concurring in the judgment.
In United States ex rel. Forman v. McCall, (“Forman I”), 709 F.2d 852 (3d Cir.1983) this court held that, as applied to petitioner, the 1979 parole guidelines were both retroactive and disadvantageous. We also held, following Geraghty v. United States Parole Commission (“Geraghty I”), 579 F.2d 238 (3d Cir.1978), vacated and remanded on other grounds, 445 U.S. 388, 100 S.Ct. 1202, 63 L.Ed.2d 479 (1980), that the guidelines should be considered “laws” within the meaning of the ex post *1165facto clause if they were “applied without substantial flexibility.” 709 F.2d at 862. We remanded for factfinding as to what extent the Parole Commission actually exercised its discretion not to adhere to its parole guidelines. Id. We did not state then, and the majority does not state now, what degree of adherence would suffice to make the guidelines “laws,” though it is now apparent that 75.4% adherence is not enough.
The majority now reverses on the ground that the district court’s factual findings of 85-90% adherence were clearly erroneous, and that 75.4% is not enough. I am not as certain as my colleagues that the district court’s findings were clearly erroneous, but I write separately to express an even more fundamental reservation. I believe that the standard announced in Forman I for determining whether the guidelines are “laws” was incorrect under ex post facto clause jurisprudence and unwise as a matter of policy. Because I also reach the conclusion that the guidelines are not “laws,” albeit by a different route, I concur in the judgment of the majority. Moreover, I recognize that under our Internal Operating Procedures we are bound to follow a prior panel decision. I urge, however, that the Forman I rule be reconsidered by the court in bane.
I.
An ex post facto law is one that,
makes an action done before the passing of the law, and which was innocent when done, criminal; and punishes such action____ [Tjhat aggravates a crime, or makes it greater than it was, when committed____ [Tjhat changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed____ [Tjhat alters the legal rules of evidence, and receives less, or different testimony, than the law required at the time of the commission of the offense, in order to convict the offender.
Calder v. Bull, 3 U.S. (3 Dall.) 386, 390, 1 L.Ed. 648 (1798). Directed principally against the tyrannical and arbitrary exercise of legislative power,1 the ex post facto clause protects against laws that are expressions of legislators’ “ambition, or personal resentment and vindictive malice,” Calder v. Bull, 3 U.S. (3 Dall.) at 389, and “upholds the separation of powers by confining the legislature to penal decisions with prospective effect and the judiciary and executive to applications of existing penal law.” Weaver v. Graham, 450 U.S. 24, 29 n. 10, 101 S.Ct. 960, 964 n. 10, 67 L.Ed.2d 17 (1981). Though the preeminent evil of ex post facto laws is the manifestation of excessive and arbitrary legislative power, courts have observed that they also weaken the system of justice by denying citizens the opportunity to know what the law requires and what punishment it exacts. “Through this prohibition, the Framers sought to assure that legislative Acts give fair warning of their effect and permit individuals to rely on their meaning until explicitly changed.” Weaver, 450 U.S. at 28-29, 101 S.Ct. at 964.2
The Supreme Court has expressly reserved judgment on whether retroactive application of parole guidelines may violate the ex post facto clause. United States Parole Commission v. Geraghty, 445 U.S. 388, 408, 100 S.Ct. 1202, 1214, 63 L.Ed.2d 479 (1980); United States, v. Addonizio, 442 U.S. 178, 184, 99 S.Ct. 2235, 2239, 60 L.Ed.2d 805 (1979). But its decisions on whether statutory changes affecting the *1166duration of a prisoner’s confinement violate the ex post facto prohibition provide some guidance. Weaver, the Court’s most recent pronouncement on the subject, establishes a two-part test for determining whether a penal statute is unconstitutional as an ex post facto law when applied to a petitioner whose crime was committed before the statute’s enactment: “it must be retrospective, that is, it must apply to events occurring before its enactment, and it must disadvantage the offender affected by it.” 450 U.S. at 29, 101 S.Ct. at 964 (footnotes omitted). I have no quarrel with Forman I’s holdings that the guidelines at issue were retrospectively applied and that they disadvantaged petitioner. Because the Parole Commission’s guidelines are administrative regulations rather than statutes, however, this court must consider a third and novel issue, whether the guidelines are “laws” within the meaning of the ex post facto clause.
