GORSUCH, Circuit Judge.

Andrew Yellowbear will probably spend the rest of his life in prison. Time he must serve for murdering his daughter. With that much lying behind and still before him, Mr. Yellowbear has found sustenance in his faith. No one doubts the sincerity of his religious beliefs or that they are the reason he seeks access to his prison's sweat lodge — a house of prayer and meditation the prison has supplied for those who share his Native American religious tradition. Yet the prison refuses to open the doors of that sweat lodge to Mr. Yellowbear alone, and so we have this litigation. While those convicted of crime in our society lawfully forfeit a great many civil liberties, Congress has (repeatedly) instructed that the sincere exercise of religion should not be among them — at least in the absence of a compelling reason. In this record we can find no reason like that.

I

Our story starts with Smith. In Employment Division v. Smith, 494 U.S. 872, 110 S.Ct. 1595, 108 L.Ed.2d 876 (1990), the Supreme Court held that the Constitution's Free Exercise Clause does not exempt religious persons from the dictates of neutral laws of general applicability. The devout must obey the law even if doing so violates every article of their faith. When Smith was handed down, some worried that it upset existing free exercise doctrine dating back to Sherbert v. Verner, 374 U.S. 398, 83 S.Ct. 1790, 10 L.Ed.2d 965 (1963). In Sherbert and its progeny the Supreme Court had suggested that no law, not even a neutral law of general applicability, may "substantially burden" the exercise of religion unless that burden amounts to the "least restrictive means" of achieving a "compelling governmental interest." Smith, 494 U.S. at 883, 110 S.Ct. 1595; id. at 899, 110 S.Ct. 1595 (O'Connor, J., concurring in the judgment). What protections Sherbert appeared to afford religious observances, Smith appeared ready to abandon.

Concerned with just this possibility, worried that Smith left insufficient room in civil society for the free exercise of religion, Congress set about the business of "restoring" Sherbert, at least as a matter of statute. It opened its efforts with the Religious Freedom Restoration Act of 1993. See 42 U.S.C. § 2000bb(b)(1). Passed nearly unanimously, RFRA was (and remains) something of a "super-statute." Michael Stokes Paulsen, A RFRA Runs Through It: Religious Freedom and the U.S.Code, 56 Mont. L.Rev. 249, 253 (1995). It instructed that all forms of governmental action — state or federal — had to satisfy Sherbert's test or risk nullification.

But as it turned out, this marked only the opening lines in what proved to be a long dialogue between Congress and the Court. In City of Boerne v. Flores, 521 U.S. 507, 117 S.Ct. 2157, 138 L.Ed.2d 624 (1997), the Court held that RFRA stretched the federal hand too far into places reserved for the states and exceeded Congress's Section 5 enforcement authority under the Fourteenth Amendment. As a result, the Court held RFRA unconstitutional as applied to the states, though still fully operational as applied to the federal government. See id. at 529-36, 117 S.Ct. 2157.

Undaunted, Congress reentered the field soon enough, this time with the Religious Land Use and Institutionalized Persons Act of 2000. In RLUIPA Congress invoked not just its Fourteenth Amendment but also its Spending Clause powers to (re)impose Sherbert's balancing test on state action — though now state action in only two specific arenas, arenas in which Congress found the record of religious discrimination particularly clear and compelling. First, in the land use context, where churches are sometimes disfavored by local zoning boards because (among other things) church members are said to generate "too much" traffic or congestion or noise when they gather for communal expressions of faith. Second, in the prison context, where it is so easy for governmental officials with so much power over inmates' lives to deny capriciously one more liberty to those who have already forfeited so many others. See Douglas Laycock & Luke W. Goodrich, RLUIPA: Necessary, Modest, and Under-Enforced, 39 Fordham Urb. L.J. 1021, 1021, 1025-41 (2012); Derek L. Gaubatz, RLUIPA at Four: Evaluating the Success and Constitutionality of RLUIPA's Prisoner Provisions, 28 Harv. J.L. & Pub. Pol'y 501, 510 & n.34 (2005). This time Congress acted unanimously and this time the Court upheld its effort, at least against a facial challenge under the Establishment Clause. See Cutter v. Wilkinson, 544 U.S. 709, 725, 125 S.Ct. 2113, 161 L.Ed.2d 1020 (2005).

That takes us to the nub of our case. Mr. Yellowbear, an enrolled member of the Northern Arapaho Tribe, seeks access to the prison's existing sweat lodge to facilitate his religious exercises. The prison has refused. The prison's sweat lodge is located in the general prison yard and Mr. Yellowbear is housed in a special protective unit (not because of any disciplinary infraction he has committed, but because of threats against him). Prison officials insist that the cost of providing the necessary security to take Mr. Yellowbear from the special protective unit to the sweat lodge and back is "unduly burdensome." Mr. Yellowbear disagrees and seeks relief under RLUIPA. For its part, the district court discerned no statutory violation and entered summary judgment against Mr. Yellowbear. Mr. Yellowbear asks us to undo that judgment so that his case might proceed to trial.

