47, at 243. Administrative regulations in Georgia, for example, call for calculating a recommended amount of time served prior to parole release by combining a “risk to re-offend score” with a “crime severity level.”298Id.; see Ga. Comp. R. & Regs. 475-3-.05(5) (2023). Georgia law is clear, however, that the parole board retains full discretion to deviate from those guidelines in individual cases and also to change any aspect of the guidelines system at any time.299See Ga. Comp. R. & Regs. 475-3-.05(5) (“The Board specifically reserves the right to exercise its discretion under Georgia Law to disagree with the recommendation resulting from application of the Parole Decision Guidelines and may make an independent decision to deny parole . . . . The Board may modify any part of the Parole Decision Guidelines system at any time.”).

Whether or not statutes are accompanied by more detailed administrative rules, the general pattern is clear: the broad discretion that most parole boards exercise today is closely akin to the broad discretion delegated to parole boards at the turn of the century. In most states, such broad delegation of discretion would have been unconstitutional absent the separate-parole principle. Now that most states have rejected the separate-parole statutes, their indeterminate sentencing systems likely violate their state constitutions. This argument notably would not apply to the handful of states with constitutional amendments that specifically allow indeterminate sentencing.300States with constitutional provisions allowing indeterminate sentences include Michigan, Idaho, Oklahoma, and Utah. See supra notes 102–03. The next Sections discuss two objections to the argument that most existing indeterminate sentences are void for uncertainty.

A. The Problem of Distinctions Across Legal Contexts

The first objection to the argument that most existing indeterminate sentences are void for uncertainty is that it fails to consider that the definition of an indeterminate sentence need not be the same across different legal contexts. Section II.B shows that states have abandoned the separate-parole principle in analyzing whether indeterminate sentences violate their respective cruel and/or unusual punishment clauses. But abandonment of the separate-parole principle in this context arguably does not entail the abandonment of the separate-parole principle in regard to the void for uncertainty doctrine. It is not uncommon in law for a term to be defined one way for the purpose of answering one kind of legal question and defined differently for the purpose of answering a different legal question. Given the path-dependency of law’s development, it is unreasonable to expect perfect consistency in definitions across legal doctrines. It is admittedly “untidy” to define indeterminate sentences as the maximum term in some contexts and not in others, but that untidiness does not render indeterminate sentences unconstitutional.

This argument is misguided because the so-called “untidiness” of having different definitions of an indeterminate sentence across contexts condones decades of potentially cruel and/or unusual incarceration. Courts rely on parole release as something that mitigates punishment in order to dismiss arguments that indeterminate sentences are cruel and/or unusual punishment, but courts do not enforce any rules to ensure that parole actually will mitigate these punishments. Despite Justice Powell’s assertion in Solem that parole-release is “regularized,”301See Solem v. Helm, 463 U.S. 277, 300–01 (1983). parole-release decisions in some states are just as discretionary and devoid of procedural protections as clemency decisions.302See Bell, supra note 18, at 465. During some periods of time, the prospect of parole-release has been slimmer than the prospect of clemency.303For example, 75% of people serving LWOP were granted release through the wholly discretionary exercise of gubernatorial clemency in Connecticut in 1981, but no more than 10% of people sentenced to LWPP in California were granted parole in any year from 1981 to 2008. Compare Conn. Bd. of Pardons v. Dumschat, 452 U.S. 458, 461 (1981), with Kathryne M. Young, Debbie A. Mukamal & Thomas Favre-Bulle, Predicting Parole Grants: An Analysis of Suitability Hearings for California’s Lifer Inmates, 28 Fed. Sent’g Rep. 268, 271 (2016). And it is not uncommon for people who are eligible for release on parole to never be granted release on parole. For example, from 2000 to 2011 in California, more people who were serving LWPP for first- or second-degree murder died in prison than were released on parole.304See Nancy Mullane, Life After Murder: Five Men in Search of Redemption 147 (2012). Among prisoners serving life sentences for first- or second-degree murder, 674 were released from prison and 775 died in prison from January 1, 2000 to May 31, 2011. Id.

Consider Mr. Palmer who was serving LWPP for a kidnapping conviction he received as a juvenile and served multiple decades over his minimum parole eligibility date.305In re Palmer, 479 P.3d 782, 785–87 (Cal. 2021); see also supra notes 256–60 and accompanying text. Suppose a person in his situation were to challenge their indeterminate sentence as void for uncertainty, and a court were to dismiss that challenge on the ground that the punishment is the maximum: incarceration until death. Suppose the person were then to argue that incarceration until death is cruel and/or unusual as applied to their offense. A court would be allowing something far worse than “untidiness” if it were to dismiss the cruel and/or unusual punishment argument on the ground that the proper understanding of the punishment is not incarceration until death but rather the less severe punishment of LWPP. In the words of Justice Powell when he was defending the separate-parole principle in his dissenting opinion in Rummel:

[H]olding that the possibility of parole discounts a prisoner’s sentence for the purposes of the Eighth Amendment [is] cruelly ironic. The combined effect of our holdings . . . allow[s] a State to defend an Eighth Amendment claim by contending that parole is probable even though the prisoner cannot enforce that expectation.306Rummel v. Estelle, 445 U.S. 263, 294 (Powell, J., dissenting).

Given that decades of incarceration are at stake, a decision to accept this bait-and-switch is not mere “untidiness”—it is cruelty under the guise of distinctions in legal contexts.

