At a December 2001 prison disciplinary hearing, Petitioner-Appellant Hubert Earl Teague, an inmate in the Texas Department of Criminal Justice, Correctional Institutions Division ("TDCJ-CID"),1 was found guilty of having violated an anti-trafficking and trading provision of the TDCJ-CID's internal rules of prisoner conduct. As part of his punishment, Teague forfeited thirty days of previously earned good-time credit. Teague subsequently filed a federal habeas corpus petition, alleging that the TDCJ-CID failed to afford him the requisite procedural due process. Citing Malchi v. Thaler2 and Richards v. Dretke,3 the district court denied Teague's petition, concluding that the loss of thirty days of previously earned good-time credit was "de minimis," and thus insufficient to command due process protection. We conclude that the district court erred—not surprisingly, as we shall show—in testing Teague's loss for whether it was de minimis and basing its ruling on a finding that it was. After visiting our dicta in Malchi and Richards, we conclude that no such "de minimis" floor is mandated by our precedent, so that no amount of previously earned good-time credit is so insignificant that it may be taken away by the institution administratively without affording the inmate due process. Stated differently, the TDCJ-CID must accord an inmate due process before depriving him of any previously earned good-time credits, however, slight; there is no exception for an amount that might otherwise be deemed de minimis. We, therefore, vacate the district court's judgment and remand for further proceedings consistent with this opinion.

In 1990, Teague pleaded guilty to a theft offense in violation of Texas state law and was sentenced to a prison term of twenty-seven years. In 1994, Teague was released on parole.

While on parole, Teague committed seven additional state offenses: six burglary offenses (committed on October 4, 1996; December 9, 1996; December 24, 1996; January 20, 1997; January 25, 1997; and February 18, 1997) and one forgery offense (committed on October 31, 1996). He pleaded guilty to five of the burglary offenses and the forgery offense, but he pleaded not guilty to the January 25, 1997 burglary offense. He was subsequently found guilty on that charge and sentenced to a forty-five year term of imprisonment. The incarceration term for each of the six guilty-plea offenses was less than forty-five years, and each was to run concurrently with the forty-five year term. Additionally, his 1994 release on parole was revoked, and his original 1990 sentence was re-imposed.

In December 2001, a prison disciplinary hearing was convened against Teague. He was charged with committing a Code 15.0 violation—Trafficking and Trading.4 It was alleged that another inmate, Melvin Jordan, had caused a $225.00 check to be deposited into Teague's inmate trust account, in violation of the TDCJ-CID's internal rules of conduct.

Prior to the disciplinary hearing, Teague was furnished written notice of the charges against him via a computer generated report. The offense description read, in part: "Teague . . . did make an unauthorized exchange to (offender Jordan . . .) by having offender (Jordan) place a sum of ($225.00) dollars onto [sic] his trust fund account." A second, hand-written notice from the charging officer specified, in its offense description section, that Teague "did make an unauthorized commodity exchange with offender Jordan . . . by having offender Jordan place a sum of $225.00 onto [sic] his trust fund account." The hand-written notice also stated, in the additional information section, that "Offender Teague did have offender Jordan place a sum of $225.00 onto [sic] his trust account...."

During the hearing, the charging officer's report, the transaction records of Teague's inmate account, the $225.00 cashier check, and Jordan's admission that he deposited the money into Teague's account were offered into evidence. Teague's defense, however, was not that the event never occurred, but that he had no knowledge of or participation in Jordan's deposit.5 No evidence was offered as to why Jordan deposited the money into Teague's account or whether Teague had any knowledge of or participation in Jordan's actions.

At the beginning of the hearing, the disciplinary officer made the following statement to ensure that Teague understood the charge: "offender Teague . . . did make an unauthorized market exchange ... by having offender Jordan place a sum of $225.00 onto [sic] his trust fund accounts. Offender Teague, do you understand the charges?" Teague responded, "Yes."

