RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 26a0034p.06
UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT
┐ CLARENCE BORNS, │ Petitioner-Appellee, │ > No. 25-1437 │ v. │ │ TROY CHRISMAN, Warden, │ Respondent-Appellant. │ ┘
Appeal from the United States District Court for the Eastern District of Michigan at Detroit. No. 2:17-cv-13694—Terrence George Berg, District Judge.
Argued: January 29, 2026
Decided and Filed: February 11, 2026
Before: McKEAGUE, GRIFFIN, and THAPAR, Circuit Judges.
_________________
COUNSEL
ARGUED: Jared D. Schultz, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellant. Matthew A. Monahan, OFFICE OF THE FEDERAL COMMUNITY DEFENDER, Detroit, Michigan, for Appellee. ON BRIEF: Jared D. Schultz, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellant. Matthew A. Monahan, OFFICE OF THE FEDERAL COMMUNITY DEFENDER, Detroit, Michigan, for Appellee. THAPAR, J., delivered the opinion of the court in which McKEAGUE, J., concurred, and GRIFFIN, J., concurred in all but Part III. No. 25-1437 Borns v. Chrisman Page 2
OPINION _________________
THAPAR, Circuit Judge. Clarence Borns was convicted in Michigan state court of committing assault with intent to murder and illegally possessing a gun. He then filed a habeas corpus petition in federal court, claiming that his attorney was ineffective because he failed to investigate and present crucial witness testimony. But Borns filed his petition over a year after the limitations period expired. And he hasn’t identified a reason to excuse his lateness. So we reverse the district court’s conditional grant of Borns’s habeas petition.
I.
In December 2012, a vandal broke the windows of two homes owned by Evelyn Hardwick. She called the police, who said there was nothing they could do. But Hardwick “wanted to find out exactly . . . why . . . someone [would] come and tear out [her] windows in mid winter for absolutely nothing.” R. 23-6, Pg. ID 1715. She suspected the reason had something to do with her son, James Rankin.
Rankin lived in one of his mother’s vandalized homes. At the time, he’d been having problems with the mother of his child, Carlesha Harris. 1 When he discovered the broken windows, Rankin believed Harris was involved. So he drove over to a house on Hubbell Street where some of her family members lived. But he saw several vehicles and a crowd of people outside the home, so he left. Rankin then met up with his mother and his two sisters. They drove back to Hubbell Street and parked several houses down from where Harris’s family lived and was gathered outside.
One of Harris’s uncles noticed Rankin pull up and walked down the street to where Rankin had parked. As he approached, Hardwick asked him, “Why would you all come and tear out my home windows?” Id. at 1722. The uncle claimed that he didn’t know anything about the
1The record isn’t clear about the names of many people involved in this case. For instance, Carlesha Harris is elsewhere referred to as “Carlyssa Borns.” The “Borns” family members occasionally go by “Barnes.” This opinion seeks to use names consistently. No. 25-1437 Borns v. Chrisman Page 3
vandalism. But the conversation quickly escalated, and Rankin, his mother, and his sisters all exited the car. While the argument continued, another man standing outside Harris’s family’s house heard the commotion down the street and yelled, “I’m the one that broke out the window.” R. 23-7, Pg. ID 1791.
Rankin began walking toward the man. As he did, the man drew a gun from his sweatshirt pocket and fired it into the air. Rankin and his family members turned and ran in the opposite direction. The shooter fired again, hitting the car’s trunk and back window as Rankin tried to open the back door and take cover. In her haste to escape, Hardwick tripped over the curb. Rankin stopped to help her up and was struck in the leg by three bullets. Meanwhile, his sisters took cover nearby and asked a neighbor to call the police. But the shooter jumped into a white car and fled the scene.
Shortly after the shooting, Rankin and his family members spoke to the police. All four of them independently viewed photo lineups and identified Borns as the shooter. Rankin knew Borns was Harris’s uncle and had seen him while spending time with Harris’s family.
Borns was charged with two counts of assault with intent to murder, one count of possessing a firearm as a felon, and one count of possessing a firearm while committing a felony. He pled not guilty, and the case proceeded to a jury trial.
At trial, the defense primarily relied on testimony from Borns’s sister, Melissa.2 She owned the Hubbell Street house near where the shooting occurred. Melissa’s trial testimony cast doubt on Borns’s identity as the shooter. She first testified that at around 5:00 a.m. on the day of the shooting, a brick crashed through her front window. She ran to the door to identify the culprit and saw Rankin walking to his van. Melissa’s brother, Carl, pursued Rankin and started shooting at him.3 According to Melissa, when Rankin returned later with his family, Carl said, “[H]e’s back.” Id. at 1902. Melissa testified that she followed Carl as he ran outside with a gun. She then went back inside to put shoes on and heard shots. A moment later, she saw Borns
2This opinion refers to the petitioner, Clarence Borns, by his last name, and to all other Borns family members by their first names. We mean no disrespect by using first names and do so only for clarity. 3Rankin denied visiting the house at that time and breaking the window. No. 25-1437 Borns v. Chrisman Page 4
inside the house. She then returned to the front lawn and saw Carl standing by the street with a gun in his hand. In contrast, she never saw Borns with a gun.
