Concurring and Dissenting.
In this capital case, the County of Fresno agreed to pay the fees of the appointed defense attorney. Under the contract, counsel received an $80,000 flat fee that included $60,000 for defense expenses; counsel was allowed to keep any of the money not spent. The contract thus had a built-in incentive for counsel to spend as little as possible on the defense so he could pocket more money. That is what happened here, where counsel spent less than 20 percent of the money allocated for defense expenses, keeping the remaining funds for himself.
One cannot tell from the record whether there were reasonable tactical decisions justifying defense counsel’s small expenditures and minimal efforts to prepare for the guilt phase of trial. Therefore, I join the majority in rejecting defendant’s claim that his right to counsel was violated at the guilt phase; defendant may, of course, raise that claim again in a habeas corpus petition, relying on evidence outside the appellate record.
With respect to the penalty phase, however, the record shows that there was no tactical justification for counsel’s substandard preparations. Consequently, unlike the majority, I am of the view that the fee agreement in question violated defendant’s right to counsel, thus requiring reversal of the sentence of death.
I
Defendant was charged in Fresno County with two counts of murder and four counts of attempted murder, as well as special circumstances that made him eligible for the death penalty. Initially the Fresno County Public Defender represented him, but when the public defender declared a conflict of interest, the trial court appointed Attorney Rudy Petilla to represent defendant.
*458To obtain the appointment, the attorney filled out a form entitled “Proposal Setting Compensation,” in which he requested a fee of $80,000; at the time, that was the amount Fresno County paid for “Category 3” cases—those with multiple victims or defendants, highly unusual publicity, complicated special circumstances, or complex factual or legal issues. The attorney estimated total defense expenses at $60,000 ($40,000 for investigation and $20,000 for expert witnesses), leaving the remaining $20,000 as his fee. The county agreed.
Counsel represented defendant at the guilt and penalty phases of the capital trial. But instead of spending the estimated $60,000 on the defense, he spent less than $9,000, keeping the remaining $71,000 of the $80,000 Fresno County had paid him.
II
Defendant contends the fee agreement between his trial counsel and Fresno County gave rise to a conflict of interest that violated his right to counsel under article I, section 15 of the state Constitution.
“The right to effective assistance of counsel, secured by the Sixth Amendment to the federal Constitution, and article I, section 15 of the California Constitution, includes the right to representation that is free from conflicts of interest.” (People v. Cox (2003) 30 Cal.4th 916, 948 [135 Cal.Rptr.2d 272, 70 P.3d 277].)
“Although the. federal Constitution . . . requires proof of an actual conflict of interest, that is, proof that counsel’s conflict adversely affected his or her performance during the proceedings [citation], under the state Constitution we have required only that the record support an ‘informed speculation’ that a ‘potential conflict of interest’ impaired the defendant’s right to effective assistance of counsel. [Citations.] Because a conflict of interest may retard counsel’s development of evidence or arguments in support of the defense ... we have retained this stricter standard in order to ‘closely guard’ the fundamental right to the assistance of counsel. ‘The very failure to produce or emphasize such information . . . produces a void and results in a record which shields the fact of any possible conflict and makes it difficult to demonstrate on appeal that a conflict did in fact exist. [Citation.] Accordingly, a [defendant] . . . need not establish that there was an actual conflict of interest, but rather it is sufficient if the record provides an adequate basis for an “informed speculation” that there was a potential conflict of interest which prejudicially affected the defendant’s right to effective counsel.’ ” (People v. Rundle (2008) 43 Cal.4th 76, 174-175 [74 Cal.Rptr.3d 454, 180 P.3d 224], italics added, fn. omitted.)
*459“Conflicts of interest may arise in various factual settings. Broadly, they ‘embrace all situations in which an attorney’s loyalty to, or efforts on behalf of, a client are threatened by his responsibilities to another client or a third person or by his own interests.’ ” (People v. Jones (1991) 53 Cal.3d 1115, 1134 [282 Cal.Rptr. 465, 811 P.2d 757], italics added.)
Here, the nature of defense counsel’s fee agreement with Fresno County gave him an incentive to spend as little as possible on the defense. Did this create a potential conflict of interest? I explore this question below.
