UNITED STATES OF AMERICA v. LUIS MIGUEL GARCÍA TORRO (2026)

Overview

Police busted García for having a machinegun while on supervised release for a similar weapons crime.¹ He eventually pled guilty under a plea deal with the government to illegally possessing that gun. See 18 U.S.C. § 922(o).² And as part of the agreement, he and the government promised to “seek a [prison] sentence within the applicable guideline[s] range” — a range that the judge set at 27 to 33 months (the deal also (i) barred García from arguing for a below-guidelines sentence and the government from pushing for an above-guidelines term and (ii) put the final sentencing decision in the judge's hands).

García consistently asked for 27 months. The government repeatedly asked for 33 months. But the judge ultimately picked 48 months.

A disappointed García now appeals, arguing that the government breached the plea agreement by giving “lip service” — and no more — “to its obligation to recommend a sentence between 27 and 33 months” (we'll explain the argument's ins and outs shortly). But we must affirm (for reasons we'll explain later too).³

Some Legal Basics

A few legal principles set the stage for this opinion.

1

Plea agreements are contracts. See, e.g., United States v. Fargas-Reyes, 125 F.4th 264, 270 (1st Cir. 2025); United States v. Rivera-Cruz, 878 F.3d 404, 408 (1st Cir. 2017). And because defendants land plea deals only by waiving many constitutional rights (like their rights to a jury trial, to confront witnesses, and to testify), we hold prosecutors “to the most meticulous standards of both promise and performance.” See Fargas-Reyes, 125 F.4th at 270 (emphasis omitted) (quoting United States v. Lessard, 35 F.4th 37, 42 (1st Cir. 2022)). Which is why the law bans not only the explicit reneging of promises but also the implicit “end-runs around them.” See United States v. Saxena, 229 F.3d 1, 6 (1st Cir. 2000) (quoting United States v. Voccola, 600 F. Supp. 1534, 1537 (D.R.I. 1985)); see also United States v. Mojica-Ramos, 103 F.4th 844, 850 (1st Cir. 2024).

Yet the law also restricts “what ․ defendant[s] reasonably may expect.” Saxena, 229 F.3d at 6. We'll give you some for-instances. Unless bound to do so under the agreement's specific terms, prosecutors needn't support a promised recommendation “enthusias[tically].” United States v. Meléndez-Rivera, 139 F.4th 83, 90 (1st Cir. 2025) (quoting United States v. Cortés-López, 101 F.4th 120, 128 (1st Cir. 2024)). And “if the agreement lets them request a sentence within a range that's ‘stiffer’ than the defense's request, they must tell the judge ‘why [their] higher sentence’ is better.” Fargas-Reyes, 125 F.4th at 270 (emphasis added) (quoting Lessard, 35 F.4th at 43) (brackets in original). In other words, they still must pass “relevant information” to the judge despite having “a corollary obligation to honor” plea-deal pledges. Saxena, 229 F.3d at 6; see also United States v. Frazier, 340 F.3d 5, 13 (1st Cir. 2003) (stating that “[t]he government is under a constant obligation to ensure the integrity of the proceedings at all levels, notwithstanding the terms of any plea agreement”). Plus they're “under no compulsion to sugar-coat the facts” when performing “these duties.” United States v. Colón-Rosario, 921 F.3d 306, 312 (1st Cir. 2019). See generally Lessard, 35 F.4th at 42 (adding that we consider all the circumstances to see if prosecutors acted impermissibly).

2

García candidly (and correctly) concedes that because he didn't raise the plea-breach issue with the judge, he must show plain error to win. That's easier said than done because he must prove not only an error, but an error that's obvious (meaning it's undeniable under governing law), affects his substantial rights, and seriously implicates the fairness of the sentencing process. See, e.g., United States v. Galíndez, 999 F.3d 60, 64-65 (1st Cir. 2021). See generally Puckett v. United States, 556 U.S. 129, 135 (2009) (holding that a claim of error “subject to reasonable dispute” doesn't satisfy the obviousness requirement); United States v. Delgado-Sánchez, 849 F.3d 1, 13 (1st Cir. 2017) (expressing that we gauge an error's obviousness from the perspective of the time of appellate consideration, not from the perspective of the time of sentencing).