Although no physical evidence tied Borns to the shooting, the prosecution relied on a recorded statement and the victims’ testimony. In a phone call from prison, Borns admitted to getting caught up in a domestic-violence dispute between Harris and Rankin. Rankin’s mother and sisters identified Borns as the shooter to police after the incident and again in court. Rankin died in an unrelated incident shortly before trial, so he didn’t testify at trial. But the government introduced transcripts from an in-court pretrial examination in which Rankin identified Borns as the shooter and testified about the incident. The jury ultimately convicted Borns of four offenses.
Borns appealed. He claimed that admitting Rankin’s pretrial examination violated his Confrontation Clause rights, and he challenged the sufficiency of the evidence that he intended to kill Rankin. People v. Borns, No. 318376, 2014 WL 7442251, at *1–3 (Mich. Ct. App. Dec. 30, 2014) (per curiam). He also attached a pro se brief, claiming that his trial counsel was ineffective and requesting a hearing to investigate that claim. That brief included affidavits from four uncalled witnesses who claimed they either (1) saw Borns in the house at the time of the shooting, or (2) saw Carl—not Borns—fire the shots. But the state courts didn’t accept his pro se brief because it didn’t comply with various procedural rules. Ultimately, the Michigan Court of Appeals affirmed, and the Michigan Supreme Court denied leave to appeal. Id. at *4; People v. Borns, 866 N.W.2d 454 (Mich. 2015) (mem.).
Borns then filed a pro se motion for relief from judgment in state court. He repeated the same ineffective-assistance claim from his pro se brief and provided the same affidavits. In relevant part, he argued that his trial counsel was ineffective for failing to contact, investigate, and present the four uncalled witnesses. He also argued that his appellate counsel was ineffective for failing to raise those issues on direct appeal. The state court denied his motion, finding that Borns’s trial counsel made a reasonable strategic decision not to call the witnesses and that his appellate counsel reasonably decided to bring only the claims he thought would prevail. Borns sought leave to appeal from the Michigan Court of Appeals and Michigan No. 25-1437 Borns v. Chrisman Page 5
Supreme Court. See People v. Borns, 911 N.W.2d 715 (Mich. 2018) (mem.). Both denied leave. Id.
Borns later filed a habeas petition in federal court, primarily based on his ineffective- assistance claims. Michigan moved to dismiss the petition as untimely under the one-year statute of limitations in the Antiterrorism and Effective Death Penalty Act (AEDPA). The district court rejected that motion and directed the state to respond to the merits of Borns’s claims. After the state responded, the district court found that the state court unreasonably applied clearly established Supreme Court precedent in rejecting Borns’s ineffective-assistance claims. So it conditionally granted the writ, directing the state to either release Borns or initiate proceedings to retry him within 120 days of the order. The state timely appealed.
II.
This case begins and ends with AEDPA’s one-year limitations period. See 28 U.S.C. § 2244(d)(1). We review a district court’s application of the statute of limitations de novo. Moss v. Miniard, 62 F.4th 1002, 1008 (6th Cir. 2023). A state prisoner must file a habeas petition within one year of “the date on which the judgment became final by the conclusion of direct review or the expiration of the time for seeking such review.” 28 U.S.C. § 2244(d)(1)(A). However, a state motion for postconviction relief tolls the statute of limitations while that motion is pending. Id. § 2244(d)(2). And even if a petition is untimely, AEDPA’s limitations period may be subject to equitable tolling. Holland v. Florida, 560 U.S. 631, 649 (2010). Borns’s state-court motion didn’t toll the statute of limitations, and he can’t show that he’s entitled to equitable tolling, so his habeas petition was filed too late.
A.
The Michigan Supreme Court denied leave to appeal on July 28, 2015. See Borns, 866 N.W.2d at 454 (mem.). Borns’s window to seek certiorari on that denial expired 90 days later, on October 26, 2015. See Sup. Ct. R. 13.1. AEDPA’s limitations period began the next day and ran until October 26, 2016. See Moss, 62 F.4th at 1010. Borns agrees with both of these dates and doesn’t dispute that he filed his habeas petition in federal court on November 2, 2017—over a year later. No. 25-1437 Borns v. Chrisman Page 6
So how does Borns attempt to get around his late filing? He points to a postconviction motion he filed in state court and argues that it tolled AEDPA’s limitations period. Motions for postconviction relief filed before the limitations period expires pause the statutory clock. Vroman v. Brigano, 346 F.3d 598, 602 (6th Cir. 2003). So the question here is whether Borns filed his motion before AEDPA’s one-year clock ended on October 26, 2016.
