Rosario v. Ercole

08-5521-pr Rosario v. Ercole 1 UNITED STATES COURT OF APPEALS 2 3 F OR THE S ECOND C IRCUIT 4 5 6 7 August Term, 2009 8 9 (Argued: November 19, 2009 Decided: April 12, 2010) 10 11 Docket No. 08-5521-pr 12 13 14 R ICHARD R OSARIO, 15 16 Petitioner-Appellant, 17 18 –v.– 19 20 S UPT. R OBERT E RCOLE, G REEN H AVEN C ORRECTIONAL F ACILITY, A TTORNEY G ENERAL 21 E LIOT S PITZER, 22 23 Respondents-Appellees. 24 25 26 27 Before: 28 C ABRANES, S TRAUB, W ESLEY, Circuit Judges. 29 30 Richard Rosario appeals from a judgment of the United 31 States District Court for the Southern District of New York 32 (Castel, J.), entered on October 23, 2008, denying his 33 petition for a writ of habeas corpus. We hold that the 34 state court’s review of Rosario’s ineffective assistance of 35 counsel claims was neither contrary to, nor an unreasonable 36 application of, Strickland v. Washington, 466 U.S. 668 37 (1984). 38 39 Affirmed. Judge Straub concurs in part and dissents in 40 part in a separate opinion. Page 1 of 69 1 2 3 J ODI K. M ILLER, Morrison & Foerster, LLP, New York, 4 NY (Carl H. Loewenson, Morrison & Foerster, 5 LLP, New York, NY, and Jin Hee Lee, NAACP 6 Defense and Education Fund, Inc., on the brief 7 ), for Petitioner-Appellant. 8 9 J OSEPH N. F ERDENZI, Assistant District Attorney, 10 Bronx, NY (Christopher J. Blira-Koessler, 11 Assistant District Attorney, Bronx, NY for 12 Robert T. Johnson, District Attorney, Bronx 13 County), for Respondents-Appellees. 14 15 16 17 W ESLEY, Circuit Judge: 18 This case requires us to examine New York law and 19 analyze one sentence in a New York Court of Appeals opinion 20 that has troubled our circuit since its publication. 21 Background 22 On June 19, 1996, George Collazo was shot and killed in 23 the Bronx while walking with his friend Michael Sanchez. 24 The daytime shooting followed an argument sparked by 25 Collazo’s racial epithet to two men as he and Sanchez passed 26 them. Sanchez later identified appellant Richard Rosario as 27 Collazo’s assailant. Robert Davis, a porter working at a 28 nearby building, witnessed the murder and also identified 29 Rosario as the shooter. A third eyewitness was also Page 2 of 69 1 present, but did not identify Rosario as a participant in 2 the crime. 3 Rosario was arrested for the murder on July 1, 1996, 4 after he voluntarily returned to New York from Florida. 5 From the time of his arrest, Rosario claimed he was in 6 Florida when Collazo was shot. Rosario provided the police 7 with a statement, maintained his innocence, and listed the 8 names of thirteen people who could corroborate his alibi. 9 Before Rosario’s trial began, he was assigned Joyce 10 Hartsfield as counsel. Hartsfield brought an application 11 before the court requesting funds for a private investigator 12 to travel to Florida and interview the potential alibi 13 witnesses. The court granted the application. Hartsfield 14 was eventually replaced as counsel by Steven Kaiser in 15 February of 1998. Kaiser had a mistaken belief that the 16 application for investigation fees had been denied. Kaiser 17 did not make a request for fees; no investigation of alibi 18 witnesses was done in Florida. 19 During the trial, the prosecution called Sanchez and 20 Porter, who identified Rosario as the shooter, and the third 21 eyewitness, who failed to identify Rosario. The defense 22 presented two alibi witnesses — John Torres, a friend of Page 3 of 69 1 Rosario, and Jenine Seda, John Torres’ fiancée. Both 2 testified that Rosario was living with them in Florida when 3 the murder occurred. They remembered the date because their 4 first child was born on June 20th, a day after the murder. 5 Rosario took the stand in his own defense and testified 6 that he was in Florida through June 30, 1996. Rosario 7 stated he lived with a woman named Shannon Beane from 8 February through April of 1996. The prosecution rebutted 9 this assertion with Rosario’s Florida arrest record, which 10 indicated that he was arrested in March of 1996 and 11 imprisoned until April of that year. The jury convicted 12 Rosario of second degree murder, and the court sentenced him 13 to 25 years to life. 14 After Rosario’s unsuccessful direct appeal of his 15 conviction, see People v. Rosario, 733 N.Y.S.2d 405 (1st 16 Dep’t 2001), leave denied 97 N.Y.2d 760 (2002), he filed a 17 motion to vacate his conviction under Section 440.10(1) 1 of 1 The relevant part of the New York statute governing a motion to vacate a judgment reads: “At any time after the entry of a judgment, the court in which it was entered may, upon motion of the defendant, vacate such judgment upon the ground that: . . . (h) The judgment was obtained in violation of a right of the defendant under the constitution of this state or of the United States.” N.Y. Crim. Proc. Law § 440.10(1). Page 4 of 69 1 the New York Criminal Procedure Law on the grounds that he 2 was deprived effective assistance of counsel at trial. The 3 Bronx County Supreme Court held a hearing, at which 4 Rosario’s attorneys (Hartsfield and Kaiser), the private 5 investigator, and seven alibi witnesses testified. 6 Hartsfield testified that she did not pursue documentary 7 records to support Rosario’s alibi defense, including 8 records from Western Union that were subsequently destroyed 9 and a police field report detailing Rosario’s stop by 10 Florida police on May 30, 1996. She also testified that, 11 though she retained a private investigator and received 12 funding from the court to send the investigator to Florida 13 to investigate the alibi witnesses, she did not instruct the 14 private investigator to do so. She conceded there was no 15 strategic reason behind that choice. 16 Kaiser, for his part, stated that he did not know where 17 he got the misimpression that the court had denied 18 investigatory funds. He testified that he did attempt to 19 locate or contact alibi witnesses in Florida, working from 20 New York alone. When asked if the two alibi witnesses he 21 called were the best witnesses, he replied “they were the 22 only two,” and he would have preferred to call additional Page 5 of 69 1 alibi witnesses. 2 Jesse Franklin, the private investigator, testified 3 that she had a meeting with Rosario where he provided her 4 with a list of names for alibi witnesses. She attempted to 5 reach all the people on the list via telephone, though it 6 was difficult to do so because many of them had moved. 7 Franklin raised these difficulties with Hartsfield, who 8 instructed her to draft an affidavit detailing her 9 difficulties for an application to the court for additional 10 investigatory funds to send Franklin to Florida. She 11 believed traveling to Florida was necessary to investigate 12 properly Rosario’s alibi. She never heard from Hartsfield 13 again about the application and assumed that it had been 14 denied. Despite not traveling to Florida, Franklin did 15 manage to contact two of the witnesses on the list, Fernando 16 and Robert Torres, both of whom told Franklin that they had 17 seen Rosario in Florida in late June of 1996. Franklin did 18 not contact those men again. However, Franklin did later 19 contact the two witnesses who were actually called at trial, 20 Jenine Seda and John Torres, and was told by John Torres 21 that he could provide the names of other alibi witnesses. 22 Franklin tried unsuccessfully to telephone other witnesses Page 6 of 69 1 that Rosario had named. 2 At the end of the hearing, the state court concluded 3 that Hartsfield and Kaiser had provided Rosario with 4 “meaningful representation” under New York law. The court 5 detailed the testimony of each witness, and concluded that 6 the two witnesses presented at trial were the “most credible 7 among the possible alibi witnesses.” Rosario v. Ercole, 582 8 F. Supp. 2d 541, 550 (S.D.N.Y. 2008). The court also 9 determined that the testimony of several of the proffered 10 alibi witnesses could have undermined Rosario’s alibi 11 defense in the eyes of the jury. 12 The state court noted that Rosario’s right to effective 13 assistance of counsel was guaranteed by both the federal and 14 state constitutions. The court contrasted the federal 15 standard set forth in Strickland with the New York standard 16 employed under the state constitution. After a lengthy 17 analysis under the New York constitutional standard, the 18 court concluded that Rosario had received “meaningful 19 representation” as required by New York’s Constitution. The 20 court also concluded that the government’s case was 21 “strong”; that the prospective alibi witnesses “were, for Page 7 of 69 1 the most part, questionable and certainly not as persuasive 2 as the two witnesses who did testify”; and that the verdict 3 was “unimpeached, and ‘amply supported by the evidence.’” 2 4 Rosario filed a petition for a writ of habeas corpus in 5 the United States District Court for the Southern District 6 of New York (Castel, J.). Rosario v. Ercole, 582 F. Supp. 7 2d 541 (S.D.N.Y. 2008). The district court requested a 8 report and recommendation from a magistrate judge (Pitman, 9 M.J.). Id. at 545. The magistrate judge and the district 10 court concluded that counsels’ performance was in fact 11 deficient under Strickland. Id. at 551. However, both 12 determined that the state court’s decision to deny Rosario’s 13 motion to vacate was not an unreasonable application of, nor 14 contrary to, clearly established federal law. Id. at 552- 15 53. This appeal followed. 16 Discussion 17 Under the Antiterrorism and Effective Death Penalty Act 2 Upon appeal, the New York Appellate Division, First Department, did not address the ineffective assistance claim. People v. Rosario, 733 N.Y.S.2d 405 (1st Dep’t 2001). The New York Court of Appeals denied leave to appeal. People v. Rosario, 97 N.Y.2d 760, 760 (2002) (Ciparick, J.). Page 8 of 69 1 of 1996 (“AEDPA”), a federal court may only grant a writ of 2 habeas corpus for a claim that has been adjudicated on the 3 merits by a state court if the adjudication of the claim: 4 (1) resulted in a decision that was contrary to, or 5 involved an unreasonable application of, clearly 6 established Federal law, as determined by the Supreme 7 Court of the United States; or 8 (2) resulted in a decision that was based on an 9 unreasonable determination of the facts in light of the 10 evidence presented in the State court proceeding. 11 28 U.S.C. § 2254(d). 12 Rosario argues that the state court decision denying 13 his claim for ineffective assistance of counsel was both an 14 unreasonable application of, and contrary to, the clearly 15 established federal standard under the first subsection of § 16 2254(d). Because the state court adjudicated the merits of 17 his claim, Rosario must prove that the state court either 18 identified the federal standard for ineffective assistance 19 but applied that standard in an objectively unreasonably 20 way, or that the state applied a rule that contradicts the 21 federal standard. Lockyer v. Andrade, 538 U.S. 63, 73, 75- 22 76 (2003); Williams v. Taylor, 529 U.S. 362, 387-89 (2000). 