B. The Problem of Judicial Sentencing Discretion

Many sentencing statutes give judges discretion to impose a determinate sentence within a very broad range and provide judges with only vague guidance on how to select a punishment within that range.307See Beckles v. United States, 580 U.S. 256, 263 (2017). Insofar as indeterminate sentencing statutes are void for uncertainty given similarly broad levels of discretion, it appears that these determinate sentencing statutes should also be void for uncertainty. Neither type of statute provides clear rules about what the punishment will be for a given crime. In both types of statutes, the legislature seems to have delegated the power to set punishment by writing an effectively blank check for nonlegislative bodies to set punishment at their discretion. The force of the objection is that statutes that allow for broad judicial discretion in imposing determinate sentences are clearly constitutional. For better or worse, these statutes have been in effect since the founding of the United States, and their constitutionality has been unquestioned in both state and federal courts.308See, e.g., Apprendi v. New Jersey, 530 U.S. 466, 481 (2000) (stating that the history of sentencing law unambiguously shows that it is permissible for judges to exercise discretion in selecting a sentence within a wide statutory range, “taking into consideration various factors relating both to offense and offender”). Given that determinate sentencing statutes are definitively constitutional despite the fact that they allow broad judicial discretion in sentencing, it seems problematic to conclude that indeterminate sentencing statutes are unconstitutional because they allow broad discretion in sentencing.

The longstanding tradition that allowed judges and juries to exercise discretion over sentencing was clear that the legislature could remove this discretionary power, but it could not delegate it outside of courts.309See generally Quinn, supra note 21. The caselaw discussed throughout Part I was clear that only judges and juries could exercise discretion over sentencing. State supreme courts agreed that if the parole board were exercising discretion over the sentence itself, indeterminate sentencing statutes would not only be void for uncertainty, but would also improperly infringe on the exclusive power of judges and juries to impose sentences.310See supra Section I.B.

An example of this reasoning is found in a 1942 Illinois Supreme Court opinion, which held unconstitutional an amendment to the state’s indeterminate sentencing law.311See People v. Montana, 44 N.E.2d 569, 575 (Ill. 1942). The preexisting indeterminate sentencing law required judges to impose a mandatory statutory minimum and maximum term and gave parole boards discretion to release within that range. Under the amended statute, the sentencing judge set a minimum and maximum term based on individual facts about the case, with the requirement that the judicially set minimum and maximum were within the statutory range. The amendment also provided that the parole board had the authority to adjust the judicially set minimum and maximum (and then allow release within the adjusted range), again with the requirement that the adjusted minimum and maximum remained within the statutory range.312See id. at 572. The court struck the statute as unconstitutional because the amendment attempted to confer upon the parole board the power to make adjustments to the judicially imposed sentence. The court reasoned that conferring such power on the parole board would infringe on an exclusive power of the judiciary: “The power to impose [a] sentence as a punishment for crime is purely judicial. The sentence must be definite and certain. As such it is final, subject only to review by higher courts.”313Id. at 575.

State court precedents are clear that insofar as the legislature does not set specific sentences but instead leaves sentences open to discretion, that discretion can be exercised only by judges and juries.314See, e.g., id. The point highlights that the void for uncertainty doctrine as applied to criminal punishment is closely intertwined with separation of powers principles. Importantly, interpretations of separation of powers principles vary across states and are not bound by U.S. Supreme Court caselaw; the U.S. Constitution does not demand any specific ordering of the separation of powers in the states.315See Rossi, supra note 14, at 1188. Therefore, arguments that indeterminate sentences are void for uncertainty will need to be assessed according to each state’s current doctrine on separation of powers.

State constitutional arguments that indeterminate sentences are void for uncertainty are likely to be more compelling in states with a constitution that includes what scholars refer to as a “strict” separation of powers clause. A strict clause not only divides powers among legislative, executive, and judicial branches, but also prohibits one branch from exercising any other branch’s powers. The Massachusetts Constitution, for example, includes such a clause:

[T]he legislative department shall never exercise the executive and judicial powers, or either of them: the executive shall never exercise the legislative and judicial powers, or either of them: the judicial shall never exercise the legislative and executive powers, or either of them . . . .316Mass. Const. pt. I, art. XXX.

The majority of states have similarly strict separation of powers clauses.317Rossi, supra note 14, at 1190. Some states, however, have a clause that divides powers but does not prohibit one branch from exercising another branch’s powers.318Id. at 1191. A handful of states have no specific separation of powers clause in their constitution.319Id. Most state supreme courts have interpreted their constitutions as requiring a more robust separation of powers than the federal government, even if the specific text of their constitution does not include a strict separation of powers clause.320See id. at 1198–200.

