Chief Judge, joined by LAY, Senior Circuit Judge, and McMILLIAN, Circuit Judge, concurring in part and dissenting in part.
James Michael Wise has been convicted on his plea of guilty to two counts of counterfeiting currency of the United States. The District Court sentenced him to prison for two years and eight months, or 32 months. The sentence was within the Guideline range of 30 to 37 months. The range was fixed by the Court’s holding that the offense level was 19 and the criminal history category I. On appeal, Wise makes two arguments: he should have been given a two-level reduction for acceptance of responsibility, and it was wrong to give him a four-level enhancement for being the leader or organizer of criminal activity with five or more participants. This Court affirms. With respect to acceptance of responsibility, I agree. But with respect to Wise’s role in the offense, I dissent. Maybe Wise deserves the four-level enhancement. Maybe the same finding would be made if proper procedures were followed. But the procedures followed here violated the Confrontation Clause of the Sixth Amendment and the Due Process Clause of the Fifth Amendment. Moreover, even if there is no constitutional violation, the judgment should be reversed. The District Court erred in placing the burden of persuasion on Wise and in basing its findings on unreliable information.
I.
First, the facts. Without the four-level adjustment for role in the offense, Wise’s offense level would be 15. His Guideline range would be 18 to 24 months. Assuming a sentence at the lower end of the range (by analogy to the sentence actually imposed, 32 months in a range of 30 to 37), his sentence would be in the neighborhood of 20 months. Thus, the enhancement in question is worth about 1 year, the difference between the 32 months actually imposed and the approximately 20 months that would have been imposed without the adjustment. Wise has served 12 additional months.
The only witness called at the sentencing hearing to support this or any other aspect of the sentence was a probation officer. He had no personal knowledge of any facts relevant to the four-level enhancement. Sentencing Transcript (Tr.) 14. The “evidence” with respect to each person poten*406tially involved with Wise in the offense does not take long to summarize.
1. Terry Lankford. The probation officer testified that Lankford was charged “in state court as a result of [his] ... involvement in this offense.” Tr. 14. Lankford is presently serving a sentence. No court records were presented to substantiate or explain this testimony. According to the probation officer, Lankford pleaded guilty to a state crime. The probation officer said that he had “information that [Lankford] admitted that he received the [counterfeit] money from Mr. Wise.” Tr. 31. So far as the transcript shows, the witness never spoke with or even saw Lankford.
2. Earl Dean Laughlin, Jr. This man also pleaded guilty in a state court. “As far as I know,” the probation officer testified, “he [Laughlin] advised that he received money from Mr. Wise.” Tr. 31 (emphasis supplied). The witness had never talked to Laughlin. He had no direct information that Laughlin even knew Mike Wise. The information the witness had about Laughlin came from a Secret Service investigative report “and conversation with the prosecuting attorney’s office.” Tr. 32. (I take it the reference to “prosecuting attorney” is to the state official who prosecuted Laughlin, though the record is not absolutely clear.) We do not know who had the “conversation” referred to, whether the Secret Service, the United States Attorney’s office, or the probation officer himself.
3 and 4. Two individuals, unnamed, on federal pretrial diversion. These two persons apparently admitted their own involvement directly to the probation officer, and, in addition, said something to implicate Wise. The probation officer refused to give their names.1
5. Another unnamed person. The witness said: “I also have an individual who had his state parole revoked as a result of his involvement in the case. He admitted that he was involved.” Tr. 31. Again, we have no state-court records to substantiate this testimony. This person, apparently, is not on federal pretrial diversion and has no other claim to anonymity, but the probation officer still failed to give his name.
In sum, we have the following state of affairs: Wise has served an additional 12 months in a federal penitentiary because a witness with no personal knowledge says that a Secret Service file, or some unnamed person in a state prosecutor’s office, says that Earl Dean Laughlin, who is not in court, says that Wise was involved with him in the offense and was its leader or organizer. This is double or triple hearsay at best. We do not know exactly even what Laughlin said, or what anyone who ever talked directly to him says he said.
