In this case, we must decide whether the current Georgia parole system, as embodied in the Georgia Constitution, the Georgia statutes, and the rules and guidelines promulgated pursuant to the statutes, creates a liberty interest in parole protected by the Due Process Clause of the Fourteenth Amendment. The district court found no protected liberty interest, and we affirm.

The Georgia parole system is set out in the Georgia Constitution, several Georgia statutes, and the rules and regulations promulgated by the state Board of Pardons and Paroles (the "Board") pursuant to those statutes. The Georgia Constitution vests the Board with the exclusive power to grant reprieves, pardons, and paroles. Ga. Const. art. IV, Sec. 2, p 2. In 1980, the Georgia legislature enacted a law requiring the Board to implement a parole guidelines system. That statute provides in pertinent part:

The board shall adopt, implement, and maintain a parole guidelines system for determining parole action. The guidelines system shall be used in determining parole actions on all inmates, except those serving life sentences, who will become statutorily eligible for parole consideration. The system shall be consistent with the board's primary goal of protecting society and shall take into consideration the severity of the current offense, the inmate's prior criminal history, the inmate's conduct, and the social factors which the board has found to have value in predicting the probability of further criminal behavior and successful adjustment under parole supervision.

O.C.G.A. Sec. 42-9-40(a). Section 42-9-40 is qualified, however, by section 42-9-42, which provides in relevant part:

No inmate shall be placed on parole until and unless the board shall find that there is reasonable probability that, if he is so released, he will live and conduct himself as a respectable and law-abiding person and that his release will be compatible with his own welfare and the welfare of society. Furthermore, no person shall be released on pardon or placed on parole unless and until the board is satisfied that he will be suitably employed in self-sustaining employment or that he will not become a public charge.

Id. Sec. 42-9-42(c).

Pursuant to these provisions, the Board adopted and maintains the Georgia Parole Decision Guidelines System (the "Guidelines").1 That system is embodied in the rules and regulations passed by the Board pursuant to O.C.G.A. Sec. 42-9-40 and in accordance with the Georgia Administrative Procedure Act, O.C.G.A. Sec. 50-13-1 et seq. The stated purpose of the Guidelines is as follows: "In the parole decision process, the Board shall utilize a set of guidelines which consider both the severity of an individual's offense and the likelihood of the individual successfully completing a term of supervision on parole." Parole Decision Guidelines System p 8-1.01.

Other rules and regulations of the Board also shed light on the purpose and function of the system. Chapter 475-3-.05, for instance, provides as follows:

The Parole Decision Guidelines System is an aid to the Board in making more consistent, soundly based and explainable parole decisions and does not create a liberty interest. 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 or establish a Tentative Parole Month at any time prior to sentence expiration.

Ga.R. ch. 475-3-.05. Moreover, Annexure 2, an essential component of the Guidelines, contains similar language:

Parole Decision Guidelines help the Board make a more consistent, soundly based, prompt, and explainable parole decision. Guidelines help the Board decide on a Tentative Parole Month for the inmate or decide that the inmate will complete his sentence without parole. When making decisions, the Board may depart from the Guidelines recommendation and make an independent decision using the full discretion given it under Georgia Law. The length of the prison sentence imposed by the court will be considered in establishing a Tentative Parole Month.

Parole Decision Guidelines System, Annexure 2.2 These provisions set forth the fundamental purpose and function of the Guidelines.

The Guidelines establish a step-by-step procedure for the Board to follow in making parole determinations for eligible inmates. First, the Board assigns the inmate a Crime Severity Level. According to the Guidelines, "[t]he Board deems the nature of the offense to be the most important element in the parole decision. Consequently, the guidelines are to be structured around a Crime Severity Index ranking crimes by increasing degree of seriousness." Parole Decision Guidelines System p 8-1.01. Using a table ranking various crimes in terms of severity, the Board assigns each eligible inmate a Crime Severity Level of I to VII.

Next, the Board assigns the inmate a Parole Success Likelihood Score. "The second component of the Parole Decision Guidelines System combines eight social and criminal history factors found to relate to the likelihood of one's success on parole, which are labeled 'Parole Success Factors.' " Id. p 8-17.01. Rating the inmate on each of these eight factors, the Board arrives at a Parole Success Likelihood Score ranging form 0 to 20 for each inmate.

The Board then uses the Parole Decision Grid to formulate the months-to-serve recommendation. The Parole Decision Grid combines the inmate's Crime Severity Level with the Parole Success Likelihood Score to arrive at a months-to-serve recommendation. Adding the months-to-serve recommendation to the date of the controlling sentence provides the inmate's Tentative Parole Month. "The Tentative Parole Month, during which the offender may expect to be released, absent new information or other cause to cancel the Board's tentative release decision, shall be calculated by adding the recommended months-to-serve to the compute-from date of the controlling sentence." Id. p 8-27.01.

A central point of contention between the parties is whether the Tentative Parole Month is a final, binding determination. The appellant argues that the Tentative Parole Month may be changed only for "new information or other cause," id., as specified in the Guidelines. The Board, on the other hand, maintains that it has the authority to exercise discretion in departing from the grid recommendation. As noted earlier, several of the rules and regulations expressly provide for departure.3 Annexure 2, which contains the Parole Decision Grid, states that "the Board may depart from the Guidelines recommendation and make an independent decision using the full discretion given it under Georgia Law." Id., Annexure 2.4 The Board thus maintains that, in arriving at the Tentative Parole Month, it considers the individual circumstances in each inmate's case, as well as the recommendation of the Parole Decision Grid.

B. Application of the Guidelines to Sultenfuss

The appellant, Stephen Sultenfuss, was convicted of two separate drug charges involving possession of cocaine. He received two concurrent sentences, the longest of which was fifteen years set to run from August 24, 1986, and set to expire on August 23, 2001.5

In order to determine Sultenfuss' Tentative Parole Month, the Board used the Guidelines. Based on his two convictions, Sultenfuss' Crime Severity Level was II. The Board then analyzed the Parole Success Factors and gave Sultenfuss a Parole Success Likelihood Score of 11. Based on these two scores, the Parole Decision Grid resulted in a recommended incarceration period of ten months before parole. Nevertheless, the Board departed from the grid recommendation. Finding that the Crime Severity Level and Parole Success Factors did not adequately reflect the true nature of Sultenfuss' case,6 the Board recommended a term of 61 months, yielding a Tentative Parole Month of January 1992.7

On July 15, 1988, Sultenfuss and several other inmates filed a pro se complaint under 42 U.S.C. Sec. 1983 in the United States District Court for the Northern District of Georgia. Claiming that the Board had violated their rights to due process and equal protection by departing from the grid recommendation in setting release dates, the inmates sought declaratory and injunctive relief as well as compensatory damages. Relying on Slocum v. Georgia State Bd. of Pardons & Paroles, 678 F.2d 940 (11th Cir.) (finding no liberty interest in Georgia parole system prior to 1980 changes), cert. denied, 459 U.S. 1043, 103 S. Ct. 462, 74 L. Ed. 2d 612 (1982), the district court sua sponte dismissed the complaint as frivolous under 28 U.S.C. Sec. 1915(d).

Sultenfuss was the only inmate to appeal the district court's dismissal. On appeal, we affirmed the dismissal of the claim for compensatory damages, but reversed the dismissal of the due process claim. Sultenfuss v. Snow, 894 F.2d 1277 (11th Cir.1990) (per curiam). We held that the district court erred by relying on Slocum because that decision had dealt with the previous parole system and did not address the issue under the current system. "While upon close examination, these changes may not have had an actual impact upon the Georgia parole system to vindicate the due process claim at issue, the changes are sufficiently significant to raise an arguable question of law so as to preclude Sec. 1915(d) dismissal." Sultenfuss, 894 F.2d at 1279. We thus remanded the case to the district court for further proceedings.