The disciplinary hearing officer found Teague guilty and assessed the following penalties: forfeiture of thirty days of previously earned good-time credit; a reduction in good-time earning rate; fifteen days solitary confinement; forty-five days commissary and recreation restrictions; reduction in custodial classification; and forfeiture of the $225.00.

After the hearing, Teague filed a step-one grievance with the TDCJ-CID, appealing the disciplinary proceeding.6 Next, Teague filed a step-two grievance.7 Finally, after fruitlessly exhausting the internal grievance procedures, Teague filed a 28 U.S.C. § 2254 petition in the district court, seeking a writ of habeas corpus.

In his habeas petition, Teague asserted that: (1) there was insufficient evidence to support the finding of guilt; (2) the prison officials failed to serve him timely with notice of the alleged violation; (3) the disciplinary hearing officer was not impartial; (4) the disciplinary hearing officer improperly denied his request to call witnesses; (5) the disciplinary hearing officer improperly denied his request to have the charging officer present at the hearing; (6) the disciplinary hearing officer stopped recording the hearing during Teague's presentation of the evidence; and (7) the TDCJ-CID improperly removed the $225.00 from his inmate trust account. Teague contended that, by these actions, he had been wrongfully deprived of his right to procedural due process under the Fourteenth Amendment.

After appointing counsel to represent Teague, the district court conducted an evidentiary hearing on Teague's petition. The district court entered a written order in which it concluded that there was no evidence to support the disciplinary hearing officer's guilty finding. The court denied the rest of Teague's claims. The district court's decision on the insufficiency of the evidence was premised on the TDCJ-CID's failure to offer any evidence that Teague "had" Jordan deposit the $225.00 into Teague's trust account. In light of this lack of evidence, the district court concluded that the disciplinary hearing officer's decision was clearly contrary to established federal law—which requires "some evidence to support the findings made in the disciplinary hearing"8—and granted Teague's habeas petition.

In response, the TDCJ-CID filed a Rule 59(e) motion to alter or amend the district court's order. The TDCJ-CID cited our decisions in Malchi9 and Richards10 and asserted that the forfeiture of thirty days of previously earned good-time credit was so "de minimis" that it was insufficient to warrant due process protection.

Understandably following the de minimis primrose path that we had laid down in Malchi's and Richards' dicta, the district court granted the TDCJ-CID's motion to amend, agreeing that the loss of thirty days good-time credit was de minimis and holding it to be insufficient to entitle Teague to due process protection. In reaching its decision, the district court calculated that thirty days represented approximately .18 of one percent of Teague's forty-five year sentence. The court conceded that it could not draw a bright line separating what does and does not constitute a de minimis amount of good-time loss, but was "relatively certain" that less than one-fourth of one percent of Teague's total sentence was de minimis in this case. At least by implication, the district court approached the question of de minimis on a relative or percentage basis, not on the basis of the absolute number of days lost.

Teague timely filed a motion for a certificate of appealability ("COA"). The district court granted the motion and certified the following issue for appeal: Whether, in this case, the forfeiture of 30 days of good-time credits pursuant to a prison disciplinary action was a de minimis loss and, as such, insufficient to trigger due process protections of the Fourteenth Amendment.

In addressing requests for habeas relief, we review a district court's findings of fact for clear error and issues of law de novo.11 Identification of a liberty interest that is protected by the Fourteenth Amendment is a question of federal constitutional law and is reviewed de novo.12 When due process is required, the standard in prison disciplinary hearings requires that there be "some evidence" to support the disciplinary decision.13 Whether there is "some evidence" is an issue of law that we review de novo.14 We may affirm a district court's decision on any basis established by the record.15

Federal habeas relief cannot be granted unless the petitioner alleges that he "`has been deprived of some right secured to him by the United States Constitution or the laws of the United States.'"16 Here, Teague's petition is premised on the claim that the Texas mandatory supervision scheme provides him with a protected liberty interest in his previously earned good-time credits, so that Texas may not deprive him of such credits without first affording him the requisite due process.

