OPINION

The Oregon State Board of Parole and PosL-Prison Supervision (“the Board of Parole”) found that Robert Lewis Himes had violated the terms of his parole and ordered him re-incarcerated to serve twenty-nine and one half additional years. The Board of Parole based its decision on parole regulations more onerous than those in place at the time Himes committed the offense for which he was incarcerated. The question for decision is whether that determination violated the Ex Post Facto Clause of the United States Constitution. U.S. Const, art. I, § 10. We conclude that it did and therefore reverse the district court’s denial of Himes’ petition for habeas corpus.

I. BACKGROUND

In 1978 Himes sexually assaulted a woman and a thirteen-year old girl at knife point. He was arrested and pleaded guilty to attempted rape, first degree rape, first degree sodomy, and first degree robbery. The state trial court imposed consecutive sentences, totaling 70 years. Fifteen years later, on April 24, 1994, the Board of Parole released Himes on parole, a decision that, according to Himes’ parole offi cer, “created a considerable stir” in the community.

A. Revocation of Himes’ Parole

On parole, Himes was subject to several “Special Conditions”: He was prohibited from having contact with minor females, required to complete successfully a sex offender treatment program, and directed to submit to random polygraph examinations.

On June 28, 1994, Himes took his first polygraph examination. During both a pre-test interview and the polygraph examination itself Himes denied haying sexual encounters with women other than his wife, following women, or entertaining sexual fantasies about women. The polygraph examiner concluded that Himes’ responses were not truthful.

In the post-test interview that followed, Himes made several admissions that contradicted his previous claims. He explained that although he loved his wife and had not engaged in sexual relations with another woman, he sometimes took unnecessary routes in order to be near attractive women while driving and shopping. He said he did this to exchange glances with the women. Following such eye contact he fantasized about sexual relationships with women.

After the examination, the polygraph examiner submitted a report to Himes’ parole officer. Based on the examiner’s report as well as on the conclusion that Himes had failed to report his behavior and fantasies' to his parole supervision team, Himes’ parole officer prepared a Violation Report recommending that the Board of Parole revoke Himes’ parole.

On August 1, 1994, less than four months after his release, the Board of Parole formally revoked Himes’ parole. After unsuccessful appeals of the revocation decision in state court, Himes filed in federal district court a petition for a writ of habeas corpus, challenging the parole revocation. The district court denied the writ, and this court affirmed in an unpublished opinion. Himes v. Thompson, 225 F.3d 662 (9th Cir.2000).

B. Denial of Rerelease

According to regulations in place at the time of Himes’ parole revocation, Himes was entitled to rerelease from parole after 90 days of re-incarceration unless the Board of Parole made a finding of “aggravation.” Or. Admin. R. 255 -075-0079(1,10) (1994); Or. Admin. R. 255 -075-0096(1) (1994). Upon a finding of aggravation, the Board of Parole could, according to the 1994 regulations, deny Himes’ rerelease altogether and require that Himes serve to his statutory “good time” date. Id.

On October 27, 1994, approximately three months after revoking Himes’ parole, the Board of Parole held a hearing to determine Himes’ eligibility for rerelease according to these criteria. Finding aggravation, the Board of Parole denied Himes rerelease and scheduled his next review for May, 2024, two years shy of his projected statutory “good-time” date. In other words, as a result of the parole violations, Himes will be required to serve a minimum of twenty-nine and one half additional years in prison. As we will develop, application of the 1978 regulations would quite likely have led to a different result.

C. Procedural History

After the Board of Parole denied Himes rerelease, Himes appealed. The Oregon Court of Appeals denied the appeal without opinion, and the Oregon Supreme Court denied review. Himes v. Bd. of Parole, 140 Or.App. 644 , 917 P.2d 77 (1996); Himes v. Bd. of Parole, 324 Or. 322 , 927 P.2d 598 (1996). Himes next filed a petition for habeas corpus relief in United States District Court, raising several claims. The district court denied the petition, rejecting some claims on the basis of procedural default and others on the merits.

Himes currently appeals only the denial of his contention that, by applying 1994 regulations to increase Himes’ incarceration period for crimes committed in 1978, the Board of Parole violated the Ex Post Facto Clause, U.S. Const, art. I, § 10. Subsequent to oral argument, we asked for and received supplemental briefing regarding the impact of the recent Oregon case of Gonzalez v. Washington, 182 Or.App. 112 , 47 P.3d 537 (2002), on the instant appeal. We now reverse.

II. PRELIMINARY CONSIDERATIONS

A. Habeas Standard

Himes filed his habeas petition on October 2, 1997, so we review his petition under the provisions of the Antiterrorism and Effective Death Penalty Act (“AED-PA”), Pub.L. No. 104-132, 110 Stat. 1214 (Apr. 24, 1996). Our review of the district court’s denial of habeas corpus relief is de novo. Bowen v. Hood, 202 F.3d 1211, 1218 (9th Cir.2000).

