GORSUCH, Circuit Judge.
Sentencing someone to prison has to be one of the district judge's toughest tasks. So much is at stake for the defendant, the victim, and the community. So much responsibility rests on the judge's shoulders, along with the high expectation that the judge will wisely weigh things that cannot be easily weighed. How much punishment is enough to protect the public? To deter future wrongdoing? To reflect the gravity of the offense? And how much punishment suffices to accomplish all these things without verging on cold revenge or needless retribution? There's rarely a single right answer to hard questions like these. So our system depends, as perhaps it must, on the discretion of thoughtful judges.
One tool district judges have to help them in their unenviable task is the advisory sentencing guidelines. The guidelines seek to supply some sense of what other courts across the country are doing in similar cases and what sentencing experts think may be appropriate. See U.S. Sentencing Guidelines Manual (U.S.S.G.) ch. 1, pt. A, subpt. 1.3 (2012); Rita v. United States, 551 U.S. 338, 349, 127 S.Ct. 2456, 168 L.Ed.2d 203 (2007). Of course, each defendant must be assessed on his or her own terms: courts are not machine presses and sentences are not widgets to be churned out on some criminal justice conveyor belt. But a properly calculated guidelines sentence provides useful data, a "starting point" or "initial benchmark," even as it remains the judge's duty to tailor every sentence to the case and defendant at hand. Gall v. United States, 552 U.S. 38, 49, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007). In this case we confront two errors in the district court's benchmark guidelines analysis.
The first arose this way. Early in the sentencing hearing the district judge noted that Mr. Sabillon-Umana was but a bit player in a larger drug operation. In that light, the judge stated that he thought a guidelines base offense level of 32 sounded about right and he asked the probation officer to offer some justification for that number. The probation officer promptly obliged. He told the court that finding Mr. Sabillon-Umana responsible for 1.5 kilograms of cocaine and 1.5 kilograms of heroin sold by the larger conspiracy would yield the court's desired base offense level. By the hearing's end, the district court adopted those findings as its own and imposed a sentence based on them.
This upended the normal course of events. When sentencing a defendant involved in a conspiracy, the district court is supposed to start by making factual findings about how much of the conspiracy's criminal activity the defendant agreed
We admit the proper order of operations we've outlined rests in part on a questionable foundation. It assumes that a district judge may either decrease or increase a defendant's sentence (within the statutorily authorized range) based on facts the judge finds without the aid of a jury or the defendant's consent. It is far from certain whether the Constitution allows at least the second half of that equation. See, e.g., Jones v. United States, ___ U.S. ___, 135 S.Ct. 8, ___ L.Ed.2d ___ (2014) (Scalia, J., dissenting from denial of certiorari). But in our case Mr. Sabillon-Umana has not challenged the district court's power to find facts at sentencing. And so long as district courts enjoy that power, their factual findings must come before — not after — their sentencing decisions.
Neither is this order of operations procedure for procedure's sake. Putting sentencing judgments first and fact-finding second risks mistakes about both. Our case illustrates the potential. The district court sought to justify a base offense level of 32. To back into that number, the probation officer suggested that the court find Mr. Sabillon-Umana responsible for selling 1.5 kilograms each of cocaine and heroin, for a total of 3 kilograms. The court could support these numbers, the probation officer suggested, with three ancillary findings: that Mr. Sabillon-Umana made $500 for each ounce of heroin or cocaine he sold, that he sold the two drugs in a 50/50 ratio, and that the $27,080 he wired to family members while apparently unemployed came almost exclusively from his drug profits. The court adopted this analysis wholesale. But the math doesn't jibe. Dividing $27,080 in total profits by $500 in profits per ounce yields 54.16 ounces, or 1.5 kilograms, of heroin and cocaine combined — not 1.5 kilograms of each drug as the court found. And if 1.5 kilograms is the total amount of drugs properly attributable to Mr. Sabillon-Umana, he is eligible for a base offense level of 30, not 32, and a correspondingly lower advisory guidelines sentence.
We do not question the distinguished district judge's intuition that Mr. Sabillon-Umana was a minor player in the drug conspiracy, or that his sentence should reflect as much. But in our legal order properly found facts drive sentencing decisions, not the other way around. Before settling on a guidelines offense level or some other sentencing conclusion, a district court must take account of the facts — whether conceded by the defendant, found by a jury, or (perhaps) found by the court. When that process is reversed, mistakes and miscalculations can creep in, and we risk sending defendants like Mr.
All the same the government argues we should affirm. Mr. Sabillon-Umana never caught the district court's math mistake so, the government tells us, he's waived any complaint about it. This is a nonsequitur. Mr. Sabillon-Umana has long and clearly argued that the district court committed reversible legal error by starting with its proposed sentence and working backwards to fit the facts to that conclusion. He's steadfastly preserved that complaint and, as we have explained, it is well taken. The only thing that has somehow eluded everyone until this appeal is the fact that the district court's math doesn't work. And this court is surely entitled to take judicial notice of that factual mistake — now acknowledged by both parties — to illustrate the sort of consequences that can flow from failing to abide the order of operations the law requires at sentencing. See United States v. Burch, 169 F.3d 666, 671 (10th Cir.1999); Fed. R.Evid. 201(b).
