joined by TACHA, KELLY, O’BRIEN, TYMKOVICH, and HOLMES, Circuit Judges.
Everyone agrees that Hydro Resources, Inc. (“HRI”) must obtain a Safe Drinking Water Act (“SDWA” or “the Act”) permit to mine its property. The only question is: from whom? The Environmental Protection Agency (“EPA” or the “Agency”), which administers the Act, has chosen to delegate its permitting authority in the State of New Mexico to the New Mexico Environment Department (“NMED”), but with one exception: EPA has not delegated its authority to issue permits for mining activities on “Indian lands.” Thinking its land hardly qualified as “Indian land”— HRI owns its property in fee, it pays county real estate taxes, the land is uninhabited, and it is not inside any Indian reservation or otherwise set aside and superintended for Indian use—the company proceeded to apply for, and obtain, a permit from NMED. Initially, EPA professed no quarrel with this, and it has never questioned NMED’s administration of the Act. But eventually a dispute broke out over the status of HRI’s land and, after years of regulatory wrangling, EPA issued a “final land status determination” expressing its judgment that HRI’s land qualifies as “Indian land.” As a result, EPA ruled, HRI must seek and obtain its SDWA permit from it rather than NMED.
How did EPA reach this conclusion? By regulation, EPA chose to define the term “Indian lands”—the only lands for which it did not cede primary permitting authority to NMED—to be synonymous with “Indian country,” as that term is defined by 18 U.S.C. § 1151. Section 1151, in turn, provides primary federal criminal jurisdiction over certain territories: “Indian reservation[s],” “dependent Indian communities,” and “Indian allotments.” So it is that, for EPA to exercise primary permitting authority in this case, the Agency had to argue that the federal government, rather than the State of New Mexico, possesses primary criminal jurisdiction over HRI’s private property. In this case, EPA took the position that HRI’s land is Indian country and subject to federal jurisdiction because it is part of a “dependent Indian communitjy]” under § 1151(b).
But whatever HRI’s land is, it can’t be that. In Alaska v. Native Village of Venetie Tribal Government, 522 U.S. 520, 118 S.Ct. 948, 140 L.Ed.2d 30 (1998), the Supreme Court identified two “requirements” of all “dependent Indian communities” under § 1151(b). First, “the land in ques*1135tion” must be an “Indian community” in the sense that it has been explicitly “set aside” by Congress (or the Executive, acting under delegated authority) “for the use of the Indians as Indian land.” Id. at 527, 531, 118 S.Ct. 948. Second, “the land in question” must be “dependent” in the sense that it is “under federal superintendence.” Id. at 527, 118 S.Ct. 948. HRI’s land—the land in question in EPA’s final land status determination—is neither of these things.
Despite this, EPA argued before a panel of this court that we should cast our gaze beyond the particular land in question. In the Agency’s view, because some sufficiently significant (though unspecified) percentage of neighboring lands—what EPA calls “the community of reference”— is Indian country, HRI’s land must be considered Indian country, too. In defense of its view, EPA pointed to certain of this circuit’s cases, most pre-F%«eiie, suggesting the approach it took. Deeming itself bound by the same authority, a panel of this court upheld EPA’s classification of HRI’s land as Indian country. Hydro Res., Inc. v. U.S. EPA, 562 F.3d 1249 (10th Cir.2009) (“HRI II ”).
HRI responded to all this with a petition for en banc review. The company argued that the “community of reference” approach advanced by EPA and certain of this circuit’s cases is inconsistent with Ven-etie. HRI submitted, too, that our cases are in conflict with each other—while some follow EPA’s approach, others after Vene-tie have abjured the “community of reference” test, as have decisions in our sister circuits. Seeking to sort all this out, we granted HRI’s request for en banc review.
Having now heard the case anew, we find ourselves compelled to vacate EPA’s final land status determination. EPA’s interpretation cannot be reconciled with the Supreme Court’s explanation of § 1151(b)’s plain meaning. Venetie explicitly rejected a Ninth Circuit test composed of the very factors used in the “community of reference” test employed by EPA and certain of our pre-Venetie cases. Neither is the amorphous “community of reference” test compatible with the history and structure of the statute we are charged to interpret, or with the Supreme Court’s longstanding direction that criminal statutes should be interpreted clearly and precisely to afford fair warning of their reach.
None of this is to say that EPA must tether its SDWA permitting authority to a statute defining the scope of the federal government’s criminal jurisdiction over Indian lands. Had EPA chosen to define its authority under the SDWA in a different way, the result in this case might have been different. But we decide the eases as they come to us. And in this case, heeding the Supreme Court’s commands in Venetie requires us to grant HRI’s petition for review and vacate the Agency’s final land status determination.
I
The history of this dispute is long and tangled. Even so, some appreciation of its twists and turns is essential. We begin by examining briefly the history of the land in question (Section I.A), the regulatory scheme governing that land and the parties before us (Section I.B), the parties’ first lawsuit before this court (Section I.C), its subsequent remand to EPA (Section I.D), and the current appeal (Section I.E), all before we turn to address our jurisdiction and standard of review (Section II) and, at last, the merits of this appeal (Section III).
A
The land at issue in this case lies in what is commonly known as the “checkerboard” region of northwestern New Mexico. See generally Pittsburg & Midway Coal Min. *1136Co. v. Yazzie, 909 F.2d 1387, 1389-92 (10th Cir.1990). This region abuts the southern and eastern boundaries of the Navajo Reservation originally created by an 1868 treaty between the United States and the Navajos. Id. at 1389. And a checkerboard it is, marked by alternating parcels of land owned by the state, the federal government, the Navajo Nation, individual Navajos, and private persons and entities. See id. at 1423 app. A (map section marked “J”); Appendix.1
The checkerboard seems to have had its start with the railroad. In the late nineteenth century, the federal government granted certain lands in the region to railroad companies in an effort to induce construction. “These grants typically consisted of alternating one-mile-square parcels on each side of the planned line for the railroad tracks....” HRI II, 562 F.3d at 1254 n. 3. From this, a checkerboard was born, aided and abetted by the fact that other tracts of land in this area, though still formally held in the public domain, were occupied by Navajos, while still others were being rapidly snapped up by white and Mexican settlers. Yazzie, 909 F.2d at 1390.
And this was just the start of the complications. As Judge Anderson explained in his thorough history of the area, by the turn of the twentieth century federal officials became concerned that the new settlers were “appropriating the limited water holes for themselves.” Id. at 1390. So, in an effort to protect the Navajo population, President Theodore Roosevelt signed two executive orders, E.O. 709 and E.O. 744, in 1907 and 1908, respectively. The combined effect of these orders was to add much of the land in this area to the Navajo Reservation. Id. at 1391. At the same time, the President’s orders expressly preserved preexisting private property rights, including the railroad land grants. Thus, “the extension to the Reservation” further complicated the variegated character of the area. Id. at 1391 n. 6.
Still more checkerboarding followed. It seems the government did not intend the area to become a permanent addition to the Reservation. See id. Instead, the plan apparently was to allow Indians2 then living in the area a brief period to claim and “receive 160-acre allotments in sever-alty without interference from whites and Mexican stockmen.” Id. at 1390. After that, any unallotted land was to revert from reservation status back to the public domain. See id. And this is exactly what transpired in 1911 by virtue of another executive order, this one issued by President Taft. Id. at 1392. Since then, land in this area has changed hands many times in still more complicating ways. In 1928, for example, Congress appropriated funds for the purchase of some privately held former railroad tracts in order that they might be held by the federal government for the benefit of the Navajo. HRI II, 562 F.3d at 1254 n. 3. Meanwhile, other parcels now belong to the New Mexico state government or remain in the hands of non-Indians.
HRI’s land falls into this final category. In 1970, the federal government sold 160 *1137acres in the southeast quadrant of “Section 8,” Township 16N, Range 16W, to the United Nuclear Corporation. See Appendix. In turn, United Nuclear later sold the land to HRI. There are no inhabitants on HRI’s land. Except for the brief period from 1907-11, it has not been set aside by Congress for Indians or placed under federal superintendence for their benefit. The remaining three quadrants of Section 8 land not owned by HRI are still owned in fee by the United States. R. 15c App. XI at para. 3. Other adjacent sections include parcels held in trust for the Navajo by the United States (Sections 9 and 17) and land owned by the State of New Mexico (Section 16). See Appendix.
Section 8 lies within McKinley County, New Mexico. The county seat resides in the city of Gallup, “approximately 11 miles southwest of the Section 8 land.” HRI II, 562 F.3d at 1254. The state and county exercise jurisdiction over private lands throughout the checkerboard area. So, for example, the State of New Mexico maintains the only road access to Section 8, State Highway 566, while McKinley County shares responsibility with the federal government for other roads in the vicinity. HRI II, 562 F.3d at 1254. The County provides essential public sendees to private lands like HRI’s, including fire, police, and emergency services. Id. The Gallup/McKinley County public school system offers public education and school transportation for those in the area. Id. And HRI pays annual property taxes on its land to McKinley County. Id.
Section 8 also falls within the boundaries of the Navajo Church Rock Chapter. The Chapter is a political and social unit of the Navajo Nation, with its boundaries and membership determined by the Tribe. See id. at 1255.3 The current boundaries as drawn by the Tribe include tracts owned in fee by the United States, privately held lands, and Navajo Nation lands. Id. At the same time, the Chapter’s boundaries exclude at least one parcel of state land, Red Rock State Park, thus creating a sort of “doughnut hole” in the middle of the Chapter. See Appendix. While most of the land in the Chapter lies north of Interstate 40, the Chapter does include a narrow traverse across the Interstate and a small tract on the highway’s south side. See id.
The political and social center of the Church Rock Chapter is the Chapter House, located in the village of Church Rock, six miles east of Gallup and about six miles south of Section 8. See id. Much of the membership of the Chapter lives in close proximity to the Chapter House, and Chapter members typically visit the Chapter House at least once a month for meetings and social activities and services. R. 40 at B-29. At the same time, nearly half *1138of the Chapter’s members are employed in Gallup and so travel there frequently, compared to just 2% who are employed within the Chapter’s boundaries. R. 40 at B-24. The Chapter recognizes that private lands within its boundaries are subject to state jurisdiction and control, R. 40 at C-16, and the Chapter provides no infrastructure or services to HRI’s portion of Section 8, R. 6 at 2. Indeed, a significant portion of the territory within the Chapter, including Section 8, consists of “rugged mountain ranges, canyons, and highlands” that, according to the Chapter, are “not suitable for community or industrial development.” R. 40 at B-40.
B
After purchasing its Section 8 land, HRI sought to mine it for uranium. In preparation, the company obtained various regulatory permits. Because HRI’s proposed mining operations contemplated the use of an underground injection system to extract the ore, SDWA regulations required the company to obtain approval of an underground injection control (“UIC”) plan aimed at mitigating the risk of contamination to potential drinking water sources.4
While EPA is responsible for administering the SDWA, Congress anticipated that the states would, at least sometimes, serve as the primary entities responsible for reviewing and granting or denying UIC permits. In the SDWA, Congress told EPA it could “either approve, disapprove, or approve in part and disapprove in part, [each] State’s” application to become the primary UIC permitting authority. 42 U.S.C. § 300h-l(b)(2). The Act then went on to direct EPA to promulgate certain standards that the state UIC regulatory programs would have to meet to achieve this distinction. See 42 U.S.C. § 300h; 40 C.F.R. § 144.1(e). Exercising these statutory authorities granted to it, EPA some time ago set standards for state UIC programs and approved NMED’s application to serve as the primary UIC permitting authority in the State of New Mexico, except with respect to underground injection wells “on Indian lands.” 40 C.F.R. §§ 147.1600-147.1601.
It is here the real regulatory complications begin. How do we know when underground injection wells lie on “Indian lands”? By regulation, EPA has chosen to define the phrase “Indian lands,” when it appears in SDWA regulations, to mean “Indian country,” as that term is defined by 18 U.S.C. § 1151. See 40 C.F.R. § 144.3. Adopted in 1948, § 1151 is part of the criminal code and circumscribes where the federal government or a tribe, rather than a state, may exercise primary criminal jurisdiction. At the same time, the statute has been used often, as EPA has chosen to use it here, to define the scope of federal authority over civil and regulatory matters. See Venetie, 522 U.S. at 527, 118 S.Ct. 948; Cohen’s Handbook of Federal Indian Law § 3.04[1] (Nell Jessup Newton et al. eds., 3d ed.2005) (hereinafter “Cohen (2005)”).5
*1139Section 1151 defines “Indian country” as encompassing three categories of land:
(a) all land within the limits of any Indian reservation under the jurisdiction of the United States Government, notwithstanding the issuance of any patent, and, including rights-of-way running through the reservation,
(b) all dependent Indian communities within the borders of the United States whether within the original or subsequently acquired territory thereof, and whether within or without the limits of a state, and
(c) all Indian allotments, the Indian titles to which have not been extinguished, including rights-of-way running through the same.
