**241 *728 *1057 A jury convicted petitioner Jarvis J. Masters of the first degree murder of Sergeant Dean Burchfield, a correctional officer at San Quentin State Prison ( Pen. Code, §§ 187, subd. (a), 189 ; further undesignated statutory references are to the Penal Code), and conspiracy to commit murder and to commit assault on correctional staff (§§ 182, 4501). The jury found true the special circumstance allegation that the murder involved the knowing and intentional killing of a peace officer engaged in the performance of his duties (§ 190.2, subd. (a)(7)). The jury returned a verdict of death, and the trial court sentenced Masters to death for the murder and to life with the possibility of parole for the conspiracy. On direct appeal, we affirmed Masters's convictions and sentence. ( People v. Masters (2016) 62 Cal.4th 1019 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 ( Masters ).)

Masters's codefendants, Andre Johnson and Lawrence Woodard, also were convicted of Burchfield's murder, and they were sentenced to life imprisonment without possibility of parole. The Court of Appeal affirmed their convictions and sentences. ( People v. Johnson (1993) 19 Cal.App.4th 778 , 23 Cal.Rptr.2d 703 .)

*1058 In 2005, while his appeal was pending, Masters filed a petition in this court seeking a writ of habeas corpus. Having found the petition stated a prima facie case for relief on several claims, we issued an order to show cause why relief should not be granted on a subset of the claims raised. After considering the return and traverse, we appointed a referee, whom we directed to take evidence and make certain findings of fact. Following an evidentiary hearing, the referee filed a report with this court, and the parties filed their exceptions to it.

We accept most of the referee's report and findings as supported by substantial evidence and discharge the order to show cause.

I. BACKGROUND

Many of the facts of the crime and proceedings in the trial court that are relevant to Masters's petition for writ of habeas corpus are set forth in our opinion on appeal. ( Masters , supra , 62 Cal.4th at pp. 1026-1041, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) We summarize those facts here.

Masters, Woodard, and other members of the Black Guerilla Family (BGF) gang were housed in the Carson section of San Quentin State Prison. According to BGF member Rufus Willis, the prosecution's main witness, Masters suggested to him, Woodard, and other BGF members that they attack prison guards. Masters, Woodard, and others decided that Sergeant Dean Burchfield would be the first target of the plot and that Johnson was to stab him with a prisoner-made weapon. Masters was to obtain a piece of metal from another BGF member, sharpen it, and pass it to Johnson. Masters also was to arrange for an inmate to signal when Burchfield was approaching the second tier of cells. Johnson, who was housed on the second tier, was to stab Burchfield when he came to Johnson's cell. After the assault, Johnson was to pass the weapon to another BGF member, who would dispose of it.

*729 On June 8, 1985, Burchfield was stabbed outside Johnson's cell during his nightly rounds; he later died of a single chest wound. At the time of the assault, Masters was housed on the fourth tier of cells.

After the murder occurred, Willis tried to contact prison officials with an offer to provide information in exchange for release from prison. Charles Numark, an investigator from the Marin County District Attorney's Office, initially suggested that Willis would be released from prison if he cooperated with the investigation. Specifically, Numark offered to help Willis secure release on parole if he testified. But the prosecutors, Deputy District Attorneys Edward Berberian and Paula Kamena, told Willis he would not be released from prison if he testified against Masters. Rather, in exchange for *1059 Willis's testimony, they offered to notify the parole board of his assistance; told Willis he would be granted immunity for the crimes he had **242 committed in prison, including his participation in Burchfield's murder; and said he would be moved to an out-of-state prison for his protection. Willis accepted the offer.

Willis gave prison officials several handwritten notes concerning the murder. Willis testified, and a handwriting expert confirmed, that at least some of the notes were in Masters's handwriting. Willis also asked Masters to write a report about the murder, which he apparently did. Masters's report implicated himself, Johnson, and others in Burchfield's murder. In a series of notes written to Willis, Johnson implicated himself in the murder.

At trial, Masters attacked Willis's testimony. While in prison, Willis had committed and ordered the stabbings of several inmates, distributed illegal drugs, and extorted prison staff. Masters also presented evidence suggesting that Willis was angry with the BGF and had planned the attack on Burchfield but set up BGF members to be blamed for the murder.