Arguments and Analysis

García spends a lot of time explaining how he's (supposedly) cleared plain error's error and obviousness hurdles. The gist of his theory is that the government (allegedly) talked out of both sides of its mouth, requesting a 33-month within-guidelines term (as permitted by the agreement) while using “wink-and-nod advocacy” to get “an upward variance.”⁴ More specifically, he faults the government for “cit[ing] case law” in its sentencing memo that “support[s] above-guideline[s]” prison stints for “firearm” crimes, “fram[ing] [his] non-violent offense against the backdrop of an unusually murderous Puerto Rico society,” painting him “— a non-violent offender — as a potential threat to that community,” and “highlight[ing]” that his “within-guideline[s] sentence” for the prior gun crime didn't “sufficiently ‘deter’ his [illegal] possessory acts in this case.”⁵ That “hollow performance,” the argument goes on, communicated “that a harsher sentence wasn't just an option — it was necessary.”

Though passionately made, García's theories don't persuade — as anyone who reads on will see (our reasoning below largely tracks the government's, FYI).

1

“Context” always “matters,” of course. See Lessard, 35 F.4th at 43 (citing Saxena, 229 F.3d at 7-8, and United States v. Canada, 960 F.2d 263, 269-70 (1st Cir. 1992)); see also United States v. Muñoz-Gonzalez, 145 F.4th 21, 26 (1st Cir. 2025). And context cuts against key aspects of García's theory, as the following examples illustrate.

a

García's sentencing memo argued (as the plea agreement permitted) for a low-end within-guidelines sentence (27 months) based on his difficult upbringing. And the government's sentencing missive argued (as the agreement also allowed) for a high-end within-guidelines term (33 months) based on many § 3553(a) factors — including deterrence, public safety, and just punishment — tied to (i) his crime when “viewed through” the lens of Puerto Rico's “firearms and violence” problem and (ii) his not staying on the right side of the law after his “prior federal weapons conviction” by once again illegally possessing a machinegun.⁶ The government did all that without belittling his mitigating circumstances.

That reality is significant for a couple of reasons. For openers, our cases say that “where a plea agreement entitle[s] the prosecutor to argue for the high end of a guideline[s] range while the defendant argue[s] for the low end of that range, the prosecutor ‘[is] within fair territory in emphasizing facts that made a sentence at the low end of that [range] inappropriate.’ ” United States v. Montañez-Quiñones, 911 F.3d 59, 65 (1st Cir. 2018) (quoting United States v. Almonte-Nuñez, 771 F.3d 84, 91 (1st Cir. 2014)) (first four bracket sets added); see also Muñoz-Gonzalez, 145 F.4th at 27. Additionally, our cases also say that “[c]ommunity-based” concerns (like crime-trend lines) “are inextricably intertwined with deterrence.” United States v. Flores-Machicote, 706 F.3d 16, 23 (1st Cir. 2013) (emphasis added); see also id. at 23, 24 (doubling down on the principle that “the incidence of particular crimes in the relevant community appropriately informs and contextualizes the relevant need for deterrence“ but noting that a “judge's resort to community-based characteristics does not relieve him or her of the obligation to ground sentencing determinations in case-specific factors” (emphasis added)).

Now it's true (as García argues) that the government's sentencing memo cited caselaw affirming above-guidelines sentences.⁷ But despite what García seemingly implies, the memo didn't seek an upwardly variant term. And the memo didn't mention — let alone highlight — the cases’ upwardly variant sentencing results in going over “deterrence.” More, the memo also indicated that García still deserved a within-guidelines sentence even after accounting for community factors. Even more, García's sentencing memo cited caselaw affirming a below-guidelines sentence (evidently neglecting the goose/gander proverb explained in footnote 6).⁸

All of this is necessary to contextualize a full picture.

b

Again recall how García blasts the government for (apparently) “highlight[ing]” that his earlier “within-guideline[s] sentence” for illegal machinegun possession didn't “ ‘deter’ ” him from illegally possessing a machinegun “in this case” (emphasis added). We must say that the italicized phrase appears nowhere on the cited-to page of the government's memo. What the government wrote is that García's “prior federal weapons conviction” didn't act “as a deterrent” against his doing the same kind of crime again. The government didn't argue that a within-guidelines sentence wouldn't suffice.