The parties dispute when exactly Borns filed that motion. Borns suggests that the motion was filed on October 24, 2016, the day he signed it. But on the state’s telling, the motion was filed the day the state court docketed it: November 4, 2016. Of course, if the state is correct, Borns filed his motion outside the one-year limitations period. Since he filed the motion in a Michigan court, we look to Michigan law to determine the date of filing. See Artuz v. Bennett, 531 U.S. 4, 8 (2000) (holding that state law governs whether a state postconviction motion is properly filed in the context of a federal habeas petition).4
Borns invokes two Michigan court rules governing prisoner filings to argue his motion was timely: Michigan’s 2021 universal prison-mailbox rule and its older appellate prison- mailbox rule. But neither rule applies.
Start with Michigan’s current prison-mailbox rule. In 2021, Michigan adopted a rule declaring that documents “must be deemed timely filed” if deposited in a prison mailbox “on or before the filing deadline.” Mich. Ct. R. 1.112. The rule covers any “pleading or other document deposited” by a pro se prisoner. Id. But Borns mailed his motion in 2016, so the 2021 rule can’t help him. See, e.g., Kincaid v. Campbell, 2:21-CV-10574 (LVP), 2024 WL 1054525, at *5 (E.D. Mich. Mar. 11, 2024) (refusing to apply the 2021 mailbox rule to a habeas petition submitted before the rule change).
4To the extent that the district court applied a federal prison-mailbox rule to find Borns’s motion timely, it erred. The district court distinguished Vroman on the basis that Ohio state courts had previously held the postconviction motion in that case untimely, and Michigan courts hadn’t done the same here. But it ignored a critical part of Vroman’s reasoning: “[T]he timeliness of [a state] prisoner’s post-conviction petition is governed by state statute.” 346 F.3d at 603; accord Davis v. Bradshaw, 900 F.3d 315, 324 (6th Cir. 2018). The Vroman court didn’t just rely on Ohio state courts’ previous rulings in that case. It also looked to Ohio law in general, including state statutes and state-court interpretations of them. So Vroman teaches that we must look to Michigan law to determine whether Borns timely filed. No. 25-1437 Borns v. Chrisman Page 7
To avoid that result, Borns asks us to apply the rule retroactively. But Michigan court rules apply only prospectively absent a clear intent for them to apply retroactively. See Johnson v. Pastoriza, 818 N.W.2d 279, 285 (Mich. 2012); see also Tyler v. Findling, 972 N.W.2d 833, 836 (Mich. 2021). And text provides the best evidence of that “intent.” Johnson, 818 N.W.2d at 287. Here, the rule’s text says nothing about retroactivity. See Mich. Ct. R. 1.112. It also carries an effective date: September 1, 2021. Id. As the Michigan Supreme Court has recognized, “providing a specific, future effective date and omitting any reference to retroactivity” indicates that a statute or rule should apply only prospectively. Brewer v. A.D. Transp. Express, Inc., 782 N.W.2d 475, 479 (Mich. 2010) (quotation omitted). In fact, that court previously held that the state’s first appellate prison-mailbox rule applied only prospectively. See People v. Lewis, 806 N.W.2d 311 (Mich. 2011) (mem.). We see no reason to apply the 2021 expansion differently.
In the alternative, Borns invokes Michigan’s earlier appellate prison-mailbox rule and urges us to look to “the purposes animating” the rule to extend its text to postconviction motions. Appellee’s Br. at 35. When Borns submitted his motion in 2016, Michigan’s prison-mailbox rule didn’t cover postconviction motions. It applied only to “appeals from administrative agencies, appeals from circuit courts . . . , and appeals from decisions of the Court of Appeals.” Mich. Ct. R. 7.105 staff cmt. to second May 1, 2010 amendment. The most natural reading of that list excludes any items not expressly enumerated—like postconviction motions for relief. Cf. Chevron U.S.A. Inc. v. Echazabal, 536 U.S. 73, 80 (2002). That’s precisely how Michigan courts read state statutes, refusing to “assume that the [drafters] inadvertently omitted” items “and then, on the basis of that assumption, apply what is not there.” Farrington v. Total Petroleum, Inc., 501 N.W.2d 76, 80 (Mich. 1993). Instead, Michigan courts interpret their court rules as written. See People v. Burkman, 15 N.W.3d 216, 225 (Mich. 2024); Tyler, 972 N.W.2d at 836. And as written, the narrower appellate prison-mailbox rule doesn’t cover Borns’s 2016 motion.
