dissenting:
In this case, we confront the question whether the government may detain a dangerous defendant before trial in order to protect society. The majority believes that the Due Process Clause of the Fifth Amendment prevents the government from detaining such a defendant for that reason for even a brief period of time — presumably even for a day. As will be seen below, I disagree. I believe that pretrial detention for dangerousness does not violate the Due Process Clause when there is clear and convincing proof that a person already under indictment for a serious crime would commit another crime if released, if the detention has not continued so long as to constitute punishment: I would affirm the judgment of the district court, and I respectfully dissent from that portion of the majority opinion that holds the Bail Reform Act unconstitutional.
I. Is the Bail Reform Act Facially Consistent With Substantive Due Process?
The Bail Reform Act of 1984, 18 U.S.C. § 3141 et seq., permits the government to move before trial for the detention of a defendant indicted for the serious crimes with which appellants are charged, on the grounds that the defendant poses a threat to the safety of the community. 18 U.S.C. § 3142(f). Following a hearing, the judicial officer must order the detention if the officer finds that no conditions of release short of confinement will avert this threat. 18 U.S.C. § 3142(e).
We recently considered this aspect of the Bail Reform Act in United States v. Melendez-Carrion, 790 F.2d 984 (2d Cir.1986). As both Judge Newman’s opinion and my concurring opinion in that case point out, at 998, 1006, this statutory scheme has serious constitutional implications. Mentally competent adults not convicted of crime obviously have a substantial liberty interest in remaining free of confinement. Pretrial detention for dangerousness curtails that interest in an untraditional manner. But the novelty of the statute does not determine its lawfulness on its face. Rather, the relevant question is whether detention for a brief period to prevent future crimes while on release comports with the fundamental fairness demanded by the substantive component of the Due Process Clause.
Congress enacted the preventive detention provisions of the Bail Reform Act out of a deep concern about “the growing problem of crimes committed by persons on release.” S.Rep. No. 225, 98th Cong., 2d Sess. 6 (1983), reprinted in 1984 U.S.Code Cong. & Admin.News at 3182, 3188-89; see United States v. Rodriguez, 794 F.2d 24, 26-27 (2d Cir.1986). Congress concluded that, at least as to “a small but identifiable group of particularly dangerous defendants,” the pressing need to protect the community called for confinement before trial. Senate Report, supra, at 6-7, 1984 U.S.Code Cong. & Admin.News at 3188, 3189. In Bell v. Wolfish, 441 U.S. 520, 534 n. 15, 99 S.Ct. 1861, 1871 n. 15, 60 L.Ed.2d 447 (1979), the Supreme Court reserved the question whether such an objective could constitutionally justify pretrial detention. In Schall v. Martin, 467 U.S. 253, 104 S.Ct. 2403, 81 L.Ed.2d 207 (1984), the Court settled the issue with respect to juvenile detainees. The Court rejected a due process challenge to New York’s scheme of detaining for a period generally not exceeding 17 days juveniles charged with crime in order to prevent them from committing crimes while on release. The Court held that the system was compatible with substantive due process because detention based on dangerousness served a “legitimate and compelling” goal and did not amount to “punishment.” Id. at 264-75, 104 S.Ct. at 2409-16.
*76It is true that Schall concerned detention of juveniles, whose liberty interest in freedom from confinement while charges are pending against them may be inferior to that of adults in a similar position. Id. at 265,104 S.Ct. at 2410. However, the societal interest identified as “compelling” by the Court only two years ago in Schall does not vary in strength with the age of the person to be detained. If anything, the need to shield the community from the hazards of pretrial crimes committed by adults is more compelling, since adults may have superior access to the means of committing more serious and far-reaching offenses. Accordingly, the government has advanced a legitimate and compelling reason for pretrial confinement of adults accused of crime. In light of this reason it is not surprising that other circuit courts that have considered the issue have held this aspect of the Bail Reform Act constitutional on its face. United States v. Portes, 786 F.2d 758 (7th Cir.1986); United States v. Accetturo, 783 F.2d 382 (3d Cir.1986); see also United States v. Edwards, 430 A.2d 1321 (D.C.App.1981) (in banc) (upholding preventive detention under District of Columbia bail statute), cert. denied, 455 U.S. 1022, 102 S.Ct. 1721, 72 L.Ed.2d 141 (1982).
This does not, however, end the due process inquiry. Since the Bail Reform Act curtails a defendant’s precious interest in liberty, the Due Process Clause requires heightened scrutiny of the statutory means used to affect such curtailment. Under 18 U.S.C. § 3142(e), the judicial officer may order a person confined for dangerousness only after a finding that no conditions of release will assure the safety of the community. Under the statute, detention is meant to be a sparingly dispensed remedy of last resort, used only when needed to protect society in cases where there is proof that less drastic measures would fail. Moreover, appellants do not make any persuasive claim that the statutory procedures required as prerequisites to a finding of dangerousness are unfair or provide insufficient “protection against erroneous and unnecessary deprivations of liberty.” Schall v. Martin, supra, 467 U.S. at 274, 104 S.Ct. at 2415.