And know this too. As part of his plea deal in the earlier case, García and the government agreed to jointly request a low-end within-guidelines sentence. See United States v. García-Toro, No. 19-cr-394 (D.P.R. Sep. 4, 2019), ECF No. 25 at 4 (plea agreement showing that the parties obligated themselves to recommending a low-end within-guidelines sentence using a category I criminal history). And it's no stretch to say that that sentence didn't deter him from further criminal conduct.

Given this wider perspective, we see nothing devious about the government's first-conviction-didn't-deter point — a point made in requesting a high-end within-guidelines term (a request permitted by the plea agreement), calculated to parry García's low-end within-guidelines bid. See United States v. Ubiles-Rosario, 867 F.3d 277, 288-89 (1st Cir. 2017) (concluding that “there was nothing sinister about the government's decision to highlight certain facts and factors in its sentencing memo[ ]” when “the government's reference[s]” were “firmly grounded in its recommendation”).

2

With the proper context set, García's appeal is ready for decision.

“The simplest way” to deal with an issue “is often the best” way, we've said time and time again. See Stor/Gard, Inc. v. Strathmore Ins. Co., 717 F.3d 242, 248 (1st Cir. 2013) (citation modified); see also, e.g., United States v. Cruz-Ramos, 987 F.3d 27, 39 (1st Cir. 2021). And fortunately there's a simple way here. So we can be (relatively) quick.

Plain-error findings occur only “sparingly.” United States v. Young, 470 U.S. 1, 15 (1985). And that's no less true in cases like ours. “Not all [plea] breaches will be clear or obvious,” the Supreme Court teaches. Puckett, 556 U.S. at 143. “Which is why plain error's second prong — requiring the complaining party to flag an undeniable mistake — often has ‘ “bite” in plea-agreement cases.’ ” Fargas-Reyes, 125 F.4th at 271 (quoting Puckett, 556 U.S. at 143).

Just so here. As we reveal in the remaining pages, any error on García's prosecutor's part — if error there was (an issue we don't decide) — wasn't clear or obvious under current controlling law. See generally United States v. Miranda-Martinez, 790 F.3d 270, 272 (1st Cir. 2015) (commenting that “[w]hile ․ plain error review applies ‘in the usual fashion’ to forfeited arguments that the government breached a plea agreement,” we know “that the fault with respect to the government's failure to uphold its end of a plea agreement ‘rests on the prosecutor, not on the sentencing judge’ ” (first quoting Puckett, 556 U.S. at 134, and then quoting Santobello v. New York, 404 U.S. 257, 263 (1971))).

a

Agreeing (or at least not disputing) that “ ‘context’ is relevant[,]” that “ ‘deterrence’ is [an] okay” consideration, and that “the word ‘upward’ ” variance never “explicitly” rolled off the prosecutor's tongue, García claims that fixating on these “truisms” obfuscates the “core” problem (at least according to him): “[w]here the agreement's within-guideline[s] recommendation reflects a mine-run case,” the government “offers no genuine justification for framing Puerto Rico as exceptionally violent and musters no explanation for how [he] or this gun-possession crime, connect to the themes it drew most attention to.” And trying to add a patina of plausibility to his claim, he cites cases saying that “[u]nmoored from any” offender- or offense-specific considerations, community concerns “cannot serve as building blocks for an upward variance.” See United States v. Rivera-Berríos, 968 F.3d 130, 137 (1st Cir. 2020); see also Flores-Machicote, 706 F.3d at 24 (ruling that judges can consider deterrence-related factors to support an upward variance if they don't “focus too much on the community and too little on the individual” defendants).

