dissenting in part.
A Fourth Amendment violation made it possible for FBI agents to enter Conrad’s father’s home, wake Conrad from his drug-induced sleep, and question him. Conrad quickly confessed to possessing and distributing child pornography and gave the agents his laptop. The agents asked Conrad to travel with them to his Chicago apartment, where he had said he had other computers. Conrad agreed and, within about fifteen minutes of the FBI agents’ entry into his father’s home, he was on his way to Chicago with two agents. Conrad sat in the backseat of their vehicle but was not handcuffed or otherwise restrained. During the hour-long ride from Geneva to Chicago he called his father, smoked a cigarette, and chatted with the agents, discussing, among other things, the type of file server programs he operated. After the ride, Conrad brought the agents into his Chicago apartment, showed one of the agents his DJ equipment, fed his cat and changed its litter, and two more FBI agents arrived. Conrad was Mirandized and signed consent-to-search forms. He again admitted to possession and distribution of child pornography and turned over more computer equipment. The district court found a Fourth Amendment violation sufficient to suppress all evidence from the Geneva home and from the car ride. But the district court did not suppress evidence from the Chicago apartment, concluding that the evidence gathered there was sufficiently attenuated from the agents’ Fourth Amendment violation. The majority agrees with the district court that a line should be drawn between (A) the evidence gathered at the Geneva home and in the agents’ car and (B) the evidence collected at the Chicago apartment. In the majority’s view, the taint of the Fourth Amendment violation dissipated even while defendant was in continuous contact and conversation with the very agents responsible for the Fourth Amendment violation. Because the majority’s position is inconsistent with precedent and dramatically lowers the standard for attenuation, I respectfully dissent.
It has long been recognized that evidence obtained by “exploitation” of a Fourth Amendment violation should be suppressed. Brown v. Illinois, 422 U.S. 590, 600, 95 S.Ct. 2254, 45 L.Ed.2d 416 (1975). It is often difficult to tell what amounts to exploitation requiring suppression. Bad faith use of a Fourth Amend*738ment violation to acquire additional evidence is sufficient to require suppression (but not necessary), see, e.g., United States v. Carter, 573 F.3d 418, 425-26 (7th Cir.2009), and a but-for causal connection between a violation and subsequently discovered evidence is necessary (but not sufficient), id. at 424 (citing Hudson v. Michigan, 547 U.S. 586, 591, 126 S.Ct. 2159, 165 L.Ed.2d 56 (2006)). Many cases, like this one, fall into the large middle-category where there isn’t clear evidence of bad faith but there is plain but-for connection. The framework laid out in Brown, 422 U.S. at 603-04, 95 S.Ct. 2254, is supposed to help us sort these middle cases. As the majority correctly noted, we are to consider (1) “temporal proximity,” (2) “the presence of intervening circumstances,” and (3) “the purpose and flagrancy of the official misconduct.” Id.
The first factor is least important. The majority’s cases, supra at 733, explain that, depending on other factors, a few minutes may be sufficient for attenuation and many hours may be insufficient. By itself, then, a two-hour lapse between the original violation and the subsequently discovered evidence — not a particularly long or short time — tells us almost nothing. The majority nevertheless concludes that two hours tips in favor of attenuation. It can reach that conclusion, however, only by collapsing the first and second considerations, analyzing under the first heading not just the time elapsed — “temporal proximity,” as Brown put it — but also the quality of that time. The quality of intervening hours or minutes is significant, of course, but it is properly analyzed as part of the intervening circumstances or, perhaps, as it relates to the purpose and flagrancy of the violation. In this case, analyzing temporal proximity on its own gets us nowhere; it doesn’t support attenuation or suppression. See United States v. Reed, 349 F.3d 457, 464 (7th Cir.2003).
