Opinion for the Court filed by Circuit Judge WILLIAMS.
Dissenting opinion filed by Circuit Judge WALD.
STEPHEN F. WILLIAMS, Circuit Judge:
A group of prisoners and publishers challenges the constitutionality of a statutory ban on the use of Bureau of Prisons funds to distribute sexually explicit material to prisoners. The statute is not enforced directly; instead, the Bureau has promulgated regulations defining the terms of the proscription and significantly narrowing its scope. The district court, analyzing the statute, ruled that it was facially invalid as a violation of the First Amendment and enjoined its enforcement. Finding that scrutiny should be directed to the substance of the regulations instead, and disagreeing with the district court’s evaluation, we reverse and remand.
* * *
Before 1996, federal regulations authorized prison wardens to reject a publication “only if it [was] determined detrimental to the security, good order, or discipline of the institution or if it might facilitate criminal activity.” 28 C.F.R. § 540.71(b). Sexually explicit material fell into this category if it “by its nature or content pose[d] a threat to the security, good order, or discipline of the institution, or facilitated criminal activity.” 28 C.F.R. § 540.71(b)(7). Under this standard, explicit heterosexual material was ordinarily admitted. See Thornburgh v. Abbott, 490 U.S. 401, 405 n. 6, 109 S.Ct. 1874, 104 L.Ed.2d 459 (1989).
In 1996 Congress passed the Ensign Amendment, which bars the use of Bureau of Prisons funds to pay for the distribution of commercial material that “is sexually explicit or features nudity.”1 See Omnibus Consolidated Appropriations Act of 1997, Pub. L. No. 104-208 § 614.2 Regulations later adopted by the Bureau assign rather narrow meanings to these terms: “nudity” means “a pictorial depiction where genitalia or female breasts are exposed”; “features” means that “the publication contains depictions of nudity or sexually explicit conduct on a routine or regular basis or promotes itself based upon such depictions in the case of individual onetime issues.” Even material that otherwise would be said to “feature nudity” is excepted if it contains “nudity illustrative of medical, educational, or anthropological content.” “Sexually explicit” means “a pictorial depiction of actual or simulated sexual acts including sexual intercourse, oral sex, or masturbation.” 28 C.F.R. § 540.72(b). Under these regulations, then, there is no restriction whatever on non-pictorial sexually explicit material.
[195]*195In 1997, three inmates, each denied receipt of either Playboy or Penthouse, filed separate suits alleging that the Ensign Amendment violated the First Amendment. Their suits were consolidated, along with similar suits filed by the publishers of those magazines and a publishing trade organization; the consolidated plaintiffs moved for injunc-tive relief. The district court, purporting to apply the test set out in Turner v. Safley, 482 U.S. 78, 107 S.Ct. 2254, 96 L.Ed.2d 64 (1987), found the Ensign Amendment facially invalid — implicitly granting plaintiffs’ motion for summary judgment (which, in fact, they never made) — and permanently enjoined its enforcement. 975 F.Supp. 365, 370 (D.D.C. 1997).
Both sides agree that Safley sets out the appropriate framework for reviewing government regulation of prisons.3 But before beginning that analysis, we must note that our first difference with the district court is as to the proper object of judicial scrutiny.
Plaintiffs ask for relief against both the Ensign Amendment and its implementing regulations. The district court seemed to assume that the statute itself has been and will be applied to these plaintiffs; accordingly, it directed its analysis primarily towards the statute. Id. at 368-70. But there is no suggestion that any warden does or will apply the statute directly; so far as appears, all enforcement is mediated through the regulations.
