On August 20, 1999, Kevin T. Hall was charged in Count One of a three-count indictment for conspiracy to distribute marijuana and possession with intent to distribute the drug in violation of 21 U.S.C. §§ 841(a)(1), 846. Hall pleaded guilty on November 22, 1999, and on February 7, 2000, the district court sentenced Hall to 37 months of imprisonment, 2 years of supervised release, and a special assessment of $100. On February 9, 2001, Hall filed a motion to vacate or set aside his conviction, pursuant to 28 U.S.C. § 2255. The district court denied the motion. After timely filing a notice of appeal, Hall filed a motion to proceed in forma pauperis on July 1, 2002. The district court construed the motion as including a request for a certificate of appealability, and on July 23, 2002, granted Hall's motion to proceed in forma pauperis, but denied certificate of appealability.

On December 12, 2002, we granted a certificate of appealability, "as to whether counsel rendered ineffective assistance in violation of the Sixth Amendment due to a conflict of interest, and whether counsel's alleged conflict adversely affected his representation of Hall during pretrial proceedings and the plea process." We now remand for an evidentiary hearing to determine whether there was an actual conflict of interest, which adversely affected the adequacy of his representation.

Kevin Hall is an inmate at the Federal Correctional Institution in Greenville, Illinois ("FCI Greenville"), and was when he was indicted. Assistant Federal Public Defender Lawrence J. Fleming was appointed as Hall's counsel on September 10, 1999. Earlier, Fleming had represented a defendant named Anthony Gignac. Gignac had pleaded guilty and had been sentenced ten days before Fleming began his representation of Hall.

On September 16, 1999, Fleming revealed his representation of Gignac to the Assistant United State's Attorney who was prosecuting Hall. He explained that Gignac had informed prison officials that one Thomas Belwood, one of Hall's co-defendants, was involved in the distribution of marijuana while serving as a corrections officer at FCI Greenville. Fleming expressed his assumption that the reports of drug dealing at FCI Greenville that Gignac had given to FBI agents in connection with his proffer still existed. Fleming concluded by noting his concern that his serving as Gignac's counsel created a conflict of interest in his representation of Hall: "If there is any possibility that Gignac will be a witness in this case, there is an obvious conflict." Br. of Petitioner-Appellant at 3.

The district court set Hall's pretrial conference for November 18, 1999, and a jury trial for December 7, 1999. Each of Hall's co-defendants pleaded guilty by the end of November 1999. During that time, Fleming had been communicating with the government in an effort to reach a plea agreement for Hall. In another letter to the AUSA, dated October 27, 1999, Fleming expressed desire to begin conversation about Hall's case and stated that, "if you will recommend the low end of the Guidelines and agree that there will be no further prosecutions for any offense now known to the Government, I will recommend that Mr. Hall plead guilty pursuant to your standard plea agreement." Br. of Petitioner-Appellant at 4.

With respect to Fleming's conflict of interest, the October 27 letter expressed Fleming's assumption that the AUSA did not perceive any conflict based on Fleming's prior representation of Gignac:

Finally, since I have not received a response to my letter of September 16, 1999, I assume you do not see any conflict on behalf of our office in this case. If you do, please let me know immediately.

Id.

In a letter from Fleming to the AUSA dated November 9, 1999, Fleming stated that he had not received a plea agreement from the AUSA despite the scheduled change of plea hearing on November 16, 1999:

I don't have a plea agreement from you yet, but if you agree with the computations set out in my letter of October 27, 1999 (copy attached) I'd appreciate you getting a draft of a plea agreement to me by Monday, November 15, 1999, so that I have a chance to review it with Mr. Hall before the 16th.

Br. of Petitioner-Appellant at 5.

The November 9 letter also stated that the government had not provided the discovery that Fleming's October 27 letter had requested:

Mr. Hall is still rather adamant about the fact that we still don't have all the discovery, so if you have anything else you can give me, please send that also.

Id.

Fleming concluded the November 9 letter noting his concern that convincing Hall to plead guilty was going to be difficult:

This is going to require a continuing sales job on my part, since my client is fairly well `institutionalized', so please give me whatever you can to answer his arguments.

Id.

Hall pleaded guilty on November 22, 1999 and was sentenced. Hall eventually received a copy of his entire case file where he learned for the first time that Fleming had represented Gignac. In addition, there is no indication in the record that Fleming informed the district court of the potential conflict.

