944 F.3d 352

1st Cir.

2019

UNITED STATES OF AMERICA, Appellee, v. STEPHEN MANTHA, Defendant, Appellant.

No. 18-1951

United States Court of Appeals For the First Circuit

December 10, 2019

Before Thompson, Boudin, and Kayatta, Circuit Judges.

APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF MASSACHUSETTS [Hon. Timothy S. Hillman, U.S. District Judge]

Elizabeth A. Billowitz for appellant.

Alexia R. De Vincentis, Assistant United States Attorney, with whom Andrew E. Lelling, United States Attorney, was on brief, for appellee.

KAYATTA, Circuit Judge. This appeal concerns the interaction between the Sentencing Guidelines’ grouping rules, the one-book and multiple-offense rules, and the U.S. Constitution‘s Ex Post Facto Clause. In calculating the offense level for an offense committed in 2001, the district court used the 2016 Guidelines Sеntencing Manual applicable to two ungrouped, later-committed offenses to which the defendant also pleaded guilty under the same indictment. The 2016 version of the manual, as compared to the version in effect in 2001, resulted in a higher Total Offense Level (TOL). In a case of first impression in this circuit, we find that application of the subsequent manuаl to the prior, ungrouped offense violated the Ex Post Facto Clause. We further find that the district court plainly erred in providing no justification for the resulting upward variance.

I.

In approximately 2001, Stephen Mantha molested a child who was then between six and eight years old. Mantha also recorded the molestation on a VHS tape. Fifteen years later, between late 2015 and early 2016, Mantha‘s employer, the U.S. Postal Service, caught him searching for and viewing child pornography on his workplace computer. A subsequent search of his home turned up the recording of the 2001 molestation and electronic storage devices containing additional child pornography. Mantha eventually entered a straight guilty plea to three offenses: (1) sexual exploitation of a child in violation of 18 U.S.C. § 2251(a) resulting from the 2001 incident; (2) access with intent to view child pornography in violation of 18 U.S.C. § 2252A(a)(5)(B), (b)(2) resulting from the 2015–2016 internet searches; and (3) possession of child pornography also in violation of 18 U.S.C. § 2252A(a)(5)(B), (b)(2) resulting from the 2016 possession of electronic storage devices.

The Presentence Investigation Reрort (PSR) prepared by the Probation Officer grouped the second (2015–2016 internet searching) and the third (2016 possession) offenses, but not the first (2001 exploitation) because the 2001 offense was insufficiently related to the more recent two offenses. See U.S.S.G. § 3D1.2 (identifying when to group closely related offenses); see also id. § 2G2.1 (providing the offense level for sexual exploitation of a minor by production of sexually explicit visual material); id. § 3D1.2(d) (excluding offenses covered by § 2G2.1 from grouping on the basis of ongoing behavior or aggregate harm). All parties agree that this grouping -- and ungrouping -- was correct. The PSR nevertheless employed the 2016 version of the Guidelines manual to calculate the applicable offense levels fоr both the two grouped offenses and for the ungrouped 2001 offense. Under the 2016 manual, by virtue of a 2004 amendment, see id. app. C, amend. 664 (effective Nov. 1, 2004), that 2001 offense generated an adjusted offense level (AOL) of 40.¹ That offense level, nine levels higher than the AOL for the two more recent grouped offenses,² drove the calculation of the TOL to 37,³ and resulted in a Guidelines Sentencing Range (GSR) of 210 to 240 months. Under the pre-amendment version of the manual in effect at the time of the 2001 offense, the AOL for that offense would have been 33, which would have resulted in a lower GSR of 121 to 151 months.

Both Mantha and the government objected to the use of the 2016 manual as applied to the 2001 exploitation offense, agreeing that, in the words of the government, “it would be a violation оf the [Ex Post Facto] clause to apply the present version of the guidelines to conduct that occurred in 2001.” The district ‍‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‍court apparently viewed the matter otherwise, stating only that “I‘ve spent a good part of the morning talking with counsel for the probation office, [and] I am going to keep the offense level and category the sаme.” The court sentenced Mantha to 196 months, a downward variance from the PSR‘s GSR but an upward variance from the calculation for which the parties advocated. When the government asked whether the sentence would have been the same under the lower GSR, the court said, “I thought about that, and I believe that would have been the sentence that I was going to impose under either scenario.” The court gave no explanation for why it chose the 196-month sentence, or for why it would have done so even if it knew it to be upwardly variant.