Due process also dictates that the government not pursue its goals through “conduct that shocks the conscience.” Ro-chin v. California, 342 U.S. 165, 172, 72 S.Ct. 205, 209, 96 L.Ed. 183 (1952). It is not necessary to be enthusiastic about comparatively brief pretrial detention of accused persons on the basis of predicted future crimes to acknowledge that the practice does not violate that norm. It may be that such predictions are less than scientific, and that it is difficult to assess whether the verifiable gains in reducing crime justify the curtailment of liberty that results. But that assessment is for the legislature, not for us, to make, so long as there is no constitutional violation. There is nothing inherently shocking to the conscience in using a prediction of future criminality to justify confinement. When a person has been convicted of crime, judges legitimately rely on such predictions in setting the length of a prison sentence. Identical considerations shape the decision whether to grant bail to a convicted person pending appeal. Carbo v. United States, 82 S.Ct. 662, 7 L.Ed.2d 769 (Douglas, Circuit Justice, 1962); United States v. Anderson, 670 F.2d 328 (D.C.Cir.1982) (per curiam); United States v. Provenzano, 605 F.2d 85 (3d Cir.1979). In both situations, a judge’s informed assessment of a person’s dangerousness may increase the length of time that the person is incarcerated. The law also acknowledges that a person arrested for crime may be summarily detained for a short while before coming before a magistrate not only to prevent flight or danger to the judicial process, but also to bar commission of future crimes. Cf. Gerstein v. Pugh, 420 U.S. 103, 114, 95 S.Ct. 854, 863, 43 L.Ed.2d 54 (1975) (once suspect is in custody “there no longer is any danger that suspect will escape or commit further crimes” while police submit evidence to magistrate). If the police may detain an arrestee for a short time to prevent danger to the community pending a *77finding of probable cause, Congress may direct judges to do the same after an indictment and pending trial.
There is no doubt, as the majority points out, that putting someone in jail simply because he is thought to be dangerous is not part of our tradition of liberty. But the issue is not as stark or as simple as that. Under the Bail Reform Act, a number of limitations come into play before a person can be confined. First, the person must be charged with a crime, and not just any crime, but a serious crime as specified in the statute. Second, the government must persuade a judicial officer by clear and convincing evidence that the defendant is a danger to the community. Third, the government must also persuade the judicial officer by the same standard of proof that no condition of release will deter the defendant from committing a crime while at liberty. Fourth, the detention is meant to be of limited duration — until it can be determined whether the defendant is guilty of the crime charged. So the constitutional issue before us is not simply whether an individual can be confined because he is dangerous. It is whether a defendant already indicted for a serious crime can be denied bail in the pretrial period because there is clear and convincing evidence that the defendant will otherwise commit another crime while on release. So stated, the proposition still raises grave constitutional issues but the answer to them is not easy. For example, if a member of a terrorist organization is indicted for blowing up an airliner for political reasons and there is clear and persuasive evidence that the defendant will do so again if not confined, it is not self-evident to me that society must nevertheless immediately release him on bail until he is tried.
The majority sees an “anomaly” in holding the Bail Reform Act constitutional because
a person who has been charged with crimes of violence but not yet convicted may be incarcerated on the ground that he will commit crimes of violence in the future, whereas persons who have been neither charged with nor convicted of violent crimes, or have served their sentence of conviction on those charges, may not be incarcerated on the basis of such a prediction.
I see no anomaly. The Bail Reform Act applies only to persons indicted for serious offenses, not to unaccused citizens plucked from the population at large. Indictment basically alters the due process inquiry by establishing probable cause to believe that the accused has committed specific acts constituting a serious crime. That legal finding gives the government a concrete basis, not available as to persons not accused, for predicting that the person will commit another crime before trial.1 In other contexts, probable cause is also sufficient to permit significant restraints on liberty. See Gerstein v. Pugh, supra, 420 U.S. at 118-20, 95 S.Ct. at 865-66 (detention of arrested person pending further proceedings); Morrissey v. Brewer, 408 U.S. 471, 485-87, 92 S.Ct. 2593, 2602-03, 33 L.Ed.2d 484 (1972) (detention of parolee pending final decision whether to revoke parole). Moreover, as noted above, the Bail Reform Act has a built-in limit on the length of the period for which an accused person may be detained. The end of that period coincides with the disposition of the underlying criminal charge against the defendant, whether by plea or by trial. While undue pretrial delay may further affect the due process equation, see Part II below, detention under the Bail Reform Act is meant to be limited to preserving the status quo, as far as the community is concerned, pending the resolution of specific legal charges.
For these reasons, I conclude that detaining indicted defendants under the Bail Re*78form Act for a limited time on the basis of clear and convincing evidence that nothing short of confinement will prevent them from violating the law while on release does not violate any norm of decency implicit in the concept of ordered liberty, and does not violate the Due Process Clause.