In reviewing the district court's order denying a petitioner's motion to vacate or set aside his conviction pursuant to 28 U.S.C. § 2255, we review all questions of law de novo. Stoia v. United States, 22 F.3d 766, 768 (7th Cir.1994). The district court's factual findings are reviewed for clear error. Id. A § 2255 petitioner is entitled to an evidentiary hearing on his claims, when he alleges facts that, if proven, would entitle him to relief. Id. Hall's § 2255 assertion that he was denied his Sixth Amendment right to effective assistance of counsel is reviewed de novo. Spreitzer v. Peters, 114 F.3d 1435, 1450 (7th Cir.1997), modified in part, 127 F.3d 551 (7th Cir.1997).

Criminal defendants are guaranteed effective assistance of counsel at all stages of the proceedings against them. Cates v. Superintendent, 981 F.2d 949, 952 (7th Cir.1992). This right includes the right to "representation that is free from conflict of interest." Spreitzer, 114 F.3d at 1450 (quoting Wood v. Georgia, 450 U.S. 261, 271, 101 S. Ct. 1097, 67 L. Ed. 2d 220 (1981)).

There are two ways to assert a claim based on counsel's conflict of interest. One, under Strickland v. Washington, 466 U.S. 668, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984), the petitioner may show that his attorney had a potential conflict of interest and that the potential conflict prejudiced his defense; or two, the petitioner may proceed under Cuyler v. Sullivan, 446 U.S. 335, 100 S. Ct. 1708, 64 L. Ed. 2d 333 (1980), where he must establish a violation "by showing that `an actual conflict of interest adversely affected his lawyer's performance.'" Stoia, 22 F.3d at 770 (quoting Sullivan, 446 U.S. at 348, 100 S. Ct. 1708) (emphasis added).

Proceeding under Sullivan places a "lighter burden" on the defendant than Strickland because demonstrating an "adverse effect" is significantly easier than showing "prejudice". Spreitzer, 114 F.3d at 1450; Stoia, 22 F.3d at 771. This more lenient standard applies because the requirements in Sullivan reflect that "prejudice is presumed" if the petitioner establishes both that his counsel had an actual conflict and that the conflict had an adverse effect on counsel's performance. Stoia, 22 F.3d at 770-71 (quoting Strickland, 466 U.S. at 692, 104 S. Ct. 2052).

Under Sullivan, an actual conflict exists if the defense counsel was faced with a choice between advancing his own interests above those of his client. Stoia, 22 F.3d at 771. We have held that an attorney's prior representation of another client leads to an actual conflict when the attorney faces the possibility of having to cross-examine his former client. Enoch v. Gramley, 70 F.3d 1490, 1496 (7th Cir.1995). "A corollary danger is that the lawyer will fail to cross-examine the former client rigorously for fear of revealing or misusing privileged information." Id. Here, Hall's defense counsel had a potential conflict of interest because of his representation of Gignac, who, according to Fleming's account, had provided information about Belwood's drug-distribution activities at FCI Greenville to FBI agents. Because Hall was charged with conspiring with Belwood, Gignac's implication of Belwood pertained to the offenses for which Hall was indicted.

In cases where there is successive representation, we have held that the defendant must show either (1) that the attorney's representation of the first client was "substantially and particularly related to his later representation of defendant," or (2) that the attorney actually "learned particular confidential information during the prior representation of the witness that was relevant to defendant's later case." Enoch, 70 F.3d at 1496-97. Here, Fleming's representation of Gignac enabled him to learn confidential information pertaining directly to Hall's case. Fleming's correspondence with the AUSA reveals his own concerns about such a conflict.

A showing of a conflict does not automatically entitle a petitioner to reversal. The Supreme Court in Mickens v. Taylor, 535 U.S. 162, 122 S. Ct. 1237, 152 L. Ed. 2d 291 (2002), requires that a petitioner also show adverse effect. However, we find no language in Mickens that requires a petitioner to engage in speculation pointing to an actual adverse effect before he has the benefit of an evidentiary hearing. A petitioner demonstrates an adverse effect by showing that there is a reasonable likelihood that his counsel's performance would have been different had there been no conflict of interest. Stoia, 22 F.3d at 771. Pursuant to this requirement, a petitioner who pleaded guilty upon the advice of an attorney with a conflict of interest is not required to demonstrate that he would have decided against pleading guilty had he been represented by a conflict-free attorney. Id. (explaining the distinction between Strickland's "but for" requirement and Sullivan's adverse impact standard). Nor does such a petitioner need to establish that a conflict-free attorney would have advised against pleading guilty. Thomas v. Foltz, 818 F.2d 476, 483 (6th Cir.1987) (explaining that the proper focus should be on whether the defense counsel's conflict affected his actions and the defendant's decision to plead guilty, not whether another attorney without conflict would have made the same recommendation).