At the end of the day, we find that's exactly the relief we must provide.1

II

RLUIPA may be a "super statute," capable of mowing down inconsistent laws, but to win its application takes no small effort. A plaintiff must carry at least two burdens, and even then can still lose if the government bears two burdens of its own.

Take the plaintiff's burdens first. RLUIPA requires us to ask whether an inmate's (1) religious exercise is (2) substantially burdened by prison policy. 42 U.S.C. § 2000cc-1(a). At the outset, then, RLUIPA requires the plaintiff to show a religious exercise. The law does not protect from governmental intrusion every act born of personal conscience or philosophical conviction. It protects only those motivated by religious faith — in recognition, no doubt, of the unique role religion, its free exercise, and its tolerance have played in the nation's history. See Wisconsin v. Yoder, 406 U.S. 205, 215-16, 92 S.Ct. 1526, 32 L.Ed.2d 15 (1972); Frazee v. Ill. Dep't of Emp't Sec., 489 U.S. 829, 833-35, 109 S.Ct. 1514, 103 L.Ed.2d 914 (1989); United States v. Meyers, 95 F.3d 1475, 1482-84 (10th Cir.1996); Michael W. McConnell, The Origins and Historical Understanding of Free Exercise of Religion, 103 Harv. L.Rev. 1409, 1488-1500 (1990).

Of course, trying to separate the sacred from the secular can be a tricky business — perhaps especially for a civil court whose warrant does not extend to matters divine. But at least one feature of the statute's "religiosity" requirement often proves relatively unintrusive in its application and not infrequently dispositive: the question of sincere belief. RLUIPA does not offer refuge to canny operators who seek through subterfuge to avoid laws they'd prefer to ignore. Like those who set up "churches" as cover for illegal drug distribution operations. Or those who, facing the difficult realities of prison life, are tempted to seek special dispensations through fraudulent assertions of faith. But in suggesting we may ask whether a claimant truly holds a religious belief isn't to suggest we may decide whether the claimant's religious belief is true. After all, "Faith means belief in something concerning which doubt is still theoretically possible." William James, The Will To Believe 90 (1897). And even if it were otherwise, federal judges are hardly fit arbiters of the world's religions. When inquiring into a claimant's sincerity, then, our task is instead a more modest one, limited to asking whether the claimant is (in essence) seeking to perpetrate a fraud on the court — whether he actually holds the beliefs he claims to hold — a comparatively familiar task for secular courts that are regularly called on to make credibility assessments — and an important task, too, for ensuring the integrity of any judicial proceeding. See, e.g., United States v. Quaintance, 608 F.3d 717, 720-23 (10th Cir.2010); Abdulhaseeb v. Calbone, 600 F.3d 1301, 1312-14 (10th Cir.2010); United States v. Seeger, 380 U.S. 163, 184-85, 85 S.Ct. 850, 13 L.Ed.2d 733 (1965); Cutter, 544 U.S. at 725 n. 13, 125 S.Ct. 2113; Donald A. Giannella, Religious Liberty, Nonestablishment, and Doctrinal Development. Part I. The Religious Liberty Guarantee, 80 Harv. L.Rev. 1381, 1417-18 (1967).

While a RLUIPA plaintiff must show a sincerely held religious belief, the statute protects considerably more than the right to hold that belief in private. RLUIPA protects religious exercises. And as the Supreme Court explained in Smith, "the `exercise of religion' often involves not only belief and profession but the performance of (or abstention from) physical acts: assembling with others for a worship service, participating in sacramental use of bread and wine, proselytizing, abstaining from certain foods or certain modes of transportation." 494 U.S. at 877, 110 S.Ct. 1595. Neither must the religious claimant prove that the exercise at issue is somehow "central" or "fundamental" to or "compelled" by his faith. Just as civil courts lack any warrant to decide the truth of a religion, in RLUIPA Congress made plain that we also lack any license to decide the relative value of a particular exercise to a religion. That job would risk in the attempt not only many mistakes — given our lack of any comparative expertise when it comes to religious teachings, perhaps especially the teachings of less familiar religions — but also favoritism for religions found to possess a greater number of "central" and "compelled" tenets. To avoid traps like these, Congress has directed courts to protect "any exercise of religion, whether or not compelled by, or central to, a system of religious belief." 42 U.S.C. § 2000cc-5(7)(A) (emphasis added). Under this standard, it isn't for judges to decide whether a claimant who seeks to pursue a particular religious exercise has "correctly perceived the commands of [his] faith" or to become "arbiters of scriptural interpretation." Thomas v. Review Bd. of Ind. Emp't Sec. Div., 450 U.S. 707, 716, 101 S.Ct. 1425, 67 L.Ed.2d 624 (1981). Even if others of the same faith may consider the exercise at issue unnecessary or less valuable than the claimant, even if some may find it illogical, that doesn't take it outside the law's protection. Instead, RLUIPA protects any exercise of a sincerely held religious belief. When a sincere religious claimant draws a line ruling in or out a particular religious exercise, "it is not for us to say that the line he drew was an unreasonable one." Id.; see also Abdulhaseeb, 600 F.3d at 1314; Steven C. Seeger, Note, Restoring Rights to Rites: The Religious Motivation Test and the Religious Freedom Restoration Act, 95 Mich. L.Rev. 1472, 1503-04 (1997).

