Sеveral months prior to Neftaly Rodriguez’s trial for murder, the prosecutor filed a motion to disqualify Joseph Brent as one of his lawyers. Brent also represented detective John McMurray in an unrelated real estate deal. The prosecutor told the judge that McMurray was an “intеgral part of the case” against Rodriguez because he had participated in the investigation of one of Rodriguez’s co-defendants. According to the prosecutor, Brent’s simultaneous representation of Rodriguez and McMurray created a “per se conflict of interest”. The thinking was not spelled out but must have been that, if McMurray appeared on the stand, Brent would treat him with kid gloves during cross-examination, lest he risk upsetting and losing •a paying client. Relying on the prosecutorial assurance that McMurray would testify, the judge held that a conflict existed and deemed inadequate Rodriguez’s offer to waive his right to conflict-free counsel. The court prevented Brent from rendering Rodriguez any further assistance. When trial arrived, however, the prosecutor failed to call McMurray as a witness. Rodriguez was convicted and оn appeal argued that Brent had been disqualified
*672
improperly. The Appellate Court of Illinois rejected this contention,
People v. Rodriguez,
Rodriguez has petitioned for a writ of habeas corpus. 28 U.S.C. § 2254. A state court’s decision may be upset on collateral review only if its decision is contrary to, or an unreasonable application of, law clearly established by the Supreme Court. 28 U.S.C. § 2254(d). Rodriguez does not contend that the state court’s decision contradicts settled law. The Appellate Court recognized, citing
Wheat v. United States,
Disqualification stemmed from the prosecutor’s assurance that MсMurray was a vital witness. But McMurray did not testify, and on collateral review Illinois does not argue that McMurray could have provided any admissible evidence against Rodriguez. The prosecutor has never explained this about-face, nor did the state’s appellate court try to mаke sense of it. All the prosecutor could manage in the state proceedings was to argue that Rodriguez had not shown that McMurray was certain not to testify. We grant the (implied) point that decisions must be made ex ante; if disqualification was proper on the basis of all information known or knowable at the time the judge acted, then later developments (such as a severance that allowed the prosecutor to withhold McMurray as a witness against Rodriguez, while using him against other defendants) would not spoil the decision. This differs, however, from an argument that Rodriguez bоre the burden of production and persuasion. How could he have known the details of the prosecution’s strategy? Under Wheat the risk of non-persuasion rests with the prosecution rather than the defendant. The state’s appellate court may have shared the prosecutоr’s error; certainly its opinion does not disclaim it or otherwise evaluate the prosecutor’s failure to put McMurray on the stand, which pulled the rug out from under the trial court’s decision.
The state’s line of argument is further embarrassed by the fact that McMurray did not testify at the co-defendant’s trial either. In this court Illinois suggests that, if severance had not occurred, McMurray might have testified at a joint trial, for his investigation indicated that the co-defendant had no alibi and likely was at the scene of the murder. But then why didn’t McMurray testify at the other trial? And why, if this were the limit of McMurray’s testimony, wоuld Brent have been inclined to cross-examine McMurray at a joint trial? It was known before the disqualification that Rodriguez and his co-defendant had made statements blaming one another for the murder (this is why their trials were severed). McMurray’s testimony about the co-defendant’s alibi either would *673 have exculpated Rodriguez or been neutral toward him. So the state’s explanation for disqualifying Brent doesn’t hang together. While Rodriguez does not accuse the prosecutor of acting in bad faith, that is not an element of the constitutional theory.