23 We review the district court’s denial of the writ de novo. 24 Jones v. West, 555 F.3d 90, 95 (2d Cir. 2009). Page 9 of 69 1 Rosario argues that the state court ran afoul of 2 federal law when it concluded that he had received effective 3 representation. In Rosario’s view, counsels’ failure to 4 investigate Rosario’s alibi witnesses and documentary 5 evidence was a violation of his constitutional right to the 6 effective assistance of counsel, and any conclusion 7 otherwise misapprehends clearly established law. 8 In Williams v. Taylor, the Supreme Court determined 9 that Strickland v. Washington, the seminal case defining the 10 contours of the right to effective assistance of counsel, 11 qualified as “clearly established law” for purposes of 12 AEDPA. 529 U.S. at 390-91. The Strickland test for 13 ineffective assistance has two necessary components: the 14 defendant must establish both that his attorney was 15 ineffective and that the attorney’s errors resulted in 16 prejudice to the defendant. Id.; see also Strickland v. 17 Washington, 466 U.S. 668, 687 (1984). Counsel is 18 ineffective when her efforts fall “‘below an objective 19 standard of reasonableness.’” Williams, 529 U.S. at 390-91 20 (quoting Strickland, 466 U.S. at 688). A defendant 21 satisfies the prejudice prong by proving that “‘there is a Page 10 of 69 1 reasonable probability that, but for counsel’s 2 unprofessional errors, the result of the proceeding would 3 have been different. A reasonable probability is a 4 probability sufficient to undermine confidence in the 5 outcome.’” Id. at 391 (quoting Strickland, 466 U.S. at 6 694). 7 When a federal court reviews a state court decision 8 under § 2254, “[t]he question is not whether a federal court 9 believes the state court’s determination under the 10 Strickland standard was incorrect but whether that 11 determination was unreasonable — a substantially higher 12 threshold.” Knowles v. Mirzayance, - - - U.S. - - - -, 129 13 S. Ct. 1411, 1420 (2009) (internal quotation marks omitted). 14 The Strickland standard itself is a “general standard,” 15 meaning its application to a specific case requires “a 16 substantial element of judgment” on the part of the state 17 court. Yarborough v. Alvarado, 541 U.S. 652, 664 (2004); 18 accord Knowles, 129 S.Ct. at 1420. Thus, state courts are 19 granted “even more latitude to reasonably determine that a 20 defendant has not satisfied that standard.” Knowles, 129 21 S.Ct. at 1420. In order to prevail, a petitioner must 22 overcome that substantial deference and establish that the Page 11 of 69 1 state court’s decision on ineffective assistance was 2 contrary to, or an unreasonable application of, Strickland. 3 To be “contrary to” clearly established law, a state 4 court must reach a conclusion of law antithetical to a 5 conclusion of law by the Supreme Court, or decide a case 6 differently than the Supreme Court has when the two cases 7 have “materially indistinguishable facts.” Williams, 529 8 U.S. at 412-13. The state court examined Rosario’s claims 9 under New York’s constitutional standard for ineffective 10 assistance. New York’s constitution, like the U.S. 11 Constitution, affords its citizens with the right to 12 competent representation by an attorney. See U.S. Const. 13 amend. VI; N.Y. Const. art. I, § 6; see also People v. 14 Baldi, 54 N.Y.2d 137, 146 (1981). However, as noted by the 15 state court, New York’s test for ineffective assistance of 16 counsel under the state constitution differs from the 17 federal Strickland standard. The first prong of the New 18 York test is the same as the federal test; a defendant must 19 show that his attorney’s performance fell below an objective 20 standard of reasonableness. People v. Turner, 5 N.Y.3d 476, 21 480 (2005). The difference arises in the second prong of 22 the Strickland test. Id. In New York, courts need not find Page 12 of 69 1 that counsel’s inadequate efforts resulted in a reasonable 2 probability that, but for counsel’s unprofessional errors, 3 the result of the proceeding would have been different. 4 Instead, the “question is whether the attorney’s conduct 5 constituted egregious and prejudicial error such that 6 defendant did not receive a fair trial.” People v. 7 Benevento, 91 N.Y.2d 708, 713 (1998) (internal quotation 8 marks omitted). Thus, under New York law the focus of the 9 inquiry is ultimately whether the error affected the 10 “fairness of the process as a whole.” Id. at 714. The 11 efficacy of the attorney’s efforts is assessed by looking at 12 the totality of the circumstances and the law at the time of 13 the case and asking whether there was “meaningful 14 representation.” Baldi, 54 N.Y.2d at 147. 15 The New York Court of Appeals clearly views the New 16 York constitutional standard as more generous toward 17 defendants than Strickland. Turner, 5 N.Y.3d at 480 (“Our 18 ineffective assistance cases have departed from the second 19 (‘but for’) prong of Strickland, adopting a rule somewhat 20 more favorable to defendants.” (citing cases)). To meet the 21 New York standard, a defendant need not demonstrate that the 22 outcome of the case would have been different but for Page 13 of 69 1 counsel’s errors; a defendant need only demonstrate that he 2 was deprived of a fair trial overall. People v. Caban, 5 3 N.Y.3d 143, 155-56 (2005). A single error by otherwise 4 competent counsel may meet this standard if that error 5 compromised the integrity of the trial as a whole. Turner, 6 5 N.Y.3d at 480. 7 For our part, we have recognized that the New York 8 “meaningful representation” standard is not contrary to the 9 Strickland standard. Eze v. Senkowski, 321 F.3d 110, 123-24 10 (2d Cir. 2003); Lindstadt v. Keane, 239 F.3d 191, 198 (2d 11 Cir. 2001). However, some of our colleagues have cautioned 12 that there may be applications of the New York standard that 13 could be in tension with the prejudice standard in 14 Strickland. Henry v. Poole, 409 F.3d 48, 70-71 (2d Cir. 15 2005). The primary source of this consternation is a 16 sentence from a New York Court of Appeals decision, 17 Benevento, which notes that “whether defendant would have 18 been acquitted of the charges but for counsel’s errors is 19 relevant, but not dispositive under the State constitutional 20 guarantee of effective assistance of counsel.” 91 N.Y.2d at 21 714. Of course, under Strickland, if a defendant would have 22 been acquitted but for counsel’s errors, that fact is both Page 14 of 69 1 relevant and dispositive because it creates more than a 2 reasonable probability of a different outcome and thus 3 soundly passes the prejudice prong of the test. See 4 Strickland, 466 U.S. at 694. 5 The problem is that focusing solely on this sentence 6 leads one to ignore the context in which it was written. 7 Benevento recognized that, like Strickland, “a claim of 8 ineffective assistance of counsel will be sustained only 9 when it is shown that counsel partook ‘an inexplicably 10 prejudicial course.’” Benevento, 91 N.Y.2d at 713 (quoting 11 People v. Zaborski, 59 N.Y.2d 863, 865 (1983)). However, 12 the New York Court of Appeals carefully noted that, prior to 13 Strickland, New York had “developed a somewhat different 14 test for ineffective assistance of counsel under article I, 15 § 6 of the New York Constitution from that employed by the 16 Supreme Court in applying the Sixth Amendment.” Id. 17 (quoting People v. Claudio, 83 N.Y.2d 76, 79 (1993)). 18 Benevento explained that in New York “‘prejudice’ is 19 examined more generally in the context of whether defendant 20 received meaningful representation.” Id. Because the 21 concept of prejudice in New York’s ineffective assistance of 22 counsel jurisprudence focuses on the quality of Page 15 of 69 1 representation provided and not simply the “but for” 2 causation chain, New York has “refused to apply the harmless 3 error doctrine in cases involving substantiated claims of 4 ineffective assistance.” Id. at 714 (citing cases). In New 5 York, even in the absence of a showing that but for 6 counsel’s errors the outcome would be different, a defendant 7 may still have an ineffective assistance claim under New 8 York’s constitution. Even if the errors are harmless in the 9 sense that the outcome would remain the same, a defendant 10 may still meet the New York prejudice standard by 11 demonstrating that the proceedings were fundamentally 12 unfair. See People v. Stultz, 2 N.Y.3d 277, 283-84 (2004). 13 This is not a novel view — New York state courts have 14 repeatedly asserted that the New York standard is, in 15 practice and in intent, more generous to defendants than the 16 federal standard. See, e.g., People v. Ozuna, 7 N.Y.3d 913, 17 915 (2006); Turner, 5 N.Y.3d at 480 (collecting cases). 18 Federal courts faced with the New York standard should view 19 it as such. 20 The concern this Court expressed in dicta in Henry v. 21 Poole about the New York state standard was misplaced. The 22 Henry panel wrote, “we find it difficult to view so much of Page 16 of 69 1 the New York rule as holds that ‘whether defendant would 2 have been acquitted of the charges but for counsel’s errors 3 is . . . not dispositive,’ as not ‘contrary to’ the 4 prejudice standard established by Strickland.” 409 F.3d at 5 71 (internal citation omitted). However, it is hard to 6 envision a scenario where an error that meets the prejudice 7 prong of Strickland would not also affect the fundamental 8 fairness of the proceeding. The very opinion from which the 9 troublesome phrase was drawn — Benevento — affirmatively 10 stated that even a “harmless error” could undermine the 11 fairness of the process in such a way that violates the 12 state’s constitutional guarantee of effective assistance. 13 See Benevento, 91 N.Y.2d at 714. What case, then, could 14 present the converse, an error so egregious that it most 15 likely influenced the outcome of the trial, but did not 16 cripple the fundamental fairness of the proceedings? We can 17 think of none. Fundamental fairness analysis by its nature 18 must always encompass prejudice. 19 The New York standard is not without its problems. In 20 defining prejudice to include “the context of whether 21 defendant received meaningful representation,” Benevento, 91 22 N.Y.2d at 713, New York has, to some degree, combined the Page 17 of 69 1 two prongs of Strickland. Prejudice to the defendant, 2 meaning a reasonable possibility of a different outcome, is 3 but one factor of determining if the defendant had 4 meaningful representation. New York courts look at the 5 effect of the attorney’s shortcomings as part of the 6 equation in deciding if the defendant received the benefit 7 of competent counsel. This approach, and the language of 8 Benevento, creates a danger that some courts might 9 misunderstand the New York standard and look past a 10 prejudicial error as long as counsel conducted himself in a 11 way that bespoke of general competency throughout the trial. 12 That would produce an absurd result inconsistent with New 13 York constitutional jurisprudence and the mandates of 14 Strickland. Properly applied, however, this standard is not 15 contrary to Strickland and, in the case before us, the court 16 properly applied the standard. 