I am astonished that a court of the United States would order anyone kept in prison on such a basis.
II.
The interpretation of any written document, including the Constitution, should begin with its words, though it will rarely end there. The Sixth Amendment says:
In all criminal prosecutions, the accused shall enjoy the right to a speedy and public trial, by an impartial jury of the State and district wherein the crime shall have been committed, which district *407shall have been previously ascertained by law, and to be informed of the nature and cause of the accusation; to be confronted with the witnesses against him; to have compulsory process for obtaining witnesses in his favor, and to have the Assistance of Counsel for his defence.
If “plain meaning” is the criterion, this is an easy case. Surely no one would contend that sentencing is not a part, and a vital one, of a “criminal prosecution.” To be sure, it is not words in isolation, but words in context, that matter, and the Framers of the Bill of Rights knew nothing of sentencing proceedings separate from the trial itself. Most sentences followed automatically from conviction. In general, the punishment for felony was death. Moreover, a literal reading of the Amendment would require jury trials of facts relevant to sentencing, and the Constitution is not generally thought to contain such a requirement. See United States v. Galloway, 976 F.2d 414 (8th Cir.1992) (en banc) (trial by jury not required); but see id. at 442-43 (Bright, J., dissenting) (trial by jury required if the facts in issue are themselves a distinct crime). On the other hand, surely defendants in sentencing proceedings enjoy the protection of those portions of the Amendment relating to notice, cf. Burns v. United States, — U.S.-, 111 S.Ct. 2182, 115 L.Ed.2d 123 (1991), the assistance of counsel, and compulsory process for the attendance of witnesses. I have heard no one argue, to take one example, that defendants have no constitutional right to counsel at the time of sentencing. So why not confrontation? It seems to me that the words of the Amendment at least point in the direction of recognizing the right to confront adverse witnesses.2
What about the intent of the Framers? If the words’ meaning isn’t plain, what evidence do we have of the Framers’ purpose? The answer seems to be, very little. What we do know, however, points towards recognizing the right of confrontation. The history is well summarized in Justice Thomas’s separate opinion in White v. Illinois, — U.S. -, -, 112 S.Ct. 736, 745-46, 116 L.Ed.2d 848 (1992):
In 16th-century England, magistrates interrogated the prisoner, accomplices, and others prior to trial. These interrogations were “intended only for the information of the court. The prisoner had no right to be, and probably never was, present.” 1 J. Stephen, A History of the Criminal Law of England 221 (1883). At the trial itself, “proof was usually given by reading depositions, confessions of accomplices, letters, and the like; and this occasioned frequent demands by the prisoner to have his ‘accusers,’ i.e., the witnesses against him, brought before him face to face_” Id., at 326. See also 5 Wigmore, supra, § 1364, at 13 (“there was ... no appreciation at all of the necessity of calling a person to the stand as a witness”; rather, it was common practice to obtain “information by consulting informed persons not called into court”); 9 W. Holdsworth, History of English Law 227-229 (3d ed. 1944). The infamous trial of Sir Walter Raleigh on charges of treason in 1603 in which the Crown’s primary evidence against him was the confession of an alleged co-conspirator (the confession was repudiated before trial and probably had been obtained by torture) is a well-known example of this feature of English criminal procedure. See Pollitt, The Right of Confrontation: Its History and Modern Dress, 8 J.Pub.L. 381, 388-389 (1959); 1 Stephen, supra, at 333-336; 9 Holds-worth, supra, at 216-217, 226-228.