As a general rule, fee agreements do not violate a defendant’s right to conflict-free counsel. “ ‘[A]lmost any fee arrangement between attorney and client may give rise to a “conflict.” An attorney who received a flat fee in advance would have a “conflicting interest” to dispose of the case as quickly as possible, to the client’s disadvantage; and an attorney employed at a daily or hourly rate would have a “conflicting interest” to drag the case on beyond the point of maximum benefit to the client, [f] The contingent fee contract so common in civil litigation creates a “conflict” when either the attorney or the client needs a quick settlement while the other’s interest would be better served by pressing on in the hope of a greater recovery. The variants of this kind of “conflict” are infinite. Fortunately most attorneys serve their clients honorably despite the opportunity to profit by neglecting or betraying the client’s interest.’ ” (Maxwell v. Superior Court (1982) 30 Cal.3d 606, 618, fn. 8 [180 Cal.Rptr. 177, 639 P.2d 248].)
Occasionally, however, a fee agreement can create a potential conflict of interest. For example, “[conflicts may . . . arise in situations in which an attorney undertakes representation of a defendant in exchange for the literary rights to a portrayal or account based on information relating to the representation.” (People v. Bonin (1989) 47 Cal.3d 808, 836 [254 Cal.Rptr. 298, 765 P.2d 460].) And in In re Gay (1998) 19 Cal.4th 771 [80 Cal.Rptr.2d 765, 968 P.2d 476], Justice Werdegar’s concurring opinion, which I signed, concluded there was a conflict of interest when trial counsel “engineered his appointment in a capital case, doing so by extraordinary, dishonest means, and for the apparent purpose of quickly obtaining a fee while expending as little time and effort on the case as possible.” (Id. at p. 833 (conc. opn. of Werdegar, J.).)
The situation here is one of those rare instances in which the fee agreement has created a “ ‘potential conflict of interest.’ ” (People v. Rundle, supra, 43 Cal.4th at p. 175.) By allowing defense counsel to keep any of the money not spent on defendant’s behalf, Fresno County gave counsel a considerable *460financial incentive to spend as little as possible on defense expenses, so counsel could keep most of the flat fee the county had paid him.1
But this potential conflict of interest does not, by itself, “support an ‘informed speculation’ that. . . [it] impaired the defendant’s right to effective assistance of counsel” (People v. Rundle, supra, 43 Cal.4th at p. 175), because most attorneys abide by their commitment to spend the allocated funds on the defense. Nor is spending only a fraction of the funds on defense expenses in itself a violation of the defendant’s right to conflict-free counsel, because tactically reasonable decisions may underlie that decision. For example, at an early stage of the proceedings, before defense counsel has spent the money allocated for investigation and experts, a defendant may be offered a highly favorable plea bargain, which counsel has reason to believe will be withdrawn if not accepted immediately. In that situation, it makes sense for counsel to advise the defendant to accept the offer; if the defendant does so, counsel has no need to spend the money allocated for investigation and experts.
If there is a plausible explanation here for defense counsel’s failure to spend the money allocated for investigation and expert testimony, or if it cannot be determined from the record whether counsel’s actions were justified, defendant’s claim must be rejected on this direct appeal, although he may still have a remedy on habeas corpus. (See p. 461, fh. 2, post.) As explained below, the record does not show that counsel’s minimal preparation for the guilt phase of trial was tactically unjustified, but there is such a showing with regard to the penalty phase.
A. Guilt Phase
According to defendant, defense investigator Jeff Gunn did only 13.5 hours of guilt phase preparation before counsel announced that the defense was *461ready for trial. But, as the majority points out, investigator Gunn “had originally worked on this case for about two and one-half months as the investigator for the Fresno County Public Defender and had conducted approximately 90 hours of investigation before the public defender declared a conflict.” (Maj. opn., ante, at p. 422.)
Defendant complains that his counsel did not hire a ballistics expert until shortly before commencement of the guilt phase. But the fee agreement’s built-in financial incentive for defense counsel not to hire experts provided no incentive for counsel to delay the retention of experts.
Defendant criticizes his counsel for not hiring an expert to perform a blood analysis or to perform a tire tread analysis. But the evidence presented at trial does not show that a blood analysis was significant to the question of guilt.