17 The trial court’s decision 3 addressing the ineffective 18 assistance of counsel claim did recite the troublesome 19 phrase from Benevento, and added a footnote that read: “The 3 Because the state court appeals did not address the ineffective assistance of counsel claim, we look to the trial court’s analysis of the issue. See Ylst v. Nunnemaker, 501 U.S. 797, 803 (1991). Page 18 of 69 1 federal standard for allegations of ineffective assistance 2 of counsel, which was set forth in Strickland v. Washington, 3 requires a showing that the attorney’s performance was 4 deficient and that, but for the attorneys[’] errors, the 5 result of the proceeding would have been different, was 6 expressly rejected in this case.” (internal quotation marks 7 and citation omitted). Rosario argues that this alone is 8 enough to establish his claim under Federal law. But as 9 noted above, New York’s rejection of Strickland was in the 10 context of recognizing a state constitutional right that is 11 more protective of a defendant’s right to an effective 12 attorney, and not because Strickland is too generous. 13 As the Henry panel recognized, this Court has 14 repeatedly held that application of the New York state 15 standard is not contrary to Strickland. See, e.g., Eze, 321 16 F.3d at 123-24. And, as the Henry panel also recognized, 17 “in the absence of a contrary decision by this Court en 18 banc, or an intervening Supreme Court decision, we are bound 19 to follow the precedents . . . that the N.Y. Court of 20 Appeals standard is not ‘contrary to’ Strickland.” Henry, 21 409 F.3d at 70. We emphasize again that the New York state 22 standard for ineffective assistance of counsel is not Page 19 of 69 1 contrary to Strickland. 2 The only avenue of reprieve available to Rosario then 3 is to establish that the state court unreasonably applied 4 Strickland. A state court “unreasonably applies” clearly 5 established law when it identifies the correct legal 6 principle from Supreme Court jurisprudence, but unreasonably 7 applies the principle to the case before it. Williams, 529 8 U.S. at 412-13. 9 In order to prevail, Rosario must first satisfy the 10 prongs of Strickland on de novo review of the merits. See 11 Henry, 409 F.3d at 67. The magistrate judge and the 12 district court concluded that Rosario had done so. We see 13 no need to quibble with those conclusions because, like the 14 magistrate judge and the district court judge, we agree that 15 the New York court’s application of Strickland — albeit in 16 the terms of New York cases — was not an unreasonable 17 application of the federal standard. 18 For us to find that the state court unreasonably 19 applied Strickland, we must uncover an “increment of 20 incorrectness beyond error.” Francis S. v. Stone, 221 F.3d 21 100, 111 (2d Cir. 2000). The increment need not be great, 22 but simply disagreeing with the outcome is insufficient. Page 20 of 69 1 Id.; see also Williams, 529 U.S. at 410. This is so even 2 if, as here, we conclude both prongs of Stickland have been 3 met. “[A] state prisoner seeking a federal writ of habeas 4 corpus on the ground that he was denied effective assistance 5 of counsel must show more than simply that he meets the 6 Strickland standard. . . . [T]he state court’s decision 7 rejecting his claim is to be reviewed under a more 8 deferential standard than simply whether that decision was 9 correct.” Henry, 409 F.3d at 67. 10 As noted above, the state court conducted an extensive 11 hearing in response to Rosario’s motion to vacate his 12 conviction under New York Criminal Procedure Law § 440.10 13 due to ineffective assistance of counsel. The hearing 14 lasted over a month. After the hearing, Justice Davidowitz 15 issued a lengthy decision, reviewing the evidence presented 16 and detailing his conclusions on Rosario’s claims. While we 17 may disagree with Justice Davidowitz’s findings (and indeed 18 our dissenting colleague does), we cannot say that he 19 unreasonably applied federal law. 20 As the district court stated: “[t]hough not delivered 21 in Strickland terminology, the state court opinion ruled 22 that 1.) Rosario was effectively represented in his alibi Page 21 of 69 1 defense, and 2.) that his representation did not undermine 2 confidence in the jury’s verdict.” Rosario, 582 F. Supp. 2d 3 at 553. Examining both the efforts of counsel and the alibi 4 witnesses presented, Justice Davidowitz concluded: “By any 5 standard, Ms. Hartsfield and Mr. Kaiser represented 6 defendant in a thoroughly professional, competent, and 7 dedicated fashion and not in accord with the issues of 8 ineffectiveness. . . . [T]he errors or omissions suggested 9 by the defendant do not alter this finding or rise to that 10 level.” (emphasis added). Justice Davidowitz noted that “an 11 investigation was conducted . . . and, most importantly, a 12 credible alibi defense was presented to the jury.” He found 13 that the two witnesses presented at trial were Rosario’s 14 best alibi witnesses. Justice Davidowitz labeled Kaiser’s 15 decision not to present the police reports detailing 16 Collazo’s fight a “perfectly reasonable and appropriate” 17 strategy. To put it in terms of Strickland, Justice 18 Davidowitz did not find that the performance of counsel was 19 objectively unreasonable. 20 Justice Davidowitz then examined in great detail the 21 testimony of the alibi witnesses presented at the hearing. 22 The court noted that the two alibi witnesses that were Page 22 of 69 1 presented at trial “had the best reason for remembering why 2 defendant was present in Florida on June 19[,] 1996 — the 3 birth of their son — an event that was more relevant for 4 them than the events relied upon by the other witnesses.” 5 He expressed skepticism as to the probative value of the 6 witnesses presented at the hearing, calling the evidence “in 7 some cases questionable and in others [raising] issues which 8 could have created questions for a deliberating jury. For 9 example, two of the witnesses — Lisette Rivero[] and Denise 10 Hernandez — could not say where the defendant was on June 19 11 and 20.” The judge “studied closely” the alibi witnesses 12 presented at the hearing, and concluded they were “for the 13 most part, questionable and certainly not as persuasive as 14 the two witnesses who did testify, and were rejected by the 15 jury” and the testimony they would have provided was 16 “largely” cumulative. In spite of the failure to call the 17 alibi witnesses, Justice Davidowitz determined “this jury 18 verdict was unimpeached and amply supported by the 19 evidence.” (internal quotation marks omitted and emphasis 20 added). Translated into the language of Strickland, Justice 21 Davidowitz concluded that there was not a reasonable 22 probability that the outcome of the trial would be different Page 23 of 69 1 but for counsel’s errors. 2 Justice Davidowitz conducted a thorough hearing, 3 assessing the credibility of the potential witnesses first- 4 hand. He concluded that the two witnesses called at trial 5 were the best witnesses to represent Rosario’s alibi 6 defense, and that the other witnesses were “questionable and 7 certainly not as persuasive as the two witnesses who did 8 testify, and were rejected by the jury.” He considered the 9 prejudicial effect of the errors, and concluded that the 10 outcome of the trial would not have been different but for 11 those errors — the guilty verdict, in his words, remained 12 “unimpeached.” He adhered to the New York state standard 13 and found counsel to have been effective. Whether our own 14 cold reading of the record would lead us to this conclusion 15 is of no moment; we must presume the state court’s findings 16 of fact are correct and can only be rebutted by clear and 17 convincing evidence otherwise. Lynn v. Bliden, 443 F.3d 18 238, 246 (2d Cir. 2006) (citing 28 U.S.C. § 2254(e)). 19 Justice Davidowitz’s analysis need not employ the 20 language of a federal court’s de novo review in order to 21 pass AEDPA muster. See Coleman v. Thompson, 501 U.S. 722, 22 739 (1991). While he did not explicitly review the evidence Page 24 of 69 1 under the Strickland standard, the import was the same. 2 Conflating the two prongs of Strickland does not violate 3 AEDPA — different is not per se unreasonable. Here, Justice 4 Davidowitz did not find that counsel’s performance was 5 objectively unreasonable, nor did he find that the 6 fundamental fairness of the trial was harmed by counsel’s 7 errors. On this record, we cannot say that the state court 8 unreasonably applied the tenets of Strickland. Therefore, 9 consistent with the standards of AEDPA, we agree with the 10 district court that the writ must be denied. 11 We have reviewed Rosario’s additional arguments and 12 find them to be without merit. 13 Conclusion 14 The district court’s judgment of October 23, 2008, 15 denying the petition for the writ of habeas corpus is hereby 16 A FFIRMED. 17 Page 25 of 69 1 S TRAUB, Circuit Judge, dissenting in part, concurring in 2 part: 3 The principal issue in this appeal is whether the state 4 court ruling on Rosario’s motion to vacate his conviction 5 pursuant to New York Criminal Procedure Law § 440.10 was 6 objectively unreasonable in holding that Rosario received 7 effective assistance of counsel in accordance with the Sixth 8 Amendment of the United States Constitution under Strickland 9 v. Washington, 466 U.S. 668 (1984). As I believe it was, I 10 must respectfully dissent. Rosario raises two additional 11 claims on appeal. Because I would conditionally grant 12 Rosario’s petition on the basis of ineffective assistance of 13 counsel, I believe it unnecessary to reach his claim under 14 Batson v. Kentucky, 476 U.S. 79 (1986). I concur only in 15 the majority’s rejection of Rosario’s actual innocence 16 claim. 17 This appeal presents an extraordinarily troubling set 18 of circumstances. During the pendency of his prosecution, 19 Rosario consistently maintained, both to the police and to 20 his criminal defense attorneys, that he was in Florida on 21 the day of the Bronx murder and on multiple occasions 22 provided a list of up to thirteen alibi witnesses to Page 26 of 69 1 corroborate this claim. Rosario’s defense attorneys 2 nevertheless failed to investigate his alibi defense 3 adequately and did not contact many of these potential 4 witnesses. They offer no strategic reason for not doing so 5 and, indeed, concede that such an investigation was 6 essential to Rosario’s defense. Their explanation for this 7 failure is that they mistakenly believed that the state 8 trial court had denied Rosario’s application for fees to 9 cover the investigatory expenses, when in fact the court had 10 clearly granted the application. Such conduct plainly falls 11 below acceptable professional standards, satisfying 12 Strickland’s performance prong. Strickland, 466 U.S. at 13 687. 