18 U.S.C. § 1151.
Having thus defined the scope of “Indian lands” under its SDWA regulations, EPA still faced the question: Who, if not the state, should enjoy primary authority to regulate wells on those lands? The SDWA entrusts primary UIC permitting authority to EPA but allows the Agency to delegate that authority to tribes, much as it does to states, at least with respect to permit applications “within the area of the Tribal Government’s jurisdiction.” 42 U.S.C. § 300j-l 1(b)(1)(B); see also 42 U.S.C. § 300h-l(e). And by regulation, EPA has indicated that a “Tribal Government’s jurisdiction” for these purposes may not extend beyond “Indian country,” as that term is (once again) defined in § 1151. See, e.g., Navajo Nation; Underground Injection Control (UIC) Program; Primacy Approval, 73 Fed.Reg. 65,556-01, at 65,558-65,560 (Nov. 4, 2008) (“EPA recognizes that 18 U.S.C. 1151 ... generally defines the limit of the area over which a Tribe may demonstrate authority.”).
In 1994, EPA chose to exercise this authority to delegate primary UIC permitting authority to the Navajo Nation for lands within the Navajo Reservation, as well as for certain other Navajo allotments and Navajo fee lands. HRI, Inc. v. EPA 198 F.3d 1224, 1232 (10th Cir.2000) PH HI I”). At the same time, however, EPA declined to approve the Tribe’s application to assume primary UIC permitting authority over all privately held fee lands in the checkerboard region where Section 8 resides. Id. at 1233. It was and is undisputed that these lands do not qualify as part of any Indian reservation within the meaning of § 1151(a), or as Indian allotments within the meaning of § 1151(c). Though the Tribe sought to persuade EPA that the lands nonetheless qualify as “Indian country” because, given social and political affinities in the area, they are part of a larger “dependent Indian community” within the meaning of § 1151(b), EPA rejected this claim. The Agency explained its view' that the Tribe had “not demonstrated that it has jurisdiction” over the lands in question, R. 13b at 239 (emphasis in original), adding that, “[bjefore it could determine if a parcel of land is part of a dependent Indian community (and therefore is Indian country), EPA would need more information about that particular parcel of land.” Id. at 238. The Agency thus left open at least the hypothetical possibility that there could be “Indian lands” within the checkerboard area over which it, rather than the Navajo Nation, might retain primary UIC regulatory authority—at least until such time as the Tribe could come forward with evidence showing that the “particular parcel of land” in question qualified as Indian country under § 1151(b).
*1140. C,
It is perhaps unsurprising that such a complex land ownership scheme, overlaid by such a complex regulatory scheme, might beget equally complex litigation. And so it did when HRI tried to ascertain which regulatory authority held the UIC permit it needed. HRI knew that the Tribe didn’t have permitting authority over its land, at least not yet. The remaining choices HRI thus confronted were EPA or NMED. Not conceiving of its land as part of a “dependent Indian community” within the meaning of § 1151(b), and absent any EPA decision holding otherwise, HRI requested a UIC permit from NMED. For their part, New Mexico state authorities agreed that HRI’s land wasn’t Indian land, reviewed HRI’s UIC application, and in 1989, approved it.
As part of the permitting process, NMED sought from EPA a mandatory “aquifer exemption” for HRI’s mining activities, because those activities contemplated the introduction of contaminants into an aquifer. Generally speaking, the SDWA prohibits contamination of underground aquifers. See 42 U.S.C. § 300h(b). But because certain aquifers “will never be used as sources of drinking water, ... EPA [has] adopted criteria for exempting [them] from SDWA requirements.” HRI I, 198 F.3d at 1233. In due course, EPA approved NMED’s requested exemption because, in EPA’s judgment, the aquifer under HRI’s land “does not currently serve as a source of drinking water” and “cannot now and will not in the future serve as a source of drinking water.” 40 C.F.R. § 146.4(a) & (b); see also HRI /, 198 F.3d at 1234; R. 15c App. II at 2 (noting that, even before any mining activity, “water quality at the Section 8 site is mineralized with naturally-occurring uranium, and uranium decay products ... exceeding U.S. EPA drinking water [standards].”).
About this time, however, a jurisdictional dispute arose regarding HRI’s planned mining operations on Section 8 and on nearby Section 17. The dispute proved protracted as state, federal, and tribal authorities wrangled over whether HRI’s UIC operations should be regulated by NMED or EPA. See HRI I, 198 F.3d at 1234-35. At some point during this back- and-forth, the Navajo Nation presented what EPA considered to be “substantial arguments to support its claim that Section 8 is within Indian country.” Id. at 1235 (quoting EPA opinion letter of July 14, 1997). Based on these assertions, EPA deemed Section 8’s Indian-land status to be “in dispute.” Id.
Eventually, in the late 1990s, HRI and NMED sought review of EPA’s assessment in this court. HRI argued that its Section 8 land wasn’t a “dependent Indian community” within the meaning of § 1151(b), and so primary UIC permitting authority rested with NMED, not EPA or the Tribe6 For its part, EPA asked the court to remand the matter because it still hadn’t reached a final decision on the question of Section 8’s status as Indian country. See HRI I, 198 F.3d at 1236 n. 6 (quoting EPA Brief). EPA explained its delay by pointing to the Supreme Court’s then-recent decision in Venetie, which indicated *1141that to qualify as Indian country under § 1151(b) “the land in question” must be set aside for Indians and federally superintended. Venetie, 522 U.S. at 531, 118 S.Ct. 948. In arriving at this holding, EPA noted, the Supreme Court expressly rejected a “more textured” balancing test adopted by the Ninth Circuit, a test the Ninth Circuit consciously modeled on preexisting Tenth Circuit jurisprudence. See State of Alaska ex rel. Yukon Flats Sch. Dist. v. Native Vill. of Venetie Tribal Gov’t, 101 F.3d 1286, 1291-93 (9th Cir.1996) (quoting Pittsburg & Midway Coal Mining Co. v. Watchman, 52 F.3d 1531, 1545 (10th Cir.1995)). EPA defended its remand request by stressing that it hadn’t yet had an adequate chance to “develop a record below with the Venetie standard in mind.” HRI I, 198 F.3d at 1236 n. 6 (quoting EPA Brief at 47).
Shortly before Venetie, this court in 1995 developed a two-step, multi-variable balancing test for identifying “dependent Indian communities” under § 1151(b), sometimes called the “Watchman test.” At its “first step,” the test required the identification of an “appropriate community of reference.” Watchman, 52 F.3d at 1543-44. When identifying an appropriate “community of reference,” we said, a court had to consider three factors: (1) “the geographical definition of the area proposed as a community,” United States v. Adair, 111 F.3d 770, 774 (10th Cir.1997); (2) “the status of the area in question as a community,” Watchman, 52 F.3d at 1543; and (3) “the community of reference within the context of the surrounding area,” id. at 1544. Within the second factor, the Watchman test called on us to inquire whether there is “an element of cohesiveness ... that can be manifested either by economic pursuits in the area, common interests, or needs of the inhabitants as supplied by that locality,” and consider whether the proposed community qualifies as “a mini-society consisting of personal residences and an infrastructure potentially including religious and cultural institutions, schools, emergency services, public utilities, groceries, shops, i*estaurants, and the other needs, necessities, and wants of modern life.” Id. at 1544. And within the third factor, we considered which public entity or entities “provide infrastructure, government, essential services, and employment” for the community. Adair, 111 F.3d at 775.
All of this, however, was just the beginning. Having identified a “community of reference,” our test then sought, at the “second step,” to determine whether that community qualified as a dependent Indian community. And this, we said, required the balancing of still more factors: “(1) whether the United States has retained title to the lands which it permits the Indians to occupy and authority to enact regulations and protective laws respecting this territory; (2) the nature of the area in question, the relationship of the inhabitants in the area to Indian tribes and to the federal government, and the established practice of government agencies toward the area; (3) whether there is an element of cohesiveness manifested either by economic pursuits in the area, common interests, or needs of the inhabitants as supplied by that locality; and (4) whether such lands have been set apart for the use, occupancy and protection of dependent Indian peoples.” Watchman, 52 F.3d at 1545 (internal quotation marks and alterations omitted). Only the first and fourth of these elements, however, added entirely new7 concepts to the mix; the second and third elements overlapped in significant measure with elements of the antecedent “community of reference” analysis. Even so, Watchman instructed the use of these elements at both steps in the analysis and held that, if both steps were satisfied, all land inside the “community of reference” qualified as “Indian country.” See id.
*1142In HRI I, a panel of this court acknowledged that Venetie altered this legal landscape and that EPA had not yet had a chance to issue a final determination about the status of HRI’s land in light of it. Accordingly, the panel held that EPA’s analysis of HRI’s Section 8 land wasn’t yet ripe for review and remanded the matter to the Agency for a final determination of the legal status of HRI’s land. HRI I, 198 F.3d at 1237, 1254. After reaching this holding, however, the panel proceeded “[i]n dicta,” State v. Frank, 132 N.M. 544, 52 P.3d 404, 408 (2002), to consider Vene-tie’s possible impact on this circuit’s Watchman test. On the one hand, the panel acknowledged that “Venetie may require some modification” to our test. 198 F.3d at 1248; see also id. at 1254. But, on the other hand, the panel also suggested that “nothing in Venetie speaks to the propriety of the first element of that test— determination of the proper community of reference.” Id. at 1248.
D
With that for guidance, EPA on remand proceeded to invite comments from interested parties. Ultimately, the Agency received comments from the State of New Mexico, McKinley County, various corporations, and more than one hundred Navajo allottees arguing that HRI’s Section 8 land should not be considered part of a dependent Indian community. At the same time, the Agency received comments from the Navajo Nation, the Navajo Church Rock Chapter, and many others arguing that the land should be considered Indian country. In addition to all this, EPA consulted the Interior Department’s Solicitor’s Office and the Navajo Nation.
At the end of its review, the Agency acknowledged that “[sjeveral commenters have suggested that the community-of-reference analysis is no longer intact.” EPA Land Status Determination, R. 44 at 4. Yet, seeming to take its cue from HRII’s intimation that Venetie had not spoken “directly” to Watchman’s threshold community of reference test, EPA concluded that the test survived Venetie—at least in the Tenth Circuit, if not elsewhere. Id. At the same time, the Agency decided that Venetie modified Watchman’s second step, replacing its four-part test with a two-part test focused on how much of the “community of reference” is set aside for Indians and federally superintended. If some sufficiently high, though unspecified, percentage of the “community of reference” met these requirements, EPA would treat all land within the community of reference as Indian country under § 1151(b). Id. at 11-12.
Turning to apply its understanding of § 1151(b) to the facts of this case, EPA concluded that the Navajo Church Rock Chapter was “the appropriate community of reference” at Watchman’s first step because the Chapter functions as a “mini-society.” Id. at 8-9. EPA did not, however, pause to consider whether the appropriate community of reference might be McKinley County, and it considered Gallup as a candidate only briefly in a footnote. Id. at 10 n. 64. Neither did EPA consider whether Section 8 might be part of no community at all; rather, its analysis seemed to presuppose that every piece of land is part of some community of reference.
After having identified what it considered to be the appropriate community of reference, EPA then applied Venetie’s set-aside and federal superintendence requirements to that community. While HRI’s Section 8 land itself was indisputably neither set aside for Indian use nor federally superintended, EPA reasoned that all of Church Rock Chapter is Indian country because a sufficiently high percentage of the land within its boundaries are set aside for Indian use and federally superin*1143tended. So it is that, by this series of steps, EPA determined that Section 8 “is within the dependent Indian community of the Church Rock Chapter and, thus, is Indian country.” Id. at 13. And so it is that EPA required HRI to file a new UIC permit application with the federal government.
E
Again HRI petitioned this court for review. The company argued, much as it had in HRI /, that Venetie abrogated the Watchman test on which EPA relied in justifying and conducting its threshold “community of reference” inquiry. In HRI’s view, Venetie said that § 1151(b) requires a court to ask only whether “the land in question” is set aside for Indian use and federally superintended, no more.
The panel, considering itself constrained by HRI Fs suggestion that Watchman’s community of reference test survived Ven-etie, rejected HRI’s argument and upheld the agency’s final land status determination. HRI II, 562 F.3d at 1261. Judge Frizzell, sitting by designation, dissented in part. He questioned wRether Section 8 is fairly included within the Church Rock community of reference, given that it is uninhabited and isolated land that the Chapter has deemed “incapable of sustaining a community.” HRI II, 562 F.3d at 1269 (Frizzell, J., concurring in part and dissenting in part). Judge Frizzell also questioned the community of reference test, noting that “[a]s long as a Chapter as a whole satisfies whatever percentage of federal set-aside and supervision a federal court deems necessary, tribal law' may itself define the boundaries of Indian country outside” reserval ions. Id. at 1270-71. Through this application “of our community of reference test,” Judge Frizzell emphasized, “we take an unprecedented step. Never before has non-Indian fee land outside the exterior boundaries of a reservation or Pueblo been held to be a dependent Indian community.” Id. at 1270.
After the panel issued its decision, HRI petitioned for rehearing en bane, asking us to tackle the one issue the panel thought it could not—'whether Watchman’s community of reference test remains an appropriate part of § 1151(b) analysis after Venetie. In support of its petition, the company suggested HRI I had opened a split of authority within this circuit: wRile HRI I suggested that the community of reference survived Venetie, in United States v. Roberts, 185 F.3d 1125 (10th Cir.1999), a previous panel of this court had analyzed a § 1151(b) claim without reference to Watchman’s community of reference analysis, asking only Venetie’s twm questions. See also Frank, 52 P.3d at 408 (noting the same tension in this circuit’s post-Venetie case law7). Though reluctant to protract even further this already aged dispute, in light of the arguably inconsistent guidance offered by different panels of this court we granted HRI’s petition for en banc review.7
*1144II
Before we can address the merits of this dispute, we must first attend to antecedent questions about our subject matter jurisdiction (Section II.A) and standard of review (Section II.B).