AEDPA requires that we give considerable deference to state court decisions. The state court’s factual findings are presumed correct. 28 U.S.C. § 2254 (e)(1). We are bound by a state’s interpretation of its own laws. Souch v. Schaivo, 289 F.3d 616, 621 (9th Cir.2002), cert. denied, 537 U.S. 859 , 123 S.Ct. 231 , 154 L.Ed.2d 98 (2002). We also defer to the state court’s determination of the federal issues unless that determination is “contrary to, or involved an unreasonable application of, clearly established Federal law.” Lockyer v. Andrade, 538 U.S.-, 123 S.Ct. 1166, 1172 , 155 L.Ed.2d 144 (2003). A state court decision is “contrary to” federal law “if the state court applies a rule that contradicts the governing law set forth in [Supreme Court] cases or if the state court confronts a set of facts that are materially indistinguishable from a decision of [the Supreme Court ] and nevertheless arrives at a result different from [Supreme Court] precedent.” Id. at 1173 (quoting Williams v. Taylor, 529 U.S. 362, 405-06 , 120 S.Ct. 1495 , 146 L.Ed.2d 389 (2000) (O’Connor, J., concurring)) (internal quotation marks omitted). A state court decision involves an unreasonable application if it correctly identifies the governing rule but unreasonably applies it to a new set of facts, see id. at 1174 (citing Williams, 529 U.S. at 413 , 120 S.Ct. 1495 ), or fails to extend a clearly established legal principle to a new context in a way that is unreasonable, see Bailey v. Newland, 263 F.3d 1022, 1028 (9th Cir.2001) (citing Van Tran v. Lindsey, 212 F.3d 1143, 1150 (9th Cir.2000)). An unreasonable application of federal law is something more than an incorrect or even clearly erroneous application; the application must be “objectively unreasonable.” Andrade , — U.S. at-, 123 S.Ct. at 1174 . Although only Supreme Court law is binding on the states, our Circuit precedent remains relevant persuasive authority in determining whether a state court decision is objectively unreasonable. See Duhaime v. Ducharme, 200 F.3d 597, 600-01 (9th Cir.1999).

Application of these standards is significantly impeded where, as here, the state court supplies no reasoned decision. Delgado v. Lewis, 223 F.3d 976, 981 (9th Cir.2000). We “cannot perform our evaluation under the models suggested by Justice O’Connor in Williams,” because we have “no basis other than the record for knowing whether the state court correctly identified the governing legal principle or was extending the principle into a new context.” See Delgado, 223 F.3d at 981-82 . Delgado therefore instructs us to perform an “independent review of the record” to ascertain whether the state court decision was objectively unreasonable. Id. at 982 . Independent review of the record is not de novo review of the constitutional issue, but rather, the only method by which we can determine whether a silent state court decision is objectively unreasonable. See id.; accord Bell v. Jarvis, 236 F.3d 149, 163 (4th Cir.2000) (en banc).

In this case, our review of the record requires an in-depth analysis of Oregon parole regulations to determine whether the change in regulations violated the Ex Post Facto Clause. Although the ultimate ex post facto question is a matter of federal law, see Lindsey v. Washington, 301 U.S. 397, 400 , 57 S.Ct. 797 , 81 L.Ed. 1182 (1937), it is axiomatic that we defer to Oregon courts regarding predicate questions related to the interpretation of the parole regulations. See Souch, 289 F.3d at 621 . Our problem here is that certain aspects of Himes’ claims involve state law questions which have not been directly addressed either in the proceedings below or in any published Oregon opinions.

We therefore must undertake our own analysis of the embedded state law question. See Luna v. Cambra, 306 F.3d 954, 960, 962-65 (9th Cir.2002) (undertaking detailed analysis of California evidentiary law to determine whether defendant was prejudiced by trial counsel’s failure to proffer certain evidence, as part of the “independent review of the record” necessitated by lack of state court rationale), as amended by 311 F.3d 928 (9th Cir.2002); Wilcox v. McGee, 241 F.3d 1242, 1245-46 (9th Cir.2001) (per curiam) (determining whether double jeopardy claim was valid by examining whether indictment “was fatally defective under state law”).

We presume, of course, that state courts “know and follow the law” and give state court decisions the “benefit of the doubt.” Woodford v. Visciotti 537 U.S. 19 , 123 S.Ct. 357, 360 , 154 L.Ed.2d 279 (2002). In analyzing the embedded state law question, we therefore vigilantly search for an interpretation of the state law question which would avoid attributing constitutional error to the state court. But we stop short of adopting an implausible or strained interpretation.

B. Procedural Default

The warden initially argued that Himes was procedurally barred from raising the ex post facto claim. The warden abandoned that claim at oral argument, however, and we therefore do not reach it.

III. EX POST FACTO VIOLATION

The Constitution provides that “No State shall ... pass any ... ex post facto Law.” U.S. Const. art. I, § 10. The Ex Post Facto Clause “is aimed at laws that retroactively alter the definition of crimes or increase the punishment for criminal acts.” Souch, 289 F.3d at 620 (quoting Cal. Dep’t of Corr. v. Morales, 514 U.S. 499, 504 , 115 S.Ct. 1597 , 131 L.Ed.2d 588 (1995)).

The Ex Post Facto Clause applies by its terms to “laws.” As such, the clause reaches “every form in which the legislative power of a state is exerted,” including “a regulation or order.” Ross v. State of Oregon, 227 U.S. 150, 162-63 , 33 S.Ct. 220 , 57 L.Ed. 458 (1913). Oregon’s Board of Parole “through its rules governing release dates” affects “the amount of freedom or punishment that a prisoner in fact receives.” Williams v. Bd. of Parole, 98 Or.App. 716 , 780 P.2d 793, 795 (1989) (citation omitted). So recognizing, Oregon courts have held that Oregon Board of Parole regulations are properly subject to ex post facto analysis, see id., as have we, see Flemming v. Bd. of Parole, 998 F.2d 721, 725-27 (9th Cir.1993). The warden does not contend to the contrary.

