concurring in part and dissenting in part.
Although I join the court’s conclusion that the plaintiffs’ “internal mail claim” falls short as a matter of law, I regret that I do not share their assessment of the “in-person association claim.” At root, this claim is about the plaintiffs’ right to treatment for their condition; it is what distinguishes them from prisoners. Even if security concerns trump almost all other constitutional interests of convicted criminals, the same is not true of those suffering from a mental disorder. As I explain below, I do not believe that the naked incantation of the word “security” is enough to relieve the staff at the Rushville Treatment and Detention Center of its duty to exercise professional judgment when it makes decisions that affect the rehabilitative aims of the facility. I would reverse the district court’s grant of summary judgment on this point, and thus, I respectfully dissent.
I
The plaintiffs are sexually violent persons (SVPs) who have been civilly committed to the Rushville facility, which is located in west central Illinois. The SVP designation means that the person “suf*885fers from a mental disorder that makes it substantially probable that the person will engage in acts of sexual violence.” 725 ILCS § 205/5(f). Recognizing that this behavior stems from a mental disease, the Illinois Department of Human Services, Sexually Violent Persons Treatment and Detention Facility has as its stated mission the goal of providing “specialized treatment that promotes a personal responsibility and pro-social behavioral change” so that “[a]ll residents [may be] released and successfully returned to their communities.” Illinois Department of Human Services, Sexually Violent Persons Treatment and Detention Facility, Resident Handbook 8 (August 2008). Unfortunately, according to plaintiffs, these laudable aims are far from being realized. Instead, plaintiffs say, Rushville has adopted a policy under which their social interaction is limited to the same 25 other patients for between 20 and 22 hours a day, for practically the entire duration of their commitment to the facility (which often runs for years, if not decades). This severe restriction on human interaction, they assert, stunts the very treatment that should be the cornerstone of one’s residency at Rushville.
In evaluating Rushville’s policy, we must not lose sight of the legal justification for the detention of its residents. As the Supreme Court has stated, civil commitment may not “become a ‘mechanism for retribution or general deterrence.’ ” Kansas v. Crane, 534 U.S. 407, 412, 122 S.Ct. 867, 151 L.Ed.2d 856 (2002) (quoting Kansas v. Hendricks, 521 U.S. 346, 372-73, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997) (Kennedy, J., concurring)). Indeed, because Illinois’s SVP procedures “recommend[ ] treatment if such is possible” and “permit[ ] immediate release upon a showing that the individual is no longer dangerous or mentally impaired,” it is beyond dispute that the purpose of the plaintiffs’ detention is not punitive. Hendricks, 521 U.S. at 368-69, 117 S.Ct. 2072. Although the plaintiffs are not free to leave Rushville until they are deemed cured, they are not prisoners in the traditional sense; rather, their commitment is meant to be rehabilitative and aimed at the goal of their ultimate release.
It is against this backdrop that I turn to the plaintiffs’ claims in this case. They assert that even though undisputed testimony has established that “the ability of residents to engage in social interaction with a broad array of other residents ... is crucial to creating a positive therapeutic atmosphere,” the Rushville facility has decided to isolate the six housing units within Rushville from each other without regard to these therapeutic effects and without input from the clinical staff. The majority has several responses to these allegations. First, the majority questions the basic premise of the plaintiffs’ argument: It questions whether Rushville’s policies really have any effect on treatment, noting that “[i]t cannot be ... that all decisions that have an impact on detainees are treatment decisions.” Ante at 882 (emphasis in original). This trivializes the plaintiffs’ point: They are not saying that every decision, no matter how inconsequential (bedtime at 9 pm, commissary hours, or the like) is automatically a treatment decision. They are saying, with the support of expert testimony, that the particular policies they are challenging have a profound impact on treatment, and they are saying that these policies have not been developed with the proper professional input.
