Dooley, J.

Plaintiffs, an inmate of the Vermont prison system who was sanctioned for disciplinary violations and the class he represents, prevailed in their claims to the Washington Superior Court that the Vermont Department of Corrections (DOC) (1) failed to adopt regulations on prison discipline in accord with the requirements of the Vermont Administrative Procedure Act; (2) denied meritorious good time credits for minor rules infractions in violation of statute; and (3) applied too low an evidentiary standard in prison disciplinary hearings. Defendant, Commissioner of Corrections, appeals from the latter holding. Plaintiffs appeal the court's refusal to award retroactive class relief and to place a higher evidentiary burden on the Commissioner in disciplinary hearings. We affirm.

The lead plaintiff in this case is Matthew LaFaso, who was disciplined in February for a minor rules infraction. His suit was subsequently certified a class action on behalf of all past, present and future inmates subject to DOC Policies 1021 and 973, and the "some evidence" rule at prison disciplinary hearings. Plaintiffs claimed that DOC is required to comply with the rulemaking procedures of the Vermont Administrative Procedure Act (APA), 3 V.S.A. §§ 801-849, in promulgating its rules and regulations. They alleged that DOC failed to comply with the APA when it established Policies 1021 and 973, the rules that cover inmate discipline, and that those rules are therefore ineffective. Plaintiffs also claimed that DOC's denial of meritorious "good time" to inmates found to have committed minor disciplinary infractions contravenes 28 V.S.A. § 853(a), which limits this sanction to cases of "serious breach of the rules." Finally, plaintiffs challenged the validity of the "some evidence" standard under which DOC determined whether inmates committed disciplinary infractions, arguing that it violates due process.

In a series of three orders, the superior court ruled in favor of plaintiffs on each of the issues. On the last issue, the court held that the "some evidence" standard violates due process and that a "preponderance of the evidence" standard is required at prison disciplinary hearings. Defendant has appealed only this latter decision. Plaintiffs agree with the decision as a matter of federal law but argue that the Vermont Constitution requires that the infractions be proved by "clear and convincing" evidence.

Following these rulings, plaintiffs sought either full or partial retroactive relief, including expungement of disciplinary convictions against class members based on Policy 1021 or the "some evidence" standard, and the award of meritorious "good time" denied under Policy 973. The court found that retroactivity would produce substantial inequity by placing "severe and destructive" burdens on DOC, and consequently denied all retroactive relief. Plaintiffs appeal this conclusion and a number of the findings supporting it.

I.

We address first the evidentiary standard necessary to satisfy due process in prison disciplinary proceedings. Policy 1021 sets forth a detailed procedure for disciplining inmates for rules infractions, derived in part from statute. DOC Policy 1021; see also 28 V.S.A. § 852 (authorizing disciplinary committee and hearing procedures). The process starts with a disciplinary report and charge brought by the reporting officer. If a major rules violation is charged, the matter goes to a hearing officer for hearing. The hearing officer is an employee of DOC appointed for that purpose by the superintendent of the institution or the DOC district manager. Also appointed is a hearing assistant, to aid the inmate to present his or her case, and a presenting officer to present the case supporting the violation. The issue here involves the burden of proof placed on DOC to prove the charge. The rule specifies that "[i]n order to determine if the inmate is guilty, the Hearing Officer need only find some evidence in the record that supports that finding." DOC Policy 1021, § IV(E)(8)(i)(5).

The trial court determined that the rule denied plaintiffs due process because it did not require that the hearing officer find guilt based on a preponderance of the evidence. It is clear that the rule allows the hearing officer to impose discipline even if the officer concludes that it is more probable than not that the inmate did not violate the rules, as long as there is some evidence of a violation. Defendant claims that the trial court erred because the United States Supreme Court has already held, in Superintendent, Mass. Correctional Inst. v. Hill, 472 U.S. 445 (1985), that prison authorities need only have "some evidence" that an inmate committed a disciplinary infraction in order to satisfy due process.

It is peculiar that there would be a difference of opinion on the nature of the constitutional mandate when there is a United States Supreme Court decision on point. In understanding how the difference arises, it is helpful to itemize what is not in dispute, as well as what is disputed. Both sides agree that an inmate can be disciplined when an independent observer, for example, this Court, finds "some evidence" supporting guilt. Plaintiffs argue, however, that in addition the hearing officer must find from that evidence that guilt is more probable than not. Defendant objects to this second requirement.

