dissenting:
I respectfully dissent from the majority’s result and from much of its analysis, largely on the basis ably expressed by Judge DeMoss in dissent. I agree, however, with the majority’s conclusion that United States v. Wilson, 36 F.3d 1298 (5th Cir.1994), is in error and must be overruled.
Maryland v. Buie, 494 U.S. 325, 110 S.Ct. 1093, 108 L.Ed.2d 276 (1990), is no exception to the longstanding view that “[t]he touchstone of the Fourth Amendment is reasonableness, and the reasonableness of a search is determined ‘by assessing, on the one hand, the degree to which it intrudes upon an individuars privacy and, on the other, the degree to which it is needed for the promotion of legitimate governmental interests.’ ” United States v. Knights, 534 U.S. 112, 118-19, 122 S.Ct. 587, 151 L.Ed.2d 497 (2001) (quoting Wyoming v. Houghton, 526 U.S. 295, 300, 119 S.Ct. 1297, 143 L.Ed.2d 408 (1999)). Buie established that a search very much like the present one was reasonable; that conclusion alone is an insufficient basis for deciding (as the panel in this case was precedent-bound to do) that the present search is presumptively invalid, no matter how reasonable.
The majority correctly identifies a number of the factors that are important to assessing the reasonableness of the officers’ decision to conduct a protective sweep. Some — such as the requirement that the search be performed for the safety of the officers; the necessity of having articulable facts from which an officer reasonably could apprehend danger; the importance of limiting the search to a cursory visual inspection of those places that could hide a person; and the cap on the duration of the search — come directly from Buie, 494 U.S. at 333-36, 110 S.Ct. 1093. Others — such as the legitimacy of the officers’ presence and purpose on the scene; the validity and scope of them consent to enter the home; the requirement that facts justify the sweep arise after officers obtain consent to enter for a conversation; and the potentially pernicious effect of allowing officers to themselves create the justification for a sweep — are reasonable and insightful attempts to compensate for the critical distinction between this case and Buie: the absence of an arrest or arrest warrant.1
A faithful application of these principles does not, however, lead to the conclusion that the protective sweep was reasonable. At best, it seems we are ill-equipped to reach that conclusion, relying as we must on nothing more than a paper record compiled under the mistaken impression that the reasonableness of the search was wholly irrelevant to its constitutionality. This matter should be remanded so that the *595rule that the majority properly crafts can be applied in a hearing convened for the purpose of elucidating those facts that bear directly on the reasonableness of the sweep.
Solely on the basis of the scant record now on appeal, the sweep was unreasonable, so the order granting the motion to suppress should be affirmed. Although I agree with most of the persuasive critiques found in Judge DeMoss’s forceful dissent, and although I share his concern that there is no such thing as valid consent where the consenting party has no idea that the officers will then be entitled to conduct a search, I write separately to focus on one particularly serious flaw in the majority’s analysis.
I start with a point also made by Judge DeMoss: The majority puffs this court’s assessment of the “knock and talk” strategy, taking what was once “not inherently unreasonable,” United States v. Jones, 239 F.3d 716, 720 (5th Cir.2001), and making it something that has “clearly been recognized as legitimate.” Slip op. at 590 (citing only Jones as authority). That is quite a transformation in only three years’ time.
I doubt even the majority would contend that this now “clearly ... legitimate” tactic, which consists primarily of approaching a suspect at his home to seek his voluntary cooperation in an investigation, presents the compelling sort of interest found in the officers’ duty to execute an arrest warrant. Officers use “knock and talk” encounters as just one of the many available investigative tools, and they do so hoping that they will be able to determine whether there even exists the probable cause that is necessary to obtain an arrest or search warrant.
In seeking the proper balance between privacy and the promotion of legitimate governmental interests, Houghton, 526 U.S. at 300, 119 S.Ct. 1297, it may well be that our decisions “mean that the police use a tactic like ‘knock and talk’ somewhat less frequently, but that may be the price of compliance with the Fourth Amendment.” United States v. Johnson, 170 F.3d 708, 718 (7th Cir.1999). There are other lawful ways for police to pursue their investigation without testing the limits of the Fourth Amendment, including — as the district court found — by returning another day when Gould was present and amenable to speaking with them.
