Gerald McCollum Clifton C. Piggie-Bey, Harry L. Greene-El, and Reynaldo Ramirez-Rodriguez v. Harold Miller, Warden

POSNER, Circuit Judge. These consolidated appeals from the denial of petitions for habeas corpus raise the question whether four inmates at the federal penitentiary at Marion, Illinois were denied due process of law in prison disciplinary proceedings, and thereby deprived of their liberty in violation of the Fifth Amendment. Three of the four received the identical statement of the charges against them: “Information has been received from various confidential sources, that during the months of June, July and August, 1980, while confined in [Marion] you pressured other inmates to pay you commissary and perform homosexual acts with you. You applied this pressure by threat of harm to their person or their friends.” The statement of charges against the other petitioner, Ramirez-Rodriguez, was almost the same, the main difference being the addition of the words “or caused pressure to be applied” after “pressured.” Ramirez-Rodriguez was thought to be the ringleader of the gang of extortionists, the other three inmates to be members. No additional particulars of the charges were furnished to the petitioners, on the ground that to do so would identify the' confidential informants and expose them to retaliation by the petitioners or their friends. A prison investigator investigated the charges, primarily by interviewing the *1046 informants, and wrote up the results of his investigation in a report that he submitted to the prison’s Institution Discipline Committee. See 28 C.F.R. §§ 541.12(b), 541.14, 541.15. On the basis of this report the Committee found the petitioners guilty of the offenses charged. The Committee’s decision in each case essentially repeats the charges and adds: “Information received from confidential sources have [sic] proven reliable ... in the past.” Neither the investigator nor any of the informants testified, and the report was not submitted under oath. The informants’ statements as summarized in the report were virtually the only evidence before the Committee when it made its decisions, for though the petitioners could have testified before the Committee if they had wanted to, and some did, they had little to offer beyond general denials. By way of punishment the Committee ordered that Ramirez-Rodriguez, the supposed ringleader, forfeit 224 days of time off for good behavior and be placed in disciplinary confinement, see 28 C.F.R. §§ 541.18, 541.19, for an indefinite period with periodic reviews, see id., § 541.18(c). The Committee recommended that his disciplinary confinement be in the prison’s “Control Unit” — a kind of halfway house to solitary confinement, see Bono v. Saxbe, 620 F.2d 609, 613 (7th Cir.1980); 28 C.F.R. § 541.40 — where he remains to this day. The Committee ordered the other three petitioners put in disciplinary confinement for 30 days, and apparently they ended up in the Control Unit too, but have since been released from it. All four filed petitions for habeas corpus in federal district court, and appeal to this court from the denial of their petitions by a federal magistrate to whom the cases were referred by agreement of the parties. The petitioners’ counsel has not been allowed to read the investigator’s report on which the Institution Discipline Committee acted. As an original matter one might doubt whether prison inmates, who have already been deprived of their liberty pursuant to a valid conviction and sentence, can complain that they are being deprived of their liberty anew, as it were, by a disciplinary sanction which merely reduces the amenities of confinement and does not increase, directly anyway, its length; or if so whether they can get relief by applying for habeas corpus, traditionally a remedy for people who are unlawfully confined, as these inmates are not. Although Ramirez-Rodriguez has lost good-time credits, there is no allegation that were it not for that deprivation he would be eligible for release, on parole or otherwise. And while he, at least, can argue that he is seeking release from custody in a sense — custody in the Control Unit, conceived as a different sort of confinement from custody in the prison at large — the other three petitioners cannot make this argument, having since been released from their disciplinary confinement. Wolff v. McDonnell, 418 U.S. 539, 556-57, 94 S.Ct. 2963, 2974-2975, 41 L.Ed.2d 935 (1974), has stilled the first of these doubts and made clear that a prisoner can complain about a deprivation of liberty brought about by a disciplinary sanction. As for whether habeas corpus is a proper remedy for someone unlawfully confined in a Control Unit or the equivalent, many cases assume without discussion that it is. See, e.g., Johnson v. Avery, 393 U.S. 483, 89 S.Ct. 747, 21 L.Ed.2d 718 (1969); Kyle v. Hanberry, 677 F.2d 1386 (11th Cir.1982). Krist v. Ricketts, 504 F.2d 887 (5th Cir.1974) (per curiam), holds explicitly that it is. We think this holding is correct, especially after Wolff. If, as Wolff holds, due process can be violated by placing a prisoner in the Control Unit, the prisoner ought to have a remedy that gets him out of it, and habeas corpus is the normal remedy for one unlawfully confined. True, release would not be into the world outside the prison. But habeas corpus can be used to get a person released on parole if his parole was revoked unlawfully, even though in modern thinking parole is a form of custody, see Jones v. Cunningham, 371 U.S. 236, 243, 83 S.Ct. 373, 377, 9 L.Ed.2d 285 (1963). This shows that habeas corpus can be used to get from a more to a less restrictive custody — which is what Ramirez-Rodriguez is seeking. *1047 But'the other three petitioners are not seeking release from custody in the Control Unit or in any other place of confinement; their principal concern appears to be that the finding by the Institution Discipline Committee of a disciplinary infraction may delay their parole. It may. See 28 C.F.R. § 2.36(a). And it