Defendant faults his counsel for not conducting DNA testing of evidence obtained from victim Espinoza’s body. But even if that evidence had not contained defendant’s DNA, defendant would not have been exculpated, because Espinoza was a prostitute and was thus likely to have had sex with other men. Also, counsel may have chosen to rush this case to trial in the expectation that the trial would occur before the prosecution’s completion of its own DNA testing; if so, counsel can hardly be faulted for not having any DNA testing done himself.
Because the record on this appeal does not show whether defense counsel’s preparations for the guilt phase were unreasonable, there can be no “informed speculation” (People v. Rundle, supra, 43 Cal.4th at p. 175) that the representation at the guilt phase violated defendant’s right to conflict-free counsel.2 Not so with respect to the penalty phase, however.
B. Penalty Phase
With respect to certain aspects of penalty phase preparation—the hiring of potential expert witnesses and the investigation of potential defense witnesses—counsel here shortchanged defendant by not using the allocated money in a reasonably competent manner.
*462Before trial, counsel budgeted $15,000 for “Background (lifetime) investigation of Defendant for penalty phase social study report” and $10,000 for a “Psychiatrist and Social Worker.” Instead, he paid a total of $4,500 for a defense psychiatrist (who testified at the guilt phase) and a psychologist (who testified at the penalty phase), and he spent no money on preparation of a social study report. There is no legitimate tactical justification for not having such a report done.
Counsel also failed to conduct an adequate investigation of potential character witnesses who could have testified on defendant’s behalf at the penalty phase. At a hearing on a defense motion for a continuance, held three days before the penalty phase was scheduled to begin, counsel told the trial court that defendant had given him a list of 16 potential character witnesses but that he had not contacted any of them. None of them testified at the penalty phase of trial, where the only defense witness was a psychologist. Counsel made the self-serving claim that he deliberately did not interview these witnesses so he could avoid furnishing the prosecution with witness discovery. But this is not a legitimate tactical justification—to determine whether these potential witnesses would be helpful to the defense, they needed to be interviewed, a task that a reasonably competent attorney would ordinarily have assigned to a hired investigator.
To prepare for the penalty phase of a death penalty trial, defense counsel must “conduct a thorough investigation of the defendant’s background.” (Williams v. Taylor (2000) 529 U.S. 362, 396 [146 L.Ed.2d 389, 120 S.Ct. 1495].) “[I]nvestigations into mitigating evidence ‘should comprise efforts to discover all reasonably available mitigating evidence and evidence to rebut any aggravating evidence that may be introduced by the prosecutor.’ ” (Wiggins v. Smith (2003) 539 U.S. 510, 524 [156 L.Ed.2d 471, 123 S.Ct. 2527].) Here, as discussed earlier, defense counsel’s investigation and preparation for the penalty phase of trial was virtually nonexistent, and he spent only a fraction of the funds Fresno County had allocated for the defense. This circumstance supports an “informed speculation” (People v. Rundle, supra, 43 Cal.4th at p. 175) that the defense has been compromised by the attorney’s conduct, resulting in a violation of defendant’s right under the California Constitution to be represented by an attorney unencumbered by a conflict of interest.
Does this violation require reversal of the sentence of death? As explained below, that question is not an easy one to resolve because case law provides no definitive answer on the test to be applied.
*463in
To determine in a particular case whether the right to conflict-free counsel requires reversal under either the state or the federal Constitution, this court has long relied on the high court’s decision in Cuyler v. Sullivan (1980) 446 U.S. 335 [64 L.Ed.2d 333, 100 S.Ct. 1708] (Sullivan). That case held that when a conflict of interest adversely affects defense counsel’s representation, reversal is required without an evaluation of whether the result of trial would have been different had there been no conflict. Put differently, there is a presumption of prejudice. (See Strickland v. Washington (1984) 466 U.S. 668, 692 [80 L.Ed.2d 674, 104 S.Ct. 2052] [Sullivan “held that prejudice is presumed when counsel is burdened by an actual conflict of interest”].)