On remand, the district court appointed counsel to represent Sultenfuss, and Sultenfuss filed a restated complaint. On August 14, 1990, the Board filed a motion for summary judgment. Finding that Georgia's parole system does not create a liberty interest protected by the Due Process Clause, the district court granted the Board's motion for summary judgment.

Sultenfuss appealed pro se to this court, and, after briefs were filed, we appointed new counsel to represent him. Reversing the district court's grant of summary judgment, a panel of this court held that the current parole system does create a protected liberty interest. Sultenfuss v. Snow, 7 F.3d 1543, 1449-50 (11th Cir.1993), vacated and reh'g en banc granted, 14 F.3d 572 (11th Cir.1994). The panel concluded that "the mandatory language of the statute and the Guidelines, combined with the particularized standards that guide the Board's parole determinations, create a liberty interest in parole release in Georgia." Id. at 1550-51. On February 4, 1994, we vacated the panel opinion and granted rehearing en banc. We now affirm the district court's grant of summary judgment and find that the Georgia parole system does not create a liberty interest protected by the Due Process Clause.

Rule 56(c) of the Federal Rules of Civil Procedure provides that the moving party is entitled to summary judgment "if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law." Fed.R.Civ.P. 56(c). When a summary judgment motion has been made and properly supported, the nonmoving party may not rest upon its pleading, but "must set forth specific facts showing that there is a genuine issue for trial." Fed.R.Civ.P. 56(e). In making this determination, the district court "should 'resolve all reasonable doubts about the facts in favor of the non-movant' and draw 'all justifiable inferences ... in his favor.' " United States v. Four Parcels of Real Property, 941 F.2d 1428, 1437 (11th Cir.1991) (alteration in original) (citations omitted). We review grants of summary judgment de novo, applying the same standards as the district court. Bannum, Inc. v. City of Fort Lauderdale, 901 F.2d 989, 996 (11th Cir.1990).

Sultenfuss alleges that the Board's departure from the grid recommendation violated his rights under the Due Process Clause of the United States Constitution. The Due Process Clause provides that "[n]o State shall ... deprive any person of life, liberty, or property, without due process of law." U.S. Const. amend. XIV, Sec. 1. The requirements of procedural due process, therefore, "apply only to the deprivation of interests encompassed by the Fourteenth Amendment's protection of liberty and property." Board of Regents of State Colleges v. Roth, 408 U.S. 564, 569, 92 S. Ct. 2701, 2705, 33 L. Ed. 2d 548 (1972). As the Supreme Court has explained,

[t]he types of interests that constitute "liberty" and "property" for Fourteenth Amendment purposes are not unlimited; the interest must rise to more than "an abstract need or desire" and must be based on more than "a unilateral hope." Rather, an individual claiming a protected liberty interest must have a legitimate claim of entitlement to it.

Kentucky Dep't of Corrections v. Thompson, 490 U.S. 454, 461, 109 S. Ct. 1904, 1908, 104 L. Ed. 2d 506 (1989) (citations omitted). In order to establish that the Board violated his rights to due process, therefore, Sultenfuss first must demonstrate that the Georgia parole system provides inmates with a liberty interest in parole.

In Greenholtz v. Inmates of Nebraska Penal & Correctional Complex, 442 U.S. 1, 99 S. Ct. 2100, 60 L. Ed. 2d 668 (1979), the Supreme Court recognized that the mere establishment of a parole system by a state does not automatically create a liberty interest in parole. "There is no constitutional or inherent right of a convicted person to be conditionally released before the expiration of a valid sentence." Id. at 7, 99 S.Ct. at 2104. This is not to say, however, that the establishment of a parole system cannot give rise to a liberty interest in parole.

When a state parole system creates a legitimate expectation of parole, an inmate has a liberty interest protected by the Due Process Clause. Id. at 12, 99 S.Ct. at 2106. In Greenholtz, for instance, the Court found a Nebraska parole statute stating that the parole board "shall order [the inmate's] release unless it is of the opinion that his release should be deferred because" one of four specified criteria is met created a protectible liberty interest in parole. Id. at 11, 99 S.Ct. at 2106. Based on the mandatory language of the statute and the presumption that parole be granted unless one of the factors is present, the Court accepted the inmates' view that the statute created an expectation of release protected by the Due Process Clause.

Likewise, in Board of Pardons v. Allen, 482 U.S. 369, 107 S. Ct. 2415, 96 L. Ed. 2d 303 (1987), the Court analyzed a Montana parole statute similar to the statute in Greenholtz. The Montana statute stated that "the board shall release on parole ... any [inmate] when in its opinion there is reasonable probability that the prisoner can be released without detriment to the prisoner or to the community." Id. at 376, 107 S.Ct. at 2420 (emphasis omitted) (first alteration in original). According to the Court, the statute placed "substantive predicates" on the discretion of the decisionmakers and used mandatory language to create a presumption that release would be granted unless the designated findings are made. Id. at 377-80, 107 S.Ct. at 2420-21. Therefore, the Court found that a liberty interest existed.

In other cases, however, the Court has found that penal regulations and parole guidelines do not implicate the Due Process Clause where no liberty interest exists. In Thompson, for instance, the Court analyzed Kentucky prison regulations to determine whether inmates had a liberty interest in visitation. There, the Court focused on the discretion vested in state decisionmakers: "Stated simply, 'a State creates a protected liberty interest by placing substantive limitations on official discretion.' " Thompson, 490 U.S. at 462, 109 S.Ct. at 1909 (quoting Olim v. Wakinekona, 461 U.S. 238, 249, 103 S. Ct. 1741, 1747, 75 L. Ed. 2d 813 (1983)). The Court also emphasized the "requirement, implicit in our earlier decisions, that the regulations contain 'explicitly mandatory language,' i.e., specific directives to the decisionmaker that if the regulations' substantive predicates are present, a particular outcome must follow." Id. at 463, 109 S.Ct. at 1910. The Court found that the regulations at issue lacked the requisite mandatory language because "[t]hey stop short of requiring that a particular result is to be reached upon a finding that the substantive predicates are met." Id. at 464, 109 S.Ct. at 1910. The Court also focused on language in the regulations in which the administrative staff reserved the right to allow or disallow prison visits.

While the Supreme Court has written extensively on whether various prison guidelines create a liberty interest, the Court has declined to set forth any definitve rules. Instead, the Court has emphasized that "whether any ... state statute provides a protectible entitlement must be decided on a case-by-case basis." Greenholtz, 442 U.S. at 12, 99 S. Ct. at 2106. In conducting this case-specific analysis, we are bound by certain principles. After reviewing the relevant caselaw from the Supreme Court and this circuit, we conclude that three, sometimes overlapping, factors are crucial in determining whether a liberty interest is created: (1) whether the system places substantive limitations on the discretion of the decisionmakers; (2) whether the system mandates the outcome that must follow if the substantive predicates are met; and (3) whether the relevant statutes and regulations contain explicitly mandatory language dictating the procedures that must be followed and the result that must be reached if the relevant criteria are satisfied. We explain these factors in more detail below as we examine their application to Georgia's parole system.

