dissenting.
Brendan Dassey confessed on videotape that he raped Teresa Halbach, helped his uncle murder her, and then burned her body in a fire pit at his uncle’s junkyard. A jury convicted Dassey of those crimes, and the Wisconsin state courts have upheld the convictions. On federal habeas corpus review, however, Dassey has persuaded the district court and now my colleagues that his confession was involuntary and his convictions invalid. I respectfully dissent. We should reverse.
To decide whether Dassey’s confession was voluntary, the state courts applied the correct but general and even indeterminate “totality of the circumstances” test. See Withrow v. Williams, 507 U.S. 680, 693-94, 113 S.Ct. 1745, 123 L.Ed.2d 407 (1993); Gallegos v. Colorado, 370 U.S. 49, 55, 82 S.Ct. 1209, 8 L.Ed.2d 325 (1962). The Wisconsin Court of Appeals upheld the trial court’s finding that Dassey’s confession was voluntary in a succinct per curiam opinion that rejected that claim in two paragraphs. That was permissible. While the majority would have preferred a more nuanced and detailed discussion of the circumstances surrounding Dassey’s confession, the Antiterrorism and Effective Death Penalty Act (AEDPA) of 1996 does not authorize federal courts to sit in judgment of the length of state court opinions. Rather, as Harrington v. Richter teaches, even unexplained decisions by state courts are entitled to deference under AEDPA. See 562 U.S. 86, 98, 131 S.Ct. 770, 178 *984L.Ed.2d 624 (2011) (“Where a state court’s decision is unaccompanied by an explanation, the habeas petitioner’s burden still must be met by showing there was no reasonable basis for the state court to deny relief.”). Under AEDPA and Richter, relief must be denied if a reasonable court could have reached the state courts’ conclusion. Id.
Habeas relief from state court convictions is rare, reserved for those unusual cases where state courts abandon their obligation to enforce federal constitutional law. See id. at 102-03, 131 S.Ct. 770 (“If [the AEDPA] standard is difficult to meet, that is because it was meant to be.... Section 2254(d) reflects the view that habe-as corpus is a ‘guard against extreme malfunctions in the state criminal justice systems,’ not a substitute for ordinary error correction, through appeal.”) (citation omitted). No Supreme Court precedent compels relief for Dassey. His petition should be denied.
Rather than show how Supreme Court precedent requires habeas relief, the majority observes: “By surveying the Supreme Court cases on the voluntariness of juvenile confessions one can see how much the unique characteristics of both the defendant and the interrogation play into the assessment of voluntariness.” Ante at 954. For this reason, the majority writes, “other cases can only act as broad guideposts.” Id.
That is exactly right, but that is also why we should reverse. Without a compelling showing based on Supreme Court precedent, habeas relief must be denied. The more a state court’s decision depends on weighing a host of factors as part of the totality of the circumstances, the harder it is to show that the decision was “contrary to, or involved an unreasonable application of, clearly established Federal law.” 28 U.S.C. § 2254(d)(1). Applying such a broad standard to a particular case leaves substantial room for judgment. “The more general the rule, the more leeway courts have in reaching outcomes in case-by-case determinations.” Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004) (reversing grant of ha-beas petition where similar fact-sensitive standard governed whether seventeen-year-old petitioner had been “in custody” during interrogation in which he confessed).
Even if we were reviewing the admissibility of Dassey’s confession de novo, great caution would be warranted. The majority’s decision breaks new ground and poses troubling questions for police and prosecutors. It calls into question standard interrogation techniques that courts have routinely found permissible, even in cases involving juveniles.
This was a relatively brief and low-key interview of a Mirandized subject who was not mistreated or threatened, whose creature comforts were satisfied, and whose parent consented. If such a gentle interrogation can be treated as unconstitutionally coercive, what should police do the next time an investigation leads to a teenager with some intellectual challenges? Few wrongdoers are eager to own up to crimes as serious as Dassey’s. The Constitution is not offended by such police tactics as encouraging the subject to tell the truth, bluffing about what the police already know, or confronting the subject with what the police know from physical evidence and with the internal contradictions and improbabilities in his story. Today’s decision will make some police investigations considerably more difficult, with little gained in terms of justice.