Insofar as plaintiffs attack the proscriptions of the statute not embodied in the regulations, they effectively pursue a pre-enforcement challenge. Even in the First Amendment context, such a challenge presents a justiciable controversy only if the probability of enforcement is “real and substantial.” Salvation Army v. Dep’t of Comm. Affairs, 919 F.2d 183, 192 (3d Cir.1990); see also Steffel v. Thompson, 415 U.S. 452, 460, 94 S.Ct. 1209, 39 L.Ed.2d 505 (1974). In the statutory borderland beyond the implementing regulations (i.e., the statute’s apparent ban on some non-pictorial material, its vaguer language, and its lack of any exception for medical or educational material), the prospect of enforcement appears completely insubstantial. It is as if the government had waived certain provisions of the law. And with such a waiver, as Salvation Army explicitly holds, there is no standing to challenge the waived provisions. See 919 F.2d at 192-93 (“[T]he district court should decline to provide an advisory opinion regarding the constitutionality of these provisions.”). See also Forsyth County, Georgia v. Nationalist Movement, 505 U.S. 123, 131, 112 S.Ct. 2395, 120 L.Ed.2d 101 (1992) (evaluation of facial challenge to statute must take into account construction by enforcing body). We therefore limit our focus to the substantive prohibitions of the regulations.4
Cases analyzing constitutional claims by those within governmental institutions such as prisons, public schools, the military, or the government workplace often open with the axiom that the boundaries of those institutions do not separate inhabitants from their constitutional rights. See, e.g., Safley, 482 U.S. at 84, 107 S.Ct. 2254 (“Prison walls do not form a barrier separating prison inmates from the protections of the Constitution.”); Connick v. Myers, 461 U.S. 138, 142, 103 S.Ct. 1684, 75 L.Ed.2d 708 (1983) (“[I]t has been settled that a State cannot condition public employment on a basis that infringes the employee’s constitutionally protected interest in freedom of expression.”); Tinker v. Des Moines Independent Community School Dist., 393 U.S. 503, 506, 89 S.Ct. 733, 21 L.Ed.2d 731 (1969) (“It can hardly be argued that either students or teachers shed their constitutional rights to freedom of speech or [196]*196expression at the sehoolhouse gate.”); General Media Communications, Inc. v. Cohen, (“GMC”) 131 F.3d 273, 276 (2d Cir.1997) (“The Constitution does not, of course, stop at the gates of a military base.”). This observation is invariably followed by the complementary principle that by their nature such environments must allow regulation more intrusive than what may lawfully apply to the general public. See Safley, 482 U.S. at 84-85, 107 S.Ct. 2254; Connick, 461 U.S. at 143, 103 S.Ct. 1684; Tinker, 393 U.S. at 507, 89 S.Ct. 733; GMC, 131 F.3d at 276. In these environments, the government is permitted to balance constitutional rights against institutional efficiency in ways it may not ordinarily do.5 See, e.g., Waters v. Churchill, 511 U.S. 661, 675, 114 S.Ct. 1878, 128 L.Ed.2d 686 (1994) (describing governmental power to restrict speech in the name of efficiency); Safley,. 482 U.S. at 88, 107 S.Ct. 2254 (noting balancing between First Amendment rights and governmental interests).
For the prison context, Safley directs courts to uphold a regulation, even one circumscribing constitutionally protected interests, so long as it “is reasonably related to legitimate penological interests.” 482 U.S. at 89, 107 S.Ct. 2254. We are to assess the overall reasonableness of such restrictions with attention to four factors: first, whether the restriction bears a “valid, rational connection” to the “legitimate governmental interest put forward to justify it,” such that the “asserted goal is [not] so remote as to render the policy arbitrary or irrational,” id. at 89-90, 107 S.Ct. 2254; second, whether inmates retain alternative means of exercising the circumscribed right, id. at 90,107 S.Ct. 2254; third, the costs that accommodating the right imposes on other inmates, guards, and prison resources generally, id.; and fourth, whether there are alternatives to the regulation that “fully accommodate[ ] the prisoner’s rights at de minimis cost to valid penological interests,” id. at 90-91,107 S.Ct. 2254. Although the factors are intended as guides to a single reasonableness standard, see id. at 89, 107 S.Ct. 2254; see also Thornburgh, 490 U.S. at 414, 109 S.Ct. 1874 (“The Court in [Safley] identified several factors that are relevant to, and that serve to channel, the reasonableness inquiry.”), the first factor looms especially large. Its rationality inquiry tends to encompass the remaining factors, and some of its criteria are apparently necessary conditions. Nothing can save a regulation that promotes an illegitimate or non-neutral goal. See Safley, 482 U.S. at 90, 107 S.Ct. 2254.