In examining these three factors in relation to the Georgia parole system, we must keep in mind that our analysis is inherently subjective. The Supreme Court has recognized that "[n]either the drafting of regulations nor their interpretation can be reduced to an exact science." Thompson, 490 U.S. at 462, 109 S.Ct. at 1909. The Georgia statutes and regulations must be read together to derive the overriding purpose and function of the parole guidelines system. With this caveat in mind, we proceed to analyze the Georgia parole system in light of the three factors set forth above.

First, we must determine whether the system places substantive predicates on the discretion of state officials. See Olim v. Wakinekona, 461 U.S. 238, 249, 103 S. Ct. 1741, 1747, 75 L. Ed. 2d 813 (1983). The inmate must establish that the statute or regulations provide particularized criteria for the decisionmaker to follow, id., and that these standards "meaningfully constrain the discretion of state officials," Allen, 482 U.S. at 384, 107 S.Ct. at 2424 (O'Connor, J., dissenting). In Slocum, we recognized the "critical distinction between 'a scheme that requires release "unless adverse findings based on [specific] criteria are made" [and] a scheme that simply obligates the board to consider such criteria in exercising its discretion.' " Slocum, 678 F.2d at 941 (quoting Boothe v. Hammock, 605 F.2d 661, 664 (2d Cir.1979)) (alteration in original).

We conclude that the Georgia parole system more closely fits within the latter category. Without a doubt, the Guidelines provide a set of particularized criteria that the Board must consider in making parole determinations. The crucial inquiry, however, is whether the relevant statutes and regulations provide standards that meaningfully limit the discretion of the decisionmakers. Not only do the Guidelines leave the Board significant discretion in applying the various factors, but the Board, in promulgating the Guidelines, expressly reserved the authority to depart from the grid recommendation. The criteria in the Guidelines provide a framework to "help the Board make a more consistent, soundly based, prompt, and explainable parole decision." Parole Decision Guidelines System, Annexure 2. The Guidelines were not intended to produce a predetermined outcome upon rote application of specific criteria.

Sultenfuss places great reliance on language from the parole statute mandating that the "guidelines system shall be used in determining parole actions on all inmates." O.C.G.A. Sec. 42-9-40. He argues that this language and the language in the Guidelines dictating the criteria to be considered "constitute specific, substantive limitations on the board's exercise of discretion." Appellant's En Banc Brief at 20-21. We disagree; this language is consistent with the substantial discretion that the Guidelines reserve for the Board. The Board's decision to depart from the grid recommendation does not mean that the Guidelines were not used in determining parole action. The Board may utilize the Guidelines as a framework for making parole determinations, while still exercising the authority to depart from the grid recommendation if other factors, not adequately considered by the Guidelines, warrant such a decision. Therefore, we find that Georgia's parole system does not place limitations on the Board's discretion sufficient to create a liberty interest in parole.

Second, we examine whether the system mandates the outcome that must be reached if the relevant criteria have been met. Thompson, 490 U.S. at 462, 109 S.Ct. at 1909. Where the statute or regulation creates a presumption that release will be granted upon a finding that the substantive predicates have been met,8 a liberty interest in parole exists. Thompson, 490 U.S. at 463-64, 109 S.Ct. at 1910. In those cases where the Supreme Court has found a liberty interest, the Court has focused on the presumption of release found in the relevant statute or guidelines. See, e.g., Allen, 482 U.S. at 376, 107 S. Ct. at 2420 ("the board shall release on parole ... any [inmate] when in its opinion there is reasonable probability that the prisoner can be released without detriment to the prisoner or to the community" (emphasis omitted) (first alteration in original)); Hewitt v. Helms, 459 U.S. 460, 470 n. 6, 103 S. Ct. 864, 871 n. 6, 74 L. Ed. 2d 675 (1983) ("If no behavior violation has occurred, the inmate must be released as soon as the reason for the security concern has abated...."); Greenholtz, 442 U.S. at 11, 99 S. Ct. at 2106 ("the Board ... shall order [the inmate's] release unless it is of the opinion that his release should be deferred because" one of four criteria is met).

We do not find such a mandate in the Georgia statutes and regulations. Neither the relevant statutes nor the Guidelines contain any language mandating the outcome that must be reached after application of the specified procedures. Conversely, the Georgia statutes actually create a presumption against parole. Section 42-9-42, O.C.G.A., provides that "[n]o inmate shall be placed on parole until and unless the board shall find that there is reasonable probability that, if he is so released, he will live and conduct himself as a respectable and law-abiding person and that his release will be compatible with his own welfare and the welfare of society." O.C.G.A. Sec. 42-9-42(c). This section must be read as a qualification of section 42-9-40, the provision requiring adoption of the parole guideline system. Thus, while the legislature has required the Board to adopt a guideline system to be used as a framework for making more consistent parole decisions, it also preserved the Board's authority to use its discretion in making final parole decisions.9 The statute and regulations, therefore, do not mandate that release be granted if the Guidelines criteria is met.

Finally, we look at the relevant statute and regulations to see if they contain explicitly mandatory language directing the decisionmaker to follow certain procedures. Thompson, 490 U.S. at 462, 109 S.Ct. at 1909-10. Although this third factor largely merges with the second, we find the use of explicitly mandatory language to be an important factor in determining whether a liberty interest exists.

It should be obvious that the mandatory language requirement is not an invitation to courts to search regulations for any imperative that might be found. The search is for relevant mandatory language that expressly requires the decisionmaker to apply certain substantive predicates in determining whether an inmate may be deprived of the particular interest in question.

Id. at 464 n. 4, 109 S. Ct. at 1910 n. 4. We recognized this crucial distinction in Staton v. Wainwright, 665 F.2d 686 (5th Cir. Unit B), cert. denied, 456 U.S. 909, 102 S. Ct. 1757, 72 L. Ed. 2d 166 (1982), where we found no protectable liberty interest in the Florida parole statutes. Although the Florida statutes were replete with mandatory language, we recognized that this language dictated what criteria should be considered, not what result must follow if those criteria are met. Id. at 688.

Similarly, as the panel in this case recognized, the Georgia statutes and regulations frequently use mandatory language. See Sultenfuss, 7 F.3d at 1547-49 (highlighting use of the word "shall" in the relevant statutes and regulations). We find no mandatory language, however, expressly dictating the outcome that must follow if the criteria are met. The statutes mandate the implementation of a guidelines system, and the Guidelines mandate certain procedures that must be followed. But the Guidelines do not mandate release if the criteria are met. As noted above, the statutory presumption is against parole unless certain subjective criteria are satisfied. See O.C.G.A. Sec. 42-9-42.

Viewing Georgia's parole system in its entirety, we conclude that no protected liberty interest in parole is created. To give rise to a liberty interest in parole, the statutes and regulations must meaningfully limit the discretion of state officials. Here, the substantial discretion reserved by the Board belies any claim to a reasonable expectation of parole. Although the Board is required to follow some relatively strict procedures, the statutes and the Guidelines, acting in conjunction, do not mandate the grant of parole if specified criteria are satisfied. Instead, the system contains a statutory presumption against parole and an explicit reservation of authority to depart from the grid recommendation, negating any reasonable claim of an entitlement to parole.