Obviously, the Constitution itself does not expressly guarantee due process protection of good-time credits attained by satisfactory behavior while in prison.17 A state statutory scheme may, however, create a right to good-time credits; and when one does so and recognizes that revocation is an authorized sanction, an inmate's interest in good-time credit is accorded due process protection.18 Specifically, an inmate is entitled to the minimum procedures appropriate under the circumstances and required by the Due Process Clause to ensure that the state-created right is not arbitrarily abrogated.19 Thus, when a state inmate enjoys a constitutional expectancy to an early release from prison based on the accumulation of good-time credits, he has a protected liberty interest and is entitled to due process before he may be deprived of such credits.20 Therefore, before we address whether Teague's thirty day loss is subject to an exception from the guarantee of due process because that loss was de minimis, we must first determine whether Texas law affords Teague a protected liberty interest.

In Texas, there are two general ways in which an inmate may become eligible for early release.21 First, an inmate may become eligible for parole; second, he may become eligible for mandatory supervised release.22

Parole, which is not at issue here, is "the discretionary and conditional release of an eligible inmate sentenced to the institutional division so that the inmate may serve the remainder of the inmate's sentence under the supervision of the pardons and paroles division."23 We have held that there is no right or constitutional expectancy of early release on parole in Texas, because parole is within the total and unfettered discretion of the State.24

In contrast, mandatory supervision in Texas is "the release of an eligible inmate sentenced to the institutional division so that the inmate may serve the remainder of the inmate's sentence not on parole but under the supervision of the pardons and paroles division."25 Prior to September 1, 1996, when a Texas prisoner's actual time served plus his accrued good-time credit equaled the term of imprisonment to which he had been sentenced, article 42.18, § 8(c) of the Texas Code of Criminal Procedure specified that "`a prisoner who is not on parole shall be released to mandatory supervision.'"26 Under this former scheme, the TDCJ-CID had no discretion in deciding when or if to release an inmate on mandatory supervision; rather, an inmate had a mandatory right to early release based solely on simple arithmetic.

Effective September 1, 1996, Texas amended its mandatory-supervision scheme to allow a parole panel a modicum of discretion in determining whether an inmate otherwise eligible for release on mandatory supervision should nevertheless remain in custody.27 Under this post-September 1, 1996 scheme, section 508.147(a) of the Texas Government Code, like its predecessor statute, mandates that "a parole panel shall order the release of an inmate who is not on parole to mandatory supervision when the actual calendar time served plus any accrued good conduct time equals the term to which the inmate was sentenced."28 A difference resulted, however, from the addition of section 508.149, which states that "[a]n inmate may not be released to mandatory supervision if a parole panel determines that: (1) the inmate's accrued good conduct time is not an accurate reflection of the inmate's potential for rehabilitation; and (2) the inmate's release would endanger the public."29 Thus, unlike its purely predictive forerunner, Texas's post-September 1, 1996 mandatory supervision scheme is mandatory in large part, but also discretionary in small part.30

In Madison v. Parker, we were called on to decide whether Texas's pre-September 1, 1996 mandatory scheme of supervised release created a constitutional expectancy of early release—an issue of first impression at the time.31 We lacked significant information as to whether the petitioner was eligible for mandatory supervised release, however, and were thus unable to resolve the constitutional question until determinative facts could be established.32 Accordingly, we deferred decision on the issue and remanded the case to the district court for further proceedings.33

Three years later, in Malchi, we again faced the issue whether Texas's pre-September 1, 1996 mandatory supervised release scheme created a constitutional expectancy of early release.34 The petitioner in Malchi had been found guilty at a prison disciplinary hearing of possessing a box of stolen envelopes, as punishment for which his earning status for good-time credit was reduced.35 This reduction was calculated to delay the petitioner's release by more than six months.36

In Malchi, we determined that the mandatory and predictive nature of the pre-September 1, 1996 scheme vested inmates with a constitutional expectancy of early release and a concomitant protected liberty interest.37 Albeit purely in dicta, we went on to temporize somewhat by offering that:

