GORSUCH, Circuit Judge.
This is the latest installment in a long running class action lawsuit about conditions inside Albuquerque’s jails. In this iteration of the case, we must answer only a single question: Does an order withdrawing approval of a class action settlement agreement qualify as a “final decision” subject to appeal under 28 U.S.C. § 1291? The answer is no. Like an order granting a new trial under Fed.R.Civ.P. 59, or an order granting relief from a judgment under Fed.R.Civ.P. 60(b), an order unraveling a class action settlement agreement is anything but a “final decision.” Such an order simply presses the reset button, vacates any prior final decision, and marks the case for renewed litigation. Without a final decision amenable to our review under § 1291, we must and do dismiss this appeal.
[1291]*1291I
While the history of this dispute is long and complex, for purposes of this appeal a thumbnail sketch will do. In 1995, the plaintiffs sued the City of Albuquerque, Bernalillo County, and various individuals involved in operating the Bernalillo County Detention Center (“BCDC”). The plaintiffs sought to represent a class of “all prisoners who are presently, or will be confined in the” BCDC. Compl. ¶ 1. Soon, other members of the class intervened to form a sub-class seeking to represent “persons with mental and/or developmental disabilities who are now, or in the future will be, detained at” BCDC. ApltApp. at 341. The plaintiffs alleged that the conditions at BCDC were constitutionally deficient due, in large part, to overcrowding.
Eventually, after the class and subclass sought and received certification, the parties negotiated a pair of settlement agreements — one for the class and another for the sub-class — which the court approved in 1997 and over which it retained continuing jurisdiction. The parties operated under these agreements until 2003 when the County built a new jail, the Metropolitan Detention Center (“MDC”), and transferred all the prisoners from BCDC to MDC. But this transfer only prompted a new dispute. The plaintiffs argued that the 1997 settlement agreements should, in effect, transfer with all of the County’s prisoners to the new MDC and govern the defendants’ conduct there. For their part, the County and the other defendants opposed extending the agreements to MDC. After much wrangling, the parties eventually reached a new pair of settlement agreements that superseded the 1997 deals. By their express terms, the new agreements governed conditions only at MDC. Pursuant to Fed.R.Civ.P. 23(e), the district court gave its approval to these new agreements in June 2005 and, as before, retained continuing jurisdiction over their implementation.
The 2005 agreements soon ran into trouble of their own. After moving all its prisoners from BCDC to MDC, the County signed an Inter-Governmental Agreement (“IGA”) with the federal government that allowed federal detainees to be housed at BCDC. Under the terms of the IGA, the County bore certain contractual responsibilities concerning the care and treatment of federal detainees. At the same time, however, the County wasn’t required to operate BCDC but was allowed instead to contract out operational obligations to a private company, Cornell Corporation. According to the plaintiffs, however, the County misrepresented this arrangement to them by suggesting that Cornell alone bore contractual duties to the federal government. And it was only because of this alleged misrepresentation, the plaintiffs say, that they agreed to restrict the coverage of the 2005 settlement agreements to the MDC facility rather than to insist on a deal that continued to address the treatment of prisoners in the BCDC facility.
After the plaintiffs claimed they learned the true nature of the County’s role in the IGA, they brought the issue to the district court’s attention. Many motions and responses and arguments followed, all of which culminated in the district court’s March 31, 2009 order finding that the County had misrepresented to the plaintiffs its IGA interactions with the federal government. Based on this finding, the district court withdrew its Rule 23(e) approval of the 2005 settlement agreements and gave the plaintiffs permission to rescind those agreements, which the plaintiffs proceeded to do. Shortly after issuing its decision, the district court judge found that she needed to recuse from the case and the matter was assigned to a new judge.
