concurring:
The majority holds that, assuming without deciding that the search of Crawford’s home and his detention there were illegal, the statements made by Crawford were nevertheless admissible. I agree.
Judge Trott would hold that the search of Crawford’s home and his detention there were constitutionally permissible, so his statements were admissible without regard to the analysis in the majority opinion. I agree.
In my view, Judge Trott’s approach is preferable, because it usefully clarifies the constitutional relationship of states and parolees, and because the Supreme Court in New York v. Harris said, “[Ajttenuation analysis is only appropriate where, as a threshold matter, courts determine that ‘the challenged evidence is in some sense the product of illegal governmental activity.’ ” 1 There was no illegal governmental activity here, and that is the end of it. Nevertheless, it is highly desirable that we issue a majority opinion, not merely a plurality opinion, so I have concurred in the majority opinion as well as Judge Trott’s regarding the suppression of Crawford’s statements.
Regarding sentencing, I concur in the result reached by the majority, vacating and remanding, but disagree with the de novo standard of review the majority applies. Standard of review has not been put at issue and my view has not been argued by any party (so we as an en banc court would not do as I suggest here without inviting further briefing), and has therefore not been a focus of our attention. But the standard of review for a sentence is a threshold question, we must view the case through the lens of the proper standard of review, and recent legislation makes the de novo standard untenable. We err by declaring that the standard of review is de novo.
I write separately for two reasons: (1) to clarify the distinction between parolees and probationers, and (2) to address the standard of review for application of the sentencing guidelines to the facts of the offense.
I.
First, parolees. The cases often speak to persons “on probation or parole.” That set includes two quite different subsets. One is entitled to less freedom than the other. Crawford himself summarized the law for parolees with near accuracy when he testified that “I just took it for granted that, you know, I’m on parole, that I don’t have no rights at all.” The difference between parolees and probationers, who *1077are near opposite ends of the punishment scale, illustrates both the correctness of Judge Trott’s analysis and the possibility of distinguishing cases relating only to probationers.
Parolees are persons who have been sentenced to prison for felonies and released before the end of their prison terms. They are held in a “variation on imprisonment” with only a “limited degree of freedom.”2 One cannot get sentenced to parole. The only way to get parole in the typical state system3 is to commit a felony, not merely a misdemeanor, and get sentenced to prison, not merely jail, for a period of time long enough to qualify for release, eventually, on parole.4 Thus, as distinguished from those not convicted of anything, those convicted of mere misdemeanors and either jailed or not jailed, and those convicted of felonies but not imprisoned for lengthy periods, parolees are persons deemed to have acted more harmfully than anyone except those felons not released on parole.
Probationers are close to the other end of the harmfulness scale. The most typical use of probation is as an alternative to jail for minor offenders, most commonly misdemeanants.5 Sometimes a first offender felon gets the lenience of probation rather than imprisonment.6 Unlike parolees, who were sent to prison for substantial terms, probationers attain that status from a judicial determination that their conduct and records do not suggest so much harmfulness or danger that substantial imprisonment is justified.
It is perfectly reasonable, and constitutionally permissible, that persons whose conduct has been so egregious as parolees’ are subject to searches on hunches, unreliable tips, general sweeps, and anything else not “arbitrary, capricious, or harassing.” 7 Thus even if the search of Crawford’s home would be violative of a probationer’s right to privacy (I do not suggest that it would be), that would not settle the question whether it would violate the right to privacy, limited as it is, of a parolee. Constitutional limits on supervision of probationers may be more extensive than those limiting supervision of parolees.
II.
Second, sentencing. During the Ulrich Street robbery, Crawford pointed a gun at a security guard, momentarily pressing it into his back, and said, “This is a holdup. Face the wall, put your hands up, and don’t move.” The district court held that this exchange sufficed for imposition of the “physical restraint” enhancement. Applying de novo review, the majority vacates and remands on the grounds that the dis*1078trict court erred by not applying United States v. Parker.8 Our court has historically held that the application of the physical restraint adjustment to an undisputed set of facts is reviewed de novo.9 We should correct our standard of review to “abuse of discretion.” I nevertheless concur in the result reached by the majority regarding sentencing.
A de novo standard was widely adopted prior to the recent revision of the relevant statute.10 The theory for the de novo standard is that in reviewing application of guidelines to the facts, we are making a determination of law.11 Our determinations of law, though, are so fact-based and fact-limited that they should more properly be viewed as judgments about facts. We have generated a vast number of decisions about how the guidelines apply to details of crimes so minute and so affected by context that the cases invite distinctions as often as application.