I. The Totality of the Circumstances
My colleagues describe the critical March 1, 2006 interview of Dassey as “intimidating and anxiety producing.” Ante at 974. I suspect the source of any anxiety *985Dassey felt was his guilt, not the circumstances of a relatively gentle and non-coercive interview. The majority focuses in painstaking detail on a few factors that weigh in favor of finding that Dassey’s confession was not voluntary. Many other factors weigh in favor of finding it was voluntary. The circumstances that have most concerned courts and that have contributed most to voluntariness jurisprudence—such as physical abuse, threatening behavior, or prolonged questioning— were simply absent here.
Consider these circumstances: the investigators did not initially consider Dassey a suspect in the murder. Still, they had good reason to think that he knew more about his uncle Steven Avery’s involvement in Teresa Halbach’s death than Dassey had told them thus far. Two days before the critical March 1 interview, Dassey had told investigators that he saw human body parts—toes, a hand, a forehead, and a stomach—in Avery’s bonfire the previous Halloween. Dassey had also said that Avery told him he stabbed Teresa. In a separate conversation that evening, Das-sey had told the investigators that he helped Avery clean a dark red stain on his garage floor.
On March 1, the investigators obtained consent from Dassey’s mother to interview him once again. They read Miranda warnings to Dassey, drove him to a local sheriffs office, and reminded him about the Miranda warnings once they arrived. They offered him snacks, beverages, and restroom breaks. During the interview, Dassey sat comfortably on a sofa. He exhibited no signs of physical distress. The investigators spoke in measured tones. They did not threaten Dassey, nor did they use intimidating or coercive language. They coaxed and encouraged him to tell the truth. They made Dassey no specific guarantees. In fact, they told him at the outset: “We can’t make any promises.... ”
The interview lasted about three hours in total. Fifty-four minutes into the conversation, Dassey told the officers that he raped Teresa Halbach the day she was murdered. Fourteen minutes later, Dassey admitted, in response to a relatively open-ended question, that he cut Teresa’s throat. The investigators soon took a thirty-minute break and then continued questioning Dassey for a little over an hour. At the conclusion of the interview, the investigators informed Dassey that they were placing him under arrest.
At times, the investigators challenged Dassey when his account seemed incomplete, did not make sense, or conflicted with physical evidence. At other points, the investigators deliberately misled Dassey by telling him they knew more than they actually did or by suggesting false facts to see if he would agree to them. (He did not.) Those are routine techniques in police interrogation. They do not transform a voluntary confession into an unconstitutional one. The investigators also repeatedly encouraged Dassey to tell the truth, and they offered vague assurances that it would be better for him if he did. Those are also routine techniques. They are not fraudulent or coercive. At no point did the investigators make the sort of specific false promises that can render a confession involuntary. The record here does not show police tactics “so offensive to a civilized system of justice that they must be condemned under the Due Process Clause of the Fourteenth Amendment.” See ante at 951, quoting Miller v. Fenton, 474 U.S. 104, 109, 106 S.Ct. 445, 88 L.Ed.2d 405 (1985).
II. AEDPA and Deference to State Court Judgments
A. The Departure from Deference
The Antiterrorism and Effective Death Penalty Act of 1996 amended the federal *986habeas corpus statute to provide that an “application for a writ of habeas corpus ... shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim—(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court ... or (2) resulted in a decision that was based on an unreasonable determination of the facts....” 28 U.S.C. § 2254(d). It is not enough that a federal court might have decided the case differently in the first instance. Rather, the federal court must be confident that the decision of the state court was so beyond the pale as to constitute an error “well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Richter, 562 U.S. at 103, 131 S.Ct. 770 (emphasis added).
My colleagues insist, repeatedly, that they have “kept the strict constraints of the AEDPA forefront” in their minds. E.g., ante at 949. Yet no Supreme Court case, no case decided in this circuit, and indeed no case cited by the parties or the majority has found a confession involuntary on facts resembling these, even where the subject was a juvenile.
Never before has the Supreme Court or this court signaled that police bluffs about what they know may render a confession involuntary. Neither the Supreme Court nor this court has ever held, as the majority seems to believe, that an investigator’s vague assurances about the value of telling the truth may amount to fraudulent promises of leniency. Nor have we held that such statements must be viewed from the subjective perspective of the suspect, no matter how distorted his perspective may be. The majority worries that Dassey may have taken as literal an investigator’s advice that honesty is the “only thing that will set you free,” transforming that biblical phrase into the “exact kind of promise of leniency that courts generally find coercive.” Ante at 961; see John 8:32. The majority reaches this conclusion in spite of our long recognition that “the law permits the police to pressure and cajole, conceal material facts, and actively mislead.” United States v. Rutledge, 900 F.2d 1127, 1131 (7th Cir. 1990).