Separate and apart from showing a sincerely motivated "religious exercise," a RLUIPA plaintiff must also demonstrate the government has imposed a "substantial burden" on that exercise. 42 U.S.C. § 2000cc-1(a). Once again, the inquiry here isn't into the merit of the plaintiff's religious beliefs or the relative importance of the religious exercise: we can't interpret his religion for him. Instead, the inquiry focuses only on the coercive impact of the government's actions. In tort law, we take plaintiffs as we find them, assessing the extent of damage done based on each plaintiff's particular attributes and circumstances. Here, we take religious claimants as we find them, assessing the coercive impact the government's actions on the individual claimant's ability to engage in a religious exercise, as he understands that exercise and the terms of his faith. This court has explained that a burden on a religious exercise rises to the level of being "substantial" when (at the very least) the government (1) requires the plaintiff to participate in an activity prohibited by a sincerely held religious belief, (2) prevents the plaintiff from participating in an activity motivated by a sincerely held religious belief, or (3) places considerable pressure on the plaintiff to violate a sincerely held religious belief — for example, by presenting an illusory or Hobson's choice where the only realistically possible course of action available to the plaintiff trenches on sincere religious exercise. Abdulhaseeb, 600 F.3d at 1315; see also Lyng v. Nw. Indian Cemetery Protective Ass'n, 485 U.S. 439, 450, 108 S.Ct. 1319, 99 L.Ed.2d 534 (1988); Thomas, 450 U.S. at 716-18, 101 S.Ct. 1425.

As the first two of these categories demonstrate, state laws or practices that require action or inaction in violation of a sincerely held religious belief amount to "substantial" burdens on the exercise of religion. As the third category shows, a burden can be "substantial" even if it does not compel or order the claimant to betray a sincerely held religious belief. It is enough that the claimant is presented with a choice in which he faces considerable pressure to abandon the religious exercise at issue. The term "substantial," after all, doesn't mean complete or total, so a "substantial burden" need not be a complete or total one. See, e.g., 17 The Oxford English Dictionary 67 (2d ed.1989) (substantial suggests an "ample or considerable amount"); see also Thomas, 450 U.S. at 718, 101 S.Ct. 1425. Take, for example, the case of the prison that fails to provide Jewish or Muslim inmates with food that satisfies their religious dietary restrictions. The prison may not formally require prisoners to violate their religious convictions — after all, the prison might say, there is no rule compelling prisoners to eat the food it provides, prisoners can purchase and supply their own — but surely the choice as presented remains a heavily freighted one. Abdulhaseeb, 600 F.3d at 1316-17; see also id. at 1325 (Gorsuch, J., concurring). Neither must the choice the government presents be so coercive as that to qualify as a substantial burden. Creating a situation that forces the religious claimant to choose between following the dictates of his faith and winning an important benefit or forgoing a considerable penalty is coercion enough. Thomas, 450 U.S. at 717-18, 101 S.Ct. 1425 ("important benefit"); Lyng, 485 U.S. at 450, 108 S.Ct. 1319 ("indirect" penalties and fines); see also Sherbert, 374 U.S. at 404, 83 S.Ct. 1790; Abdulhaseeb, 600 F.3d at 1315-16.

In this case, it doesn't take much work to see that Mr. Yellowbear has satisfied his obligations under both the sincere religious exercise and the substantial burden tests. After all, at the summary judgment stage — where we find ourselves, procedurally speaking in this case — Mr. Yellowbear is obliged to show merely that a reasonable fact finder could rule his way when viewing the evidence in the record in the light most favorable to him. And no one in this case even tries to question that Mr. Yellowbear's Northern Arapaho faith constitutes a religion. No one disputes Mr. Yellowbear's faith is sincerely held. No one questions either that access to a sweat lodge is a form of religious exercise for Mr. Yellowbear, or indeed for many who share his faith tradition. As Mr. Yellowbear has explained, it is his religious belief that a sweat lodge "is used to cleanse and purify our mind, our spirit, and our bodies ... upon leaving it is said that you are born again physically and spiritually." This court and others have long recognized that sweat lodges play just this role in many Native American religions — before RLUIPA did away with the inquiry we even held that the use of sweat lodges is "central and fundamental" to those faith traditions. Werner v. McCotter, 49 F.3d 1476, 1480 (10th Cir.1995) (collecting cases).2

Neither do we doubt a reasonable finder of fact could conclude the prison has substantially burdened Mr. Yellowbear's religious exercise. As Mr. Yellowbear understands his faith, it requires at least some access to a sweat lodge. The prison refuses any access. This isn't a situation where the claimant is left with some degree of choice in the matter and we have to inquire into the degree of the government's coercive influence on that choice. The prison's policy here falls easily within Abdulhaseeb's second category — flatly prohibiting Mr. Yellowbear from participating in an activity motivated by a sincerely held religious belief.

