dissenting:
While I agree with the majority’s holding that Estep’s counsel was ineffective for repeatedly failing to object to inadmissible character evidence, I disagree with the conclusion that there has been no preju*242dice. The majority accurately states that the standard of review under AEDPA demands a high level of deference when considering ineffective assistance of counsel on federal habeas review. But, while AEDPA sets a high standard, it does not set an impossible one. I believe that the extensive, repetitive and pivotal use of prohibited evidence in this case meets the AEDPA standard.
The State’s sole discernible theme at trial — that Estep killed a “kind and gentle man” who was a virtual saint in his community — was a calculated attempt to use inadmissible evidence to pull at the heart strings of the jury. The framing of the theme early in opening argument makes the State’s objective clear: “After you discover the facts about this kind and gentle man, you’re going to find it especially painful to think that he lost his life while he was attempting to help a stranger.” This framing shows that the State made a flagrant appeal to the jury’s emotions as they relate to the victim’s character. The prosecution directly and unmistakably communicated to the jury that it should make a decision based on facts which should never have been admitted.
The pervasive, strategic use of prohibited good character evidence of the victim makes this case exceptional. I could find no case — and neither the majority nor the State points to any case — that approaches this level of abuse for the relevant eviden-tiary rules. Instead, the majority engages in mathematical calisthenics in an attempt to minimize the gravity of the disputed evidence. Such a simplistic approach demeans the spirit of the Sixth Amendment protections at issue. Prejudice is not a question of whether 10 percent or 25 percent or 50 percent of the prosecution’s evidence was inadmissible. It is a question of impact on the jury. When the prosecution communicates three paragraphs into its opening argument that it intends to persuade the jury using impermissible evidence and then successfully introduces and re-visits that impermissible evidence at each and every stage of the trial, it is unreasonable to conclude that the jury did not respond accordingly.
Here, the State elevated the victim to virtual sainthood. The prosecution continually illustrated its theme by introducing facts that presented the victim as: a “mentor”; the most “humble” of men; a “hardworking, honest, patriotic, patient” man; a man who “never complained”; a man “loved by his family”; a “generous” man; a man who carried on as a farmer only so that he could give “literally tons of food to people”; a man who cut firewood so that he could give it to people for free; a man who took care of his “frail and bed ridden” mother, tied her shoes, fed her, kept her company, and was her sole support and companionship. None of this information was even arguably admissible. But, the State took full advantage of Estep’s non-responsive counsel. With the reins of evi-dentiary law cast aside, the State took the opportunity to create a detailed portrait of the victim designed to win over the emotions of the jury. It would be unreasonable for any court to conclude that there was not, at very least, a reasonable probability that this portrait of a saintly man did not have a determinative effect on the jury’s mercy decision.
The majority adopts the district court’s tenuous reasoning that there was no prejudice because “the crime itself was particularly brutal.” This borders on tautology. Murder is particularly brutal by nature. What matters for our purposes is the jury’s perception of brutality. Any attempt to place this crime on some imaginary spectrum between “brutal” and “humane” would be inextricably entangled with the inadmissible character evidence. The conclusion that a jury would not con*243sider the murder of a saint-like man whom his family and community depended on more brutal than the murder of a relatively anonymous victim ignores the basic human capacity for empathy.
The majority also reasons that there is no prejudice in this trial because “the good character evidence did not detract from the defense’s ability to present a strong affirmative case for mercy.” Indeed, Es-tep presented evidence that showed he was just eighteen-years old when he committed his crime, that he was clinically depressed because he had recently lost his mother and brother, that he had begun to abuse prescription drugs that his brother had given to him, that he did not intend to kill his victim, and that he took responsibility for the crime. But, the majority pretends this evidence somehow works against a finding of prejudice. Quite to the contrary, this evidence only moves the case closer to the tipping point on the question of mercy. It makes it all the more likely that the inadmissible character evidence was outcome determinative.
The majority also believes its holding sound because Estep did not immediately display remorse after his crime was complete, but instead made several purchases of trivial items with the money he had stolen. There are two problems with this line of reasoning. First, remorse is not a time-limited emotion. While Estep may not have felt remorse in the hours or days after his crime, that does not preclude a finding that he was remorseful at trial. See United States v. Rodriguez, 959 F.2d 193, 197 (11th Cir.1992) (explaining that conduct illustrating remorse for the purpose of mitigation may take place “prior to, during, and after the trial”). The second, related problem with the majority’s reasoning is that Estep’s petty purchases after the crime serve to reinforce the pro-mercy argument that Estep was young and immature at the time.
This case does present a challenge in that West Virginia law does not provide concrete factors to guide jury deliberation on mercy. See State v. Miller, 178 W.Va. 618, 363 S.E.2d 504, 508-09(1987). Instead, the jury considers the evidence as a whole in deciding whether to recommend mercy. Id. While this legal framework makes review under AEDPA difficult, it cannot completely insulate a state court. Such a conclusion would mean that a defendant’s Sixth Amendment right to effective counsel only adheres when the legal underpinnings provide for clear and convenient review.
Our precedent instructs us to look at the totality of the evidence when considering prejudicial effect on habeas review. Elmore v. Ozmint, 661 F.3d 783, 868 (4th Cir.2011). The totality of the evidence before us in this case establishes — at very minimum — a reasonable probability that a jury would have made a recommendation for mercy if Estep’s counsel properly objected to the good character evidence of the victim. I do not believe that it is reasonable to conclude otherwise. From soup to nuts, the State’s case emphasized and relied upon inadmissible character evidence. Our complicity in this sort of egregious abuse of the rules of evidence only serves as a signal to certain prosecutors that they are free to play fast and loose when the opportunity presents itself.
One court recently noted in an en banc opinion that “if we succumb to the temptation to abdicate our responsibility on habe-as review, we might as well get ourselves a big, fat rubber stamp, pucker up, and kiss the Great Writ good-bye.” Doody v. Ryan, 649 F.3d 986, 1003 (9th Cir.2011). I fear the inkpad may be opening.