The more deferential abuse of discretion standard would prevent the proliferation of not-very-useful appellate decisions which district judges may overlook, and afford appropriate deference to the sentencing judgments of district judges. The more deferential standard would also establish needed play in the joints of the sentencing guidelines, so that district judges can smooth out the variations generated by different prosecutors and probation officers in similar cases. Most important, though the old de novo standard was a permissible construction (though I think it was erroneous) of the “due deference” language in the statute before the recent amendment, it cannot be squared with the statute as it now stands.
Congress has provided the appellate court with statutory direction regarding the standard of review for different sentencing decisions by a district court. In 18 U.S.C. § 3742(e), we are instructed by Congress to “give due deference to the district court’s application of the guidelines to the facts.” With the addition of the PROTECT Act amendments last year,12 *1079the statute expressly sorts out three different standards of review, (1) “clearly erroneous” for credibility determinations; (2) “due deference” for application of the guidelines to the facts; (3) “de novo” for application of guidelines to facts for “3(A) or 3(B)” determinations, i.e., departures for improper reasons and departures without written statements of reasons:
The court of appeals shall give due regard to the opportunity of the district court to judge the credibility of the witnesses, and shall accept the findings of fact of the district court unless they are clearly erroneous and, except with respect to determinations under subsection (3)(A) or (3)(B), shall give due deference to the district court’s application of the guidelines to the facts. With respect to determinations under subsection (3)(A) or (3)(B), the court of appeals shall review de novo the district court’s application of the guidelines to the facts.13
We are erroneously applying “de novo” review where the statute expressly commands us to apply “due deference” review.
There is no way to infer that Congress meant for us to apply de novo review to all “application of the guidelines to the facts,” where it expressly distinguished between “due deference” and “de novo” review and explicitly limited de novo review to the designated classes of departures from the guidelines. Where we once may have been able to say that the “deference due” other applications of guidelines to the facts was “none” — and then review de novo— that interpretation is now precluded for applications of the facts to the guidelines other than for (3) (A) or (3)(B) determinations. Congress has said, in effect, review decisions in class A for clear error, class B with due deference, and class C de novo, yet we review class B as well as class C decisions de novo. That is plainly a misreading of the statute.
Though “due deference” is a different verbal formula from “abuse of discretion,” it is best interpreted as meaning the same thing in a case such as Crawford’s. In Koon v. United, States, the Court directed appellate courts to review a district court’s decision to depart from the Guidelines range for abuse of discretion rather than de novo.14 The Court acknowledged that “the deference that is due [a district court under § 3742(e) ] depends on the nature of the question presented.”15 Where the appellate court “will be in as good a position to consider the question as the district court was in the first instance,” as when the claim on appeal is that the district court made some kind of mathematical error, “[t]he district court may be owed no deference.”16
In distinguishing the departure decision from questions of law that require no deference, the Court in Koon relied on the factual nature of departure decisions, decisions that require a district court to “make a refined assessment of the many facts bearing on the outcome” and that involve “the consideration of unique factors that are little susceptible of useful generalization.” 17 The Court reasoned that we cannot avoid the factual nature of the departure inquiry by taking it to a higher level of generality.18 Rather than asking whether a factor is within the heartland as a general proposition, a court asks wheth*1080er the factor is within the heartland given all the facts of the case.19
More recently, the Supreme Court rejected a defendant’s argument that because no facts were in dispute in the decision to be reviewed by the appellate court, de novo review was appropriate.20 In Buford v. United States, the Court upheld the Seventh Circuit’s deferential review of the district court’s decision that certain past convictions were not consolidated for sentencing when it determined that Buford was a career offender.21 Relying heavily on Koon, the Court held that “the district court is in a better position than the appellate court to decide whether a particular set of individual circumstances demonstrates ‘functional consolidation.’ ” 22 Additionally, the Court pointed out that “factual nuance may closely guide the legal decision, with legal results depending heavily upon an understanding of the significance of case-specific details.”23
Whether the “physical restraint” enhancement, § 2B3.1(b) (4)(B), applies in a given case is analogous to the kind of question for which the Supreme Court has required deference to the judgment of the district court. A comprehensive, nuanced, contextualized consideration of the crime, the criminal, and the victim, and all the surrounding circumstances, may affect the decision whether the victim was physically restrained. A raised fist might very effectively restrain an elderly, frail victim but might not at all restrain a victim larger and stronger than the perpetrator, and the effectiveness of restraint might vary with time of day, visibility of the crime to passersby, proximity of refuge, and all sorts of other things. Thus, application of the “physical restraint” guideline to the facts is not a question “readily resolved by reference to general legal principles and standards alone.”24 Rather, it is a question “that grows out of, and is bounded by, case-specific detailed factual circumstances ... [which] limits the value of appellate court precedent.”25 We are not in as good a position as the district court to evaluate the facts in light of the record, so we should largely get out of the business of doing so, except when the district court determination, even when viewed deferentially, cannot be deferred to.