In one telling departure from AEDPA deference, the majority cites a law review article to observe: “Experts on confessions have noted that ‘though courts are reluctant to find that police officers have overwhelmed a child’s will by repeatedly admonishing the child to “tell the truth,” many children will eventually hear “tell the truth” as, “tell me what I want to hear.” ’ ” Ante at 963 (citation omitted). The majority then suggests that “Dassey found ‘the truth’ either by stumbling upon it or by using the information the investigators had fed him,” and asserts boldly that it is “impossible to read or view Dassey’s interrogation and have any confidence that Dassey’s confession was the product of his own free will rather than his will being overborne.” Ante at 963-64. The majority invites the reader to scrutinize Dassey’s confession with this “key” in hand.
I read (and see) the evidence quite differently: Dassey’s confession appears to have been the product of a guilty conscience, coaxed rather gently from him with standard, non-coercive investigative techniques. Even assuming, however, that the majority’s interpretation is plausible, our job as a federal court reviewing a state conviction under § 2254(d) is not to consult scholarly literature in search of new best practices.
Our narrower task is to determine whether the state court decision was based either on an unreasonable application of clearly established law as handed down by *987the Supreme Court or on an unreasonable view of the facts. Apart from the uncontroversial observation that juvenile confessions should be treated with care, see J.D.B. v. North Carolina, 564 U.S. 261, 269, 131 S.Ct. 2394, 180 L.Ed.2d 310 (2011) (direct appeal of Miranda custody decision), the majority cites no Supreme Court authority in support of its interpretive “key.”1
B. Deference or Critiquing Opinions?
Early in its opinion, the majority writes that the “state appellate court did not identify the correct test at all and did not apply it correctly.” Ante at 947. The criticism is misplaced. The state court correctly recognized that (1) a confession’s voluntariness turns on the “totality of the circumstances” and (2) the analysis involves a “balancing of the defendant’s personal characteristics against the police pressures used to induce the statements.” That standard fits comfortably with the Supreme Court’s explanation in Withrow: “courts look to the totality of circumstances to determine whether a confession was voluntary. Those potential circumstances include ... the crucial element of police coercion; the length of the interrogation; its location; its continuity; the defendant’s maturity; education; physical condition; and mental health. They also include the failure of police to advise the defendant of his rights to remain silent and to have counsel present during custodial interrogation.” 507 U.S. at 693-94, 113 S.Ct. 1745 (citations omitted). This fact-sensitive balancing test applies whether the subject is a mature adult or an intellectually challenged high-school student. See Gilbert v. Merchant, 488 F.3d 780, 793 (7th Cir. 2007) (“[I]t is the totality of the circumstances underlying a juvenile confession, rather than the presence or absence of a single circumstance, that determines whether or not the confession should be deemed voluntary.”) (collecting cases).
The majority’s real concern seems to be that the Wisconsin Court of Appeals only paid lip service to the correct standard but did not apply it seriously. The majority writes that the state appellate court “listed Dassey’s age, education and IQ, but it never ... evaluated those factors to determine whether they affected the voluntariness of Dassey’s confession.” Ante at 950. Likewise, the majority writes that the state court “analyzed some of the investigators’ interrogation techniques, but it never evaluated or assessed how those techniques affected the voluntariness of [Dassey’s] confession.” Id. Elsewhere the majority complains that “the state appellate court addressed the voluntariness of *988the confession in two short paragraphs.” Ante at 960. The majority also writes that the less a state court says, “the less a federal court can ascertain that the state actually applied a totality of the circumstances evaluation.” Ante at 960-61. The majority seems to expect longer, more detailed, and perhaps more anguished opinions from the state courts in such cases. Those expectations do not call for habeas relief.
Under § 2254(d), federal courts do not judge the length or brevity of opinions issued by state courts with dockets far more crowded than ours. Federal courts have “no authority to impose mandatory opinion-writing standards on state courts.... The caseloads shouldered by many state appellate courts are very heavy, and the opinions issued by these courts must be read with that factor in mind.” Johnson v. Williams, 568 U.S. 289, 300, 133 S.Ct. 1088, 1092, 1095-96, 185 L.Ed.2d 105 (2013) (footnote omitted) (reversing habeas relief; federal court erred by finding that state court overlooked petitioner’s federal claim and by then reviewing that claim de novo); see also Wright v. Secretary for Dep’t of Corrections, 278 F.3d 1245, 1255 (11th Cir. 2002) (“Telling state courts when and how to write opinions to accompany their decisions is no way to promote comity.”). Where the last state court to review a claim reaches a decision and offers reasons, its decision is entitled to the same deference whether the court states its reasons succinctly in an unpublished order or expounds at length in a landmark opinion.