We conclude that Hill described the appropriate standard for judicial review of the actions of prison authorities, not the proof necessary for a fact-finder to find that an inmate violated a disciplinary rule. Thus, the decision resolved only the point on which the parties agree in this case — there must be some evidence of guilt as found by an independent observer. Hill did not resolve the point of disagreement.

Although the Hill opinion is not a model of clarity, several factors support this conclusion. First, the Court stated that "[t]he fundamental fairness guaranteed by the Due Process Clause does not require courts to set aside decisions of prison administrators that have some basis in fact." Id. at 456 (emphasis added). The Court's statement focuses on the level of due process required of a reviewing court, not that required of prison disciplinary officers. Allowing the reviewing court to employ such a highly deferential standard of review as "some evidence" may be justified by the lower levels of due process required in "the distinctive setting of a prison," see id. at 454-55, and the need for conservation of judicial resources. A deferential standard of review is necessary, "[o]therwise [courts] would assume the task of retrying all prison disciplinary disputes." Willis v. Ciccone, 506 F.2d 1011, 1018 (8th Cir. 1974); see also Smith v. Rabalais, 659 F.2d 539, 545 (5th Cir. 1981). We may fairly infer from the Supreme Court's analysis that disciplinary proceedings require more extensive evidentiary consideration than the "some evidence" rule for judicial review or at least that the Court left this question open.

Further, the Court stated that its "some evidence" standard "does not require examination of the entire record, independent assessment of the credibility of witnesses, or weighing of the evidence." Hill, 472 U.S. at 455. This statement shows that the Court was contemplating a standard of review. We find incredible the suggestion that a de novo proceeding intended to determine the guilt or innocence of any individual could dispense with these procedures and retain a semblance of "fundamental fairness."

Finally, the questions for which the Court granted certiorari in Hill concerned judicial review of prison disciplinary board findings, including "whether the standard of review applied by the [Massachusetts] court was more stringent than is required by the Due Process Clause." Id. at 449 (emphasis added). Despite some confusing language in the opinion, it is clear that the Court was not asked to determine the standard of proof the prison disciplinary board was required to meet when imposing discipline.

Consistent with our conclusion, a number of federal and state courts have explicitly recognized that Hill addressed only a standard of review question. See, e.g., Brown v. Fauver, 819 F.2d 395, 399 n.4 (3d Cir. 1987); Kodoma v. Johnson, 786 P.2d 417, 420 (Colo. 1990); Harper v. State, 397 N.W.2d 740, 743 (Iowa 1987). Similarly, our own case law has cited Hill in a review context. See In re Nash, 151 Vt. 1, 2, 556 A.2d 88, 89 (1988).

Defendant is correct in stating that the "some evidence" rule has widespread support in the case law, but he fails to recognize that virtually all of these cases address only the standard applied by the courts in reviewing facts found in a prison disciplinary hearing. We expressly decline to follow the small minority of decisions that appear to endorse the "some evidence" rule as an evidentiary standard of proof applicable to the fact-finder.

We disagree with defendant that Hill resolved the standard-of-proof issue in his favor. The safest reading of the Supreme Court's ambiguous analysis is that Hill does not purport to resolve the question one way or the other. Thus, we must ourselves perform the balancing that is required to determine the mandate of due process. In Santosky v. Kramer, 455 U.S. 745, 754 (1982), the United States Supreme Court stated that a determination as to "whether a particular standard of proof in a particular proceeding satisfies due process" should be governed by the factors set forth in Mathews v. Eldridge, 424 U.S. 319, 335 (1976). Based on an application of the Mathews criteria to the present facts, but not limiting our holding to the federal constitution, we conclude that due process requires prison authorities to prove inmate disciplinary infractions by a preponderance of the evidence.

The Mathews test specifies several factors that a court should balance in determining whether a given procedure satisfies due process. Id. The first factor is the private interest that will be affected by the state action; here, it is the prisoner's interest in not being unjustly disciplined. This is a highly important interest, with both direct and indirect consequences. The possible direct consequences are itemized in the regulation as "placement in disciplinary segregation, loss of good time, restitution, loss of privileges, reprimand, extra work, forfeiture of funds, apology, written essay or any combination of sanctions." DOC Policy 1021, § IV(B). The indirect consequences relate to the inmate's classification and programming, as well as the length of the sentence, as a discipline record may serve to deprive an inmate of any or all opportunity for early release or parole.