But García's efforts hit a snag. And that is that despite having the burden of proving plain error, see Greer v. United States, 593 U.S. 503, 508 (2021), he spotlights no “binding on-point” authority holding that prosecutors breach plea deals by focusing as much on community concerns as his prosecutor did in explaining where within the guidelines range a sentence should fall, see United States v. Langston, 110 F.4th 408, 419 (1st Cir.), cert. denied, 145 S. Ct. 581 (2024); see also United States v. Jones, 748 F.3d 64, 70, 73 (1st Cir. 2014) (affirming that a plain error must be “indisputable,” adding also that a “no-plain-error holding” isn't a “ ‘ruling on the merits’ ” and so doesn't “whisper[ ] even the faintest hint of how we might someday rule” on a preserved challenge (quoting United States v. Caraballo-Rodriguez, 480 F.3d 62, 70 (1st Cir. 2007))).⁹

García says that his case is like Mojica-Ramos. The prosecutor there definitely only paid “lip service” to his duty to request a within-guidelines sentence. See 103 F.4th at 850 (quoting Almonte-Nuñez, 771 F.3d at 89). Quite problematically, he flashed “250 photos” of guns and drugs plus “an unanalyzed video” of a person “ ‘resembling’ ” defendant as “ ‘additional’ ” uncorroborated “ ‘evidence’ of [defendant's] ‘likely’ participation in ‘other criminal behavior beyond the machinegun count charged.’ ” See id. And calling the charged offense “ ‘exceptional,’ ” id., he also claimed that defendant's “conduct” constituted “ ‘a big part of the’ ” violent-crime “ ‘problem’ ” on the island, see id. at 851. Which triggered our prosecutorial-breach finding. See id. at 853.

But night-and-day differences separate García's situation from Mojica-Ramos. García's prosecutor affirmatively requested a within-guidelines sentence without labeling the crime “exceptional” or discussing his “ ‘likely’ ” involvement in other crimes. And while he “generally discussed the prevalence of firearm violence in Puerto Rico,” the prosecutor only offered two gun-related photos (not 250) and never suggested that García's “conduct was itself violent or played a ‘big part’ in such violence.” See United States v. Aponte-Colón, 104 F.4th 402, 413 (1st Cir. 2024) (distinguishing Mojica-Ramos on similar grounds). What's more, his memo's gun-violence breakdown reads like a cut-and-paste of a colleague's memo in Aponte-Colón — where we found no plea breach. See id. Compare United States v. García-Toro, No. 19-cr-394 (D.P.R. Sep. 27, 2024), ECF No. 40 at 4-9, with United States v. Aponte-Colón, No. 19-cr-394 (D.P.R. May 2, 2022), ECF No. 42 at 5-11.

b

Still clinging to the idea that Mojica-Ramos helps with the obvious-error showing, García also (and somewhat relatedly) argues that language in the government's memo saying that machineguns aren't self-defense weapons went too far by “insinuat[ing] unproven, offensive intentions” — “veiled speculation [that] signaled that within-guideline[s] sentencing, explicitly endorsed by the plea agreement, was insufficient.” But the Mojica-Ramos prosecutor “insist[ed] that the [judge] consider unproven conduct,” see 103 F.4th at 853 (emphasis added), “offer[ing]” — as noted in our lettered sub-subsection a — “ ‘additional evidence’ of [defendant's] ‘likely’ participation in ‘other criminal behavior beyond the machinegun count charged,’ ” id. at 850 (emphases added). But that's hardly what the prosecutor did here, meaning this aspect of García's argument isn't a needle-mover either.

c

Again quoting the government's sentencing memo, García elsewhere insists that “[t]he prosecution further urged the [judge] to craft” a message-sending sentence to other “ ‘people’ ” showing “that they'd ‘face serious consequences’ ” if they commit “this possession-only” crime — “echoing the type of coded advocacy that ․ constitutes breach” (to be absolutely clear, only the text inside the internal quotation marks comes from the government's dispatch). And he claims that our Canada case supports his thesis. It doesn't.¹⁰

The Canada prosecutor “affirmatively” failed to “recommend” an agreed-to “36 month” sentence “within a context suggesting that she had” a “greater” term “in mind.” See 960 F.2d at 270. Agreeing on the record that she was “stuck with the plea agreement,” she urged (in her own words) a “lengthy period of incarceration” to deliver “a very strong message.” See id. at 269. All this while adopting a “grudging and apologetic” tone regarding the plea agreement. See id. Which sparked our prosecutorial-breach holding. See id. at 271.