[1292]*1292II
The defendants now seek to appeal the district court’s March 31, 2009 order. In their appeal, the defendants raise a host of arguments for reversal, including a contention that the district court judge should have recused herself before issuing the March 31, 2009 order and so allowed another judge to consider the plaintiffs’ complaint that they had been misled. But before the defendants can get to any of that, they bear the burden of establishing this court’s authority to hear their appeal. This burden, the defendants say, they can carry for a variety of reasons: the March 31, 2009 order they seek to challenge came after a final judgment; the order is appealable under the collateral order doctrine; and the order implicates the jurisdictional authority of the district court. We address each of these theories in turn.
A
The courts of appeals are creations of Congress and the boundaries of their jurisdiction are staked by statute. When it comes to when federal appellate courts may take a case, Congress has said that we may usually hear appeals only from “final decisions of the district courts of the United States.” 28 U.S.C. § 1291. And a final decision does not normally occur “until there has been a decision by the District Court that ‘ends the litigation on the merits and leaves nothing for the court to do but execute the judgment.’ ” Van Cauwenberghe v. Biard, 486 U.S. 517, 521-22, 108 S.Ct. 1945, 100 L.Ed.2d 517 (1988) (quoting Gatlin v. United States, 324 U.S. 229, 233, 65 S.Ct. 631, 89 L.Ed. 911 (1945)). A final decision is, put differently, one by which the district court “disassociates itself from a case.” Swint v. Chambers Cnty. Comm’n, 514 U.S. 35, 42, 115 S.Ct. 1203, 131 L.Ed.2d 60 (1995); Mohawk Indus., Inc. v. Carpenter, — U.S. -, 130 S.Ct. 599, 604-05, 175 L.Ed.2d 458 (2009).
Undergirding the final decision rule is Congress’s implicit judgment that the district judge is best positioned to bear primary responsibility for policing the tactics and maneuvers of the litigants in ongoing litigation — and that the district judge can better exercise this responsibility if appellate courts do not regularly intervene and second-guess the district judge’s rulings before he or she has reached a final decision on a particular matter. Better to wait until the district judge has made up his or her mind than to intervene when things remain in flux and the district court could still reverse course and provide the very relief a complaining party might seek to achieve in an appeal. See RichardsonMerrell, Inc. v. Roller, 472 U.S. 424, 436, 105 S.Ct. 2757, 86 L.Ed.2d 340 (1985); United States v. Wampler, 624 F.3d 1330, 1334-35 (10th Cir.2010).
The defendants say their appeal satisfies the final decision rule because the district court’s 1997 order (and perhaps its 2005 order) approving the parties’ settlement agreements were final judgments for purposes of Fed.R.Civ.P. 58. In the defendants’ view, because an appealable final decision (or decisions) preceded the March 31, 2009 order, this makes the March 31, 2009 order a final decision, too. Put simply, in the defendants’ estimation, any post-judgment order is automatically subject to appellate review.
This is mistaken. While the defendants are correct that a final judgment is the paradigmatic “final decision” appeal-able under § 1291, see Midland Asphalt Corp. v. United States, 489 U.S. 794, 798, 109 S.Ct. 1494, 103 L.Ed.2d 879 (1989), it doesn’t follow from this fact that every case with a final judgment in it is appeal-able. The defendants are too quick to generalize and in doing so run afoul of the logical fallacy of accident. See generally Nicholas Bunnin & Jiyuan Yu, The Black[1293]*1293well Dictionary of Western Philosophy 248 (2004). Just because all the people you’ve met lately are kind doesn’t mean all people are kind. In much the same way, just because final judgments can be appealed doesn’t mean everything in every case that has a final judgment may be appealed.
Fact is, every post-judgment decision must be assessed on its own terms to determine whether it is a final decision amenable to appeal. The plain language of § 1291 tells us as much, authorizing this court to entertain only “final decisions”— that is, decisions that are themselves final — not any old decision in any case where a final judgment happens to be knocking about. And while it is surely true That many post-judgment orders qualify as final decisions on their own terms, not every post-judgment order does. See 19-202 Moore’s Federal Practice § 202.13.