A Texas prisoner does not necessarily have a constitutional expectancy of release on a particular date. For example, it is possible that a de minimis delay of a few days in a prisoner's mandatory supervision release would not give rise to a constitutionally cognizable claim. In the present case, the evidence shows that the prison calculated that the subject disciplinary action delayed [the petitioner's] release for more than six months as a result of the change of status . . . . We hold that such delay is more than de minimis.38

The parties have assumed, without argument, that the pre-September 1, 1996 version applies here and, as such, that Teague has a protected liberty interest under Malchi, leaving only the de minimis question for our review. This assumption, however, is flawed. First, as Teague was convicted of eight offenses—one of which occurred prior to September 1, 1996 and seven of which occurred after September 1, 1996— it is unclear whether and to what extent the former and present schemes apply.

Essentially, then, what we must determine is whether the post-September 1, 1996 addendum to Texas's mandatory supervision scheme, with its narrowly limited modicum of discretion, works to deprive all inmates of their constitutional expectancy of early release. We hold that it does not. Therefore, we need not determine which scheme applies here, as we conclude that Teague has a protected liberty interest under either version.

Texas's current mandatory supervision scheme is virtually identical to those at issue in Greenholtz v. Inmates of the Nebraska Penal & Correctional Complex39 and Board of Pardons v. Allen40 and, as such, warrants the same result. In Greenholtz, the issue before the Supreme Court was whether a Nebraska parole statute created a protected liberty interest.41 Like the present Texas mandatory scheme of supervised release at issue here, the Nebraska statute was discretionary in part and predictive in part, specifying that the state parole board "shall order" the release of an eligible inmate, unless one of four specific designated reasons was found to be present.42 Relying on the structure and language of Nebraska's parole statute, the Court held that it created a presumption that parole release would be granted and thus a constitutional expectancy of early release, entitling inmates to due process protection.43

Likewise, the Court in Allen determined that a Montana parole statute created a constitutional expectancy of early release.44 The Montana scheme provided that the state parole board "shall release" an eligible inmate on parole when it finds that there is a reasonable probability that the inmate may be released without detriment to himself or the community and that he is able and willing to fulfill the obligations of a law-abiding citizen.45 Comparing the Montana statute to the Nebraska statute in Greenholtz, the Court held that the Montana statute created the same presumption of parole release and thus a constitutional expectancy of early release, necessarily entitling inmates to due process protection.46

Juxtaposing the structure and language of Texas' post-September 1, 1996 mandatory supervision scheme with those at issue in Greenholtz and Allen leads us to the same conclusion. Each of these statutes affords inmates the right to early release, then ever so slightly diminishes that right with the potential for the exercise of official discretion in limited instances.47 The Court has made clear that the presence of this type of discretion will not deprive an inmate of an otherwise protected liberty interest. Moreover, unlike the Court in Greenholtz,48 we have the benefit of Texas caselaw on this issue, which uniformly holds that the State's current scheme does create a protected liberty interest.49 All this convinces us that Texas' post-September 1, 1996 mandatory supervision scheme creates a constitutional expectancy of early release and, as such, a protected liberty interest in previously earned good-time credits.

As Teague thus has a protected liberty interest in his previously earned good-time credits under either the pre or post-September 1, 1996 mandatory supervision scheme, we need not determine which scheme applies to him. He is entitled to due process protection under both. Instead, we now address the certified question of whether the state's taking of thirty days of previously earned good-time credit without affording Teague the otherwise requisite due process is not error because such a loss could be deemed de minimis.