Our error in applying a de novo standard of review generates a large number of cases for which our review makes no very valuable contribution. According to the United States Sentencing Commission, in fiscal year 2001, there were 14,000 Guidelines offenders sentenced in federal courts in the Ninth Circuit.26 Of that number, 935 appealed either their sentences or their sentences and convictions.27 This is almost as many cases as the number of immigration cases we handled in 2001.28 An abuse of discretion standard of *1081review would generate far more unpublished dispositions and far fewer published opinions than a de novo standard.
The district court found that Crawford pointed a gun directly at a bank employee, told him this was a robbery, made him turn around to face the wall, and forced him to remain there during the entire robbery. The majority holds that the district court correctly held that physical contact with the victim was unnecessary, under United States v. Thompson,29 but that the district court could not impose the “physical restraint” adjustment without expressly finding, under our decision in United States v. Parker30 a “sustained focus on the restrained person.” Both cases are, in my view, unnecessary elaborations of the guidelines to which we were led by a mistaken standard of review. Were the district court simply to apply the guidelines as they are without the cases, guidance would be quite adequate, and would probably suggest not imposing the adjustment.
Under sentencing guideline § 2B3.1(b)(2), using or brandishing a firearm during a robbery gets a five or six level increase. Abduction and physical restraint are dealt with separately by four and two level adjustments in § 2B3.1(b)(4). The commentary following the guideline refers to physical restraint “by being tied, bound, or locked up.” We eliminated the simplicity and clarity of this arrangement by generating needless common law around it. First, we construed the physical restraint enhancement to include pointing a gun at someone in Thompson, even though pointing a gun at someone gets the five or six level adjustment under subsection (b)(2) rather than the mere two level adjustment for physical restraint under subsection (b)(4).31 Then we distinguished and limited Thompson in Parker by holding that just pointing a gun at someone does not amount to physical restraint without a “sustained focus” on the victim.32 We could have stayed out of this thicket, and left the district courts with the greater clarity of the guidelines unadorned, had we reviewed merely for abuse of discretion. This pair of cases shows how de novo review encrusts the guidelines with needless additional complexity. We did not generate a common law of sentencing before the guidelines, because sentences were generally unappealable. The guidelines are so elaborate a codification that encrustation by an extensive body of common law is unhelpful to sentencing judges and does not contribute to just sentences any more than abuse of discretion' review would.
The distinctions upon which the cases have focused have salience in particular cases, but as legal generalizations they are arbitrary. Whether a gun touches a person or not is simply a silly question, since the whole idea of guns is that they can function without touching — that ’ is why they have replaced swords. As for whether the felon spends a whole lot of time focusing on one of the individuals in the bank, that is but one of many contextual considerations that affect how restrained victims feel. A more appropriate focus of inquiry would be upon whether restraint by a gun is covered by the “using” and “brandishing” adjustments, and whether it is like the tying up, binding, or locking up exam-*1082pies in the application note for “physical restraint.” But that sensible line of inquiry is foreclosed by our needless common law.
I concur in our result vacating and remanding for resentencing, though I differ as to the proper standard of review. Because we as an en banc panel can easily correct mistaken language in prior law, we should consider the effect of the PROTECT Act and revise our standard of review. Doubtless in some subsequent case the issue of standard of review will be squarely set before us by the briefs.
. New York v. Harris, 495 U.S. 14, 19, 110 S.Ct. 1640, 109 L.Ed.2d 13 (1990).
. Pa. Bd. of Parole v. Scott, 524 U.S. 357, 365, 118 S.Ct. 2014, 141 L.Ed.2d 344 (1998).
. Parole was abolished in the federal system by the Sentencing Reform Act of 1984, Pub.L. No. 98-473, 98 Stat.1987, tit. II, ch. II, § 212(a)(2).
. See, e.g., Cal.Penal Code §§ 3000, 3040 et seq. (imposing mandatory parole for persons imprisoned for more than one year).
. See, e.g., U.S.S.G. § 5B1.1 (Imposition of a Term of Probation).