García's case and Canada are apples and oranges. While the prosecutor here suggested that García deserved “very serious consequences,” he did so against the backdrop of affirmatively requesting a high-end within-guidelines sentence (as opposed to some other term on that range) twice in the memo and twice more at the sentencing — a request perfectly in sync with the plea agreement. And he didn't express any regret over “the plea agreement's terms or otherwise suggest that [he] would seek a different sentence if free to do so.” See Aponte-Colón, 104 F.4th at 410-11 (quoting United States v. Irizarry-Rosario, 903 F.3d 151, 154 (1st Cir. 2018), and distinguishing the situation before it from Canada).

d

Continuing his efforts to (somehow) surmount plain error's obviousness prong, García cites United States v. Gonczy, 357 F.3d 50 (1st Cir. 2004). But Gonczy isn't the difference-maker he thinks it is. As promised in the plea agreement, the Gonczy prosecutor requested a low-end within-guidelines sentence. Id. at 51. But after saying how the defendant had (among other things) “ruined many lives,” she stated that he “at a minimum deserve[d] what the guidelines provide for and those are his just deserts.” Id. at 53-54. The trouble (as we saw it) was that despite pledging to back a low-end within-guidelines sentence, the prosecutor basically said that the full guidelines range — spanning 17 months — constituted the least amount of time that he should get. Nothing resembling that occurred here, however, thus removing García's case from Gonczy’s reach. See generally Fargas-Reyes, 125 F.4th at 272 n.9 (distinguishing Gonczy for similar reasons).

Upshot

All that's left to say is that we affirm García's sentence and judgment.

FOOTNOTES

1.   García's full name appears in our caption (and we use “García,” per Spanish-naming norms).

2.   We take the facts from different materials before the district judge — including the indictment, the plea agreement, the uncontested parts of the presentence report, the sentencing memos, and the sentencing transcript. See, e.g., United States v. Diaz-Serrano, 77 F.4th 41, 44 (1st Cir. 2023); United States v. LeBlanc, 169 F.3d 94, 94 (1st Cir. 1999).

3.   A couple FWIWs: García doesn't attack the substantive reasonableness of his sentence. See generally United States v. Ponzo, 171 F.4th 507, 512 n.3 (1st Cir. 2026) (observing that a substantive-reasonableness challenge “focuses on the sentence's length” (emphasis omitted)). And he doesn't appeal the judge's order revoking his supervised release and imposing a consecutive 15-month revocation prison sentence.

4.   A “variance” is a sentence outside the guidelines range based on the weighing of statutory factors — seriousness of the offense, need for deterrence, achieving fair punishment, and the like. See, e.g., United States v. Miranda-Díaz, 942 F.3d 33, 40 (1st Cir. 2019) (discussing 18 U.S.C. § 3553(a)).

5.   He calls the government's sentencing document “the primary delivery vehicle for its sentencing-advocacy payload.”

6.   The maxim “what's good for the goose is good for the gander” springs to mind. See generally Ireland v. Digitraidix, LLC, No. 24-23069, 2025 WL 2719998, at *9 n.8 (S.D. Fla. July 18, 2005) (explaining that the goose-gander principle means that what applies to one should apply to both). García, however, rejects any “effort to equate” his bid for a low-end within-guidelines sentence with the government's push for a high-end within-guidelines term because of the prosecutor's (so-called) “coded invitations” for a higher sentence (emphasis added). Unfortunately for him, his coded-invitations claim flunks plain-error review — as the rest of this opinion confirms.

7.   Garcia focuses most particularly on the government's citing United States v. Viloria-Sepúlveda, 921 F.3d 5, 10 (1st Cir. 2019), United States v. Narváez-Soto, 773 F.3d 282, 286 (1st Cir. 2014), and Flores-Machicote, 706 F.3d at 23.

8.   He relied on United States v. Martin, 520 F.3d 87 (1st Cir. 2008), a case upholding the judge's decision that the relevant § 3553(a) factors called for a serious “downward“ variance from the guidelines sentencing range. See id. at 89-90, 98 (emphasis added).

9.   It goes without saying (but we say it anyway) that developing a sustained argument out of ․ legal precedents” is a party's “job,” not ours. See Town of Norwood v. FERC, 202 F.3d 392, 405 (1st Cir. 2000).

10.   García also cites Mojica-Ramos. But we've already shown that his circumstances fall outside that case's compass.

THOMPSON, Circuit Judge.