Despite acknowledging in Malchi the existence of a protected liberty interest in good-time credits, we there mused in dicta that a delay in an inmate's mandatory supervised release might be so insignificant ("de minimis") that depriving him of his recognized protected liberty interest need not comply with the requirement of due process.50 Four years later, in Richards v. Dretke, we touched on the de minimis concept, again in dicta.51 In footnote 5 of our Richards opinion, we speculated that:

A 30-day delay of a mandatory supervision release might be de minimis and therefore not give rise to a due process claim. The Malchi court held that while a few days might be de minimis, six months was not. That issue, however, is not before us as it has not been raised. . . .52

Since Richards, numerous district courts have assumed that, in the context of good-time deprivation, Malchi and Richards constitute precedent for the de minimis exception to the due process entitlement and have proceeded to address what quantum of good-time credit loss is de minimis.53 Now is the time for us to disabuse the courts of this circuit of the belief that entitlement to due process is subject to a de minimis floor in the context of disciplinary loss of good-time credit. In so doing, we reject out of hand the concept of a de minimis loss for good-time credit (and thus any exception to entitlement to due process protection) that has seeped interstitially into our lexicon, presumably from treating our dicta on the subject as precedent. Acknowledging that Texas inmates have long had protected liberty interests in any amount of previously-earned good-time credit, we hold today that none may be taken away by an administrative tribunal without affording the inmate due process, regardless of the absolute number of days forfeited or the percentage of the sentence (or the remaining balance thereof) represented by the number of days lost.

As we have implied, the de minimis concept is a creature of spurious conception. In Malchi, we neither cited to nor relied on any authority for the existence of the de minimis principle.54 In Richards, we relied solely on Malchi for the existence of the de minimis exception, but merely hypothesized whether thirty days was de minimis.55 We offered no jurisprudential or analytical buttress for our "un-adoption" of the de minimis concept. It appears instead that the concept was simply spawned parthenogenetically and slipped into our storehouse of obiter dicta. It is wholly unclear where the concept originated or why, but it is time to cast this troublemaker from our jurisprudence in the context of the issues before us. Today we exorcize the exception for de minimis deprivation of the liberty interest and hold that, when a state has created such a liberty interest, no amount of good-time credit, however slight, may be stripped from an inmate without affording him the protection of due process.

In Sandin v. Conner, the Supreme Court held that state created liberty interests are generally limited to freedoms from the restraints that impose "atypical and significant hardship on the inmate in relation to the ordinary incidents of prison life."56 In particular, the Court concluded that an inmate is not entitled to due process protection before the imposition of thirty days solitary confinement, because this punishment was not an atypical or significant hardship.57

In justifying its decision, the Court noted, "[n]or does [the petitioner's] situation present a case where the State's action will inevitably affect the duration of his sentence."58 In contrast, Teague's situation, and that of any other Texas inmate who is deprived of previously earned good-time credits, is one in which the State's action inevitably will affect the duration of his sentence.

Under Texas law, once an inmate's good-time credits are forfeited, they can never be restored.59 Thus, once Teague's thirty days of good-time credit are taken away without due process protection, his sentence inevitably will be thirty days longer, as proscribed by the Court in Sandin.

Neither is the possibility that Teague may gain or lose additional good-time credits in the future of any relevance. Whatever additional good-time credit Teague may eventually gain or lose will not affect his previously forfeited time. For instance, if Teague should subsequently forfeit an additional thirty days of previously earned good-time credit, these additional thirty days would be cumulated with the thirty days that he previously lost, making Teague's total loss sixty days. It would be a consecutive loss, not a concurrent one. Conversely, awards of new good-time credit would be subtracted from the previously lost thirty days. Neither future awards nor future forfeitures have relevance to a present forfeiture.

In addition to conflicting with Supreme Court precedent, legitimizing the de minimis exception would work serious practical and equitable problems. As indicated earlier, two methods could be used in determining whether a loss of good-time credit is de minimis. The first is a comparative or percentile approach, similar to that employed by the district court here. Under this method, the number of forfeited days is divided by the total number of days to which the inmate was actually sentenced to produce a fraction or percentage of loss. Depending on the court's subjective view of whether the resulting quotient is de minimis, the inmate might or might not be entitled to due process.

The other potential method is the absolute approach, which was employed in Malchi. Under this approach, some number of days is arbitrarily selected by the court, the deprivation of less than which is deemed de minimis and thus not subject to due process protection; deprivation of that or any greater number of days would be sufficient to warrant due process protection.

