dissenting.
The Supreme Court has held that in order to establish ineffective assistance of counsel based on multiple representation, “a defendant who raised no objection at trial must demonstrate that an actual conflict of interest adversely affected his lawyer’s performance.” Cuyler v. Sullivan, 446 U.S. 335, 348, 100 S.Ct. 1708, 1718, 64 L.Ed.2d 333 (1980). It is not enough for the defendant to prove the existence of a possible or potential conflict of interest. Id. This court has held that where a defendant pursues a common defense rather than one adverse to his code-fendant, the defendant, to establish more than merely a potential conflict, “must show specific instances where [his] attorney could have, and would have, done something different if that attorney had represented only [the] defendant.” United States v. Cirrincione, 780 F.2d 620, 630-31 (7th Cir.1985) (emphasis added). The majority applies these standards and decides that attorney Marvin Goldenhersh’s representation of Lee Griffin violated the Sixth Amendment. I conclude, however, that Griffin has failed to demonstrate anything more than a possible conflict and therefore respectfully dissent from the majority’s decision.
The relevant issue in this case boils down to whether an attorney representing Griffin alone would have presented the innocent bystander defense instead of the alibi defense at trial. Much of the majority’s analysis, however, focuses not on the choice of defenses but rather on the pre-trial discussions that Goldenhersh had with Griffin regarding his defense. Relying completely on the absence of evidence in the record, the majority concludes that no consideration or discussion of possible defenses occurred. Yet it is the defendant’s burden to prove that his attorney had an actual conflict of interest in advising him. See, e.g., Cuyler, 446 U.S. at 348, 100 S.Ct. at 1718; Cirrincione, 780 F.2d at 630-31 n. 4. Accordingly, “[t]he mere possibility of divided loyalties is insufficient to sustain the defendant’s burden.” Id. The district court did not find that Goldenhersh failed to examine with Griffin the weaknesses of the alibi defense and the possibility of presenting an innocent bystander defense. The record on this point is indeed inconclusive.1 I fear the majority treads on dangerous ground by inferring the absence of specific discussions from an indeterminate record developed over ten years after the fact.
The record does make clear that Griffin told his previous attorney, Ralph Derango, that he was an innocent bystander, that Der-ango attempted to negotiate a agreement with the police where Griffin would testify against co-defendant Jimmy Smith, and that the police weren’t buying Griffin’s story. Confronted with this situation, Griffin’s family decided to retain Goldenhersh and discharge Derañgo. And, although the exact nature of the conversations that took place between Griffin and Goldenhersh regarding Griffin’s defense is unclear, we do know that, at trial, Goldenhersh presented the alibi defense, and Griffin testified under oath that he and Smith were at a tavern playing cards when the murders occurred. Thus, the facts of this case are analogous to those in United *892States ex tel. Cole v. Lane, 752 F.2d 1210 (7th Cir.1985), where the habeas petitioner abandoned a defense of accusing his eo-de-fendant and testified under oath that neither he nor his co-defendant was involved in the crime. In Cole, we stated:
[W]hen Cole was presented with an opportunity to accuse Cade of committing the armed robbery, Cole denied under oath telling the arresting officers that Cade had committed the crime. It is logical to assume from the record in this case, rather than demonstrating an actual conflict of interest, that in all probability the defendants agreed prior to trial upon a trial strategy of neither defendant testifying against the other. Because there is no evidence in the record that Cole ever pursued the defense of accusing Cade of committing the armed robbery and, to the contrary, disclaimed the defense under oath, we hold that Cole abandoned the defense of accusing Cade. Decisions to pursue or abandon defenses are tactical choices of trial strategy.... Since the decision to abandon the defense of accusing Cade was an obvious strategical and tactical decision, it may not be reviewed by this court on appeal.
752 F.2d at 1220. Where a defendant testifies under oath to the truth of a particular defense, we therefore presume that the defendant made a tactical decision to present the defense consistent with his testimony. Such a tactical decision cannot provide the foundation for a claim of actual conflict of interest. Id. Of course, the defendant can overcome this presumption of trial strategy by showing that an unconflicted counsel would have pursued a different course of action. See, e.g., Cirrincione, 780 F.2d at 631.