What’s more, commentary to Michigan’s 2021 universal prison-mailbox rule states that it “expan[ded]” the existing appellate version. Mich. Ct. R. 1.112 staff cmt. to 2021 adoption. Expansion logically implies that the new rule covers additional items not previously included. No. 25-1437 Borns v. Chrisman Page 8
See Expand, Merriam-Webster’s Collegiate Dictionary (11th ed. 2004) (defining “expand” as “to increase the extent . . . or scope of”). And postconviction motions weren’t listed in the older version of the rule. See Mich. Ct. R. 7.105 staff cmt. to second May 1, 2010 amendment. That’s further evidence that Michigan courts wouldn’t have applied the prison-mailbox rule to Borns’s motion at the time he filed it. See Robinson v. Romanowski, No. 2:14-CV-10617 (GAD), 2014 WL 5480808, at *2 n.2 (E.D. Mich. Oct. 29, 2014).
And even if either prison-mailbox rule applied, Borns hasn’t made the required showing that he deposited his motion in a prison mailbox on time. He argues only that he signed the motion within the limitations period. But the 2021 rule requires “[p]roof” of when the prisoner deposited the document in the mail, such as mailing receipts, prepaid postage, or postmarks. Mich. Ct. R. 1.112; see, e.g., Garrett v. Braman, No. 1:23-CV-12499 (TLL), 2025 WL 725249, at *2 n.1 (E.D. Mich. Mar. 6, 2025) (clarifying that under the 2021 rule, the filing’s “outgoing postmark date”—not date of signature—“controls”). Likewise, the older prison-mailbox rule required a sworn statement including the date of deposit and prepaid postage. See Mich. Ct. R. 7.205(A)(3) (2010) (repealed Sep. 23, 2018); Mich. Ct. R. 7.204(A)(2)(e) (2010) (repealed Sep. 23, 2020); see also, e.g., Robinson, 2014 WL 5480808, at *2 n.2. Borns offers nothing more than his signature date. That’s not enough, even if either rule applied.
Because neither prison-mailbox rule applies, Borns’s motion wasn’t filed within AEDPA’s one-year limitations period. Absent guidance from the prison-mailbox rules, we look to general Michigan court rules to determine when Michigan courts deem a motion filed. See Artuz, 531 U.S. at 8; Davis, 900 F.3d at 324. Recall Borns’s theory: He argues his motion for relief from judgment was filed when he signed it on October 24, 2016, with two days to spare. He doesn’t dispute that the state court docketed that motion on November 4, 2016—nine days too late.
In all cases when the prison-mailbox rule doesn’t apply, the Michigan Supreme Court considers a document “filed” “only when it is actually received by the Clerk’s Office in Lansing.” Filing Requirements, Mich. Sup. Ct., https://perma.cc/Y2CS-ZQ2Z. A document delivered by mail, like Borns’s motion, “is ‘filed’ when it is received and accepted by the clerk with the intent to enter it in the record.” Mich. Sup. Ct. I.O.P. 7.202(4)-1; accord Hussain v. No. 25-1437 Borns v. Chrisman Page 9
Barrett, No. 2:16-CV-11342 (GAD), 2016 WL 6995738, at *3 n.1 (E.D. Mich. Nov. 30, 2016); Mich. Ct. R. 7.102(6) (defining “date of filing” as “the date of receipt of a document by the ‘clerk’”). That also fits with Michigan’s limited guidance in the specific context of postconviction motions. See Mich. Ct. R. 6.503(A) (directing petitioners to “file a motion . . . with the clerk of the court,” who then “shall file it”); accord Walker-Bey v. Dep’t of Corr., 564 N.W.2d 171, 171–72 (Mich. Ct. App. 1997) (per curiam) (refusing to apply a prison-mailbox rule to a court rule requiring “filing . . . with the court clerk”). These state rules reflect the ordinary meaning of “filed.” See Artuz, 531 U.S. at 8 (collecting definitions). So we agree with the state that Borns filed his motion on November 4, 2016, meaning that his motion for relief from judgment didn’t toll AEDPA’s one-year limitations period.
Because his state-court motion didn’t toll AEDPA’s clock, Borns’s federal habeas petition came over a year too late: on November 2, 2017. Although AEDPA’s statute of limitations isn’t jurisdictional, Holland, 560 U.S. at 645, Michigan appropriately invoked it here. So unless equitable tolling applies, we must enforce it. Cf. Gonzalez v. Thaler, 565 U.S. 134, 146 (2012) (“[C]alling a rule nonjurisdictional does not mean that it is not mandatory or that a timely objection can be ignored.”); Fort Bend County v. Davis, 587 U.S. 541, 549 (2019) (“A claim-processing rule may be ‘mandatory’ in the sense that a court must enforce the rule if a party ‘properly raise[s]’ it.”).