A
Federal courts do not wield plenary jurisdiction over every slight or suit. Instead, our authority is restricted in ways small and large by constitutional and statutory design. Because of this, the task of ensuring ourselves of our own subject matter jurisdiction “is not a mere nicety of legal metaphysics,” but essential to the rule of law in “a free society.... The courts, no less than the political branches of government, must respect the limits of their authority.” U.S. Catholic Conference v. Abortion Rights Mobilization, Inc., 487 U.S. 72, 77, 108 S.Ct. 2268, 101 L.Ed.2d 69 (1988); see also In re C & M Properties, L.L.C., 563 F.3d 1156, 1161 (10th Cir.2009).
Before the panel, EPA challenged HRI’s standing under Article III of the Constitution, arguing that its final land status determination imposed no constitutionally cognizable injury on HRI. After all, EPA said, that determination “merely implicates which regulator (NMED or EPA) will enforce UIC regulations,” and does nothing to alter the substantive SDWA threshold HRI or anyone else must clear in order to obtain a UIC permit. HRI II, 562 F.3d at 1258 (emphasis in original). In support of its argument, EPA emphasized New Mexico’s representation that its state UIC permitting process “is more stringent in some respects than the Federal program,” and the absence of any evidence that EPA’s process or permit would involve “more onerous terms than NMEDf’s].” EPA’s Merits Brief at 19-20, HRI II, 562 F.3d 1249 (10th Cir.2009).
Under Article III, federal courts have jurisdiction only to decide “Cases” and “Controversies.” U.S. Const, art. Ill, § 2. One “essential and unchanging part of the case-or-controversy requirement” is the concept that the plaintiff must have “standing,” which in turn requires the presence of “three elements.” Lujan v. Defenders of Wildlife, 504 U.S. 555, 560, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992). A party has standing to pursue a claim in federal court only if: (1) it “suffered an ‘injury in fact’—an invasion of a legally protected interest which is (a) concrete and particularized, and (b) actual or imminent, not conjectural or hypothetical”; (2) that injury is “fairly traceable to the challenged action of the defendant” rather than some third party not before the court; and (3) that injury is likely to be “redressed by a favorable decision.” Id. at 560-61, 112 S.Ct. 2130 (internal quotation marks, alterations, and citations omitted).
Before us, EPA has disputed only the first element—injury in fact.8 The Agency’s challenge, however, cannot succeed. Even if, as EPA would apparently have us assume, its UIC permitting process is no stricter than New Mexico’s, the panel opinion in this case correctly explained that “the outlay of funds necessary to secure” a second UIC permit from EPA, on top of the one HRI has already secured from NMED, amply qualifies as a concrete and particularized, actual and imminent injury. HRI II, 562 F.3d at 1259. As we have previously explained, “the out-of-pocket cost to a business of obeying a *1145new rule of government!,] ... whether or not [there may be] pecuniary loss” associated with the new rule, suffices to establish an “injury in fact:,” Nat’l Collegiate Athletic Ass’n v. Califano, 622 F.2d 1882, 1386 (10th Cir.1980). EPA’s final land status determination requires HRI to undergo the UIC permit process for a second time—this time with the federal authorities—before it can mine its property. There is nothing hypothetical or conjectural about that, or about the fact that such a “redo” would impose on HRI some additional administrative costs. Maybe those costs wouldn’t break HRFs bank, but that’s hardly required to constitute a constitutionally cognizable injury. See id. at 1389 (“Certainly the cost of obeying the regulations constitutes injury.”).9
B
Assured of HRI’s constitutional standing to bring this appeal, before reaching the merits it remains to ask whether and to what degree we are statutorily empowered to review EPA’s decision.
The SDWA authorizes us to review “final actions” taken by EPA in its administration of the statute. 42 U.S.C. § 300j-7(a)(2). The parties before us agree that EPA’s final land status determination qualifies as such a final action, and we can see no basis on which we might disagree. But though the SDWA grants us the power to review EPA’s action in this case, it does not tell us what standard of review we should use in doing so. When the legislation at hand doesn’t supply a standard of review for us to apply, the Administrative Procedure Act (“APA”) provides the default, filling the gap and telling us, among other things, that we “shall ... hold unlawful and set aside agency action, findings, and conclusions, found to be arbitrary and capricious, an abuse of direction, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A).
It is this last phrase—“otherwise not in accordance with law”—most directly at issue here. See FCC v. NextWave Pers. Commc’ns Inc., 537 U.S. 293, 300, 123 S.Ct. 832, 154 L.Ed.2d 863 (2003) (“The Administrative Procedure Act requires federal courts to set aside agency action that is ‘not in accordance with law’—which means, of course, any lawr, and not merely those laws that the agency itself is charged with administering.” (citation omitted)). EPA’s final land status determination represents the Agency’s interpretation of its earlier regulations affording NMED primary authority to regulate UIC wells in New Mexico “except on Indian lands,” 40 C.F.R. §§ 147.1600-147.1601, and then defining “Indian lands” to mean “Indian country” as that term is used in 18 U.S.C. § 1151, id, § 144.3. It is by dint of these regulatory choices that, in the end, EPA faced the purely legal task of interpreting § 1151 in its final land status determination. And it is by dint of this that we must ask whether EPA’s interpretation of § 1151 is or is not in accordance with the statute.
Of course, courts afford considerable deference to agencies interpreting ambiguities in statutes that Congress has delegated to their care, see Chevron, U.S.A., Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), including statutory ambiguities af*1146fecting the agency’s jurisdiction, see Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 844, 106 S.Ct. 3245, 92 L.Ed.2d 675 (1986); see also Miss. Power & Light Co. v. Mississippi ex rel. Moore, 487 U.S. 354, 381, 108 S.Ct. 2428, 101 L.Ed.2d 322 (1988) (Sealia, J., concurring in the judgment); Teamsters Local Union No. 523 v. NLRB, 590 F.3d 849, 850-51 (10th Cir.2009). Courts do not, however, afford the same deference to an agency’s interpretation of a statute lying outside the compass of its particular expertise and special charge to administer. See Metro. Stevedore Co. v. Rambo, 521 U.S. 121, 137 n. 9, 117 S.Ct. 1953, 138 L.Ed.2d 327 (1997) (no deference given to agency interpretation of statute, in part, because the agency was not “charged with administering” it); Adams Fruit Co. v. Barrett, 494 U.S. 638, 649, 110 S.Ct. 1384, 108 L.Ed.2d 585 (1990) (“A precondition to deference under Chevron is a congressional delegation of administrative authority.”); see also Crandon v. United States, 494 U.S. 152, 174, 110 S.Ct. 997, 108 L.Ed.2d 132 (1990) (Sealia, J., concurring in the judgment) (“The law in question, a criminal statute, is not administered by any agency but by the courts.”).
Section 1151 quite clearly does not fall within EPA’s particular expertise or charge to administer. It is not a statute specially involving environmental regulation, but one all and only about the geographic parameters of federal and tribal criminal prosecutorial authority. Even so, we need not decide whether EPA’s interpretation of the statute is entitled to deference because, throughout the proceedings before the panel and now the en banc court, EPA itself hasn’t claimed any entitlement to deference. In these circumstances, when the agency doesn’t ask for deference to its statutory interpretation, “we need not resolve the ... issues regarding deference which would be lurking in other circumstances,” Estate of Cowart v. Nicklos Drilling Co., 505 U.S. 469, 477, 112 S.Ct. 2589, 120 L.Ed.2d 379 (1992), and so may proceed to review EPA’s interpretation of § 1151(b), as the Agency would have us and the panel did, de novo.10
*1147ill
With this much resolved, we finally reach the merits. The dispute here is a purely legal one: Does the Watchman community of reference test remain viable following the Supreme Court’s decision in Venetie? EPA argues it does; HRI says it does not.
EPA reads § 1151(b)’s three operative words, “dependent Indian communities,” to require a two-step, multi-factor balancing test. First, and following Watchman, EPA says we must identify an appropriate “community of reference” by weighing three factors in balance: (1) the geographic definition of the proposed community; (2) the status of the area in question as a community; and (3) the community in the context of the surrounding area. Within the second of those factors, and again following Watchman,, EPA says we must inquire whether there is “cohesiveness that can be manifested either by economic pursuits in the area, common interests, or needs of the inhabitants as supplied by that locality,” and consider “whether the community is more than an economic pursuit, and whether it qualifies as a mini-society consisting of personal residences and an infrastructure potentially including religious and cultural institutions, schools, emergency services, public utilities, groceries, shops, restaurants, and the other needs, necessities, and wants of modern life.” Land Status Determination at 5 (internal alterations and quotation marks omitted). And within the third factor, EPA says, we must also ask “which government or governments provide the infrastructure and essential services for the community.” Id. at 6. In EPA’s view, none of the analysis to this point is affected by Venetie.
Having balanced all these competing considerations to identify some “community of reference,” EPA moves to the second Watchman step, which it does view as modified by Venetie. Before Venetie, the Watchman test sought to determine whether a community of reference qualified as a dependent Indian community by considering four factors: (1) whether and to what degree the United States has retained title to the lands in the community, (2) the nature of the area and the relationship of the inhabitants in the area to Indian tribes and to the federal government, (3) whether there is “an element of cohesiveness ... manifested either by economic pursuits in the area, common interests, or needs of the inhabitants as supplied by that locality”; and (4) “whether such lands have been set apart for the use, occupancy and protection of dependent Indian peoples.” Watchman, 52 F.3d at 1545. EPA suggests that, after Venetie, the second and third of these factors are no longer operative, and that the first and fourth essentially track Venetie’s set-aside and *1148federal superintendence requirements. Accordingly, EPA submits, if some sufficiently high (though unspecified) percentage of the community of reference is set aside for Indians and federally superintended, then all the land inside that community should be treated as Indian country regardless whether any particular tract is set aside and superintended. Weighing all the foregoing considerations, EPA argues that HRI’s Section 8 land, though itself neither set aside nor superintended for Indians, is Indian country nonetheless because it lies within the Church Rock Chapter community of reference and, by the Agency’s calculation, enough of that land is set aside and superintended for Indians.
For its part, HRI submits that Venetie leaves no room for EPA’s reading of § 1151(b). After Venetie, HRI urges, the appropriate § 1151(b) test asks only two straightforward questions: (1) Has Congress explicitly set aside the land in question for Indian use? (2) Does the federal government superintend the land in question? Unless the answer to both questions is “yes,” the land in question is not within a dependent Indian community. HRI stresses that Venetie expressly rejected the Ninth Circuit’s § 1151(b) test, a test that was consciously patterned on this court’s pre-Venetie jurisprudence and employed the same factors found in the Watchman, community of reference test. In addition, HRI submits that the fact-intensive and multi-factored community of reference balancing test leaves the scope of federal criminal jurisdiction impermissi-bly uncertain and unpredictable, contravening the Supreme Court’s direction in Venetie and its repeated admonitions elsewhere that criminal statutes merit more concrete and precise constructions.
We are constrained to agree with HRI. We hold that Watchman's community of reference test did not survive Venetie and that dependent Indian communities under § 1151(b) consist only of lands explicitly set aside for Indian use by Congress (or its designee) and federally superintended. We reach this result in light of Venetie’s exposition of the statute’s plain meaning (Section III.A), the statute’s history (Section III.B), and the statute’s structure (Section III.C), and in doing so we bring the law of this circuit in line with the recent decisions of our sister circuits (Section III.D).
A
In Venetie, the Supreme Court explained that “dependent Indian communities” under § 1151(b) embrace “a limited category of Indian lands that are neither reservations nor allotments” encompassed by § 1151(a) and (c), respectively. 522 U.S. at 527, 118 S.Ct. 948. The Court then identified two necessary “requirements” for lands falling into § 1151(b)’s “dependent Indian communities” category, explaining that, much like reservations or allotments, “first, they must have been set aside by the Federal Government for the use of the Indians as Indian land; second, they must be under federal superintendence.” Id.
What does it mean for the federal government to set aside land for Indian use and to superintend it? The Court noted that the set-aside requirement means that there must be “some explicit action by Congress (or the Executive, acting under delegated authority) ... to create or to recognize” the “land in question” as part of a federally recognized and dependent Indian community. 522 U.S. at 531 n. 6, 118 S.Ct. 948. Through an Act of Congress or some equally explicit executive action, then, the federal government must identify the land as “set apart for the use of the Indians as such,” Id. at 529, 118 S.Ct. 948 (internal quotation marks omitted) *1149(emphasis in original). So, for example, land simply conveyed by Congress to individual Indians or tribes that they are then “free to use ... for non-Indian purposes” or sell as they wish does not qualify. Id. at 533, 118 S.Ct. 948. While groups of Indians may very well live on such lands in socially and politically discrete communities, they do not live in “Indian country” because the land in question has not- been explicitly set aside by Congress for use as a “dependent Indian community.” The superintendence requirement means that the federal government currently must be “actively confronting] the lands in question, effectively acting as a guardian for the Indians.” Id. This requirement, too, necessarily excludes lands that the government has conveyed without restriction to Indians or others because such lands do not implicate any sense of “guardian[ship],” “wardshipt,] or trusteeship.” Id. (internal quotation marks omitted).