In states with less strict doctrines governing the separation of powers, the historic arguments that indeterminate sentencing is void for uncertainty are likely to be less forceful. The arguments may nevertheless be sufficiently strong to prevail. California, for example, does not require a formal separation by which one branch is prohibited from exercising any of the powers of another branch.321See Jonathan Zasloff, Taking Politics Seriously: A Theory of California’s Separation of Powers, 51 UCLA L. Rev. 1079, 1087 (2004). Powers that are primarily vested in one branch may be shared with another branch, so long as the sharing of power does not “defeat or materially impair the inherent functions of another branch.”322In re Rosenkrantz, 59 P.3d 174, 208 (Cal. 2002) (citing Obrien v. Jones, 999 P.2d 95 (Cal. 2000)). California also has statutes that attach mandatory indeterminate sentences to certain convictions; kidnapping with the purpose of robbery, for example, is punished by a mandatory life sentence with the possibility of parole after seven years.323See, e.g., Cal. Penal Code § 209(b)(1) (2023) (providing a mandatory life sentence with the possibility of parole for kidnapping for the purpose of robbery); id. § 3046(a) (providing that a person serving a life sentence shall not be paroled until he or she has served seven years unless a higher minimum term is specified by law). Judges have no discretion in imposing such sentences, and insofar as parole release is understood as ending a sentence, the parole board is deciding whether the sentence will be seven years, thirty-three years, or any other period until death. Empirical research on California parole-release decisions shows that some people serving seven-to-life have been retained in prison for decades longer than people serving twenty-five-to-life for crimes like first-degree murder.324See Bell, supra note 18, at 507 (showing that juvenile lifers eligible for parole in California who were convicted of nonhomicide, nonsexual crimes served up to thirty-five years, whereas those convicted of first-degree murder served a minimum of nineteen years); id. at 495 (stating that the punishment for first-degree murder in California is twenty-five-to-life, whereas the punishment for kidnapping—a nonhomicide, nonsexual crime—is seven-to-life). As a functional matter, the tail is wagging the dog; the sentence becomes what the parole board makes it up to be, rather than what a judge imposes upon conviction or what the legislature sets in statute. Even under California’s less strict doctrine on separation of powers, the state’s indeterminate sentencing statutes are arguably unconstitutional because they delegate to the parole board powers that “defeat or materially impair” the core functions of the legislature to set punishment and the judiciary to impose punishment.325See In re Rosenkrantz, 59 P.3d at 208 (citing Obrien, 999 P.2d 95).

This constitutional argument would be less strong if judges or juries had the power to meaningfully narrow the range of years over which the parole board has discretionary power to release. Several states have indeterminate sentences with a broad statutory range between the minimum and maximum terms but give judges discretion to impose a narrower range within statutory limits.326See Reitz, Rhine, Lukac & Griffith, supra note 15, at 32. Such statutes arguably provide for a sharing of sentencing power across branches that may pose less constitutional concern in states that lack a strict separation of powers doctrine. In contrast, statutes like California’s that delegate to parole boards nearly all sentencing power pose serious constitutional concern regardless of whether a state adheres to a strict separation of powers doctrine.

Arguments that indeterminate sentences are void for vagueness will need to be tailored to each state’s particular sentencing statutes and separation of powers doctrine. Indeterminate sentences may withstand constitutional scrutiny in some jurisdictions and will certainly withstand scrutiny in states with constitutional provisions explicitly allowing indeterminate sentencing. The next Part considers whether indeterminate sentences may be void for vagueness under the U.S. Constitution even if they pass muster under state constitutions.

IV. Indeterminate Sentencing Analyzed Under the Federal Vagueness Doctrine

This Part provides background about the contemporary federal vagueness doctrine, followed by discussion of how the doctrine compares to the historic state constitutional doctrines and how it applies to parole-release statutes. The discussion shows that although there is a strong argument that parole-release statutes are void for vagueness under the U.S. Constitution, state constitutional arguments likely have more precedential support.

A. Background on the Federal Vagueness Doctrine

The U.S. Supreme Court has explained that the Due Process Clause of the Fifth Amendment is violated when “a criminal law [is] so vague that it fails to give ordinary people fair notice of the conduct it punishes, or so standardless that it invites arbitrary enforcement.”327Johnson v. United States, 576 U.S. 591, 595 (2015) (citing Kolender v. Lawson, 461 U.S. 352, 357–58 (1983)). In the context of criminal law, the Court has primarily applied the void for vagueness doctrine to statutes that define crimes. In 1972 in Papachristou v. City of Jacksonville, for example, the Court struck a statute that criminalized vagrancy.328See Papachristou v. City of Jacksonville, 405 U.S. 156, 156 n.1, 162 (1972) (striking statute that criminalized vagrancy, defined as, among other things, “[r]ogues and vagabonds, or dissolute persons[,] . . . common drunkards, common night walkers, . . . common railers and brawlers, . . . habitual loafers, [and] disorderly persons”). Most relevant to this Article, the Court has recently applied the vagueness doctrine to statutes that set punishments in Johnson v. United States, Sessions v. Dimaya, and United States v. Davis.329Johnson, 576 U.S. 591; Sessions v. Dimaya, 138 S. Ct. 1204 (2018); United States v. Davis, 139 S. Ct. 2319 (2019).

The statutory text struck as void for vagueness in Johnson increased the sentencing range for illegal gun possession if a person had been convicted of three or more violent felonies.330Johnson, 576 U.S. at 593 (citing 18 U.S.C. § 924(e)(1)). The statutory definition of a “violent felony” included the following residual clause: “conduct that presents a serious potential risk of physical injury to another.”331Id. at 593–94 (quoting 18 U.S.C. § 924(e)(2)(B)). Prior caselaw established that this residual clause was to be interpreted using the “categorical approach.”332Id. at 596 (citing Taylor v. United States, 495 U.S. 575, 600 (1990)). The categorical approach instructs a judge to imagine what the crime looks like in “the ordinary case” and then decide whether the ordinary case “presents a serious potential risk of physical injury.”333Id. at 596.

The Court reasoned that the combined effect of two features in the residual clause rendered it void for vagueness. The first feature is that the statute provided no guides, standards, or reliable ways to determine what constitutes an “ordinary case” or how such a case plays out over time.334Id. at 597. One judge, for example, might imagine that the ordinary “attempted burglary” results in someone spotting the burglar, giving chase, and ultimately ending in a violent encounter. Another might imagine that the ordinary “attempted burglary” plays out in a more humdrum manner of a homeowner yelling, “Who’s there?” and the burglar running away.335Id. at 597–98. The second feature of the residual clause that rendered it void for vagueness was its imprecision regarding what poses a “serious potential risk.”336Id. at 598. The Court cited many statutes that are constitutional despite their inclusion of imprecise risk thresholds such as “substantial risk,” “grave risk,” and “unreasonable risk.”337Id. at 603. The downfall of the residual clause was that it tied the imprecise risk threshold to the imaginary “ordinary case” rather than conduct that had actually occurred.338See id. at 603–04.