is irrelevant that they may not be entitled to an immediate parole if the finding of infraction is expunged; habeas corpus has been held to be the proper method of challenging the duration of imprisonment. Preiser v. Rodriguez, 411 U.S. 475, 487-88, 93 S.Ct. 1827, 1835, 36 L.Ed.2d 439 (1973). But unlike Preiser there is no automatic relationship in this case between the finding of infraction and the length of imprisonment. Cf. 411 U.S. at 487, 93 S.Ct. at 1835. It is in the Parole Board’s discretion whether a disciplinary infraction shall delay parole, and if so by how long. See 28 C.F.R. § 2.36(b). That makes this case somewhat similar to Williams v. Ward, 556 F.2d 1143, 1151 (2d Cir.1977), which held that a “petition challenging parole procedures but not demanding release or the granting of parole” could not be brought under the habeas corpus statute because, “Although the relief sought by petitioner may improve his chances for parole, the question of his release and of the length of his confinement still lies within the sound discretion of the [parole] board .... ” The proper remedy was therefore an action under 42 U.S.C. § 1983. See 556 F.2d at 1150-51. Williams, however, may be distinguishable from a case such as the present, where the regulations governing parole specifically authorize, although they do not compel, the Parole Board to delay parole because of disciplinary infractions, and the prisoner is challenging his “conviction” for such an infraction. It was entirely uncertain in Williams whether a successful challenge to the Parole Board’s procedures would result in the petitioner’s being paroled; but it is likely, though not certain, that getting a disciplinary finding expunged will accelerate a prisoner’s eligibility for parole and hence the date when he is paroled. The present case may therefore be closer to cases which hold that habeas corpus is an appropriate remedy “where a prisoner in custody pursuant to one conviction might receive credit on that sentence if he prevails in his attack on a prior, fully served, and unrelated sentence,” Harrison v. State of Indiana, 597 F.2d 115, 117 (7th Cir.1979), than it is to Williams. But we do not think we should decide this question on a record as scanty as the one in these cases. It contains no evidence from which one can infer that if the disciplinary proceedings against these three petitioners are set at naught the length of their imprisonment may — with sufficient probability to warrant judicial review on habeas corpus of the Institution Discipline Committee’s findings — be reduced. Ramirez-Rodriguez’s situation is different, not only because he is still in the Control Unit and habeas corpus is the appropriate procedure for getting out, but also because like the petitioner in Preiser he lost good time by reason of the infraction. But the other three cases must be remanded for a determination of the probable impact of vacating the Institution Discipline Committee’s findings on the length of imprisonment. If the evidence developed on remand shows that the outcome of the disciplinary proceedings against these three inmates is reasonably likely to delay their parole and thus lengthen their imprisonment, then presumably, by analogy to Preiser and Harrison, their suits can be maintained as habeas corpus proceedings. Another issue may become material on remand. The petitioners (other than Ramirez-Rodriguez) allege other consequences from the finding of disciplinary infraction besides delay in parole — unspecified “collateral consequences.” If they filed their habeas corpus petitions while they were in disciplinary confinement, a collateral consequence — even something so hypothetical and remote as the possibility that the record of that confinement might result in a stiffer sentence if the petitioners were later convicted of an unrelated offense — would be enough to prevent their release from mooting the petitions. See Carafas v. La *1048 Vallee, 391 U.S. 234, 237, 88 S.Ct. 1556, 1559, 20 L.Ed.2d 554 (1968); United States ex rel. Grundset v. Franzen, 675 F.2d 870, 873 (7th Cir.1982). But if they were released from disciplinary confinement before they filed their petitions for habeas corpus, those petitions were moot when filed and cannot be revived by collateral consequences. E.g., Harvey v. State of South Dakota, 526 F.2d 840 (8th Cir.1975) (per curiam). The record does not indicate when these inmates’ petitions were filed. That is another issue for the district court to consider on remand should it find that the action of the Institution Discipline Committee did not lengthen these petitioners’ imprisonment; and it may not be a distinct-issue. Even if the district court finds that the effect of the Committee’s action on the petitioners’ parole date is too remote for their petitions to be deemed an attack on the legality of their custody, that effect would be a collateral consequence sufficient to prevent release from disciplinary confinement from mooting their attack on the lawfulness of that confinement — provided they had filed their petitions for habeas corpus before being released from that confinement. On the merits of this appeal, then, the only issue is whether Ramirez-Rodriguez received due process of law. To decide this we must weigh the costs of additional procedural safeguards against the benefits, in reduced errors, which those safeguards would yield, bearing in mind that the benefits are a function both of the probability of error in the absence of a given safeguard and the cost of that error to the’ person claiming the safeguard. Mathews v. Eldridge, 424 U.S. 319, 335, 96 S.Ct. 893, 903, 47 L.Ed.2d 18 (1976); see Wolff v. McDonnell, supra, 418 U.S. at 567, 94 S.Ct. at 2980. Ramirez-Rodriguez argues that he should have received both a more detailed notice of the charges against him and more safeguards against erroneous findings at the hearing. The notice he received was so general that it was difficult for him to prepare any defense. But unfortunately the costs of additional notice