The standard for determining whether a law or regulation violates the Ex Post Facto Clause has two components. First, the regulations must have been applied retroactively to the defendant. See Weaver v. Graham, 450 U.S. 24, 29 , 101 S.Ct. 960 , 67 L.Ed.2d 17 (1981). Second, the new regulations must have created a “sufficient risk” of increasing the punishment attached to the defendant’s crimes. Morales, 514 U.S. at 509 , 115 S.Ct. 1597 . Both conditions are met here. The Oregon courts’ contrary decision was objectively unreasonable.

A. Retroactive Application

In determining retroactivity, the “critical question is whether the [regulations] changef ] the legal consequences of acts completed before [the] effective date[of the regulations.]” Weaver, 450 U.S. at 31 , 101 S.Ct. 960 , 67 L.Ed.2d 17 (1981). Parole eligibility affects the length of a prison term and therefore affects the measure of punishment attached to the original crime. See id. at 31-32 , 101 S.Ct. 960 . In this case, although the new parole regulations were applied in a 1994 hearing, they affected the punishment Himes received for crimes committed in 1978. Therefore, the regulations were applied retroactively. See id.; see also Flemming, 998 F.2d at 722-24 (applying the analysis of Weaver, 450 U.S. at 32 , 101 S.Ct. 960 , and holding that parole regulations enact ed after the date of conviction, but before the Board of Parole hearing in which they were applied, were applied retroactively). Again, the warden does not contend to the contrary.

B. Substantial Disadvantage

The Ex Post Facto Clause is violated if a change in law creates “a sufficient risk of increasing the measure of punishment attached to the covered crimes.” Morales, 514 U.S. at 510 , 115 S.Ct. 1597 (emphasis added). The risk is apparent from the face of the changed regulations if, after comparing the two regulatory schemes as a whole, it is apparent that the new regulations are detrimental. Whether an individual can show definitively that he would have received á lesser sentence is not determinative. See Miller v. Florida, 482 U.S. 423, 432 , 107 S.Ct. 2446 , 96 L.Ed.2d 351 (1987); Nulph v. Faatz, 27 F.3d 451, 455-56 (9th Cir.1994). In other words, “[t]he inquiry looks to the challenged provision, and not to any special circumstances that may mitigate its effect on the particular individual.” Weaver, 450 U.S. at 33 , 101 S.Ct. 960 .

Our inquiry focuses on the parole regulations. The warden emphasizes that in both 1978 and in 1994, the Board of Parole had statutory authority to require an inmate to serve his full sentence. See Or. Rev.Stat. § 144.343(2)(b). That authority, however, was channeled by Or.Rev.Stat. § 144.395, requiring the Board of Parole to “adopt rules ... relating to the rerelease of persons whose parole has been revoked.” We must therefore consider whether the 1994 regulations thus adopted, as compared to the 1978 regulations, created a significant risk of a more onerous sentence.

We focus on the regulations even though neither the 1978 nor the 1994 regulations completely circumscribed the Board of Parole’s ability to deny Himes’ rerelease pursuant to Or.Rev.Stat. § 144.343(2)(b). The principle that legislation that sufficiently channels official discretion may violate ex post facto principles, even if some discretion is retained, was definitively established in Miller, 482 U.S. at 425 , 107 S.Ct. 2446 .

In Miller , the Supreme Court considered a change in Florida’s sentencing guidelines. See id. Under the Florida guidelines, the trial judge could fix a sentence within a presumptive range without providing a written explanation, but had to provide “clear and convincing” reasons for deviating from the presumptive range. Id. at 426 , 107 S.Ct. 2446 . Between the time Miller committed his offense and the time he was sentenced, changes in the guidelines increased the “presumptive” sentence range applicable to his offenses. See id. at 426-27 , 107 S.Ct. 2446 . The trial judge applied the new, more onerous guidelines and sentenced Miller within the new presumptive range. See id. at 427-29 , 107 S.Ct. 2446 . Miller could not “show definitively that he would have gotten a lesser sentence” under the old guidelines, because the trial judge could have reached the same result by departing from the presumptive range. Id. at 432 , 107 S.Ct. 2446 . Nonetheless, the Court determined that Miller “plainly [had] been substantially disadvantaged by the change in sentencing laws.” Id. (internal citation and quotation omitted). The guidelines did more than “provide flexible ‘guideposts’ they created a “high hurdle” before the judge could exercise discretion. Id. at 435 , 107 S.Ct. 2446 . The Court therefore held that the new sentencing rules exposed Miller to a substantial risk of an increased sentence, in violation of the Ex Post Facto Clause.

As Miller demonstrates, changes in sentencing rules can violate the Ex Post Facto Clause when the rules sufficiently circumscribe official discretion, even if the change does not automatically lead to a more onerous result than would have occurred under the prior law. See Miller, 482 U.S. at 432-33 , 107 S.Ct. 2446 . Similarly, a change in law that increases the mandatory minimum sentence attached to a crime increases the measure of punishment even if a petitioner cannot show that he would have received a more lenient sentence under the old scheme. See Lindsey, 301 U.S. at 400-01 , 57 S.Ct. 797 .

This is not to say, however, that every “mechanical change[ ]” in parole procedure violates the Ex Post Facto Clause. See Morales, 514 U.S. at 508-09 , 115 S.Ct. 1597 (risk that less frequent parole hearings could produce an increased term of confinement was “speculative” and “attenuated,” especially in light of the parole board’s practice of granting earlier review in appropriate cases). Whether or not a legislative adjustment creates a “sufficient” rather than an “attenuated” risk of increased punishment is largely “a matter of ‘degree.’” Id. at 509 , 115 S.Ct. 1597 (citation omitted).