D. Deference to Administrative Interpretations

Our conclusion that the Georgia parole system does not create a protected liberty interest is strengthened by another factor, the Board's own interpretation of the Guidelines. It is a well-established rule of statutory construction in Georgia that the interpretation of a statute or regulation by an administrative agency responsible for enforcement of the provision is entitled to great deference, unless clearly erroneous. E.g., Hospital Auth. v. State Health Planning Agency, 211 Ga.App. 407, 408, 438 S.E.2d 912, 914 (1993), cert. denied (Ga.1994); National Advertising Co. v. Department of Transp., 149 Ga.App. 334, 337, 254 S.E.2d 571, 573 (1979); Mason v. Service Loan & Fin. Co., 128 Ga.App. 828, 831, 198 S.E.2d 391, 394 (1973); Belton v. Columbus Fin. & Thrift Co., 127 Ga.App. 770, 772, 195 S.E.2d 195, 197 (1972). Here, the Board is the administrative agency responsible for implementation of the statute and Guidelines under consideration. Under O.C.G.A. Sec. 42-9-40, the Board has the responsibility for adopting, implementing, and maintaining a parole guidelines system. Pursuant to that statute, the Board adopted the Guidelines under consideration here and is the sole administrative agency responsible for administering the parole system. Therefore, unless its interpretation is unreasonable, it deserves great deference.

The Board has unequivocally evinced its opinion that the statute and Guidelines do not create a liberty interest in parole. In several portions of the rules and regulations promulgated pursuant to O.C.G.A. Sec. 42-9-40, the Board has expressly reserved the right to depart from the grid recommendation. See Ga.R. ch. 475-3-.05. In fact, the portion of the Guidelines that contains the parole decision grid has the following "NOTICE" in capital letters at the top of the page:

THE BOARD SPECIFICALLY RESERVES THE RIGHT TO EXERCISE ITS DISCRETION UNDER GEORGIA LAW TO DENY PAROLE EVEN THOUGH GUIDELINES CRITERIA ARE MET BY AN INMATE. IT IS NOT THE INTENTION OF THE BOARD TO CREATE A LIBERTY INTEREST OF THE TYPE DESCRIBED IN GREENHOLTZ V. NEBRASKA PENAL INMATES 442 U.S. 1 (1979).

Parole Decision Guidelines System, Annexure 2. The Board's unequivocal interpretation of the statute and Guidelines, therefore, is that no protected liberty interest is created.10 Because we find this interpretation to be reasonable, we owe it deference. We agree with the Board that Georgia's parole system does not create a liberty interest in parole protected by the Due Process Clause.

The district court found that Georgia's parole system does not create a liberty interest in parole implicating the protections of the Due Process Clause. Georgia's parole system contains a statutory presumption against parole and fails to limit meaningfully the discretion of state officials. We therefore agree with the district court that Georgia inmates do not have a legitimate expectation of parole. Because the protections of the Due Process Clause do not arise without a protectable liberty interest, the district court properly granted summary judgment to the Board.

AFFIRMED.

I dissent from the Court's failure to certify to the Georgia Supreme Court the unsettled questions of state law which control the disposition of this case.

The majority opinion and Judge Clark's dissenting opinion disagree not so much about applicable federal constitutional law as about applicable state law. The disagreement is over whether the relevant Georgia statutes permit the Parole Board to reserve to itself essentially unfettered discretion, which is what it has attempted to do with the adoption of Ga.R. ch. 475-3-.05 and Annexure 2 to the Parole Decision Guidelines System. The state law validity of those two Board-adopted provisions is a premise essential to the majority's conclusion that the Georgia parole system has not created a liberty interest. Likewise, the state law invalidity of those two provisions is a premise essential to Judge Clark's conclusion that a liberty interest has been created. The competing syllogisms are each logical: each one's conclusion follows from its premises. I agree with the majority that if the two provisions in question are valid under state law, so that the Board does have the substantial discretion it claims, then no liberty interest has been created. I also agree with Judge Clark that if the two provisions are not valid under state law, so that the Board lacks the substantial discretion it claims, then a liberty interest has been created.

It is the accuracy of the competing state law premises that is in dispute. Only one of them can be correct, and the case turns on which one is. The majority makes a strong argument that the Board's adoption of Rule 475-3-.05 and Annexure 2 was within its authority under Georgia law. Judge Clark, joined by three other judges of this Court, makes an equally strong argument to the contrary. I do not know whether the majority or Judge Clark is right about the Georgia law question, but I do know where we can, and should, turn for the answer.

We have discretion to certify controlling but unanswered questions of Georgia law to the Georgia Supreme Court. See GA. CONST. art. VI, Sec. 6, para. 4; O.C.G.A. Sec. 15-2-9 (1994); GA.SUP.CT.R. 37. We have not hesitated to do so in the past. In the last five years, we have certified state law questions to the Georgia Supreme Court in no fewer than 19 cases.* As our frequent resort to certification evidences, it is a useful adjudication tool. Only through certification can federal courts get definitive answers to unsettled state law questions. Only a state supreme court can provide what we can be assured are "correct" answers to state law questions, because a state's highest court is the one true and final arbiter of state law. From our perspective, state law is what the state supreme court says it is, and a state supreme court's pronouncements on the subject are binding on every state and federal judge. By contrast, when we write to a state law issue, we write in faint and disappearing ink: what we write does not bind any state court judge, and even as to the federal judges in this Circuit, once the state supreme court speaks the effect of anything we have written vanishes like the proverbial bat in daylight, only faster. For every reason that it is important to decide controlling issues correctly and definitively, certification of unsettled state law questions is important.

Certification is especially important here, because this case "present[s] difficult questions of state law bearing on policy problems of substantial public import whose importance transcends the result in [this] case." Colorado River Water Conservation District v. United States, 424 U.S. 800, 814, 96 S. Ct. 1236, 1244, 47 L. Ed. 2d 483 (1976). The difficulty of the unsettled state law questions in this case is evident from the fact that this Court is sharply divided over those questions. Six judges answer the state law questions one way, and four judges answer those questions the other way. Likewise, the public importance of the questions transcends the result in this one case. The inmate plaintiff has been released already, anyway, see majority op. at 1498 n. 5, but the decision in this case will determine when scores, if not hundreds, of other Georgia prison inmates will be released. The operation of the Georgia prison system--including the determination of which inmates will be released on parole and when--is a matter of utmost importance to the State of Georgia and all of its people.

The Court's unwillingness to certify the important state law questions in this case stands in sharp contrast to its willingness to certify questions of far less public importance in previous cases. The following are some examples of questions this Court has thought sufficiently important to certify to the Georgia Supreme Court in recent cases:

Under the facts of this case, when the insurance policy application established the policy's delivery to and acceptance by the applicant as a condition precedent to the formation of the insurance contract, but the issued policy specified a date certain on which coverage was said to be effective and from which future premium payments were to be calculated, is the failure of actual delivery of the policy of insurance fatal to contract formation so as to render the coverage ineffective?

Middle Georgia Neurological Specialists v. Southwestern Life Insurance Co., 946 F.2d 776, 779-80 (11th Cir.1991).

Whether the insured's tardy forwarding of suit papers is cured by the plaintiff's voluntary dismissal and refiling of an identical lawsuit; or, can the insurer rely on the tardy forwarding in the first suit to avoid liability even though the insured immediately forwarded the papers served in the second suit.

Granite State Insurance Co. v. Nord Bitumi U.S., Inc., 959 F.2d 911, 915 (11th Cir.1992).

Whether, with respect to a automobile insurance policy which covers vehicles principally garaged and used in another state but which is sold and delivered to a resident of Georgia, O.C.G.A. Sec. 33-7-11 acts to invalidate an underinsured coverage exclusion which attempts to limit coverage because the insured was injured in a vehicle not covered by the policy.

Amica Mutual Insurance Co. v. Bourgault, 979 F.2d 187, 190 (11th Cir.1992). Every case matters to the parties, and we should do our utmost to decide every case correctly. Even so, the state law questions involved in this case, and the proper decision of this case, are of greater public import than most, if not all, of the questions that we have certified to the Georgia Supreme Court in the past.