The statutory text at issue in the subsequent cases of Sessions v. Dimaya and United States v. Davis mirrored the residual clause at issue in Johnson. In each case, the text at issue had been interpreted using the categorical approach discussed in Johnson.339See Sessions v. Dimaya, 138 S. Ct. 1204, 1215 (2018); United States v. Davis, 139 S. Ct. 2319, 2326 (2019). In Dimaya, the text was part of the definition of an “aggravated felony,” and a noncitizen who had been convicted of an aggravated felony was rendered automatically removable.340Dimaya, 138 S. Ct. at 1210. The Immigration and Nationality Act defines an aggravated felony as including, among other things, a “crime of violence” as defined by 18 U.S.C. § 16. 8 U.S.C. § 1101(a)(43)(F). A “crime of violence” is defined as including “any other offense that is a felony and that, by its nature, involves a substantial risk that physical force against the person or property of another may be used in the course of committing the offense.” 18 U.S.C. § 16(b) (emphasis added). The text at issue in Davis defined a “crime of violence,” and a federal statute authorized heightened punishment for an individual who used a firearm during such a crime.341Davis, 139 S. Ct. at 2324; 18 U.S.C. § 924(c)(3) (defining a crime of violence as including any felony “that by its nature, involves a substantial risk that physical force against the person or property of another may be used in the course of committing the offense”). In each case, the Court held that the text at issue shared the same two features as the text in Johnson: it called for judges to imagine an ordinary case and then decide whether that imagined case presented a sufficient amount of risk. The combination of those two features made the statutes void for vagueness.342See Dimaya, 138 S. Ct. at 1216; Davis, 139 S. Ct. at 2326.

B. Comparing Federal and State Vagueness Doctrines as Applied to Punishment

One notable difference between the federal and state doctrines of vagueness as they apply to punishment is that the contemporary federal doctrine is focused almost exclusively on statutory text. State constitutional law applied the vagueness doctrine to the text that defined crimes, but the state doctrine was never exclusively focused on text.343See, e.g., People v. Belcastro, 190 N.E. 301, 303–04 (Ill. 1934) (using textual analysis in striking a statute that criminalized vagrancy under the state constitution’s due process clause). The historic state court cases discussed in Part I apply the vagueness doctrine to the structure of criminal punishment. The central questions in Johnson, Dimaya, and Davis regarded whether the statutory text gave judges too much discretion in interpreting what is a “violent felony” or “crime of violence.” In contrast, the central state constitutional question discussed in Part I is whether a sentence can be structured to end by way of a discretionary judgment.

Despite this difference in focus, the doctrines bear considerable similarity in their underlying principles. In his concurring opinion in Dimaya and majority opinion in Davis, Justice Gorsuch traced the historic pedigree of the void for vagueness doctrine to “the twin constitutional pillars of due process and separation of powers.”344Davis, 139 S. Ct. at 2325 (citing Dimaya, 138 S. Ct. at 1224–28 (Gorsuch, J., concurring)). With respect to the due process pillar, Justice Gorsuch emphasized the common law principle of fair notice. He cited Justice Blackstone in explaining that “[c]riminal indictments at common law had to provide ‘precise and sufficient certainty’ about the charges involved,” and that statutes had to be written with clarity in order to give people fair notice about the law.345Dimaya, 138 S. Ct. at 1225 (Gorsuch, J., concurring) (quoting 4 William Blackstone, Commentaries on the Laws of England 301 (1769)). In regard to the second pillar of separation of powers, Justice Gorsuch explained that the Constitution assigns all legislative power to Congress and that legislators cannot “abdicate their responsibilities for setting the standards of the criminal law.”346Id. at 1227 (quoting Smith v. Goguen, 415 U.S. 566, 575 (1974)). When legislators write vague laws, they unconstitutionally delegate their responsibilities to judges who decide what crimes or punishments are actually included in a vague phrase.347See id. at 1228.

Notably, fair notice and separation of powers were the very same principles that state courts relied on in analyzing early parole systems. The issue of fair notice is particularly clear in the cases that distinguished credit systems from parole systems. Because credit systems were specific enough to provide fair notice about how prison time could be deducted from the maximum term, they could “inhere” into the definition of a sentence without making the sentence void for uncertainty.348See supra Section I.A. In contrast, because no one could know when, if, or why parole release might be granted prior to the maximum term, parole release had to be separate from the legal sentence on pain of making the sentence void for uncertainty.349See supra Section I.A. In regard to separation of powers, courts agreed that indeterminate sentences would be an abdication of the legislative duty to make law about punishment unless the sentences were defined as the maximum term.350See supra Section I.B; Kerr, supra note 115, at 731–32.

Importantly, although both state and federal doctrines share roots in fair notice and separation of powers principles, states differ in their interpretation of separation of powers in at least two ways. First, states generally insist on a stricter separation of powers that can lead to a more robust application of the vagueness doctrine.351See supra text accompanying notes 310–15. Second, state courts have been more broad in articulating how separation of powers principles apply in criminal punishment. In Dimaya, Justice Gorsuch diagnoses the separation of powers issue in vague statutes primarily as a problem of legislative abdication.352See Dimaya, 138 S. Ct. at 1227–28 (Gorsuch, J., concurring). In addition to this problem of legislative abdication, state courts also recognized other separation of powers problems with indeterminate sentences. They grappled with whether indeterminate sentences infringed on the governor’s exclusive power to grant clemency or on the judiciary’s unique power to use sentencing discretion.353See supra text accompanying notes 118–33. By including consideration of not only legislative but also judicial power and gubernatorial power, the state constitutional analysis offers relatively more breadth than its federal analog.