The set-aside and superintendence requirements, the Court explained, derive from the statute’s plain language—“dependent Indian communities.” The set-aside requirement “ensures that the land in question is occupied by an ‘Indian community.’” Id. at 531, 118 S.Ct. 948. That is, the boundaries of the Indian community are demarcated by and delimited to those lands that are explicitly set aside by legislation or executive action for Indian use. The federal superintendence requirement “guarantees that the Indian community is sufficiently ‘dependent’ on the Federal Government that the Federal Government and the Indians involved, rather than the States, are to exercise primary jurisdiction over the land in question.” Id.
In our case, it is undisputed that HRI’s Section 8 land hasn’t been explicitly set aside by Congress (or the Executive) for Indian use since the brief period when it was appended to the Navajo Reservation nearly a century ago. See supra Section I .A. It is likewise undisputed that the land isn’t under federal superintendence, and hasn’t been since the government sold it in 1970. See id. McKinley County and the State of New Mexico provide all essential public services to HRI’s Section 8 land, including roads, law enforcement, and emergency and school services. The Navajo Church Rock Chapter recognizes that private lands within its boundaries, like HRI’s, are subject to state jurisdiction and control. And state authorities have long assessed tax on HRI’s property. Under Venetie's interpretation of § 1151(b), it would seem unavoidable that the land in question is not Indian country.
Of course, EPA seeks to avoid just this conclusion by expanding the focus from HRI’s particular tract to a wider “community of reference.” According to the Agency, only after ascertaining some appropriate community of reference, by balancing various social, political, and geographic factors, should one then turn to the questions prescribed by the Supreme Court in Vene-tie, asking whether some significant percentage of the community of reference is set aside for Indian use and federally superintended.
This misreads Venetie. The Supreme Court did not direct its set-aside and superintendence inquiries toward some abstract “community of reference.” Instead, the Court told us (repeatedly) to ask whether the “land in question” is explicitly set aside for Indian use and federally superintended. See, e.g., Venetie, 522 U.S. at 530, 531, 533, 118 S.Ct. 948. Before us, the only land in question in EPA’s challenged final land status determination is HRI’s segment of Section 8. And so it is only that land that is subject to, and the focus of, Venetie’s set-aside and superintendence requirements.
Tellingly, EPA adopted just this approach to § 1151(b) before Hill I, when it considered the Navajo Nation’s request for *1150SDWA authority over private lands in the checkerboard area. See supra Section I.B. EPA rejected the Navajo Nation’s request, explaining that “[b'Jefore it could determine if a parcel of land is part of a dependent Indian community (and therefore is Indian country), the Agency would need more information about that particular parcel of land.” R. 13b at 238. EPA itself, then, once focused its § 1151(b) analysis on the “particular parcel of land” in question. And it seemingly came to eschew this approach in favor of the community of reference test only after the panel in HR] I suggested in dicta that it should.11
Of course, EPA now tells us that focusing narrowly on the status of the “land in question” fails to give full vent to the statutory term “communities.” The Agency stresses that § 1151(b) speaks of “dependent Indian communities,” not parcels of land. And the Agency submits that, while Venetie defined the statutory terms “dependent” and “Indian,” it did not consider “community,” the final statutory term. The dissents offer this same critique. See, e.g., Principal Dissent at 1175-76; Separate Dissent at 1183.
This, too, is in error. The Venetie Court did address the term “community,” and did so in light of the statutory terms modifying it. The Court expressly held that “dependent Indian communities” are composed of those lands Congress (or the Executive, by delegation) has explicitly set aside and superintended for Indian use. As the Court clearly explained, “[t]he federal set-aside requirement ensures that the land in question is occupied by an ‘Indian community.’ ” Venetie, 522 U.S. at 531, 118 S.Ct. 948 (emphasis added). And the federal superintendence requirement, the Court emphasized, “guarantees that the Indian community is,” in turn, “sufficiently ‘dependent’ on the Federal Government that” it should be subject primarily to federal, not state, jurisdiction. Id. (emphasis added).
In adopting the approach it did, moreover, the Court rejected the idea that the boundaries of a federally dependent Indian community should be determined by a sort of judicially administered census study of the nature of “the Indian tribe inhabiting” the area. Id. at 530 n. 5, 118 S.Ct. 948. The right question, the Court held, is instead whether Congress has taken some action to designate and maintain the land in question for Indian use. It is Congress’s action alone that demarcates the boundaries of a dependent Indian community. Id.; see also United States v. McGowan, 302 U.S. 535, 538, 58 S.Ct. 286, 82 L.Ed. 410 (1938) (“Congress alone has the right to determine the manner in which this country’s guardianship over the Indians shall be carried out.”); cf. Rosebud Sioux Tribe v. Kneip, 430 U.S. 584, 586, 97 S.Ct. 1361, 51 L.Ed.2d 660 (1977) (“In determining whether ... Reservation boundaries were subsequently diminished[,] ... [t]he underlying premise is that congressional intent will control.”).
In this way, the Court explained, the federal set-aside and superintendence requirements respect and give meaning to an important feature of our constitutional order—namely, “the fact that because Congress has plenary power over Indian affairs, see U.S. Const, art. I, § 8, cl. 3, some explicit action by Congress (or the Executive, acting under delegated authority) must be taken to create or to recognize *1151Indian country.” Venetie, 522 U.S. at 531 n. 6, 118 S.Ct. 948. While social or political communities of Indians can exist anywhere in our country, and surely do in a great many places outside Indian country, federally “dependent Indian communities” exist only where and when Congress has said so.
Simply put, Venetie held that Congress—not the courts, not the states, not the Indian tribes—gets to say what land is Indian country subject to federal jurisdiction. It is long settled that Congress does so by declaring land to be part of a reservation, or by authorizing its distribution as Indian allotments. And so it is the case that Congress must take some equally “explicit action ... to create or to recognize” dependent Indian communities. Id. When seeking to identify a § 1151(b) “dependent Indian community,” we must ask whether Congress has explicitly set aside the “land in question” for Indian use and put it under federal superintendence. If Congress hasn’t declared the land set aside for the establishment of a federally dependent Indian community, we are powerless to do so ourselves.12
Applying these principles in Venetie, the Supreme Court held that none of the 1.8 million acres of land granted by Congress in fee to the Neets’aii Gwieh’in Indians in Alaska constituted a § 1151(b) “dependent Indian community.” Id. at 523-24, 118 S.Ct. 948. It was undisputed that these lands were home to what EPA (and the dissents) would consider to be distinct Indian “communities of reference.” After all, robust and well-defined Indian villages occupied the land. See Venetie, 101 F.3d at 1300. Despite such social, political, and geographic affinities, the Supreme Court held that a § 1151(b) “dependent Indian community” didn’t exist on any portion whatsoever of the 1.8 million acres of tribal lands because none of that land was con-gressionally set aside or superintended for Indian use. And that was because, when it passed the Alaska Native Claims Settlement Act (“ANCSA”), Congress deliberately revoked all existing Indian reservations in Alaska “set asida by legislation or by Executive or Secretarial Order for Native me ” and granted tribal corporations fee title to those lands. Venetie, 522 U.S. at 532, 118 S.Ct. 948 (quoting 43 U.S.C. § 1618(a)). In taking this step, Congress contemplated that the tribes could sell any or all of the lands to non-Indians, and could make use of them as they wished for any purpose, Indian or non-Indian, without federal approval. “In no clearer fashion,” the Supreme Court explained, “could Congress have departed from its traditional practice of setting aside Indian lands.” Id. Where there is no eongressionally approved set-aside for Indian use, and no federal superintendence, the Court underscored, there can be no dependent Indian community for purposes of § 1151(b).13
*1152The Watchman community of reference test is inconsistent with this direction. Instead of asking whether Congress (or the Executive, through delegated authority) has taken some explicit action to set aside the land in question and whether the federal government superintends that land, it first conducts a threshold inquiry involving multiple competing factors that have no basis in Venetie or the text or history of § 1151(b).14 Only after conducting this inquiry does the community of reference test turn to Venetie's prescribed set-aside and superintendence questions. And even then the community of reference test asks only about the degree and extent to which some other, larger area is set aside and superintended. In this way, EPA’s test effectively “redueles] the federal set-aside and superintendence requirements to mere considerations”—something Venetie expressly warned us against. 522 U.S. at 531 n. 7, 118 S.Ct. 948.
In this same way, the community of reference test also disregards and regularly overrides Congress’s plenary authority in charting the extent of Indian country. Applying the test here, EPA held HRI’s land to be Indian country even though Congress has not explicitly set aside the land for Indian use and the most recent federal action with respect to this land was an executive order in 1911 removing it from Indian country. See supra Section I.A. It seems to us that, just as in Vene-tie, there could be “no clearer fashion” in which the federal government could have extinguished the Indian country status of HRI’s land. Venetie, 522 U.S. at 532, 118 S.Ct. 948.
By disregarding Congress’s authority, the community of reference test contemplates the possibility that even land never set aside by Congress for Indians can become Indian country simply because of its proximity to other lands that are federally set aside and superintended. This is so despite the fact that neither EPA nor the principal dissent has pointed us to a single case in the history of the Supreme Court or this court reaching such a result. Consider what happens when a tribe unilaterally redefines its borders to take in just a *1153bit more land (like, say, the state park at the center of the Church Rock Chapter, or strips of private land currently outside but along the edges of the Chapter). See Appendix. Assuming the social and political characteristics of this expanded community of reference remain more or less constant, land that once wasn’t Indian country becomes Indian country by tribal preference or judicial decree rather than congressional action.
Neither does the community of reference test’s disregard for Congress’s authority work only to expand Indian communities. It can operate just as well the other way around—and would surely do so increasingly as time wears on and non-Indian communities encroach on Indian lands. What would happen, for example, if Gallup grows and dilutes adjacent Indian populations, so that the Chapter might no longer be said to constitute its own independent “mini-society,” but only part of a greater Gallup “community of reference”? Presuming that this new greater Gallup community of reference wouldn’t include a sufficient percentage of set-aside and superintended land, no land within its limits would constitute “Indian country” under § 1151(b). Under EPA’s approach, land expressly set aside for Indian use and superintended by the federal government could and would lose its status as Indian country whenever social and political boundaries shift—and all this would happen even in the face of express and contrary congressional direction.15 Venetie ties the jurisdictional determination to the proper hitch: the will of Congress. The same cannot be said of the community of reference test.16
Any remaining question about the vitality of Watchman’s community of reference test is answered by the fact that it would require us to revive the same factors the Supreme Court rejected in Venetie. Before the case reached the Supreme Court, the Ninth Circuit looked to this circuit’s *1154then-governing precedent for guidance in determining whether the land in question before it constituted Indian country under § 1151(b). In fact, the Ninth Circuit applied a six-factor test that it called “virtually identical” to the second step of our Watchman analysis. Venetie, 101 F.3d at 1294; see id. at 1291-92. The Supreme Court, of course, proceeded to reject this “textured” balancing test. Venetie, 522 U.S. at 531 n. 7, 118 S.Ct. 948. In particular, the Court criticized reliance on three factors—“the nature of the area,” “the relationship of the area inhabitants to Indian tribes and the federal government,” and “the degree of cohesiveness of the area inhabitants”—as “extremely far removed from” the more concrete set-aside and superintendence requirements. Id.
As we have noted previously, many of the same factors appear in both steps of the Watchman analysis. See supra Section I.C. And the three specific factors the Supreme Court held impermissible in Venetie are among these reappearing factors. Put more pointedly, the factors Ven-etie expressly rejected are found not just in Watchman’s second step; they also form the backbone of Watchman’s first step: defining the community of reference. For example, considering the “status of the area in question,” as EPA did in its community of reference analysis, Land Status Determination at 5, is no different from looking at the “nature of the area,” a consideration the Supreme Court rejected in Venetie. 522 U.S. at 531 n. 7, 118 S.Ct. 948. Searching for “the existence of an element of cohesiveness,” as EPA did, Land Status Determination at 5, is indistinguishable from seeking to discern “the degree of cohesiveness of the area inhabitants,” which the Supreme Court held impermissible. Venetie, 522 U.S. at 531 n. 7, 118 S.Ct. 948. And focusing “on which government or governments provide the infrastructure and essential services for the community,” as EPA would have us do, Land Status Determination at 6, necessarily requires consideration of “the relationship of the area inhabitants to Indian tribes and the federal government,” again something Venetie ruled out-of-bounds. 522 U.S. at 531 n. 7, 118 S.Ct. 948. In EPA’s application of Watchman’s community of reference test, then, the three factors Venetie discredited reemerge, like a phoenix, from their own demise. And in some ways, the community of reference test favored by EPA is even more problematic than the Ninth Circuit’s approach. At least in the Ninth Circuit’s test the three factors rejected by the Supreme Court were balanced in fair competition against the set-aside and superintendence requirements. In EPA’s application of Watchman’s community of reference test, the rejected factors actually receive a promotion to an outcome-determinative threshold inquiry. The Agency offers us no credible explanation how we might lawfully revive, and then elevate, the very same factors the Supreme Court has expressly rejected.17
*1155B
The Venetie Court didn’t create its § 1151(b) test out of whole cloth, but consciously grounded its test in the language and history of the statute. On the latter score, the Court noted that subsection (b)’s use of the phrase “dependent Indian communities” was designed to codify language from the Supreme Court’s decisions in United States v. Sandoval, 231 U.S. 28, 34 S.Ct. 1, 58 L.Ed. 107 (1913), and United States v. McGowan, 302 U.S. 535, 58 S.Ct. 286, 82 L.Ed. 410 (1938). See Venetie, 522 U.S. at 528-31, 118 S.Ct. 948; see also 18 U.S.C. § 1151 History; Ancillary Laws and Directive Notes (“Definition is based on latest construction of the term by the United States Supreme Court in U.S. v. McGowan following U.S. v. Sandoval.” (citations omitted)). For that reason, Vene-tie examined and relied on these earlier decisions to give content and meaning to subsection (b), recognizing that when a legal concept like this “is obviously transplanted from another legal source, whether the common law or other legislation, it brings the old soil with it.” Felix Frankfurter, Some Reflections on the Reading of Statutes, 47 Colum. L.Rev. 527, 537 (1947).