In this case, both proponents in congressional debate, and the government in its briefs here, assert as the goal an interest in the rehabilitation of prisoners. See, e.g., 142 Cong. Rec. H8261 (daily ed. July 24, 1996) (“Congress should not be fueling the sexual appetites of offenders, especially those who have been convicted of despicable sex offenses against women and children. Magazines that portray and exploit sex acts have no place in the rehabilitative environment of prisons, nor should we pay Bureau of Prison[s] staff to distribute them.”) (statement of Rep. Ensign); id. at H8262 (“The infamous serial killer Ted Bundy ... stated before his death his belief that pornographic materials directly contributed to his violent crimes. While a number of factors determine whether a prisoner will become a law-abiding citizen upon release from prison, cutting prisoners off from their sexually explicit magazines will certainly do no harm.”) (statement of Rep. Ensign). The next question is whether rehabilitation is a legitimate and “neutral” government interest within the meaning of Safley.
The legitimacy of the rehabilitative purpose appears indisputable. Indeed, the Supreme Court has often characterized rehabilitation as one of the primary goals of penal institutions. See, e.g., O’Lone v. Estate of Shabazz, 482 U.S. 342, 348, 107 S.Ct. 2400, 96 L.Ed.2d 282 (1987) (describing rehabilitation [197]*197as “valid penological objective[ ]”); Procunier v. Martinez, 416 U.S. 396, 413, 94 S.Ct. 1800, 40 L.Ed.2d 224 (1974) (identifying “substantial government interests of security, order, and rehabilitation”). At least since the mid-19th century, it is hard to imagine a prison system that did not seek to reduce the likelihood that prisoners would again transgress society’s norms. The inventors of the prison in New York and Pennsylvania framed its goals in these terms:
The penitentiary, free of corruptions and dedicated to the proper training of the inmate, would inculcate the discipline that negligent parents, evil companions, taverns, houses of prostitution, theaters, and gambling halls had destroyed. Just as the criminal’s environment had led him into crime, the institutional environment would lead him out of it.
David J. Rothman, The Invention of the Asylum: Social Order and Disorder in the New Republic 82-83 (rev. ed.1990).
In turning to “neutrality,” the district court looked at the statute itself, not the goal, and found it non-neutral. “[T]he Ensign Amendment is a content-based statute with a sole focus on the sexual nature of the publications it seeks to prohibit.” 975 F.Supp. at 368. The conclusion is correct — the Ensign Amendment is formulated in terms of content. But we do not understand that characteristic to run afoul of Safley’s neutrality requirement. Despite the apparent similarity in terms, “neutrality” in the Safley sense is quite different from the familiar First Amendment notion of “content-neutrality.” The Court most recently explained in Thornburgh that “neutral” here means no more than that “the regulation or practice in question must further an important or substantial governmental interest unrelated to the suppression of expression.” 490 U.S. at 415, 109 S.Ct. 1874 (quoting Martinez). There the Court found neutrality simply because in application of the regulations the prison administrators were to “draw distinctions between publications solely on the basis of their potential implications for prison security.” Id. It pointed out that the regulations struck down in Martinez, in contrast, barred writings that “unduly complain[ed]” or “magnified] grievances,” or expressed “inflammatory” views, thus inviting the prison authorities “ ‘to apply their own personal prejudices and opinions as standards for prisoner mail censorship.’ ” 490 U.S. at 416 n. 14, 109 S.Ct. 1874 (quoting Martinez); see also Martinez, 416 U.S. at 415, 94 S.Ct. 1800 (striking down regulations because no legitimate interest at all was offered as justification). Thus the Martinez regulations were “decidedly not ‘neutral’ in the relevant sense.” Thornburgh, 490 U.S. at 416 n. 14, 109 S.Ct. 1874.