C. Application of the Federal Vagueness Doctrine to Parole-Release Statutes

Given that the federal vagueness doctrine focuses on statutory text, application of that doctrine should begin with the text of statutes that govern indeterminate sentences. The Appendix provides the text of the statutory standard that parole boards apply in deciding whether to grant discretionary release on parole in the thirty-four states that rely heavily on indeterminate sentencing.354See infra Appendix; supra note 290 and accompanying text. In the vast majority of states with discretionary parole-release systems, the parole statutes share the same two features that made the residual clauses void for vagueness in Johnson, Dimaya, and Davis. First, the standards call for parole boards to imagine an abstraction: what will the parole candidate be like in the future if released from prison and placed on parole? Second, the standard asks parole boards to consider whether there is “reasonable probability,” “substantial risk,” or “likelihood” as to whether that future person will violate the law.355See infra Appendix, which cites statutes in Alabama, Alaska, Arkansas, Colorado, Connecticut, Georgia, Iowa, Maryland, Massachusetts, Michigan, Missouri, Montana, Nevada, New Hampshire, New York, Rhode Island, South Carolina, Tennessee, and Texas. The text of the California parole standard does not show this two-part inquiry on its face, but the California Supreme Court has interpreted it in this fashion. See In re Lawrence, 190 P.3d 535, 560 (Cal. 2008) (explaining that the ultimate inquiry for the parole-release decision is whether an incarcerated person, if released, will “pose an unreasonable risk to public safety”).

The New York parole statute is representative of this two-part inquiry. It calls for consideration of a hypothetical: whether “there is a reasonable probability that, if such [an] incarcerated individual is released,” the hypothetically released person will “remain at liberty without violating the law.”356See N.Y. Exec. Law § 259-i(2)(c)(A) (McKinney 2023) (emphasis added) (providing for release on parole “if there is a reasonable probability that, if such incarcerated individual is released, he or she will live and remain at liberty without violating the law”). Other statutes do not call for imagining the hypothetical of releasing the person but are even more vague. For example, statutes in Idaho, Pennsylvania, Vermont, and West Virginia ask whether parole release is in the common good.357See Idaho Code § 20-1005(5) (2023) (release if “it is in the best interests of society”); 61 Pa. Cons. Stat. § 6137(a)(1)(iii) (2022) (release if “the interests of the Commonwealth will [not] be injured”); Vt. Stat. Ann. tit 28, § 502a(b)(2) (2023) (release if “without detriment to the community”); W. Va. Code § 62-12-13(a) (2023) (release if in “the best interests of the state”). There is a strong argument that the language used in nearly all state parole-release statutes should violate the federal vagueness doctrine under Johnson, Dimaya, and Davis.358But see Okla. Stat. tit. 57, § 332.7(R)(1) (2023). Although the Oklahoma statute grants the parole board broad discretion in parole-release decision-making, the statutory language does specifically direct the use of that discretion. The statute provides that release decisions are to be made based on whether the parole candidate substantially complied with the case plan, has not incurred disciplinary infractions in prison, and no victim has submitted an objection to release. Id.

An initial hurdle with the argument is that, under Greenholtz, the Due Process Clause of the U.S. Constitution does not necessarily apply to parole-release decisions.359See Greenholtz v. Inmates of Neb. Penal & Corr. Complex, 442 U.S. 1, 7 (1979); Thompson v. Bd. of Pardons & Paroles, 806 So. 2d 374, 375 (Ala. 2001) (rejecting a federal vagueness challenge to the Alabama parole statute on the ground that the Due Process Clause does not apply to the statute). The reasoning underlying Greenholtz, however, squarely relies on adherence to the separate-parole principle.360See supra text accompanying notes 274–79. The Court characterizes parole in Greenholtz as a “mere hope” that is independent from the judicially imposed sentence.361Greenholtz, 442 U.S. at 11. Given that the Court has subsequently, and repeatedly, held that parole is an integral part of the legal definition of a sentence, Greenholtz lacks support. Scholars have also questioned the holding on several other grounds and have argued that the Court ought to overrule Greenholtz.362See Ball, supra note 35, at 944–48; Thomas & Reingold, supra note 47, at 251.

Assuming that federal vagueness challenges to parole statutes pass the Greenholtz hurdle, they are likely to face another hurdle. Lower courts that have considered federal vagueness challenges to parole statutes prior to the Johnson trilogy have applied a standard that is less searching than what is used in criminal cases.363See, e.g., Lee v. Withrow, 76 F. Supp. 2d 789, 793 (E.D. Mich. 1999); Glauner v. Miller, 184 F.3d 1053, 1054–55 (9th Cir. 1999); Hess v. Bd. of Parole & Post-Prison Supervision, 514 F.3d 909, 913–14 (9th Cir. 2008). These courts have reasoned that parole statutes are not “penal” statutes, and they therefore apply a less-searching vagueness standard that is used in civil statutes.364See, e.g., Hess, 514 F.3d at 913–14 (“The Due Process Clause does not require the same precision in the drafting of parole release statutes as is required in the drafting of penal laws.” (citing Glauner, 184 F.3d at 1055)). The U.S. Supreme Court’s reasoning in Dimaya, however, should change this approach. In Dimaya, the Court applied the heightened vagueness standard when analyzing text that dealt with immigration consequences.365See Sessions v. Dimaya, 138 S. Ct. 1204, 1213 (2018). Insofar as immigration consequences are sufficiently punitive to call for a heightened vagueness inquiry, so too should parole-release decisions. Justice Gorsuch also suggested in his concurring opinion in Dimaya that the heightened standard should be applied to civil and criminal statutes alike.366See id. at 1228 (Gorsuch, J., concurring).