As we have held, a defendant can establish ineffective assistance of counsel by showing that his attorney pressured him to plead guilty because of the attorney's conflict of interest. Daniels v. United States, 54 F.3d 290, 295 (7th Cir.1995). Hall argues that Fleming improperly influenced him to plead guilty and rushed him through that process in order to avoid the conflict between his successive representation of Gignac and then Hall. Hall has presented record evidence, in the form of the correspondence between Fleming and the AUSA, which supports this position. Hall produced what an individual in his position could have under these circumstances. It would have been practically impossible for Hall to establish that Fleming would have had trouble examining Gignac or that he would have failed to turn the case over to a colleague. At an evidentiary hearing, Hall can attempt to show adverse effect by establishing that Fleming improperly pressured Hall to accept the plea in order to avoid proceeding to trial. This is consistent with the requirement of Sullivan that a defendant show "that conflict of interest actually affected the adequacy of his representation." Sullivan, 446 U.S. at 348, 100 S. Ct. 1708. If Hall can show at a hearing that counsel neglected preparation for trial while pressuring Hall to take a plea because of a conflicting interest, he will have established a conflict that affected the adequacy of representation.

While Mickens contains dicta that can be read to cast doubt on the application of the Sullivan framework in cases of successive representation, see Mickens, 535 U.S. at 175-76, 122 S. Ct. 1237, the facts of this case are distinguishable. The conflict here involved successive representations that were just ten days apart and were also closely interrelated. It is unclear whether Mickens would be applicable to such a situation. In any case, Enoch, applying Sullivan to cases of successive representation, remains controlling in this circuit and we see no reason to ignore it here. Cf. Moss v. United States, 323 F.3d 445, 459-63 (6th Cir.2003) (explaining the effect of Mickens in cases of successive representation and ultimately applying Sullivan because the likelihood of prejudice under the circumstances exceeded that of a "traditional" successive representation).

In Rosenwald v. United States, cited in the dissent, we stated that the fact that Rosenwald did not go to trial did not undermine his allegations of prejudice. 898 F.2d 585, 587 (7th Cir.1990). We noted that even if the attorney took steps in advance of trial to avoid having to cross-examine a witness, the client was nonetheless the victim of his attorney's divided loyalties. In that case, as in this one, the record did not conclusively establish that his attorney's representation was unaffected by his conflict of interest and we remanded for an evidentiary hearing.

In addition, Rosenwald was a party to a civil case. We stated, "The pragmatic pressure on counsel in cases such as these is purely financial — the lawyer does not want to lose a client whether that client is seeking advice on civil or criminal matters." Id. We mentioned the attorney's financial interest to illustrate that the attorney's conflict in a civil case does not lessen the ethical quagmire merely because it does not possess the "seriousness" of a criminal proceeding. We said, "the ethical dilemma is also the same — the attorney must still guard secrets and confidences and must seek to promote the client's interests whether the client is being represented in a civil or a criminal matter." Id.; See ABA Model Code of Professional Responsibility Conduct Rule 1.6(a) (confidentiality), 1.7(b) (conflict of interest). Rosenwald does not stand for the notion that an attorney's interest is only financial nor does it suggest that the petitioner must show exactly what his attorney's motivation was when the attorney proceeded despite the conflict. More importantly, Rosenwald drives home the point that secrets between an attorney and his client are sacred (as were those between Fleming and Gignac) and that a conflict of interest can taint an attorney's decisions before, during and after trial.

Also the dissent states that Hall has not shown adverse effect. We believe Hall has put forth record evidence to show adverse effect. He has certainly put forth all someone in his position could without the benefit of an evidentiary hearing. The facts alleged are enough to warrant an evidentiary hearing; they show that Fleming was laboring under an actual conflict of interest and may have induced Hall to plead guilty in an effort to avoid the conflict. Hall was unaware of the conflict and he was entitled to have all the relevant information his attorney had in which to assist in his own defense. This includes both possible information Fleming had with regard to Gignac as well as the knowledge of the conflict itself.

Under Sullivan, Hall need not show prejudice if he can show an actual conflict existed that had an adverse effect on his representation. We think, given a chance, that Hall may uncover evidence at a hearing that would show an adverse effect. Such a hearing may include testimony from Fleming, the AUSA, Hall himself, etc.

