This petition for collateral relief has been decided three times by the district court and twice by this one; it has been briefed four times in this court. Today’s decision is our third encounter with the subject.
A state judge disqualified one of Neftaly Rodriguez’s retained lawyers. After a federal district court concluded not only that the disqualification had been a mistake, but also that any such error automatically leads to a writ of habeas corpus, we reversed. 382 F.3d 670 (7th Cir.2004). We held that, because Rodriguez was represented by one lawyer of his choice, he had to show prejudice from the disqualification of another. We wrote: “The [effect] does not have to be great enough to undermine confidence in the outcome ... but it must be enough to show that the defendant’s representation suffered a setback from the disqualification.” Id. at 675.
On remand, the district court denied the petition after concluding that Rodriguez has not established that the erroneous disqualification had affected his trial. Five days later, the Supreme Court held in United States v. Gonzalez-Lopez, 548 U.S. 140, 126 S.Ct. 2557, 165 L.Ed.2d 409 (2006), that a mistaken refusal to allow a defendant’s chosen lawyer to represent him at trial is a “structural” error that requires relief without regard to prejudice. The district court then reversed course and issued a writ of habeas corpus. The parties’ appellate briefs concentrated on the question whether Gonzalez-Lopez applies when a defendant hires more than one lawyer and is represented at trial by at least one of his chosen counsel. They did not discuss whether Gonzalez-Lopez applies to prosecutions that were finally decided by state courts before it was released.
Before argument, we directed the parties to file supplemental briefs on that subject. We then held that, although Gonzalez-Lopez applies to multi-lawyer defense teams, it is not retroactive. 492 F.3d 863 (7th Cir.2007). This left the question posed by our first decision and answered by the district court’s second: Whether Rodriguez suffered a setback from the erroneous disqualification. We directed the parties to brief the appeal for a fourth time. It is at last ready for decision. Long overdue, actually. The parties have our apologies for the unnecessary delay.
Rodriguez was convicted of murder and sentenced to 29 years’ imprisonment. (He was released on parole after serving only 10 years, but this does not moot his request for collateral relief. See Spencer v. Kemna, 523 U.S. 1, 7, 118 S.Ct. 978, 140 L.Ed.2d 43 (1998).) After a bench trial, the judge concluded that Rodriguez was one of three gang members who battered a member of a rival gang with blunt objects (such as baseball bats), causing his death. *550After receiving Miranda warnings and waiving his right to counsel, Rodriguez confessed. At trial multiple witnesses identified Rodriguez as one of the aggressors. Counsel tried and failed to have the confessions suppressed; trial counsel (Perry Grimaldi) also cross-examined the witnesses vigorously in an attempt to undermine confidence in their identifications by bringing out differences in their descriptions of the events. Grimaldi had Rodriguez examined by mental-health specialists, hoping to set up a diminished-capacity defense or show that Rodriguez lacked the mental capacity to waive his right to counsel, though the defense ultimately did not present any expert evidence.
Joseph Brent was Grimaldi’s co-counsel for several months, until his erroneous disqualification. Our first opinion posed the question whether the disqualification adversely affected the defense. We suggested that this might be shown “if Brent had expertise that [Rodriguez’s] other lawyer lacked, or if Brent had planned a line of defense that co-counsel was unable to sustain on his own.” 382 F.3d at 675. On remand in the district court, Brent filed an affidavit that did not pursue either of these possibilities. Instead, Brent said, the defense suffered for two reasons: First, he “would have added another perspective” (Brent is a former prosecutor, while Grimaldi is not); second, he “would have assumed some of the responsibilities in investigating [and trying] the case”. At a deposition, Brent stated that there are intangible benefits of multiple lawyers and observed that he wanted to have Rodriguez evaluated by a psychiatrist in order to explore any options that such an evaluation might yield. Reminded that Grimaldi had done exactly this, and asked whether he would have pursued differently any issues arising from Rodriguez’s mental capacity, Brent said no. The district judge then concluded that no material dispute remained, because Grimaldi had the same skills as Brent (who conceded in his deposition that Grimaldi, his classmate in law school, is a first-rate trial lawyer) and Brent’s participation would not have changed either the nature of the defense or the probability of acquittal.
Rodriguez’s final brief devotes most of its space to arguing that we erred in 2004 when articulating the constitutional rule that governs the case, and erred again in 2007 when holding that Gonzalez-Lopez does not apply retroactively. That’s water under the bridge, however, as far as this court is concerned. Rodriguez, who does not suggest any reason to depart from the law of the case, has done what is necessary (and more than is required) to preserve his legal arguments for the Supreme Court.
He does present one legal argument not resolved in either 2004 or 2007: Who bears the burden of persuasion on the question whether the defense suffered a setback from the erroneous disqualification? The district court assigned that burden to Rodriguez, who maintains that it belonged on the prosecution’s side. But we need not decide who bears the burden under our (superseded) 2004 standard. Let us suppose that the burden should have been assigned to the state. That would not lead to a remand for a third round of proceedings in the district court. Rodriguez could not prevail under pre-Gonzalez-Lopez law without showing that any error “had substantial and injurious effect or influence in determining” the state judge’s decision. Brecht v. Abrahamson, 507 U.S. 619, 638, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993). This standard governs even though the state court thought that no error had occurred and therefore did not apply any standard of harmless-error review. Fry v. Pliler, 551 U.S. 112, 127 S.Ct. 2321, 168 L.Ed.2d 16 (2007); Johnson v. Acevedo, 572 F.3d 398, 403-04 (7th Cir.2009).
*551Brent’s affidavit and deposition are enough to establish that his disqualification did not cause a “substantial and injurious effect” at Rodriguez’s trial. Neither Brent nor Grimaldi contends that the two lawyers differed materially in knowledge, skills, or strategy. Rodriguez’s position thus boils down to the generic contention that two lawyers are better than one, which may be true but is not a constitutional proposition. The criminal justice system reflects the assumption that one competent lawyer normally is enough. (This was not a capital case, and even in federal capital prosecutions the entitlement to a second lawyer is statutory. 18 U.S.C. § 3005.) The prosecution was not so document-heavy that only a team of lawyers could keep up. Because the record would not support a conclusion that going to trial with only one lawyer “had a substantial and injurious effect” on the outcome, the state judiciary’s erroneous decision to disqualify Brent is harmless, and Rodriguez is not entitled to collateral relief.
Rodriguez contends that we should not apply Brecht because it would duplicate the prejudice aspect of the substantive claim. When a petitioner must show prejudice, as when arguing that counsel furnished ineffective assistance, see Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), it is unnecessary to show prejudice a second time through the lens of Brecht. That makes sense (and it reflects the Supreme Court’s assumption in Williams v. Taylor, 529 U.S. 362, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000)), but it does not assist Rodriguez. If the applicant for collateral relief must demonstrate a setback for the purpose of our 2004 decision, that avoids Brecht but leads to an adverse decision because neither Brent nor Grimaldi articulated any concrete setback from Brent’s disqualification. If the applicant does not bear such a burden, then Brecht does apply, and Rodriguez loses for the reasons we have given. One way or the other, the applicant must show injury; Rodriguez has not done so.
The judgment is reversed, and the case is remanded with instructions to deny the petition.