The majority also suggests that even if the amount of social interaction has an effect on the plaintiffs’ treatment (a point we must assume as true at this stage of the litigation), the “Rushville detainees have opportunities to associate in-person with approximately one-hundred and fifty detainees each week.” Ante at 881-82. This statistic, however, masks a much *886more troubling reality: For all hours of the day except for, at most, four hours, the plaintiffs spend all of their time with the same 25 people (their pod-mates). Then, for two to three hours per day, the plaintiffs get “yard” or “gym” time, which they share with their unit. Practically, during this recreational time the residents’ social circle is expanded to include only the other two 25-person pods in their unit. The patients also have small group therapy for an average of less than one hour a day. But once again, these episodic sessions are limited to same small set of patients each time, and given the philosophy dictating unit assignment, they probably do not represent an increase in social interaction. The far more sporadic encounters with people outside the unit described by the majority (such as a movie every other month, three summer picnics, and a Christmas concert) add little, as the majority acknowledges. Ante at 881-82.
Importantly, as I have already noted, the plaintiffs have presented unrebutted expert evidence suggesting that the amount and nature of the actual (and undisputed) social interaction Rushville offers is inadequate to promote their “release[] and successful ] return[ ] to their communities.” It is also telling that the efforts by Rushville’s own clinical staff to increase the amount of social interaction between patients have been quashed by the facility’s administrative staff. This raises a crucial threshold question that the majority’s analysis has overlooked: Who is entitled to decide whether a policy is (1) purely related to treatment, (2) purely related to security, or (3) related to both?
One consequence of the important distinction between punitive incarceration and civil commitment is that decisions that have more than an incidental impact on the rehabilitative aims of a civil commitment facility must be made with input from clinical professionals (in addition to, rather than to the exclusion of, others such as security experts). This requirement comes straight from Youngberg v. Romeo, 457 U.S. 307, 102 S.Ct. 2452, 73 L.Ed.2d 28 (1982), the case that sets the rules for assessing the validity of the policies set by civil commitment facilities. Youngberg noted that the federal Constitution requires “that the courts make certain that professional judgment in fact was exercised.” 457 U.S. at 321, 102 S.Ct. 2452 (emphasis added). Following that lead, we have similarly emphasized the need for professional judgment in balancing “progress toward that goal that justified plaintiffs’ commitment” and “institutional security.” Johnson by Johnson v. Brelje, 701 F.2d 1201, 1209 (7th Cir.1983); see also Allison v. Snyder, 332 F.3d 1076, 1079 (7th Cir.2003).
The majority argues that our decision in West v. Schwebke, 333 F.3d 745 (7th Cir.2003), obviated the need for any such clinical input; it focuses on the statement that “ ‘if at trial defendants can establish that their use of seclusion was justified on security grounds, they will prevail without regard to the question whether extended seclusion is justified as treatment.’ ” Ante at 883 (quoting Schwebke, 333 F.3d at 748). But this overlooks the fact that in Schwebke we also emphasized the need for “considered judgment,” 333 F.3d at 748, and so we actually remanded to the district court to determine whether the policy at issue was, in fact, justifiable on either security or treatment grounds. 333 F.3d at 748-49 (noting conflicting evidence on whether policy was “appropriate from a security perspective”). Because it was an open question whether the policy was justifiable on security or treatment grounds, we had no occasion to reach the issue of who was entitled to make such a determination and whether a policy affecting treatment can be re-branded — without any input from a clinical professional' — as one that is only *887security-related. One powerful reason to reject a reading of Schwebke that implies that the word “security,” once uttered by the administrative staff of a civil commitment facility to justify a policy, trumps all treatment needs, is that such a ruling would be inconsistent with the Supreme Court’s directives in Crane and Youngberg. This is not to say that security does not have an important role to play in facilities such as Rushville; it is merely to recognize that security and treatment goals both have a part to play. The job of reconciling these two interests does not belong to the courts. To the contrary, it is precisely because courts are ill-equipped to “second-guess the expert administrators on matters on which they are better informed,” Youngberg, 457 U.S. at 323, 102 S.Ct. 2452 (quoting Parham v. J.R., 442 U.S. 584, 607, 99 S.Ct. 2493, 61 L.Ed.2d 101 (1979)), that the Supreme Court held that qualified professionals from both the security and the treatment sides must have a say in matters that simultaneously implicate both treatment and security.