Turning to the Board of Parole’s regulations, we find that the change in Oregon’s parole regulations created a substantial— rather than attenuated or speculative— risk of increasing Himes’ incarceration, and therefore violated the Ex Post Facto Clause.

1. The Change In the Method For Determining “Aggravation.”

The change in the parole regulations could have disadvantaged Himes in two interrelated ways. First, in making the finding of aggravation necessary to its decision to deny his rerelease, the Board of Parole relied on the 1994, rather than the 1978, methodology for determining aggravation. Although we are troubled by this application, giving due deference under AEDPA to the conclusions of the Oregon state courts we find that this change, standing alone, created no ex post facto violation that would support granting the habeas petition.

The 1994 Board of Parole made its aggravation finding by comparing Himes’ conduct to a list of factors (“Exhibit H”), each of which indicated aggravation. Or. Admin. R. 255 -075-0096(1) (1994). In denying rerelease, the Board of Parole relied on two of those factors: (1) “Repetition of type [of] conduct associated with commitment of offense or past conditions”; and (2) “Less than three months until first difficulty.”

In 1978, by contrast, the Board of Parole would have compared Himes’ conduct to a different list of non-exclusive aggravation factors contained in “Exhibit G.” Former Or. Admin. R. 254 -70-042(3) (1978). While similar to Exhibit H, Exhibit G to the 1978 regulations did not contain one of the factors relied on by the Board in its aggravation finding, namely, “Repetition of type [of] conduct associated with commitment offense or past conditions.” Exhibit G did contain as a factor, “Less than 6 months to first difficulty,” so even under the 1978 regulations at least one aggravating factor would have applied. Himes maintains, however, that had the Board applied the 1978 regulations, it would have been less likely to make a finding of aggravation, because one rather than two aggravation factors would have been pertinent.

The warden counters that any disadvantage to Himes caused by the addition of the “repetition” factor in 1994 is fully offset: In 1978, Exhibit G contained a factor involving “resistance to parole supervision” which, the warden argues, is the counterpart of the “repetition” factor applied to Himes in 1994. Therefore, the warden reasons, under either scenario the Board would have found two aggravating factors applicable to Himes’ conduct.

We do not share the warden’s certainty that the Board would have found that Himes was “resisting” parole supervision. Although Himes did violate certain parole conditions, the record also reflects that he fervently sought employment, actively sought guidance from his pastor and parole officer, and voluntarily revealed his questionable conduct after probing by the polygraph officer, all in an effort to comply with parole. As the “resisting parole” factor does focus on somewhat different considerations than the “repetition” factor, there is the distinct possibility that, although the Board of Parole perhaps could have found the “repetition” factor applicable, it would not have done so.

Nonetheless, we could not say that Oregon unreasonably applied the federal Ex Post Facto Clause to the change in factors if we considered that change in isolation. The Board of Parole need not find more than one factor applicable before making a finding of aggravation. And, because the 1978 factors were non-exclusive, the Board could have made an aggravation finding with regard to the repetition of conduct related to the crime, even if that consideration had not been identified as a listed factor. Accord Moore v. Oregon State Bd. of Parole, 54 Or.App. 369 , 635 P.2d 3, 5 (1981) (holding, in a related context, that the Board may rely on aggravation factors not listed in its guidelines). Therefore, even if we assume that the Board in 1978 would have found only one enunciated aggravating factor applicable, the Board could well have made an aggravation finding anyway.

The Oregon Court of Appeals has held, in two analogous situations, that a change in non-exclusive “factors” did not alter the Board of Parole’s ability to exercise its discretion to such a degree as to warrant finding an ex post facto violation. See Thierman v. Bd. of Parole, 134 Or.App. 304 , 894 P.2d 1250, 1252 (1995); Carroll v. Bd. of Parole, 124 Or.App. 180 , 859 P.2d 1203, 1203 (1993). Giving those opinions the deference due under AEDPA and assuming that a similar analysis was applied here, we cannot say that this logic is an objectively unreasonable application of Supreme Court precedent. The Oregon courts could have determined that the factors did not sufficiently bind Board of Pa role discretion to trigger the principles of Miller , concluding instead that under Morales , the increased risk of punishment was not so large as to trigger Ex Post Facto Clause concerns. See Andrade, — U.S. at-n. 1, 123 S.Ct. at 1174 n. 1 (holding that where state court is presented with issue falling between two Supreme Court precedents, state court’s decision to apply one precedent over the other is neither contrary to nor an unreasonable application of Supreme Court law).

2. The Change in the Presumption of Punishment

The change in the applicable factors did not, however, stand alone. That change was accompanied by a fundamental alteration in the regulatory scheme. Application of the 1994 regulations as a whole substantially increased the risk of a longer sentence by instructing the Board of Parole, once it made a finding of aggravation, to deny rerelease entirely, thereby requiring Himes to serve out the remainder of his sentence. See Nulph, 27 F.3d at 455 (explaining that ex post facto inquiry requires an evaluation of the regulatory scheme “in toto”) (quoting Dobbert, 432 U.S. at 294, 97 S.Ct. 2290 ). In contrast, the 1978 regulations would almost surely have resulted in a more moderate re-incarceration period. Because our conclusion requires a careful analysis of the relevant regulations, we set them out in some detail.