III

Though Mr. Yellowbear has met his summary judgment burden, that isn't enough to guarantee him victory. Instead, the burden of persuasion now shifts to the government. If it can show that the burden it has imposed on Mr. Yellowbear serves a "compelling governmental interest" and "is the least restrictive means" of furthering that interest, it can prevail yet. 42 U.S.C. § 2000cc-1(a). At summary judgment, of course, the government must satisfy this test when the facts in the record are viewed in the light most favorable to Mr. Yellowbear, as the non-movant. Abdulhaseeb, 600 F.3d at 1311. The district court held that the prison had succeeded in doing just this, but it is our duty to decide the question again de novo, and we find ourselves compelled to disagree. Id. While RLUIPA anticipates that its solicitude for religious exercise must sometimes yield to other competing state interests, the prison hasn't yet shown our case qualifies.

Perhaps the most important question we confront in the compelling interest inquiry concerns the level of generality at which our analysis should proceed. When a prison declines to accommodate a religious exercise, citing security and cost concerns (as the prison does here), should we assess the "compellingness" of those interests in the abstract? Or should we assess the "compellingness" of those interests (say) only as they appear in light of the particular circumstances of the case before us?

The more abstract the level of inquiry, often the better the governmental interest will look. At some great height, after all, almost any state action might be said to touch on "one or another of the fundamental concerns of government: public health and safety, public peace and order, defense, revenue," and measuring a highly particularized and individual interest "directly against one of these rarified values inevitably makes the individual interest appear the less significant." J. Morris Clark, Guidelines for the Free Exercise Clause, 83 Harv. L.Rev. 327, 330-31 (1969); see also Michael W. McConnell & Richard A. Posner, An Economic Approach to Issues of Religious Freedom, 56 U. Chi. L.Rev. 1, 53 (1989) ("[A] common pitfall [in religious liberty cases] is to consider the two sides of the balance at different levels of generality."); Richard H. Fallon, Jr., Strict Judicial Scrutiny, 54 UCLA L.Rev. 1267, 1323-25 (2007).

The text of RLUIPA and Supreme Court guidance, however, do not support such a lopsided inquiry. The statute says the government must prove the "compellingness" of its interest in the context of "the burden on that person" — suggesting an inquiry that proceeds in light of the particular burden the government has placed on the particular claimant before us. 42 U.S.C. § 2000cc-1(a) (emphasis added). The Supreme Court has confirmed this understanding, explaining that virtually identical statutory language in RFRA requires courts to "look[] beyond broadly formulated interests" and "scrutinize[] the asserted harm of granting specific exemptions to particular religious claimants." Gonzales v. O Centro Espirita Beneficente Uniao Do Vegetal, 546 U.S. 418, 431, 126 S.Ct. 1211, 163 L.Ed.2d 1017 (2006). Put simply, we must examine both sides of the ledger on the same case-specific level of generality: asking whether the government's particular interest in burdening this plaintiff's particular religious exercise is justified in light of the record in this case. Id.; see also O Centro Espirita Beneficiente Uniao Do Vegetal v. Ashcroft, 389 F.3d 973, 1019 (10th Cir.2004) (McConnell, J., concurring), aff'd sub nom. Gonzales v. O Centro Espirita Beneficente Uniao Do Vegetal, 546 U.S. 418, 126 S.Ct. 1211, 163 L.Ed.2d 1017 (2006).

So with that guidance in mind, what's the prison's compelling interest at play in this case? We see three possibilities.

A

The prison begins by suggesting that it cannot provide any access to a sweat lodge because sweat lodges involve the use of hot coals and fire and are in this way inherently unsafe for use in a prison environment (or at least they cannot be made safe at a feasible cost). By way of support, the prison points to cases in which other penitentiaries have refused to build or provide access to sweat lodges and other courts have upheld those decisions over religious liberty objections. See, e.g., Fowler v. Crawford, 534 F.3d 931 (8th Cir.2008); Allen v. Toombs, 827 F.2d 563 (9th Cir.1987).