My disagreement with current Third Circuit law centers around how to conduct this inquiry. In Geraghty I and Forman I, the court established an essentially statistical test — how many times do the Parole Commission’s parole determinations coincide with its guidelines? If the two factors— outcome and guideline — are congruent on a sufficient number of occasions, the court suggested, insufficient administrative flexibility would be demonstrated, and the “guidelines” would perforce be “laws.” Forman I, 709 F.2d at 862; Geraghty I, 579 F.2d at 266-67. The court has not yet stated what percentage of compliance transforms a guideline into a law.3 But were it to do so, we would have a simple numerical test. After that, the assumption seems to be, the district courts need only periodically check what the current figure is to see whether the Parole Commission is acting as a quasi-legislature. In my view this test is insufficient to determine whether an administrative guideline is a law within the meaning of the ex post facto clause; a statistical tendency does not clothe a guideline with the trappings of a law. As the Supreme Court stated in an analogous context:
No matter how frequently a particular form of clemency has been granted, the statistical probabilities standing alone generate no constitutional protections; a contrary conclusion would trivialize the Constitution. The ground for a constitutional claim, if any, must be found in statutes or other rules defining the obligations of the authority charged with exercising clemency.
Connecticut Board of Pardons v. Dumschat, 452 U.S. 458, 465, 101 S.Ct. 2460, 2464, 69 L.Ed.2d 158 (1981) (emphasis added).4
I certainly subscribe to what I understand to be the underlying premise of Geraghty I: that Congress cannot, by delegation, escape constitutional limitations on its power. 579 F.2d at 266. As I interpret this principle, it requires us to examine the scope of the power delegated to the Parole Commission by Congress, rather than the Commission’s actual practices. For if Congress intended that the Commission should have relatively unfettered discretion to grant or deny parole in individual cases, it does not matter — for purposes of ex post facto analysis5 — that in the great majority *1167of cases the Commission conforms its decisions to the narrower range of options found in the guidelines. It is the intent of Congress, not the practices of the Commission, that determines whether the guidelines are to have the force and effect of laws. Accordingly, I would examine the parole statute and its legislative history to determine whether it established substantial constraints on the Commission’s decisionmaking in individual cases. As I discuss in Part II, I find that Congress intended to place no such constraints on the discretion of the Parole Commission.
II.
Parole was first instituted in this country in the 1870’s; the Sixty-First Congress established the federal parole system in 1910. Prior to then, legislators viewed the primary purpose of incarceration as retribution and punishment, and prescribed the period of incarceration for each crime with specificity. United States v. Grayson, 438 U.S. 41, 45-46, 98 S.Ct. 2610, 2613, 57 L.Ed.2d 582 (1978). Parole was initially a system for showing clemency to prisoners. Those who behaved well were granted the privilege of supervised release. See Comptroller General of the United States, Federal Parole Practices: Better Management and Legislative Changes are Needed 2 (July 16, 1982) [hereinafter cited as '“Federal Parole Practices ”]. In 1930, Congress created the United States Board of Parole. Act of May 13, 1930, ch. 255, 46 Stat. 272 (1931). By that time, the legislators had begun to adopt a new penal philosophy. This “medical model” viewed the criminal as ill, imprisonment as a period of cure, and parole as the point at which the prisoner’s supervisors deemed him or her sufficiently rehabilitated to be safely released.6
This approach ... assumed that because it was impossible to accurately predict how long the cure would take, judges should set only the outside limits of the prison term. The parole board would assess the progress of the offender toward rehabilitation and decide when the offender should be released. Parole boards were granted wide discretion to make predictions about whether a cure had taken place and whether the offender could safely be released into society.
Federal Parole Practices at 2. See also, Warren v. United States Parole Commission, 659 F.2d at 189-190.