14 As a result of this colossal failure, Rosario’s trial 15 counsel presented a relatively weak alibi defense, 16 consisting of only two alibi witnesses who were subject to 17 impeachment as interested witnesses because they were close 18 friends with Rosario. It is now clear that had Rosario’s 19 defense attorneys followed through in investigating his 20 alibi defense, they would have had the opportunity to call 21 at least seven additional alibi witnesses at trial. These 22 witnesses would have provided corroboration and supplied Page 27 of 69 1 distinct facts relating to Rosario’s presence in Florida on 2 and around the day of the murder, adding further context and 3 credibility to his alibi defense; moreover, a number of 4 these additional witnesses would not have been as vulnerable 5 to impeachment as interested witnesses as were the two trial 6 witnesses because they are not as close with Rosario. 7 Moreover, the prejudice in this case is worsened because the 8 only evidence of Rosario’s guilt was the testimony of two 9 stranger eyewitnesses. There is no question, in my opinion, 10 that had the additional alibi witnesses who were presented 11 in connection with Rosario’s post-conviction motion 12 testified at trial, there is a reasonable probability that 13 the jury’s verdict would have been different, satisfying the 14 prejudice prong of the Strickland analysis. Id. 15 While the majority appears to agree with this much of 16 the analysis, our opinions diverge where I further conclude 17 that the state court’s holding to the contrary was not 18 merely error, but an unreasonable application of Strickland. 19 I come to this conclusion, as I must, because there exists 20 too much alibi evidence that was not presented to the jury, 21 and too little evidence of guilt, to now have any confidence 22 in the jury’s verdict. In sum, I would conditionally grant Page 28 of 69 1 the petition because it was objectively unreasonable both to 2 sanction counsel’s failure to investigate Rosario’s alibi 3 defense as reasonable and to find no reasonable probability 4 that the verdict would have been different if the jury had 5 heard the significant alibi evidence that Rosario’s defense 6 attorneys neither uncovered nor presented. 7 I. Ineffective Assistance of Counsel 8 The majority does not dispute that Rosario received 9 constitutionally ineffective assistance of counsel under 10 Strickland, but views the state court’s decision to the 11 contrary as within the bounds of permissible error. 12 Engaging in the Strickland analysis is helpful to underscore 13 why I must disagree with the majority’s conclusion that the 14 state court did not unreasonably apply the precedent. 15 Under Strickland, to establish ineffective assistance 16 of counsel, Rosario “must (1) demonstrate that his counsel’s 17 performance fell below an objective standard of 18 reasonableness in light of prevailing professional norms; 19 and (2) affirmatively prove prejudice arising from counsel’s 20 allegedly deficient representation.” Carrion v. Smith, 549 21 F.3d 583, 588 (2d Cir. 2008) (internal quotation marks 22 omitted). “To satisfy the first prong – the performance Page 29 of 69 1 prong – the record must demonstrate that ‘counsel made 2 errors so serious that counsel was not functioning as the 3 “counsel” guaranteed the defendant by the Sixth Amendment.’” 4 Wilson v. Mazzuca, 570 F.3d 490, 502 (2d Cir. 2009) 5 (quoting Strickland, 466 U.S. at 687). “[S]trategic choices 6 made after thorough investigation of law and facts relevant 7 to plausible options are virtually unchallengeable,” 8 Strickland, 466 U.S. at 690, and even “strategic choices 9 made after less than complete investigation do not amount to 10 ineffective assistance – so long as the known facts made it 11 reasonable to believe that further investigation was 12 unnecessary,” Henry v. Poole, 409 F.3d 48, 63 (2d Cir. 2005) 13 (citing Strickland, 466 U.S. at 690-91), cert. denied, 547 14 U.S. 1040 (2006). By contrast, “omissions [that] cannot be 15 explained convincingly as resulting from a sound trial 16 strategy, but instead arose from oversight, carelessness, 17 ineptitude, or laziness,” may fall below the constitutional 18 minimum standard of effectiveness. Wilson, 570 F.3d at 502 19 (alteration in original) (quoting Eze v. Senkowski, 321 F.3d 20 110, 112 (2d Cir. 2003)). To satisfy the second prong – the 21 prejudice prong – a “defendant must show that there is a 22 reasonable probability that, but for counsel’s Page 30 of 69 1 unprofessional errors, the result of the proceeding would 2 have been different. A reasonable probability is a 3 probability sufficient to undermine confidence in the 4 outcome.” Strickland, 466 U.S. at 694. 5 A. Performance Prong 6 Defense counsel has a “duty to make reasonable 7 investigations or to make a reasonable decision that makes 8 particular investigations unnecessary.” Wiggins v. Smith, 9 539 U.S. 510, 521 (2003) (quoting Strickland, 466 U.S. at 10 690). Rosario’s pre-trial and trial counsel did neither. 11 From his arrest to the present, Rosario has consistently 12 maintained that he was in Florida on the day of the murder. 13 At every juncture of this case, he has disclosed the 14 substance of his alibi defense and the names of the 15 individuals who could corroborate it, including in his post- 16 arrest statement on the day he voluntarily surrendered to 17 the police and to both of his defense counsel thereafter. 18 Nevertheless, his attorneys abdicated their duty to 19 investigate a majority of these individuals because of their 20 mistaken belief that the state trial court had denied the 21 application for fees to cover the expenses of such an 22 investigation. This clearly satisfies the deficient Page 31 of 69 1 representation prong of Strickland. 2 To be more specific, the record is undisputed that 3 Rosario’s first counsel, Joyce Hartsfield, retained 4 investigator Jessie Franklin, and, after Franklin’s 5 unsuccessful attempt to contact several potential alibi 6 witnesses by telephone, concluded that an on-the-ground 7 investigation in Florida was necessary. Accordingly, 8 Hartsfield applied to the trial court for fees to cover the 9 cost of sending Franklin to Florida. The court ultimately 10 granted the application, but Hartsfield failed to disclose 11 this fact to Franklin. Franklin assumed the court had 12 denied the application because Hartsfield never informed her 13 otherwise and never ordered her to conduct the 14 investigation. Steven Kaiser, Rosario’s second counsel, 15 similarly labored under the erroneous impression that the 16 court had denied the application and neglected to make any 17 further inquiry into the matter. Whatever their reasons for 18 harboring this mistaken belief, an on-the-ground 19 investigation in Florida was never conducted. The direct 20 and proximate result of this mistake was that Rosario’s 21 defense team never contacted most of Rosario’s alibi 22 witnesses. Page 32 of 69 1 To be clear, neither Hartsfield nor Kaiser claim that 2 the failure to conduct this investigation was strategic; 3 they admit it was a mistake. Hartsfield testified that in 4 this case it was “critical” for the investigator to be able 5 to meet the witnesses “in person and have a face-to-face 6 conversation,” and that had Hartsfield realized that the 7 application for fees had been granted she would have asked 8 Franklin to go to Florida. Hartsfield unequivocally 9 confirmed that her failure to interview additional witnesses 10 was not strategic. Kaiser likewise testified that he relied 11 upon the erroneous belief that the fee motion had been 12 denied in limiting his investigation of Rosario’s alibi to 13 evidence that could be gathered from New York, and 14 repeatedly testified to the effect that he would have “loved 15 to” call additional alibi witnesses if only they had been 16 available to him. 17 Under these circumstances, there is simply no question 18 that this mistake on the part of Rosario’s defense attorneys 19 – and their resulting failure to investigate Rosario’s alibi 20 properly – was constitutionally deficient under the Sixth 21 Amendment. See, e.g., Williams v. Taylor, 529 U.S. 362, 396 22 (2000) (concluding that counsel’s failure to uncover and Page 33 of 69 1 present voluminous mitigating evidence at sentencing could 2 not be justified as a tactical decision to focus on the 3 defendant’s voluntary confessions because counsel had not 4 “fulfill[ed] their obligation to conduct a thorough 5 investigation of the defendant’s background”); Pavel v. 6 Hollins, 261 F.3d 210, 220 (2d Cir. 2001) (noting that “an 7 attorney’s failure to present available exculpatory evidence 8 is ordinarily deficient, unless some cogent tactical or 9 other consideration justified it”); Maddox v. Lord, 818 F.2d 10 1058, 1061-62 (2d Cir. 1987) (concluding that counsel would 11 be constitutionally deficient if he “was aware of – but 12 failed” for non-strategic reasons “to interview – a 13 potential witness . . . who was prepared to testify . . . 14 that he had diagnosed [petitioner] as being extremely 15 emotionally disturbed prior to, and during, the commission 16 of the crime”); Garcia v. Portuondo, 459 F. Supp. 2d 267, 17 287-88 (S.D.N.Y. 2006) (“[T]here is no reasonable trial 18 strategy that would have excluded at least conducting 19 interviews of the alibi witnesses to determine whether they 20 could provide exculpatory evidence.”). 4 4 Kaiser’s deficiency extended beyond his failure to investigate. In his limited investigation, Kaiser was able Page 34 of 69 1 B. Prejudice Prong 2 I also conclude that Rosario has satisfied the 3 prejudice prong of Strickland. Because of defense counsel’s 4 failure to properly investigate Rosario’s alibi defense, the 5 only two alibi witnesses presented at trial were John Torres 6 and Jenine Seda, who both testified that Rosario stayed with 7 them in Deltona, Florida from approximately the end of April 8 or beginning of May until about June 20, 1996. 9 Specifically, they testified that Rosario was in Florida on 10 June 19, 1996, the day of the murder, and that they 11 remembered this because it was the day before the birth of to contact a few individuals in Florida beyond the two witnesses he actually presented at trial. Specifically, he spoke with Fernando Torres – who, it will be seen, could have been an important witness – perhaps Fernando’s wife Margarita, and others whose names Kaiser could not recall. Kasier would have liked to call some of these witnesses; the limited recollection of the conversations he had, however, was that those he spoke with could not afford to come to New York and may have been reluctant to testify at least in part for financial reasons. Kaiser was unaware of a New York state statute providing reimbursement of certain expenses of out-of-state witnesses and, in any event, operating under the belief that the state court had denied the motion for fees to send Franklin to Florida, assumed that the court would have likewise declined to reimburse the witnesses any expenses. Thus, Kaiser’s decision not to pursue additional witnesses was also based on an erroneous belief rather than on any “plausible strategic calculus or an adequate pre- trial investigation” of the facts and law. Pavel, 261 F.3d at 222. Page 35 of 69 1 their son. John 5 further explained that on June 19, his car 2 broke down and he spent the day with Rosario looking for car 3 parts before they returned to his apartment together. Seda 4 also testified that Rosario was at her apartment to see the 5 baby on June 21, 1996, when she returned from the hospital. 