B.
Alternatively, Borns asks us to apply equitable tolling to excuse his late filing. This court applies equitable tolling “sparingly.” Robertson v. Simpson, 624 F.3d 781, 783 (6th Cir. 2010). Borns bears the burden of showing that (1) he diligently pursued his rights and (2) some extraordinary circumstance prevented him from filing on time. Holland, 560 U.S. at 649; see also Robertson, 624 F.3d at 784. He can’t show either.
Borns’s request for equitable tolling largely repeats his arguments as to why we should deem his habeas petition timely. As to diligence, he argues that he “signed and mailed his state postconviction motion in enough time to toll the statute of limitations.” Appellee’s Br. at 38. But that doesn’t explain why he waited 364 days after his conviction became final to mail his No. 25-1437 Borns v. Chrisman Page 10
motion for relief. The Michigan Supreme Court denied leave in Borns’s direct appeal on July 28, 2015. See Borns, 866 N.W.2d at 454. And he offers no explanation for his failure to file a petition for certiorari or do anything else to diligently pursue his rights until October 24, 2016. Borns had previously tried to present the four new affidavits and raise his mistaken-identity theory on direct appeal. So his twelve-month delay in presenting the exact same claim on collateral review shows a lack of diligence. In fact, we’ve denied equitable tolling in cases when petitioners sat idle for far less time. See, e.g., Hall v. Warden, Lebanon Corr. Inst., 662 F.3d 745, 752 (6th Cir. 2011) (five months); Allen v. Yukins, 366 F.3d 396, 403 (6th Cir. 2004) (seven months); Greene v. Lafler, 457 F. App’x 485, 487 (6th Cir. 2012) (per curiam) (five months). Thus, Borns can’t show diligence.
He also can’t show an extraordinary circumstance prevented him from filing. Borns claims that “the unsettled nature of Michigan’s prison-mailbox rule in the 2010s presents an extraordinary circumstance.” Appellant’s Br. at 38. But state law wasn’t unsettled at the time. It didn’t cover motions for postconviction relief. See supra Section II.A; Mich. Ct. R. 7.105 staff cmt. to second May 1, 2010 amendment. To the extent Borns claims he was confused, “[t]he settled state of the law at the relevant time belies any claim to legal confusion.” Lawrence v. Florida, 549 U.S. 327, 336 (2007). And to the extent Borns claims ignorance of the applicable deadlines, a petitioner’s “pro se status and lack of knowledge of the law are not sufficient to constitute an extraordinary circumstance.” Keeling v. Warden, Lebanon Corr. Inst., 673 F.3d 452, 464 (6th Cir. 2012). In short, Borns hasn’t identified any extraordinary circumstances that would warrant equitable tolling.
Even if we agreed that Borns was diligent or faced extraordinary circumstances, his claim for equitable tolling fails for a third reason. That’s because petitioners must show “a causal link between” the extraordinary circumstance and “untimely filing.” Ata v. Scutt, 662 F.3d 736, 742 (6th Cir. 2011). Borns doesn’t allege that he waited over a year to file his federal habeas petition because of the prison-mailbox rule or any related confusion about the filing deadline. He doesn’t even argue that he believed the rule applied to his filing at the time. So even if we agreed that unsettled law could constitute extraordinary circumstances, he hasn’t shown that those circumstances caused his late filing. That’s independently fatal to his claim. No. 25-1437 Borns v. Chrisman Page 11
In sum, Borns hasn’t established diligent pursuit of his rights or extraordinary circumstances. So he isn’t entitled to equitable tolling. Because his federal habeas petition was untimely under AEDPA’s one-year statute of limitations, we reverse the district court’s grant of his petition.
III.
Even if we assume Borns’s federal habeas petition was timely, it would fail on the merits. The state argues we can’t reach that question because Borns procedurally defaulted his claims. But procedural default doesn’t affect our jurisdiction. Day v. McDonough, 547 U.S. 198, 205 (2006). So we can address the merits of Borns’s claim without answering the more complex question of procedural default. Lambrix v. Singletary, 520 U.S. 518, 525 (1997). Because we can easily resolve the claim against Borns, we proceed to the merits rather than addressing procedural default.
We review Borns’s ineffective-assistance claim through AEDPA’s deferential lens. A Michigan state court already reviewed the claim and found it meritless. So AEDPA authorizes relief only if the Michigan court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1). Review under section 2254(d)(1) is “highly deferential” and “demands that state-court decisions be given the benefit of the doubt.” Cullen v. Pinholster, 563 U.S. 170, 181 (2011) (quotation omitted).