Another reason we should allow the Georgia Supreme Court to answer the unsettled questions of state law upon which this case turns is that the questions involve the division of power and authority between the Georgia Legislature and the Georgia Board of Pardons and Paroles, an executive agency. The way in which a state allocates powers between the executive and legislative branches of its government is a decision of the most fundamental sort for that state; it goes to the very " 'heart of representative government.' " Gregory v. Ashcroft, 501 U.S. 452, 458-61, 111 S. Ct. 2395, 2400-01, 115 L. Ed. 2d 410 (1991) (quoting Sugarman v. Dougall, 413 U.S. 634, 647, 93 S. Ct. 2842, 2850, 37 L. Ed. 2d 853 (1973)). "Through the structure of its government, and the character of those who exercise governmental authority, a State defines itself as a sovereign." Id. 501 U.S. at 460, 111 S. Ct. at 2400. We ought to leave to the Georgia Supreme Court the business of interpreting Georgia legislation that may, or may not, effectively curtail the power of an executive agency that is specifically provided for in the Georgia Constitution. We need not, and we should not, get involved in deciding how Georgia has distributed its sovereign powers among the branches of its government. The majority and the other dissenters do just that by venturing to answer unsettled questions about whether the two Board-adopted provisions conflict with the relevant statute, a matter which has implications sounding in Georgia constitutional law.

If the Georgia parole guidelines statute does prohibit the Board from adopting Rule 475-3-.05 and Annexure 2, as four members of this Court believe, then a question arises as to whether that legislative restriction on the Board's discretion violates Georgia's Constitution. The Georgia Constitution authorizes the Board to grant paroles, but allows the legislature to limit the Board's discretion in certain enumerated circumstances, none of which are relevant here. GA. CONST. art. 4, Sec. 2, p 2. It may be that any other legislative limitation on the Board's discretion, such as requiring it to adopt and employ a set of parole guidelines based on legislatively enumerated criteria, violates the separation of powers clauses of the Georgia Constitution. See id. art. 4, Sec. 7, p 2; id. art. 1, Sec. 2, p 3. Maybe not. The Georgia courts have not squarely addressed the constitutionality of the parole guidelines statute. Compare Charron v. State Bd. of Pardons & Paroles, 253 Ga. 274, 319 S.E.2d 453, 455 (1984) (holding that "an independent Board of Pardons and Paroles is envisioned under our State Constitution," and suggesting that substantive legislative limits on the Board's discretion would be unconstitutional) and Stephens v. State, 207 Ga.App. 645, 428 S.E.2d 661, 663 (1993) (holding that a sentencing court's attempt to impose conditions on a criminal defendant's parole unconstitutionally infringed on the Board's authority) with Freeman v. State, 264 Ga. 27, 440 S.E.2d 181, 184 (1994) (holding that a statute authorizing a sentence of "life without parole" did not unconstitutionally infringe on the Board's discretion). The existence of that state constitutional question, which lurks in the shadows of this case, is another reason we should let the Georgia Supreme Court decide the state law issues.

To the casual reader it will appear that Sultenfuss, or the inmates for whom he is a proxy, are the losers in this case. They are, but they are not the only ones who have lost. The people of Georgia and the principles of federalism also have lost. The people of Georgia have lost some measure of their right to govern themselves, because close and important issues of their state law have been decided not by the court they have established as supreme in such matters, but instead by a federal court which should have deferred to their supreme court. Federalism has lost because some well-meaning federal judges have determined important matters involving Georgia's government that could and should have been left to the state and its supreme court. It sometimes seems as though we federal judges treat federalism like the flag. We salute it, pledge allegiance to it, and like to talk about how important it is. However, in popular parlance: it's easy to talk the talk, but the time has come to walk the walk. Because the Court fails to do so, I dissent.

CLARK, Senior Circuit Judge, dissenting, in which KRAVITCH, HATCHETT and BARKETT, Circuit Judges, join:

The majority errs in denying due process to Georgia prisoners entitled to parole consideration. The court fails to follow controlling United States Supreme Court precedent. Then, it holds that a Georgia administrative agency (the Parole Board) can ignore a controlling statute passed by the Georgia legislature.

The Supreme Court, as will be discussed below, has unequivocally held that a state by legislative action may create a due process right in its parole system. A state does not have to establish a parole system; when it does, it may elect whether to provide for due process protection or leave parole to the unguided discretion of the paroling authority, who may be the Governor, an agency appointed by the Governor, or an agency staffed as prescribed by the legislature.

Due process protects prisoners entitled to parole consideration from decisions of a paroling authority mistakenly made, infected by discrimination or lack of equal protection, resulting from bribery or political influence, or from some other unjustifiable cause. Due process provides protection from unaccountable arbitrary action on the part of government (invisible people), which is what this country is all about.

Some of these evils led the Governor and legislature of Georgia in 1980 to change the existing system. For reasons not clear, the majority of the court decides to ignore the unequivocal change made in the 1980 law (O.C.G.A. Sec. 42-9-40). The majority holds that the Parole Board has the same unbridled and arbitrary discretion that it had before the 1980 legislation.

The 1980 statute mandates implementation of a parole guidelines system, mandates the specified criteria that the system must take into consideration, and mandates use of the system in determining parole actions. In response to the mandates of O.C.G.A. Sec. 42-9-40, the Board promulgated Guidelines that require application of particularized criteria, which determine an inmate's Tentative Parole Month. The Guidelines describe the Tentative Parole Month as the month "during which the offender may expect to be released."1 Notwithstanding the mandatory language, the particularized criteria that guide decisionmaking, and the explicit language of expectancy of release, the majority concludes that the Georgia parole scheme does not give rise to a liberty interest in parole. In so concluding, the majority makes three fundamental mistakes, which I discuss separately below. First, the majority fails to compare the Georgia parole scheme with the parole schemes at issue in the two seminal Supreme Court cases, Greenholtz v. Inmates of Nebraska Penal and Correctional Complex,2 and Board of Pardons v. Allen.3 Such a comparison demonstrates that the Georgia parole scheme necessarily creates a liberty interest in parole. Second, the majority erroneously interprets O.C.G.A. Sec. 42-9-42 as a presumption against parole, rather than as a further indication of the legislative intent to cabin the discretion of the Board. Finally, the majority erroneously defers to the Board's purported reservation of unfettered discretion to ignore the Guidelines for any reason or for no reason at all. The Board's position is nonsensical, patently unreasonable, and contrary to the history, the purpose, and the mandates of O.C.G.A. Sec. 42-9-40. Accordingly, I dissent.

I. Under Greenholtz and Its Progeny, the Georgia Parole

Scheme Creates a Liberty Interest In Parole

A state creates a protected liberty interest "by placing substantive limitations on official discretion."4 The most common means by which a state creates such an interest is "by establishing 'substantive predicates' to govern official decisionmaking, and, further, by mandating the outcome to be reached upon a finding that the relevant criteria have been met."5 Thus, "the repeated use of explicitly mandatory language in connection with requiring specific substantive predicates demands a conclusion that the State has created a protected liberty interest."6 The "substantive predicates" necessary to create a liberty interest have been described as " 'particularized standards or criteria [that] guide the State's decisionmakers.' "7

Greenholtz is the seminal case on liberty interests in parole. The Nebraska parole statute at issue in Greenholtz provided, in pertinent part:

"Whenever the Board of Parole considers the release of a committed offender who is eligible for release on parole, it shall order his release unless it is of the opinion that his release should be deferred because:

"(a) There is a substantial risk that he will not conform to the conditions of parole;

"(b) His release would depreciate the seriousness of his crime or promote disrespect for the law;

"(c) His release would have a substantially adverse effect on institutional discipline; or

"(d) His continued treatment, medical care, or vocational or other training in the facility will substantially enhance his capacity to lead a law-abiding life when released at a later date."8

The statute further provided a list of 15 factors that the Board was obligated to consider in reaching a parole decision. Factor number 15 read: "Any other factors the board determines to be relevant."9 The Supreme Court concluded that this statute gave rise to an "expectancy in release"10 protected by the due process clause.