(a) The Regulations

The 1994 regulations applicable to Himes compelled the board to deny rere-lease upon a finding of aggravation. Or. Admin. R. 255 -075-0079 (1994) states in relevant part:

In turn, Or. Admin. R. 255 -075-0096(1) (1994), entitled “Denial of Rerelease Consideration” (emphasis added), states:

The 1994 regulations gave the Board of Parole a binary choice between the 90 days contemplated in Or. Admin. R. 255 -075-0079(1) (1994), and the outright denial of rerelease contemplated by Or. Admin. R. 255 -075-0079(10) (1994) and 255-075-0096(1) (1994). Once the Board found aggravation, and, as a result, determined to extend the revocation period beyond the 90 days otherwise applicable, it had no choice but to deny Himes rerelease on parole.

Unlike the 1994 regulations, the 1978 regulations relevant to Himes’ parole rere-lease did not mandate outright denial of rerelease as the only available aggravation penalty. Former Or. Admin. R. 254 -70-042 (1978) (“Return for Technical Violation”) stated:

While Or. Admin. R. 254 -70-042 allows the Board of Parole to deviate from the presumptive ranges, it contains no requirement that the Board deny rerelease altogether once it so deviates. The “unless” clause allows deviation from the presumptive ranges for either aggravation or mitigation, indicating that the clause operates to allow the board to increase or decrease the resulting incarceration period, commensurate with the degree of aggravation or mitigation found. Additionally, there was no reference in the 1978 regulations, as there was in 1994, to denying rerelease altogether, or to serving to the statutory good time date. While those actions may have been permissible under the 1978 regulations, there is no reasonable reading of the regulations mandating complete denial of rerelease if the Board wished, upon a finding of aggravation, to impose an incarceration period longer than the presumptive terms set out in former Or. Admin.' R. § 254-070-042 (1978). So the 1978 regulations created a continuum of sanctions, depending on the level of aggravation or mitigation, not a binary choice.

The two different sets of regulations therefore produce startlingly divergent results. In 1994, if the Board wished to impose an aggravation penalty, it was compelled to re-incarcerate an inmate for the remainder of his term — in Himes’ case, over twenty-nine and a half years. In 1978, the Board would have chosen from a continuum of sanctions: anywhere from a few months to the entirety of the prison term. For prisoners like Himes, for whom the remaining prison term was quite lengthy, the Board of Parole in 1978 would likely have imposed the entire prison term only under extraordinary circumstances. So the change in regulatory regime, viewed in its entirety, significantly increased the possibility of serving a lengthy re-incarceration period under the new regime.

The warden urges otherwise, maintaining that Or. Admin. R. 255 -075-0079(10) (1994) permitted sanctions that exceeded 90 days, yet fell short of re-imprisonment for the entire statutory term. This reading of the regulations is not supported by Oregon case law and is simply not plausible.

First, Or. Admin. R. 255 -075-0096(1) (1994) provides that a finding of aggravation “requires the affirmative vote of a majority of members” — a heightened voting standard. Compare with Or. Admin. R. 255 -75-0096(3) (1994) (majority vote of full board not required if offender is re-incarcerated within guideline ranges set forth in Or. Admin. R. 255 -075-0096(1-2) (1994)). This heightened voting standard indicates the gravity of the mandatory penalty imposed under Or. Admin. R. 255 -075-0096(1) (1994). It is unlikely the regulations would impose such a heightened standard to extensions of periods of incarceration for a few months beyond the presumptive period.

Second, unlike the range specified in 255-075-0079(l)(a) (1994), which allows re-imprisonment for “up to 90 days,” section 255-075-0096(1) (1994) simply states that if the Board should “deny” rerelease the prisoner must serve “to the statutory good time date.” Id (emphasis added). The absence of the “up to” language suggests that only one outcome is possible.

Subsequent amendments to 255-075-0096(1) reinforce this point. In 1997, the regulation was changed to read:

Id. (emphasis added). The amended language was added to “conform the rules to practice and eliminate confusion or unintended early release.” Or. Bull., April 1, 1997, at 100. “Unintended early release” could result when a prisoner was released on rather than before his statutory good time date. See Erbs v. Bd. of Parole, 90 Or.App. 253 , 752 P.2d 318, 319-20 (1988) (if prisoner serves until his “good time” date, he is unequivocally entitled to release upon good time release date, and Board is thereby divested of all post-prison supervision authority). To avoid this result, the Board of Parole, “following the Erbs decision” adopted “a practice of setting a release date in advance of the good time date, in order to assure that the inmate would serve some period of supervised parole before final discharge.” Bollinger v. Bd. of Parole, 142 Or.App. 81 , 920 P.2d 1111, 1113 (1996) (“Bollinger I”) aff'd by Bollinger v. Bd. of Parole, 329 Or. 505 , 992 P.2d 445 (1999) (“Bollinger II ”).

The 1997 amendment thus indicates that the board felt that it needed to “amendf ] its rules relative to setting a release date when parole violators are denied rere-lease,” to clarify its authority to release such inmates a few days prior to, rather than on, the statutory good time date. Or. Bull, April 1, 1997, at 100. If the warden’s interpretation of the 1994 regulations were credited, this amendment would not have been necessary, as any period short of the statutory good time date would already have been valid. Moreover, as Bol-linger I and the regulatory changes make clear, the changes codified the practice of setting a rerelease date just shy of the statutory good time date. See Bollinger I, 920 P.2d at 1113 (release date set one month in advance of good time date to avoid early release); Kendrick v. Bd. of Parole, 146 Or.App. 756 , 934 P.2d 607, 607 (1997) (two days prior to good time date) Kessler v. Bd. of Parole, 145 Or.App. 584 , 931 P.2d 801, 802 (1997) (same); Luckey v. Bd. of Parole, 150 Or.App. 480 , 946 P.2d 361, 361 (1997) (same). That in none of these eases was any intermediate release date set (other than the short period necessary to avoid the Erbs result) indicates an understanding that “denial of rere-lease” means precisely what it said, and was mandatory, not permissive.