Another BGF member, Bobby Evans, also testified against Johnson, Masters, and Woodard. After Burchfield was murdered, Johnson, Masters, and Woodard were transferred to the Adjustment Center, the section of the prison where Evans was housed. According to Evans, each of them separately told Evans about their respective actions in the assault. Those accounts generally were consistent with Willis's testimony. Specifically, Evans testified that Masters had been transferred to the Adjustment Center "around" August 1985 and told him "around" September that he voted in favor of killing Burchfield.

At trial, Masters attacked Evans's credibility by presenting evidence of his extensive criminal history. Evans had been convicted of four burglaries and an attempted robbery, had stabbed numerous inmates, and had supervised the BGF's street crimes. Masters also presented evidence that Evans was testifying for the prosecution to reduce his sentence in his own criminal proceedings. During the guilt phase, Evans admitted that while on parole, he had pleaded guilty to attempted robbery and was awaiting sentencing in Alameda County. During the penalty phase, Evans testified that he had been hired on several occasions to shoot people and had shot six people, though none died.

Unlike Willis, however, Evans did not testify under a grant of immunity. Rather, after Evans had pleaded guilty, he contacted James Hahn, a parole agent for the then-Department of Corrections (now Department of Corrections and Rehabilitation), and offered to disclose information in exchange for protection from the BGF. Hahn made no guarantees but said he might be able *1060 to do a favor for Evans "sometime down the line." According to Evans, before Masters's trial, he had provided *730 information to Hahn on only one other occasion.

Evans testified that in an effort to reduce the amount of time he would serve in state prison, he wanted to spend as much time as possible before sentencing in local custody. Evans's sentencing hearing was repeatedly postponed. During the guilt phase deliberations, Masters learned that after Evans testified, Evans had been granted probation at his sentencing hearing.

Hahn testified about actions he took on Evans's behalf. He spoke to Evans around 10 times between June 1989 and October 30, 1989, the date Evans first testified in Masters's case. In June, Evans told Hahn that he was facing prison time after pleading guilty in Alameda County to attempted robbery and that he did not want to return to prison because the BGF had threatened to kill him. Hahn told Evans he would "take care" of Evans's safety and security but could not make any promises regarding Evans's Alameda County case or that Evans would receive any benefit for providing information. Hahn also told Evans that if he did have to return to prison, Hahn would try to arrange it so Evans could serve his sentence in another state. Hahn also said he would try to place Evans and his family in a witness relocation program. Alameda County prosecutors testified that at Hahn's behest they twice requested that Evans's sentencing hearing be postponed.

During the penalty phase, the jury learned that an inmate, David Jackson, was stabbed to death on an exercise yard at San Quentin.

**243 According to Johnnie Hoze, a BGF member, Masters told him that Masters stabbed Jackson.

II. HABEAS CORPUS PROCEEDINGS

Masters filed a petition for writ of habeas corpus raising numerous issues. We found the petition stated a prima facie case for relief as to some claims that asserted his innocence or challenged the veracity of the evidence presented at trial, and we ordered the Director of Corrections and Rehabilitation (now the Secretary of the Department of Corrections and Rehabilitation) to show cause why relief should not be granted because "(1) material false evidence was admitted at the guilt phase of his trial; (2) newly discovered evidence casts fundamental doubt on the prosecution's guilt-phase case; (3) [Masters]'s trial was fundamentally unfair because prosecution witness Rufus [Willis's] testimony was unreliable due to improper coercion by the prosecution[;] (4) the prosecution violated Brady v. Maryland (1963) 373 U.S. 83 , 83 S.Ct. 1194 , 10 L.Ed.2d 215 by failing to disclose the promises of leniency to prosecution witness Bobby Evans and other facts bearing on [Evans's] credibility that *1061 have come to light after the judgment was imposed[;] (5) the prosecution knowingly presented the false testimony of Bobby Evans[;] (6) [Masters]'s trial was fundamentally unfair because Bobby [Evans's] testimony was unreliable due to improper coercion by the prosecution[;] (7) material false evidence - the testimony of [Johnnie] Hoze - was admitted at the penalty phase regarding [Masters]'s participation in the murder of David Jackson; and (8) newly discovered evidence regarding Hoze's testimony casts fundamental doubt on the accuracy and reliability of the penalty-phase proceedings ...."