Some years later, in Board of Pardons v. Allen, the Supreme Court passed upon the Montana parole statute, which provided, in pertinent part:

"Prisoners eligible for parole. (1) Subject to the following restrictions, the board shall release on parole ... any person confined in the Montana state prison or the women's correction center ... when in its opinion there is reasonable probability that the prisoner can be released without detriment to the prisoner or to the community[.]

. . . . .

"(2) A parole shall be ordered only for the best interests of society and not as an award of clemency or a reduction of sentence or pardon. A prisoner shall be placed on parole only when the board believes that he is able and willing to fulfill the obligations of a law-abiding citizen."11

Relying on the use of "mandatory language" and "substantive predicates," the Supreme Court concluded that this statute, like the Nebraska statute at issue in Greenholtz, created a liberty interest in parole. As further support for its conclusion, the Court pointed to the legislative history of the Montana statute: the statute quoted above was enacted in 1955 to replace a 1907 statute that had granted absolute discretion to the Board. The Court saw the 1955 change in the law as an "indication of a legislative intent to cabin the discretion of the Board."12

It is inconceivable that the Nebraska and Montana statutes at issue in Greenholtz and Allen create a liberty interest in parole while the Georgia parole scheme does not. The Georgia statute, which is entitled "Parole guidelines system," provides:

(a) The board shall adopt, implement, and maintain a parole guidelines system for determining parole action. The guidelines system shall be used in determining parole actions on all inmates, except those serving life sentences, who will become statutorily eligible for parole consideration. The system shall be consistent with the board's primary goal of protecting society and shall take into consideration the severity of the current offense, the inmate's prior criminal history, the inmate's conduct, and the social factors which the board has found to have value in predicting the probability of further criminal behavior and successful adjustment under parole supervision.

(b) The guidelines system required by subsection (a) of this Code section shall be adopted by rules or regulations of the board. The rules or regulations shall be adopted in conformity with Chapter 13 of Title 50, the "Georgia Administrative Procedure Act."13

Certainly the language of this statute is no less "mandatory" than that of the Nebraska or Montana statutes at issue in Greenholtz and Allen. The Georgia statute mandates implementation of a parole guidelines system, mandates use of the system in making parole decisions, and mandates the criteria that the system must take into consideration.

Likewise, the criteria that guide the Board's decisionmaking under the Georgia parole scheme are no less "particularized"14 than the criteria in the Nebraska or Montana statutes at issue in Greenholtz and Allen. Under the Georgia scheme, the Board must arrive at each parole decision by following a step-by-step procedure. First, the Board must assign the inmate a Crime Severity Level, which is determined by reference to a table that lists dozens of specific crimes, with each crime assigned a level between I and VII.15 Second, the Board must assign the inmate a Parole Success Likelihood Score. This score is calculated based on the following eight factors:

(a) Age at First Commitment;

(b) Prior Juvenile and Adult Convictions;

(c) Prior Incarcerations Since Age 17;

(d) Probation/Parole Failure;

(e) Use, Possession, or Attempt to Obtain Heroin or Opiate Drugs;

(f) Commitment Offense Involved Burglary or Forgery;

(g) Fully Employed During the Six Months Prior to Current Offense.16

Each of these eight factors is designated a numerical scoring system designed to reflect the inmate's success or lack thereof as to that factor. The numerical scores for each of the eight factors are added together to arrive at the Parole Success Likelihood Score, which ranges from zero to 20.17 Third, the Board must take the Crime Severity Level and the Parole Success Likelihood Score and apply the Parole Decision Grid to determine the number of months the inmate must serve:

The [Parole Success Likelihood Score] shall be used in conjunction with the Crime Severity Level finally to determine a recommended number of months to serve from the Parole Decision Guidelines grid.

. . . . .

PAROLE DECISION GRID. The Parole Decision Grid is the final component of the Parole Decision Guidelines, the application of which determines the months-to-serve recommendation.18

As this detailed procedure demonstrates, the criteria that guide the Board under the Georgia scheme are far more detailed and particularized, and go much farther in limiting the Board's discretion, than the criteria in the Nebraska and Montana statutes at issue in Greenholtz and Allen. For example, under the Nebraska statute, the Board could consider "[a]ny ... factors the board determines to be relevant,"19 while the criteria set out in the Georgia statute and Guidelines grant the Board no such discretion. Likewise, the criteria in the Montana statute were very few and very general, thus permitting the Board far more discretion than the very particularized criteria in the Georgia scheme.

Finally, the Georgia Guidelines, like the Nebraska and Montana statutes, contain mandatory language that " 'creat[es] a presumption that parole release will be granted' when the designated findings are made."20 The Guidelines provide:

TENTATIVE PAROLE MONTH. The Tentative Parole Month, during which the offender may expect to be released, absent new information or other cause to cancel the Board's tentative release decision, shall be calculated by adding the recommended months-to-serve to the compute-from date of the controlling sentence.21

The Greenholtz decision turned upon the Supreme Court's finding that language of the Nebraska statute created an "expectancy of release" on parole.22 The provision quoted above, coupled with the mandated procedure by which the Board must arrive at the recommended months-to-serve, creates more of an "expectancy of release" than the Nebraska and Montana statutes at issue in Greenholtz and Allen. Other provisions in the Guidelines confirm this "expectancy of release." For example, in a section setting out the benefits of the system, the Guidelines provide: "By eliminating the uncertainty of release dates, inmate morale and sincere participation in rehabilitation programs is enhanced."23 The Guidelines eliminate uncertainty by mandating that an inmate's Tentative Parole Month, during which he may expect to be released, be set by applying the Parole Decision Grid to the inmate's Crime Severity Level and Parole Success Likelihood Score. Thus, like the statutes at issue in Greenholtz and Allen, the Georgia parole written procedures use mandatory language that requires the Board to apply particularized criteria to arrive at a parole decision. Under Greenholtz and Allen, the Georgia scheme creates a liberty interest in parole.

The majority's reasoning is not grounded in the Greenholtz and Allen decisions. I give three specific examples. First, the majority begins its discussion of the Georgia parole system with the statement: "we must keep in mind that our analysis is inherently subjective."24 The majority does not cite and I am not aware of any authority to support the proposition that a Greenholtz-type analysis is "inherently subjective." Under Greenholtz and its progeny, we must look to the language of the state parole scheme and determine whether that language places "substantive limitations on official discretion."25 I see nothing "inherently subjective" in this analysis.