Moving to the next factor, “intervening circumstances,” the majority, endorsing the district court’s analysis, identifies three facts favoring attenuation: (1) Conrad’s consents to search and Miranda waiver, (2) the evidence that was not suppressed was obtained at a location different than the evidence that was suppressed, and (3) Conrad was allowed to call his father during the car ride with the agents. These facts carry some weight, to be sure, but, in my view, much less than the majority believes. First, consent and Miranda waivers are important — particularly insofar as voluntariness is a threshold requirement for the admissibility of any confession, supra at 733 n. 3 — but, however important, they cannot amount to independent intervening circumstances. See, e.g., Taylor v. Alabama, 457 U.S. 687, 690, 102 S.Ct. 2664, 73 L.Ed.2d 314 (1982); Reed, 349 F.3d at 464. United States v. Robles-Ortega, 348 F.3d 679, 684 (7th Cir.2003), is instructive. In Robles-Ortega we focused on whether a voluntary consent to search was an intervening circumstance for the purpose of attenuation. We noted that “[a]s with confessions given after Miranda warnings, ... consent alone does not necessarily purge the taint of the illegal action.” Id. We then rejected the government’s attempt to construe United States v. Liss, 103 F.3d 617 (7th Cir.1997), as holding “that a consent is an independent intervening event that breaks the causal chain stemming from the illegal search.” Id. The inquiry in Liss and Robles-Ortega had to do with what else happened; those cases concerned whether consent was obtained by means “sufficiently distinguishable” from Fourth Amendment violations to have been purged of the primary taint. Id. at 683. When other circumstances surrounding a voluntary consent to search or voluntary confession indicate that a prior illegality was not crit*739ical to obtaining the confession or consent to search, then Miranda warnings and voluntariness of consent will be decisive. But where, as here, the Fourth Amendment violation resulted in a voluntary confession and a voluntary surrender of incriminating evidence that the district court saw fit to suppress, attenuation should require more than just another voluntary confession and voluntary consent. The illegality must cease to be a path-breaking event that makes subsequent discoveries all but inevitable. In this case, we have a confession and consent given to an agent and, after about an hour in the presence of that same agent, another confession and consent given to the same agent (and others). As the Supreme Court concluded in a different legal context, “[i]t would have been reasonable to regard the two sessions as parts of a continuum, in which it would have been unnatural to refuse to repeat at the second stage what had been said before.” Missouri v. Seibert, 542 U.S. 600, 616-17, 124 S.Ct. 2601, 159 L.Ed.2d 643 (2004). Without additional intervening factors, avowals of voluntariness are a technical break that have little value for an individual caught-up in successive (even non-custodial) interrogations on the same subject with the same agents.
But the majority thinks that there were additional intervening circumstances. The consents and waivers occurred in different locations — at the Geneva house and at the Chicago apartment. Yet the change in scenery cannot count for much here, where the most salient feature in Conrad’s environment was the continuous presence of FBI agents. Consider United States v. Fazio, 914 F.2d 950 (7th Cir.1990), a two-location case similar to this one but with at least one major difference: In Fazio, after a Fourth Amendment violation at his restaurant, defendant drove himself to a meeting where he made incriminating statements. Id. at 952. We emphasized this fact in deciding that the initial illegal search was sufficiently attenuated. Id. at 958; Reed, 349 F.3d at 464 (citing Fazio as an example of a case with sufficient intervening circumstances and specifically noting that defendant drove his own vehicle to the meeting where he made incriminating statements). If the agents would have asked Conrad to drive his own car (he had a Porsche parked in the driveway of the Geneva home) to the Chicago apartment, this would be a very different case. Given some time away from the agents, Conrad may or may not have decided to go to the Chicago apartment and confess. He may have taken more time to talk to his father (and it would have been a more private conversation without agents listening to Conrad’s side of the call), he may have made other calls — we can’t know. Whatever he might have done, if he did show up at his Chicago apartment an hour or two later to consent to a search and confess his crimes, the search and the confession would have retained a but-for connection to the initial Fourth Amendment violation, but it would have separated the two events enough to make the second (if it still occurred) an independent act, freely undertaken and not one made virtually inevitable given what he’d already shared with the agents. See Fazio, 914 F.2d at 958. But the agents did not take the chance that Conrad wouldn’t show up or would think better of continuing to talk to them. They kept him in their presence, in the backseat of their car, kept him talking, and kept him comfortable — they kept him on the line and then continued reeling him in.
Finally on this second factor, the majority lists Conrad’s phone call to his father while riding in the backseat of the agents’ car as a significant intervening circumstance. This fact recalls Taylor, 457 U.S. at 691, 102 S.Ct. 2664, where a defendant was given the chance to speak privately for *740five to ten minutes with his girlfriend and a male companion before making his confession. That much longer (and more private) conversation, even paired with repeated Miranda warnings, wasn’t enough to “purge the primary taint.” Id. In this case, the short call Conrad made from the agents’ car, in the agents’ presence, did little to interrupt the seamless interaction between Conrad and the agents. Moreover, it’s hard to understand how the call could have much value as an intervening circumstance when it was part of the evidence suppressed in the wake of the initial Fourth Amendment violation.