After considering the return and traverse, we appointed a referee to answer these questions: (1) "Was false evidence regarding [Masters]'s role in the charged offenses admitted at the guilt phase of [Masters]'s trial? If so, what was that evidence?"; (2) "Is there newly discovered, credible evidence indicative of [Masters]'s not having been a participant in the *731 charged offenses? If so, what is that evidence?"; (3) "What, if any, promises or threats were made to guilt phase prosecution witness Rufus Willis by District Attorney Investigator Charles Numark or Deputy District Attorneys Edward Berberian or Paula Kamena? Was Willis's trial testimony affected by any such promises or threats, and, if so, how?"; (4) "Were there promises, threats or facts concerning guilt phase prosecution witness Bobby Evans's relationship with law enforcement agencies of which Deputy District Attorneys Berberian and Kamena were, or should have been, aware, but that were not disclosed to the defense? If so, what are those promises, threats or facts?"; (5) "Did Deputy District Attorneys Berberian and Kamena knowingly present false testimony by Bobby Evans? If so, what was that testimony?"; (6) "What, if any, promises or threats were made to Bobby Evans by District Attorney Investigator Numark, Department of Corrections Investigator James Hahn, or Deputy District Attorneys Berberian and Kamena? Was Evans's trial testimony affected by any such promises or threats, and, if so, how?"; (7) "Did penalty phase prosecution witness [Johnnie] Hoze provide false testimony regarding [Masters]'s involvement in the murder of inmate David Jackson? If so, what was that false testimony?"

At the reference hearing, several BGF members testified, and other BGF members' statements were introduced into evidence. One of the prosecutors from Masters's trial and various law enforcement officials testified, as did Masters's investigators and trial counsel. Two additional witnesses testified as experts.

Two former San Francisco police officers testified about the investigation of the 1988 killing of James Beasley. The officers testified that Evans, a BGF member and a key prosecution witness, was a possible suspect in Beasley's killing, but they had never contacted him. They denied forgoing investigating Evans in exchange for his testimony at Masters's trial.

*1062 James Hahn had worked for what is now the Department of Corrections and Rehabilitation. Hahn identified and tracked gang members, including parolees. Evans was among the gang members whom Hahn **244 tracked. Prior to Masters's trial, Evans had provided Hahn with information on multiple occasions, but most of it was of little value. Hahn also arranged for Evans to work for other law enforcement agencies as an informant, and Evans often was paid for his efforts. The testimony of two former Oakland police officers generally corroborated the preexisting, ongoing working relationship between Evans and Hahn. Hahn believed that Evans was a "professional liar," as he had provided inaccurate information on multiple occasions.

Hahn testified that Evans approached him with information about Sergeant Burchfield's murder and that Hahn made no offer to Evans for his testimony other than to help keep him secure. After Evans testified at Masters's trial, Hahn arranged to have Evans placed on parole in Texas. Evans wanted to enter the federal witness protection program. Hahn explained that a person on parole could not enter the program, so he wrote to Texas's parole board on Evans's behalf, noted how Evans had assisted law enforcement, and urged the board to end his parole.

Hahn testified that he became aware that Evans was a suspect in Beasley's killing, but he could not recall exactly when he learned this. Hahn testified that an investigator from San Francisco had asked him questions about Evans. An Oakland police officer testified he had told Hahn that Evans was a suspect. Hahn could not recall whether he informed the *732 prosecutors that Evans was a suspect in Beasley's killing.

Edward Berberian, then a Marin County Deputy District Attorney, prosecuted Burchfield's murder. Berberian testified he was not aware of anyone from his office directing Hahn's actions in regard to the investigation of Burchfield's murder. Berberian testified that he did not make any threats or promises in exchange for Evans's testimony at Masters's trial. David Gasser, who had served as the prosecution's investigator, recalled some mention of Evans being a potential suspect in Beasley's killing. Both Hahn and Gasser acknowledged they were in regular contact with each other during the investigation of Burchfield's murder.

Due to concerns about Evans's health, the referee presided over Evans's deposition, which occurred before the reference hearing. The parties stipulated that Evans's deposition could be used as evidence at the reference hearing, and Evans did not testify at the hearing. During his deposition, Evans stated that he had never spoken to Masters. Evans said he did not know if Masters was involved in Burchfield's murder. Evans acknowledged he had testified to the contrary at Masters's trial.

*1063 Evans said he had worked regularly as a paid informant for various law enforcement agencies and had previously provided information to Hahn. Evans admitted that some of the information he had provided to Hahn was false.

Evans said that Hahn and Gasser supplied him with information about Burchfield's murder in order for him to implicate Masters. Evans said Hahn and Gasser assured him that he would receive a sentence of less than year on his own pending charges in Alameda County if he testified against Masters. Evans also said that Hahn, prior to Masters's trial, had told him to underplay his informant work so that he could later resume his work as an informant.