The second factor is the risk of erroneous deprivation of the affected private interest through the procedures used. See Mathews, 424 U.S. at 335. Defendant's only relevant argument on this point is that DOC provides significant procedural protections for inmates accused of violating prison rules. Although these procedures are indeed important, they are targeted largely at informing an inmate of the charge and subsequent basis for the decision, and assisting the inmate in presenting evidence at the hearing. See DOC Policy 1021, § V (disciplinary process). The procedures are of little help when prison authorities may impose discipline despite evidence whose cumulative force indicates the inmate's innocence. We also note that the relaxed admissibility rules for prison disciplinary hearings mean that the "some evidence" may be nothing more than the reporting officer's report. See DOC Policy 1021, § V(E) (hearing officer to receive investigating officer's report, "as well as any other evidence in support of the violation").

It is difficult to conceive of an aspect of disciplinary procedure with a greater impact on the accuracy of fact-finding than the evidentiary standard on which the ultimate conclusion must be based. The standard adopted by defendant allows erroneous determinations as long as any evidence supports a violation. We conclude there is a very significant risk of erroneous discipline of an innocent inmate under a "some evidence" standard of proof.

The final Mathews factor is the government's interest in the process at issue. See Mathews, 424 U.S. at 335. The state has an undeniably strong interest in the orderly administration of its prisons, including inmate discipline and rehabilitation. Especially in the prison setting, the state has an interest in swift and certain punishment for violations of disciplinary rules. See Wolff v. McDonnell, 418 U.S. 539, 563 (1974). For these reasons, prisoners are not entitled to the full range of procedural protections afforded other citizens. See id. at 569-70.

We find it highly unlikely, however, that adherence to a preponderance-of-the-evidence standard will interfere with the enforcement of disciplinary regulations in Vermont prisons. This standard will not impose an undue hardship on the state, and any burden created by the need to produce additional evidence any burden created by the need to produce additional evidence in close cases will be offset by the benefit to the government, and to our society, of the greater assurance that disciplinary cases will be correctly decided. See United States ex rel. Miller v. Twomey, 479 F.2d 701, 718 (7th Cir. 1973) ("[N]either the state nor the inmate has any valid interest in treating the innocent as though he were guilty."), cert. denied, 414 U.S. 1146 (1974). Further, it is unlikely that a preponderance standard will impose any significant fiscal burden. We find greatly overstated defendant's claim that a preponderance-of-the-evidence standard "would require increased staff resources in the investigation and prosecution of prison DR's" and the "discipline process would slow down measurably" given that the presenting officer is currently required to compile and present "all relevant, reasonably available" documentary and witness evidence. DOC Policy 1021, § V(E)(7).

We recognize fully the difficulties inherent in maintaining prison discipline, but these problems cannot justify our approval of a standard of proof — already scaled far back from the "reasonable doubt" standard necessary for the original criminal conviction — that allows imposition of discipline in the face of probable innocence. The one case that has given the state's interest sufficiently heavy weight to justify the lower standard, Goff v. Dailey, 991 F.2d 1437, 1442 (8th Cir. 1993), is unpersuasive. The court's opinion focused almost exclusively on the state's interest and failed to consider all the Mathews factors; thus, its analysis is, at best, incomplete, and as the accompanying dissent points out, "has neither precedential nor academic support." Id. at 1443 (Heaney, J., dissenting in part). Moreover, the state's approach may have a negative impact on society's legitimate interest in prisoner rehabilitation. Respect for our system of justice is not enhanced by punishing innocent individuals, especially prisoners who have a heightened need to develop such respect.

As noted above, inmates accused of violating prison rules may face serious consequences if the disciplinary proceedings result in an adjudication of guilt. Nonetheless, we decline plaintiffs' invitation to require that such an adjudication be based on "clear and convincing" evidence. As a general rule, the preponderance standard applies in administrative adjudications in this state. See Harrington v. Department of Employment Training, 152 Vt. 446, 448-49, 566 A.2d 988, 990 (1989). In view of this general rule, we conclude that the Legislature adopted this standard in establishing the procedure for prison discipline adjudication. See 28 V.S.A. § 852(c) (discipline is imposed if charge "is sustained").