This brings us to the third factor, “purpose and flagrancy.” The majority carefully and accurately recites the law, observing that our cases consider “both the conduct before and after the constitutional violation.” Supra at 735. The majority also points out that “actions ... undertaken in an effort to advance the investigation” cut against attenuation. Supra at 736 (quoting Reed, 349 F.3d at 465). Agreeing with the district court, the majority concludes, however, that “what began as purposeful conduct morphed into conduct consistent with attenuation by the time of the Chicago Apartment.” Supra at 736. To support this result, the majority refers to the agents’ “professional” behavior at the Chicago apartment, that Conrad was advised of his Miranda rights, and that he was told in writing of his right not to allow the agents’ search. I have already said why I believe that “reliance on the giving of Miranda [v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966) ] warnings is misplaced.” Taylor, 457 U.S. at 691, 102 S.Ct. 2664. And the same goes for reliance on defendant’s consent to search. Robles-Ortega, 348 F.3d at 684. The bottom line for the district court and the majority, then, incorporating their emphasis on Miranda warnings and consent, is that the agents’ professional behavior and the strangely cordial atmosphere were sufficient to dissipate the taint of the Fourth Amendment violation. I disagree. As we explained in Reed, there is no attenuation by friendliness:
In determining the purpose and flagrancy of the official misconduct, the district court held that the factor weighed against suppression because [defendant’s] interrogation was conducted congenially and the police judiciously administered Miranda warnings, suggesting that the manner of the illegal arrest and subsequent interrogation was not “calculated to cause surprise, fright, or confusion.” E.g., Brown, 422 U.S. at 605, 95 S.Ct. 2254. But that inquiry, although relevant, is not complete, because “purposeful and flagrant” misconduct is not limited to situations where the police act in an outright threatening or coercive manner....
349 F.3d at 465. In this case, the agents only came into contact with Conrad because of their Fourth Amendment violation. That got them in the house and got them their first interview with Conrad. It turned out to be a very advantageous time, at least from the agents’ perspective, for the interview to occur because moments before it began, it appears that Conrad was in the depths of a drug-aided sleep. Once roused by the agents, Conrad promptly confessed to a very serious crime and gave them, in addition to his statement, incriminating physical evidence. Having learned about the Chicago apartment, the agents asked Conrad to travel there with them to continue his cooperation. By the time they got to Chicago, an hour later, the taint (according to the majority) had dissipated. The only pause in the continuous conversation with the FBI agents was Conrad’s short call to his dad. Did that do it? Or was it the length of the drive itself? Would the situation have *741“morphed” into conduct consistent with attenuation if the house and apartment had been just twenty minutes apart? Was it the extra time for friendly banter about stereo equipment that made the difference?
In my view, nothing of significance separated the Fourth Amendment violation, the collection of evidence that was suppressed from the Geneva house through the drive to Chicago, and the collection of additional evidence at the Chicago apartment. It was a single continuum of events. Yet the majority holds that the taint of a Fourth Amendment violation can dissipate even while a defendant is in continuous contact and conversation with the very agents responsible for the Fourth Amendment violation. I realize that letting Conrad off the hook after the Fourth Amendment violation — like arranging to meet him later — would have jeopardized the agents’ nearly certain collection of the incriminating evidence they set out to get when they went to the Geneva house. But it is difficult to see any consequential separation between what happened at that house and the completion of that process at the Chicago apartment. Of course, the agents treated Conrad well enough; they chatted him up; there was no need to be aggressive, they were smarter than that. But at no point did they disengage, and at no point did they provide Conrad a real break in the continuous interaction that started with a confession in Geneva and ended with another in Chicago. It would seem that after this case not much is required for attenuation. The takeaway for law enforcement is that if you think you may have gone beyond what the Fourth Amendment allows, not to worry, just don’t let the suspect out of your sight, be congenial, and all will be forgiven. Because I think that allowing attenuation to mysteriously arise during a nonstop interaction with a single group of agents is inconsistent with precedent and gives far too much weight to the role of Miranda warnings and consent, and because I think the agents improperly exploited a Fourth Amendment violation to investigate Conrad’s crimes and to gather evidence, on this issue I respectfully dissent.