Evans said that he changed his mind about testifying at Masters's trial, but then Hahn, Hahn's partner, and Gasser threatened Evans with prosecution in numerous cases - including Burchfield's murder - if he did not implicate Masters. Evans also testified that the Alameda County District Attorney had threatened to charge him under a recidivist statute, with a possible prison sentence of 18 years, if he did not cooperate.

Evans denied any involvement in Beasley's killing and denied being questioned by the police about the killing. Evans admitted that he knew of Beasley, that he had worked for Beasley's son as an "enforcer," and had received payments from the son.

Graham McGruer, a former correctional officer, testified as an expert on California prisons. McGruer had reviewed San Quentin's records and concluded that Masters was not sent to the Adjustment Center until December 1985. (Evans had testified at Masters's **245 trial that the two of them discussed Burchfield's murder in the center around September 1985.) McGruer testified that an attack on a prison guard would normally have been ordered only by the highest echelon of a gang's leadership. If the attack had not been sanctioned by the gang's leaders, McGruer opined, the leadership quickly and personally would have investigated the matter.

Willis, a BGF member and the main prosecution witness at Masters's trial, invoked his privilege against self-incrimination at the reference hearing. Willis later indicated he wished to testify. The referee advised Masters that she would allow him to call Willis as a witness. Masters declined the referee's offer, accepted the referee's *733 initial ruling that Willis was unavailable to testify, and instead offered into evidence prior statements made by him. For example, in 2001, Willis had declared under penalty of perjury that Masters played no role in the attack on Burchfield. Willis further declared that Masters did not author two of the notes that were admitted at trial but merely copied their contents. Willis also declared that he told Berberian he *1064 did not want to testify at Masters's trial and that Berberian told him if he did not testify, he would be returned "right to San Quentin," which Willis interpreted as "a death threat." In 2002, however, Willis declared that he "never lied during [Masters's] trial ...." In 2010, Willis recounted to representatives of the Attorney General his involvement in the murder conspiracy, some of which was consistent with his trial testimony.

Robert Leonard, a linguistics professor, testified as an expert. Dr. Leonard compared how language was used in two documents that were introduced at Masters's trial against 14 other documents authored by Masters. Notably, one of the two documents compared against the others was the report about Burchfield's murder that Willis had asked Masters to write. Dr. Leonard testified that in his opinion it was more likely than not that these two documents were not authored by the same person who authored the other 14 documents.

Woodard, a BGF member and Masters's codefendant, testified that he had assumed leadership of the BGF cell in Carson section after its previous leader was sent to the Adjustment Center. Woodard testified that in April or May 1985, prior to his assuming command of Carson section, Willis had suggested that the BGF attack other gangs. Woodard's predecessor and Willis then changed the plan to attack Burchfield. Masters was present when this new plan was discussed but disagreed with it. Woodard testified that after he had assumed command, he relieved Masters of his responsibilities in the BGF due to Masters's disagreement with the plan to attack Burchfield. Woodard also testified that he had punished Masters by assigning him to perform extra exercise and to write an essay about his insubordination.

Woodard testified that the weapon used to stab Burchfield was crafted on the second tier. Woodard also testified that Masters was not good at making weapons. Woodard testified that he ordered Masters and their codefendant Andre Johnson to neither discuss the case nor testify at trial. Woodard threatened to kill Masters if he did not obey, and the two had some physical altercations while their trial was proceeding.

Michael Rhinehart, a BGF member who was housed on the second tier in Carson section, testified that he learned of the plan to attack a prison guard from Woodard, Willis, and another BGF member, Harold Richardson. Rhinehart testified that his own participation in the attack was limited to passing notes and that Masters did not participate at all. Rhinehart testified that Masters had voted against the plan to assault a guard. As a result of Masters's opposition, Rhinehart said, Woodard became hostile toward Masters. Rhinehart denied any participation in passing the weapon to Johnson but claimed it was made on the second tier. Rhinehart testified that he told Evans the details of Burchfield's murder.

*1065 Welvie Johnson ("Welvie") was the overall third-in-command of the BGF gang and a "shot caller," that is, part of the process that could authorize BGF attacks. Welvie did not know Masters before the attack on Burchfield, and he testified the attack was not sanctioned by the BGF's governing body.