The clear-and-convincing-evidence standard imposes a higher burden on the administrative agency than normal, and it shifts the balance of interests assessed under the Mathews test. It might be justified by looking at the prisoner's interest alone because liberty is involved and serious consequences may result from the adjudication. However, it intentionally errs on the side of avoiding discipline in some cases where the fact-finder determines that the prisoner violated the disciplinary rule.

Although we have not always agreed with the United States Supreme Court in weighing the competing interests in a prison setting, see State v. Berard, 154 Vt. 306, 310-12, 576 A.2d 118, 120-21 (1990), we have recognized the great interest of the state in the "safe and orderly operation of Vermont's prisons." Id. at 311, 576 A.2d at 121. The swiftness and certainty of punishment has a greater weight in this setting. The disciplinary process involves "confrontations between inmates and authority and between inmates who are being disciplined and those who would charge or furnish evidence against them" such that "reasonable personal safety for guards and inmates may be at stake" and the fact-finding process becomes difficult. Wolff, 418 U.S. at 562. On balance, we conclude that proof of a disciplinary violation by a preponderance of the evidence is appropriate.

Prisoners accused of disciplinary infractions may not be punished for such actions unless their guilt can be shown by a preponderance of the evidence. Anything less does not comport with the due process protections of the Fourteenth Amendment. We also hold that, in this context, the concept of due process as embodied independently in Chapter I, Article 10 of the Vermont Constitution requires the preponderance rule.

II.

We next address plaintiffs' claim that the trial court improperly denied the request of the plaintiff class for retroactive relief. As set forth above, this action involves three separate claims: an improper standard of proof in disciplinary hearings, an illegal policy which provided that meritorious good time could be denied for minor disciplinary infractions, and failure to go through the Administrative Procedure Act procedures in issuing disciplinary regulations. For each of these claims plaintiffs sought retroactive class relief in their complaint. For example, plaintiffs sought "that all disciplinary convictions obtained under written policies . . . which were promulgated without compliance with the Administrative Procedures Act, be vacated and expunged from the files of those prison inmates, and that said inmates be relieved of all direct and collateral consequences of said disciplinary convictions."

Following the court's decision that plaintiffs prevailed on the merits, and the issuance of an injunction reinstating the named plaintiff's meritorious good-time credits and prospectively invalidating the involved policies, plaintiffs proposed a limited retroactive relief order that would cover the 1024 currently incarcerated prisoners, approximately 200 prisoners in community settings, and 246 persons on parole. The order would require the immediate restoration of statutory good time to each prisoner but would not affect classification until the next reclassification and would affect only future parole decisions. Because of the lack of records on awarding of meritorious good time, the proposal made certain assumptions about how DOC may have acted in determining whether to increase the amount of meritorious good time in any month in which the inmate was disciplined.

DOC calculated the cost of full retroactive relief for all past prisoners would be $758,000. It opposed full or limited retroactive relief to the class.

The trial court found the propriety of retroactive class relief to be governed by the test developed by the United States Supreme Court in Chevron Oil Co. v. Huson, 404 U.S. 97 (1971), for determining whether new law announced in civil decisions is to be applied retroactively. See American Trucking Ass'ns v. Conway, 152 Vt. 383, 391-92, 566 A.2d 1335, 1339-40 (1989); Solomon v. Atlantis Dev., Inc., 145 Vt. 70, 74-75, 483 A.2d 253, 256-57 (1984). That decision authorized the denial of retroactive relief if three considerations are present: (1) the decision to be applied only prospectively establishes a new principle of law, either by overruling clear past precedent or by deciding an issue of first impression not clearly foreshadowed by precedent; (2) the purpose and effect of the rule of law will be furthered by nonretroactive application; and (3) retroactive application would cause substantial inequitable results. Chevron, 404 U.S. at 106-07. The trial court held that the ruling on each of plaintiffs' claims decided an issue of first impression, the resolution of which was not clearly foreshadowed by prior decisions, and that retroactive relief would be inequitable.

On appeal, plaintiffs argue that the court had no discretion to deny retroactive class relief, the Chevron test requires retroactive class relief, the trial court abused its discretion in failing to analyze plaintiffs' limited retroactive remedy proposal, and certain of the facts found by the court are clearly erroneous. We take these claims in order.