AEDPA deference still applies when a state court offers no reasons, facially defective reasons, or incomplete reasons for its decision. Where a state court provides no explanation, “the habeas petitioner’s burden still must be met by showing there was no reasonable basis for the state court to deny relief.” Richter, 562 U.S. at 98, 131 S.Ct. 770. “Under § 2254(d), a habeas court must determine what arguments or theories supported or ... could have supported, the state court’s decision; and then it must ask whether it is possible fairmind-ed jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of this Court.” Id. at 102, 131 S.Ct. 770 (emphasis added); see also Williams, 133. S.Ct. at 1094 (“Although Richter itself concerned a state-court order that did not address any of the defendant’s claims, we see no reason why the Richter presumption should not also apply when a state-court opinion addresses some but not all of a defendant’s claims.”).
Similarly, even where the last state court to render a decision offered a faulty reason for its decision, “although we would no longer attach significance to the state court’s expressed reasons, we would still apply AEDPA deference to the judgment,” turning to the “remainder of the state record, including explanations offered by lower courts.” Whatley v. Zatecky, 833 F.3d 762, 775 (7th Cir. 2016) (citation omitted); Brady v. Pfister, 711 F.3d 818, 827 (7th Cir. 2013) (“A state court could write that it rejected a defendant’s claim because Tarot cards dictated that result, but its decision might nonetheless be a sound one.”).
And by the reasoning of Richter and Williams, deference likewise applies where a state court “gave some reasons for an outcome without necessarily displaying all of its reasoning.” Hanson v. Beth, 738 F.3d 158, 164 (7th Cir. 2013); see also Jardine v. Dittmann, 658 F.3d 772, 777 (7th Cir. 2011) (per curiam) (“This court must fill any gaps in the state court’s discussion by asking what theories ‘could have supported’ the state court’s conclusion.”) (citation omitted).
Since AEDPA deference applies when a state court offers no reasons, faulty rea*989sons, or incomplete reasons, such deference must surely be due where, as here, the state court offers a terse explanation for a reasonable result. This is not to suggest that a state court may evade habe-as review by merely incanting the correct test (in this case, “totality of the circumstances”). The majority is correct in saying that “if a court can merely state the generic Supreme Court rule without any analysis, then no federal court could ever find that ‘a decision ... involved an unreasonable application of clearly established Federal law.’ ” Ante at 947 (citation omitted). AEDPA review is deferential but not toothless. Federal courts are charged with reviewing state court records to assess the reasonableness of state court decisions. We grant relief in a small but non-trivial portion of cases, at least at the appellate level. The issue is not whether the state court might have overlooked something but whether the bottom-line result is “beyond any possibility for fairminded disagreement.” Richter, 562 U.S. at 103, 131 S.Ct. 770.
What federal courts may not do is infer that a decision was unreasonable based on the lack of explanation. As a reader of judicial opinions, I too would have appreciated more context and development in the opinion of the Wisconsin Court of Appeals. I cannot, however, hold that use of Das-sey’s confession was unconstitutional merely because the state court did not say more about all the relevant factors. The overall mix of relevant factors here simply does not dictate a finding that his confession was involuntary.
III. Doctrinal Developments
Showing a lack of the required deference to the state courts, the majority breaks new doctrinal ground in three significant respects: redefining what counts as a false promise of leniency, relying on police bluffs in the interrogation to find the confession was involuntary, and departing from a series of our court’s habeas cases denying relief to juveniles who were subjected to much more pressure than Dassey was.
First, what counts as a false promise of leniency? The majority opinion loses sight of the difference between general assurances of better treatment, which are permitted even when made to juveniles, and factually false promises, which are not. We have long recognized that “a false promise of leniency may render a statement involuntary” but that “police tactics short of the false promise are usually permissible.” United States v. Villalpando, 588 F.3d 1124, 1128 (7th Cir. 2009). In Villalpando, we rejected a claim that a police detective made a false promise of leniency where she offered to “go to bat” for the defendant and said she would “sit down” with law enforcement and probation to “work this out,” and where she also remarked, “we don’t have to charge you.” Id. at 1129.