A more difficult hurdle arises due to Beckles v. United States, in which the Court held that the federal sentencing guidelines are not subject to a vagueness challenge.367See Beckles v. United States, 580 U.S. 256, 263 (2017). The Court’s reasoning hinged on the fact that the sentencing guidelines are not mandatory but instead guide judicial discretion in selecting a sentence within the statutory range.368See id. at 263. The Court relied on its tradition of upholding the practice of judicial sentencing discretion, stating “our cases have never suggested that a defendant can successfully challenge as vague a sentencing statute conferring discretion to select an appropriate sentence from within a statutory range, even when that discretion is unfettered.”369Id. at 264. The Court distinguished Johnson on the ground that the text at issue in Johnson defined primary conduct, which in turn mandated a difference in the range of punitive consequences.370Id. at 262–63.

Unlike the statutes at issue in Johnson, Dimaya, and Davis, parole-release statutes do not specify some conduct that mandates a change in permissible punishment. Instead, parole-release statutes are structured to give the parole board wide discretion in deciding whether to incarcerate a person past their parole-release eligibility date.371See supra text accompanying notes 290–99. In this sense, parole statutes are more akin to sentencing statutes that give judges broad discretion in selecting a sentence within a statutory range. The holding and reasoning in Beckles would therefore seem to imply that parole-release statutes are likewise not amenable to federal vagueness challenges.372See generally Beckles, 580 U.S. 256.

This Article does not argue, however, that parole-release statutes are immune from federal vagueness challenges under Beckles. Parole-release statutes are distinguishable from the sentencing guidelines at issue in Beckles on the ground that they concern parole boards exercising sentencing discretion rather than judges or juries. Another distinction is that parole-release discretion is exercised at the back end of a sentence that has already been imposed, rather than during the decision about what sentence to impose in the first instance. These arguments merit closer analysis, particularly as applied to parole statutes governing release of people serving life sentences for juvenile convictions.373Vague parole statutes raise heightened constitutional concern as applied to juvenile life sentences because these sentences have been likened to death sentences. See Graham v. Florida, 560 U.S. 48, 69, 75, 82 (2010); Miller v. Alabama, 567 U.S. 460, 474–75 (2012). As Beckles acknowledged, the Court has struck sentencing factors as vague when those factors were used in imposing death sentences. See Beckles, 580 U.S. at 268. Closer analysis ultimately may show that parole statutes do violate the federal vagueness doctrine, but Beckles presents a serious hurdle to such arguments. In contrast, Beckles presents no hurdle to the argument that indeterminate sentences are void for uncertainty under state constitutions.

V. Constitutional Paths Forward for Indeterminate Sentencing Statutes

State courts in jurisdictions that have indeterminate sentencing statutes and do not have constitutional amendments insulating these sentences have multiple alternatives if litigants challenge those statutes as void for uncertainty. First, courts could return to the separate-parole principle and overrule any caselaw that had rejected it. Second, courts could strike indeterminate sentencing statutes as void for uncertainty unless or until legislative reform restricts the discretionary powers of parole boards. This Part describes each of these paths and argues for the latter.

A. Path 1: Return to the Separate-Parole Principle

A return to the separate-parole principle would avoid the cruel “bait-and-switch” problem discussed in Section III.A. Readopting the separate-parole principle would also require courts to revisit and possibly overrule opinions that relied on the integrated-parole principle. Courts would need to revisit several lines of caselaw, including cases rejecting cruel and/or unusual punishment challenges on the ground that the sentences at issue were less severe because they included the possibility of parole.374Another option might be to let these cases stand but provide for a post-conviction mechanism for people who are denied parole to raise a proportionality claim based on actual time served. See, e.g., In re Dannenberg, 104 P.3d 783, 804 (Cal. 2005) (holding that those denied parole may challenge time served as constitutionally excessive in habeas petitions). This avenue is limited, however, because unlike a proportionality challenge on direct appeal, there is no constitutional right to appointed counsel in post-conviction proceedings, and successful challenges require finding and presenting considerable data about comparable punishments in other jurisdictions. See Pennsylvania v. Finley, 481 U.S. 551, 555 (1987) (holding that the constitutional right to counsel extends to first appeal by right and no further); In re Rodriguez, 537 P.2d 384, 395 n.18 (Cal. 1975) (“Were unrepresented prisoners required to take the initiative by seeking relief at such time as they believed their continued imprisonment to be constitutionally impermissible, . . . courts would . . . receive inadequate petitions unaccompanied by necessary supporting data.”). For example, courts would need to revisit any reliance on Rummel and Ewing because these cases relied on the idea that the possibility of parole makes a sentence less severe.375See Rummel v. Estelle, 445 U.S. 263, 280–81 (1980); Ewing v. California, 538 U.S. 11, 22 (2003); Graham, 560 U.S. at 70 (citing Rummel, 445 U.S. at 280–81).