Borns argues the Michigan court’s decision unreasonably applied Supreme Court precedent. An “unreasonable application” of Supreme Court precedent is more than simply “incorrect or erroneous.” Wiggins v. Smith, 539 U.S. 510, 520–21 (2003). Instead, Borns must demonstrate that the Michigan court “applie[d] Supreme Court precedent in a way that no fair- minded judge could accept.” Rogers v. Mays, 69 F.4th 381, 389 (6th Cir. 2023) (en banc).
The Michigan court identified the correct legal principle governing Borns’s claim: Strickland’s dual requirements of deficient performance and prejudice. Strickland v. Washington, 466 U.S. 668, 687 (1984). Strickland directs that “[j]udicial scrutiny of counsel’s performance must be highly deferential” and “indulge a strong presumption” of adequate No. 25-1437 Borns v. Chrisman Page 12
assistance. Id. at 689. When reviewing a Strickland claim under section 2254(d)(1), AEDPA adds a second layer of deference. Harrington v. Richter, 562 U.S. 86, 105 (2011). So Borns faces a “substantially higher threshold” than if we reviewed his ineffective-assistance claim de novo. Knowles v. Mirzayance, 556 U.S. 111, 123 (2009) (quotation omitted).
Borns’s trial attorney didn’t perform deficiently under Strickland. Deficient performance occurs when “counsel’s actions ‘so undermined’ the trial that it could not have produced a just result.” Rogers, 69 F.4th at 389–90 (quoting Strickland, 466 U.S. at 686). And in this procedural posture, Borns can’t succeed if “there is any reasonable argument that counsel satisfied Strickland’s deferential standard.” Harrington, 562 U.S. at 105.
An attorney’s failure to investigate witnesses isn’t always ineffective assistance. Rather, counsel may choose not to make certain investigations—so long as that decision is “reasonable” within the circumstances of the case. Strickland, 466 U.S. at 691. To make that determination, we consider only “the objective reasonableness of counsel’s performance, not counsel’s subjective state of mind.” Harrington, 562 U.S. at 110. That means we must “affirmatively entertain the range of possible reasons” for counsel’s actions. Pinholster, 563 U.S. at 196 (quotation omitted). Viewed through Strickland and AEDPA’s “doubly” deferential lens, a fair- minded judge could conclude that Borns’s trial counsel didn’t perform deficiently. Harrington, 562 U.S. at 105 (quotation omitted).
1.
To start, Borns’s claim suffers from serious factual deficiencies. In Strickland cases, the defendant bears the burden of proof, and “the absence of evidence cannot overcome the strong presumption” of adequate performance. Burt v. Titlow, 571 U.S. 12, 22–23 (2013) (quotation omitted). The record doesn’t clearly indicate that Borns told trial counsel about the four alibi witnesses at the time of trial.5 That’s relevant to deficient performance because “what
5At Borns’s initial arraignment hearing, Borns’s counsel mentioned calling “all sorts of witnesses.” R. 23- 3, Pg. ID 1526. But there’s no indication of who those witnesses were nor what they planned to testify about. Borns’s attorney also included “Mr. O.J. Heils” on an initial witness list. R. 9-10, Pg. ID 844. Assuming that “O.J. Heils” is the same person as Othello Nails, that means counsel knew about one of the alibi witnesses before trial No. 25-1437 Borns v. Chrisman Page 13
investigation decisions are reasonable depends critically on” “information supplied by the defendant.” Strickland, 466 U.S. at 691. What’s more, Borns alleges that he told counsel about the additional witnesses six months after his arraignment. This curiously long delay to present such critical evidence raises further doubts about Borns’s claims.
There are other reasons to doubt Borns’s story. For instance, he never raised the issue when counsel didn’t include three of the four alibi witnesses in the list of witnesses he planned to call at trial. And at trial, he was content to “follow counsel[’s] advice” and declined to testify in his defense. R. 23-7, Pg. ID 1922. Surely a defendant who knew about four uncalled alibi witnesses would have taken any opportunity to mention them. But he didn’t. And Borns again declined to mention them at sentencing. Years later, Borns submitted an email purporting to show that he informed his counsel about the witnesses before trial. But the email was sent to his sister’s email address—not his lawyer’s—and was timestamped almost three years after his trial. So it doesn’t provide any additional support for his claim.