This does not ignore the holding in Mickens, if Hall can prove at a hearing that Fleming's conflict adversely affected his representation then Mickens says he is entitled to relief. Stoia and Enoch are not disturbed by the holding in Mickens. In Stoia we held that a petitioner shows adverse effect by showing there is a reasonable likelihood that the attorney would have acted differently without a conflict. Stoia v. United States, 22 F.3d 766, 768 (7th Cir.1994). In that case, after an evidentiary hearing, we denied relief because the adverse effect was not shown. If at the hearing it is found that Hall cannot show adverse effect, then he must proceed under the ordinary Strickland standard of prejudice where he must show that but for counsel's errors, the outcome of the proceeding would have been different.

Hall is entitled to an evidentiary hearing to determine whether there was an actual conflict of interest that produced an adverse effect.

We REVERSE and REMAND to the district court for that purpose.

EASTERBROOK, Circuit Judge, dissenting.

Kevin Hall pleaded guilty to a drug conspiracy. Now he wants collateral relief on the ground that his lawyer formerly represented Anthony Gignac, who might have been called as a witness had a trial occurred, and therefore could not represent him adequately. The conclusion that these events call into question the validity of the plea is incompatible with Mickens v. Taylor, 535 U.S. 162, 122 S. Ct. 1237, 152 L. Ed. 2d 291 (2002), and would be hard to justify even under pre-Mickens decisions.

After Hall was indicted, the Federal Defender's office for the Southern District of Illinois assigned Lawrence Fleming to represent him. Fleming previously had furnished legal assistance to Gignac, and given the posture of these proceedings I must assume that Gignac might have been a witness had a trial been held — though the United States Attorney denies that Gignac would have been called, and he should know. (The prosecutor says that Gignac knows nothing material to the charge against Hall.) Hall's theory is that Fleming strong-armed him to plead guilty in order to avoid the problem that would ensue if Gignac took the stand. This is implausible on its own terms. Why should Fleming have cared? The Federal Defender in the Southern District has five other lawyers on the staff. If the case had gone to trial, and the prosecutor had shown any inclination to call Gignac, Fleming could have turned the representation over to one of his colleagues. Hall proceeds as if Fleming had received a hefty fee that he feared having to return if required to withdraw. See Rosenwald v. United States, 898 F.2d 585, 587 (7th Cir.1990) ("The pragmatic pressure on counsel in cases such as these is purely financial; the lawyer does not want to lose a client."). Yet Fleming, as a public employee, had no financial stake and thus no reason other than his client's best interests to urge Hall to plead guilty.

Hall received a sentence of 37 months' imprisonment, the bottom of the guideline range (itself reduced by the acceptance of responsibility that Hall now wants to repudiate). Five defendants were charged in the indictment. The others — all represented by conflict-free counsel — likewise pleaded guilty. Two of these four entered pleas in advance of Hall's and were available to testify against him; the remaining two pleaded guilty at the same time Hall did. If the four defendants with non-conflicted counsel had gone to trial (better, had gone to trial and won), while only Hall had capitulated, one might see a point; but given what actually happened Hall's plea is best seen as a good deal in light of mounting evidence.^†

My colleagues say, however, that a conflict of interest not only excuses any need to show prejudice to the defense — in other words, a "reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different", Strickland v. Washington, 466 U.S. 668, 694, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984); see also Woodford v. Visciotti, 537 U.S. 19, 123 S. Ct. 357, 154 L. Ed. 2d 279 (2002) — but also eliminates any need to establish adverse effect. They rely on Enoch v. Gramley, 70 F.3d 1490 (7th Cir.1995), and Stoia v. United States, 22 F.3d 766 (7th Cir.1994), which read Cuyler v. Sullivan, 446 U.S. 335, 100 S. Ct. 1708, 64 L. Ed. 2d 333 (1980), to require collateral relief whenever counsel had a conflict of interest that created a potential for an effect on the conduct of the defense.

Mickens rejects that understanding of Sullivan and its predecessor Holloway v. Arkansas, 435 U.S. 475, 98 S. Ct. 1173, 55 L. Ed. 2d 426 (1978). The Court concluded in Mickens that those decisions deal with situations in which the trial judge should have inquired into the conflict (as when one lawyer represents two defendants with inconsistent positions) or the judge rebuffs a protest by a lawyer assigned to represent multiple defendants. When no inquiry was required at trial, Mickens holds, then unless the petitioner establishes an adverse effect from the conflict no collateral relief is appropriate. What my colleagues have to say about Mickens is this:

A showing of a conflict does not automatically entitle a petitioner to reversal. The Supreme Court in Mickens v. Taylor, 535 U.S. 162, 122 S. Ct. 1237, 152 L. Ed. 2d 291 (2002), requires that a petitioner also show adverse effect. However, we find no language in Mickens that requires a petitioner to engage in speculation pointing to an actual adverse effect. A petitioner demonstrates an adverse effect by showing that there is a reasonable likelihood that his counsel's performance would have been different had there been no conflict of interest. Stoia, 22 F.3d at 771.