If courts follow the separate-parole principle to its logical conclusion, doing so would have an especially significant impact on caselaw relating to juvenile sentencing. In Graham and Miller, the U.S. Supreme Court put constitutional restraints on imposing LWOP for convictions of individuals under age eighteen.376See Graham, 560 U.S. at 82; Miller, 567 U.S. at 479. Under the separate-parole principle, those same restraints ought to extend to juveniles serving LWPP. Just as LWOP is unconstitutional for juveniles convicted of nonhomicide offenses under Graham, so too LWPP should be unconstitutional for juveniles convicted of nonhomicide offenses.377See Graham, 560 U.S. at 82. Likewise, the prohibition in Miller against imposing mandatory LWOP on juveniles convicted of homicide offenses should extend to a prohibition against imposing mandatory LWPP on juveniles convicted of homicide offenses.378Miller, 567 U.S. at 479–80. Michigan has held that LWPP for second-degree murder as a juvenile violates its state constitution. See People v. Stovall, No. 162425, 2022 WL 3007491, at *10 (Mich. July 28, 2022).

A return to the separate-parole principle is better than the status quo because it addresses the cruelty of the “bait-and-switch” approach. This Article, however, argues for a different approach. A return to the separate-parole principle is a return to a legal fiction—something that is accepted by courts to achieve a legal aim (such as upholding the constitutionality of indeterminate sentences), regardless of whether it is an accurate reflection of reality. As critics have noted, the practical reality is that being released on parole marks a substantial change in a person’s punishment, not a continuation of the same punishment.379See Ball, supra note 35, at 957 & n.334. By willfully ignoring this reality, the separate-parole principle has allowed the judiciary to largely abdicate oversight over parole, and left parole boards free to determine the practical meaning of most punishments. As Sanford Kadish wrote, the resulting indeterminate sentencing system subordinated criminal law’s fundamental principle of legality in favor of vesting individual officials with “the greatest degree of uncontrolled power over the liberty of human beings that one can find in the legal system.”380See Kadish, supra note 55, at 916. The result has been over a century of widespread arbitrariness in parole-release decisions, exactly what the Michigan Supreme Court feared when it struck the first indeterminate sentencing statute over a century ago.

B. Path 2: Moving Forward Under the Integrated-Parole Principle

Adopting the integrated-parole principle is supported by a more practical understanding of how indeterminate sentences function. Under the integrated-parole principle, the legal definition of an indeterminate sentence mirrors how that sentence functions: the sentence is the minimum term of imprisonment plus however much time the parole board decides to require up to the maximum term. Defining sentences in this way walks directly into the argument that many indeterminate sentencing statutes will be unconstitutionally void for uncertainty. As this Section explains, however, embracing the integrated-parole principle does not require that sentences that include the possibility of release on parole be thrown into an unconstitutional waste heap. If legislatures make substantial reforms to restrict the breadth of parole-release discretion, parole-eligible sentences should stand on sound constitutional ground.

The historic jurisprudence discussed in Part I was clear that when punishment is defined with a proviso that allows it to change over time, the punishment is constitutional so long as the legislature provides clear rules about what makes the punishment longer or shorter.381See supra text accompanying notes 85–93. Courts held that good-time credit statutes were constitutional because they included clear rules that provided notice about what conduct can lead to reduced incarceration, clarity in the definition of that conduct, and stability over time in how that conduct is defined.382See supra text accompanying notes 85–93.

The same type of reasoning can be applied to parole systems, as illustrated by a 1910 opinion by the Kentucky Supreme Court.383See Wilson v. Commonwealth, 132 S.W. 557, 562 (Ky. 1910). The court considered an argument that the Indeterminate Sentence Act of 1910 was unconstitutional because, among other things, it delegated “unlimited power to the Board of Penitentiary Commissioners to discharge persons confined in the penitentiary without fixing a standard by which they shall be entitled to such discharge.”384Id. at 561. The court did not strike the statute as the Michigan and Tennessee courts had done, nor did it follow the majority of courts in adopting a fiction that parole was separate from the sentence itself.385See supra text accompanying notes 85–93. Instead, the Kentucky Supreme Court upheld the indeterminate sentencing statute by finding that, like a good-time credit statute, it included specific rules about release.386Wilson, 132 S.W. at 562.

The Kentucky statute at issue provided that a person is not eligible for parole unless the person has served the minimum term, has incurred no rules violations for at least nine months prior to the date of parole, and has a written contract guaranteeing at least six months of employment at a sustainable wage.387Id. at 561–62. The Kentucky court interpreted the statute to mean that those who comply with the conditions are entitled to release on parole and that the Board has a legally enforceable duty to release them.388Id. at 562. In upholding the statute, the court emphasized that the Board’s authority was not arbitrary but substantively restrained by an enforceable standard.389Id. It also emphasized that the clarity of the standard allowed people to know what they had to do to be entitled to parole: “[T]he standard fixed by these sections is so unmistakable in meaning and specific in character, as to enable the convict to know what his conduct must be to entitle him to the parole or discharge, and also to enable the board to know . . . .”390Id.

In a subsequent case, the court reiterated that upon completion of the standard for parole, a person is entitled to parole as a matter of right.391See Bd. of Prison Comm’rs v. De Moss, 163 S.W. 183, 186 (Ky. 1914). The Kentucky Supreme Court’s clear language of a right to parole stands in stark contrast to most other state supreme courts, which have relied on the separate-parole principle in treating parole as a favor and not a right.392See, e.g., State v. Tyree, 78 P. 525, 526 (Kan. 1904). However, the court’s holding was unfortunately short-lived. The court limited the applicability of its holding years later with little explanation or reasoning, coinciding with the national trend of strong public support for discretionary parole-release systems.393See Bd. of Prison Comm’rs v. Smith, 159 S.W. 960, 962 (Ky. 1913). In Smith, the Kentucky Supreme Court distinguished between two types of parole-eligible sentences. In the first type, the sentence pronounced in the judgment is a fixed term of years, such as twenty-one years. A general parole law gave a commission authority to grant parole to a person prior to the end of that term of years. See id. at 961. In the second type, which was created by the Indeterminate Sentence Act of 1910, the sentence pronounced in the judgment is a maximum and minimum term, and the parole board is authorized to grant parole between the maximum and minimum. See id. Whereas the Indeterminate Sentence Act upheld in Wilson set forth specific conduct that entitled a person to parole as a matter of right, the general parole law gave the board a prerogative to grant or deny parole at the board’s discretion. See id. at 961–62. The court upheld the general parole law without analyzing whether giving standardless discretion to the parole board rendered the sentences uncertain. See id. at 962.