The allegation that trial counsel failed to investigate the witnesses suffers from the same problems. Only Borns and the four witnesses say that trial counsel didn’t contact them or otherwise look into their claims. And it’s impossible to judge their credibility on a paper record. See Blackledge v. Allison, 431 U.S. 63, 82 n.25 (1977). The alleged witnesses also stayed silent through Borns’s trial, conviction, and sentencing. Cf. McCray v. Vasbinder, 499 F.3d 568, 473 (6th Cir. 2007) (asking why family members provided no “good explanation for why they took so long to come forward with evidence that their relation stands wrongly accused of murder”). That silence places Borns’s legal claim on shaky factual ground. At the end of the day, Borns asks to overcome Strickland’s “strong presumption” of adequate performance based on his own word and four bare affidavits from his friends and family members. 466 U.S. at 689. But our caselaw demands more, especially when granting the extraordinary relief of habeas corpus.6
(and chose not to call him for some reason). But Borns bears the burden of establishing the factual basis for his ineffective-assistance claim. Burt, 571 U.S. at 22–23. Simply put, Borns had to do more to rebut Strickland’s weighty “presumption of effectiveness.” Id. at 23. 6The district court didn’t grapple with any of these inconsistencies and seemed to credit Borns’s story wholesale. In part, it found that the affidavits lacked certain attributes that usually indicate unreliability. But that’s not enough. See Burt, 571 U.S. at 22–23. Federal habeas relief “must be tailored” to remedy a legal injury “without No. 25-1437 Borns v. Chrisman Page 14
2.
Even assuming Borns’s allegations are true, trial counsel’s decision not to call the four alibi witnesses falls within the broad range of reasonable decisions Harrington protects. Under Strickland, we don’t ask whether counsel could have reasonably—or rightly—chosen a different strategy. We ask whether counsel could have reasonably chosen the one he did. See Pinholster, 563 U.S. at 196. And we can think of plenty of reasons why Borns’s counsel chose to proceed as he did.
For one, the additional testimony may have been cumulative. Borns’s sister, Melissa, testified at trial that she saw Carl outside with a gun and that Borns was in the house when she heard gunshots. That testimony includes everything the additional witnesses would have said. It’s true that multiple witnesses testifying to the same fact isn’t necessarily cumulative. See Vasquez v. Jones, 496 F.3d 564, 576 (6th Cir. 2007). But at the same time, a single witness—or even no eyewitnesses at all—can support a conviction. See, e.g., United States v. Underwood, 129 F.4th 912, 939 (6th Cir. 2025) (one); James v. Corrigan, 85 F.4th 392, 395 (6th Cir. 2023) (zero). The value of witness testimony in any given case instead depends on a number of factors, including the witness’s credibility. Cf. Penn. R.R. Co. v. Chamberlain, 288 U.S. 333, 338 (1933) (“[W]here there is a direct conflict of testimony upon a matter of fact, the question must be left to the jury to determine, without regard to the number of witnesses upon either side.”). So the district court’s conclusion that four additional defense witnesses could “strengthen[] the defense” turned the inherently unscientific and case-specific question of witness credibility into a simple game of nose-counting. R. 30, Pg. ID 3344–45. Such nose-counting ignores that calling multiple witnesses carries both benefits and risks. For instance, witnesses’ stories could contradict each other and impact the strength of the case before the jury. Strickland demands far more deference to counsel’s professional judgment in weighing these costs and benefits.7 Here,
unnecessarily infringing on competing interests of comity, federalism, and finality.” Ewing v. Horton, 914 F.3d 1027, 1032–33 (6th Cir. 2019) The district court’s acceptance of Borns’s claims based on such shaky factual grounds without an evidentiary hearing undermines the “deference and respect owed to state courts” on federal collateral review. Id. at 1033. 7Imagine the consequences if failing to call a second, third, fourth, or fifth witness to offer the same testimony were automatically ineffective assistance of counsel. Two options would emerge: Either trials would No. 25-1437 Borns v. Chrisman Page 15
counsel could have reasonably chosen to call one witness instead of four others with the same story—and with associated risks.
More testimony about identity could have helped the defense—but more testimony could also have hurt it, because the testimony came from people close to the defendant. Each of the four affiants had a relationship with Borns: Patricia was his sister, Othello Nails was Patricia’s boyfriend, Gregg Marshall was a family friend, and Lashanti Bovans was Borns’s niece. Borns argues that these relationships made their testimony more reliable because the affidavits implicated Carl, another family member. But family feuds are common. So counsel could have reasonably decided that the affiants’ specific relationships to Carl and Borns undercut their testimony. Or counsel could have worried that their testimony seemed suspiciously coordinated. See United States v. Foreman, 323 F.3d 498, 503 (6th Cir. 2003) (holding that counsel wasn’t ineffective for declining to call a witness whose testimony seemed “fabricat[ed]” even though it corroborated the defendant’s alibi). And counsel could have decided that probing these relationships and possible biases four more times would cause doubt about their stories and unnecessary delay—as the state court recognized.
Trial counsel also could have worried about the affiants’ ability to recall and describe relevant events. The Borns family had gathered the night before the shooting to celebrate Melissa’s birthday, returning to her house after visiting a club. The party “continue[d]” there and the group “stayed up” all night. R. 23-7, Pg. ID 1899–1900. Perhaps trial counsel was concerned about the reliability of the affiants’ memories after a night of partying and little sleep. Trial counsel could have reasonably preferred to call one witness instead of providing multiple witnesses whose recollections might not match and could easily be disputed on cross- examination.