An examination of these precedents, the soil in which § 1151(b) was germinated, confirms the implausibility of a community of reference test. Sandoval dealt with Congress’s constitutional authority to include pueblos within “Indian country.” The defendant in that case was prosecuted for bringing alcohol into the Santa Clara Pueblo. He was charged under a federal statute that “ma[de] it a punishable offense to introduce intoxicating liquor into the Indian country,” and another statute specifically providing that all of the Santa Clara Pueblo constituted “Indian country.” Sandoval, 231 U.S. at 36-37, 34 S.Ct. 1. Faced with these statutes, the Supreme Court was asked to decide whether Congress had the power to deem the pueblos part of Indian country. The Court answered in the affirmative, reasoning that the Indian Commerce Clause of the Constitution and “an unbroken current of judicial decisions have attributed to the United States ... the power and the duty of exercising a fostering care and protection over all dependent Indian communities within its borders.” Id. at 46, 34 S.Ct. 1 (emphasis added). While not a formal Indian reservation or allotment that might fall into today’s § 1151(a) and (c) categories, the Court noted that Congress had “recognized the Pueblos’ titles to their ancestral lands by statute,” executive orders had “reserved additional lands ‘for the [Pueblos’] use and occupancy,’ ” and “Congress had enacted legislation ... ‘in the exercise of the Government’s guardianship over th[e] [Indian] tribes and their affairs’ ... including federal restrictions on the land’s alienation.” Venetie, 522 U.S. at 528, 118 S.Ct. 948 (quoting Sandoval, 231 U.S. at 39, 48, 34 S.Ct. 1) (emphasis added). In this way, the Court held, Congress had taken deliberate and independent actions to set aside the land in question and guarantee its federal superintendence, thereby rendering it a federally dependent Indian community.
McGowan is to the same effect. The case arose when the United States sought the forfeiture of two automobiles used to introduce alcohol into the Reno Indian Colony, which consisted of a single tract of land bought and owned by the federal government as a “permanent settlement” for “needy Indians scattered over the State of Nevada.” McGowan, 302 U.S. at 537, 58 S.Ct. 286. The question presented to the Court was whether the Colony constituted “Indian country” under the same criminal statute at issue in Sandoval The Court concluded that the Colony, while neither a reservation nor an allotment, constituted a “dependent Indian commun-it[y],” id. at 538, 58 S.Ct. 286, because all *1156of its land had been “validly set apart for the use of the Indians” and remained “under the superintendence of the government,” id. at 539, 58 S.Ct. 286. As Venetie explained, the McGowan Court stressed that “like Indian reservations generally, the colony had been ‘validly set apart’ ” because the “Federal Government had created the colony by purchasing the land with ‘funds appropriated by Congress,’ ” and it was federally superintended because “the Federal Government held the colony’s land in trust for the benefit of the Indians residing there.” Venetie, 522 U.S. at 529, 118 S.Ct. 948 (quoting McGowan, 302 U.S. at 537, 539, 58 S.Ct. 286).18
Nothing in Sandoval or McGowan suggests that the metes and bounds of “dependent Indian communities” should be determined by a court’s perceptions about local social, political, or geographic affinities. Rather, in both cases, the Supreme Court identified “dependent Indian communities” based on congressional intent, as expressed in independent statutory declarations. Both decisions emphasized that “the questions whether, to what extent, and for what time [Indians] shall be recognized and dealt with as dependent tribes requiring the guardianship and protection of the United States are to be determined by Congress, and not by the courts.” Sandoval, 231 U.S. at 46, 34 S.Ct. 1 (emphasis added); see also McGowan, 302 U.S. at 538 & n. 9, 58 S.Ct. 286 (citing Sandoval, 231 U.S. at 46, 34 S.Ct. 1). And both explained Congress must express its intent to make land a “dependent Indian community” by explicitly setting aside the land for Indian use and authorizing the federal government to superintend the land for that purpose. Cf. Felix S. Cohen’s Handbook of Federal Indian Law, Ch. 1, § 3 at 7 (Robert L. Bennett & Frederick M. Hart eds., 1971) (reprinting original 1942 edition) (hereinafter “Cohen (1942)”) (summarizing Sandoval as explaining “that the term Indian country as applied to the Pueblos means any lands occupied by ‘distinctly Indian communities’ recognized and treated by the Government as ‘dependent communities’ entitled to its protection”).
Venetie, thus, is nothing new under the sun. It simply reaffirms what the Supreme Court has always held to be the case: the dependent Indian community inquiry centers on whether Congress (or the Executive, by delegation) has taken some explicit action to set aside the land in question for Indian use and whether the land remains federally superintended. As always, we do well to follow the Supreme Court’s guidance—most especially when that guidance has received the blessing of Congress through codification. Yet, EPA (like the dissents) makes no attempt to grapple with any of § 1151(b)’s considerable history and would instead have us view it in a vacuum, extirpated from the soil in which it grew.
Though not dispositive to our analysis, it also merits noting that the historic meaning of the statute recently received further congressional confirmation. Seeking to salve a dispute over the status of privately held lands within New Mexico pueblos, Congress passed the Indian Pueblo Land *1157Act Amendments of 2005, Pub.L. No. 109-133, 119 Stat. 2573, codified at 25 U.S.C. § 331 Note. The law provides for federal and tribal criminal jurisdiction over offenses committed involving Indians “anywhere within the exterior boundaries of any grant from a prior sovereign, as confirmed by Congress or the Court of Private Land Claims[, an Article I court established by Congress,] to a Pueblo Indian tribe of New Mexico.” Id. If the community of reference test is correct, this law is surplusage, a waste of everyone’s time and effort in writing, passing, and signing. Under the community of reference test, if and when a pueblo qualifies as an “appropriate community” (and surely that would be most of the time), everything in it, including privately held lands, is automatically subject to federal or tribal jurisdiction under § 1151(b). No additional law was or is or will ever be needed. The fact that Congress thought otherwise is one more (not inconsiderable) piece of evidence against the community of reference test— as is Congress’s continued focus on boundaries “confirmed by Congress” or its desig-nee, rather than on boundaries set by litigation over the latest social statistics.
C
The structure of § 1151 further confirms our understanding. The notion that some explicit, congressionally approved action is required to create “Indian country” is hardly foreign to the statute’s design. Subsection (a) classifies reservations as Indian country, and reservations are traditionally created by and delineated according to boundaries Congress has set or sanctioned. Subsection (c) recognizes allotments as Indian country, and allotments, too, are a product of congressional action, either directly or through delegation. Subsection (b) serves as something of a catch-all provision, encompassing “a limited category of Indian lands that are neither reservations nor allotments” but that are still explicitly set aside for Indians by congressional mandate and superintended by the federal government. Venetie, 522 U.S. at 527, 118 S.Ct. 948. In all three of § 1151’s subsections, then, the creation of Indian country hinges on some explicit action by Congress (or the Executive, acting under delegated authority). Without some textual indication to think otherwise, it would be anomalous to think that subsection (b) alone was designed to permit the judiciary to commission or decommission Indian country by means of an atextual, multi-factor balancing test.
EPA attempts to overcome this by arguing that § 1151 as a whole is imbued with an “anti-checkerboard” purpose. The Agency worries that looking to the eon-gressionally approved status of each parcel of land will result in a patchwork, with some lands falling under federal jurisdiction while neighboring ones are subject to state jurisdiction. In support of this argument, EPA cites Seymour v. Superintendent of Washington State Penitentiary, 368 U.S. 351, 358, 82 S.Ct. 424, 7 L.Ed.2d 346 (1962), and the 2005 edition of Cohen’s Handbook of Federal Indian Law. See also Principal Dissent at 1172-73; Separate Dissent at 1183-84. But this argument overstates the statutory case against checkerboarding. In Seymour, the Court simply observed the obvious: subsection (a), by its express terms, includes within the definition of Indian country all lands within the congressionally prescribed boundaries of a reservation, including private fee lands. The 2005 edition of Cohen’s Handbook adds nothing new to the mix, citing only Seymour. Meanwhile, the 1982 edition of the Handbook on which the principal dissent relies expressly acknowledges that § 1151 “does not eliminate all forms of ‘checkerboarding.’ ” See Felix S. Cohen’s Handbook of Federal Indian Law, Ch. 1, § D3c, at 39 & n.105 (Rennard Strickland et al. eds.1982) (hereinafter “Cohen (1982)”).
*1158By indicating in subsection (a) that all lands within reservations are Indian country, Congress undoubtedly did something to mitigate the potential for checkerboard jurisdiction, as EPA observes. But Congress did not pursue this goal single-mind-edly throughout § 1151 at the expense of all other purposes. Indeed, as this court has previously explained, “Congress has authorized checkerboard jurisdiction under its definition of Indian eountry in 18 U.S.C. § 1151. Although subsection 1151(a) clarifies that checkerboard titles within an existing reservation do not affect the status of an Indian reservation as reservation, subsections 1151(b) and (c) allow checkerboard jurisdiction outside reservation boundaries.” Yazzie, 909 F.2d at 1421-22 (emphasis omitted). In this respect, then, § 1151 is just one more example of the fact that “[n]o legislation pursues its purposes at all costs” without consideration of competing goals and concerns. Pension Benefit Guar. Corp. v. LTV Corp., 496 U.S. 633, 646, 110 S.Ct. 2668, 110 L.Ed.2d 579 (1990) (quoting Rodriguez v. United States, 480 U.S. 522, 525-26, 107 S.Ct. 1391, 94 L.Ed.2d 533 (1987) (per curiam)); see also John F. Manning, Textualism, and the Equity of the Statute, 101 Colum. L.Rev. 1, 18 (2001) (“Because statutory details may reflect only what competing groups could agree upon, legislation cannot be expected to pursue its purposes to their logical ends; accordingly, departing from a precise statutory text may do no more than disturb a carefully wrought legislative compromise.”).19
In truth, Congress has authorized some degree of checkerboard jurisdiction in all three § 1151 categories of Indian country. As EPA stresses, subsection (a) evinces the strongest effort to minimize checker-boarding by declaring all lands within Indian reservations, including private fee lands, to be Indian eountry. But even this rule has its exceptions. So, for example, in the Indian liquor laws, Congress has excluded from Indian country any private “fee-patented lands in non-Indian communities,” even if they are located within reservations. 18 U.S.C. §§ 1154(c), 1156; see also United States v. Mazurie, 419 U.S. 544, 95 S.Ct. 710, 42 L.Ed.2d 706 (1975).20
*1159Checkerboarding is also an unavoidable byproduct of subsection (c), as Yazzie and Cohen’s Handbook note. See Yazzie, 909 F.2d at 1421-22; Cohen (1982) Ch. 1, § D3c, at 39 n.105 (“[A]n irregular arrangement of Indian country can exist through the inclusion of off-reservation allotments in its definition”). By definition, allotments outside reservations are stray plots, pieces of land, parcels. Set aside and superintended for Indians, they are Indian country even while neighboring territory is not. The “checkerboard region” of New Mexico bears witness to all this, with Indian allotments lying cheek by jowl with lands owned by private persons or the state or the federal government. See Appendix. To be sure, Congress sought to mitigate, to a degree, the checkerboarding created by the allotment system by extending federal jurisdiction over “all Indian allotments, the Indian titles to w’hieh have not been extinguished, including rights-of-way running through the same.” 18 U.S.C. § 1151(c) (emphasis added). But this hardly eliminates the checkerboard jurisdiction that allotments inevitably create. Indeed, nothing wre might do today will ameliorate the checkerboard nature of the area in wfhieh HRI’s land lies. The allotment system of subsection (c) bears the lion’s share of blame for the area’s jurisdictional patchwork, and no interpretation of subsection (b) will change that unavoidable fact.
While no interpretation of § 1151(b) can eliminate the area’s checkerboarded character, Venetie’s set-aside and superintendence requirements at least ensure that the boundaries of dependent Indian communities will be precisely and predictably defined. Under the community of reference approach favored by EPA, these boundaries wmuld be defined in a more convoluted and less predictable manner. And even after a “community of reference” is defined, its boundaries would likely be irregular, jigsawed, and themselves often hard to discern, as the boundaries of many communities are.