The nature of the defendants’ appeal in this case illustrates as much. Before this court, the defendants seek to challenge only the March 31, 2009 order, not the 1997 or 2005 orders that, they contend, constitute final judgments.1 The defendants’ arguments before us thus aren’t aimed at attacking any final judgment. Just the opposite, in fact: the defendants seek the benefit of those putative final judgments, challenging only the March 31, 2009 order that effectively undid them. So it is that, when assessing our authority to hear this appeal, our focus naturally turns, as it must, to the district court order that the defendants’ appeal seeks to attack, not other orders the appeal doesn’t seek to upset.
Properly framed, then, this appeal raises the question whether an order withdrawing a settlement approval is an appealable final decision. And the answer to that question must be no. A district court’s order in these circumstances doesn’t “disassociate” the court from the case; it doesn’t “end the litigation on the merits.” Just the opposite: the order ensures litigation on the merits will continue in the district court. If the 1997 and 2005 orders approving the parties’ settlement agreements amounted to final judgments, the March 31, 2009 order was their antithesis, obliterating any finality the case might have once enjoyed, “vacatfing] [the] judgment and setfting] the stage for farther trial court proceedings.” 15B Charles A. Wright, Arthur R. Miller & Edward H. Cooper, Fed. Prac. & Proc. § 3916, p. 367 (2d ed. 1992) (“Wright & Miller”). The district court’s challenged order essentially indicates that any prior final decision is defunct, gone, and further litigation on the merits must resume, and the usual rule that we must not interfere with ongoing district court proceedings governs.
That this is so even the defendants cannot help but confirm backhandedly, complaining to us as they do that the district court’s March 31, 2009 order means they will “face[ ] the very serious and unwholesome prospect of protracted litigation of a class action lawsuit.” Aplt. Br. in Supp. of Appellate Jurisdiction at 18. Whether or not the renewed proceedings in district court will be unwholesome or protracted we don’t know. But the defendants are surely right that, under the order’s terms, they face more litigation on the merits in the district court. And that irrepressible fact dictates our conclusion that the March 31, 2009 order is no final decision.
Our holding finds compelling parallels in other contexts. In a pair of decisions for [1294]*1294this court, Judge Brorby has explained that an order granting a motion under Fed.R.Civ.P. 60(b) reopening a judgment and an order declining to enter a proposed consent decree are not appealable under § 1291. This is because in both situations “the district court’s order ensures that litigation [on the merits] will continue in the district court.” Utah v. Kennecott Corp., 14 F.3d 1489, 1492 (10th Cir.1994); see also Stubblefield v. Windsor Cap. Grp., 74 F.3d 990, 996 (10th Cir.1996). Likewise, we have long held that a district court’s post-judgment order granting a new trial pursuant to Fed.R.Civ.P. 59 isn’t an appealable final decision for the same reason. See, e.g., Delano v. Kitch, 663 F.2d 990, 1001 (10th Cir.1981); Kanatser v. Chrysler Corp., 195 F.2d 104, 105 (10th Cir.1952); see also Allied Chem. Corp. v. Daiflon, Inc., 449 U.S. 33, 34, 101 S.Ct. 188, 66 L.Ed.2d 193 (1980). A final appealable judgment may well already exist in a case before a motion under either Rule 59 or Rule 60(b) is filed. But the fact a preexisting judgment exists doesn’t mean an order granting relief under those provisions is itself also appealable. Far from it, such orders generally aren’t appealable because, just as here, they set aside or undo a judgment and so settle nothing with finality except the fact that more litigation is on the way.