*408Apparently in response to such abuses, a common-law right of confrontation began to develop in England during the late 16th and early 17th centuries. 5 Wigmore, supra, § 1364, at 23; Pollitt, supra, at 389-390. Justice Story believed that the Sixth Amendment codified some of this common law, 3 J. Story, Commentaries on the Constitution of the United States 662 (1833), and this Court previously has recognized the common-law origins of the right. See Salinger v. United States, 272 U.S. 542, 548, 47 S.Ct. 173, 175, 71 L.Ed. 398 (1926) (“The right of confrontation did not originate with the provision in the Sixth Amendment, but was a common-law right having recognized exceptions”). The Court consistently has indicated that the primary purpose of the Clause was to prevent the abuses which had occurred in England. See Mattox v. United States, 156 U.S. 237, 242, 15 S.Ct. 337, 339, 39 L.Ed. 409 (1895) (“The primary object of the [Confrontation Clause] was to prevent depositions or ex parte affidavits, such as were sometimes admitted in civil cases, being used against the prisoner in lieu of a personal examination and cross-examination of the witness ... ”); California v. Green, 399 U.S. [149] at 156, 90 S.Ct. [1930] at 1934 [26 L.Ed.2d 489 (1970)] (“It is sufficient to note that the particular vice that gave impetus to the confrontation claim was the practice of trying defendants on ‘evidence’ which consisted solely of ex parte affidavits or depositions secured by the examining magistrates, thus denying the defendant the opportunity to challenge his accuser in a face-to-face encounter in front of the trier of fact”); id., at 179, 90 S.Ct., at 1946 (Harlan, J., concurring) (“From the scant information available it may tentatively be concluded that the Confrontation Clause was meant to constitutionalize a barrier against flagrant abuses, trials by anonymous accusers, and absentee witnesses”); Dutton v. Evans, 400 U.S. [74] at 94, 91 S.Ct. [210] at 222 [27 L.Ed.2d 213 (1970)] (Harlan, J., concurring in result) (the “paradigmatic evil the Confrontation Clause was aimed at” was “trial by affidavit”).
What is happening to Wise in this case is not exactly the same thing, but it’s close enough to cause me to suspect that the Framers would easily recognize some of the abuses that concerned them. Wise has been tried, so far as sentencing goes, on the basis of prosecution documents: the presentence report (which relies heavily on prosecution sources), investigative reports of the Secret Service, composed of employees of the adverse party, and so forth. Some of these documents he has not even seen. Some of the crucial people mentioned in them are not even named. Such a procedure attacks the core values of the Confrontation Clause.
About all this the Court says nothing. It does not quote the Sixth Amendment, nor does it mention the Framers’ intent. Its reliance, instead, is on recent history. Traditionally, the Court says, procedure at sentencing has been informal, and confrontation has not been required. The argument depends entirely on Williams v. New York, 337 U.S. 241, 69 S.Ct. 1079, 93 L.Ed. 1337 (1949), and collapses on close examination. Williams, in the first place, is not a Confrontation Clause case at all. It is a due-process case from a state court, decided before the Confrontation Clause was held, in Pointer v. Texas, 380 U.S. 400, 85 S.Ct. 1065, 13 L.Ed.2d 923 (1965), applicable to the states. And William’s whole underpinning is the system then obtaining of discretionary sentencing, a system believed, in those days, to represent the best of modern penology — the very system the Sentencing Reform Act of 1984 was passed to destroy! Passages can be cited that seem, if one reads nothing else, to support the Court’s position, but such an approach ignores the most basic principle of legal analysis: Case law is based on reason, and if the reason justifying the result of a case ceases to exist, the case loses its authority. Ces-sante ratione, cessat ipsa lex. If this is not enough, consider two more aspects of Williams. First, it was decided in a gentler age, when probation officers were supposed “to aid offenders,” 337 U.S. at 249, 69 S.Ct. at 1084 (emphasis added), not com*409pile evidence or appear as witnesses against them. Second — and this has been called “the most important line in the opinion,” Saltzburg, Sentencing Procedures: Where Does Responsibility Lie?, 4 Fed. SentRep. 248, 248 (1992) — Williams did not contest any of the facts allegedly shown by the hearsay used against him! “The accuracy of the statements made by the judge as to appellant’s background and past practices was not challenged by appellant or his counsel, nor was the judge asked to disregard any of them or to afford appellant a chance to refute or discredit any of them by cross-examination or otherwise.” Williams, 337 U.S. at 244, 69 S.Ct. at 1081-82 (emphasis added).