The rehabilitative interest offered by the government here meets this rather thin neutrality requirement. The Court cannot have meant to insist upon neutrality in its classic First Amendment sense, as embodied, for example, in the dictum of Gertz v. Robert Welch, Inc., 418 U.S. 323, 339, 94 S.Ct. 2997, 41 L.Ed.2d 789 (1974), that “[ujnder the First Amendment, there is no such thing as a false idea.” In its role as an operator of prisons, the government surely does not lapse from the required neutrality if it bans “how-to” manuals on fraud and embezzlement (“Fraud for Dummies”). The idea that committing such crimes is not proper may not be “false” in the Gertz sense, but in the prison context the government may oppose it rather absolutely. This authority cannot be explained away on the theory that such a ban could withstand even the strictest scrutiny and that its validity therefore tells us nothing of the “neutrality” requirement; as voiced by the Court, the “neutrality” requirement is an absolute prerequisite. Accordingly, we think the Court’s actual application of the requirement, with its focus on the existence of some legitimate goal and on assuring that rules are in fact not framed so as to advance illegitimate or unvetted goals, must control our understanding of its meaning.
That government power to inculcate values conflicts with the libertarian premises of the First Amendment is clear, but such power is well established. Even outside special governmental institutions, the state has some authority to become a player in the marketplace of ideas. It may commission advertising urging children to stay in school or off drugs; it can sponsor anti-violence campaigns. See Regan v. Taxation with Repre[198]*198sentation, 461 U.S. 540, 548-49, 103 S.Ct. 1997, 76 L.Ed.2d 129 (1983); see generally Rust v. Sullivan, 500 U.S. 173, 193, 111 S.Ct. 1759, 114 L.Ed.2d 233 (1991) (“The Government can, without violating the Constitution, selectively fund a program to encourage certain activities it believes to be in the public interest, without at the same time funding an alternate program which seeks to deal with the problem in another way.”).
And within its own institutions, government’s power to pursue its legitimate goals is elevated: within reasonable limits, it may attach sanctions to or exclude speech that threatens its goals — even if those goals include promotion of particular values. That is the general lesson of cases such as Waters, 511 U.S. 661, 114 S.Ct. 1878, 128 L.Ed.2d 686 (upholding dismissal of employee for speech that might have caused disruption); Bethel School Dist. No. 403 v. Fraser, 478 U.S. 675, 685, 106 S.Ct. 3159, 92 L.Ed.2d 549 (1986) (upholding suspension of student who used indecent language in assembly speech); and Greer v. Spock, 424 U.S. 828, 96 S.Ct. 1211, 47 L.Ed.2d 505 (1976) (upholding exclusion from military base of civilians who wished to distribute political literature).
Of course this does not mean by a long shot that every government institution can pursue any value. Our century has seen more than enough of governments trying to act as engineers of human souls. But inculcation of values cannot be characterized as a suppression of expression in every context. The Supreme Court has in fact taken the position that democratic society depends on inculcation of democratic values. See, e.g., Arnbach v. Norwich, 441 U.S. 68, 76-77, 99 S.Ct. 1589, 60 L.Ed.2d 49 (1979) (stating that public schools are important “in the preparation of individuals for participation as citizens, and in the preservation of the values on which our society rests”); Brown v. Board of Education, 347 U.S. 483, 493, 74 S.Ct. 686, 98 L.Ed. 873, (1954) (“[Public education] is a principal instrument in awakening the child to cultural values, in preparing him for later professional training, and in helping him to adjust normally to his environment.”). The generality of Safley's reasonableness test, and common sense as well, suggest that the flexibility open to the political branches must be at its apogee in institutions for the care and custody of those who have already transgressed society’s norms. If “promoting respect for authority and traditional values” among schoolchildren is a legitimate interest of government and not inherently a form of speech suppression, see Board of Education, Island Trees Union Free School Dist. No. 26 v. Pico, 457 U.S. 853, 864, 102 S.Ct. 2799, 73 L.Ed.2d 435 (1982) (plurality opinion), surely the same is true of promoting such a goal among criminals.