Second, the majority concedes that "the Guidelines provide a set of particularized criteria that the Board must consider in making parole determinations."26 Nevertheless, the majority concludes that this "set of particularized criteria" does not indicate the existence of a liberty interest because "the Guidelines leave the Board significant discretion in applying the various factors.... The Guidelines do not envision the rote application of specific criteria and a predetermined outcome if those criteria are met."27 In so concluding, the majority ignores the teaching of Allen, in which the Supreme Court distinguished between two types of discretion:

In essence, the Court [in Greenholtz ] made a distinction between two entirely distinct uses of the term discretion. In one sense of the word, an official has discretion when he or she "is simply not bound by standards set by the authority in question." R. Dworkin, Taking Rights Seriously 32 (1977). In this sense, officials who have been told to parole whomever they wish have discretion. In Greenholtz, the Court determined that a scheme awarding officials this type of discretion does not create a liberty interest in parole release. But the term discretion may instead signify that "an official must use judgment in applying the standards set him [or her] by authority"; in other words, an official has discretion when the standards set by a statutory or regulatory scheme "cannot be applied mechanically." Dworkin, supra, at 31, 32; see also id., at 69 ("[W]e say that a man has discretion if his duty is defined by standards that reasonable [people] can interpret in different ways"). The Court determined in Greenholtz that the presence of official discretion in this sense is not incompatible with the existence of a liberty interest in parole release when release is required after the Board determines (in its broad discretion) that the necessary prerequisites exist.28

The discretion upon which the majority relies in concluding that the "particularized criteria" in the Guidelines do not indicate the existence of a liberty interest is clearly the second type of discretion described by the Court in Allen. The majority fails to recognize that this type of discretion is not incompatible with the existent of a liberty interest. Just as a liberty interest may exist when "the standards set by a statutory or regulatory scheme 'cannot be applied mechanically,' "29 so a liberty interest may also exist when "[t]he Guidelines do not envision the rote application of specific criteria...."30

Finally, the majority fails to recognize the significance of the 1980 change in the Georgia law. In Allen, the Supreme Court relied on a change in the Montana parole statute to support its conclusion that the new statute created a liberty interest in parole. Noting that the old statute granted absolute discretion to the Board, the Court found that the new statute provided an "indication of a legislative intent to cabin the discretion of the Board."31 Similarly, prior to 1980 in Georgia, there was no statutory requirement that the Board exercise its parole powers in accordance with any specific guidelines or regulations. The Georgia statutes did not place limits on the Board's discretion to determine when an eligible inmate was to be released on parole. In 1980, however, the Georgia legislature enacted O.C.G.A. Sec. 42-9-40, thereby mandating that such parole determinations be made in accordance with a parole guidelines system. Under Allen, this change in the law is at least some indication of legislative intent to "cabin the discretion of the Board."32

The majority is able to reach the conclusion it does only by ignoring both the statutes at issue in Greenholtz and Allen and the teachings of those two cases. Under Greenholtz and its progeny, the Georgia parole scheme gives rise to an expectancy of parole that is protected by the due process clause.

II. O.C.G.A. Sec. 42-9-42 Is Consistent with a Liberty

The majority relies heavily on O.C.G.A. Sec. 42-9-42, which it interprets as creating a "presumption against parole."33 A cursory review of the Supreme Court's decision in Allen indicates that this interpretation is flawed. O.C.G.A. Sec. 42-9-42 is very similar to a portion of the Montana statute at issue in Allen. O.C.G.A. Sec. 42-9-42 provides, in part:

No inmate shall be placed on parole until and unless the board shall find that there is reasonable probability that, if he is so released, he will live and conduct himself as a respectable and law-abiding person and that his release will be compatible with his own welfare and the welfare of society.

Similarly, the Montana statute at issue in Allen provided, in part:

A prisoner shall be placed on parole only when the board believes that he is able and willing to fulfill the obligations of a law-abiding citizen.34

O.C.G.A. Sec. 42-9-42 further provides, in part:

[N]o person shall be released on pardon or placed on parole unless and until the board is satisfied that he will be suitably employed in self-sustaining employment or that he will not become a public charge.

Similarly, the Montana statute provided, in part:

A parole shall be ordered only for the best interests of society and not as an award of clemency or a reduction of sentence or pardon.35

The Supreme Court in Allen did not read the Montana statute as creating a presumption against parole. To the contrary, it read the portions of the statute quoted immediately above as a further constraint on the discretion of the Board, preventing a decision in favor of release when a prisoner is unable "to fulfill the obligations of a law-abiding citizen."36 Likewise, O.C.G.A. Sec. 42-9-42 can only be interpreted as a further constraint on the discretion of the Board, not as a presumption against release. Consistent with the purpose and intent of the Georgia parole guidelines system, O.C.G.A. Sec. 42-9-42 prohibits the Board from releasing a prisoner who cannot "live and conduct himself as a respectable and law-abiding person." Thus, O.C.G.A. Sec. 42-9-42 is compatible with the liberty interest in parole created by the Georgia parole scheme. Any other conclusion flies in the face of the Supreme Court's decision in Allen.

III. The Board's Position that It Retains Absolute Authority

to Ignore the Guidelines is Contrary to the

of the Georgia Parole Guidelines System

In addition to O.C.G.A. Sec. 42-9-42, the majority also relies heavily on Georgia Rule 475-3-.05(5) and on a portion of the Guidelines that the majority refers to as Annexure 2. Relying solely on these two provisions, the majority states that "several of the rules and regulations expressly provide for departure [from the grid recommendation]."37 The majority is correct to the extent that Georgia Rule 475-3-.05(5) purports to reserve the Board's discretion to "disagree" with the Guidelines, and Annexure 2 purports to reserve the Board's decision to "depart" from the Guidelines.38 Nevertheless, neither of these provisions nor any other provision in the Guidelines even mention either a procedure to follow or criteria to consider in effecting such a departure. Thus, the Board purports to reserve for itself the discretion to make a decision independent of the Guidelines, that is, to simply ignore the Guidelines, for any reason or for no reason at all. I cannot accept, as the majority does, the Board's expansive interpretation of its own discretionary powers.

The Board's position, when considered in the context of the Georgia parole scheme as a whole, is nonsensical. The Georgia legislature, with O.C.G.A. Sec. 42-9-40, mandated that the Board implement and use Guidelines that take into consideration certain specified factors. The Guidelines promulgated pursuant to this mandate set out a detailed procedure, including the application of numerous particularized criteria, by which the Board must arrive at an inmate's Tentative Parole Month. The Guidelines also set out a procedure by which the Board may adjust the Tentative Parole Month prior to an inmate's release; such adjustments may be based on the inmate's institutional behavior or on new information not previously available.39 The Guidelines do not, however, provide a procedure for departures. That is, there is absolutely nothing in the Guidelines to indicate when or how the Board is to go about disagreeing with, or departing from, the Tentative Parole Month. Neither do the Guidelines indicate what factors or criteria the Board should consider in determining whether departure is appropriate. Rather, the Guidelines simply declare, in Annexure 2, that "the Board may depart from the Guidelines recommendation and make an independent decision...." It is inconceivable that the Georgia legislature intended to establish a detailed and particularized system for arriving at an inmate's Tentative Parole Month, "during which the offender may expect to be released,"40 only to grant the Board the unfettered discretion to completely disregard the system.

Other provisions of the Guidelines are inconsistent with the Board's position that it has unfettered discretion to depart. For example, the Guidelines provide a procedure by which an inmate may request reconsideration of his months-to-serve calculation.41 The Guidelines specifically provide, however, that such requests "are limited to alleged errors in the Crime Severity Level and/or the Parole Success Factor scores."42 Thus, under the Guidelines, an inmate may request reconsideration of the two factors that determine his Tentative Parole Month, but he may not request reconsideration or otherwise contest the Board's decision to depart from this Tentative Parole Month. It makes no sense to provide a procedure for contesting the determination of the Tentative Parole Month but not provide for contesting the decision to depart therefrom.