Problems inhere in the application of both methods. For instance, how and why would we draw the line at a particular fraction or any particular number of days? What would be the justification and reasoning to label, say, .18% de minimis, but not .25%; or to label thirty days or less de minimis, but thirty-one days or more not de minimis?

Difficulties with uniformity and equity in both methods exist as well. An inmate sentenced to five years who commits a trafficking offense could receive due process protection if deprived of thirty days (thirty days is approximately 1.7% of five years), but an inmate sentenced to forty-five years who commits the same offense and loses the same thirty days would be entitled to no due process protection, as thirty days is but .18% of forty-five years.60 It impresses us as patently unfair to afford one inmate due process, but not another, when all factors are the same except the length of their initial sentence or the balance remaining to be served.

Interestingly, application of the relative approach could implicitly overturn Malchi. There, we concluded that six months, in and of itself, is "more than de minimis."61 Under a percentage or relative approach, however, even six months could be considered de minimis. For example, if an inmate were sentenced to sixty years incarceration, the forfeiture of six months good-time credit would be less than one percent of his sentence. We would have to ask whether six months, equaling less than one percent of the actual sentence, should nevertheless be considered de minimis. If so, how could this example comport with our holding under Malchi's absolute approach that six months "is more than de minimis?" For Teague, the question would be whether the deprivation of thirty days—regardless of the length of his actual sentence—is always going to be de minimis.

As the de minimis concept is contrary to Supreme Court precedent and is a Pandora's Box, the opening of which would let loose myriad difficulties and inequities, we today hold that no amount of previously earned good-time credit, however slight, can ever be deemed de minimis and, more importantly, that any loss of such credits that extends the inmate's expectation of release is never subject to a test for de minimis in the context of procedural due process. The TDCJ-CID must afford its inmates procedural due process before depriving them of any good-time credit. If, as argued by counsel for the TDCJ-CID, this would open the floodgates of disciplinary appeals, the prisons of Texas can either use any of the innumerable alternative punishments that do not offend due process or see to it that their inmates receive the process that is due before taking away any good-time credits.

As an alternative basis for affirming the district court's judgment, the TDCJ-CID advances that the Code 15.0 trafficking and trading offense of which Teague was found guilty is a strict liability offense and that, as such, there was no need to introduce any evidence of Teague's knowledge of or participation in Jordan's deposit. We disagree.

At the time of Teague's offense, Code 15.0 punished "[t]he unauthorized buying, selling, exchange or transfer of any commodity from any individual, other than making authorized purchases from the commissary (evidence may include an excessive inventory of marketable items)." Nothing in the language of this provision justifies a conclusion that it is a strict liability offense, obviating the need for the TDCJ-CID to present evidence of Teague's knowledge or participation. Purchases, sales, exchanges, or transfers are bilateral transactions that necessarily require at least the knowledge of both parties, if not the active participation of each. This provision simply cannot be read to permit the TDCJ-CID to punish an inmate who has no knowledge of or participation in an unauthorized deposit into his trust account.

We perceive no error in the district court's original judgment based on the TDCJ-CID's failure to offer any evidence of Teague's knowledge of or participation in Jordan's deposit. Our review of the record from the disciplinary hearing confirms that the TDCJ-CID did establish that Jordan made an unauthorized deposit; that record is devoid, however, of any evidence that Teague had Jordan deposit the check or even knew about it.

Teague had a constitutional expectancy of early release to mandatory supervision and thus a protected liberty interest in his previously earned good-time credits. This entitled him to due process protection before being deprived of any of his credits through administrative or disciplinary proceedings. No exception exists for a loss that is judicially perceived to be de minimis, so none could even be considered. Accordingly, we vacate the district court's amended judgment holding otherwise, affirm the district court in all other respects, and remand with instructions that the district court re-enter its original judgment to the extent that it granted habeas relief.

VACATED IN PART, AFFIRMED IN PART, AND REMANDED WITH INSTRUCTIONS.