Prisoners of course differ from the other examples of individuals entangled with the government in that they have no recourse to a private sphere. For them, there is no outside. Judges plainly must bear in mind the total occupation of prisoners’ lives by the state. But prisoners find themselves in the maw of the state not because they have been drawn there by the promise of government wages, nor because a statute imposes a universal requirement of education, but because they have broken the law. Moreover, even government workers have suffered a similar eclipse of the private sphere; the government’s ability to fire them for speech that threatens workplace disruption is not limited to speech within the workplace. See, e.g., Pickering v. Board of Education, 391 U.S. 563, 88 S.Ct. 1731, 20 L.Ed.2d 811 (1968) (analyzing letter to the editor). See generally Note, The Costs of Agencies: Waters v. Churchill and the First Amendment in the Administrative State, 106 Yale L.J. 1233, 1239-40 (1997) (discussing scope of governmental authority). Accordingly, we conclude that rehabilitation, and such character-moulding as may be implicit therein, constitute legitimate and neutral goals as those are understood in Safley.
We thus turn to whether there is a “valid rational connection” between the interest and the regulations. Prison jurisprudence is not well enough developed to indicate precisely how demanding the requirement of rational means-end connection is. But the similarity between Safley’s phrasing and the language of rational basis review suggests to us that, as far as the [199]*199means-end fit is concerned, Safley’s standard is, if not identical, something very similar. Compare, e.g., Nordlinger v. Hahn, 505 U.S. 1, 11, 112 S.Ct. 2326, 120 L.Ed.2d 1 (1992) (“[T]he relationship of the classification to its goal is not so attenuated as to render the distinction arbitrary or irrational.”) with Safley, 482 U.S. at 89-90, 107 S.Ct. 2254 (“[T]he logical connection between the regulation and the asserted goal is so remote as to render the policy arbitrary or irrational.”).
The legislative judgment is that pornography adversely affects rehabilitation. It does not matter whether we agree with the legislature, only whether we find its judgment rational. The question for us is not whether the regulation in fact advances the government interest, only whether the legislature might reasonably have thought that it would. See, e.g., Nordlinger, 505 U.S. at 11, 112 5.Ct. 2326.
We think that the government could rationally have seen a connection between pornography and rehabilitative values. Congress might well perceive pornography as tending generally to thwart the character growth of its consumers. One current exposition of this view sees pornography as treating women purely as objects of male sexual gratification. See, e.g., Catharine A. MacKinnon, Only Words 108-10; MacKinnon, “Francis Biddle’s Sister: Pornography, Civil Rights, and Speech,” in Feminism Unmodified 163, 174 (1987); Martha C. Nussbaum, “Objectification,” 25 Phil. & Pub. Affairs 249, 283-86 (1995); Cass Sunstein, The Partial Constitution 257-90 (1993). But this viewpoint shares at least a core with ideas that have a lineage of a few centuries, perhaps millennia, stressing the desirability of deferring gratification, ■ of sublimation of sexual impulses, of channeling sexual expression into long-term relationships of caring arid affection, of joining eros to agape. The supposition that exclusion of pornography from prisons will have much of an impact in this direction may be optimistic, but it is not irrational.
There is, of course, no “record evidence,” and certainly no sophisticated multiple regression analyses or- other social science data, to support this belief — a fact our dissenting colleague finds fatal. See Dissent at 207-08. We do not think, however, that common sense must be the mere handmaiden of social science data or expert testimonials in evaluating congressional judgments. Quite the opposite: scientific studies can have a corrective ■ effect by establishing an apparently implausible connection or refuting an apparently obvious one, but, subject to such corrections, conformity to commonsensical intuitive judgments is a standard element of both reasonableness and rationality. Compare D.N. McCloskey, The Rhetoric of Economics 44 (1985) (“Not all regression analyses, are more persuasive than all moral arguments; not all controlled experiments are more persuasive than all introspections. People should not discriminate against propositions on the basis of epistemological origin. There are some subjective, soft, vague propositions that are more persuasive than some objective, hard, precise propositions.”). Nor do we think Safley requires more. In that case, the Court scoured the record for evidence of a rational link between the asserted security interests and the marriage ban because common sense does not suggest any. See Safley, 482 U.S. at- 97-100, 107 S.Ct. 2254. Here, the regulations restrict prison consumption of publications that implicitly elevate the value of the viewer’s immediate sexual gratification over the values of respect and consideration for others. Common sense tells us that prisoners are more likely to develop the now-missing self-control and respect for others if prevented from poring over pictures that are themselves degrading and disrespectful.