Similarly, in United States v. Rutledge, a police officer asked the defendant whether he. would be willing to give a post-arrest statement. The officer advised the defendant that “all cooperation is helpful.” 900 F.2d at 1128. We noted that one “interpretation of the officer’s statement ... is that it promised ... a net benefit from spilling the beans,” and if the officer made such a promise without intending to keep it, “the statement was fraudulent.” Id. at 1130-31 (emphasis omitted). “But it was the sort of minor fraud that the cases allow. Far from making the police a fiduciary of the suspect, the law permits the police to pressure and cajole, conceal material facts, and actively mislead—all up to limits not exceeded” in that case. Id. at 1131; see also Fare v. Michael C., 442 U.S. 707, 727, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979) (police “did indeed indicate that a cooperative attitude would be to [sixteen-year-old’s] bene*990fit,” but their “remarks in this regard were far from threatening or coercive”).
The majority acknowledges that in Das-sej^s case, the investigators “never made the type of explicit and specific promise of leniency that an adult of ordinary intelligence might understand as a promise.” Ante at 974. That’s right. The investigators’ statements were comparable to those permitted in Villalpando and Rutledge. The investigators made vague assurances that honest cooperation would make things easier for Dassey “if this goes to trial”; that “the honest person is the one who’s gonna get a better deal out of everything”; and that honesty is the “only thing that will set you free.” One investigator said at the very beginning of the interview, before Dassey had confessed to anything, that “from what I’m seeing ... I’m thinking you’re all right. OK, you don’t have to worry about things.” But the other then cautioned: “We can’t make any promises but we’ll stand behind you no matter what you did.”
At no point did the investigators assure Dassey that he would escape prosecution or receive some other specific benefit if he cooperated or confessed. Cf. Sharp v. Rohling, 793 F.3d 1216, 1235 (10th Cir. 2015) (subject’s will was overborne where detective promised her she would not go to jail if she admitted to her participation in crime); Henry v. Kernan, 197 F.3d 1021, 1027 (9th Cir. 1999) (subject’s will was overborne where officer falsely informed him that what he said “can’t be used against you right now”).2
The majority insists, however, that whether police have made an impermissible false promise of leniency (or of anything else) depends on the subjective perception of the suspect, no matter how distorted or inaccurate his perception might be. Thus, to Dassey—with his borderline IQ and suggestible personality— the investigators’ vague assurances had in the majority’s view the “same effect” as a fraudulent promise. Ante at 974. ⅛
The Supreme Court’s “totality of the circumstances” test takes account of the subjective characteristics of the defendant (e.g., his age, health, and education). Yet no Supreme Court case has held that a confession should be deemed involuntary if the subject believed—however improbably or baselessly—that he had been promised a get-out-of-jail-free card. No case requires the reviewing court to disregard what police actually said (on a video recording, no less) in favor of what the defendant, with the benefit of time, hindsight, and savvy counsel, says he thought the police said. At a minimum, reasonable jurists could disagree whether the abstract assurances by the investigators here were, in context, false and fraudulent. That alone should defeat any claim for habeas relief.
Even if we were to approach the question de novo, there is good reason to review any alleged promises by investigators from an objective point of view, at least when we have hard evidence of what was said (and what was not). People who commit brutal crimes of the sort Dassey was convicted of committing tend to be malad*991justed and detached from social norms. It should come as no surprise that a juvenile who helps to rape a helpless victim, caps off that experience by watching television and chatting with his uncle, and then helps to murder their victim, as Dassey said he did, lives with a distorted worldview. Das-sey’s subjective impression of what police told him should not be decisive.3
Second, the majority suggests that Das-sey was at greater risk of being misled by the investigators’ vague moral support because they repeatedly told him that they “already knew” what happened. As the majority construes these statements, Das-sey could have believed that—so long as he was honest—nothing bad would happen to him. See ante at 975. The majority cites no case from the Supreme Court or any other court holding that such bluffing by police about what they know could render a confession involuntary. On the contrary, we have recognized that “a lie that relates to the suspect’s connection to the crime is the least likely to render a confession involuntary.” United States v. Ceballos, 302 F.3d 679, 695 (7th Cir. 2002) (emphasis added) (citation omitted); see also United States v. Sturdivant, 796 F.3d 690, 697 (7th Cir. 2015) (“[W]e have repeatedly held that a law-enforcement agent may actively mislead a defendant in order to obtain a confession, so long as a rational decision remains possible.”) (alteration in original), quoting Conner v. McBride, 375 F.3d 643, 653 (7th Cir. 2004).