Rather than focus on these parallels, defendants ask us to look to an unpublished (and so non-binding) decision, Roose v. Patrick, 98 Fed.Appx. 719 (10th Cir. 2004), that, they say, should lead us to conclude all post-judgment orders are appealable final decisions. Examining Roose, however, confirms the opposite conclusion. After the district court entered a judgment dismissing her case, Karen Roose filed a motion for leave to file a second amended complaint. Id. at 721. The district court denied the motion and Ms. Roose sought to appeal this post-judgment order. Id. We permitted the appeal only because the court’s order finally decided the amended complaint issue and no further proceedings were contemplated in the district court. Id. at 722-23. In this way, Roose is similar to the situation where a district court denies a post-judgment motion to reopen a final judgment under Rule 60(b), or where it denies a motion for a new trial under Rule 59— decisions we’ve said usually qualify as final decisions for purposes of § 1291 because they usually signal that the district court’s business is done, that it has disassociated itself from the case, that we may act without stepping on the district court’s toes. See, e.g., Stubblefield, 74 F.3d at 993; see also Wright & Miller § 3916 at 362 (“The general rule that appeal can be taken from final denial of a motion to vacate a judgment is well settled.”). But none of this speaks to the converse situation now before us, where a post-judgment order grants relief vacating a judgment. Indeed, Roose only serves to highlight the profound analytical difference between post-judgment orders that decline to undo the judgment (a situation which usually marks the end of the parties’ road in district court) and those that grant such relief (and so signal only that litigation in the district court is far from done).
B
While the district court’s March 31, 2009 order may not disassociate the district court from the case, the defendants reply that it does at least finally decide that the district court’s earlier 1997 and 2005 decisions aren’t final. Put differently, the defendants argue that the March 31, 2009 order deprives them of the repose secured by the parties’ earlier settlement agreements — the right to avoid trial or at least renewed litigation. And this right, they say, will be irretrievably' — or finally — lost if appellate review is deferred.
But that is not enough. Many interlocutory orders can be characterized as “final” [1295]*1295in the sense that they “irretrievably” decide a question about the scope and duration of the district court’s proceedings. Should this deposition proceed? Should that interrogatory be allowed? How about this or that question in cross-examination? The district court’s decisions on any of these questions can be “final” in the sense that they are definitive and not practically reviewable later. But they rarely qualify as final appealable decisions for purposes of § 1291 because they do not signal the litigation is over; neither do they disassociate the district court from the case. To decide such questions would require us to do exactly what Congress has told us not to do — enmesh ourselves in the job of policing the parties’ tactics and maneuvers in ongoing district court litigation. Neither, in any event, are the district court’s decisions themselves always as final as they seem: the parties remain free to seek, and in most cases the district court remains free to revisit and reconsider, interlocutory rulings on questions like whether a deposition should be allowed or a settlement agreement disallowed. To be a final decision in the relevant sense for § 1291, then, it is generally not enough that a decision might seem at one point to “irretrievably decide” the scope and duration of litigation in the district court; usually it must end that litigation.
The defendants urge that a contrary result is compelled by Cohen v. Beneficial Industries Loan Corp., 337 U.S. 541, 69 S.Ct. 1221, 93 L.Ed. 1528 (1949), which held that certain otherwise interlocutory orders can qualify for appeal under § 1291 when they concern some “collateral” issue and have “practically final” effects on the parties. Id. at 546, 69 S.Ct. 1221. The undoing of a settlement agreement is not among those matters appealable under Cohen, however.
Neither can our conclusion on this score come as a surprise. We said as much long ago, in Desktop Direct, Inc. v. Digital Equipment Corp., 993 F.2d 755 (10th Cir. 1993). There, a party attempted to appeal a district court’s post-judgment order rescinding a final judgment that was predicated on a settlement agreement between the parties. The district court had rescinded the final judgment because, much as here, the parties’ settlement agreement had been induced by a misrepresentation. Id. at 756. Before this court, the putative appellant argued that, unless its appeal was heard immediately, it would forever lose the right — guaranteed to it by the parties’ settlement agreement — not to endure further litigation in the district court. Id. at 758. Writing for this court, Judge Logan declined to permit an appeal under Cohen. Id. at 760. The Supreme Court then granted certiorari and proceeded to affirm Judge Logan’s decision unanimously. Digital Equip. Corp. v. Desktop Direct, Inc., 511 U.S. 863, 884,114 S.Ct. 1992, 128 L.Ed.2d 842 (1994). In doing so, the Supreme Court explained that “Congress’s final decision rule would end up a pretty puny one” if immediate appellate review, whether under Cohen or otherwise, were available every time a party claimed that some interest, such as the right not to stand trial, would be “irretrievably lost” due to an erroneous district court decision. Id. at 872, 114 S.Ct. 1992. So it is that, as Desktop Direct held and we reaffirm today, a settlement agreement’s promise against future litigation, in whatever form, is insufficient to warrant an appeal under the Cohen doctrine. It is insufficient here in the civil context when a party seeks to enforce a rescinded settlement agreement, and it is equally insufficient in the criminal context when a party seeks to enforce a rejected plea deal, see Wampler, 624 F.3d at 1335.