Woolstrum v. Bd. of Parole, 141 Or.App. 332 , 918 P.2d 112, 116 (1996), confirms our reading of Or. Admin. R. 255 -075-0096(1-2). Woolstrum’s parole was revoked. In a separate future disposition hearing in June, 1994, Woolstrum was denied rere-lease pursuant to Or. Admin. R. 255 -075-0096 and had his rerelease date set at his statutory good time date. See Woolst-rum, 918 P.2d at 114 . Woolstrum argued that because he received no notice of the future disposition hearing (although he had been notified of the prior revocation hearing), his substantial rights were violated. The court agreed, stating: “The Board proceeded under OAR 255-75-096, which carries the consequence of imprisonment to the statutory good time date, and the record does not show that petitioner had notice of that consequence.” Id. at 116 (emphasis added).

Gonzalez v. Washington, 182 Or.App. 112 , 47 P.3d 537 (2002), does not affect our understanding of Oregon law. That case considered the Board of Parole’s ability to revoke parole and impose a re-incarceration period longer than the “sanctions” contemplated by certain regulations in place in 1986. The court first observed that the pertinent statutes and regulations distinguished between being “continue[d]” on parole with sanctions and having parole “revokefd].” Id. at 539-40. The 1986 regulations relied upon by Gonzalez circumscribed the Board’s authority only with respect to the former situation, so the court concluded in Gonzalez that those particular regulations were inapplicable where parole was revoked. See id.

As to the Board of Parole’s authority to impose intermediate periods of re-incarceration subsequent to revoking parole, the court looked to two statutes. Under Or.Rev.Stat. § 144.390, if the Board elected to revoke parole, the prisoner “shall serve out the sentence.” Section 144.395, however, then as now instructed the Board to “adopt rules ... relating to the re-release of persons whose parole had been revoked.” Gonzalez approved, as consistent with the statute, the petitioner’s subsequent rerelease on parole, but noted that “the board’s discretion in resetting appellant’s release date ... is circumscribed only by [Or.Rev.Stat] § 144.390 and by any rules promulgated under [Or. Rev.Stat.] § 144.395.” Id. at 541. Noting that the defendant in that case pointed to no pertinent rules promulgated under Or. Rev. S. § 144.395, Gonzalez expressly did “not decide if any other rules not relied on by him could apply.” Id. at 541 n. 7.

The warden argues that, under Gonzalez , the board has discretion to impose intermediate incarceration terms subsequent to parole revocation. Gonzalez , however, held that to be true only in the absence of rules promulgated to circumscribe that discretion. In 1994, Or. Admin. R. 255 -075-096 so circumscribed the Board’s discretion with respect to its authority to rerelease prisoners after parole violations, specifically instructing the Board to “deny rerelease ” (emphasis added) after a finding of aggravation. Thus, far from supporting the warden’s reading of the regulations, Gonzalez makes clear that the Board is free to limit its statutory discretion through properly promulgated regulations. Under the regulations in place in 1994, the Board did so.

Finally, the warden points to other portions of the 1994 regulations to support his contention that the 1994 regulations allowed the Board of Parole to impose incremental reincarceration periods. The provisions relied on, however, do not mitigate the impact of Or. Admin. R. 255 -075-0079(10) (1994) and 255-075-0096 (1994), but rather illustrate why our reading of those rules is the correct one.

Rule 255-075-0072(2) (1994) allows the Board of Parole, at the time of the revocation decision, to defer a rerelease decision pending a future disposition hearing. The Board must hold this future disposition hearing, however, within 60 days of revoking parole. Or. Admin. R. 255 -075-0097(2) (1994). This separate future disposition hearing is not only authorized but also necessary when the Board intends to invoke the weighty consequence of denying rerelease. See Or. Admin. R. 255 -075-0096(2) (1994) (“Denial of rerelease on parole requires a future disposition hearing”).

Rule 255-075-0072(2) (1994), then, does not allow the Board of Parole to set an intermediate release date between the 90-day period prescribed in Rule 255-075-0079(1) (1994) and the denial of rerelease. Rather, it merely instructs the Board to either: (1) make a decision to continue parole or set a rerelease date at the initial revocation hearing or (2) decide that a future disposition hearing is necessary, thereby deferring a rerelease decision for up to 60 days.

Rule 255-075-0079(9) (1994) doesn’t help the warden’s argument either. Rule 255-075-0079(9) (1994) allows the Board of Parole to “postpone rerelease on parole pursuant to Divisions 50 and 60 of this chapter.” Division 50 relates to prisoners engaged in serious misconduct while incarcerated. See generally Or. Admin. Rs. 255-50-0005, 255-50-0010 (1994). Division 60 refers to the Board’s authority to defer parole release up to 90 days when the parole plan is deficient. See Or. Admin. R. 255 -60-0008(2) (1994). These provisions are limited to specific situations not applicable to most prisoners, including Himes. Moreover, by referring to circumstances in which rerelease can be set at some period beyond the technical violation period but short of the statutory good time date as “deferring” rerelease, those sections reinforce the conclusion that when Rule 255-075-0096 (1994) refers to “denial” of rerelease, it means exactly that, and not some intermediate period of incarceration such as that permissible under Division 50 and 60.