Third, in concluding that Dassey’s confession was involuntary, the majority effectively departs from a string of our habeas decisions involving confessions by juveniles who were denied relief despite being subjected to far greater pressures than Das-sey was.
For instance, in Etherly v. Davis, 619 F.3d 654, 657 (7th Cir. 2010), we reversed habeas relief for a petitioner with no prior criminal justice experience who at age fifteen was taken from his home before dawn and interviewed by police several hours later without the consent, let alone the presence, of a parent or other friendly adult. Like Dassey, Etherly had borderline intellectual abilities; like the investigators here, the police in Etherly assured the juvenile that it would “go better for him in court” if he cooperated. Id. at 658.
In Carter v. Thompson, 690 F.3d 837, 839 (7th Cir. 2012), we denied relief to a habeas petitioner who at age sixteen endured an interrogation lasting fifty-five hours in total. During gaps in the interrogation, the petitioner slept on a bench, without a pillow, a blanket, or a change of clothes. Id. at 841; see also Murdock v. Dorothy, 846 F.3d 203, 210 (7th Cir. 2017) (denying relief to sixteen-year-old who was interrogated over seven-hour period); Gilbert, 488 F.3d at 784-86 (denying relief to fifteen-year-old who was kept from his mother and interrogated over nine-hour period); Hardaway v. Young, 302 F.3d 757, *992766 (7th Cir. 2002) (denying relief to fourteen-year-old who was interviewed over sixteen-hour period and abandoned for lengthy intervals in interrogation room).
The majority describes these cases but makes no real effort to reconcile them with the relief it grants Dassey. Instead, it criticizes the Wisconsin Court of Appeals for failing to elaborate on all the factors the majority considers important. See ante at 956. As explained above, § 2254(d) does not authorize federal courts to critique state court opinions so closely. It is enough that the state court identified the correct legal standard and applied it reasonably to the facts of the case. Just as police investigators will be left scratching their heads after this decision, state and federal courts will be flummoxed as they attempt to reconcile our grant of habeas relief to Dassey with the line of cases pointing the other way.
IV. The Details of Dassey’s Confession
Having replaced deference to the state court with what amounts to de novo review, and having redefined what counts as a false promise of leniency, the majority evaluates Dassey’s confession in the light most favorable to him. The majority opinion highlights the moments when Dassey seemed most hesitant or ambivalent.
. I have no quarrel with the majority’s consideration of those moments. We need to consider Dassey’s strongest arguments as well as the strongest arguments advanced by the State. At a few points, the investigators’ questions were so assertive and leading that it is difficult to tell whether Dassey made an honest attempt at a truthful answer or simply offered up the answer he believed the investigators were fishing for.-
A good example: the investigators believed that Teresa Halbach had been shot in the head, a detail that had not been reported in the media. (A burnt fragment of her skull recovered from the fire pit had traces of lead on it.) If Dassey knew that Teresa had been shot in the head, that knowledge would tend to corroborate his story. The investigators asked Dassey, “[Wjhat else did you do? Come on. Something with the head.” Dassey floundered, volunteering that his uncle Avery cut off some of Teresa’s hair and punched her in the head and that he—Dassey—slit Teresa’s throat. Apparently exasperated, one investigator said: “All right, I’m just gonna come out and ask you. Who shot her in the head?” Avery did, Dassey replied, adding that he did not volunteer the information because he “couldn’t think of it.” It’s reasonable to be skeptical about Dassey’s response to such a leading question, at least taking the response in isolation.
But for every point when Dassey seemed uncertain or confused, at many other points Dassey gave specific and incriminating answers to open-ended questions. Most important, Dassey volunteered specific and incriminating details about what he did, what he saw, what he heard, and even what he smelled.
Early in the interview, the investigators asked Dassey what Avery told him and showed him after he arrived at Avery’s trailer. Dassey said: “He showed me the knife and the rope.” They then asked Das-sey where he saw Teresa. Dassey said she was lying dead in the back of her jeep and that Avery told him he stabbed her. They asked why Avery had invited Dassey over. Dassey said, “Probably to get rid of the body.” When the investigators asked what happened next, Dassey admitted that he helped his uncle move Teresa’s body to the burn pit. When they asked Dassey to describe Teresa’s injuries, he said she had been stabbed in her stomach, a detail he repeated several times. (The condition of Teresa’s remains made it impossible to confirm or refute that fact.)