371 F.3d at 974 (emphasis in original). By saying that the need to show adverse effect does not entail a need to show "actual adverse effect" — and adding that a "reasonable likelihood" of any effect on counsel's performance will do — my colleagues depart from Mickens. Any conflict creates (a reasonable probability of) some effect. So does the fact that the lawyer caught a cold. That is not a sensible understanding of the sixth amendment's right to the assistance of counsel. Effects are ubiquitous. A koala's choice among tasty eucalyptus leaves in Australia could change the weather in Alaska. (This is the butterfly effect of chaos theory.) But we need to ask whether a given effect was both likely and injurious. Only someone who lacks "counsel" within the meaning of the Constitution is entitled to relief. When the Justices spoke of "adverse effect" they meant a real rather than a hypothetical adverse effect (or, as my colleagues would have it, no adverse effect at all). That's what the debate was about.

The sixth circuit concluded that Mickens failed to show an adverse effect from the conflict; the Supreme Court held that this was enough to warrant the denial of the petition when there had been no reason for the trial court to make an inquiry at the time. The reason inquiry was unnecessary was that counsel did not represent multiple defendants. Instead he had represented Mickens's victim. So there was a potential for a problem if the trial led toward information that counsel held in confidence from the former representation. But there was no reason for the judge to know this, and when Mickens failed to establish a concrete adverse effect he lost. Exactly the same may be said about Hall: the representation was sequential, there was no occasion for the judge to inquire, and so "adverse effect" must be shown.

Four Justices dissented. Two of them (Ginsburg and Breyer) did so principally because Mickens was a capital case, and they favor a special standard for such prosecutions. Justice Souter stated in dissent that he supports the "adverse effect" test for situations in which the trial judge bears no fault, and his opinion contemplates a demonstrable adverse effect. He wrote that "a defendant can fairly be saddled with the characteristically difficult burden of proving adverse effects of conflicted decisions after the fact when the judicial system was not to blame in tolerating the risk of conflict" (535 U.S. at 203, 122 S. Ct. 1237) — and to say that a defendant must prove something is to say that the adverse effect must be realized (= actual) rather than potential. Only Justice Stevens thought that conflict plus any effect on performance should be enough. See 535 U.S. at 186-87, 122 S. Ct. 1237. He suggested, however, that proof of adverse effect should be required when the trial judge lacked knowledge of the conflict. Id. at 187 & n. 12, 122 S. Ct. 1237. Why should this circuit adhere to a position that has the support of one Justice at best?

One more observation about "actual" adverse effect. Although the phrase "actual adverse effect" does not appear in Mickens, the word does appear in Sullivan (in a passage quoted by the Mickens majority, 535 U.S. at 171, 122 S. Ct. 1237). In Sullivan the Court concluded that it would be enough to prove that the conflict "actually affected the adequacy of [the] representation". 446 U.S. at 349-50, 100 S. Ct. 1708. That showing would suffice for dual-representation situations in lieu of Strickland prejudice. In the seventh circuit, however, no "actual" effect on counsel's adequacy need be shown. My colleagues require "a reasonable likelihood that ... counsel's performance would have been different" but not that this "difference" undermined the defense or had any other adverse consequence. To equate "difference" with "adverse effect" is to elevate pre-Mickens decisions of this circuit over recent instructions from our superiors.

Part III of Mickens, 535 U.S. at 174-76, 122 S. Ct. 1237, observes that courts of appeals have extended Sullivan beyond simultaneous-representation cases to conflicts of all sorts. The Justices remarked that they had followed this approach arguendo in Mickens (a sequential-representation situation), and applied the "adverse effect" standard rather than insisting on proof of "prejudice," because the question presented by the petition was so limited. But Mickens shows that the Court is skeptical about this extension and would welcome an opportunity to decide whether Holloway and Sullivan apply beyond simultaneous-representation situations. My colleagues call Part III of Mickens dictum, and technically it is, but together with the rest of the opinion it tells us that a majority of the Justices believe that petitioners should be required to prove real adverse effects if not Strickland "prejudice" in most conflict situations. Instead of fobbing Part III off as dicta to be ignored, we should use the discussion to illuminate the rest of the opinion. It shows that the Court's references to "adverse effect" must be taken seriously rather than dismissed because the Court did not prepend the word "actual." It ill behooves a court of appeals to say: "We see what you are driving at, but until you use the magic word we shall do the opposite."