The reason for this rule, the Supreme Court has explained, lies in the fact that the only time a claimed “right not to [1296]*1296stand trial” will justify immediate appellate review under Cohen is when a “statutory or constitutional” provision guarantees that claimed right. Digital Equip., 511 U.S. at 874, 114 S.Ct. 1992; Desktop Direct, 993 F.2d at 760. And this is because, as the Supreme Court has frankly put it, statutory and constitutional guarantees against having to stand trial are more important than analogous legal entitlements arising from other instruments, including settlement or plea agreements. Digital Equip., 511 U.S. at 877-78, 114 S.Ct. 1992. Settlement and plea agreements are, after all, the creation of the parties and may bear only what force of law a court, in its judgment, will allow. See Wampler, 624 F.3d at 1338. Meanwhile, a statutory (or, of course, a constitutional) entitlement that includes a right not to be tried — a right that by its terms contemplates vindication prior to trial — is different in kind because courts in our constitutional order are bound to find some way to reconcile it with § 1291’s competing statutory command: “Where statutory and constitutional rights [not to be tried] are concerned ... the collateral order doctrine might ... be understood as reflecting the familiar principle of statutory construction that, when possible, courts should construe statutes (here § 1291) to foster harmony with other statutory and constitutional law.” Digital Equip., 511 U.S. at 879, 114 S.Ct. 1992; Wampler, 624 F.3d at 1336.
If there is to be any change to the Supreme Court’s bright line rule disallowing immediate appeals based solely on a settlement or plea agreement purporting to grant a right not to stand trial, the Court has indicated it must not be through litigation in this court but through a different avenue altogether. While Cohen was once essentially the only game in town for securing appellate review of “practically” final but otherwise interlocutory district court orders, Congress has since amended the Rules Enabling Act to allow the Supreme Court to prescribe rules “defining] when a ruling of a district court is final for the purposes of appeal under section 1291.” 28 U.S.C. § 2072(c). Thus, a congressionally prescribed means now exists for defining when a matter is sufficiently or “practically” final to warrant appellate review. (And, of course, Congress is and has always been free itself to create exceptions to the finality rule — a freedom Congress has demonstrated its willingness to exercise when it thinks appropriate, see Wampler, 624 F.3d at 1338 n. 4). Given this development, it makes sense that “[a]ny further avenue for immediate appeal” of interlocutory rulings not amenable to Cohen, including those involving a putative right not to stand trial guaranteed by a plea or settlement agreement, “ ‘should be furnished, if at all, through rulemaking, with the opportunity for full airing it provides,’ not through further [ad hoc] case-by-case expansion of the Cohen doctrine.” Wampler, 624 F.3d at 1338 (quoting Mohawk Indus., Inc., 130 S.Ct. at 609).2
[1297]*1297c
Unsuccessful in their main efforts at securing an appeal, the defendants seem at times to try still a different approach. They sometimes appear to suggest we should allow their appeal under § 1291 because the district court lacked jurisdiction to withdraw its previous order granting Rule 23(e) approval of the parties’ settlement agreements. Or at least because this particular district judge lacked legal authority to act because she should have (in the defendants’ view) disqualified herself before issuing the March 31, 2009 order. But whether the March 31, 2009 order was entered with or without jurisdiction or in error has nothing to do with the question whether that order was a final decision for purposes of § 1291. Whatever problem may or may not lurk in the district court’s March 31, 2009 decision, it wasn’t a final one. And that’s the dispositive point.