Mantha timely appealed. We review de novo a preserved claim that application of a particular version of the Guidelines violated the Ex Post Facto Clause. United States v. Goergen, 683 F.3d 1, 3 (1st Cir. 2012).

II.

A.

We look first to see if the Guidelines themselves support the approach taken by the district court, apart from any limitations imposed by the Ex Post Facto Clause. The Guidelines adopt what we call the “one-book rule“: “The Guidelines Manual in effect on a particular date shall be applied in its entirety.” U.S.S.G. § 1B1.11(b)(2).⁴ The Guidelines also set forth what we

call the “multiple-offense rule“: “If the defendant is convicted of two offenses, the first committed before, and the second after, a revised edition of the Guidelines Manual became effective, the revised edition of the Guidelines Manual is to be applied to both offenses.” Id. § 1B1.11(b)(3). The commentary to the multiple-offense rule states that “the apрroach set forth in [§ 1B1.11(b)(3)] should be followed regardless of whether the offenses of conviction are the type in which the conduct is grouped under § 3D1.2(d).” Id. § 1B1.11 cmt. background. Viewed in the aggregate, these rules call for the approach taken by the district court, subject to one caveat: The Guidelines also warn that (of course) the manual in effect at the time оf sentencing should not be used if doing so “would violate the ex post facto clause.” Id. § 1B1.11(b)(1).

The Constitution states that “[n]o . . . ex post facto Law shall be passed.” U.S. Const. art I, § 9, cl. 3. The Supreme Court has held that application of a version of the Sentencing Guidelines adopted after an offense is committed violates the Ex Post Facto Clause where the newеr Guidelines result in a higher GSR than the version in effect at the time the offense was committed, even under the post-Booker advisory Guidelines regime. Peugh v. United States, 569 U.S. 530, 549–50 (2013); see United States v. Booker, 543 U.S. 220, 245 (2005).

In a case decided shortly after Peugh, our circuit nevertheless held that application of the one-book and multiple-offense rules to a series of grouped offenses does not violate the Ex Post Facto Clause even if the eаrlier grouped offense occurred before an amendment to the manual increasing the offense level for that offense. United States v. Pagán-Ferrer, 736 F.3d 573, 598 (1st Cir. 2013). In that case, the defendant was convicted of civil-rights violations and obstruction of justice for lying to investigators in 2008 (post-amendment) about a 2003 assault (pre-amendment). Id. 595–96. We reasoned as follows:

The Sentencing Guidelines’ one book and grouping rules placed [the defendant] on notice that if he committed a closely related offense in the future, his sentence for both offenses would be calculated pursuant to the Guidelines in effect at the time of that later, related offense conduct. . . . Accordingly, the change in [the defendant‘s] offense level is proрerly viewed not as a consequence of an ex post facto violation, but as the direct result of his decision to engage in closely related offense conduct [after the amendment].

Id. at 599. Every circuit except the Ninth agrees with this holding. See id. at 598 (collecting cases); United States v. Siddons, 660 F.3d 699, 706–07 (3d Cir. 2011); see also United States v. Wijegoonaratna, 922 F.3d 983, 992–93 (9th Cir. 2019).

Mantha‘s case differs from Pagán-Ferrer in a crucial way: his offenses were not grouped. The Guidelines commentary ‍‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‍states that this distinction is irrelevant, reasoning that “[t]he

ex post facto clause does not distinguish between groupable and nongroupable offenses.” U.S.S.G. § 1B1.11 cmt. background. But in Pagán-Ferrer we adopted the fiction that prior notice sufficient to satisfy the Ex Post Facto Clause could be presumed in the case of groupable offenses because such offenses, by definition, are “closely related.” See Pagán-Ferrer, 736 F.3d at 599; U.S.S.G. § 3D1.2. So to concede that the offenses are not groupable is to concede that they are not “closely related,” thereby attenuating the connection that served in Pagán-Ferrer to justify what would have otherwise seemed to be a clear ex post facto violation.

Arguably, the offenses here, while not closely related (they were, after all, fifteen years apart and did not involve the same victim), could be seen as related in some sense; one involves abuse of a child and the other possession of visual depictions of another person‘s abuse of a child. But as we move down the spectrum from “closely rеlated” to “related,” the fiction of notice in the case of groupable offenses, however plausible, approaches utter fantasy, and would seem to have no stopping point. After all, almost all crimes committed by the same person are related in some significant sense.