The warden’s strongest argument in support of the conclusion that the Board of Parole had the authority to set an intermediate date is that the Board scheduled a hearing for May 2024, two years shy of Himes’ projected “good time” date. However, an inmate’s “good time” date — the day on which the inmate’s sentence legally expires — is subject to change. See Erbs, 752 P.2d at 319 (noting that Or.Rev.Stat. § 421.120(h) grants the Board of Parole “a measure of discretion” in allowing good time and “may, pursuant to its rules, allow, retract and restore ‘good time’ ”). The board did not direct that Himes be released prior to his statutory good time date; it merely scheduled a hearing. The obvious explanation is that the May 2024 hearing was necessary to recalculate Himes’ “good time” date based on ensuing events and otherwise to formulate plans for Himes’ release. Cf. Bollinger II, 329 Or. 505 , 992 P.2d 445, 446 (noting that the Board on several occasions adjusted release date to exactly two days prior to good time date).

(b) The Ex Post Facto Violation

The 1994 regulations removed the Board of Parole’s ability to grant a continuum of sanctions for aggravated violation of parole, in favor of an all-or-nothing choice between a 90-day sanction and outright denial of release. This change disadvantaged any inmate whose conduct warranted a finding of intermediate aggravation (e.g. any aggravation that merited some upward departure, yet did not rise to the extreme level required to impose the outermost sanction on the continuum, outright denial of rerelease). Under the 1978 regulations, a finding of intermediate aggravation could result in an intermediate re-incarceration sanction. In 1994, any finding of aggravation finding could result only in the denial of rerelease. This change drastically restricted the opportunity to re-qualify for parole for a substantial class of inmates, thereby increasing the incarceration period attached to the original crime.

It is true that the change in parole regulations may have marginally helped a different class of inmates. However, these ameliorative effects are small in comparison to the possibility — a reality for inmates such as Himes — that postrevocation rerelease eligibility would be severely curtailed. Overall, the change in parole regulations greatly increased the risk that an inmate would serve a longer sentence for his crime. See Weaver, 450 U.S. at 34-36 , 101 S.Ct. 960 (holding that an overall reduction in parole eligibility under new regulations did increase the punishment originally attached to the crime, notwith standing that some prisoners might benefit from the changes).

In summary, the new regulations created a new substantive formula for the calculation of parole rerelease. See Weaver, 450 U.S. at 26-27 , 101 S.Ct. 960 . The “new restrictions on eligibility for release” were disadvantageous. Id. at 34 , 101 S.Ct. 960 . The re-incarceration “presumption,” Miller, 482 U.S. at 434-35 , 107 S.Ct. 2446 , for technical violators with aggravation switched from a flexible continuum to a compelled determination that the inmate be returned for his entire remaining sentence. In other words, this switch increased the “mandatory minimum” punishment for a particular category of inmates, Lindsey, 301 U.S. at 400-01 , 57 S.Ct. 797 , creating a “sufficient risk” of increasing the measure of punishment attached to Himes’ crime. Morales, 514 U.S. at 509 , 115 S.Ct. 1597 . See also Nulph, 27 F.3d at 456 (change in Oregon Parole regulations which increased “benchmark punishment” for a certain class of inmates violated the Ex Post Facto Clause).

Overall, the change in the measure of punishment for parole revocations between the time of Himes’ offense and those in effect when his parole was revoked was extreme. We conclude that it was objectively unreasonable for the Oregon Courts to decide that there was no ex post facto violation. The petition is granted. Himes is entitled to have his parole rerelease considered under the guidelines in place in 1978. See Murtishaw v. Woodford, 255 F.3d 926, 966 (9th Cir.2001).

IV. CONCLUSION

We REVERSE the district court’s order denying Himes’ petition for habeas corpus, and REMAND with instructions to grant the writ unless the Oregon Board of Parole reconsiders Himes’ parole rerelease eligibility within a reasonable time.