The theory that parole release is the proper result of rehabilitation has remained dominant.7 But in 1967, the President’s Commission on Law Enforcement and the Administration of Justice criticized the manner in which the federal parole system actually worked. The Commission’s two central concerns were that there were no explicit standards for deciding whether and when to parole a prisoner, and that the prisoner received no reliable, written information about his or her release. President’s Commission on Law Enforcement and the Administration of Justice, The Challenge of Crime in a Free Society 179-83 (1967). In response to these criticisms, the Parole Board in 1973 instituted *1168parole guidelines to be followed in making all federal parole decisions. 38 Fed.Reg. 31,942 (1973).8
In 1976, Congress made the guideline system a legislative requirement in the Parole Commission and Reorganization Act (“PCRA”), Pub.L. No. 94-233, § 2, 90 Stat. 219 (1978) (codified at 18 U.S.C. § 4203 (1982)). “This legislation was an effort to constrain and guide parole discretion through more rational, consistent, and equitable decisionmaking.” Federal Parole Practices at 3. In 1984, Congress, still dissatisfied with both the medical model and the system of sentencing and parole release that the model engendered, passed the Sentencing Reform Act of 1984, Pub.L. No. 98-473, 1984 U.S.Code Cong. & Ad. News (98 Stat.) 1837, 1987. This law provides for the future abolition of parole and the Parole Commission.9
Throughout the evolution of the parole system, courts have viewed parole decisions as essentially discretionary. As recently as 1979, the Supreme Court, discussing Nebraska’s parole system, wrote,
[i]n parole releases, like its siblings probation release and institutional rehabilitation, few certainties exist. In each case, the decision differs from the traditional mold of judicial decisionmaking in that the choice involves a synthesis of record facts and personal observation filtered through the experience of the decisionmaker and leading to a predictive judgment as to what is best both for the individual inmate and for the community.
Greenholtz v. Inmates of the Nebraska Penal and Correctional Complex, 442 U.S. 1, 8, 99 S.Ct. 2100, 2104, 60 L.Ed.2d 668 (1979) (footnote omitted).
In my view, Congress’ purpose in passing the PCRA was to promote uniformity in parole decisionmaking and to moderate, though not eliminate, the influence of the medical model by emphasizing notions of just punishment. The PCRA shifted ultimate parole decisionmaking power from individual parole officials to a central Parole Commission, with five regional offices. Congress believed that although “[tjhere is no body of competent empirical knowledge upon which parole decision-makers can rely, ... it is important for the parole process to achieve an aura of fairness by basing determinations of just punishment on comparable periods of incarceration for similar offenses committed under similar circumstances.” H.R.Conf.Rep. No. 838, 94th Cong., 2d Sess. 26, reprinted in 1976 U.S.Code Cong. & Ad.News 351, 358 [hereinafter cited as “Conference Report”].10
In order to bring about greater parity in punishment, Congress required the Parole Commission to establish guidelines and procedural rules for parole determinations. The guidelines are intended to be followed by hearing examiners in making recommendations to one of the five regional commissioners, who may adopt the recommendation and enter a decision, or make his or her own decision. Conference Report at 22; 1976 U.S.Code Cong. & Ad.News at 354-55; 18 U.S.C. § 4203(a), (b), (c) (1982).
*1169Notwithstanding the goal of reducing the disparity of prison terms for like offenders and offenses, and the establishment of a guideline system toward this end, Congress did not establish substantial constraints on the Parole Commission’s discretion in individual cases. In discussing parole release criteria, the Conference Committee stated,
First, it is the intent of the Conferees that the Parole Commission reach a judgment on the institutional behavior of each prospective parolee____
Second, it is the intent of the Conferees that the Parole Commission review and consider both the nature and circumstances of the offense and the history and characteristics of the prisoner.
Conference Report at 25; 1976 U.S.Code Cong. & Ad.News at 358 (emphasis added).
In addition to providing that parole depends in the first instance on the individual’s prison behavior and individual characteristics, and the circumstances of the offense, the PCRA also allows the Commission to grant or deny parole notwithstanding the guidelines, as long as there is good cause to do so and the prisoner is given a written explanation of the decision. 18 U.S.C. § 4206(c) (1982). The Conference report states that,
[t]he definition of what constitutes good cause to go outside the established guidelines can not be a precise one, because it must be broad enough to cover many circumstances____
... By focusing on the justifications for exceptions to the guidelines, subsequent administrative review ... will be facilitated and there will be more uniformity____
If decisions to go above or below parole guidelines are frequent, the Commission should reevaluate its guidelines.