*993The investigators suspected Dassey had left out some important information. They asked how Dassey knew Teresa was already dead when he saw her in the jeep. Dassey volunteered that he heard screaming while riding his bike outside. He then admitted that he entered Avery’s trailer and saw Teresa. He said that Teresa was handcuffed to Avery’s bed. When the investigators asked Dassey what Avery told him, Dassey said: “That he never got some of that stuff so he wanted to get some,” adding that Avery “wanted to f*** her so hard.”
While it took more than a little coaxing from the investigators before Dassey admitted that he too raped Teresa, Dassey soon provided quite specific details about his role in the crime. He said that Teresa begged him to do the “right thing”; that Avery, conversely, praised him for doing a “good job”; that he helped Avery tie up Teresa; and that he slit her throat and cut her hair. Dassey described the brutal cremation, recalling how he and Avery carried Teresa’s body to the burn pit and covered her with branches and tires.
When the investigators asked Dassey how he and Avery cleaned the crime scene, he recounted their efforts: “We threw gas on [a pool of blood] so he could get it off. Then he tried paint thinner and then he went to bleach to get it off and ... he went like he was spraying it... .1 thought he got it on the floor and it splashed up on my pants.... ” The investigators retrieved Dassey’s pants from his home. Sure enough, they were stained with bleach.
In addition to answering open-ended .questions in specific and incriminating detail, Dassey resisted several lines of inquiry. Those points of resistance gave the state courts substantial reason to find that Dassey’s will was not overborne. Recall that the investigators were keenly interested in any information Dassey could offer about how and when Teresa Halbach was shot. They asked him how many times he shot Teresa. “Zero,” he replied. He added that he “didn’t even touch the gun,” explaining that he had been unable to shoot ever since his mother’s ex-boyfriend had shot their sick cat.
After Dassey admitted that he cut Teresa’s hair at Avery’s urging, the investigators asked what had become of the hair. Dassey insisted that he did not know and did not have the hair. Even when the investigators warned Dassey that they would find the hair if he had kept it, he insisted, “I don’t got none of the hair.”
At another point in the interview, the investigators asked Dassey whether he saw Avery rape Teresa. Three times Das-sey said no. They repeatedly asked Dassey whether he and Avery had used wires hanging in the garage to harm Teresa; Dassey insisted they had not. He rejected their suggestion that he and Avery might have hung Teresa from a rafter, even after the investigators pointed out that the “worst” was over and nothing he said would surprise them.
In one of the most direct tests of Das-sey’s suggestibility, the investigators told him falsely that Teresa had a tattoo on her stomach and asked him if he remembered it. Dassey said no. They pressed Dassey, asking if he disagreed with them. Dassey replied: “No but I don’t know where it was.” If Dassey were as overwhelmed by the police questioning as the majority seems to believe, surely he would have simply agreed that Teresa had a stomach tattoo—and that he had kept her hair— and that he had hung her from the rafters, and so on.
To be sure, Dassey’s confession was not a smooth and consistent story. There were holes in the narrative. Dassey waffled and backtracked. The sequence of events was not always clear. The majority, reviewing the interview with its defense-friendly *994“key” in hand, takes these inconsistencies as proof that Dassey was not recounting real memories but only telling the investigators what he believed they wanted to hear.
As an alternative “key” for reviewing Dassey’s confession, one might consider that the sixteen-year-old subject was wracked by guilt and was finally coming to grips with the gravity of his crimes. He had been led to do things so awful that, in the months following the crimes, he stayed silent but lost forty pounds and had fits of uncontrolled sobbing.
Owning up to what he did proved difficult for Dassey, as it surely would for anyone with a trace of a conscience. He had trouble getting the words out. Given the vagaries of human memory, it is not surprising that some details and sequences had become garbled as he replayed those violent and grisly images over and over in his mind for four months. It is easy to understand why, by the time of the March 1 interview, Dassey was not sure about everything that had happened and in what order.
While Dassey’s recollection of the sequence of events was hazy, he remembered some details vividly. He remembered colors, sounds, and smells. He remembered his uncle standing in the doorframe in his white shirt and red shorts, beckoning him inside. He remembered Teresa Halbach, lying alive on his uncle’s bed and later dead in the back of the jeep. He remembered her screams. He remembered her telling him he did not have to rape her and he should do the right thing. He remembered her blood pooling on the garage floor. He remembered the odor of her burning flesh. And he remembered why he committed the cruel acts he described: he “wanted to see how [sex] felt.”