I make this point only to highlight a significant principle that the majority opinion recognizes but fails to invoke: However high the government’s interest in protecting its officers, there must be some other legitimate purpose for which officers secure themselves. See slip op. at 590-91. A search that does nothing more than allow the officers safely to remain in a place where they have no reason or right to be will, of necessity, be unreasonable in all but the rarest of circumstances. The majority’s assessment that the police have a legitimate interest in pursuing “knock and talk” encounters suffices to create a justification for the officers’ presence in Gould’s trailer, and it plays a large role in the eventual conclusion that this search was reasonable in light of all the circumstances.
Yet, even assuming the majority correctly assesses the legitimacy of the “knock and talk” technique, a reasonable officer would have known, before entering Gould’s bedroom, that the original purpose of the encounter would not be realized that day. At best, from the officers’ perspective, Gould was not home and was unable to discuss the allegations made against him. At worst, he was hiding and did not wish to speak with them.2
*596As Judge DeMoss rightfully recognizes, slip op. at 601 (DeMoss, J., dissenting), the majority glosses over this error by assessing the legitimacy of the officers’ purpose and the reasonableness of their fear at two different points in time. Slip op. at 590-91, 591-93. It is true that at one point, the officers were in the mobile home with a valid purpose: to discuss with Gould the serious allegations against him. It is equally true that the officers were, at another point, in the bedroom with a legitimate fear: that Gould was hiding in a closet and posed a threat to their safety. But there is no consanguinity between these points. The legitimate purpose of the encounter had all but evaporated by the time the majority concludes the officers possessed a valid fear.
The officers had no reason to enter Gould’s bedroom if Gould was not therein, voluntarily cooperating. An empty room serves no investigative purpose where the entire focus of the investigation is on having a conversation. This fact is illustrated by the action taken by the officers as soon as the room was secure: They left it. Inasmuch as the sweep served no purpose other than to secure a room in which the officers had nothing to do, it was unreasonable and in violation of the Fourth Amendment.
The majority has a rejoinder to that argument: Regardless of whether the officers should have known that their quest for a “knock and talk” encounter had been rendered fruitless, they nonetheless possessed a compelling interest in securing the mobile home so they could safely depart from it. Slip op. at 591-93. I agree that this is one of two articulated justifications for the sweep in Buie,3 and, if supported by the record, conceivably could serve to make the sweep reasonable as well. The record, however, flatly refutes that view. Moreover, the majority’s assertion to the contrary is based in large part on a factual finding that it previously overturns as being clearly erroneous.
There is no dispute that Cabral lacked even the apparent authority to consent to the entry into Gould’s bedroom. Slip op. at 588-89. As a result, the sweep must be justified on the basis of the threats facing the officers at the instant before they entered that room. Id. That is to say, once the officers observed that Gould was not in his bedroom (which they could not enter without his consent, nor had an investigative reason to enter without his presence), their decision to enter and conduct a sweep is reasonable only if they would have faced a greater danger by not entering. The majority appears to conclude that it was not only safer, but obviously so, for the police to enter the room that potentially housed a danger, than it was to retreat the few feet toward the doorway they had used only an instant before.
We are not faced here with Daedalus’s Labyrith or the Minotaur lurking somewhere inside. The officers — one of whom already had his gun drawn — were in a fourteen-by-sixty-foot trailer home, and there is every indication that they had as plain a view of their path to retreat as they did of the empty bedroom.4 The govern*597ment has not even argued, in its briefs, that the officers were unable to depart safely.
Instead, the government’s posture throughout this case is that there was no need to flee, because the officers still were conducting a valid “knock and talk” investigation. As I have shown, and as the majority tacitly concedes, slip op. at 591-93, that claim is inconsistent with the scope of the “knock and talk” technique, which has, as its central premise, the presence of a voluntarily cooperating witness. To compensate for the absence of any investigative purpose to the sweep, the majority instead has adopted the notion that there was a greater danger in retreating than there was in sweeping, a claim not supported by the district court’s findings of fact, to which we ordinarily should defer.