6 Rosario took the stand on his own behalf and testified that 7 he was in Florida on the day of the murder and was staying 8 with John and Seda for most of June 1996. 9 The prosecution successfully discredited the alibi 10 defense presented at trial by convincing the jury that John, 11 Seda and Rosario were lying. The first words from the 12 prosecution during its summation were: “You [the jury] have 13 to determine which witnesses were credible, which witnesses 14 were believable, which witnesses had an interest in the 15 outcome of this case.” The prosecution went on to argue, 16 Rosario’s “saying he was in Florida. Look at the testimony 17 to determine, can you rely on it? Is it believable? Is it 18 credible?” Discrediting Seda and John, the prosecution 19 argued: 5 Because a number of relevant witnesses share the surnames “Torres” and “Ruiz,” I shall refer to those individuals by their first names. Page 36 of 69 1 First the two witnesses we heard. 2 Jenine Seda and John Torres, the 3 defendant’s friends. I would suggest to 4 you, ladies and gentlemen, that those 5 witnesses are interested witnesses, 6 interested because they have an interest 7 in the outcome of the case. They don’t 8 want to see their friend go to jail. They 9 don’t want to see their friend in 10 trouble. 11 12 With respect to Rosario’s testimony, the prosecution noted 13 at the outset that “the Judge will instruct you he is an 14 interested witness.” The proseuction also emphasized that 15 Rosario lied about staying with a woman in Florida during 16 March and April of 1996 named Shannon Beane, whom he claimed 17 to have been with every day, when in fact he had been 18 incarcerated between March 13 and April 12. The prosecution 19 argued: 20 Ladies and gentlemen, he took the 21 stand. He put his hand on the Bible. He 22 swore to tell the truth, and he told you 23 I was with Shannon [B]ean[e]. I was with 24 her daily, every day, and we know, ladies 25 and gentlemen, from Captain Bolton that’s 26 not true. 27 28 Ladies and gentlemen, I would 29 suggest to you he doesn’t want you to 30 know the truth about June 19th because to 31 know the truth is to know that he was on 32 White Plains Road, to know that he was on 33 Turnbull Avenue, and to know that he was 34 pumping a bullet [into] the head of 35 George Collazo, and ending his life. Page 37 of 69 1 2 Ask yourself to what extent he would 3 go to preventing you from knowing the 4 truth. If he didn’t want you to know 5 where he was in March and April of 1996, 6 a time period which is insignificant 7 since it has nothing to do with the 8 commission of this crime, what would he 9 do for the time period that really 10 matters? 11 12 . . . . 13 14 Ladies and gentlemen, use your 15 common sense. Keep in mind that 16 [Rosario] has an interest in this case. 17 18 The prosecution thus presented the jury with a choice: 19 it could choose to believe two, disinterested eyewitnesses, 20 or it could believe Rosario and his two good friends. It 21 was a credibility battle. It is not shocking, therefore, 22 that the prosecution secured a conviction. I conclude, 23 however, that if the jury had been presented with the 24 additional alibi evidence unearthed only by Rosario’s post- 25 conviction team, there is a reasonable probability that the 26 outcome at trial would have been different. 27 Rather than the slim alibi defense actually presented 28 at trial, the jury would have been presented with a much 29 stronger and more credible account of Rosario’s presence in 30 Florida on the day of the murder and in the immediately Page 38 of 69 1 surrounding period. Instead of disbelieving two alibi 2 witnesses who were good friends with Rosario and Rosario 3 himself, the jury would have had to discredit at least seven 4 additional witnesses, who would have corroborated Rosario’s 5 alibi, provided further context to his defense and testified 6 to additional facts that had not been elicited at trial. 7 Moreover, many of the additional witnesses are less 8 interested in the outcome of the trial than were the trial 9 witnesses and thus would have been less vulnerable to 10 impeachment as interested witnesses. 11 The following alibi evidence was presented in 12 connection with Rosario’s post-conviction motion. First, 13 Chenoa Ruiz, neighbor of John and Seda and wife of John’s 14 brother Robert Torres, testified that she saw Rosario about 15 five times a week when he was living with John and Seda in 16 June 1996, but that he moved out of their house and in with 17 a friend named Ray who lived nearby when the baby was born. 18 Chenoa testified that on the night of June 18, 1996 (the 19 night prior to the murder), John, Seda, Robert and Rosario 20 were at her and Robert’s apartment when Seda began to have 21 contractions. Chenoa and another woman took Seda to the 22 hospital without John, who chose instead to remain with Page 39 of 69 1 Robert and Rosario. Seda was not kept in the hospital that 2 night, but was told to return the next day for a scheduled 3 appointment. The next morning (June 19, the day of the 4 murder), Chenoa saw Rosario when she arrived at John and 5 Seda’s apartment to take Seda to the doctor and saw him 6 again when she and Seda returned home several hours later. 7 Chenoa testified that she remembers this day in particular 8 because she was annoyed that John was “hanging out” with 9 Rosario instead of tending to his pregnant girlfriend. 10 Chenoa would have provided less interested testimony than 11 John and Seda because she did not consider Rosario a friend. 12 Second, Fernando Torres, John’s father, testified that 13 Rosario lived with John and Seda around the time that his 14 grandson was born on June 20, 1996. Fernando testified that 15 he was with Rosario in Florida on June 19, 1996, the day of 16 the murder, because John’s car broke down and Fernando 17 accompanied John and Rosario to purchase car parts. 18 Fernando also saw Rosario in Florida the following morning: 19 early on June 20, Fernando went to his son’s house and 20 learned from Rosario that John and Seda were at the 21 hospital. Finally, Rosario was again present at John and 22 Seda’s apartment when Fernando met his grandson for the Page 40 of 69 1 first time on June 21, the day Seda returned from the 2 hospital. Fernando invited Rosario to church that day. In 3 addition to providing additional facts not supplied by 4 either John or Seda, a jury may have found Fernando more 5 credible because he was not a friend of Rosario and thus 6 undoubtedly a less interested witness. Additionally, 7 Fernando is a generation older than Rosario, John and Seda, 8 who were all in their twenties, which may have further 9 bolstered his credibility over the trial witnesses. See, 10 e.g., United States v. Liporace, 133 F.3d 541, 545 (7th Cir. 11 1998) (approving instruction to jury that it may consider a 12 witness’s age in assessing that witness’s credibility); cf. 13 Washington v. Schriver, 255 F.3d 45, 59-60 & n.10 (2d Cir. 14 2001) (implicitly approving same). 6 15 A third witness – Michael Serrano, a corrections 6 Margarita Torres, Fernando’s wife and John’s mother, filed an affidavit in connection with the post-conviction hearing stating that she saw Rosario in Florida on June 19, 1996, the day of the murder, and again when Seda came home from the hospital with her grandson. Along with Fernando, Margarita invited Rosario to church that day. Though she did not testify at the post-conviction hearing, she indicated that she would be “more than willing” to testify on Rosario’s behalf. As with Fernando, had Margarita testified, a jury may have viewed her testimony as more credible than either John’s or Seda’s because she was not a friend of Rosario and is a generation older than they. Page 41 of 69 1 officer – testified that in June of 1996, he saw Rosario two 2 or three times a week in the apartment complex where John, 3 Seda, Robert and Chenoa lived, including in the days prior 4 to the birth of John and Seda’s child. Though he did not 5 know Rosario’s whereabouts on the day of the murder, Serrano 6 testified that on the night that the baby was born (i.e., 7 the day after the murder), Serrano and several other people, 8 including Rosario, held an impromptu celebration in the 9 parking lot of the apartment complex to congratulate John 10 when he came home from the hospital. 7 As with Ruiz, Serrano 11 did not consider himself to be good friends with Rosario. 12 Fourth, Denise Hernandez, Rosario’s ex-girlfriend, 13 testified that she saw Rosario in Florida around the time of 14 the murder because they were dating throughout June 1996, 15 and recalled in particular a big argument at some point in 16 the middle to end of that month. Hernandez explained that 17 one day, she and her friend were at her friend’s house 18 getting ready to go out to a movie when Rosario took her 7 According to Serrano, John alone came home briefly to get a change of clothes before returning to the hospital that night. Accordingly, his testimony does not contradict Fernando’s account that Seda and the baby remained in the hospital until the following day. Page 42 of 69 1 car, without her permission, on a “joyride.” Hernandez was 2 particularly upset because this incident occurred a few days 3 before her sister’s birthday, which is on June 26, and her 4 sister’s birthday present was in the car. As a result of 5 this and other issues in their relationship, Hernandez broke 6 up with Rosario at some point between her sister’s birthday 7 and when Rosario returned to New York. It is true that 8 Hernandez has maintained a close relationship with Rosario, 9 even visiting him in prison on several occasions, and thus 10 the prosecution presumably would have attacked her as an 11 interested witness. Nevertheless, her testimony would have 12 provided additional and distinct facts relating to Rosario’s 13 whereabouts around the date of the murder and would have 14 provided further context to his alibi defense. 