In short, Williams was decided in the context of a sentencing “system that fo-cuse[d] on subjective assessments of rehabilitative potential....” Saltzburg, supra, 4 Fed.Sent.Rep. at 250. Sentencing was intuitive and impressionistic. The world is very different now. For most defendants in the federal courts, sentencing is what the case is really about. And the process is much more fact-driven, in the sense that specific, identifiable increments of punishment are a legal consequence of specific past conduct, other than the offense of conviction itself. Some discretion remains, it is true. The court is free to choose within the Guidelines range — but the range is fairly narrow; the upper limit may not exceed the lower by more than 25 per cent. The court may even depart from the Guidelines — but only if certain findings are made, the propriety of which, like almost every other question under the Guidelines, is reviewable on appeal at the instance of either side. And the importance of the sentence imposed by the court is much greater. Parole has been abolished. Sentences are diminished only by the accrual of good time, which is limited to slightly less than 15 per cent. The world of sentencing has in fact been “revolutionized.” The word is not mine, but the Supreme Court’s. Burns v. United States, — U.S. at-, 111 S.Ct. at 2184.
Williams is obsolete for other reasons as well. Trial judges no longer have unlimited discretion in either the range of information they may consider or the punishment they may impose. Not only do the Guidelines attach a mandatory penalty to proof of particular facts, but they greatly restrict the type of information a judge may consider. Under U.S.S.G. Ch. 5, Pt. H, age, education, mental, emotional, and physical condition, employment history, and family and community ties are no longer relevant to sentencing.3 The Court decides this is not of great significance since the “sentencing judge still considers information not strictly relevant to a defendant’s guilt or innocence, such as uncharged conduct, when fashioning a sentence.” Ante at 398. As to this category, the judge “still requires full and complete information.” Ante at 399. Therefore, the Court reasons, the Guidelines have not “fundamentally transformed the search for information at the sentencing phase.” Id. I disagree. What the Court appears to be saying is that as long as we still conduct a broad, unlimited search for the information the Guidelines do allow, it is irrelevant that we have eliminated some categories of allowable information. Somehow the deck of available facts seems a bit more stacked against the defendant under this regime. Before, both the defendant and the prosecution contributed information to the process. The defendant played his cards — employment history, mental, physical, and emotional state, potential for rehabilitation, and other favorable policy arguments — and the prosecutor played his. The judge would then consider all of the information and impose a sentence which was fair and reasonable under those particular circumstances. The Court now says there is no need to change the rules, even though the prosecutor has all the cards and the judge is required by law to impose a sentence depending not upon his independent analysis of the facts, but *410upon what cards the prosecutor chooses to play. In general, the sentencing factors so important under the new scheme all point in one direction: towards longer sentences. The major exception is cooperation of the defendant with the prosecution, and in many cases this fact is relevant only on the prosecution’s motion.
Next, the Guidelines have replaced the goals of rehabilitation and individualized sentencing with punishment and uniformity. See 18 U.S.C. § 3553(a)(2) and U.S.S.G. § 1A3. The Court admits this, but states that since the sentencing system still places a premium on obtaining all evidence relevant to the new goals, the absence of the right of confrontation still stands. Ante at 399. Once again, the Court misses the mark. We are no longer fitting the punishment to the criminal, but only to the crime. The information obtained is not used to decide how best to rehabilitate the defendant — a goal which is arguably shared by the defendant — but to determine how long he will be imprisoned. Allowing a sentencing judge to conduct an inquiry free from evidentiary strictures was perhaps permissible in achieving this mutual goal and sentencing the defendant individually. It has no place in a system where the goal is no longer mutual. • When you change the prosecution’s goals to retribution and uniform sentencing, you should add procedural safeguards to ensure that the proceeding remains fair to the defendant.