**246 The governing body investigated *734 Burchfield's murder, and Welvie learned of no information indicating that Masters was involved. Welvie also testified that a "short-timer" like Johnson would not have been selected to carry out the attack. Welvie further testified it would have been a breach of BGF protocol to pass a weapon between tiers or to allow a non-BGF member to pass a weapon. Welvie testified that he had been questioned by Hahn about Burchfield's murder, and Hahn confirmed this. Welvie denied that he had previously told prison officials that Woodard ordered Johnson and Masters to attack Burchfield, and he did not recall telling officials that Masters had killed David Jackson.

Hoze, a BGF member who testified at Masters's trial, was subpoenaed to testify at the reference hearing but ultimately was not called as a witness. Statements previously made by Hoze exonerating Masters in Jackson's killing were received into evidence.

Correctional officers had interviewed Richardson in 1986 about his involvement in gang activities, including Burchfield's murder. (See Masters , supra , 62 Cal.4th at p. 1054 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) In his statements to correctional officers, Richardson implicated himself in the attack but did not mention Masters. (See id. at pp. 1054-1058, 199 Cal.Rptr.3d 85 , 365 P.3d 861 [ruling that the trial court did not abuse its discretion by excluding these statements because they contained inadmissible hearsay].) Richardson did not testify at the reference hearing. Reports detailing Richardson's interview, as well as a letter written by Richardson to a prison official, were admitted into evidence at the reference hearing. During the course of these proceedings, Masters submitted to this court additional reports from prison officials about Richardson's debriefing.

III. REFEREE'S REPORT

The referee found, as a general matter, that it was likely that some false testimony was offered at Masters's trial. The referee specifically found that the prosecution's key witnesses, Willis and Evans, had recanted their trial testimony. But the referee also found that both were "liars with highly unreliable and selective memories. [¶] ... [¶] Evans and Willis are utterly lacking in credibility. Both are career criminals whose word, under oath or otherwise, means nothing. Both are well-known snitches. Both would say anything to save their own hide - and both have so admitted. Both are manipulative and unreliable." (Again, Willis did not testify at the reference hearing, and the referee presided over Evans's deposition, which was presented in lieu of his testifying at the reference hearing.)

*1066 The referee further found that every BGF member who testified at the reference hearing had lied during Masters's trial, this proceeding, or both. "All of them, as members of the same prison gang, have a motive now to give testimony favorable to Masters," the referee found.

The referee also noted that the testimony of the BGF members at the reference hearing often was contradictory. For example, there was no clear agreement among the witnesses as to who was in charge, who ordered and organized Burchfield's killing, why the killing was ordered, who made the weapon, who stabbed Burchfield, and whether there was a backup plan.

The law regarding our review of the referee's report is settled. " 'The referee's factual findings are not binding on us, and we can depart from them upon independent examination of the record even when the evidence is conflicting. [Citations.] However, such findings are entitled to great weight where supported by substantial *735 evidence. [Citations.]' [Citations.] 'On the other hand, any conclusions of law or resolution of mixed questions of fact and law that the referee provides are subject to our independent review.' " ( In re Cowan (2018) 5 Cal.5th 235 , 243-244, 234 Cal.Rptr.3d 75 , 419 P.3d 535 ( Cowan ).)

" '[T]he referee is entitled to discredit portions of a witness's testimony while finding the witness credible in other particulars. [Citation.] Thus, the fact that the referee expressly or impliedly disbelieved a witness in some respects, or that portions of a witness's testimony seem unlikely on their face, does not mean that any finding based solely or primarily on the same witness's testimony **247 on other matters is without substantial support.' [Citation.] 'Deference to the referee is particularly appropriate on issues requiring resolution of testimonial conflicts and assessment of witnesses' credibility, because the referee has the opportunity to observe the witnesses' demeanor and manner of testifying.' " ( Cowan , supra , 5 Cal.5th at pp. 244-245, 234 Cal.Rptr.3d 75 , 419 P.3d 535 .)

A. Question One

Our first question to the referee asked, "Was false evidence regarding [Masters]'s role in the charged offenses admitted at the guilt phase of [Masters]'s trial? If so, what was that evidence?"

1. Referee's findings

The referee's answer to this question focused on Willis and Evans. Although Willis did not testify at the reference hearing, the referee found that he initially recanted his trial testimony but then later recanted his recantation. The referee further found that Willis was neither coerced into testifying *1067 against Masters nor coerced into recanting his testimony. The referee ultimately did not believe Willis's recantation of his trial testimony: "The jury saw Willis testify. They knew of his murder conviction. They knew he was given immunity. They heard about his disappointment in not being released from prison in exchange for giving State's evidence. The jury was in the very best position to evaluate him as a witness."