But we've just witnessed one difficulty with this line of argument. It is our statutory duty to decide whether the prison's claimed safety and cost interests qualify as compelling in the context of particular cases, not in the abstract. Just because other prisons may have had a compelling interest in denying access to a sweat lodge in other circumstances doesn't necessarily prove, without more, that all prisons have a compelling interest in denying access to sweat lodges in all circumstances. Certainly, the prison hasn't given us any reason to think that conclusion follows here. Neither do the cases it cites: they hold only and more prosaically that the particular access sought by the particular claimants in each dispute was not feasible on the particular set of facts at issue.

A second and separate difficulty also comes quickly clear. Our prison has a sweat lodge. Before this appeal, the prison expressed no dissatisfaction with its operation. In all the summary judgment record amassed in the district court, we can find not a clue suggesting a hint of trouble. As we will see later, the prison even considered building a second sweat lodge in the protective custody unit, ultimately rejecting it for reasons having nothing to do with the expected safety of its operation. Put plainly, the argument advanced by the prison's lawyers on appeal about the "inherent dangers" of sweat lodges finds precisely no support in the evidence given by the prison's officials in district court. And such factually unsupported "post-hoc rationalizations" aren't the stuff of summary judgment victories in RLUIPA cases (or in most any other). Abdulhaseeb, 600 F.3d at 1318.

While saying this much, it's perhaps just as important to observe what we are not saying. We don't take any issue with the cases the prison cites rejecting RLUIPA challenges to sweat lodge access. Other prisons very well may have compelling reasons to refuse to build sweat lodges given their particular resource limitations and safety challenges. Neither do we mean to suggest that just because a prison chooses to open a sweat lodge for some prisoners it must forever maintain the lodge or provide unfettered access to it. Surely the granting of a religious accommodation to some in the past doesn't bind the government to provide that accommodation to all in the future, especially if experience teaches the accommodation brings with it genuine safety problems that can't be addressed at a reasonable price. If the rule were otherwise, it would only invite the unwelcome side effect of discouraging prison officials from granting the accommodation in the first place — a result directly at odds with RLUIPA's stated purpose. Instead, we mean to suggest here only that the appropriate level of generality requires us to assess the government's compelling interest assertions in light of each case as it comes — and that the summary judgment record in this case lacks any evidence suggesting that the current operation of this prison's sweat lodge is in any way problematic.

B

Retreating to the argument it did pursue in the district court, the prison argues that its ban on Mr. Yellowbear's access to the sweat lodge is warranted because moving him back and forth between the protective custody unit and the sweat lodge would require officials to "lock[] down a significant portion of the facility" to ensure Mr. Yellowbear isn't placed in "contact with other non-protective custody inmates" who might seek to do him harm. And this ferrying business, the prison says, has cost and administrative implications. Of course, cost and administrative considerations are often intertwined and they certainly are here. This isn't (again) a case in which the prison has evidence suggesting an inability to provide adequate security at any price. In the prison's view, a lock down can supply an acceptable level of security — but only at some marginal cost it considers too high.

Two problems, it seems to us, arise here, preventing us from saying that the government has carried its summary judgment burden of establishing a compelling interest as a matter of law.

In the first place, the prison does not even attempt to quantify the costs it faces, let alone try to explain how these costs impinge on prison budgets or administration. Instead, the prison simply asserts, flatly and without more, that the marginal costs are "unduly burdensome." But conclusory legalese (borrowed from far-flung substantive due process doctrine, no less) does no more to prove a compelling interest than post-hoc rationalizations unsupported by record evidence. RLUIPA's compelling interest test is a strict one: Congress borrowed its language from First Amendment cases applying perhaps the strictest form of judicial scrutiny known to American law. That test isn't traditionally the sort of thing that can be satisfied by the government's bare say-so. See, e.g., City of Richmond v. J.A. Croson Co., 488 U.S. 469, 500-01, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989).

To be sure, the Supreme Court has suggested that RLUIPA's "compelling governmental interest" test holds an unusual twist in the prison context. Though RLUIPA uses the same linguistic formulation for prison and land use cases, though it uses the same language found in RFRA, the Court has told us that "context matters." Cutter, 544 U.S. at 723, 125 S.Ct. 2113 (internal quotation marks and brackets omitted). So while courts do not usually afford much deference to the government when assessing whether its claimed interest is a "compelling" one in RFRA or even land use cases arising under RLUIPA, see, e.g., O Centro, 546 U.S. at 432-33, 126 S.Ct. 1211, in the prison context the Supreme Court has instructed us to apply "due deference to the experience and expertise of prison and jail administrators in establishing necessary regulations and procedures to maintain ... security ... consistent with consideration of costs and limited resources," Cutter, 544 U.S. at 723, 125 S.Ct. 2113. How all this fits together raises some interesting questions. See, e.g., Marci A. Hamilton, The Establishment Clause During the 2004 Term: Big Cases, Little Movement, 2004-2005 Cato Sup.Ct. Rev. 159, 169 (wondering whether Cutter seeks to reestablish "the low-level scrutiny" the Court applied to prison cases before RLUIPA).