Borns asks us to overlook these reasonable explanations because of the affidavits’ alleged substantive reliability. The district court agreed with him that the affidavits “bear certain earmarks of reliability.” R. 30, Pg. ID 3345. It first noted that none of the witnesses recanted prior statements. That’s true—because none of them came forward until after trial, so they gave
become bloated and inefficient, with defendants calling every available witness to prove every contested point of fact, or defendants would call one witness and immediately claim ineffective assistance if they lost. No. 25-1437 Borns v. Chrisman Page 16
no testimony to recant. It’s also irrelevant. While recanting affidavits are indeed considered with “extreme suspicion,” non-recanting affidavits don’t get a corresponding credibility boost. Houston v. Tanner, 160 F.4th 683, 694 (6th Cir. 2025) (quotation omitted). The district court also found that the affidavits were promptly offered, in contrast to “‘11th hour’ affidavits” that warrant “a fair degree of skepticism.” R. 30, Pg. ID 3346 (quoting Herrera v. Collins, 506 U.S. 390, 423 (1993) (O’Connor, J., concurring)). That’s debatable: The first three witnesses came forward over six months after Borns’s sentencing. That seems close to the 11th hour—they came forward well after Borns’s arrest, over six months of trial preparation, and his eventual guilty verdict and sentence. And the district court’s analysis is largely beside the point. Whether the district court—or this court—thinks counsel made a bad strategic decision doesn’t matter. Under Harrington, we ask only “whether there is any reasonable argument that counsel satisfied Strickland’s deferential standard.” 562 U.S. at 105. There are many such arguments, so Borns’s ineffective-assistance claim fails.
Strickland also requires petitioners to “affirmatively prove prejudice.” 466 U.S. at 693. That’s a high bar, demanding that Borns show “a reasonable probability that, but for counsel’s unprofessional errors, the result” of his trial “would have been different.” Id. at 694. A reasonable probability requires more than mere doubt. Rather, “the likelihood of a different result must be substantial, not just conceivable.” Harrington, 562 U.S. at 112.
The prosecution presented a strong case based on eyewitness testimony that Borns was the shooter. Although Rankin died before trial, he identified Borns as the shooter from a photographic lineup and in court at his pretrial examination. Rankin’s identification was especially reliable because he knew Borns before the shooting as the uncle of Carlesha Harris, who was the mother of Rankin’s child. Rankin’s mother and two sisters also identified Borns. They did so in photographic lineups shortly after the shooting. They did so again in open court. And they testified at length—subject to cross-examination—about what Borns was wearing, what he said to them, and how the shooting unfolded. No. 25-1437 Borns v. Chrisman Page 17
The state court concluded that “the jury simply believed” the government’s witnesses, not Borns’s. R. 23-12, Pg. ID 2078. Borns argues that no fair-minded jurist could agree with the state court’s conclusion because the jury didn’t convict him on all counts. But the jury could have believed the prosecution’s witnesses over the defense’s and still rendered a split verdict. For example, the jury found Borns guilty of a lesser included offense as to Evelyn Hardwick: assault with intent to do great bodily harm less than murder. But that verdict still required the jury to find that Borns was the shooter—and thus believe the state’s eyewitnesses. Likewise, even though the jury acquitted Borns of any assault charges against Rankin’s sister Latisha, that verdict doesn’t cast doubt on whether the jury believed the state’s eyewitnesses. That’s because Rankin’s sisters were hiding under cover farther away from the shooter, which may have made it harder for the state to prove that the shooter intended to kill them. So a fair-minded jurist could still agree with the Michigan courts that there was no reasonable probability that additional witnesses would have changed the result of Borns’s trial.
The Supreme Court has repeatedly emphasized that under Strickland, “even a strong case for relief does not mean the state court’s contrary conclusion was unreasonable.” Harrington, 562 U.S. at 102. The district court here focused more on the strength of Borns’s case for relief than on the deferential standard that Strickland and AEDPA require. And “[f]aithful application of those standards sometimes puts federal district courts . . . in the disagreeable position of having to deny relief in cases they would have analyzed differently if they had been in the shoes of the relevant state court.” Klein v. Martin, 607 U.S. ---, 2026 WL 189976, at *1 (Jan. 26, 2026). But “federal courts are dutybound to comply with AEDPA,” no matter how disagreeable they may find it. Id. (collecting examples of summary relief in AEDPA cases). Because a fair- minded jurist could conclude the state court resolved Borns’s claims correctly, the district court erred by conditionally granting his habeas petition.
* * *
We reverse.