II. Does the Statute as Applied Comport with Due Process?
Up to this point, I have considered whether the Bail Reform Act is unconstitutional on its fa.ce. The conclusion that.it is not does not determine whether the statute as applied comports with due process. The government’s powers of preventive detention in a given case are not limitless. Every other appellate court that has examined the lawfulness of this practice under the Bail Reform Act has indicated that pretrial detention to prevent future crimes may be invalid if unduly prolonged. Portes, supra, 786 F.2d at 768 & n. 14; Accetturo, supra, 783 F.2d at 388; see also United States v. Theron, 782 F.2d 1510, 1516 (10th Cir.1986); cf. United States v. Edwards, supra, 430 A.2d at 1333 (stressing 60-day time limit of D.C. preventive detention scheme determined to be constitutional). This reflects widespread agreement that the passage of time can tip the scales of the balance mandated by due process.
To put the proposition another way, even legitimate and compelling objectives cannot justify detaining a person indefinitely before trial. At some point in time, the harsh burdens of extended confinement will exceed the bounds of due process. As I stated in my concurring opinion in Melendez-Carrion, at 1006-09, lengthy pretrial incarceration can run afoul of the rule that the government may not punish a person not convicted of crime. Bell v. Wolfish, 441 U.S. 520, 535-36, 99 S.Ct. 1861, 1871-73, 60 L.Ed.2d 447 (1979). Whether a measure is punitive in nature depends both on whether the action is imposed for the purpose of punishment and whether it is excessive in relation to the legitimate, nonpunitive aims assigned to it. Schall, supra, 467 U.S. at 269, 104 S.Ct. at 2412; Bell v. Wolfish, supra, 441 U.S. at 538-39, 99 S.Ct. at 1873-74.
In Melendez-Carrion, Judge Newman and I were willing to assume that Congress did not intend preventive detention under the Bail Reform Act to be a punitive measure. However, the second branch of the due process inquiry demands that confinement before trial not be excessive in relation to the statute’s goals. Accordingly, undue prolongation of detention before trial can transform what is initially a valid regulatory measure into punishment prohibited by the Due Process Clause. That is so because the traditional test for distinguishing regulatory from penal measures set forth in Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168-69, 83 S.Ct. 554, 567-68, 9 L.Ed.2d 644 (1963), and reaffirmed in Bell v. Wolfish, supra, looks to whether a measure “involves an affirmative disability or restraint” and whether “it has historically been regarded as a punishment.” Detention before trial undoubtedly imposes a major disability and restraint on the defendant. As I explained in Melendez-Carrion, detention for a long period of time founded on a person’s danger to the community is also so harsh that it comes to resemble a traditional prison sentence. Since none of the traditionally “regulatory” reasons for jailing a person without trial, such as the defendant’s propensity to flee or to tamper with witnesses, justify confinement in such a case, such lengthy detention would historically be seen as a punishment. Accordingly, unduly extended incarceration for general dangerousness of persons never convicted of any crime can cross the line separating a valid regulatory measure from punishment imposed in violation of the Due Process Clause. Melendez-Carrion, supra, at 1008-09.
In Melendez-Carrion, I took the position that the length of preventive detention requires assessment on a case-by-case basis, since due process does not necessarily set a bright line limit for length of pretrial con*79finement. Here, Salerno and Cafaro have been detained for dangerousness for a little over three months. Arguably, time alone should not always determine whether a given instance of detention for dangerousness has become punitive. Though length of time may be the best measure of the impact of incarceration on a person accused of crime, perhaps due process judgments should take into account the reasons for pretrial delay in unusual circumstances. Relevant factors might include the complexity of pretrial preparations for the underlying case, and whether the defendant or the government has added needlessly to that complexity or has egregiously delayed the proceedings. See Accetturo, supra, 783 F.2d at 388. I cannot conclude on this record that the restriction on appellants’ liberty has degenerated into punishment. Were we to affirm, I would not foreclose appellants from renewing their constitutional claims at a later date.2 But at this point, there is no constitutional basis on which to set aside Salerno’s and Cafaro’s detention. We should affirm. I respectfully dissent.
. Also inapposite is the analogy to those who have served sentences for past offenses but have not been again indicted. Double jeopardy considerations apply to them. In addition, a principal rationale for sending convicted offenders to prison is that their sentence will rehabilitate them and deter them from future misconduct. Neither consideration applies to those not yet convicted of a crime.
. Pretrial detention under any circumstances is an extraordinary remedy with serious due process implications. If the government seeks the benefit of that remedy it must exert itself to accelerate the date for a trial. Quite often, reliance on the usual processes of trial administration will not be enough. If this is so, the government must take whatever special steps are needed to move the trial date forward. In complex multi-defendant cases, that may require that the government seek a severance. In cases involving many hours of taped intercepted material, the government may have to arrange for swift and reliable transcription, by extraordinary means if necessary, before moving for detention or immediately after obtaining it. Should the government be unwilling or unable to shoulder the cost of these procedures, pretrial detention may not be available.
The Speedy Trial Act, 18 U.S.C. § 3164, does mandate that cases involving detained persons be given priority. On June 4, the trial judge apparently set a date shortly sifter October 1, 1986 for submission of pretrial motions in this case. Such a leisurely pace in a criminal case involving pretrial detention is difficult to justify.