But this much is clear and clearly enough for our purposes: the deference this court must extend the experience and expertise of prison administrators does not extend so far that prison officials may declare a compelling governmental interest by fiat. At summary judgment, the government must do more than assert an "undue burden" or a "compelling interest." When weighing the existence of a compelling interest, the deference due prison administrators may be enough to nudge a questionable case across the line, but it doesn't mean prison officials get to recognize compelling interests on their own. If that were the case, RLUIPA's supposedly strict judicial scrutiny would become no judicial scrutiny at all. See Werner, 49 F.3d at 1480 ("[T]he state must do more than simply offer conclusory statements that a limitation on religious freedom is required for security, health or safety in order to establish that its interests are of the highest order." (internal quotation marks omitted)); Koger v. Bryan, 523 F.3d 789, 800 (7th Cir.2008) ("We can only give deference to the positions of prison officials as required by Cutter ... when the officials have set forth those positions and entered them into the record."); Spratt v. R.I. Dep't of Corr., 482 F.3d 33, 40 (1st Cir.2007) ("[T]o prevail on summary judgment, [a prison] `must do more than merely assert a security concern.'" (quoting Murphy v. Mo. Dep't of Corr., 372 F.3d 979, 988 (8th Cir.2004))); Shakur v. Schriro, 514 F.3d 878, 890 (9th Cir.2008) (denying summary judgment because the record "contains no competent evidence as to the additional cost of providing Halal or kosher meat to ... Muslim prisoners").

Neither is this the end to the problems confronting the government's lock down rationale. It faces at least one more. Mr. Yellowbear has attested that prison-wide lock downs already occur on a daily and sometimes hourly basis to transport "specially housed" inmates — inmates in the geriatric unit, the women's unit, and (like himself) the protective custody unit — to other parts of the prison, including the medical wing. Mr. Yellowbear has presented affidavits, as well, from other inmates stating much the same thing. All of these uncontested statements we must take as true at summary judgment. And they can't help but suggest this question: If lock downs can be accomplished on a daily basis consistent with the prison's administrative concerns to facilitate inmates' medical needs, what compelling interest is served by refusing any lock downs ever to facilitate an inmate's religious needs? Put differently, why is this religious exemption offensive to the prison's putatively compelling no-lock-down interest when other secular exemptions are not?

A law's underinclusiveness — its failure to cover significant tracts of conduct implicating the law's animating and putatively compelling interest — can raise with it the inference that the government's claimed interest isn't actually so compelling after all. As the Supreme Court has said, it's sometimes hard to see how a law or regulation can "be regarded as protecting an interest of the highest order," as serving a compelling interest, "when it leaves appreciable damage to [the] supposedly vital interest unprohibited." Church of the Lukumi Babalu Aye, Inc. v. City of Hialeah, 508 U.S. 520, 547, 113 S.Ct. 2217, 124 L.Ed.2d 472 (1993) (internal quotation marks omitted); see also O Centro, 546 U.S. at 433, 126 S.Ct. 1211; United States v. Friday, 525 F.3d 938, 958 (10th Cir. 2008); Fallon, supra, at 1327; Eugene Volokh, Freedom of Speech, Permissible Tailoring and Transcending Strict Scrutiny, 144 U. Pa. L.Rev. 2417, 2420 (1996) (explaining how in the free speech context a "law's underinclusiveness ... may be evidence that an interest is not compelling, because it suggests that the government itself doesn't see the interest as compelling enough to justify a broader statute"). Evidence that the prison grants secular exceptions more readily than religious exemptions to a putatively compelling policy can raise the inference, too, that its most compelling interest may actually be discrimination against, or at least indifference to, the religious liberties of incarcerated persons — precisely the scenario RLUIPA identified as too prevalent in our society and sought to redress. See, e.g., Lukumi, 508 U.S. at 537-38, 113 S.Ct. 2217; Fraternal Order of Police Newark Lodge No. 12 v. City of Newark, 170 F.3d 359, 365 (3d Cir.1999); Fallon, supra, at 1327; Volokh, supra, at 2423 ("Underinclusiveness might suggest ... that the government's real interest wasn't the stated one but was rather just a desire to favor one form of speech over another, or to suppress offensive or otherwise disfavored speech.").

At the same time, it is important to acknowledge that inferences like these are not inevitable or irrebuttable. We know that few statutes pursue a single purpose at any cost, without reference to competing interests. See Genova v. Banner Health, 734 F.3d 1095, 1099 (10th Cir. 2013); John F. Manning, What Divides Textualists from Purposivists?, 106 Colum. L.Rev. 70, 74-75 (2006). Given this, it would be odd if the mere fact that a law contains some secular exceptions always sufficed to prove the government lacked a compelling interest in avoiding another exception to accommodate a claimant's religious exercise. If that were the case, the compelling interest test would seem nearly impossible to satisfy and RLUIPA's suggestion that religious accommodations must sometimes give way to other competing governmental interests would appear close to a dead letter.