One glance at a map of the Church Rock Chapter confirms as much. See Appendix. The Chapter includes within its boundaries a narrow jog across Interstate 40 and a small chunk of land south of the highway. And there it includes a long strip of private fee lands along the Chapter’s southeastern boundary, yet (without explanation) not other immediately adjacent private fee lands. At the same time, in the center of the Chapter lies Red Rock State Park, wdiich the Chapter excludes from its boundaries, creating a sort of doughnut hole (and raising the perplexing question: to wrhat—noncontiguous—“com-munity of reference” might the park belong?). If that weren’t enough, and seemingly at odds with the exclusion of this particular piece of state-owned property, the Chapter purports (again without ex*1160planation) to include other state-owned lands within its boundaries. Far from curing a checkerboard problem, the upshot of the community of reference test is to invite an odd sort of parquetry all its own.21
Worse still, the community of reference test creates multiple, unpredictable, and shifting checkerboards—and does so in the context of a criminal statute, in tension with the “basic principle” of due process “that a criminal statute must give fan-warning of the conduct that it makes a crime.” Bouie v. City of Columbia, 378 U.S. 347, 350, 84 S.Ct. 1697, 12 L.Ed.2d 894 (1964); cf. Flores-Figueroa v. United States, — U.S. -, 129 S.Ct. 1886, 1895, 173 L.Ed.2d 853 (2009) (Scalia, J., concurring in part and concurring in the judgment) (calling expansions of criminal liability based on factors outside the statute’s text “not unlike the practice of Caligula, who reportedly wrote his laws in a very small character, and hung them up upon high pillars, the more effectually to ensnare the people” (internal quotation marks omitted)).
The Supreme Court has repeatedly warned us against the “judicial expansion of narrow and precise statutory language” in the criminal context because “a deprivation of the right of fair warning can result not only from vague statutory language but also from an unforeseeable and retroactive judicial expansion” of statutes. Bouie, 378 U.S. at 352, 84 S.Ct. 1697.22 This is a principle we cannot help but bear in mind even here, in a civil case. It remains, after all, a criminal statute we are expounding. And this is no mere “technical[ity]” or “hypothetical.” Principal Dissent at 1170-71; Separate Dissent at 1183-84. Though EPA has chosen, for now at least, to link its SDWA permitting authority to the scope of a criminal statute, it may not have to live with this regulatory decision forever. Our interpretation of the statute, meanwhile, will apply to all criminal cases arising under § 1151(b) and cannot so easily be discarded.23
*1161Venetie supplies a test that is consistent with the statute’s meaning, history, and structure; and asking whether a piece of land has been explicitly set aside for Indian use and is federally superintended is a reasonably precise and clear task. The community of reference test, by contrast, is anything but these things. Consider the couple who pops a bottle of champagne to celebrate their vacation on private land in New Mexico that is neither set aside nor superintended for Indians. Say it turns out that the land beneath their picnic blanket is within the political boundaries a Navajo Chapter has asserted for itself. Has the couple just committed a federal offense by introducing alcohol into a dependent Indian community? See 18 U.S.C. §§ 1154, 1156. Under EPA’s approach, no one could be sure until a court has conducted a multi-factor community of reference test and tallied up the set-aside and superintendence percentages. Under that test, the legality of one’s actions "would thus become a post hoc guessing game. Which nearby community might the court pick as its community of reference? Is the community’s population cohesive enough? Are there enough businesses in town to make it a true “mini-society”? What percentage of the land within the community’s boundaries is federally set aside and superintended? Is that percentage enough to make it a dependent Indian community? One wrong guess, and the pop of a champagne cork makes our couple federal offenders.
The case before us illustrates just how hard all that guessing can be. EPA determined that the appropriate community of reference in this case was the Church Rock Chapter, despite evidence that most of the Chapter’s residents work in Gallup; most services and infrastructure are provided by the city, county, or state; the entire area around Section 8 is uninhabited and, in the Chapter’s words, “not suitable for community ... development”; and EPA itself has concluded that no community is ever likely to use the aquifer under HRFs land as a source of drinking water. On a record like this, it seems one might just as easily have determined that the “appropriate” community of reference was larger than the Chapter, perhaps embracing Gallup or all of McKinley County. Or, perhaps just as plausible, an area smaller than the Chapter, comprising only those outlying “rugged mountain ranges, canyons, and highlands” that the Chapter has indicated aren’t suitable for development. See HRI II, 562 F.3d at 1269 (Frizzell, J., dissenting). Or perhaps one might have even concluded that HRI’s land just isn’t part of any community at all. The whole exercise takes on the feel of Goldilocks, searching for an “appropriate” community of reference that feels “just right.”
And that’s just the start of things. Having defined a community of reference, it still remains under the approach urged by EPA and the dissents to ask whether a sufficient degree or percentage of the land in that community is set aside and federally superintended. But the requisite degree or percentage, of course, is specified nowhere in the text of the statute, our pre-Veuetie precedent, EPA’s land status determination, or the dissenting opinions. It must be, then, that this is a sort of I-know-it-when-I-see-it test. Cf. Jacobellis v. Ohio, 378 U.S. 184, 197, 84 S.Ct. 1676, 12 L.Ed.2d 793 (1964) (Stewart, J. concurring). EPA and the principal dissent tell us that 78% set-aside and 92% federal superintendence is sufficient. Land Status Determination at 11-12; Principal Dissent at 1181-82. But one can’t help to wonder: Would 70% set-aside and 90⅜ superintendence cut it? And what happens when those percentages fall a bit farther still? Or if the numbers are reversed? Such extra-statutory guess*1162work is hardly the stuff on which criminal determinations should turn.24
Neither would the scope of “appropriate communities of reference” remain stable under EPA’s approach. With such nebulous criteria to employ, different courts would surely reach different conclusions about whether the same parcel of land falls within one community of reference or another—and thus reach competing holdings about which sovereign enjoys primary criminal law authority. Worse still, and as we have already noted, even the same court may take a different view about a community of reference’s boundaries at different times—shrinking or expanding those boundaries depending on social or political changes in the area. See supra Section III.A. So just because a piece of land was (or wasn’t) Indian country at the time of the last judicial decision, that’s no guarantee the result would be the same next time. No one could be sure.
The dissents’ approach would complicate things further still by introducing the possibility of special purpose communities of reference. Given the “environmental context” in which this case arises, the principal dissent would have us consider presumed “common hydrologfieal]” links between HRI’s land and other parts of the Chapter. Principal Dissent at 1174, 1179-80. But one would think that, in non-environmental cases, courts would not be obliged to look at the hydrological “context” when trying to determine the appropriate community of reference. Who, after all, would fault the parties and court for failing to consider hydrological details in, say, a more conventional § 1151(b) case involving a conspiracy to commit various burglaries? Presumably, under the dissents’ approach, a court would instead focus on whatever the relevant “context” of the case at hand may be. So, to take the burglary case, a court would presumably take account of the “burglary context” in which the case arises. If strong links are found to exist between the housing markets in Church Rock and Gallup, the court might then conclude that the “appropriate” community of reference in that “context” encompasses both. The appropriate community of reference in an environmental case thus could be different than the appropriate community of reference in a burglary case, and both might be different than the right community of reference in still other kinds of cases. So it is that the same piece of land could be both Indian country and not Indian country even in the same court at the same time—depending, that is, on the “context” of the case.25
*1163In this way, the community of reference test invites not just a checkerboard, but a virtually three-dimensional checkerboard, a sort of ever-shifting jurisdictional Rubik’s cube, “jettisoning relative predictability for the open-ended rough-and-tumble of factors” that assures “complex argument[s] in [the] trial court and ... virtually inevitable appealfs].” Jerome B. Grubart, Inc. v. Great Lakes Dredge & Dock Co., 513 U.S. 527, 547, 115 S.Ct. 1043, 130 L.Ed.2d 1024 (1995). And precisely because of the test’s complexity, this dispute has dragged on, year after year, decade after decade, through regulatory proceedings, an appeal, a remand, and an appeal yet again. Since NMED granted HRI a UIC permit in 1989, the lawyers have fought, the parties have lobbied, and an agency and the courts have stewed over the mushy ingredients in the community of reference soup, all in an effort to get the recipe “just right.” So it is that this dispute has nearly reached the age of majority waiting to be served the perfected answer.
Despite all this, the principal dissent levels the remarkable accusation that our decision today “introduc[es] confusion” into an area of law “settled for decades” and “overturn[s] decades of our precedent.” Principal Dissent at 1168; see also id. at 1181. But not only is the atextual community of reference test the source of confusion in this area of law, it is hardly as old or as venerated as the dissents would have it. In fact, the community of reference test is younger than this dispute; it’s been around for only fifteen of the sixty-two years since Congress enacted § 1151(b).26 And in that time, the test has always been a work in progress, anything but “settled.”
In the first place, the test has been applied only a few times since its genesis, and sporadically at that. After Watchman created the test in 1995, we applied it in a published opinion only once before Venetie came along in 1998. See Adair, 111 F.3d at 774-75.27 When the Supreme Court speaks, it (of course) supercedes our prior-case law. And in 1999, in the first § 1151(b) ease following Venetie, we did not use the community of reference test, but straightforwardly applied Venetie’s two requirements. See Roberts, 185 F.3d at 1133. In a second post -Venetie case that followed several years later, we appeared to split the difference, saying “[w]e examine the entire Indian community,” but conducting no community of reference analysis and holding only that “lands within the exterior boundaries of a Pueblo land grant, to which the Pueblo hold title, are Indian country.” United States v. Arrieta, 436 F.3d 1246, 1250-51 (10th Cir.2006) (emphasis added). Thus, the only Tenth Circuit case clearly to employ the community of reference test after Venetie is this case, which created an intracircuit split with Roberts, a fact that led us to grant en banc review.
In the second place, in each of the (few) cases we’ve applied the test, we’ve changed it. Literally. When creating the test in *1164Watchman, we identified “two organizing principles”—“the states of the area in question as a community” and “the context of the surrounding area.” 52 F.3d at 1543-44. Two years later, in Adair, we added another factor concerned with “the geographical definition of the area proposed as a community.” 111 F.3d at 774. And in this case the principal dissent would have us reconfigure the factors in its balancing test yet again. See supra note 17. In its view, the test should now ask “whether the land ... possesses a reasonable degree of coherence” and “whether the uses to which the land is put and the people inhabiting the land possess a reasonable degree of coherence.” Principal Dissent at 1179.
Truth be told, then, the confusion in this area of law has been sown by the community of reference test. Its lack of any basis in the statute’s text, history, or structure, its multifarious and incommensurable competing factors, the unpredictable results it yields, and its constant judicial reworking—all of this has left the law and litigants confused. Rather than adding to the confusion, our decision today eschews yet more tinkering in favor of the simple and predictable test the Supreme Court has told us to use.28
D
Though it does not control our outcome, it’s worth noting that our decision today brings the law of this circuit into harmony with the law developed in our sister circuits after Venetie. In Blunk v. Arizona Department of Transportation, 177 F.3d 879 (9th Cir.1999), the Ninth Circuit acknowledged that “Venetie controlled] [its] decision,” and then concluded that the Navajo fee land in question before it was not Indian country under § 1151(b) because the land was neither set aside nor superintended by the federal government. Id. at 883-84. In reaching its conclusion, the Ninth Circuit rejected the notion, lying at the heart of EPA’s position here, that the land in question might be considered Indian country “because of its proximity or importance to the Navajo Reservation.” Id. at 884. The court conducted no community of reference analysis and considered no factors other than Venetie’s two x*equirements. Id. at 883-84.
In Yankton Sioux Tribe v. Podhradsky, 577 F.3d 951 (8th Cir.2009), the Eighth Circuit conducted a similarly straightforward application of Venetie. Considering whether scattered trust lands qualified as Indian country under § 1151(b), the *1165Eighth Circuit concluded that all of these lands “easily meet [Venetie’s ] definition.” Id. at 971. At no point did the court seek to place these various parcels within a larger community of reference, nor did it analyze the particular parcels at issue in light of any factors other than the two Venetie prescribes. See id. Had the Eighth Circuit applied the community of reference test urged by EPA, the result in Yankton might have been quite different— perhaps the non-Indian lands interspersed among the scattered trust lands might have become “Indian country,” or perhaps, if the non-Indian land had predominated the community, the scattered trust lands might have ceased to be “Indian country” under § 1151(b).29
A state court within our jurisdiction has also rejected this circuit’s community of reference jurisprudence in favor of Vene-tie’s two-part analysis. In State v. Frank, 132 N.M. 544, 52 P.3d 404 (2002), the Supreme Court of New Mexico held that the § 1151(b) inquiry must focus solely on the land in question, not on some broader community of reference. Id. at 409 (“In light of the clear guidelines in the Venetie opinion, we decline to incorporate a community of reference inquiry into our case law.”). Indeed, the court acknowledged, as we have today, that “the six-factor test that was rejected by the Supreme Court in Venetie used essentially the same factors as those in” this court’s community of reference test. Id.; see also id. (“The Vene-tie two-prong test redirects our attention ... away from the more nebulous issue of community cohesiveness.” (internal quotation marks omitted)).30
*1166IV
Ultimately, Venetie compels us to abandon the community of reference test. Under the proper test we adopt today, only two questions are relevant in assessing claims of jurisdiction under § 1151(b): (1) Has Congress (or the Executive, acting pursuant to delegated authority) taken some action explicitly setting aside the land in question for Indian use? (2) Is the land in question superintended by the federal government? Because the parties agree that the land in question in EPA’s final land status determination—HRI’s segment of Section 8—is neither explicitly set aside for Indian use nor federally superintended, it follows that, as a matter of law, the land does not qualify as Indian country under § 1151(b). And from this it follows that EPA’s final land status determination holding otherwise must be vacated as inconsistent with the statute. While the outcome may have been different under this circuit’s Watchman test, Venetie is now the law and neither we nor EPA may ignore it.