15 Furthermore, a fifth witness, Hernandez’s friend 16 Lyssette Rivera, testified that she was present when Rosario 17 took Hernandez’s car on the joyride and recalled the ensuing 18 argument between Hernandez and Rosario and its proximity to 19 Hernandez’s sister’s birthday (recalling that the argument 20 occurred between five days and a week prior to the sister’s 21 birthday). Thus, to the extent that Hernandez would have 22 been subject to impeachment in light of her relationship Page 43 of 69 1 with Rosario, defense counsel could have corroborated her 2 testimony with that of Rivera, who – though she also had 3 communicated with Rosario since his incarceration – did not 4 have as close a relationship with him. 5 Sixth, Ricardo Ruiz, the brother of Chenoa, testified 6 that he saw Rosario at John and Seda’s apartment during the 7 month of June 1996 “[a]ll the time,” including before and 8 after their baby was born. In particular, he testified that 9 Rosario was in Florida “[a]t the time that [Seda] gave birth 10 to [the baby].” He also testified that after Rosario moved 11 out of John and Seda’s apartment, Rosario moved in with a 12 friend named Ray, who lived across the street from John and 13 Seda. 14 The seventh witness – Minerva Godoy, Rosario’s ex- 15 fiancee – testified that Rosario left New York for Florida 16 in May 1996, to relocate and find a job, and she did not see 17 him again until the morning of July 1, 1996, when he claims 18 to have returned to New York. Godoy testified that she was 19 in regular contact with Rosario while he was in Florida, 20 calling him at Fernando’s Florida telephone number and once 21 wiring him money in Florida via Western Union. In 22 particular, she testified that Rosario called her from Page 44 of 69 1 Florida the day after Seda gave birth and told her that he 2 was going to go see the baby. 8 3 Because the prosecution’s case hinged so much on 4 discrediting Rosario’s alibi defense, these additional 5 witnesses could have made all the difference in the world. 6 Godoy could have provided the necessary context by 7 testifying about Rosario’s departure from New York to 8 Florida in May 1996, essentially serving as the first 9 chapter of his alibi defense, and then about their meeting 10 on July 1, 1996, providing the final chapter immediately 11 prior to his surrender to the police. All of the other 12 witnesses discussed above would have filled in the middle by 13 testifying that they saw Rosario in their Florida community 14 throughout June of 1996. They would have provided specific 15 facts regarding where he lived and what he was doing at that 16 time. Several witnesses could have corroborated each 17 other’s testimony that Rosario was in Florida on the exact 18 day of the murder and in the immediately surrounding days. 19 Chenoa would have testified that she saw Rosario both the 8 Another potential alibi witness – Jeremy David Guzman – filed a written statement in connection with the post- conviction hearing stating that he had spent “hours” with Rosario in Florida on June 19, 1996. Page 45 of 69 1 night prior to the murder, when she took Seda to the 2 hospital, and twice throughout the day of the murder, both 3 before and after Seda’s doctor’s appointment. John’s father 4 Fernando would have placed Rosario in Florida on three 5 consecutive days beginning with the day of the murder and 6 would have corroborated John’s testimony that Rosario was 7 with him looking for car parts on the nineteenth. From 8 Chenoa and Fernando alone, the jury would have been provided 9 additional concrete facts that Rosario was in Florida the 10 night prior to, at various points the day of, and the 11 morning following the murder – indisputably critical data 12 points in establishing that Rosario was in Florida, and not 13 over 1000 miles away in New York, when the victim was 14 murdered. 15 Additionally, Serrano would have testified that he was 16 with Rosario in the parking lot of John and Seda’s apartment 17 complex on the night after the murder; Hernandez and Rivera 18 would have provided consistent testimony about the fight 19 between Hernandez and Rosario around the date of the murder; 20 and Ricardo could have further corroborated Rosario’s 21 general presence in Florida throughout June. 22 This additional evidence that the jury never heard Page 46 of 69 1 would have provided the necessary context and corroboration 2 for Rosario’s alibi defense. Moreover, as discussed, many 3 of these witnesses were not vulnerable to impeachment as 4 interested witnesses because they were not close friends 5 with Rosario. 9 6 I conclude that this evidence, taken together, clearly 7 establishes a reasonable probability that the outcome of the 8 trial would have been different had defense counsel 9 investigated and presented this additional alibi evidence, 10 satisfying Strickland’s prejudice prong. “Overall,” as 11 Rosario argues, “if presented with the additional evidence 12 at trial, a jury must disregard nine witnesses, as opposed 13 to two, as mistaken or lying about seeing Rosario in Florida 14 on and about June 19, 1996, before convicting him of the 15 Bronx murder.” Brief for Rosario at 34. See Stewart v. 16 Wolfenbarger, 468 F.3d 338, 359 (6th Cir. 2006) (finding 17 prejudice when defense counsel failed to call two additional 18 alibi witnesses to corroborate the one alibi witness called 19 at trial who was impeached because of his close association 9 Nor would have impeachment for criminal history been an issue. Notably, Ricardo was the only witness at Rosario’s post-conviction hearing with any criminal record, consisting solely of misdemeanor convictions. Page 47 of 69 1 with the defendant); Washington v. Smith, 219 F.3d 620, 634 2 (7th Cir. 2000) (finding prejudice when defense counsel 3 failed to call three additional alibi witnesses to 4 corroborate the one alibi witness at trial who had knowledge 5 of the defendant’s whereabouts during the robbery, 6 particularly when none of the additional witnesses, unlike 7 the trial witness, had a criminal record); Montgomery v. 8 Peterson, 846 F.2d 407, 415 (7th Cir. 1988) (finding 9 prejudice in failure to call additional, disinterested alibi 10 witness, noting that “the jury might well have viewed the 11 otherwise impeachable testimony of the twelve witnesses who 12 were presented at the . . . trial in a different light had 13 the jury also heard the testimony of this disinterested 14 witness”). 15 Further highlighting the prejudicial effect of defense 16 counsel’s error in this case is the paucity of the 17 prosecution’s case, which consisted of only two stranger 18 eyewitnesses. We have consistently acknowledged that this 19 sort of evidence is “proverbially untrustworthy.” Kampshoff 20 v. Smith, 698 F.2d 581, 585 (2d Cir. 1983); see also Gersten 21 v. Senkowski, 426 F.3d 588, 613 (2d Cir. 2005) 22 (characterizing direct evidence consisting only of Page 48 of 69 1 eyewitness testimony as “underwhelming”), cert. denied sub 2 nom., Artus v. Gersten, 547 U.S. 1191 (2006); Lyons v. 3 Johnson, 99 F.3d 499, 504 (2d Cir. 1996) (“[T]his court has 4 noted on more than one occasion that eyewitness testimony is 5 often highly inaccurate.”). Indeed, each year thousands of 6 defendants in the United States are convicted for crimes 7 that they did not commit, and many experts estimate that 8 eyewitness error plays a role in half or more of all 9 wrongful felony convictions. Richard A. Wise, Clifford S. 10 Fishman & Martin A. Safer, How to Analyze the Accuracy of 11 Eyewitness Testimony in a Criminal Case, 42 C ONN. L. R EV. 12 435, 440 & n.12 (2009) (citing study showing that eyewitness 13 error accounts for nearly sixty percent of all wrongful 14 convictions). 15 In this case, there are reasons to be concerned with 16 the two eyewitnesses’ accounts: the porter, Robert Davis, 17 saw the shooter at a distance of more than two car lengths 18 for only a few seconds, and although Michael Sanchez 19 testified that he got a good look at the shooter, it was 20 only for a short moment under very stressful conditions. 10 10 A third eyewitness, Jose Diaz, believed that he might be able to identify the shooter, but failed to identify Page 49 of 69 1 This is of course not to say there was insufficient evidence 2 to convict Rosario. But Strickland makes clear that “a 3 verdict or conclusion only weakly supported by the record is 4 more likely to have been affected by errors than one with 5 overwhelming record support.” 466 U.S. at 696. Such is the 6 case here. See Lindstadt v. Keane, 239 F.3d 191, 204-05 (2d 7 Cir. 2001) (finding prejudice and reversing denial of writ 8 of habeas corpus where trial counsel failed to investigate 9 evidence that could have corroborated the petitioner’s alibi 10 claims, and where the prosecution’s case rested on only two 11 eyewitnesses and limited corroborating evidence); see also 12 Espinal v. Bennett, 588 F. Supp. 2d 388, 402, 407-08 13 (E.D.N.Y. 2008) (granting habeas relief when defense counsel 14 failed to investigate a statement provided by a potential 15 alibi witness who might have corroborated the petitioner’s 16 own testimony regarding his whereabouts on the day of the 17 murder in a prosecution consisting primarily of two 18 eyewitnesses, one of whose credibility was impeached), 19 aff’d, 342 F. App’x 711 (2d Cir. Aug. 18, 2009) (unpublished 20 disposition). Rosario in court. Page 50 of 69 1 II. Habeas Corpus Standards 2 The majority essentially concedes a Strickland 3 violation and that Rosario would be entitled to relief if 4 this case arose on direct review but denies the writ out of 5 deference to the state court. Pursuant to 28 U.S.C. § 2254, 6 a federal court may not grant a writ of habeas corpus to a 7 state prisoner “with respect to any claim that was 8 adjudicated on the merits” by the state court unless the 9 state court’s decision “was contrary to, or involved an 10 unreasonable application of, clearly established Federal 11 law, as determined by the Supreme Court of the United 12 States.” 28 U.S.C. § 2254(d)(1). Under this principle of 13 deference, habeas relief may not be granted merely upon a 14 “conclusion that counsel’s performance was constitutionally 15 inadequate.” Carrion v. Smith, 549 F.3d 583, 591 n.4 (2d 16 Cir. 2008). Rather, “petitioner must identify some 17 increment of incorrectness beyond error in order to obtain 18 habeas relief.” Jones v. West, 555 F.3d 90, 96 (2d Cir. 19 2009) (quoting Sorto v. Herbert, 497 F.3d 163, 169 (2d Cir. 20 2007)). Moreover, as the majority notes, “because the 21 Strickland standard is a general standard, a state court has 22 even more latitude to reasonably determine that a defendant Page 51 of 69 1 has not satisfied that standard.” Knowles v. Mirzayance, 2 556 U.S. __, 129 S. Ct. 1411, 1420 (2009). Nevertheless, 3 “the increment of incorrectness beyond error need not be 4 great; otherwise, habeas relief would be limited to state 5 court decisions so far off the mark as to suggest judicial 6 incompetence.” Georgison v. Donelli, 588 F.3d 145, 154 (2d 7 Cir. 2009) (internal brackets omitted) (quoting Hoi Man Yung 8 v. Walker, 468 F.3d 169, 176 (2d Cir. 2006)). 9 A close review of the state court’s decision makes it 10 entirely clear, however, that – even affording the state 11 court its due deference – its decision rejecting Rosario’s 12 claim was an unreasonable application of Strickland and 13 should not stand. 