When exactly is (and isn't) underinclusiveness enough to undermine the credibility of a claimed compelling interest? That question takes us into territory not yet fully charted, but at least this much we can say with confidence. A government can rebut an argument from underinclusion by showing that it hasn't acted in a logically inconsistent way — by (say) identifying a qualitative or quantitative difference between the particular religious exemption requested and other secular exceptions already tolerated, and then explaining how such differential treatment furthers some distinct compelling governmental concern.

Take an example. This court has said that the government's "supposedly vital interest" in preventing eagle deaths isn't undermined simply because the government has restricted intentional eagle killings more than accidental ones. Friday, 525 F.3d at 958-59. After all, trying to avoid accidental eagle deaths from electrocution by the power lines that crisscross the West is enormously more difficult than deciding to cancel a hunting expedition. Id. at 959. And surely the government has a compelling interest in not subjecting citizens to laws they can't realistically avoid breaking. So it is that in these circumstances, we have explained, a law's underinclusiveness might bespeak neither a shaky commitment to the asserted compelling interest nor any discriminatory intent. See id.; see also Fegans v. Norris, 537 F.3d 897, 906-07 & n. 4 (8th Cir.2008); Republican Party of Minn. v. White, 416 F.3d 738, 778 (8th Cir.2005).

Problem is, in this case the prison has made no effort (of any kind) to respond to Mr. Yellowbear's underinclusiveness evidence. Mr. Yellowbear's evidence about the regularity of secular lock downs dangles out there unanswered, inviting a reasonable fact finder to infer that the prison's desire to avoid lock downs isn't really as compelling as it insists, and that the prison may even harbor some animus or at least indifference to his religious exercise. To be sure, we can imagine responses the prison could have attempted that would have done much to address Mr. Yellowbear's underinclusiveness argument. What if, for example, the prison's resources are already stretched thin by lock downs for prisoners' medical needs, and the marginal cost of any further lock down would be the proverbial straw, overwhelming the prison's resources? What if not all lock downs are created equal, and those needed to facilitate sweat lodge access (say) require considerably more personnel for longer periods than lock downs needed to facilitate medical treatment? Maybe evidence along these or other directions could have mitigated or undermined the sort of inferences that arise from Mr. Yellowbear's underinclusiveness argument. But in this case we have none of that.

C

In a final effort to establish a compelling governmental interest, the prison aims in a different direction still. Even if it can easily accommodate Mr. Yellowbear, the prison insists that granting his request will lead other specially housed inmates to flood it with similar requests. As the prison puts it, Mr. Yellowbear's request "would be just the tip of the iceberg." And avoiding a slippery slope down to submerged troubles just out of present view, the prison suggests, amounts to a compelling interest all its own.

Again, however, the prison gives us too little to work with. It identifies nothing in the summary judgment record suggesting that there are untold numbers of specially housed prisoners lined up waiting to join Mr. Yellowbear. Or that accommodating them would stretch prison resources too far. Describing (and rejecting) a similarly speculative slippery slope argument in the RFRA context, the Supreme Court said that the claim "echoe[d] the classic rejoinder of bureaucrats throughout history: If I make an exception for you, I'll have to make one for everybody, so no exceptions." O Centro, 546 U.S. at 436, 126 S.Ct. 1211. The whole point of RFRA and RLUIPA is to make exceptions for those sincerely seeking to exercise religion. Id. ("RFRA operates by mandating consideration, under the compelling interest test, of exceptions to rules of general applicability." (internal quotation marks and brackets omitted)). It can't be the case that the speculative possibility that one exception conceivably might lead to others is always reason enough to reject a request for the first exception. Instead, the Court tells us, the feasibility of requested exceptions usually should be assessed on a "case-by-case" basis, taking each request as it comes: accommodations to avoid substantial burdens must be made until and unless they impinge on a demonstrated compelling interest. Id. The Court has recognized the possibility that there may be cases where that approach is inappropriate and the "need for uniformity precludes the recognition" of any exceptions to anyone, or where a well-documented slippery slope argument could support such a uniform rule. Id. But one thing is certain. To warrant any of that, the law requires considerably more than milquetoast musing that granting one request might lead to others.

IV

It turns out the prison has failed to carry not just one of its statutory burdens in this case, but both. Even assuming (without granting) that the prison could prove a compelling interest in denying Mr. Yellowbear any access to the sweat lodge (say, some amalgam of cost and security concerns), RLUIPA would still require it to show that its policy of no access, ever, represents the least restrictive means of accomplishing that interest. And here the prison falls short.