Furthermore, sentencing is now fully adversarial. Rule 32 “contemplates full adversary testing of the issues relevant to a Guidelines sentence.” Burns, — U.S. at -, 111 S.Ct. at 2185. “[T]he facts that are now relevant to sentencing are far more susceptible to trial-type proof than those facts relevant under the rehabilitative scheme, because they implicate claims about the offense of conviction or other related offenses rather than claims about the offender’s background.” Note, An Argument for Confrontation Under the Federal Sentencing Guidelines, 105 Harv. L.Rev. 1880, 1886 (1992). (Remember that Congress has now expressly disavowed rehabilitation as a purpose of sentencing. 28 U.S.C. § 994(k)). The Guidelines themselves recognize that a significant change has occurred. “The court's resolution of disputed sentencing factors will usually have a measurable effect on the applicable punishment. More formality is therefore unavoidable if the sentencing process is to be accurate and fair.” U.S.S.G. § 6A1.3, commentary. In sum, “sentencing proceedings are arguably the most important judicial business conducted by Article III judges.” Becker & Orenstein, The Federal Rules of Evidence After Sixteen Years, 60 Geo.Wash.L.Rev. 858, 889 (1992). No reasonable effort should be spared to make these proceedings fair and accurate in perception and reality.
The position I advocate has significant support in pre-Guidelines Supreme Court case law. The closest case is Specht v. Patterson, 386 U.S. 605, 87 S.Ct. 1209, 18 L.Ed.2d 326 (1967). There, confrontation rights were required in the context of a state sentencing system that increased the maximum sentence from ten years to life upon a finding by the court that the offender, on the basis of past conduct, “if left at large, constitutes a threat of bodily harm to members of the public, or is an habitual offender and mentally ill.” Id. at 607, 87 S.Ct. at 1211. The system contemplated an adversary proceeding and “a new finding of fact that was not an ingredient of the offense charged.” Id. at 608, 87 S.Ct. at 1211-12. The Supreme Court held that the defendant was entitled to “an opportunity to be heard, be confronted with the witness against him, have the right to cross-examine, and ... offer evidence of his own.” Id. at 610, 87 S.Ct. at 1212. Williams was easily distinguished.
The Confrontation Clause applies, and the case should be remanded for resentenc-ing in a manner consistent with the Clause.
III.
The Court suggests, ante at 398 n. 2, that the issue might better be analyzed in terms of due process than under the Confrontation Clause. In my view, this case would come out the same way under the *411Due Process Clause. The right to confront and cross-examine witnesses is part of the process that is due before a defendant may be deprived of his liberty for an additional period of time.
First, there is surely no doubt that what the defendant lost in the present case is “liberty” within the meaning of the Due Process Clause of the Fifth Amendment. The Court’s opinion assumes as much, and it would be hard to deny. Wise has spent 12 more months in jail because of the contested facts. He has lost liberty in the most basic sense. He has been in jail instead of free. When this sort of loss occurs as punishment for crime or some other sort of infraction, one has been deprived of the sort of “liberty” that derives its protection directly from the Constitution. The “liberty interest” analysis of more recent case law is unnecessary. See Jones v. Mabry, 723 F.2d 590 (1983).
Even if a liberty-interest analysis had to be undertaken, the same result would follow. Our opinion in United States v. Galloway, 976 F.2d 414, 425 (8th Cir.1992) (en banc), so holds. “ ‘Once convicted, a defendant has a liberty interest in the correct application of the Guidelines within statutory limits....’” Ibid, (quoting United States v. Mobley, 956 F.2d 450, 455 (3d Cir.1992)).
The real question is, what process is due? Neither McMillan v. Pennsylvania, 477 U.S. 79, 106 S.Ct. 2411, 91 L.Ed.2d 67 (1986), nor Galloway decides the issue with respect to the right of confrontation. McMillan holds that due process does not require the government to prove a sentencing factor by clear and convincing evidence. A preponderance is enough to satisfy due process. And Galloway holds that the use of relevant conduct in sentencing does not trigger the rights to indictment, jury trial, and proof beyond a reasonable doubt. There is a distinction between the rights denied in Galloway and the right at issue here. The Galloway Court held that uncharged conduct is a sentencing factor, not a new element of the offense and not a separate offense. Therefore, it reasoned, indictment and a jury trial are not necessary. Additionally, the standard of proof is elastic and stretches to fit a particular situation depending upon a balancing of the competing interests. Unlike the right to confrontation, the question with standard of proof is not whether one exists, but where on the sliding scale it lies. The Court in Galloway held that a lesser standard of proof was adequate to insure fairness in sentencing. The right of confrontation and cross-examination, on the other hand, has no sliding scale. It is an all-or-nothing proposition. Since Galloway itself says that the defendant has a liberty interest in a correct application of the Guidelines, we should be wary of giving him nothing. The very fact that the burden of proof offers little protection to the defendant is one of the best reasons to allow him to cross-examine the prosecutor’s sources of information.