14 At the outset, I note that the state court’s use of the 15 “meaningful representation” standard led it to focus on 16 certain factors that have little bearing on a proper 17 Strickland analysis. And it appears to have done so at the 18 expense of determining whether the undisputed mistakes made 19 by Rosario’s defense counsel fell below objectively 20 reasonable standards and, moreover, whether they caused him 21 prejudice, as required under Strickland. Indeed, the state 22 court relied heavily upon its finding that Rosario’s pre- Page 52 of 69 1 trial and trial attorneys “represented [him] in a thoroughly 2 professional, competent, and dedicated fashion.” It 3 emphasized that “[b]oth attorneys filed all appropriate 4 motions; within the scope of the information that was then 5 available to them, an investigation was conducted; witnesses 6 were examined and cross-examined adeptly, professionally and 7 with clarity; Mr. Kasier’s opening and closing statements 8 were concise and to the point; and, most importantly, a 9 credible alibi defense was presented to the jury.” The 10 state court went on to emphasize that counsel’s mistake as 11 to the denial of the application for investigative fees “was 12 not deliberate” and “does not alter the fact that both 13 attorneys represented defendant skillfully, and with 14 integrity and in accordance with the standards of 15 ‘meaningful representation’ defined by [the New York state] 16 appellate courts.” It wrote: 17 Defendant has tried to second-guess 18 his trial counsel at almost every level 19 of their representation. He has 20 questioned the depth of their 21 investigation, the scope and focus of 22 cross-examination and argued that his 23 alibi defense could have been better if 24 they had only followed through on [the 25 state trial court’s fee] order. His 26 criticisms ignore the fact that Ms. 27 Hartsfield and Mr. Kaiser ably, and Page 53 of 69 1 professionally represented him at every 2 stage of the case with integrity and in 3 ways that were consistent with the 4 standards of ‘meaningful representation’ 5 described above. 6 7 . . . And Mr. Kaiser at trial was 8 prepared, skillful, purposeful, 9 thoughtful and creative. 10 11 This type of analysis is entirely at odds with 12 Strickland and is not dispositive of whether Rosario’s 13 defense counsel were ineffective under the Sixth Amendment. 14 It is axiomatic that, even if defense counsel had performed 15 superbly throughout the bulk of the proceedings, they would 16 still be found ineffective under the Sixth Amendment if 17 deficient in a material way, albeit only for a moment and 18 not deliberately, and that deficiency prejudiced the 19 defendant. See, e.g., Henry v. Poole, 409 F.3d 48, 72 (2d 20 Cir. 2005) (“[R]eliance on counsel’s competency in all other 21 respects, . . . fail[s] to apply the Strickland standard at 22 all.” (internal citation and quotation marks omitted)), 23 cert. denied, 547 U.S. 1040 (2006); cf. Kimmelman v. 24 Morrison, 477 U.S. 365, 386 (1986) (noting that while “[i]t 25 will generally be appropriate . . . to assess counsel’s 26 overall performance throughout the case in order to 27 determine whether the identified acts or omissions overcome Page 54 of 69 1 the presumption that a counsel rendered reasonable 2 professional assistance,” a “failure to make reasonable 3 investigations or to make a reasonable decision that makes 4 particular investigations unnecessary,” may be 5 constitutionally deficient irrespective of trial performance 6 (internal quotation marks omitted)). 7 It is far from clear whether the state court realized 8 this basic principle. In fact, the state court noted in a 9 footnote that New York case law, in particular People v. 10 Benevento, 91 N.Y.2d 708 (1998), “expressly rejected” 11 Strickland’s requirement “that, but for the attorneys[’] 12 errors, the result of the proceeding would have been 13 different.” This footnote, viewed in context with the 14 entirety of the court’s decision, begs the question whether 15 the state court understood that New York state’s 16 “ineffective assistance cases have departed from the second 17 (‘but for’) prong of Strickland,” only to “adopt[] a rule 18 somewhat more favorable to defendants.” People v. Turner, 5 19 N.Y.3d 476, 480 (2005) (emphasis added) (citing People v. 20 Caban, 5 N.Y.3d 143, 155-56 (2005); People v. Stultz, 2 21 N.Y.3d 277, 284 (2004); Benevento, 91 N.Y.2d at 713-14). 22 That is, it is unclear whether the state court appreciated Page 55 of 69 1 that even if prejudice in the Strickland sense is not shown, 2 a defense attorney can be found ineffective under the New 3 York State Constitution if his performance was so below par 4 that he did not provide “meaningful representation” to his 5 client. See Caban, 5 N.Y.3d at 156 (“[U]nder our State 6 Constitution, even in the absence of a reasonable 7 probability of a different outcome, inadequacy of counsel 8 will still warrant reversal whenever a defendant is deprived 9 of a fair trial. . . . [O]ur state standard thus offers 10 greater protection than the federal test . . . .”). 11 On a different note, at one point in the decision the 12 state court sharply detoured into an analysis regarding 13 newly discovered evidence. It wrote: 14 In order to prevail on a motion for 15 a new trial based on a claim of newly 16 discovered evidence, a defendant must 17 establish by a preponderance of the 18 evidence that evidence has been 19 discovered since the trial which could 20 not, with due diligence, have been 21 produced at trial, and which is of such a 22 character that, had it been presented at 23 trial, there is a probability that the 24 verdict would have been more favorable 25 for him . . . . 26 27 . . . . 28 29 . . . the existence of these 30 witnesses was not new evidence discovered Page 56 of 69 1 since the trial. They were known to 2 defendant, who immediately gave their 3 names to the police after his arrest, to 4 his attorneys at their first and 5 subsequent meetings, and to Jesse 6 Franklin. Efforts were made to speak, 7 and interview them and the substance of 8 their testimony was known to the parties 9 before the trial began. 10 11 It is unclear when, if ever, the court returned to the 12 ineffective assistance of counsel analysis, and, more 13 importantly, to what extent this detour infected that 14 analysis. If this newly discovered evidence analysis did in 15 fact bleed over to the ineffective assistance of counsel 16 analysis, the harmful effect is patent, considering the 17 obvious tension between a newly discovered evidence claim 18 and an ineffectiveness claim based on an attorney’s failure 19 to investigate an alibi that was disclosed to him by his 20 client prior to trial. 21 It is true that a New York state court’s application of 22 the meaningful representation standard does not necessarily 23 result in error affording a petitioner habeas relief because 24 the standard, properly construed, is more favorable to 25 defendants. See Henry v. Poole, 409 F.3d at 68-71. It is 26 also true that we do not grant habeas relief when a state 27 court is merely inartful or unclear in its reasoning. But, Page 57 of 69 1 in this case, it is entirely unclear to what extent the 2 state court abandoned the Strickland analysis for a rule 3 less favorable to defendants. Such an error would clearly 4 be “contrary to” Strickland. 28 U.S.C. § 2254(d)(1). 5 The majority aptly pinpoints the “danger” of New York’s 6 “meaningful representation” standard: though generally more 7 protective of defendants’ rights than Strickland, it risks 8 leading a court that “misunderstand[s] the New York 9 standard” to “look past a prejudicial error as long as 10 counsel conducted himself in a way that bespoke of general 11 competency throughout the trial.” Ante at 18-19. The state 12 court’s opinion provides strong indications that this is 13 precisely what happened here. Yet the majority fails to 14 address the very real likelihood that the state court fell 15 victim to the danger it identified, merely concluding that, 16 in general, when properly applied, the New York standard is 17 not contrary to Strickland. Id. at 19. 18 Nevertheless, I “need not make a determination under 19 the ‘contrary to’ clause, for [I] conclude that the . . . 20 Court’s rejection of [Rosario’s] 21 ineffective-assistance-of-counsel claim was at least an 22 objectively unreasonable application of Strickland.” Henry, Page 58 of 69 1 409 F.3d at 71. It is clear from the record that the state 2 court not only unreasonably focused on counsel’s overall 3 performance and minimized their mistakes, but also 4 unreasonably discounted the alibi evidence adduced at the 5 post-conviction hearing and thus undervalued its prejudicial 6 effect. 7 In terms of Strickland’s performance prong, the state 8 court recognized that counsel’s failure to complete their 9 investigation was neither strategic nor the result of any 10 sound trial strategy, but rather a “mistake.” The state 11 court – as well as the majority – appears to excuse this 12 mistake because it was “not deliberate,” counsel’s 13 performance was otherwise “skillful[],” and counsel 14 conducted some investigation leading to the presentation of 15 a putatively “credible” alibi defense. But none of this 16 excuses the fact that counsel essentially turned a blind eye 17 to the existence of substantial potentially excuplatory 18 evidence of which it was aware and, moreover, did so not on 19 the basis of any “reasonable professional judgment,” 20 Strickland, 466 U.S. at 690, but rather as a result of pure 21 inadvertence. Such conduct clearly falls below the 22 threshold of minimal competence and, to the extent the state Page 59 of 69 1 court found otherwise, I conclude that was an unreasonable 2 application of Strickland. 3 With respect to prejudice, in relevant part, the state 4 court reasoned: 5 [A]n alibi defense was presented 6 through the two witnesses who had the 7 best reason for remembering why defendant 8 was present in Florida on June 19[,] 1996 9 – the birth of their son – an event that 10 was more relevant for them than the 11 events relied upon by the other witnesses 12 . . . . Moreover, the alibi evidence 13 offered by defendant at the hearing was 14 in some cases questionable and in others 15 raised issues which could have created 16 questions for a deliberating jury. For 17 example, two of the witnesses – Lisette 18 Rivero [sic], and Denise Hernandez – 19 could not say where defendant was on June 20 19 and 20. And Fernando Torres, when 21 questioned about the purchase of auto 22 parts years later, changed the date to 23 three or four days before his grandson 24 was born . . . . 25 26 . . . It may not be cumulative to 27 evidence presented at the trial – which 28 largely was the case herein – and it must 29 not be merely impeaching evidence . . . . 30 31 32 For instance, Chenoa Ruiz recalled 33 defendant’s presence in the Torres’ 34 apartment on June 18 and 19, the two days 35 prior to the birth of their child. And, 36 Fernando Torres testified that he was 37 with defendant and his son the day before 38 his daughter-in-law gave birth. That 39 testimony was cumulative to his son Page 60 of 69 1 John’s trial testimony. 2 3 . . . . 4 5 An investigator was not sent to 6 Florida to interview witnesses. 7 Nevertheless, the fact remains that the 8 People’s case was strong, which was 9 acknowledged by the Appellate Division 10 when it affirmed the conviction herein. 