. Both parties briefed and argued the issue with respect to regulations in place in 1997, ' regulations similar in most but not all respects to the regulations in place in 1994. For precision’s sake, we reference the exact language of the 1994 regulations. . The “good time” date refers to an inmate’s entitlement to a reduction in prison term if "the inmate faithfully has observed the rules of the institution.” Or.Rev.Stat. § 421.120(1). Once an inmate’s good time date arrives, the inmate is entitled to unconditional release, and the Board of Parole loses jurisdiction over the inmate. Erbs v. Bd. of Parole, 90 Or.App. 253 , 752 P.2d 318, 320 (1988). At the time Himes' parole was revoked, his good time date was projected as May 8, 2026. . The "independent review” of the record required when a state court supplies no ratio decidendi must be carefully distinguished from "independent review” of the constitutional question as that term was employed in Van Tran, 212 F.3d at 1153-55 . Van Tran held that we first apply an "independent” (e.g. "de novo”) review to a reasoned state court decision to ascertain whether the state court committed constitutional error, as a first step to determining whether the error, if any, was objectively unreasonable. That approach has been overruled. Andrade , -U.S. at- -, 123 S.Ct. at 1172-74. Neither Van Tran nor Andrade , however, considered the difficulty we face here: the need to review a state court decision silent as to the federal question. Nor is this a case such as Early v. Packer, 537 U.S. 3 , 123 S.Ct 362 , 154 L.Ed.2d 263 (2002), in which the state court addressed the pertinent federal issues by reference to analogous state law, without citing federal cases. Here, we face a summary adjudication which supplies no indication at all regarding the basis for the decision. A state court’s summary adjudication of federal claims creates unique difficulties. See Ylst v. Nunnemaker, 501 U.S. 797, 803 , 111 S.Ct. 2590 , 115 L.Ed.2d 706 (1991) ("[M]any for-mulary orders are not meant to convey anything as to the reason for decision. Attributing a reason is therefore both difficult and artificial."). As we cannot attribute reason to a silent opinion, we have no source other than the record upon which to base our analysis. . We note that our analysis is also consistent with Garner v. Jones, 529 U.S. 244, 120 S.Ct. 1362 , 146 L.Ed.2d 236 (2000). Because Himes' case was adjudicated prior to the decision in Gamer, the requirements of 28 U.S.C. § 2254 (d)(1) would bar the application of Gamer if Gamer established a "new rule” of constitutional law according to the principles of Teague v. Lane, 489 U.S. 288, 310 , 109 S.Ct. 1060 , 103 L.Ed.2d 334 (1989) and its progeny. A rule qualifies as an "old rule” under Teague jurisprudence if it satisfies the "clearly established” requirement of 28 U.S.C. § 2254 (d)(1). See Williams v. Taylor, 529 U.S. at 412, 120 S.Ct. 1495 . The caveat, of course, is that unlike Teague , § 2254(d)(1) restricts the source of the clearly established law to Supreme Court jurisprudence. Pre-Gamer principles, specifically those announced in Morales, Weaver, Miller v. Florida, 482 U.S. 423 , 107 S.Ct. 2446 , 96 L.Ed.2d 351 (1987), and Lindsey compel the resolution of this case. We therefore need not consider whether, as Himes maintains, the warden waived the Teague issue. See Caspari v. Bohlen, 510 U.S. 383, 390 , 114 S.Ct. 948 , 127 L.Ed.2d 236 (1994) (holding Teague rule inapplicable when result compelled by precedent in place prior to conviction). . Because we hold that the risk of increased punishment is facially apparent, we do not consider whether Himes could alternatively prove an individual ex post facto violation by demonstrating that it can be " 'said with assurance' that he would have received less severe punishment under the prior scheme.” Nulph, 27 F.3d at 456 (quoting Dobbert v. Florida, 432 U.S. 282, 294 , 97 S.Ct. 2290 , 53 L.Ed.2d 344 (1977)). . Contrary to the warden's assertion, Himes did not waive this aspect of the ex post facto argument. Himes' briefing before the district court stated that "[b]y using the 1994 parole regulations, the Oregon Board of Parole denied him re-release following a revocation based on a standard which did not exist in 1978.” In describing the 1994 regulations, Himes specifically noted that "the regulations allowed the Board to deny rerelease entirely.” His briefing before this court is similar: He set out the new regulations in detail and stated that "[t]he 1994 parole regulations applied to Mr. Himes changed the formula by which he was denied re-release and it was thus unconstitutional to apply them to Mr. Himes” and specifically noted "if certain aggravating factors are present, the 1994 regulations allow the Board to deny re-release entirely ” (emphasis added). Sufficiently embedded within these statements was the complaint that the change in regulations as a whole disadvantaged Himes by, in part, "allowfing] the Board to deny re-release entirely.” . Himes asserts, and the warden does not contest, that his parole was revoked for “technical violations.” Although Himes' parole violation report suggested that his behavior violated Oregon laws related to “stalking,” no such finding was made at Himes’ parole hearings. . "Crime severity” is a measure of the seriousness of Himes’ original crime. See Anderson v. Bd. of Parole, 303 Or. 618 , 740 P.2d 760, 763 (1987) (discussing the crime severity metric). Because Himes’ crime severity had been variously calculated as a five or a six, it is unclear which sub-section of former Or. Admin. R. 254 -70-042 would have applied. . The practice observed in Kendrick, Kessler, and Luckey , of setting the parole rerelease date at exactly two days' shy of the good time date suggests that the. "up to” language in the 1997 amendments to Or. Admin. R. 255 -075-0096(1) means that the Board can set a parole date no earlier than two days prior to the good time date. In that sense, the "up to” language in Or. Admin. R. 255 -075-0096(1) (1997) may differ from the "up to” language used earlier in the regulations; the statement in Or. Admin. R. 255 -075-0079(1) allowing imprisonment for "up to 90 days” presumably means any period no longer than 90 days. For present purposes, however, the critical point is that in 1994, the "up to” language simply did not appear in the pertinent regulation. . Although the opinion does not indicate which version of Or. Admin. R. 255 -075-0096 was used, as the parole rerelease decision was made in 1994, we infer that the 1994 version applied to Himes was also applied to Woolstrum. Where Woolstrum relied on non-current regulations it explicitly so stated. See, e.g., 918 P.2d at 114 n. 1. . Section 144.390 has since been repealed. See 1989 Or. Laws ch. 790, § 47(a). Similar language appears, however, in § 144.343(2)(b), which states that upon revocation of parole "[t]he board may ... [rjevoke parole and require that the parole violator serve the remaining balance of the sentence as provided by law.” . Section 255-075-0072 (1994), entitled “Rerelease Order After Revocation” provides: (1) At the time of a revocation decision, the Board shall make an order concerning rere-lease. (2) In the rerelease order, the Board may: (a)Continue parole or post-prison supervision pursuant to 255-075-0075 or 255-075-0080; or (b) Set the rerelease date in accordance with rule 255-075-0079; or (c) defer the rerelease decision pending a future disposition hearing (3)Upon notification that parole or post-prison supervision has terminated by operation of ORS 144.345(2), the Board shall apply subsection (2) of this rule. . For example, an inmate whose parole was revoked, but for whom the Board of Parole did not find aggravation, could have received only a 90-day re-incarceration sanction under the 1994 regulations, as compared to the four to ten month sanction provided for in the 1978 regulations.