If the rule were otherwise — if a party could win admission into the court of appeals by, say, characterizing a district court’s challenged action as lacking a “jurisdictional” basis — no doubt the only thing that’d follow would be a flurry of “imaginative attempts to [rejcharacterize asserted errors as matters of district court power.” Wright & Miller § 3916, at 371. Instead of focusing on whether the district court’s decision was final, the parties would spend their time arguing, and we would spend our time asking, whether the district court’s decision is fairly characterized as jurisdictional. And that’s often a slippery question. The word “jurisdiction” has “many, too many meanings,” Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 90, 118 S.Ct. 1003, 140 L.Ed.2d 210 (1998) (internal quotation omitted), it is “capable of different interpretations,” Rumsfeld v. Padilla, 542 U.S. 426, 434 n. 7, 124 S.Ct. 2711, 159 L.Ed.2d 513 (2004), and it has sometimes been used with “excessive! ] exuberance]” to encompass “things other than the [true] absence of constitutional or statutory power to adjudicate a matter,” Flitton v. Primary Residential Mortg., 614 F.3d 1173, 1182 (10th Cir.2010) (Gorsuch, J., concurring in part and dissenting in part).
Even more problematic for defendants than the practical problems associated with their proffered test, it asks the wrong question. In § 1291, Congress told us to ask only whether a district court’s decision is final and Congress’s direction demands our respect, not our rewriting. Neither is there any need for rewriting. [1298]*1298Congress has already provided a way for parties to challenge a district court’s erroneous assertion of jurisdiction before the entry of a final judgment. That path is paved by 28 U.S.C. § 1651(a) and its approval of writs of mandamus. See, e.g., In re C & M Properties, 563 F.3d 1156, 1163 (10th Cir.2009). One of the primary functions of writs of mandamus has long been, and still is, to address situations where district courts exceed the bounds of their congressionally prescribed jurisdiction. Ours would be an exceedingly odd jurisprudence if we could rewrite Congress’s direction in § 1291 to allow the defendants to raise a jurisdictional complaint in this court on easier and different terms than Congress itself has authorized in § 1651.3
Ill
This case is fifteen years old.and, the defendants worry, marked for still more years dragging before the courts. No doubt that’s why they have tried so hard to secure appellate review now, in an effort to bring some measure of certainty and closure to this proceeding. We are hardly impervious to the defendants’ concerns: the delays and costs associated with civil litigation in modern America are substantial and worrisome, and even the most hard-boiled litigator may raise an eyebrow at a case lasting as long as this one. We are no less hopeful than the defendants that the able district court judge newly assigned to this case will facilitate its expeditious progress toward a lasting final judgment. Neither will the defendants be left without any means for raising the points of error they seek to pursue in this appeal: a district judge may revisit and reconsider not only his or her own interlocutory rulings but also those of a disqualified predecessor; and, of course, there will come a day when this case will reach final judgment and so become amenable to appellate review. Likewise, if this case languishes, the defendants, like all litigants, have (and always had) the ability to seek mandamus, which functions not only to confine district courts to their jurisdiction but also to “compel [them] to exercise [their] authority when it is [their] duty to do so.” In re Antrobus, 519 F.3d 1123, 1124 (10th Cir.2008) (internal quotation omitted). But one thing we may never do is disregard the bounds of our legal authority and assert § 1291 jurisdiction over an appeal where it doesn’t exist. To do so in this case would compound any error the defendants imagine with an impropriety of our own, making matters worse not better. It is, after all, a “central principle of a free society that courts,” no less than the other [1299]*1299branches of government, “have finite bounds of authority.” U.S. Catholic Conf. v. Abortion Rights Mobilization, Inc., 487 U.S. 72, 77, 108 S.Ct. 2268, 101 L.Ed.2d 69 (1988). We must respect that principle and those bounds no less when it is hard to do so than when it is easy. This appeal is Dismissed.