We have specific guidance from the Supreme Court on how to decide whether due process requires a given procedure. In Mathews v. Eldridge, 424 U.S. 319, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976), the Court held that three factors must be considered: the private interest at stake; the government’s interest, including the cost of the additional procedural requirement; and the extent to which more accurate fact-finding would result from the additional procedural requirement.4 A weighing of these factors points decisively in favor of the right of confrontation. The private interest — the interest in being free of unjust imprisonment for crime — is among the highest recognized by the law. In fact, it is the highest, I think, with two exceptions: the interest in not being unjustly convicted in the first place, and the interest in not being unjustly put to death. The government has an interest in preventing defendants from unfairly escap*412ing appropriate punishment, but this interest is not at all likely to be harmed by allowing defendants a right of confrontation. The government’s major interest is parallel to the defendant’s, if not identical: the interest in seeing that the sentence imposed is based on facts accurately found. It is by now a commonplace that the interest of the United States “in a criminal prosecution is not that it shall win a case, but that justice shall be done.” Berger v. United States, 295 U.S. 78, 88, 55 S.Ct. 629, 633, 79 L.Ed. 1314 (1935) (Sutherland, J.). The only real down side from the government’s viewpoint is that sentencing will cost it more time and money. In most cases, though — and this seems to be true here — the additional burden would be light. It could be satisfied by invoking firmly rooted hearsay exceptions or by calling witnesses who will often be either government employees {e.g., Secret Service agents) or other persons easily available {e.g., co-defendants on pretrial diversion). Add to all this the significant increase in accuracy that confrontation and cross-examination bring. In Dean Wigmore’s phrase, cross-examination is “the greatest legal engine ever invented for the discovery of truth.” 5 Wigmore, Evidence 32 (Chadbourn rev. ed. 1974).
The right of confrontation is worth the cost. It is, after all, not a “technicality” serving some extraneous purpose (a view that some take of the Fourth Amendment Exclusionary Rule, for example). It bears directly on and significantly advances the truth-seeking function of sentencing hearings. Confrontation is required by due process. See Note, supra, 105 Harv.L.Rev. at 1898.
IV.
Even if I agreed with the Court about the Constitution, I could not vote to affirm this sentence. The procedure followed was indefensible (indeed, the Court does not really try to defend it). And the evidence underlying the District Court’s finding was not reliable, if that word is to retain any substance.
The District Court began the sentencing hearing by announcing that it would not read the defendant’s objections to the pre-sentence report, even though the objections were timely. It then called as a witness an official of its own, the probation officer who wrote the presentence report, and told the defendant he could cross-examine. The Court warned, however, in a quite extraordinary statement, that
I’m going to assume that the factual statements [in the presentence report] are correct and accurate unless you are able to convince the Court that they are not.
Sent. Tr. 6.
This Court today reaffirms the holding of United States v. Streeter, 907 F.2d 781, 791-92 (8th Cir.1990), that a “presentence report is not evidence and is not a legally sufficient basis for making findings on contested issues of material fact.” Ante at 404. It also repeats the rule that facts relied on at sentencing must be proved by a preponderance of the evidence. Ante at 400. The District Court overtly violated these two rules. The burden was put on the defendant to disprove the probation officer’s boilerplate assertion that his own presentence report was correct! There is such a thing as harmless error, of course, but the Court's opinion does not say the error was harmless. It does not say that the evidence with respect to the sentencing factor in question was so overwhelming that misplacement of the burden of persuasion clearly made no difference. It says simply that there was “no functional difference between the procedure that was followed and the result that would have been reached” if correct procedures had been followed. Can this be so? Is it really true now in this Circuit that it doesn’t matter which side has the burden of proof at sentencing?