Summary judgment in the defendants’ favor can be justified only if the clinical staff at Rushville either (1) acceded to the view that the facility’s social interaction policy does not implicate substantial rehabilitative concerns (a professional judgment that we would review under Young-berg's deferential standard) or (2) have had an appropriate amount of input into a decision-making process that balances the rehabilitative concerns against the security and administrative interests of the facility. A policy that emerged from such a process would pass muster so long it complied with minimum professional standards and was otherwise constitutional. Thus, although it may be true that “[sjeeurity decisions do not violate Youngberg just because they restrict treatment options,” ante at 883, security decisions that restrict treatment options without any input from clinical staff do violate Youngberg. The majority appears to concede that the clinical staff neither agreed that the social interaction policy had nothing to do with rehabilitation nor did it have any input into the policy. Ante at 882-83 (“[Tjhere is no serious argument that the contested restrictions are the product of [the exercise of professional judgment]”); ante at 883 (adopting flat rule that security decisions never violate Youngberg only because “they restrict treatment options”). That is why this aspect of the case should be moving forward.
The undisputed facts in the record before us are not sufficient to permit a judgment on the question whether the policies challenged in this case were valid. There is a material question of fact on the question whether Dr. Jumper, Rushville’s Clinical Director, made a reasoned professional judgment that he and his staff had no need to participate in the process that set the policy regarding social interaction among units. Dr. Jumper insisted in his deposition that he “is not involved in the decisions regarding which units use yard,” that “no one on the clinical staff is involved in that decision” and that he is “not involved in any of the decisions regarding who goes on the yard.” He has changed his tune a bit in his brief, where he now says that his “decision not to be involved ... was in fact evidence of his exercise of discretion.” But it is well-established that people cannot rewrite or revise their depositions in later affidavits or briefs. In addition, the plaintiffs have presented other evidence that the clinical staff tried to change the social interaction policies but their efforts were rebuffed. Contradictorily, Dr. Jumper testified that “clinical reasons” are “irrelevant” because “the operational needs of the facility would supercede.” This evidence might be interpreted by a reasonable jury in either of two ways: They might conclude that Dr. Jumper did not consider the clinical impact of the security policy, but they might equally infer that *888Rushville had put him in his place, and he knew that he and his clinical staff were excluded from the decision-making process by its operational staff.
The extent of the clinical staffs participation in the decision-making process is also highly contested. The plaintiffs have presented significant evidence suggesting that clinical staff had little, if any, input into policy decisions regarding social interaction, and that their suggestions often fell on deaf ears. They have evidence suggesting that no clinical staff member was involved in yard time decisions, that clinicians expressed serious doubts about the administrative personnel’s decisions to decrease unit interaction, and that their suggestions to increase social interaction were unapologetically rejected by security staff. In response, Rushville argues that the clinical staffs input into placing new residents into their respective housing units and pods is an adequate substitute for input into setting the overall policies. Again, a trier of fact might ultimately conclude (depending on how much meaningful interaction occurred) that such an initial assessment is adequate to compensate for the clinical staffs ongoing inability to effect change in the social interaction policies, but it is not our role to resolve that issue. In light of this evidence, I cannot conclude that summary judgment was appropriate on this claim. A reasonable fact-finder could find that Dr. Jumper and his clinical staff were excluded from a decision-making process to which they should have been party. This claim should proceed to trial.
II
As I noted at the outset, I join my colleagues’ assessment of the plaintiffs’ “internal mail” claim. The requirement to use the United States Postal Service for inter-unit mail is not a sufficiently substantial “impingement” of the detainees’ constitutional rights. This is not a case like Lindell v. Frank, 377 F.3d 655 (7th Cir.2004), where the challenged policy (there, a publishers-only restriction) clearly violates an expressive right. Nothing in this policy limits the inmates’ ability to communicate, or imposes a content- or viewpoint-based restriction on their correspondence.
For these reasons, I respectfully Dissent from the court’s resolution of the “in-person association” claim, which I would remand for further proceedings.