Nor can we find irrationality in a congressional fear that pornography may set back rehabilitation in the narrower sense, more directly related to recidivist activity. The, briefs, particularly that of the. amicus curiae, call our attention to studies purporting either to document or to demolish the causal link between pornography and sex crimes.6 [200]*200Without attempting a comprehensive summary, we note a few. One concludes that exposing “already angered men” (probably not especially rare in prison) to nonviolent pornography can lead to short-term increases in their aggressiveness. Edward Donnerstein, et al., The Question of Pornography 40-48 (1987). Another finds that “pornography provides ideational support for rape_” Larry Baron & Murray A. Straus, Four Theories of Rape in American Society 185 (1989). Baron & Straus’s primary evidence for this claim is an observed correlation between the circulation rates of sex magazines and reported rape rates. While they are hesitant to defend a “cause-effect” link between pornography and rape, they do see the correlation as explained by “the presence of a hypermaseuline or macho culture pattern.” Id. at 186. This cultural pattern, Baron and Straus hypothesize, causes both higher incidence of rape and higher consumption of pornography. Of course, it seems quite likely that a culture and its manifestations have a mutually reinforcing relationship, so that a prohibition of pornography is a reasonable element of a struggle against machismo and its ill effects.
Finally, there is a significant body of research showing that long-term exposure to pornography, particularly pornography containing scenes of “aggressive sexuality,” can make its (male) audience more aggressive, more tolei'ant of violence against women, and more susceptible to myths about rape, such as the notion that women can enjoy being raped. See Neil Malamuth, “Aggression against Women: Cultural and Individual Causes,” in Malamuth & Donnerstein, 19, 32-39; see also Mike Allen et al., “Exposure to Pornography and Acceptance of Rape Myths,” 45 J. Communication 5 (1995) (“Although the experimental studies demonstrate that violent pornography has more effect [inducing rape myth acceptance] than nonviolent pornography, nonviolent pornography still demonstrates an effect.”) See also Richard A. Posner, Sex and Reason, 366-71 (1992) (noting inconclusiveness of social science findings on net effects of pornography on incidence of rape; pornography may tend to increase rape by its ideological and aphrodisiac effects, but may tend to decrease it by enhancing a substitute (masturbation), the balance of these effects being unknown).
The point of this is not to suggest that a causal link has been shown. The array of academic authority on the other side is at least as substantial, and quite possibly more so. But even undertaking to weigh the competing scholarship would misconceive the judicial role. Dealing with legislative judgments about rehabilitation, the Supreme Court has said that “[w]hen Congress undertakes to act in areas fraught with medical and scientific uncertainties, legislative options must be especially broad and courts should be cautious not to rewrite legislation, even assuming, arguendo, that judges with more direct exposure to the problem might make wiser choices.” Marshall v. United States, 414 U.S. 417, 427, 94 S.Ct. 700, 38 L.Ed.2d 618 (1974) (emphasis in original); see also Kansas v. Hendricks, 521 U.S. 346, 117 S.Ct. 2072, 2081 n. 3, 138 L.Ed.2d 501 (1997). And in upholding a statute barring provision of sexually explicit material to minors, the Court noted that “while these studies all agree that a causal link [between exposure to explicit material and impaired ethical and moral development] has not been demonstrated, they are equally agreed that a causal link has not been disproved either.” Ginsberg v. New York, 390 U.S. 629, 642, 88 S.Ct. 1274, 20 L.Ed.2d 195 (1968) (internal quotation omitted); see also American Booksellers Ass’n v. Hudnut, 771 F.2d 323, 329 n. 3 (7th Cir.1985). The same uncertainty prevails here, and suffices to place the legislative judgment within the realm of reason under the standards applicable to the political branches’ management of prisons.