The Court also labels the evidence “reliable.” I respectfully disagree. In the first place, there is nothing reliable about a system in which a court purports to judge the credibility of one of its own employees. I concede that the record shows no reason to suppose the probation officer would lie. *413But what about unintentional mistakes, faulty memory, inaccurate repetition of what others had told him? The Court says the source of the probation officer’s information, “the Secret Service and the personnel in the prosecutor’s office — gives the information an aura of authenticity....” Ante at 404. The Court has a touching faith in governmental authority which I do not share. Have members of a police agency or of a prosecution force ever exaggerated the facts? Have they ever, perhaps, remembered only those portions of a conversation helpful to their position? Have they ever gone so far as to lie? Of course they have. They are not immune from human frailty or free of selfish interest. No doubt outright lying is rare. I have never known the Department of Justice to misrepresent facts in court. But if the mere fact that information comes from the government gives it “an aura of authenticity that renders it sufficiently reliable as the basis of a finding of fact,” ante at 404, why bother to have any sentencing hearings? Let’s just hear the government’s statement of facts and dispense with formalities like testimony and burden of proof.
I would reverse.
. The Court “assume[s],’’ ante at 403, "that any self-incriminatory statements made by [these two persons] ... fell within the definition of a statement against penal interest excepted from the exclusions of the hearsay rule by Federal Rule of Evidence 804(b)(3) (we put aside that Rule’s precondition of the witness’s unavailability).” To the extent that the two unnamed persons incriminated themselves, this may be true. But the important point here is not these persons’ own guilt, but whether another person, Wise, was also involved in the same offense, and, more important still, whether that other person was the leader or organizer of the offense. One may assume that people will not untruthfully incriminate themselves. But persons on pretrial diversion have a strong incentive to incriminate others. If they tell the authorities what they want to hear about Wise, they may hope for lenient treatment for themselves, perhaps even complete dismissal of charges. Maybe this evidence is "more than unsubstantiated rumor,” ante at 403-04, but how anyone can characterize it as "more than ... second-hand knowledge,” ibid., is beyond me. The best that can be said is that it is "second-hand knowledge.”
. See Middendorf v. Henry, 425 U.S. 25, 38, 96 S.Ct. 1281, 1289, 47 L.Ed.2d 556 (1976) (summary courts martial are not "criminal prosecutions" under the Sixth Amendment because they are not adversarial); Morrissey v. Brewer, 408 U.S. 471, 480, 92 S.Ct. 2593, 2599-2600, 33 L.Ed.2d 484 (1972) (parole-revocation hearings are not within the Sixth Amendment because "[p]arole arises after the end of the criminal prosecution, including imposition of sentence" (emphasis added)); Mempa v. Rhay, 389 U.S. 128, 134, 88 S.Ct. 254, 256-57, 19 L.Ed.2d 336 (1967) (presence of counsel required at sentencing because sentencing is a “stage of a criminal proceeding where substantial rights of a criminal accused may be affected”); United States v. Garbett, 867 F.2d 1132, 1136 (8th Cir.1989) (discussing application of Fed.R.Crim.P. 32 “at the sentencing phase of a trial" (emphasis added)).
. See Daniel J. Freed, Federal Sentencing in the Wake of Guidelines: Unacceptable Limits on the Discretion of Sentencers, 101 Yale LJ. 1681, 1715 (1992) (arguing that Ch. 5, Pt. H is inconsistent with 18 U.S.C. § 3661, providing that "[n]o limitation shall be placed on the information” to be considered by the sentencing court).
. In Medina v. California, — U.S. -, 112 S.Ct. 2572, 120 L.Ed.2d 353 (1992), the Court held that the Mathews analysis does not apply to questions of criminal procedure in the state courts. The Court’s reasoning draws heavily on considerations of federalism, id. — U.S. at -, 112 S.Ct. at 2577, and it seems that Mathews is still generally applicable under the Due Process Clause of the Fifth Amendment.