As the majority correctly observes, the officers knew Gould had a violent past and was alleged to have been making threats against government officials. But all those facts were known to the officers before they decided even to enter the house. If that alone placed them in an unjustified state of danger, that was so as much at the time they elected to enter the mobile home as when they chose to sweep. If the majority’s new rule is taken to countenance a sweep whenever police seek voluntary consent to enter a building that they already perceive to be intolerably dangerous, Judge DeMoss is surely correct to assign this investigative technique the new moniker: “knock; enter, maybe talk, and search.” Slip op. at 601 (DeMoss, J., dissenting).5
Apparently, however, this is not the point the majority is attempting to make, for it unambiguously requires officers to justify the sweep on the basis of evidence that was discovered after they obtained consent to enter. Slip op. at 590. Nevertheless, though the majority laudably imposes this limitation on its new rule, the majority has not faithfully applied that rule to the present record.
The only fact to which the majority points for its explanation of how the officers went from the point of being safe enough to enter the room to the point of being threatened enough to justify a sweep, is the finding that Gould was not in the bedroom where Cabral said he would be. The most natural inference to draw from that fact is that Gould was not home, or that if he was home, he wanted nothing to do with the officers.
Even assuming the reasonableness of the belief that Gould instead was preparing to ambush the officers as they left the scene — something he chose not to do when they entered the home, and was far more likely to do when the officers drew nearer in their search for him — ’that would pose a threat to the officers only when they were outside the bedroom, if they also believed *598Gould was in possession of a firearm. This is the unambiguous basis on which the district court determined that “the officers were justified in viewing the defendants as a danger to their safety,” a statement that immediately follows the paragraph in which the court states that the officers knew, before they arrived at the scene, that Gould had a firearm. But the majority cannot possibly reach the same conclusion, because its opinion also states that the district court clearly erred when it found that the officers knew about the weapon! Slip op. at 591 n.13.
As a result, there is no basis in the record for the majority’s contention that it was more dangerous for the officers to leave the room instead of entering a confined area that they suspected housed a threat, and start poking around. That is a theory that was manufactured out of whole cloth at the en banc oral argument.
If the majority genuinely suspects that this might have been the case, the best it can do is remand so the record can be developed with an eye to the correct governing legal standard. As the court correctly determines today, that standard is not just whether the sweep was made incident to arrest (as Wilson erroneously led the district court to believe), but rather whether the sweep was a reasonably necessary, minimally intrusive means of securing an area in which the officers needed to perform a task of compelling importance.
The majority recites, then loses sight of, the well-established maxim that “physical entry of the home is the chief evil against which the wording of the Fourth Amendment is directed.” Payton v. New York, 445 U.S. 573, 585, 100 S.Ct. 1371, 63 L.Ed.2d 639 (1980). Because the majority thereby gives insufficient respect to the constraints of the Fourth Amendment, I respectfully dissent.
. See slip op. at 590-91 (legitimacy of purpose); id. at 590 (validity and scope of consent); id. at 590 (concern that sweeps will be attempted after obtaining consent but before new facts indicate a heightened danger); id. at 590-92 (potential that sweep would be improper if officers unnecessarily created dangerous situation).
. The fact that Gould was found hiding in the woods, wearing only his boxer shorts, adequately attests to the fact that the latter of *596these two possibilities was the more realistic that day.
. See Buie, 494 U.S. at 335-36, 110 S.Ct. 1093 ("The sweep lasts no longer than is necessary to dispel the reasonable suspicion of danger and in any event no longer than it takes to complete the arrest and depart the premises.”).
. Here again, the sparse record inhibits the court's ability truly to assess whether the sweep was a reasonable alternative to a safe retreat. At en banc oral argument, significant time was devoted to the question whether this was a single-wide or double-wide trailer home, and what effect that might have on the officers ability to leave the scene safely. It is apparent that such questions became relevant only after the court determined that Wilson *597was not good law, and there should be no great surprise in finding that a district court's memorandum addressing an entirely different question provides poor fodder on which to graze.
. There is no basis for the majority's conclusion, slip op. at 584-85, that this situation was inherently dangerous, as the Supreme Court described the in-home arrest in Buie. In Buie, 494 U.S. at 333, 110 S.Ct. 1093, the Court recognized the danger an officer faces when forced to effect a confrontational encounter on his "adversary's 'turf.' ” Here, although the majority correctly recognizes that a "knock and talk” encounter does not include the potentially explosive confrontation of an arrest, slip op. at 584-85, it nevertheless concludes that the encounter is dangerous by virtue of being on the adversary's turf. Id.
This completely mistakes the fact that in a "knock and talk” encounter, there is no adversary. The whole point is to approach a citizen and learn something through voluntary cooperation.