Conference Report at 27; 1976 U.S.Code Cong. & Ad.News at 359-360.
To me, the PCRA’s provisions, viewed together, indicate that the Congress took steps toward more uniformity of prison terms, and away from the medical model, but was unwilling to abandon altogether the discretionary determinations about individual prisoners that have historically characterized the parole system.11 There are no sanctions for not following the guidelines. Instead, Congress established a mechanism that both requires a set of guidelines, and encourages their alteration in the event the Parole Commissioners do not follow them.12 In my view, this deliberate flexibility precludes characterizing the guidelines as laws within the meaning of the ex post facto clause. See also Ruip v. United States, 555 F.2d 1331 (6th Cir. 1977); Roth v. United States Parole Com-
*1170mission, 724 F.2d 836 (9th Cir.1984); Dufresne v. Baer, 744 F.2d 1543 (11th Cir.1984).
III.
We have now ruled twice that the Parole Commission’s practice of applying its regular parole guideline revisions to all current prisoners instead of only to those prisoners who committed offenses on or after the effective date of the latest revision, may violate the Constitution because “if applied without substantial flexibility, the parole guidelines constitute ‘laws’ within the meaning of the ex post facto clause.” For-man I, 709 F.2d at 862; see also Geraghty I, 579 F.2d at 267-68. Notwithstanding these two rulings, we have never actually held that the Parole Commission applies its guidelines without the requisite flexibility and do not do so in this case. Indeed, the meaning of the phrase “substantial flexibility” is so ambiguous that I cannot foresee when we would ever do so. Moreover, as the majority notes, the Forman I test has been rejected by every other circuit that has addressed the issue. See Majority Opinion at footnote 3. To permit such an ambiguity to remain at the center of the daily operation of the United States parole system, in this circuit, would ill serve the parole commissioners and hundreds of parole officers who seek to do an honest job in complying with the law and upholding the Constitution, It would also-ill serve the district courts of this circuit, which may now have to reexamine Parole Commission practices each time a prisoner who has been denied parole asserts that he or she has not been treated with the necessary flexibility.
It is for these judicial policy reasons, as well as my belief that the Parole Commission’s guidelines are not “laws” within the meaning of the ex post facto clause, that I concur only in the judgment of Forman II, and urge that the rationale of Forman I, and its predecessor, Geraghty I, be considered in banc.
. See The Federalist No. 44 (J. Madison) & No. 84 (A. Hamilton). See generally Warren v. United States Parole Commission, 659 F.2d 183, 186-89 (D.C.Cir.1981).
. Critical to relief under the Ex Post Facto Clause is not an individual’s right to less punishment, but the lack of fair notice and governmental restraint when the legislature increases punishment beyond what was pre-
scribed when the crime was consummated. Thus, even if a statute merely alters penal provisions accorded by the grace of the- legislature, it violates the Clause if it is both retrospective and more onerous than the law in effect on the date of the offense.
Weaver, 450 U.S. at 30-31, 101 S.Ct. at 965 (footnote omitted).
. The Geraghty I court did, however, suggest numerical possibilities:
[I]f in practice the parole authorities found good cause to deviate from the guidelines in 60% of the cases, for example, it might be argued that their discretion is, in fact, unfettered. However, ... [one] decision of this Court took judicial notice of estimates of compliance with the guidelines ranging from 88% to 94%. It thus appears that the "channel for discretion” provided by the guidelines is in actuality an unyielding conduit.
Geraghty /, 579 F.2d at 267 (footnote omitted).
. In Dumschat, the Court held that the fact that the Connecticut Board of Pardons granted approximately three-fourths of the applications for communication of life sentences created no "liberty interest” in life-terms inmates that would require the Board to explain its reasons for denying applications. The Court also noted that "[a] commutation decision ... shares some of the characteristics of a decision whether to grant parole." 452 U.S. at 464, 101 S.Ct. at 2464.
. It is important to keep in mind that we are not dealing with a contention that the delegation of authority to promulgate guidelines was uncon*1167stitutional, or with a challenge to either the validity of the guidelines on their face or their prospective application. We are concerned only with their retroactive revision.