Another example of the Guidelines' incompatibility with the Board's position is p 8-12.01, which provides that an inmate's Crime Severity Level shall be determined "based on actual convictions of record, without reference to original arrest or booking charges, or allegations in accusations or indictments for which an actual conviction was not obtained."43 Notwithstanding this provision, it is the Board's position that it may consider any information in deciding, in its unfettered discretion, whether to depart from the Guidelines. Sultenfuss is a case in point. On Sultenfuss's parole rating summary sheet are the following comments:

[M]ajor drug dealer of cocaine in Macon for years and always claims entrapment. Will have numerous lawyers telling us he's a great fellow. Paroled on cocaine charges revoked based on new cocaine charge received maximum sentence of 15 years to serve. This is a small quantity case which is out of character for him.44

Thus, the Board certainly considered "accusations ... for which an actual conviction was not obtained" in reaching its decision to depart from the Guidelines in Sultenfuss's case. To prevent the Board from considering certain information during the calculation of the Tentative Parole Month, when the inmate may challenge the information, but then permit the Board to consider the very same information during the departure decision, when the inmate has no recourse, defies logic and common sense. The Board's interpretation of its own discretion is contrary to the construction of the Guidelines as a whole.

The Board's position is completely at odds with the history and the purpose of the 1980 change in Georgia law governing parole decisions. A bit of background here is helpful.45 In 1978, a federal investigation disclosed widespread corruption in the Tennessee parole system. The scandal reached the Governor, Ray Blanton, and three individuals from his office were eventually convicted of accepting bribes for influencing the granting of pardons and paroles.46 As this scandal became national news, the institution of parole came "under siege."47 Some suggested abolishing parole altogether, while others suggested strengthening the standards that govern the way parole boards operate.48

A year later, in 1979, the Georgia Board of Pardons and Paroles was widely criticized for a number of its decisions.49 One of the Board's decisions even led to a grand jury investigation of the Board's actions and, later, to a state court ruling that certain Board proceedings were unconstitutional.50 Specifically, the state court held that the Board must cease its practice of conducting secret proceedings.51

In 1980, in the wake of the Tennessee scandal and the controversies over the Georgia Board's decisions, the Georgia legislature enacted O.C.G.A. Sec. 42-9-40, mandating that the Georgia Board establish and use a guidelines system in rendering parole decisions. This new system was touted as a "scientific approach" to parole decisions. A contemporaneous newspaper report reads:

Gov. George Busbee announced Thursday that a new "scientific approach" will be implemented to help determine whether inmates in Georgia prisons ought to be paroled, thus avoiding the controversy that surrounded such paroles as the recent release of convicted kidnapper Gary Steven Krist.

"No longer," said Busbee, "will the parole board's decision-making process be a myth to the public.... The (new) parole rating guidelines will bring the Pardons and Paroles Board into the 20th Century."

. . . . .

"The parole rating guidelines will fulfill two critical needs in the criminal justice system," said the governor. "It will bring about greater fairness, consistency and openness in the parole decisions-making process and will allow us to expand the use of paroles as an alternative to incarceration."

. . . . .

The new program will involve "the careful development of two charts as grids," Busbee explained.

"One chart--the crime-severity index--will rank in fine detail the seriousness or severity of each category of crime as determined by input from board members. The second chart takes into account social factors, including work history and criminal history, which serve as indicators of the offender's chance for success on parole," he said.

Once in place, the criteria used in making decisions will be available for public review, the governor said. This will increase "the credibility of the system," he said.52

Publicly, at least, the Board purported to embrace the new guidelines system as a means to repair its tainted image. Another contemporaneous newspaper report reads:

To repair its image, the board set about codifying its parole procedures. Reviews of past cases had revealed wide disparities in the board's decisions. "We were not the most consistent people in making decisions," [board director] Morris said. "As we looked at cases we would find that, for example, a person serving for an offense would be paroled. Four months later, a person serving for the same offense with the same background would not be paroled."

To deal with charges of capriciousness, the board and its staff ... began developing a new way of dealing with the cases. Advertised as a "scientific" way of managing parole, the plan took effect in January.53

Thus, by mandating the implementation and use of the guidelines system, the legislature intended to prevent the perversion of the system that had occurred in Tennessee and instill public confidence in the Georgia Board by insuring that parole decisions were made equitably, objectively, and openly. This intent is embodied in the Guidelines. For example, p 8-1.02 provides, in pertinent part:

The Parole Decision Guidelines System shall be utilized to promote the following policy goals of this Board:

(a) Objective and equitable use of the clemency power in determining appropriate lengths of imprisonment for individual inmates;

(b) Equitable and consistent terms of imprisonment for all inmates sharing similar personal behavior histories and convictions for similar crimes....

The Board's position that it retains unfettered discretion to depart from the Guidelines for any reason or for no reason at all is at odds with these stated goals. Rather than encouraging equitable, objective, and consistent decisions under the guidelines system, the Board's position permits arbitrary departures.

By enacting O.C.G.A. Sec. 42-9-40, the Georgia legislature mandated that the Board implement the guidelines system and use the system in determining parole actions. The majority concludes that the Board's position is not inconsistent with this mandate because "[t]he Board's decision to depart from the grid recommendation does not mean that the Guidelines were not used in determining parole action."54 The majority fails to acknowledge, however, that under the Board's interpretation, the Guidelines are meaningless. If the Board has the unfettered discretion to arbitrarily depart from the Guidelines, then the Guidelines play no meaningful role in the final parole decision; and if the Guidelines play no meaningful role in the decision, they are not "used in determining parole actions" as mandated by the statute. The Board's purported retention of unfettered discretion to arbitrarily depart from the Guidelines, without notice or response from the inmate, is nothing more than a cavalier attempt to avoid the mandate of the Georgia legislature.

This purported retention of discretion is not the only cavalier response the Board has had to the legislature's mandate. O.C.G.A. Sec. 42-9-40(b) specifically requires that the guidelines system be adopted by rules or regulations of the Board and that these rules or regulations "be adopted in conformity with Chapter 13 of Title 50, the 'Georgia Administrative Procedure Act.' " Under the Georgia Administrative Procedure Act, the Secretary of State "shall compile, index, and publish all rules adopted by each agency...."55 Rules may be omitted from the Secretary of State's official compilation provided the "compilation contains a notice stating the general subject matter of rules so omitted and stating how copies thereof may be obtained."56 Notwithstanding these requirements, the Guidelines discussed in this opinion are not published in the Secretary of State's official compilation of rules, and there is no notice "stating how copies thereof may be obtained." It is the Board's position that the Guidelines are a part of an "internal operating manual"57 that has not been made available to this court for review. By failing to make the Guidelines available to the public as mandated by law, the Board does nothing to dispel the stereotype of the parole board as a shadowy body making arbitrary and capricious decisions outside the realm of public scrutiny.

While an agency's interpretation of a statute it is charged with implementing is entitled to deference, "administrative construction should be restricted to cases in which the meaning of the statute is really doubtful and must be disregarded where its invalidity is apparent."58 Here, the Board's self-serving interpretation of the Georgia parole scheme is contrary to the Georgia statute and, therefore, invalid. The majority errs in deferring to the Board instead of following the law established by the Georgia legislature.

Under Greenholtz and its progeny, the language of the Georgia parole statutes and Guidelines creates a liberty interest in parole. Nothing in the Georgia parole scheme is inconsistent with this liberty interest except the Board's self-serving attempt to reserve for itself the unfettered discretion to arbitrarily depart from the Guidelines. The Board's interpretation is without any basis; indeed, it is contrary to the mandates of the Georgia legislature and is, therefore, entitled to no deference. I dissent.