. In the federal system, 18 U.S.C. § 3561(a)(1) prohibits a sentence of probation for individuals who commit Class A or B felonies. But a person who commits a lesser felony may be sentenced to probation if his offense level and criminal history level otherwise place him in Zone A of the Sentencing Table (where the applicable range is 0-6 months), or Zone B and the court also imposes a condition or combination of conditions requiring intermittent confinement, community confinement, or home detention. U.S.S.G. § 5Bl.l(a).
.People v. Reyes, 19 Cal.4th 743, 80 Cal. Rptr.2d 734, 968 P.2d 445, 450 (1998).
. United States v. Parker, 241 F.3d 1114, 1119 (9th Cir.2001).
. See Thompson, 109 F.3d at 640("We review de novo the court’s interpretation of the Guidelines, and its factual determinations for clear error.”).
. See Parker, 241 F.3d at 1118-19(failing to articulate a standard of review but according no apparent deference to the district court’s decision to apply the enhancement for physical restraint); Thompson, 109 F.3d at 640; see also United States v. Plenty, 335 F.3d 732, 734 (8th Cir.2003) ("The district court’s imposition of the enhancements is based on factual findings subject to review for clear error. The district court’s interpretation of the United States Sentencing Guidelines and application of those Guidelines to the facts of the case are reviewed de novo.” (internal citations omitted)); United States v. Drew, 200 F.3d 871, 880 (D.C.Cir.2000) ("Because the facts on this issue are not significantly in dispute, the issue is primarily a question of law and therefore review closer to de novo is required.”); United States v. Perkins, 89 F.3d 303, 307 (6th Cir.1996) ("Since Perkins challenges only whether the district court properly applied the Sentencing Guidelines and does not challenge the district court’s findings of fact, the issues on appeal are questions of law, which we review de novo.”); United States v. Jones, 32 F.3d 1512, 1517-18 (11th Cir.1994) ("Whether a particular Guideline applies to a given set of facts is a legal question subject to de novo review.”); United States v. Stokley, 881 F.2d 114, 115-16 (4th Cir.1989) (performing de novo review where the defendant did “not really attack the factual undergirding of his sentence but rather contended] that his behavior did not fall within the legal definition of the term 'physically restrained’ ”).
. United States v. Anglin, 169 F.3d 154, 164 (2d Cir.1999).
. Pub.L. No. 108-21, 117 Stat. 650 (Apr. 30, 2003). See United States v. Phillips, 356 F.3d 1086, 1098-1100 (9th Cir.2004), as amended by 367 F.3d 846, 2004 Daily Journal D.A.R. 5444 (9th Cir. May 06, 2004) (holding that the PROTECT Act amendments to the standard of review apply retroactively).
. 18 U.S.C. § 3742(e).
. Koon v. United States, 518 U.S. 81, 99-100, 116 S.Ct. 2035, 135 L.Ed.2d 392 (1996).
. Id. at 98, 116 S.Ct. 2035.
. Id.
. Id. at 98, 99, 116 S.Ct. 2035 (internal quotations omitted).
. Id. at 99, 116 S.Ct. 2035.
. Id.
. Buford v. United States, 532 U.S. 59, 63-64, 121 S.Ct. 1276, 149 L.Ed.2d 197 (2001).
. Id. at 66, 121 S.Ct. 1276.
. Id. at 64, 121 S.Ct. 1276.
. Id. at 65, 121 S.Ct. 1276.
. Id.
. Id. at 65-66, 121 S.Ct. 1276.
. "Table 2: Guidelines Offenders in Each Circuit and District, Fiscal Year 2001,” U.S. Sentencing Commission, Sourcebook of Federal Sentencing Statistics, available at http:// www.ussc.gov/ANNRPT/2001/ table2.pdf (last visited March 31, 2004).
. "Table 55: Types of Appeals in Each Circuit and District, Fiscal Year 2001” U.S. Sentencing Commission, Sourcebook of Federal Sentencing Statistics, available at http:// www.ussc.gov/ANNRPT/2001/table55.pdf (last visited March 31, 2004).
. In the 12-month period ending September 30, 2001, the Ninth Circuit commenced 1,150 "administrative appeals.” See "Table B-l U.S. Courts of Appeals-Appeals Commenced, Terminated, and Pending, by Circuit, During *1081the 12-Month Period Ending September 30, 2001,” Administrative Office of the United States Courts, Judicial Business of the United States Courts: 2001 Annual Report of the Director, available at http://jnet.ao.dcn/img/as-sets/4825/b01sep01.pdf (last visited March 31, 2004).
. Thompson, 109 F.3d at 641.
. Parker, 241 F.3d at 1118.
. Thompson, 109 F.3d at 641.
.Parker, 241 F.3d at 1118.