. Federal Parole Practices at 2; Grayson, 438 U.S. at 45-46, 98 S.Ct. at 2613.
. In 1978, the Supreme Court observed that "[t]he evolutionary development of sentencing and incarceration practices continues to engage attention. Increasingly, there are doubts concerning the validity of the earlier, uncritical acceptance of the rehabilitation model.” United States v. Grayson, 438 U.S. at 47 n. 6, 98 S.Ct. at 2614 n. 6 (citations omitted). Subsequently, the Court, while conceding that "[n]o ideal, error-
free way to make parole-release decisions has been developed,” insisted that,
[i]t is important that we not overlook the ultimate purpose of parole which is a component of the long range objective of rehabilitation. The fact that anticipations and hopes for rehabilitation programs have fallen far short of expectations of a generation ago need not lead states to abandon hopes for those objectives.... The objective of rehabilitating convicted persons to be useful, law-abiding members of society can remain a goal no matter how disappointing the progress.
Greenholtz v. Inmates of the Nebraska Penal and Correctional Complex, 442 U.S. 1, 13, 99 S.Ct. 2100, 2107, 60 L.Ed.2d 668 (1979).
. The Parole Board’s guidelines for parole release consideration set out customary prison terms for a list of offenses. The terms vary in duration, according to both the severity of the offense and the attributes of the offender, including employment history and prior criminal records. See Forman I, 709 F.2d at 854 n. 5.
. Pub.L. No. 98-473, § 218(a)(5), 1984 U.S.Code Cong. & Ad.News (98 Stat.) at 2027. Current parole laws, 18 U.S.C. ch. 311 (1982), are to remain in effect for five years after the Sentencing Reform Act’s effective date. Pub.L. No. 98-473, § 235(b)(1)(A), 1984 U.S.Code Cong. & Ad.News (98 Stat.) at 2032. See S.Rep. No. 225, 98th Cong., 1st Sess. 189, reprinted in 1984 U.S. Code Cong. & Ad.News 3182, 3372. For discussion of Congress' reasons for not retaining the Parole Commission, see S.Rep. at 52-58, 164-165; 1984 U.S.Code Cong. & Ad.News at 3236-3241, 3347-3348.
. The Conference Report stated that parole serves various objectives of the correctional system: ”[i]n the first instance, parole has the practical effect of balancing differences in sentencing policies and practices between judges and courts____ In performing this function, the parole authority must have in mind some notion of the appropriate range of time for an offense which will satisfy the legitimate needs of society to hold the offender accountable for his own acts." Conference Report at 19; 1976 U.S.Code Cong. & Ad.News at 352.
. The primary reason for the 1984 parole reforms was the failure of the previous reforms, including the PCRA, to eliminate disparities in the prison terms of similar offenders:
Even two such offenders who are sentenced to terms of imprisonment for similar offenses may receive widely differing prison release dates____
These disparities ... can be traced directly to the unfettered discretion the law confers on those judges and parole authorities responsible for imposing and implementing the sentence. This sweeping discretion flows from the lack of any statutory guidance or review procedures to which courts and parole boards might look.
S.Rep. No. 225, 98th Cong., 1st Sess. 38 (1983), reprinted in 1984 U.S.Code Cong. & Ad.News 3182, 3221 (emphasis added) (footnote omitted).
The Committee also pointed out that, in part because "the parole guidelines frequently fail in practice to achieve their goal of reducing unwarranted sentencing disparities,” S.Rep. No. 225 at 48; 1984 U.S.Code Cong. & Ad.News at 3231, the Parole Commission’s efforts in this direction "unfortunately contribute to a second grave defect of present law: no one is ever certain how much time a particular offender will serve if he is sentenced to prison." S.Rep. No. 225 at 49; 1984 U.S.Code Cong. & Ad.News at 3232. The Congress clearly does not view the Parole Commission guidelines as having the force and effect of laws.
. One problem with Forman /'s "substantial flexibility” test is that it does not recognize that conformity with the guidelines may be the consequence, not the cause, of individual decisions, as guidelines are revised to reflect actual practice. Cf. Benedict v. United States Parole Commission, 569 F.Supp. 438, 444 (E.D.Mi.1983).