The evidence against Griffin left Golden-hersh with two options: he could have presented the alibi defense or the innocent bystander defense. The majority attempts to distinguish Cole by concluding that an uncon-flicted counsel representing Griffin would have presented a defense based on inconsistencies in the testimony identifying him as a shooter. Yet the majority fails to concretely evaluate the relative strengths of the two potential defenses that were available to Griffin. Only by comparing the strengths and weaknesses of the two defenses can we decide whether an unconflicted counsel would have presented the innocent bystander defense rather than the alibi defense.2 Certainly, the evidence supporting the alibi defense was weak at best. The alibi was only partially corroborated by Priscilla Smith, and extensive evidence placed both Griffin and Jimmy Smith at the scene of the crime. Furthermore, the defense had the disadvantage of connecting Griffin’s story to that of Jimmy Smith, who was consistently identified by witnesses as a shooter. This is not to say that the alibi defense lacked any advantages. We have noted that “Cjjoint representation is a means of insuring against reciprocal recrimination. A common defense often gives strength against a common attack.” Cole, 752 F.2d at 1220 (quoting Holloway v. Arkansas, 435 U.S. 475, 482-83, 98 S.Ct. 1173, 1177-78, 55 L.Ed.2d 426 (1978)). Although it may have been unlikely that Jimmy Smith would have testified against Griffin if they had been afforded separate trials, by presenting the alibi defense Goldenhersh eliminated this possibility entirely.
The majority posits that Griffin could have pursued a defense based on the inconsistencies identifying him as a shooter. I agree with the majority’s assessment that the evidence at trial did not conclusively demonstrate that Griffin was a shooter. Yet the prosecution’s ease against Griffin was broader, implicating him as at least an accomplice to the murders. Both Robinson and Kellick clearly placed Griffin at the scene of the crime in a group of three men who committed the murders. To respond to this evidence, Goldenhersh could have presented the innocent bystander defense — Griffin could have taken the stand and testified that, al*893though he was at the scene of the crime, he had nothing to do with the murders. Accomplice liability in Illinois is broad, however, and this defense would not have provided much hope for Griffin. Under Illinois law, a person is legally accountable for the conduct of another when “[e]ither before or during the commission of an offense, and with the intent to promote or facilitate such commission, he solicits, aids, abets, agrees or attempts to aid, such other person in the planning or commission of the offense.” 720 ILCS 5/5-2. Although a person’s mere presence at the scene of a crime is insufficient to impose criminal liability, active participation in a crime is not necessary for liability to be based on a theory of accountability. People v. Davis, 254 Ill.App.3d 651, 193 Ill.Dec. 636, 651, 626 N.E.2d 1187, 1202 (1993). The Illinois Supreme Court has stated:
If the proof shows that a person was present at the commission of the crime without disapproving or opposing it, it is competent for the trier of fact to consider this conduct in connection with other circumstances and thereby reach a conclusion that such person assented to the commission of the crime, lent to it his countenance and approval and was thereby aiding and abetting the crime. Stated differently, circumstances may show there is a common design to do an unlawful act to which all assent, and whatever is done in furtherance of the design is the act of all, making each person guilty of the crime.
People v. Morgan, 67 Ill.2d 1, 7 Ill.Dec. 69, 73, 364 N.E.2d 56, 60 (1977) (quoting People v. Washington, 26 Ill.2d 207, 186 N.E.2d 259, 261 (1962)). Other factors that Illinois courts have used to determine whether a defendant can be liable on an accountability theory include his association with the co-defendants (e.g., common arrival at the scene), flight from the scene, continued association with the perpetrators after the criminal act, and failure to report the incident. See Morgan, 7 Ill.Dec. at 73, 364 N.E.2d at 60; Davis, 193 Ill.Dec. at 651, 626 N.E.2d at 1202; People v. Jones, 12 Ill.App.3d 643, 299 N.E.2d 77 (1973) (holding defendant accountable for murder committed by co-defendant based on evidence that they arrived at scene together and fled together after murder).