Plaintiffs' first claim is that retroactive relief is required because all of defendant's actions are based on regulations that were never promulgated in accordance with the Administrative Procedure Act. Such a failure "shall prevent a rule from taking effect." 3 V.S.A. § 846(a). Since no rules took effect, plaintiffs argue that the consequences of application of the rules must be ineffective.

Plaintiffs confuse the rights of class members with the remedies the court will provide for violation of those rights. As set forth below, the remedy of retroactive class relief lies in the discretion of the trial court, subject to review for abuse of that discretion.

Second, plaintiffs argue that the Chevron test requires retroactive relief in this case. This issue has become more complicated because the United States Supreme Court has abandoned Chevron, in part, and required that federal constitutional decisions be given "full retroactive effect in all cases still open on direct review." Harper v. Virginia Dep't of Taxation, ___ U.S. ___, ___, 113 S. Ct. 2510, 2517 (1993). We need not decide in this case the extent to which Harper has changed the applicable law on retroactivity for constitutional and nonconstitutional questions. We can assume that the ruling in this case for plaintiff LaFaso is fully retroactive to any other case which is properly before us. The real question we must address is whether the claims of others are properly before us through the procedural device of a class action.

Restated in terms by which it has been addressed by other courts, the question is whether the court should extend the equitable remedy of a mandatory injunction to a class of plaintiffs, and, if so, how that relief should be structured. See, e.g., Shannon v. United States Civil Serv. Comm'n, 444 F. Supp. 354, 368-69 (N.D. Cal. 1977), modified on other grounds, 621 F.2d 1030 (9th Cir. 1980). "[I]n wielding equity power, [the court] must weigh competing claims and determine where a preponderance of the equities lies." Rothstein v. Wyman, 467 F.2d 226, 234 (2d Cir. 1972); see also Archer v. District of Columbia Dep't of Human Resources, 375 A.2d 523, 528 (D.C. 1977) (consideration of retroactive class relief results in "a delicate balancing of the equities"). The applicable principles were recently recognized by this Court in American Trucking Associations:

Whether to fashion a remedy in a particular case so that it applies prospectively only is a decision within the discretion of the trial court. The United States Supreme Court emphasized this in its decision in Lemon v. Kurtzman, 411 U.S. 192 (1973). . . . The Court . . . noted that "[i]n shaping equity decrees, the trial court is vested with broad discretionary power; appellate review is correspondingly narrow." . . . Moreover, "equitable remedies are a special blend of what is necessary, what is fair, and what is workable."

152 Vt. at 378, 566 A.2d at 1332 (citations omitted).

We look at the equities on each of the substantive issues involved with the understanding that the standard of review is abuse of discretion. See id. at 379, 566 A.2d at 1333. The trial court ruled that the governing statute did not permit meritorious good time to be denied as punishment for a minor disciplinary rule infraction. The Department's Policy 973, however, provided that an inmate could not earn meritorious good time in any month that the inmate commits an act subject to disciplinary sanctions. DOC Policy 973, V(E). Thus, over the years DOC has improperly denied meritorious good time as a punishment for minor disciplinary infractions in months where it might otherwise be awarded.

Meritorious good time is awarded only when the inmate has rendered special or unusual service. 28 V.S.A. § 811(b). Any order of retroactive class relief for this violation must surmount DOC's lack of record keeping and show what meritorious good time the inmate would otherwise have received. Plaintiffs suggested resolving the deficiency by awarding five days of meritorious good time, the maximum allowable, in any month in which the inmate earned the full amount of statutory good time, but earned no meritorious good time, if the inmate earned some meritorious good time in any of the six months preceding or following the month in question.

Based on the evidence received, the court determined that a fair consideration of how much meritorious good time to award to each inmate would require four hours of work on each inmate file, and that the cost of this work for the entire caseload would be $116,000. This expense would "have negative effects on the Department's budget, manpower and ability to achieve its statutory mandate." Thus, the court ruled that retroactive relief would produce substantial inequitable results and work a hardship on DOC.