The majority writes that “the lack of physical evidence was the weakest part of the State’s case.” Ante at 981-82. The physical evidence does not prove or disprove Dassey’s guilt or the accuracy of his confession. Still, the State offered substantial evidence that tended to corroborate some details of his confession. Examples include handcuffs and leg irons found in Avery’s bedroom (corroborating Dassey’s description of Teresa’s rape); a charred shovel, rake, and car seat (corroborating Dassey’s description of the crude cremation of Teresa’s body); and a stipulation by a family friend that he saw Avery and Dassey standing by a bonfire on Avery’s property on Halloween night in 2005, the same night that Teresa and her SUV vanished after she headed to an appointment to take photographs at Avery’s junkyard.
We also should not lose sight of the most damning physical evidence: the bones of Teresa Halbach, broken and charred, buried in the ashes of Avery’s burn pit. The corpus delicti does not point inexorably to Dassey. But it is grim corroboration for much of the story he told the investigators.
V. Conclusion
All agree that the governing constitutional standard for the voluntariness of a confession depends on the totality of the circumstances. The state courts recognized that standard and applied it reasonably to the facts before them. As in most cases on voluntariness of confessions, relevant factors point in conflicting directions. A few factors and passages from Dassey’s confession support the majority’s view that the confession was not voluntary. Many other factors and passages support the state courts’ view that, overall, the confession was voluntary. The Wisconsin Court of Appeals could have been much more thorough in its discussion, but its conclusion *995was within the bounds of reason. It was not contrary to or an unreasonable application of controlling Supreme Court precedent. We should reverse the district court’s grant of the writ of habeas corpus.
. The majority supports the need for special care in juvenile confession cases by citing studies of exonerated defendants showing that false confessions are more common by juveniles and mentally ill or intellectually deficient suspects. Ante at 952-53. False confessions are a real phenomenon, and even one is veiy troubling. Yet we should not conclude from these studies of exonerated defendants that there is an epidemic of false confessions. The more relevant denominator in the fraction is all confessions. That number is not easy to estimate, but we can estimate a conservative lower boundary for it. Bureau of Justice Statistics reports on Felony Defendants in Large Urban Counties tally violent felony convictions by guilty plea in just the nation’s 75 largest counties. (The most recent report is Brian A. Reaves, U.S. Dep't of Justice, Bureau of Justice Statistics, Felony Defendants in Large Urban Counties, 2009—Statistical Tables (2013), https://www.bjs.gov/ content/pub/pdf/fdluc09.pdf.) The majority's statistics report 227 demonstrably false confessions from 1989 to 2016. From the BJS reports, we can estimate there were more than 1.5 million guilty pleas to violent felonies over that period. So for every one demonstrably false confession over those years, there were more than 6,500 guilty pleas to violent felonies in just those 75 largest counties.
. In oral argument, we asked Dassey’s counsel to identify' a case—any case—in which a habeas petitioner was granted relief due to police representations similar to those made here. Counsel cited AM. v. Butler, 360 F.3d 787 (7th Cir. 2004), a split panel decision that is readily distinguishable and illustrates how much of a stretch Dassey's claim is. In A.M., the subject was just eleven years old, and he was not properly Mirandized. Id. at 793. He testified at trial that the interviewing officer made him a specific false promise: that if he confessed to beating and stabbing to death his elderly neighbor, “God and the police would forgive him and he could go home in time for his brother’s birthday party.” Id. at 794. Investigators made no such false promise to Dassey.
. Dassey brought his involuntary confession claim under both 28 U.S.C. § 2254(d)(1), decisions contrary to or unreasonably applying clearly established federal law, and (d)(2), decisions based on unreasonable factual determinations. The majority and I both focus on the Supreme Court's "totality of the circumstances” test and related doctrinal considerations under (d)(1). The majority also says in several places that the state courts made unreasonable factual findings under (d)(2) but acknowledges that the analyses under (d)(1) and (2) overlap here. Ante at 950. There is no dispute about what the investigators actually said, and the discussion in this section shows why the claim should also fail under (d)(2). The state courts’ finding that the investigators made no false promises is best understood as a finding that they made no legally relevant false promises, i.e., no specific false promises of leniency, as distinct from vague assurances that cooperation would be in Dassey's best interests. Dassey has not shown by clear and convincing evidence that the finding was wrong. See § 2254(e)(1).