In light of the law of accountability and the evidence against Griffin, the innocent bystander defense was far from promising. Robinson testified that Griffin and another man entered the apartment during the first set of murders, then drove away in Griffin’s car, and returned to the apartment together immediately before Walker was murdered. These facts, combined with no evidence (apart from Griffin’s presumed testimony) showing that Griffin attempted to disassociate himself from the crime or call the police, would have been severely damaging to a claim by Griffin that he was merely an innocent bystander. See, e.g., Morgan, 7 Ill.Dec. at 73, 364 N.E.2d at 60; Davis, 193 Ill.Dec. at 641, 626 N.E.2d at 1202 Jones, 12 Ill.App.3d 643, 299 N.E.2d 77. Moreover, the testimony identifying Griffin as one of the shooters would only have increased the likelihood of the jury finding him guilty of murder. See, e.g., Davis, 193 Ill.Dec. at 641, 626 N.E.2d at 1202 (finding some evidence that defendant may have actually shot at victim supported guilty verdict based on accountability theory).
Comparing the likelihood of success of the two defenses that were available to Golden-hersh, I cannot conclude that an unconflieted counsel would have presented-the innocent bystander defense. Given the evidence in this ease, Griffin was fighting an uphill, and ultimately losing, battle. Although the probability of a successful alibi defense appears close to nil, the likelihood of a successful innocent bystander defense was not significantly greater.3 Griffin has therefore failed to demonstrate specific instances where an unconflieted counsel representing him would have acted differently.4 Griffin cannot sus*894tain his burden of demonstrating an actual conflict by showing that it was possible that Goldenhersh’s-choice of the alibi defense was a product of divided loyalties. See, e.g., Cirrincione, 780 F.2d at 630-31 n. 4. For that reason, I have determined that Griffin’s Petition for Writ of Habeas Corpus should not be granted.
. As the majority points out, Goldenhersh's memory was less than clear regarding the extent to which Goldenhersh discussed the innocent bystander defense with Griffin. Yet Griffin’s petition for habeas relief alleged that Goldenhersh erroneously advised him that the innocent bystander defense would fail, not that Goldenhersh neglected to discuss the defense with him. Moreover, the record is silent as to whether Goldenhersh explored with Griffin the dangers of linking his defense to Jimmy Smith’s.
. Much of the majority’s opinion focuses on the relative strength of the cases against Griffin and Jimmy Smith. The fact that the case against Jimmy Smith was stronger than that against Griffin is relevant to evaluating the risks of linking Griffin’s defense to that of Smith. It does not, however, directly determine the ultimate issue of whether an unconflicted counsel would have presented the innocent bystander defense.
. The majority distinguishes Cole as a case in which “the avenues of inquiry closed off by the joint representation were of little consequence.” Maj. Op. at 889. Yet there is no reason to believe that Griffin’s innocent bystander defense would have had a better chance of success than the potential defense that was foreclosed in Cole.
. To support its finding of conflict, the majority also points to Goldenhersh’s use of the scapegoat argument and his confusion of the defendants at trial. Although the scapegoat argument may have been unhelpful, and even prejudicial, to the defendants, it appears to have been the result of *894bad trial strategy rather than any conflict of interest. Given the weakness of both of Griffin’s potential defenses, Goldenhersh very likely would have made the same argument regardless of the defense chosen. Similarly, Goldenhersh's confusion of the two defendants and the evidence against each of them may exhibit incompetence, but it does not demonstrate a conflict of interest. Griffin has not made a claim of ineffective assistance of counsel under Strickland v. Washington, 466 U.S. 668, 687, 104 S.Ct. 2052, 2064, 80 L.Ed.2d 674 (1984), which provides relief for defendants who can demonstrate that their counsel’s deficient performance prejudiced their defense.