We conclude that the ruling was within the court's discretion. In balancing the equities, the court could consider the administrative burden of granting retroactive class relief. See Wolff, 418 U.S. at 574 (new procedural rules for prison disciplinary hearings not retroactive in part because of "significant impact . . . on the administration of all prisons in the country" and "burden [on] federal and state officials"); Cornell v. Cupp, 550 P.2d 1386, 1388 (Or. Ct. App. 1976) (declining to apply procedural due process requirements retroactively regarding segregation of prisoner because of burden on correctional system). Also to be considered is the fairness of such relief in light of the inadequate records to show who should receive meritorious good time and in what amounts. Plaintiffs' proposal would reduce somewhat the cost of compliance and respond to the record deficiencies by awarding the maximum amount of good time to some inmates who never would have received any meritorious good time or would not have received that amount. The trial court certainly had the discretion to deny such a broad and overinclusive remedy. See Klaips v. Bergland, 715 F.2d 477, 485 (10th Cir. 1983) ("administrative burdens and the uncertain efficacy of relief" are valid reasons for limiting retroactive class relief).

The second substantive violation was the use of the "some evidence" standard for administrative fact-finding, as set forth in Part I of this opinion. The court found that the policy was first introduced in 1989, but "[t]here is insufficient evidence that plaintiff was convicted on less than preponderance of the evidence at a DR hearing." Plaintiffs seek relief that would expunge the results of any disciplinary hearing for current inmates or parolees since the standard-of-proof policy took effect.

The extent of the administrative costs in correcting the records under plaintiffs' modified retroactive relief proposal is disputed. Two additional considerations, however, support the denial of retroactive class relief in the manner proposed by plaintiffs. Initially, the trial court concluded that Hill authorized the lower evidentiary standard. In its August 3, 1990 decision, the court changed its mind and found the regulation improper but would not award plaintiffs relief because DOC asserted that the hearing officers actually used a preponderance-of-the-evidence standard. The court issued a prospective injunction in February 1991 but never was able to resolve the extent to which the lower evidentiary standard was actually used.

As set forth above, the appropriate sanction for a major infraction is always determined after a hearing. The extent to which there may be factual disputes in such a hearing will necessarily vary. If any other inmate had brought suit seeking expungement of a disciplinary sanction issued as a result of a hearing, the court would have determined the extent to which the improper standard affected the outcome. Plaintiffs' proposal avoided this step, assuming that the evidentiary standard determined the outcome in every case. Thus, the relief sought was overbroad, and the court had the discretion to deny this relief.

As more fully discussed infra, we are also concerned with the timeliness of the attack of the class members on the disciplinary sanction, as well as the finality of administrative adjudication. Class members whose disciplinary hearings occurred before this case was filed could have appealed the result of the hearing to superior court pursuant to Rule 75 of the Vermont Rules of Civil Procedure. They failed to do so. Except for the possibility of a habeas corpus action, see Shuttle v. Patrissi, 158 Vt. 127, 605 A.2d 845 (1992), the failure to act may foreclose them from relief. See LD MD, Inc. v. State, 154 Vt. 384, 387-88, 576 A.2d 1244, 1245 (1990) (plaintiffs may not use nullity of agency's substantive rule to avoid compliance with procedural rules governing appeals); see also Haynes v. State Commercial Fisheries Entry Comm'n, 746 P.2d 892, 894 (Alaska 1987) (administrative appeals not filed within thirty days will be dismissed). Although considerations of administrative finality may not control in every individual case, they do make inappropriate blanket relief for all class members in which finality is ignored. See Merrilees v. Treasurer, 159 Vt. 623, 624-25, 618 A.2d 1314, 1316 (1992); Chittenden Trust Co. v. MacPherson, 139 Vt. 281, 283-84, 427 A.2d 356, 358 (1981).

The third violation was the failure to follow the procedures of the Administrative Procedure Act (APA) in promulgating the discipline policy. For this violation, plaintiffs seek the retroactive cancellation of all disciplinary action with respect to current prisoners or parolees if such action occurred after December 31, 1981, the date the policy was adopted. The court refused the relief for cost reasons and because it would make the classification system "chaotic," and necessitate changes in "custody levels, release dates, parole eligibility dates as well as involve transfer between the Department's six facilities" and the use of a subjective and inappropriate classification system.