Cinching a conclusion thаt' the state court’s decision is unreasonable is that there never was much chance (viewed objectively) that Brent would be placed in a conflicted or compromised position. That is so even if McMurray was likely to testify. Rodriguez had two lawyers and offered to have Brеnt’s co-counsel conduct any cross-examination of McMurray. Illinois does not contend that Brent learned from McMurray privileged information that he could have passed on to co-counsel (if there had been such privileged information,
both
of Rodriguez’s attorneys would have been disqualified). Having co-counsel cross-examine McMurray would have eliminated all risks; and this easy solution (which the state judiciary ignored) makes it unreasonable for the state to have denied Rodriguez the benefit of Brent’s services. A state court can be erroneous without bеing “unreasonable.” See
Yarborough v. Alvarado,
— U.S. -,
Illinois contends that any error is harmless. Rodriguez enjoyed the services of one attorney of his own choosing even after Brent departed, and he does not contend that his trial counsel’s performance at trial was deficient in any respect, let alone constitutionally ineffective. See
Strickland v. Washington,
The Supreme Court has never decided whether prejudice is a prerequisite to reversal of a judgment following erroneous disqualificatiоn of counsel. In
Flanagan v. United States,
The norm in cases on collateral review is that the petitioner must show that the error had a substantial and injurious effect on the outcome. See
Brecht v. Abrahamson,
It is hard to see why violations of the qualified right to counsel of choice should lead to аutomatic reversal, when deprivation of the absolute right to a competent attorney leads to relief only if prejudice is demonstrable. An attorney is vital to a reliable trial. A competent attorney is less so—one must show that, but for counsel’s bungling, the outcome likely would have differed. Having one’s preferred attorney is icing on the cake. A defendant with an inept attorney is in a more precarious position than one with a competent lawyer who is the defendant’s second or third choice. A rule of automatic reversal when the defendant does not get his first-choice lawyer, but requiring proof of prejudice when the defendant does not get even a competent lawyer, would not make sense—particularly in cases such as this where the defendant goes to trial with another lawyer of his own choice. Wе are aware that several circuits have held that the denial of the right to choice of counsel never may be deemed harmless. See
United States v. Panzardi Alvarez,
This leaves unanswered the question posed in Santos—how can a defendant prove prejudice when his substitute counsel provided constitutionally adequate assistance? Maybe the answer is that, when the defendant cannot prove prejudice, the judgment always stands. So the Supreme Court has held even when the right at issue affects the assistance of counsel.
United States v. Morrison,
Some kind of effect must be shown. But
what
kind? Losing the services of one’s preferred lawyer can be similar to receiving the services of a lawyer with a concealed conflict: in either situation trial counsel may well do just fine, but there may be hard-to-uncover shortcomings. The Supreme Court resolved the difficulties facing defendants represented by a conflicted attorney by holding that relief is possible if the conflict “adversely affects” counsel’s performance at trial. See
Mick-ens,
The adverse-effect standard lies midway between automatic reversal and requiring proof of a likely difference in the litigation’s outcome. Until the Supreme Court settles the question it left open in
Richardson-Merrell,
this middle ground seems the closest match to situations in which the court wrongly strips the defendant of his preferred lawyer. Adverse effect in this situation means an identifiable difference in the quality of representation between the disqualified counsel and the attorney who represents the defendant at trial. The difference does not have to be great enough to undermine confidence in the outcome—that is the standard under
Strickland,
see
Has Rodriguez shown an adverse effect? This might be done if Brent had expertise that his other lawyer lacked, or if Brent had planned a line of defense that co-counsel was unable to sustain on his own. So far, however, the record is silent. Brent has not filed an affidavit detailing what he would have done differently had he conducted the trial or how the defense otherwise might have been affeсted by his absence. It is hard to blame Rodriguez for this. Illinois argued for application of the prejudice standard from ineffective-assistance cases, which would have the practical effect of eliminating relief in all *676 cases when the' defendant loses his preferred lawyer, and thus of making Wheat a dead letter. Rodriguez argued for automatic reversal, the opposite extreme. Now that we have settled on a middle ground, both sides should have an opportunity to produce evidence and craft arguments bearing on the applicable standard. See 28 U.S.C. § 2106. Accordingly, the judgment of the district court is vacated, and the case is remanded for proceedings consistent with this opinion. .