would have been great. The essential information that Ramirez-Rodriguez needed to prepare his defense was the time and place of each alleged act of extortion. Any information falling short of this would have added nothing of value to the uninformative statement of charges which he did receive. But the additional information would have tipped him off to the names of all or most of the informants, whether the informants were the alleged victims of extortion, or witnesses to the alleged acts of extortion, or some of each. Such a tip-off could be lethal. Marion is the successor to Alcatraz as the maximum-security federal penitentiary. Not only does it have the most dangerous federal offenders, but it takes in, by way of assistance to the very overcrowded state prisons of the area, some of the most dangerous state offenders. Violent crime directed at other inmates is a common experience in our prisons today, and “rats” are frequent targets. See, e.g., United States v. Bruscino, 687 F.2d 938, 939 (7th Cir.1982). In the short run revealing the names of informants against a gang of extortionists and homosexual molesters could lead to the death or serious injury of some or all of the informants; in the long run it would dry up the supply of informants and allow extortion to rage unchecked through Marion. These costs outweigh in our judgment the benefits, substantial as they would be, of giving inmates accused of serious offenses the information they need to prepare an effective defense. But without such notice the adversary hearing so prized in American procedure is likely to have little meaning. The inmates will not know what the evidence is against them, so they will not be able to counter it with evidence of their own. The disciplinary proceeding- will be inquisitorial. The conditions in Marion today make that inescapable and it is therefore consistent with due process. But if the usual safeguards of an adversary procedure are unavailable it is all the more important that there be other safeguards, and we find none in this case. Cf. Helms v. Hewitt, 655 F.2d 487, 502 (3d Cir.1981). The report of *1049 the investigator is persuasive in its detail, but an investigative report, however vivid and apparently true, is not, as the magistrate thought, self-validating. The investigator was not called as a witness, although his identity is not confidential. He was not asked to swear to the truth of the report. None of the informants testified before, or was interviewed by, the Institution Discipline Committee. The Committee would not even vouch for the credibility of the investigator or his informants. All it said was that information received from confidential sources had proved reliable in the past — not necessarily information from these sources, compiled by this investigator. Not all prison inmates who inform on other inmates are telling the truth; some are enacting their own schemes of revenge; and though it is unlikely that all or most of the informants interviewed for the investigative report were lying, if some were that could have affected the severity of the sanction that the Committee meted out to Ramirez-Rodriguez. Since the procedures used by the Committee in this case carried with them a significant risk of error, and the consequences of error in this setting are serious for Ramirez-Rodriguez, it remains only to consider the costs of extending to him some additional procedural safeguards. We cannot determine these costs on the scanty record before us. One safeguard Ramirez-Rodriguez is seeking is to let his lawyer read the investigative report. We have no reason to believe that his counsel would give Ramirez-Rodriguez the names of the informants, or information from which those names could be deduced, but we do not know whether it would be safe to allow inmates’ counsel access to such reports as a general rule and we do not think that courts or prison officials should try to decide which lawyers are trustworthy. It should be perfectly feasible to insist that the Institution Discipline Committee require that its investigative reports be under oath and that the investigator appear in person and be available for cross-examination; but we are not sure whether it is feasible to go further and require the Committee to interview — in camera, of course— some or all of the informants. Maybe that is too dangerous. Maybe the prison grapevine is so efficient that Ramirez-Rodriguez would learn who had been interviewed by the Committee — or maybe, whether or not it is that efficient, informants would be afraid to give testimony other than through the investigator. Cf. Wolff v. McDonnell, supra, 418 U.S. at 568, 94 S.Ct. at 2980. You need a better feel for Marion than we have to make judgments in these sensitive matters; and they happen to be matters of life and death. Thus, while we believe that Ramirez-Rodriguez did not receive the process he was due under the Fifth Amendment, we do not know how much more process he could have been given without jeopardizing the lives, as well as the willingness to inform, of other inmates. We therefore remand No. 81 — 2496 (Ramirez-Rodriguez) for a hearing in the district court at which officials of Marion can testify on the feasibility of the various procedural safeguards discussed in this opinion. We direct that the proceedings on remand be conducted by Chief Judge Foreman in person in view of his long experience in prisoner litigation out of Marion and the delicacy and importance of the issues. The three other cases are remanded for a determination as to whether they can be maintained as habeas corpus cases. If so, Chief Judge Foreman will proceed to the merits of those cases. In deciding the threshold merits question of whether confinement to disciplinary segregation is a deprivation of “liberty” within the meaning of the due process clause of the Fifth Amendment, Chief Judge Foreman will want to consider the possible bearing of the Supreme Court’s recent decision in Hewitt v. Helms,-U.S.-, 103 S.Ct. 864, 74 L.Ed.2d 675 (1983).