We explained in People v. Easley (1988) 46 Cal.3d 712, 725 [250 Cal.Rptr. 855, 759 P.2d 490]: “It is important to recognize that ‘adverse effect on counsel’s performance’ under Sullivan ... is not the same as ‘prejudice’ in the sense in which we often use that term. When, for example, we review a ‘traditional’ claim of ineffective assistance of counsel (i.e., one involving asserted inadequate performance as opposed to ‘conflicted’ performance), we require the defendant to show a reasonable probability that the result . . . would have been different. [Citations.] This, however, is not the inquiry called for under Sullivan.” Instead, reversal is required when “counsel ‘pulled his punches,’ i.e., failed to represent the defendant as vigorously as he might have had there been no conflict.” (Ibid.; see also People v. Mroczko (1983) 35 Cal.3d 86, 104, fn. 16 [197 Cal.Rptr. 52, 672 P.2d 835] [“the Supreme Court’s formulation seems to envision an analysis of whether there has been some identifiable prejudice to the right of effective representation, but not an analysis of whether that prejudice affected the outcome of the case”]; Maxwell v. Superior Court, supra, 30 Cal.3d at p. 612 [“When a conviction is attacked validly on the ground that an appointed lawyer was influenced by conflict of interest the appellate court may not ‘ “indulge in nice calculations as to the amount of [resulting] prejudice ....”’ ”].)
In Sullivan, the alleged conflict of interest arose from defense counsel’s multiple concurrent representation; that is, counsel simultaneously represented more than one client. (Sullivan, supra, 446 U.S. at pp. 337-339.) Rather recently, the United States Supreme Court noted in Mickens v. Taylor (2002) 535 U.S. 162, 174 [152 L.Ed.2d 291, 122 S.Ct. 1237] (Mickens), that several federal circuit courts have applied Sullivan’s presumption of prejudice standard in cases in which a defense attorney’s representation implicated “counsel’s personal or financial interests.” But, Mickens pointed out, “the language of Sullivan itself does not clearly establish, or indeed even support, such expansive application.” (Id. at p. 175.) Mickens concluded that “[w]hether Sullivan should be extended to such cases remains, as far as the *464jurisprudence of this Court is concerned, an open question” (id. at p. 176), one that it did not resolve in that case.
Thus, it is not clear what test of prejudice to apply where, as here, the defense attorney’s professional duty to competently represent the client collides with the attorney’s personal or financial interests. I note that after the high court’s decision in Mickens, supra, 535 U.S. 162, this court in People v. Rundle, supra, 43 Cal.4th at pages 168-176, did not apply Sullivan’s presumption of prejudice standard. In Rundle, the attorney’s narrow conflict of interest pertained only to an allegation of jury misconduct, not to any other aspect of the case. Here, by contrast, defense counsel’s conflict of interest adversely affected the entire penalty phase, and it was the government (Fresno County) that, through its fee agreement, created the conflict.
The nature of the fee agreement was such that the less money counsel spent on the defense, the more money he could pocket for himself, thus creating a strong financial incentive to enrich himself at the expense of the defense. Because it was the government’s fee agreement that made this possible, I would apply the standard of presumed prejudice that the high court established in Sullivan, supra, 446 U.S. 335. And I would conclude, based on an “informed speculation” (People v. Rundle, supra, 43 Cal.4th at p. 175) supported by the record, that Fresno County’s fee agreement with counsel did adversely affect his representation at the penalty phase of trial, thus requiring reversal of the judgment of death.
IV
I question the soundness of the majority’s reasoning in rejecting defendant’s conflict of interest claim.
Disapproving decades of this court’s case law, the majority blithely discards a well-established rule, which goes back some 40 years, that a criminal defendant need only show “that the record support an ‘informed speculation’ that a ‘potential conflict of interest’ impaired the defendant’s right to effective assistance of counsel.” (People v. Rundle, supra, 43 Cal.4th at p. 175; see, e.g., People v. Frye (1998) 18 Cal.4th 894, 998 [77 Cal.Rptr.2d 25, 959 P.2d 183]; People v. Mroczko, supra, 35 Cal.3d at p. 105; People v. Cook (1975) 13 Cal.3d 663, 670 [119 Cal.Rptr. 500, 532 P.2d 148]; People v. Chacon (1968) 69 Cal.2d 765, 776, fn. 2 [73 Cal.Rptr. 10, 447 P.2d 106].) The majority’s conclusory explanation is that “a precise definition of our informed speculation concept has proven elusive and the concept has been somewhat variously applied” (maj. opn., ante, at p. 419), and that the standard “is too amorphous to provide meaningful guidance to either the bench or bar” (id. at p. 421). How so? The majority does not say. It simply *465cites six decisions of this court, each containing in brackets a quoted sentence fragment describing, each in a slightly different manner, the informed speculation standard. (Maj. opn., ante, at pp. 419-420.) But those quoted sentence fragments offer no insight into the majority’s reasons for tossing the informed speculation standard on the rubbish heap of discarded legal principles.