The Kentucky court’s reasoning suggests a promising way forward for putting parole statutes on sound constitutional ground. A sentence can be defined as including the possibility of parole—without rendering that sentence constitutionally uncertain—as long as the conditions governing parole are specific enough to provide fair notice about what a person needs to do (or not do) to be granted parole. Parole-eligible sentences pass constitutional muster if state legislatures do what they are responsible for doing: passing laws that define criminal punishment in clear terms. State courts can review parole statutes to ensure that the rules for release are specific enough to provide notice and to be legally enforceable. Individuals who have satisfied the rules for release should have a protected legal right to release, and courts will need to provide meaningful judicial review of parole-release decisions to protect that right.

Legislators do not need to reinvent the wheel in order to craft parole statutes that provide clear rules governing parole release. In a comprehensive report describing a variety of prison-release mechanisms, Kevin R. Reitz, Edward E. Rhine, Allegra Lukac, and Melanie Griffith highlight what they classify as “administrative parole” statutes.394See Reitz, Rhine, Lukac & Griffith, supra note 15, at 47. The basic design of administrative parole statutes is to establish predetermined, objective criteria at the start of a sentence.395See id. at 47. If an eligible prisoner has served the minimum required time and satisfied these criteria, the individual is released on parole, and parole officials have no discretion to deny release.396See id. The clarity and limited discretion in these administrative parole-release statutes makes them somewhat similar to credit systems. North Carolina, for example, requires release with no parole board consideration if an eligible person has served 80% of the minimum term, has completed risk-reduction programs designated by the Department of Corrections, and has no disciplinary infractions.397See id. at 50.

Twelve states have adopted some form of administrative parole statute.398See id. at 49–51. Unlike the statutes compiled in the Appendix, however, these administrative parole statutes do not apply to the general population of people who are serving indeterminate sentences. Instead, the administrative parole statutes apply to a relatively small subset of individuals who meet strict eligibility requirements.399See id. Most states that operate administrative parole systems exclude people with serious violent convictions.400See id. at 52. Extending administrative parole to the full population of incarcerated people in every state would be a dramatic shift for millions of carceral sentences.

If state legislatures undertook this type of reform to parole statutes, parole-eligible sentences would avoid being void for uncertainty under state constitutions. In addition, the reforms likely would call for a significant change to how the Due Process Clause applies to parole-release decisions. In Greenholtz, the Court held that the Due Process Clause applies to parole-release decisions only if a state statute creates a presumption in favor of parole.401See Greenholtz v. Inmates of Neb. Penal & Corr. Complex, 442 U.S. 1, 11–12 (1979). And where the Due Process Clause does apply, it requires only minimal procedural protections: an in-person hearing at the initial consideration of parole release and a statement of reasons for the decision.402See id. at 14–16. The Court reasoned that more meaningful procedural protections were not required because of the “necessarily subjective” nature of parole-release decision-making.403See id. at 13. The Court characterizes parole release as a “discretionary assessment of a multiplicity of imponderables, entailing primarily what a man is and what he may become rather than simply what he has done.”404Id. at 10 (quoting Sanford H. Kadish, The Advocate and the Expert—Counsel in the Peno-Correctional Process, 45 Minn. L. Rev. 803, 813 (1961)).

This reasoning would be inapplicable if statutes were reformed in a way that imposed clear rules governing parole-release. Parole-release decisions would no longer be a subjective assessment of a person’s future but rather a factual inquiry as to whether a parole candidate completed the goals they were assigned at the start of the sentence. Insofar as decisions turn on a factual inquiry about conduct, the Due Process Clause ought to require meaningful procedural protections and judicial review of parole-release decisions.

Conclusion

This Article argues that state constitutional law provides a strong reason for states to “legalize” their parole systems, that is, to enact clear rules about what is required to earn an enforceable right to release on parole. The Article has not explored the important policy question of whether reforms designed to legalize parole would meaningfully improve operation of parole systems in practice. A great deal would depend on what rules legislatures decide to enact to govern parole release. If they enact rules that make parole incredibly difficult to achieve, legalizing parole could result in fewer people being granted parole and increased prison populations. If legislatures enact rules that put parole clearly within reach, legalizing parole could lead to consistently high rates of parole-release and substantial decarceration over time.

Thus far, legislatures have avoided taking action in either direction by simply failing to enact any clear laws governing parole-release decisions. Aside from the handful of administrative parole systems described in Section V.B, the vast majority of parole-release decisions have been left to the discretion of parole boards without any clear substantive rules. Courts have condoned legislative abdication of parole-release decisions by way of a formal legal fiction for over a century. Now that courts have abandoned that principle, they ought to insist that legislatures take seriously their responsibility to make law about punishment. Legislatures need to make clear rules governing not only how punishment is to start, but also how punishment is to end. Insofar as legislatures continue to leave the end of an indeterminate sentence to the unfettered discretion of parole boards, courts should strike those sentences as void for vagueness.