11 The prospective witnesses now before the 12 court, studied closely, were, for the 13 most part, questionable and certainly not 14 as persuasive as the two witnesses who 15 did testify, and were rejected by the 16 jury. 17 18 First, the state court’s finding that “a credible alibi 19 defense was presented to the jury” is hardly relevant to 20 whether there is a reasonable probability of a different 21 result had defense counsel presented a substantially more 22 credible alibi defense. Second, the state court’s 23 recognition that “an alibi defense was presented through the 24 two witnesses who had the best reason for remembering why 25 defendant was present in Florida on June 19, 1996 – the 26 birth of their son – an event that was more relevant for 27 them than the events relied upon by the other witnesses” 28 also misses the point. It overlooks the fact that John and 29 Seda were subject to impeachment as interested witnesses, 30 and at least seven additional witnesses were available, a Page 61 of 69 1 number of whom were less interested in the outcome of the 2 trial, to corroborate their testimony, as well as add 3 additional facts. 4 Third, although the court did find that “the alibi 5 evidence offered by defendant at the hearing was in some 6 cases questionable and in others raised issues which could 7 have created questions for a deliberating jury,” it provided 8 just three examples from a voluminous record in support of 9 this finding, none of which bear scrutiny. It noted that 10 “two of the witnesses – Lisette Rivero [sic], and Denise 11 Hernandez – could not say where defendant was on June 19 and 12 20.” But, as discussed, these witnesses testified to 13 additional, non-cumulative facts that placed Rosario in 14 Florida around the day of the murder. See ante at 46-48. 15 The relevancy of this evidence is indisputable. The court 16 also noted that “Fernando Torres, when questioned about the 17 purchase of auto parts years later, changed the date to 18 three or four days before his grandson was born.” This is 19 simply not supported by the record. In fact, when asked 20 whether he told Rosario’s post-conviction counsel that he 21 went looking for car parts with his son and Rosario three or 22 four days before his grandson was born, Fernando responded, Page 62 of 69 1 “No, I don’t recall that at all,” and maintained that the 2 excursion occurred on June 19. 3 Fourth, the state court found that the additional alibi 4 witnesses were “largely . . . cumulative.” To the extent 5 that the additional alibi evidence corroborated John’s and 6 Seda’s testimony, it is only reasonable to conclude that 7 this militates in favor of a showing of prejudice. Again, 8 John’s and Seda’s credibility was attacked by the 9 prosecution. Corroboration was thus desperately needed. 10 See, e.g., Washington v. Smith, 219 F.3d 620, 634 (7th Cir. 11 2000) (“Evidence is cumulative when it ‘supports a fact 12 established by existing evidence,’ Black’s Law Dictionary 13 577 (7th ed. 1999), but Washington’s whereabouts on the day 14 of the robbery was far from established – it was the issue 15 in the case. The fact that Pickens had already testified to 16 facts consistent with Washington’s alibi did not render 17 additional testimony cumulative.”). 18 Finally, the state court characterized the People’s 19 case as “strong.” But, the fact remains that it was based 20 solely on the eyewitness accounts of two strangers – the 21 type of evidence that this Court has repeatedly 22 characterized as weak. Page 63 of 69 1 At bottom, the problem with the state court’s decision 2 is its application of the reasonable probability standard. 3 Contrary to the state court’s apparent belief, this standard 4 does not require that the reviewing court be convinced of 5 Rosario’s alibi defense. “[T]he reasonable-probability 6 standard is not the same as, and should not be confused 7 with, a requirement that a defendant prove by a 8 preponderance of the evidence that but for error things 9 would have been different.” Wilson v. Mazzuca, 570 F.3d 10 490, 507 (2d Cir. 2009) (quoting United States v. Dominguez 11 Benitez, 542 U.S. 74, 83 n.9 (2004) (citing Kyles v. 12 Whitley, 514 U.S. 419, 434 (1995))). “A reviewing court 13 looks instead to whether the probability of a different 14 result is sufficient to undermine confidence in the outcome 15 of the proceeding.” Id. (internal quotation marks omitted) 16 (quoting Dominguez Benitez, 542 U.S. at 83 (quoting 17 Strickland, 466 U.S. at 694)); see also Porter v. McCollum, 18 558 U.S. __, 130 S. Ct. 447, 455-56 (2009) (per curiam) (“We 19 do not require a defendant to show ‘that counsel’s deficient 20 conduct more likely than not altered the outcome’ of his 21 penalty proceeding, but rather that he establish ‘a 22 probability sufficient to undermine confidence in [that] Page 64 of 69 1 outcome.’” (alteration in original) (quoting Strickland, 466 2 U.S. at 693-94)). 3 Under the present circumstances, it is unreasonable to 4 conclude that the probability of a different result is not 5 sufficiently likely so as to undermine the confidence in the 6 verdict. Defense counsel failed to investigate Rosario’s 7 alibi defense and, as a result, did not call at least seven 8 additional alibi witnesses. Instead, they proceeded with 9 only two witnesses, both of whom were impeached as 10 interested. In a credibility battle, such as this case, 11 there is, to some extent, power in numbers – that is, if 12 presented with the additional evidence at trial, the jury 13 would have had to disregard a total of at least nine defense 14 witnesses claiming to have seen Rosario in Florida on and 15 around the day of the murder, as opposed to just two 16 interested witnesses. As discussed, the additional alibi 17 witnesses would have provided further context and 18 corroboration of Rosario’s alibi defense, would have 19 testified to non-cumulative facts, and a number of them 20 would have been less subject to impeachment than John and 21 Seda. 22 The prosecution’s principal argument is that the Page 65 of 69 1 additional alibi witnesses are not as reliable or credible 2 as John and Seda. It emphasizes that Fernando, Chenoa, 3 Rivera and Godoy provided less detailed accounts of their 4 recollection during interviews prior to the 440.10 hearing 5 than they did on the stand during the actual hearing. We 6 have noted, however, that such “silence is so ambiguous that 7 it is of little probative force.” Victory v. Bombard, 570 8 F.2d 66, 70 (2d Cir. 1978) (quoting United States v. Hale, 9 422 U.S. 171, 176 (1975)). The prosecution also emphasizes 10 that Chenoa did not recollect certain facts, such as when 11 Rosario traveled back and forth between Florida and New York 12 during his previous trips and the precise date he left 13 Florida at the end of June 1996. The fact that witnesses do 14 not remember all relevant details is hardly surprising and 15 certainly not dispositive as to whether they are reliable 16 witnesses to the ultimate fact at issue, such as Rosario’s 17 whereabouts on or about June 19, 1996 – particularly where, 18 as here, there is a significant independent event to anchor 19 memories surrounding the relevant date. The prosecution 20 also argues that any harm created by defense counsel’s 21 failure to call additional alibi witnesses is overwhelmed by 22 the harm that Rosario caused himself by what it Page 66 of 69 1 characterizes as lying on the stand when he did not disclose 2 that he was incarcerated for part of March and April of 3 1996. This argument seems to cut the other way, however. 4 That is, to the extent that the jury believed that Rosario 5 was being deliberately deceptive, additional alibi witnesses 6 were all the more necessary. 7 At bottom, the prosecution’s brief takes each witness’s 8 testimony in isolation, picks it apart, and makes an 9 assessment as to whether there is a reasonable probability 10 that the inclusion of that particular witness’s testimony 11 would have affected the outcome of the trial. We cannot 12 engage in such a piecemeal analysis. Rather, we must 13 analyze the cumulative effect of counsel’s failure to call 14 any of the additional alibi witnesses. See Lindstadt v. 15 Keane, 239 F.3d 191, 199 (2d Cir. 2001) (“Strickland directs 16 us to look at the ‘totality of the evidence before the judge 17 or jury’ . . . . We therefore consider these errors in the 18 aggregate.” (quoting Strickland, 466 U.S. at 695-96)). This 19 principle, which the majority’s analysis seems to overlook, 20 is essential to the proper application of Strickland, as we 21 were yet again reminded by the Supreme Court in Porter v. 22 McCollum, 558 U.S. __, 130 S. Ct. 447, 453-54 (2009) (per Page 67 of 69 1 curiam). 2 I find defense counsel’s performance and the resulting 3 prejudice in this case very troubling. “[T]here is nothing 4 as dangerous as a poorly investigated alibi. An attorney 5 who is not thoroughly prepared does a disservice to his 6 client and runs the risk of having his client convicted even 7 where the prosecution’s case is weak. A poorly prepared 8 alibi is worse than no alibi at all.” 2 G. Schultz, Proving 9 Criminal Defenses ¶ 6.08 (1991), quoted in Henry v. Poole, 10 409 F.3d 48, 65 (2d Cir. 2005), cert. denied, 547 U.S. 1040 11 (2006); cf. United States v. Parness, 503 F.2d 430, 438 (2d 12 Cir. 1974) (“It is axiomatic that exculpatory statements, 13 when shown to be false, are circumstantial evidence of 14 guilty consciousness and have independent probative 15 force.”), cert. denied, 419 U.S. 1105 (1975). Defense 16 counsel put forth a half-baked alibi defense, leaving 17 substantial additional alibi evidence unexplored, and 18 Rosario is paying the price. For all the foregoing reasons, 19 I would grant the writ of habeas corpus on a conditional 20 basis, providing the State with sufficient opportunity to 21 commence a new prosecution against Rosario prior to his 22 ordered release. Accordingly, I respectfully dissent. Page 68 of 69 1 I note that I agree with the majority’s implied denial 2 of habeas relief on the basis of Rosario’s actual innocence 3 claim. While I conclude it is unreasonable to hold that 4 defense counsel performed adequately and that there is no 5 reasonable probability that the verdict would have been 6 different had the additional alibi witnesses testified at 7 trial, I do not think that Rosario has surmounted the 8 extraordinary hurdle required to succeed on an actual 9 innocence claim, assuming such a claim exists under federal 10 law. Finally, I would not so quickly dismiss Rosario’s 11 claim of racial discrimination in the prosecutor’s use of 12 peremptory challenges; however, I need not reach the merits 13 of this claim, because I would grant a conditional writ of 14 habeas corpus based upon Rosario’s receipt of ineffective 15 assistance of counsel, which would warrant a new trial or 16 his release from custody – the same or greater relief that 17 would be provided by a successful Batson challenge. Page 69 of 69