A number of considerations support the trial court's decision. Putting aside the financial costs, about which there is dispute, the effect of the requested relief is to eliminate any consideration of acts of misconduct, whether serious or not serious, in deciding placement, programming, parole eligibility, and release date. As the court found, DOC relies heavily on the inmates' track record within the institution in making these decisions. Plaintiffs' relief would destroy that classification system, leading to a wholly subjective replacement or to inappropriate decisions. These negative consequences support the denial of relief.

We must look closely at the rule of law involved in determining whether to give class retroactive relief. The legal violation was the failure of DOC to go through a notice-and-comment process in proposing its regulations and to allow review by a legislative committee. On the other hand, the Legislature has commanded that DOC adopt a code of disciplinary conduct and implement sanctions for its violation. See 28 V.S.A. § 102(c)(5) (commissioner is charged with responsibility "[t]o prescribe rules and regulations for the maintenance of discipline and control at each correctional facility"). As a result, plaintiffs' relief would trade the violation of one legislative directive for the violation of another.

We do not believe this is the kind of violation that warrants retroactive class relief. It is entirely speculative whether any inmate would have avoided disciplinary sanctions if DOC had followed the APA procedures in adopting regulations. Many of the decisions plaintiffs would reverse could have been taken in the absence of regulations either because they were directly authorized by statute, see 28 V.S.A. § 812 (good time may be denied for "any offense"), or because DOC or the Parole Board, see 28 V.S.A. § 501(c) (in determining parole, board may consider "inmate's conduct and employment at the correctional facility"), had the broad authority to consider an inmate's actions, even in the absence of a specific conduct code.

Retroactive class relief also offends principles of finality. Many of the class members went through formal disciplinary proceedings, failed to raise the invalidity of the regulations and did not appeal. Many of the class claims are stale. This is an equity proceeding in which the doctrine of laches applies. That doctrine bars relief where a plaintiff "has failed to assert his right for an unreasonable and unexplained period of time and where the delay has been prejudicial to the defending party." American Trucking Ass'ns, 152 Vt. at 382, 566 A.2d at 1334. Claims that would be affected by plaintiffs' requested relief go back as much as eight years before the filing of this complaint. If the matter had been litigated earlier, or any inmate had raised it on appeal of a disciplinary decision, many of the negative consequences of class relief would have been avoided.

Other courts have refused to give retroactive class relief for defects in the adoption of disciplinary rules, stressing considerations like those discussed above. See Guerrero v. Department of Corrections, 418 N.W.2d 685, 687 (Mich. Ct. App. 1987) (retroactive relief would severely disrupt correctional classification process, creating administrative chaos for prison officials); Department of Corrections v. McNeil, 506 A.2d 1291, 1294 (N.J. Super. Ct. App. Div. 1986) (retroactive invalidation of disciplinary standards inappropriate because "many of these proceedings resulted in the loss of commutation time or otherwise reflected on the suitability of numerous inmates for parole" so that retroactive relief would "trigger the early inappropriate release of large numbers of inmates"). We find these decisions to be well reasoned, and they support our decision to affirm the trial court in this case.

In each of the above instances, we have stressed the considerations supporting the trial court's decision without discussing in detail those that point in the opposite direction. We have done so because we are reviewing for abuse of discretion, and the trial court could find these opposing considerations inadequate to carry the day.

Finally, plaintiffs argue that the trial court gave insufficient consideration to their proposal to limit class relief and thereby reduce the administrative burden on DOC, and that many of the court's findings, particularly those related to cost, are clearly erroneous. Much of plaintiffs' argument goes to the issue of the cost of compliance. While we agree that cost is relevant to the availability of class relief, particularly in these days when it is well known that DOC has inadequate resources to fulfill its mandate, we have not stressed that consideration in affirming the trial court decision. We believe that denial of retroactive class relief is supported even if such relief bore no direct financial cost.

Similarly, a substantial part of the argument relates to the evaluation of evidence about legal advice defendant received on whether the Administrative Procedure Act applied to the disciplinary regulations. We do not view the nature of the legal advice defendant received as a significant factor.

The remaining findings plaintiffs dispute are not clearly erroneous and will stand. See Klein v. Klein, 150 Vt. 466, 469, 555 A.2d 382, 384 (1988); V.R.C.P. 52(a)(2). We find no error in the refusal of the court to accept plaintiffs' alternative proposal for relief.

Affirmed.