It may be that Venetie complicates to some degree EPA’s efforts to regulate activities affecting underground water sources, which of course don’t follow neat land survey lines. But that is because the Agency has chosen, in an exercise of its considerable discretion under the SDWA, to limit its primary regulatory authority over water quality to those Indian lands encompassed by § 1151, a criminal jurisdictional statute. While § 1151 does its job of assigning prosecutorial authority over particular tracts of land tolerably well, it is perhaps unsurprising that it may prove less satisfactory when it comes to allocating regulatory authority over aquifers running beneath those lands. Crimes, after all, usually occur on land, not in aquifers.
Someday, EPA may seek to avoid these difficulties by unhitching its UIC permitting authority from § 1151. We do not purport to pass on the propriety or wisdom of such a move. For now, we concern ourselves only with EPA’s current choice to confine its authority in New Mexico to what § 1151 deems Indian country. And on that score, we cannot help but conclude that EPA’s final land status determination under review is inconsistent with the statute’s terms as a matter of law and cannot stand.
The panel opinion is vacated, the petition for review is granted, and the EPA’s final land determination is vacated.
APPENDIX
*1167[[Image here]]
. The Appendix reproduces a map of the Church Rock Chapter and its boundaries. The map was prepared as part of the Church Rock Chapter’s Land Use Plan and included in the record before the Agency. See R. 16b; R. 40 at B-39.
. We recognize that “most tribal members do not refer to themselves as generic ‘Indians’ or ‘Native Americans,’ but rather as constituents of particular groups, such as Hopi or Cheyenne” or Navajo. Cohen’s Handbook of Federal Indian Law, § 3.01 at 134 n. 1 (Nell Jess-up Newton et al. eds., 3d ed.2005). We use the general term "Indian” throughout this opinion in light of its use in the relevant statutory language and precedents.
. The EPA’s final land status determination states that the “Church Rock Chapter was first established in 1927 by the United States as a subdivision of the Navajo Nation government, to facilitate local Navajo self-government and to foster improved communications between Navajos and federal agencies." R. 44 at 8. HRI disputes this conclusion. First, HRI argues that the 1927 date makes no sense, because it was a year before the government began repurchasing for Indian use much of the land that now comprises the Chapter. Second, HRI argues that the record contains no support for the assertion that the federal government established the Church Rock Chapter. Upon review of the source cited by EPA, it appears HRI may have the better of the argument. While the source suggests that the federal government played some undefined role in helping to establish Navajo chapters beginning in 1927, it does not mention the Church Rock Chapter, let alone indicate that the federal government established it. See Robert Young, Navajo Yearbook 191 (1958). The record, at least as briefly adduced by the parties in this litigation, thus suggests that the Navajo Nation, not the federal government, established the Chapter.
. "Underground injection . .. means the subsurface emplacement of fluids by well injection.. ..” 42 U.S.C. § 300h(d)(l)(A). Basically, HRI seeks to pump water into the ground and circulate that water in order to pull uranium to the surface where it can be easily collected. This method is apparently less onerous than manually excavating the rock from underground and removing the uranium later. R. 15c App. II at 1, 3-4 ("Instead of manually excavating the rock from underground as in conventional mining and placing it in large piles on the surface, water wells are used, very much like those for a home. Ox \ - gen is added to the native ground water from the ore body, and that water is continuously circulated until most of the uranium is recovered.”).
. While statutory criminal jurisdiction has been confined to “Indian country” as defined in § 1151, it has been said that Congress enjoys “plenary power over Indian affairs,” Venetie, 522 U.S. at 531 n. 6, 118 S.Ct. 948, *1139by virtue of Article I, Section 8, Clause 3 of the Constitution, which ascribes to Congress exclusive authority over “Commerce ... with the Indian Tribes.’’ Whether and how far Congress's constitutional authority to authorize federal "Indian country" jurisdiction might proceed beyond the limits of § 1151 is not before us in this ease.
. HRI also sought review of EPA’s assertion of jurisdiction over its mining activities on Section 17. See HRI I, 198 F.3d at 1243-44. Unlike its Section 8 land, HRI owns only the mineral rights for Section 17, while the federal government holds the Section 17 land in trust for the Navajos. The panel in HRI I held that Section 17 was Indian country under § 1151(a) and thus HRI’s mining of that land was subject to EPA’s regulatory jurisdiction. Id. at 1249-54. HRI did not seek review of that issue with the ett banc court or file a petition for certiorari with the Supreme Court. Consequently, all that remains in dispute is HRI’s Section 8 land.
. Pending before die court are several motions seeking leave to file amicus briefs. Because the movants possess an adequate interest and present arguments that are useful to this court, we grant the motions of the Pueblos of Santa Clara, Sandia, Isleta, and Zia and the United Nuclear Corporation. We deny the motion of the American Indian Law Professors for leave to file an amicus brief only because granting the motion would cause one or more members of this court to recuse themselves from the matter. See 16AA Charles Alan Wright et al., Federal Practice and Procedure § 3975, at 318-19 (4th ed. 2008) ("Some circuits will restrict amicus filings in order to avoid disqualifying a member ... of the en banc court.... ”). The states of Colorado, Kansas, New Mexico, Utah, and Wyoming also filed an amicus brief, which Federal Rule of Appellate Procedure 29(a) allows them to do without requesting the leave of this court.
. EPA hasn’t challenged the remaining elements of constitutional standing for good reason. HRI's claimed injury—having to incur the expenses associated with a second UIC application—is directly traceable to EPA's final land status determination. And that injury is fully redressable by a favorable decision of this court vacating EPA’s decision as incompatible with the law.
. HRI notes this court’s previous suggestion that "[cjompulsion by unwanted and unlawful government edict is injury per se." Califano, 622 F.2d at 1389; see also 13A Charles Alan Wright et al., Federal Practice and Procedure § 3531.4, at 249 (3d ed. 2008) (“[ I Injury sufficient to support standing may be found in subjection to unwanted procedures.”). The panel did not reach or rest its standing holding on this ground, however, and neither must we, given that EPA's rulings will indisputably impose some out-of-pocket costs on URL
. Though claiming no entitlement to deference for its legal interpretation of § 1151 (b), EPA does seek deference for its factual findings in applying its interpretation of the statute to this case. See 5 U.S.C. § 706(2)(E) (establishing "substantial evidence” review); Allentown Mack Sales & Serv., Inc. v. NLRB, 522 U.S. 359, 366, 118 S.Ct. 818, 139 L.Ed.2d 797 (1998). As will be apparent in the analysis that follows, however, only purely legal questions, not factual ones, are in dispute before us.
The principal dissent agrees that EPA is not entitled to Chevron deference, but argues that the Agency deserves Skidmore deference. Principal Dissent at 1170 n.2. Of course, an agency's interpretation of a statute merits deference under Skidmore only in "proportion) ] to its 'power to persuade,’ ” United States v. Mead Corp., 533 U.S. 218, 235, 121 S.Ct. 2164, 150 L.Ed.2d 292 (2001) (quoting Skidmore v. Swift & Co., 323 U.S. 134, 140, 65 S.Ct. 161, 89 L.Ed. 124 (1944)); see also id. at 250, 121 S.Ct. 2164 (Sealia, J., dissenting) (Skidmore deference is a "statement of the obvious: A judge should take into account the well-considered views of expert observers."). Whether or not Skidmore compels us to do so, we certainly seek in this opinion to take into account EPA’s considered views. But we decline to decide the question whether we are obliged by Skidmore to afford even this modicum of deference, for a few reasons. First, EPA has not sought Skidmore deference, and when a party chooses not to pursue a legal theory potentially available to it, we generally take the view that it is "inappropriate” to pursue that theory in our opinions. United States v. Int'l Bus. Mach. Corp., 517 U.S. 843, 855, 116 S.Ct. 1793, 135 L.Ed.2d 124 (1996); see also Estate of Cowart, 505 U.S. at 476-77, 112 S.Ct. 2589 (declining to pass on deference question because the agency requested no deference); cf. Spector Motor Serv., Inc. v. Walsh, 139 F.2d 809, 823 (2d Cir.1943) (L.Hand, J., dissenting) (“'[It is not] desirable for a lower court to embrace the exhilarating opportunity of anticipating a doctrine which may be in the womb of time, but whose birth is distant ....”), vacated sub nom. Spector Motor Serv., Inc. v. McLaughlin, 323 U.S. 101, *114765 S.Ct. 152, 89 L.Ed. 101 (1944). Our caution in this respect flows front a recognition of our dependence on the adversarial process to test the issues for our decision and from concern for the affected parties to whom we traditionally extend notice and an opportunity to be heard on issues that affect them. Second, Skidmore deference traditionally has been justified, at least in part, on an assumption that the agency in question has “specialized experience and broader investigations and information available to” it than do judges. Mead, 533 U.S. at 234, 121 S.Ct. 2164 (internal quotation marks omitted). EPA has claimed, however, no comparative experience or expertise over us in rendering a legal interpretation of a criminal statute that Congress has never charged the Agency to administer and which we have long experience applying. Third, the principal dissent offers no indication what impact Skidmore deference makes to its analysis. If it makes no difference, then deciding the question is just an exercise in dicta. But if it does make a difference, then presumably the dissent would agree with us that, at least if viewed de novo, EPA's final land status determination is contrary to law.
. It is notable, too, that post-Venetie another arm of the government has taken the position, in a criminal matter before this court, that the proper focus is on the particular piece of land in question, explaining that “lljederal jurisdiction is absent over crimes which occur on land which has transferred from Indian ownership to non-Indian ownership." Brief of the United States at 7, United States v. Arrieta, 436 F.3d 1246 (10th Cir.2006).
. The principal dissent errs when it suggests that under our reading of Venetie “title alone is determinative of whether a parcel of land is ‘Indian country’ under § 1151(b).” Principal Dissent at 1172. The test we adopt at Vene-tie’s direction focuses on set-aside and superintendence, not title. While the results of the principal dissent’s imagined title test would surely sometimes coincide with Venetie’s set-aside and superintendence test, the two are distinct. It is, after all, possible that Congress might set aside and guarantee federal superintendence for privately titled land, much as it holds certain privately titled lands within reservations to be Indian country. See infra Section III.B (discussing a 2005 law treating privately titled land within pueblos as Indian country). What the constitutional limits may be to Congress’s authority to declare land titled to non-Indians to be part of Indian country is not before us. See supra note 5.
. The principal dissent emphasizes that Ven-etie considered the status of the Tribe’s 1.8 million acres of ANCSA land, though the dispute originally arose when the Tribe claimed taxing authority over a small strip of land on which the State of Alaska planned to build a *1152school. Apparently the principal dissent believes this suggests that the Court implicitly thought of all 1.8 million acres of ANCSA land as the appropriate "community of reference”; that is, the Court really used a community of reference analysis, it just forgot or failed to say so. See Principal Dissent at 1175-76. Nothing of the kind took place. Before the Ninth Circuit, the Tribe categorically claimed "inherent authority to tax activities occurring” anywhere "within its [ANCSA] territory." Venetie, 101 F.3d at 1290. The Ninth Circuit agreed, holding that the Tribe enjoyed such taxing authority throughout the ANCSA territory. By dint of that holding, all 1.8 million acres were in dispute when the case came before the Supreme Court. The Court did not explicitly, implicitly, or otherwise hold that the whole 1.8 million acres, or any portion of it, qualified as an appropriate community of reference. Instead, the Court did what it instructed other courts facing § 1151(b) cases to do; it simply and straightforwardly analyzed whether the land in question in the litigation at hand—all 1.8 million acres of the Tribe’s ANCSA land—was set aside or superintended. And it held that none of that land met those requirements.
. Indeed, for EPA’s understanding of § 1151 (b) to be correct, the statute would have to read very differently than it does, perhaps something like this: "All land must be part of some community of reference that is sufficiently cohesive economically and socially to be a mini-society. If the land in question is part of a community of reference that contains a sufficiently high percentage of Indian people and lands set aside for Indians and federally superintended, it is Indian country.” What the statute actually says, of course, is that Indian country includes dependent Indian communities—that is, land that has been federally set aside for use by an "Indian community” that remains "dependent” on the federal government through its superintendence. See Venetie, 522 U.S. at 530-31, 118 S.Ct. 948.