As part of its burden to show that its policy represents the least restrictive means available to further its putatively compelling interest, the government must of course "refute ... alternative schemes" suggested by the plaintiff to achieve that same interest and show why they are inadequate. United States v. Wilgus, 638 F.3d 1274, 1289 (10th Cir.2011). For his part, Mr. Yellowbear has offered (at least) two less restrictive alternatives that, he says, should satisfy the prison's related cost and security concerns. First, Mr. Yellowbear has presented an affidavit suggesting officials could avoid the need for any lock down at all by holding sweat lodge ceremonies early in the day, before the main yard (where the sweat lodge stands) opens to general population inmates. Second, he has suggested that the prison might build a second sweat lodge in the protective custody unit. Both alternatives, he argues, would allow the government to achieve its interest in avoiding any additional lock downs while facilitating his religious exercise.

For its part, the prison offers no meaningful reply to the first alternative. In some places, the prison's brief seems to suggest that it considered the idea and rejected it — and that this, by itself, is enough to discharge any burden under the least restrictive means test. But the government's burden here isn't to mull the claimant's proposed alternatives, it is to demonstrate the claimant's alternatives are ineffective to achieve the government's stated goals. In other places, the prison's brief takes a different tack, asking us to make much of the fact that the prison offered other alternatives to Mr. Yellowbear that he rejected (like a transfer to an out-of-state prison, away from friends and family). But this isn't sufficient either to discharge the government's burden. In fact, it is logically irrelevant: the claimant's rejection of alternatives the government offers doesn't address the question whether his suggested alternatives suffice to achieve the government's asserted compelling interest. Cf. Arthur Ernest Davies, A Text-Book of Logic 573 (1915) ("In order to refute an assertion, Aristotle says we must prove its contradictory; the proof, consequently, of a proposition which stood in any other relation than that to the original, would be an ignoratio elenchi.").3

As to Mr. Yellowbear's second proposed alternative (building a new sweat lodge in the protective custody unit), the prison has presented no evidence that the building or maintenance costs would be prohibitive or that the operations would be unsafe (quite at odds, again, with its lawyers' assertions about the inherent safety problems of sweat lodges). Instead, the prison argues that the only viable site in the protective custody unit would be so close to the general prison yard that Mr. Yellowbear and others would be seen by general population inmates as they enter and exit the lodge. And this, the prison contends, implicates another and new putatively compelling interest — this one in preventing Mr. Yellowbear from being seen by general population inmates because they may attempt to attack him if they can identify him.

With the introduction of a new and different claimed compelling interest, it seems we must return to the compelling interest inquiry we've already discussed, see supra pp. 56-62, to see how it holds up. In doing so, we quickly confront reason to question it: Mr. Yellowbear and other prisoners have attested, again without reply from the prison, that he and the other protective custody inmates are routinely seen by general population inmates at meal times. So the prison policy is underinclusive in yet another significant and potentially relevant way — the prison allows an exception for meals but not for religious exercises. And this raises by now familiar questions: Do prison officials really have a compelling interest in preventing Mr. Yellowbear from being seen by other inmates slightly more frequently than he already is? Is there any relevant difference between the exception for meal times and the exception Mr. Yellowbear has requested? Once again, the prison fails to provide any evidence or argument in response to these questions, leaving us at a loss to see how it could meet its compelling interest burden as a matter of law. Besides, Mr. Yellowbear has presented evidence that any anonymity interests the prison might have could be resolved less restrictively by holding the ceremonies in a protective custody unit sweat lodge on weekends, when general population prisoners (he says, again without dispute) won't be around to observe those who enter or leave the lodge — a potential less restrictive means the prison has (again) failed to refute.

V

So far the parties have contended at the level of absolutes. Mr. Yellowbear has sought some access to a sweat lodge. The prison has refused any access. But what happens if (or when) the discussion turns to questions of degree? What happens if the marginal burden on Mr. Yellowbear's religious exercise becomes much less significant — the prison allows him periodic access, say, but not as often as his faith suggests — and the marginal cost to the prison much greater — because ever more visits mean ever more expense? Surely the relative strength of the two parties' interests would appear quite different in that new light. Surely as well RLUIPA must account for that. Even accepting that the imposition on Mr. Yellowbear in this scenario might still qualify as a "substantial burden" (the government's policy would still prohibit that which his faith compels), the change in the relative strength of the parties' positions would reveal itself at later stages in the doctrinal analysis, with the government's interests now arguably appearing more "compelling" and its policies more "narrowly tailored." For now, however, these subtler (and admittedly more difficult) questions remain for the parties and district court to consider on remand. As litigated to us, the burden on Mr. Yellowbear's religious exercise is high (no access of any kind, ever, to a religious exercise) and the cost to the prison left undefined by the record and thus presumably low. In these circumstances, we don't doubt a reasonable trier of fact could find a RLUIPA violation.

The district court's grant of summary judgment on Mr. Yellowbear's RLUIPA claim seeking prospective injunctive relief under Ex parte Young is vacated and this case is remanded for further proceedings consistent with this opinion.