The evidence is simply not conclusive on the efficacy of a ban on pornography in pro[201]*201moting prisoners’ rehabilitation. For judges seeking only a reasonable connection between legislative goals and actions, scientific indeterminacy is determinative.
The remaining Safiey factors further substantiate the overall reasonableness of the regulations. The second involves the prisoners’ alternative means of exercising the right at stake. Of course if the “right” at stake is defined in terms of the materials excluded by the ban, any regulation will come up short. The Court accordingly observed in Thorn-burgh that the relevant right “must be viewed sensibly and expansively,” 490 U.S. at 417,109 S.Ct. 1874, though it stated no method for actually carrying out the definitional process. In Thornburgh, where the regulation barred all “sexually explicit material which by its nature or content poses a threat to the security, good order, or discipline of the institution, or facilitates criminal activity,” id. at 405 n. 5, 109 S.Ct. 1874, the Court found the second factor “clearly satisfied” because “the regulations [there at issue] permit a broad range of publications to be sent, received, and read,” id. at 418, 109 S.Ct. 1874. In so doing, it made no inquiry into the scope of sexually explicit material not covered by the ban. Thus, unless there is some minimum entitlement to smut in prison, the origins of which must be obscure, this factor can hardly be fatal.7
The third Safiey factor concerns the adverse impact “on guards and other inmates, and on the allocation of prison resources” of accommodating the right claimed. 482 U.S. at 90,107 S.Ct. 2254. It thus seems in part a restatement of the deferential balancing called for under the first factor: if Congress may reasonably conclude that pornography increases the risk of prison rape, then the adverse impact is substantial. Accommodating the right poses a threat to the safety of guards and other inmates. The issue of the allocation of prison resources we take up below; we think the third and fourth factors are strongly interrelated.
The fourth Safiey factor poses the question whether there are alternatives that can accommodate that right “at de minimis costs to valid penological interests.” Safiey, 482 U.S. at 91, 107 S.Ct. 2254. This is not, the Court emphasized, a “ ‘least restrictive alternative’ test.” Id. at 90, 107 S.Ct. 2254. The most obvious alternative is a detailed prisoner-by-prisoner (and presumably publication-by-publication) sifting to determine whether a particular publication will harm the rehabilitation of a particular prisoner. The costs of this approach seem far from de minimis. Apart from the purely administrative burden of such a process, the exercise of such broad discretion seems highly conducive to just the sort of conflicting decisions that give bite to vagueness concerns. And the reality is that prisoners are likely to pass pornographic materials about, so that prisoner-by-prisoner determinations may run aground on the “ripple effect” that the Court mentioned in Saf-iey, 482 U.S. at 90, 107 S.Ct. 2254. Even if pornography could be directed only to those not likely to be adversely affected, it could find its way to others, interfering with their rehabilitation and increasing threats to safety.
Of course the flip side of avoiding the costs and hazards of a grant of broad discretion to wardens is the possibility of preventing delivery of materials that, either generally or for a specific set of (effectively segregated) prisoners, pose little or no threat of the harm Congress sought to diminish. Outside prison walls this concern is addressed by the over-breadth doctrine. Here it seems to us captured in part by the demand for a rational relation between the regulation and the goal, and by consideration of alternatives under the fourth “factor.” Thus, the objections to a more flexible standard — such as one that would vest discretion in wardens to prevent delivery of materials in any case where in his judgment it would impair a specific prisoner’s rehabilitation — go a long way to rebut the claim.