. This result, virtually inevitable under the community of reference test, sits uneasily with the Supreme Court's direction that the diminishment of reservations under § 1151(a), like additions to them, is Congress's prerogative alone. See, e.g., Solem v. Bartlett, 465 U.S. 463, 470, 104 S.Ct. 1161, 79 L.Ed.2d 443 (1984) ("The first and governing principle is that only Congress can divest a reservation of its land and diminish its boundaries."); id. at 471-72, 104 S.Ct 1161 (treating history of land after enactment of a statute allegedly effecting reservation diminishment as "relevant” only to "decipher Congress' intention,” as an "additional clue as to what Congress expected would happen”); Rosebud Sioux Tribe, 430 U.S. at 588 n. 4, 97 S.Ct. 1361 ("The focus of our inquiry [in reservation diminishment cases) is congressional intent.”); Yazzie, 909 F.2d at 1393 ("Congress must clearly evince the intent to reduce [reservation] boundaries.”).
. The dissents offer no response to this except to suggest that Venetie permits us to ask about the "degree” or "extent” of federally set aside and superintended land in the proximate area, relying on footnote 7 of the Court's opinion. Principal Dissent at i 176-77 (quoting Cohen (2005) § 3.04[2][c][iii], at 194). But rather than support the dissent’s view, the cited footnote does just the opposite. The footnote was expressly aimed at rejecting the Ninth Circuit's use of the factors that form the basis of the community of reference test, calling them "extremely far removed" from the statute's language, history, and mandate. Venetie, 522 U.S. at 531 n. 7, 118 S.Ct. 948. The Court indicated that the Ninth Circuit's examination of the “degree" or "extent” of set-aside and superintended land in the general area were "more relevant” inquiries, but the Court did not endorse this aspect of the Ninth Circuit's approach either. Id. Instead, in formulating its own test the Court repeatedly held that dependent Indian communities constitute only those lands explicitly set aside by Congress for Indians and superintended by the government for their benefit. See, e.g., id. at 527, 118 S.Ct. 948 (stating that the land “must have been set aside” and “must be under federal superintendence” (emphasis added)).
. Neither do the dissents. Instead, the principal dissent proposes a new community of reference test different from the one this court used in Watchman or EPA used in its final land status determination. The principal dissent would still apparently uphold EPA's determination, though, rather than remand the case for reconsideration under its newly formulated version of the community of reference test. And it would do so even though its proposed test arguably might require additional fact finding by the Agency. Perhaps this is because, while the details are slightly changed, the essence of the test is not. Like the Watchman test EPA employed, the principal dissent's modified test would still have us look generally (and impermissibly) to the "coherence” of the area and its inhabitants and their social and political affinities. See Principal Dissent at 1179-81.
. The fact that Sandoval and McGowan rested on an outdated sense of cultural superiority, viewing the Indian communities at issue as dependent and in need of paternalistic guidance by the federal government, may be further reason to tread carefully before haphazardly expanding the subsection (b) class of “dependent Indian community” lands without express congressional direction. See, e.g., James E. Lobsenz, “Dependent Indian Communities: ” A Search for a Twentieth Century Definition, 24 Ariz. L.Rev. 1, 2 (1982) ("The ‘quasi-sovereign’ or ‘dependent’ status of the Indian tribe is inextricably linked to past concepts of the Indian as an uncivilized savage who was to be gradually elevated to the level of a civilized human being.”).
. The dissents say we should "listen” to the Cohen Handbook’s purported condemnation of checkerboarding. Separate Dissent at 1183-84; see also Principal Dissent at 1171-72, 1176-77. Of course, we have and do listen to anyone who offers some valuable insight. But: the various versions of the Handbook have hardly been as clear on this point as the dissents suggest. The 1982 edition acknowledged, as we did in Yazzie, that § 1151 “does not eliminate all forms of 'checkerboarding.'” Cohen (1982) Ch. 1, § D3c, at 39 n.105. The original edition of the Handbook—the only one authored by Professor Cohen's hand—was precise in reading Sandoval and McGowan, the progenitors of the term “dependent Indian communities,” as including within Indian country only lands "which have been validly set apart for the use and occupancy of Indians." Cohen (1942) Ch. 1, § 3, at 8. And the language from the 2005 edition on which the dissents rely appears in a single sentence, citing and discussing only a § 1151(a) case. Surely, a stray sentence in one edition of a treatise—a sentence that cites no § 1151 (b) case for support, that was authored many decades after the passage of the statute, and that is seemingly at odds with previous editions—cannot be the alpha and omega of this court's inquiry into statutory meaning.
. The dissents read Mazurie to suggest that the Supreme Court has adopted an approach similar to the community of reference test for evaluating cases under 18 U.S.C. § 1154(c)’s “non-Indian communities” exception. See Principal Dissent at 1173-74; Separate Dissent at 1183. But Mazurie deals with an entirely different statutory provision removing from “Indian country” certain lands within the boundaries of a reservation—land that Congress has otherwise directed should be treated as “Indian country” under § 1151(a). Although the terms “non-Indian communities” and “dependent Indian *1159communities” both concern communities, they deal with markedly different types of communities. And so the analysis to determine the bounds for each unsurprisingly differs. As Venetie explains, § 1151(b) requires a showing that Congress has designated the land as an Indian community (set-aside) and that the community is federally dependent (federal superintendence), while § 1154(c) non-Indian communities plainly require neither of these characteristics. Thus, although Mazurie tells us something about how to identify non-Indian communities under § ! 154(c), it tells us nothing about how to identify dependent Indian communities under § 1151(b). The Supreme Court itself noted just this point. See Mazurie, 419 U.S. at 553 n. 10, 95 S.Ct. 710 (“We note that the § 1154(c) exception is available for fee-patented lands which are in non-Indian communities, rather than for those which are not in Indian communities. This fact renders irrelevant the inability of prosecution witnesses to specify precise boundaries of the Fort Washakie Indian community.’').
. EPA and the principal dissent fail to explain how the Red Rock doughnut hole might permissibly be excluded from their community of reference, while "a vacant lot in the middle of a developed neighborhood” must be part of the surrounding community. Principal Dissent at 1172. It seems the only possible explanation they might: offer is that the Navajo Nation has chosen to exclude the Red Rock State Park from the Church Rock Chapter boundaries. And this highlights one way the community of reference can go wrong, by impermissibly delegating to a tribe Congress’s authority to define "Indian country.” See Venetie, 522 U.S. at 531 n. 6, 118 S.Ct. 948.
. See also Brendale v. Confederated Tribes & Bands of the Yakima Indian Nation, 492 U.S. 408, 425 n. 8, 109 S.Ct. 2994, 106 L.Ed.2d 343 (1989) (White, J., for the Court in part and dissenting in part) (rejecting a proposed jurisdictional approach because the "uncertainty that would result from the necessarily case-by-case determination of which regulatory body (or bodies) has zoning jurisdiction over such land” (citations omitted)); Ute Indian Tribe of the Uintah & Ouray Reservation v. State of Utah, 114 F.3d 1513, 1527 (10th Cir. 1997) (''[Tjhe task of allocating jurisdiction necessarily involves line-drawing, and in an area where there is a compelling need for uniformity, there must be a single bright line.”).
.The Supreme Court recently reminded us, moreover, that even in civil cases “administrative simplicity is a major virtue in a jurisdictional statute.” Hertz Corp. v. Friend, - U.S. -, -, 130 S.Ct. 1181, 1193, — L.Ed.2d - (2010); see also id. (explaining that "[sjimple jurisdictional rules .. . promote greater predictability), which] is valuable to corporations making business and investment decisions”). This is because “[cjomplex jurisdictional tests complicate a case, eating up time and money as the parties litigate, not the merits of their claims, but which court is the right court to decide those claims.” Id. As a result, such tests "produce appeals and reversals, encourage gamesmanship, and, again, diminish the likelihood that results and settlements will reflect a claim’s legal and factual merits.” Id.
. EPA can't even be sure of the percentages on which it relies. As the principal dissent acknowledges, the Agency’s figures "do not add up exactly to 100%." Principal Dissent at 1168 n. 1. We are told not to worry, though, because the various percentages in the record are “within 2.4 %" of each other, id., and this margin of error is—we are left to presume— no big deal.
. The dissents' approach rests on the factual premise that Section 8 shares a common "hydrology [with] the entire Church Rock Chapter,” such that “any pollution into the aquifers" under Section 8 “would likely affect much of the Chapter population.” Principal Dissent at 1180. EPA, however, disputes this premise, having concluded that any aquifer affected by HRI's operations on Section 8 "cannot now and will not in the future serve as a source of drinking water.” See supra Section l.C. And the dissents cite no record authority to support their competing claim. While this court is no less sensitive than the dissents to the potential contamination of drinking water, we can claim no expertise on the hydrology of McKinley County and have no idea whether EPA’s scientific assessment (unchallenged by the parties before us), or the dissents', is correct. We can be certain, however, that the plain language of § 1151(b) doesn’t require us to become hydrological experts to apply the law.
. The principal dissent pegs the test’s age at twenty, citing the Yazzie decision from 1990. But that's mistaken. Yazzie adopted a holding under § 1151(a), not § 1151(b), and thus said nothing about a community of reference test. The principal dissent's citation to Yazzie points to that decision’s appendix, where this court merely reproduced (without endorsement) the district court opinion discussing § 1151 (b) in a consolidated case. That’s why, five years later in Watchman, this court called the community of reference question “a question of first impression.” 52 F.3d at 1543.
. In an unpublished decision also preceding Venetie, a panel of this court acknowledged the existence of the community of reference test but declined to apply it because it "was not addressed by the parties or the district court.” Eaves v. Champion, 1997 WL 291186, at *2 n. 4 (10th Cir.1997) (unpublished).
. The separate dissent’s suggestion that our decision disregards the “effects of a mining operation that may greatly impact the surrounding lands," is likewise unwarranted. Separate Dissent at 1184. No one questions that mining operations can pose "grave consequences." Id. It is this court's hope, no less than the hope of the dissents, that the appropriate agencies will do their assigned jobs in protecting the environment. See, e.g., Morris v. U.S. Nuclear Regulatory Comm’n, 598 F.3d 677 (10th Cir.2010) (upholding Nuclear Regulatory Commission licensing of HRI’s uranium mining plans on Section 8). But we are judges, not environmental regulators. As such, “[o]ur charge," and all this court has sought or may properly seek to do, "is to give effect to the law Congress enacted." Lewis v. City of Chicago, 560 U.S. -, 130 S.Ct. 2191, 2200, — L.Ed.2d - (2010).
The separate dissent is likewise mistaken when it suggests that today's decision means that HRI’s activities will no longer be "subject to federal environmental regulations.” Separate Dissent at 1182, 1184-85. As we have explained, the SDWA, a federal statute, provides the regulatory framework for all UIC permits nationwide. EPA regulations govern when those permits may be issued. Thus, although EPA may have delegated to NMED authority to issue permits in New Mexico, NMED may issue its permits only in compliance with federal law (SDWA) and federal (EPA) regulations. Indeed, EPA suggested that no case or controversy exists before us *1165precisely because of this. See supra Section II.A.
. For this reason, EPA and the principal dissent are wrong when they suggest that Blunk and Yankton shouldn’t be read to reject the use of a threshold community of reference test because they didn't consider it. See Principal Dissent at 1177-78. The fact is that each court put every parcel of land at issue through the Venetie two-part test and required that each, not some "percentage" or “degree,” satisfy those requirements. In Blunk, none of the Navajo fee land met Venetie's requirements, while in Yankton, all of the miscellaneous trust lands met the two requirements. If the community of reference test were an essential threshold inquiry, as EPA and the principal dissent argue, it's unclear how these courts could have conducted a proper § 1151 (b) analysis without it.
. The New Mexico Supreme Court took a somewhat different approach when dealing with private land within pueblo boundaries. See State v. Romero, 140 N.M. 299, 142 P.3d 887, 895 (2006). In Romero, the court held that private land within the original land grant of the Taos pueblo is Indian country and therefore beyond the state’s criminal jurisdiction. The Romero court distinguished its case from Frank on the grounds that pueblos have long been recognized as Indian country. See id. at 892. Although the court appeared to treat the pueblo as a dependent Indian community under § 1151(b), much of its analysis focused on comparing pueblos to reservations under § 1151(a), See, e.g., id. at 894 (“We determine that a pueblo satisfying § 1151(b) is sufficiently similar to a reservation in § 1151(a) to merit identical treatment for the purposes of criminal jurisdiction.”). Along these lines, the court declined to conduct a community of reference analysis, and instead analyzed the pueblo as it would a reservation. It is not entirely dear how to reconcile Frank and Romero, and to the extent that the two cases conflict, we agree with Frank. Land not explicitly set aside and superintended is not Indian country under § 1151(b). Neither is it clear whether or what remaining force Romero bears today, given that Congress has now spoken directly to the criminal jurisdictional status of privately titled lands within pueblo boundaries in New Mexico. See Indian Pueblo Land Act Amendments of 2005, Pub.L. No. 109-133, 119 Stat. 2573, codified at 25 U.S.C. § 331 Note; see also supra Section III.B. Romero was obliged to analyze the land in question under § 1151 alone rather than in light of Congress's new statute because the statute didn’t apply retroactively to the case. See Romero, 142 P.3d at n. 1; see also Robert L. Lucero, Jr., State v. Romero: The Legacy of Pueblo Land Grants and the Contours of Jurisdiction in Indian Country, 37 N.M. L.Rev. 671, n.6 (2007) (explaining the history and structure of the Act’s jurisdictional grant). And considering that Romero explicitly declined to overrule Frank, there can be little *1166question that Frank remains governing law in New Mexico. See Romero, 142 P.3d at 892.