Retreating, the government suggests there's still another reason why we shouldn't reverse. There's no need to reverse, the government says, because there's another way we might still get to the same base offense level (32) the district court did. To manage this feat, the government urges this court (yes, the court of appeals) to find as a factual matter that all of the drugs Mr. Sabillon-Umana sold were heroin, rather than a mix of heroin and cocaine as the probation officer suggested and the district court found. With this new factual finding in hand, the government promises, a base offense level of 32 would surely follow. The government's proposed procedure, however, not only calls on this court to become the fact-finder — a strange enough invitation — it also asks us to commit the same legal error the district court did: to back into a finding about the nature of Mr. Sabillon-Umana's drug trafficking activities only to support a judgment about an appropriate sentence. That we refuse to do.
But that doesn't end our encounter with this case. A second and separate error remains for us to address. At the sentencing hearing, the prosecutor proposed a downward departure from the advisory guidelines range in recognition of Mr. Sabillon-Umana's assistance to the government. See U.S.S.G. § 5K1.1. But when Mr. Sabillon-Umana argued that the district court should grant an even greater departure, the prosecutor shot back that the district court lacked the lawful authority to grant such relief. The district court agreed, noting that "it's really the Government rather than the Court that evaluates the value of the substantial assistance." That is incorrect. Section 5K1.1 does not grant prosecutors the power to control the length of a defendant's sentence. Rather, it is emphatically for the court, not the government, to determine the appropriate sentencing reward for the defendant's assistance. See id.; United States v. Krejcarek, 453 F.3d 1290, 1300-01 (10th Cir. 2006).
On appeal the government concedes the error it invited but strains again to avoid a remand. In the first place, the government says a remand isn't required because the district court caught and corrected the problem before issuing its sentence. But this much we just don't see. The transcript of the sentencing hearing shows the government misstating the law. The transcript then shows the court signaling its agreement with the government's mistaken view of the law. After that, however, the transcript never shows the government or the district court disavowing the error.
Both before and after United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005), this court has recognized that an obvious misapplication of the sentencing guidelines will usually satisfy the third and fourth elements of the plain error test. See, e.g., United States v. Rosales-Miranda, 755 F.3d 1253, 1259-60, 1262-63 (10th Cir.2014); Figueroa-Labrada, 720 F.3d at 1268-69; United States v. Meacham, 567 F.3d 1184, 1191 (10th Cir. 2009); United States v. Johnson, 414 F.3d 1260, 1263-65 (10th Cir.2005); United States v. Smith, 919 F.2d 123, 124 (10th Cir.1990). Other circuits have reached similar conclusions or even adopted an explicit presumption that a clear guidelines error will satisfy the latter two steps of plain error review. See, e.g., United States v. Vargem, 747 F.3d 724, 728-29 (9th Cir.2014); United States v. Wernick, 691 F.3d 108, 117-18 (2d Cir.2012); United States v. Slade, 631 F.3d 185, 191-92 (4th Cir.2011); United States v. Story, 503 F.3d 436, 440-41 (6th Cir.2007); United States v. Baretz, 411 F.3d 867, 877 & n. 7 (7th Cir.2005); United States v. Knight, 266 F.3d 203, 206-07 & n. 7 (3d Cir.2001).
This presumption is sound. If the guidelines form the essential starting point in any federal sentencing analysis (and they do), it follows that an obvious error in applying them "runs the risk of affecting the ultimate sentence regardless of whether the court ultimately imposes a sentence within or outside [the] range" the guidelines suggest. Rosales-Miranda, 755 F.3d at 1259. As the Third Circuit has well said: "[I]t is beyond cavil that the Guidelines are intended to, and do, affect sentencing. Indeed, that is their very raison d'etre." Knight, 266 F.3d at 207 (footnote omitted). In the language of plain error's third prong, the whole point of the guidelines is to affect the defendant's "substantial rights" by guiding the district court's analysis of how much of his liberty he must forfeit to the government. When the court's starting point is skewed a "reasonable probability" exists that its final sentence is skewed too. United States v. Uscanga-Mora, 562 F.3d 1289, 1295 (10th Cir.2009) (quoting United States v. Cook, 550 F.3d 1292, 1298 (10th Cir.2008)). And turning to plain error's fourth prong, what reasonable citizen wouldn't bear a rightly diminished view of the judicial process and its integrity if courts refused to correct obvious errors of their own devise that threaten to require individuals to linger longer in federal prison than the law demands?
Of course, presumptions can be overcome and the presumption that obvious guidelines errors meet the latter elements of the plain error test can be too. In some cases, the record will reveal a "fortuitous comment" from the sentencing judge making clear that its error in applying the guidelines didn't adversely affect the defendant's ultimate sentence. Knight, 266 F.3d at 207. For example, a district judge might proceed to analyze a case under alternative theories — one permissible, the other obviously mistaken — and arrive at the same sentencing conclusion either way. In cases along these and similar lines, this court and others have sometimes declined to remand for resentencing in the face of an obvious guidelines error, reasoning that a new sentencing proceeding would not help the defendant or enhance the integrity of judicial proceedings. See, e.g., United States v. Westover, 435 F.3d 1273, 1276-79 (10th Cir.2006); United States v. Brown, 316 F.3d 1151, 1161-63 (10th Cir. 2003).
This case, however, falls within the heartland of the presumption, not any exception. At the government's urging, the district court obviously erred in construing its authority under the guidelines' instructions in § 5K1.1. And whether a court clearly miscalculates the advisory guidelines range or clearly mistakes its entitlement to depart from that range under § 5K1.1, a defendant's substantial rights and the integrity of the judicial process are surely at risk: in either event the benchmark for the entire sentencing process rests on an obviously mistaken premise. We lack as well any fortuitous comment from the district court that might
The case is remanded for resentencing.