The Guidelines’ commentary justifies the applicаtion of a later and stiffer Guidelines manual to an earlier ungrouped offense by observing that misconduct predating a manual change may be considered in sentencing for a post-amendment offense if the

earlier misconduct qualifies as “relevant conduct” in its relationship to the more recent offense. See U.S.S.G. § 1B1.11 cmt. background (citing the consideratiоn of pre-amendment uncharged embezzlement in sentencing for related post-amendment embezzlement). But conduct is treated as “relevant conduct” precisely because it is closely related to the primary offense. See id. § 1B1.3 (defining relevant conduct); see also id. § 3D1.2(c). So the analogy drawn by the commentary fits poorly the case at hand -- the treatment of prior, largely unrelated conduct. Moreover, the commentary, drafted before Peugh, seems to overlook the difference between considering prior conduct in sentencing for a later offense and (as in this case) actually sentencing for the prior conduct as an offense.

In any event, even if we were to accept the notion that a stiffеr, amended manual could be applied to an earlier offense based on the fact that the underlying conduct could have alternatively qualified as relevant conduct to a later offense, that notion would find no application in this case. Here, the government does not argue that Mantha‘s 2001 conduct could have been considered as relevant conduct in sentencing for the 2015–2016 possession and access offenses. Nor do we see why it would have been. So the Guidelines commentary does not solve the constitutional problem at issue here.

In finding the tenuous relatedness between Mantha‘s earlier and later crimes to be material, we also have in mind the fact thаt the Ex Post Facto Clause advances purposes beyond that of providing notice. It serves the cause of “fundamental fairness” by “having the government abide by the rules of law it establishes to govern the circumstances under which it can deprive a person of his or her liberty or life.” Peugh, 569 U.S. at 544 (quoting Carmell v. Texas, 529 U.S. 513, 533 (2000)). Simply telling a person that those rules may change should not suffice to circumvent the ex post facto bar. Otherwise, that bar could be effectively eliminated altogether by the enactment of a broad, catch-all caveat.

Three of the four circuits that have addressed the question now before us have held that application of the one-book and multiple-offense rules to ungrouped offenses constitutes an ex post facto violation. See United States v. McMillian, 777 F.3d 444, 449 (7th Cir. 2015); United States v. Saferstein, 673 F.3d 237, 244 (3d Cir. 2012); United States v. Lacefield, 146 ‍‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‍F. App‘x 15, 22 (6th Cir. 2005). The only circuit court opinion to the contrary predates Peugh and relied almost exclusively on the Guidelines commentary in reaching that result. See United States v. Butler, 429 F.3d 140, 153–54 (5th Cir. 2005). Our focus on groupability in Pagán-Ferrer invites the approach adopted by the majority of these other circuits, and we now join them expressly.

Our holding today is a narrow one. We leave undisturbed the rule laid out in Pagán-Ferrer. Furthermore, we do not say that application of the multiple-offense rule to ungrouped conduct will pose a problem in all circumstances. Different permutations may necessitate a different analysis. We hold only that, under the present circumstances, where the TOL is raised by aрplication of a Guidelines amendment to a pre-amendment offense based solely on the existence of post-amendment offenses that are not closely related to the earlier offense, use of the post-amendment Guidelines is unconstitutional.

B.

We now turn to the government‘s argument for why Mantha‘s sentence should be upheld notwithstanding thе ex post facto violation. The government relies on the district court‘s statement that “I believe [196 months] would have been the sentence that I was going to impose under either scenario.”⁵ The government

argues, citing Peugh, 569 U.S. at 550 n.8, that the ex post facto error was therefore harmless beyond a reasonable doubt. Cf. United States v. Acevedo-Hernández, 898 F.3d 150, 172 (1st Cir. 2018); United States v. Tavares, 705 F.3d 4, 25 (1st Cir. 2013).

In making this harmless-error argument, the government presumes that the district court‘s sentence stood independently on each of two legs: a nonvariant (or downwardly variant) application of the new Guidelines manual, and an upward variance from the GSR calculated under the older manual. Thus, reasons the government, it makes no difference that the former leg was infirm.