The dissent, nonetheless, seems to regard only such a case-by-case approach as consti[202]*202tutional, and makes much of various hypothetical situations in which the categorical ban could unneccessarily exclude certain publications. See Dissent at 210-213. To the extent that these hypothetical assume direct enforcement of the Ensign Amendment, the likelihood of such enforcement is, as we have noted, quite remote and not before us in this case. Further, the actions of the Bureau of Prisons also suggest that the likelihood of this sort of exclusion under the actual regulatory scheme is remote. The Bureau’s Program Statement, for example, listing examples of permissible publications, cites National Geographic; Our Bod[ies], Our Selves; Sports Illustrated (Swimsuit Edition); and the Victoria’s Secret Catalog. P.S. 5266.07, at 7. We find it all but impossible to believe that the Swimsuit Edition and Victoria’s Secret pass muster while Michelangelo’s David or concentration camp pictures fail; nor has there been any suggestion that any prison official has attempted to implement such a bizarre interpretation. As there is no allegation that women in segregated women’s prisons are demanding Playboy, neither the possibility of such demand nor the Bureau’s possible response assists plaintiffs in their overbreadth claim.8
Leaving aside these possible fringe applications of the regulation, we again note that the regulation by its terms only restricts pictures; a prisoner may read anything he pleases. The dissent’s appeal to the value of ideas, pointing by way of example to the vistas opened for Malcolm X by his prison reading, see Dissent at 210-211, thus makes little sense here. Congress could, we think, reasonably be skeptical of the proposition that study of the pictures in Penthouse would help a prisoner become “an intelligent customer in the marketplace of ideas.” If anything, the ban could strengthen the impact of literature by removing the distraction of pictures that in almost every case would be pornographic.
In any event, even under conventional overbreadth doctrine outside of prison, over-breadth claims by those on the margins of pornography have fared poorly. See, e.g., Young v. American Mini Theatres, Inc., 427 U.S. 50, 61, 96 S.Ct. 2440, 49 L.Ed.2d 310 (1976) (“Since there is surely a less vital interest in the uninhibited exhibition of material that is on the borderline between pornography and artistic expression than in the free dissemination of ideas of social and political significance ... we think this is an inappropriate case in which to adjudicate the hypothetical claims of persons not before the Court.”) And regardless of the type of speech at issue, narrow or improbable excesses of a statute are not a ground for invalidation before application shows then-reality:
[Wjhere a statute regulates expressive conduct, the scope of the statute does not render it unconstitutional unless its over-breadth is not only “real, but substantial as well, judged in relation to the statute’s plainly legitimate sweep.” Even where a statute at its margins infringes on protected expression, “facial invalidation is inappropriate if the ‘remainder of the statute ... covers a whole range of easily identifiable and constitutionally proseribable ... conduct....’”
Osborne v. Ohio, 495 U.S. 103, 112, 110 S.Ct. 1691, 109 L.Ed.2d 98 (1990) (citations omitted) (ellipses in original). Accordingly, such problems of application can be left to the time when, if ever, they arise.
We find that the statute and regulation satisfy Safley’s demand for reasonableness, scoring adequately on all four factors.
Plaintiffs argue that if we are unpersuaded by the district court’s analysis we should remand the case to allow them to introduce evidence supporting their position. But that suggestion misconceives the legal issue under Safley. The question is not whether curtail[203]*203ment of pictorial smut will advance the prisons’ rehabilitative project, but whether Congress could reasonably have believed that it would do so. The studies cited above, coupled with Congress’s implicit appeal to ethical norms against the undue stimulation of carnal appetites, indicate the reasonableness of such a belief. If the Bureau of Prisons applies the regulations in a manner that would violate the Safley priciples, of course, that would be another case. Accordingly, the district court should have granted the defendants’ motion to dismiss plaintiffs’ complaint insofar as that complaint rested upon Safley.
Besides the overbreadth issue, which we have already addressed in the Safley context, plaintiffs have raised two additional theories — excessive vagueness and a claim that the regulations’ distinction between pictorial and verbal works is content-based. In fact, the distinction between pictorial and verbal works seems more directed to form than to content, and in any event we have seen that Safley’s neutrality requirement both displaces conventional First Amendment strictures on content-based limits and is satisfied by these regulations. Although Safley may well function as an all encompassing free speech test for the circulation of reading materials in prison, supplanting otherwise applicable First Amendment doctrine, it may be that plaintiffs’ vagueness claim has independent force. If it did, of course, the role of the Bureau of Prisons, as manager rather than as law enforcer, might well play an important role in resolution of the claim. Accordingly, as the district court has not addressed these vagueness issues, we remand the case for it to do so. In the meantime, we lift the permanent injunction that it has imposed on enforcement of the regulations.
So ordered.