This court does not lightly cast aside established principles of law. When it does, it is generally because the old standard has proven unworkable or inequitable, or because it has led to inconsistent results. (See, e.g., In re Lawrence (2008) 44 Cal.4th 1181, 1218, 1220 [82 Cal.Rptr.3d 169, 190 P.3d 535] [abandoning the “minimum elements inquiry” for appellate review of parole decisions because it proved to be “unworkable in practice” and “substantially undermine[d] the rehabilitative goals of the governing statutes”].) Here, the majority makes no attempt to show that the informed speculation standard is unworkable in practice or that it has led to inconsistent results.
The prejudice standard the majority prefers, and adopts, for cases (like this one) in which the conflict of interest does not stem from an attorney’s concurrent representation, is the one that the high court established in Strickland v. Washington, supra, 466 U.S. 668 (Strickland), a case involving incompetent representation. The inquiry under that test is whether there is “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” (Id. at p. 694.) Applying that test here, and assuming for the sake of argument that Fresno County’s fee agreement with defense counsel gave rise to a conflict of interest, the majority concludes that any violation of defendant’s right to counsel arising from that conflict was harmless. (Maj. opn., ante, at pp. 429-430.)
Application of the Strickland test would indeed lead to the majority’s conclusion, for defendant has not shown that he suffered any prejudice from the violation of his right to counsel. This is why: The record does not show what mitigating evidence would have been offered had defendant been represented by an attorney unencumbered by the conflict of interest at issue here.
But, for reasons discussed earlier (see ante, pp. 463-^164), that is not the test I would use here. Because it was the government (Fresno County) that, through the particular fee agreement it had with defense counsel, created the conflict of interest, I would apply the presumption of prejudice standard that the high court established in Sullivan, supra, 446 U.S. 335.
*466Conclusion
For the reasons given above, I would reverse the judgment of death. Werdegar, J., concurred.
Appellant’s petition for a rehearing was denied March 25, 2009. Kennard, J., and Werdegar, J., were of the opinion that the petition should be granted.
Recently, flat fee contracts of this nature were criticized in a report by the California Commission on the Fair Administration of Justice, a group that was chaired by former Attorney General John Van de Kamp, and whose members included the current Attorney General; the District Attorneys of Santa Clara, San Mateo, and Ventura Counties; several law enforcement officers; and members of the criminal defense bar. The report recommended that “legislation be enacted to provide that when Counties contract for indigent defense services in criminal cases, the contract shall provide separate funding for . . . investigators and expert witnesses.” (Cal. Com. on Fair Admin, of Justice, Final Rep. (2008) Professional Responsibility and Accountability of Prosecutors and Defense Lawyers, Rep. on Funding of Defense Services in Cal., p. 13.)
In light of the above-mentioned problems associated with this type of fee agreement, this court should exercise its supervisory powers to prospectively declare fee agreements of this type invalid. (See generally People v. Barboza (1981) 29 Cal.3d 375, 381 [173 Cal.Rptr. 458, 627 P.2d 188] [announcing a “ ‘judicially declared rule of criminal procedure’ ” invalidating fee agreements between counties and public defender’s offices that create a financial incentive for the public defender not to declare a conflict of interest].)
Rejection of such a claim on direct appeal does not foreclose defendant from making a more persuasive showing on habeas corpus, based on matters outside the appellate record. Such a showing may well be possible: Defendant’s habeas corpus petition (pending in this court), alleges that while representing defendant, counsel incurred huge gambling debts; that to pay those debts, counsel borrowed money without intending to repay it, which led the State Bar in 2001 to temporarily suspend his right to practice law; and that in 2004 he resigned from the State Bar with additional charges pending. But these matters are not fully included in the appellate record and therefore may not be considered on this direct appeal.