But this reasoning only works if the second leg itself suffers from no disqualifying infirmity. And it clearly does suffer from such an infirmity because the record contains no statement of reasons for the upward variance. When sentencing, a court must “state in open court the reasons for its imposition of the particular sentence,” and for sentences outside the GSR, the stated reasons must be “specific.” 18 U.S.C. § 3553(c)(2); see also Gall v. United States, 552 U.S. 38, 46 (2007). “[T]he greater a deviation from the GSR, the more compelling the sentencing court‘s justification must be.” United States v. Del Valle-Rodríguez, 761 F.3d 171, 177 (1st Cir. 2014). Here, the district court gave no explanation whatsoever for why it would impose a 196-month sentence even if the upper end of the GSR were forty-five months less than that. The government contends that the

reasons for the sentence can be “inferred from the record,” United States v. Rivera-Gonzalez, 809 F.3d 706, 712 (1st Cir. 2016), because the government provided defendant-specific reasons for a harsher sentence at the sentencing hearing. The record lacks any indication that the district court was adopting those reasons, and in the context of a substantially upwardly variant sentence we decline to impute the governmеnt‘s justification to the court and thereby relieve the court of its statutory obligation.

The government questions whether Mantha preserved in the district court his objection to the adequacy of the court‘s explanation for the upward variance. But even assuming plain-error review applies, Mantha would prevail on this point. See id. at 711 (setting out the plain-error standard). Section 3553(c) is unambiguous and its requirement for justifying a variant sentence ‍‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‌‍is well known. Indeed, we have previously found similar omissions in upward-variance cases to constitute plain error. See United States v. Montero-Montero, 817 F.3d 35, 37 (1st Cir. 2016) (vacating a 60-month sentence where the GSR was 6–12 months and the court gave “no coherent explanatiоn” for the variance); Rivera-Gonzalez, 809 F.3d at 712 (vacating a 360-month sentence where the GSR was 60 months and the court gave “no explanation” for the variance). We have upheld, under plain-error review, “short and simple” explanations for small upward variances, United States v. González-Rodríguez, 859 F.3d 134, 136, 139 (1st Cir. 2017)

(affirming a 33-month sentence where the GSR was 24–30 months and the court explained that the defendant‘s “conduct [flouted] the law and . . . represent[ed] a risk to the community” (alterations in original)), but here there is no explanation at all. The possibility of prejudice and unfairness in this situation cannot be discounted. See Montero-Montero, 817 F.3d at 38; Rivera-Gonzalez, 809 F.3d at 712. We cannot rule out the possibility of a lower sentence until the district court explains its reasons for selecting what it will now know is an upwardly variant sentence.

None of this is to say that the sentence assigned by the district court is substantively precluded as excessive. For that reason, we reject Mantha‘s argument that a 196-month sentence would necessarily represent an abuse of discretion even if properly explained.

III.

For the foregoing reasons, we vacate the district court‘s sentence and remand for resentencing consistent with this opinion.

Notes

As relevant here, the amendment raisеd the base offense level (BOL) under § 2G2.1(a) from 27 to 32 and created a new two-level enhancement, § 2G2.1(b)(2)(A), for offenses involving sexual contact. Two other enhancements, for four and two levels, respectively, applied under both the pre-2004 and post-2004 manuals. See U.S.S.G. § 2G2.1(b)(1)(A), (5) (2016).

By operation of § 3D1.4 (instructing how to combine ungrouped offenses), the more recent offenses did not affect Mantha‘s TOL in any way other than by bringing the newer version of the mаnual into play.

Mantha received a three-level reduction for acceptance of responsibility under § 3E1.1(a), (b).

The one-book rule also states that, “if a court applies an earlier edition of the Guidelines Manual, the court shall consider subsequent amendments, to the extent that such amendments are clarifying rather than substantive changes.” Id.; see also . There is no indication here that the district court was using the 2016 manual only for this purpose.

In its written statement of reasons, the court stated that “196 months is sufficient, but not greater than necessary, to provide adequate punishment,” which merely tracks the language of 18 U.S.C. § 3553(a). The court then reiterated that it “would impose this sentence whether an earlier version of the Guideline Manual was applicable, as 196 months is the appropriate sentence based upon the crimes committed.” Although required by statute, see id. § 3553(c)(2); 28 U.S.C. § 994(w)(1), the written statement-of-reasons form “serves a largely administrative purpose” by “facilitat[ing] data collection,” . We therefore place no weight on the district court‘s written statement.

Read the detailed case summary