The referee similarly found Evans's testimony at his deposition to be "spectacularly unreliable." Specifically, the referee disbelieved Evans's testimony that Hahn had coerced him into implicating Masters.

2. Masters's exceptions

Masters takes exception to the referee's findings. Preliminarily, he contends that Willis at most only partially disavowed his initial recantation. More importantly, Masters contends the totality of Willis's statements about Burchfield's murder shows that Willis has been so inconsistent that none of his statements is worthy of belief and therefore cannot be the basis for Masters's conviction. Masters also notes that the referee found that some false evidence likely had been admitted at his trial and that Willis and Evans were both liars. From this, Masters concludes that Evans's and Willis's trial testimony implicating him in the conspiracy to attack Burchfield was false.

Masters's exceptions lack merit. "It has long been recognized that 'the offer of a witness, after trial, to retract his [or her] sworn testimony is to be viewed with suspicion.' " ( In re Roberts (2003) 29 Cal.4th 726 , 742, 128 Cal.Rptr.2d 762 , 60 P.3d 165 .) We agree with the referee that there is no sound reason to credit Willis's recantation of his trial testimony. (See id. at p. 743, 128 Cal.Rptr.2d 762 , 60 P.3d 165 [declining to "disturb the jury's verdict based upon a recantation that must be viewed with suspicion and was subsequently disavowed"].)

*736 To the extent Masters relies on Willis's other posttrial statements that are inconsistent with his trial testimony, we agree with the referee that those statements are unbelievable due to Willis's lack of credibility.

Masters also contends that Willis's trial testimony about the two notes that he said Masters sent to him was false. But at trial Willis testified that he sent notes to Masters, requesting him to write reports about the attack. Willis further testified at trial that Willis received two notes in response, both of which were written in Masters's handwriting. There is no evidence suggesting that Willis's testimony on this subject was false.

Masters nonetheless contends that Willis's testimony falsely implied that Masters authored the notes. The referee accepted, and we accept as well, Dr. Leonard's opinion that it was more likely than not that these two *1068 documents were not authored by the same person who authored the other 14 documents offered at the reference hearing. But the referee also found that whether Masters wrote the two notes in his own words or in the words of another did not exonerate him from the conspiracy to attack Burchfield.

Overall, the referee found that the jury had sufficient information to gauge Willis's credibility. We agree. Masters here mounts a generalized attack on Willis's credibility, but this general attack does not render false any **248 particular aspect of Willis's trial testimony about Masters's role in the murder of Burchfield.

Evans, in his deposition, also retracted his trial testimony. But the referee had the opportunity to observe Evans's demeanor and, as noted, found his recantation to be "wholly incredible." We accept the referee's finding regarding Evans's lack of credibility and her ultimate finding rejecting Evans's recantation.

Masters notes that the trial testimony concerning Evans and Hahn's pre-existing, ongoing working relationship was incomplete. The referee found this contention "appear[s] to be true," and the Attorney General concedes that Evans did lie at Masters's trial about the number of meetings he had with Hahn. But the referee further found that "Evans had many contacts with Hahn, although it is unlikely that Evan gave useful information to Hahn more than a few times." We accept the finding that Evans testified falsely at Masters's trial regarding his relationship with Hahn. This finding, however, goes to Evans's general credibility, which defense counsel vigorously attacked at trial. ( Post , 250 Cal.Rptr.3d at pp. 741-742, 446 P.3d at p. 252.) And Evans's false testimony about his relationship with Hahn does not render false any particular aspect of Evans's trial testimony concerning Masters's role in the attack on Burchfield.

Masters further notes that Evans testified at trial that the two of them discussed the attack on Burchfield around September 1985 in the Adjustment Center. The evidence adduced at the reference hearing showed that Masters was not transferred to the Adjustment Center until December 1985. From this, Masters infers that Evans's trial testimony about it was false. But even if Evans's testimony about the timing of the conversation was false, this does not necessarily mean his testimony about the content of the conversation was false. In any event, Masters's trial counsel was aware of this discrepancy as to when the conversation occurred and argued this point to the jury during guilt phase closing arguments.

Masters also notes that Rhinehart and Woodard testified at the reference hearing that Masters was not involved in the conspiracy to attack Burchfield. But the referee *737 discounted the testimony of all BGF members because all of *1069 them had lied at some point and now had a motive to testify in Masters's favor. We accept the referee's finding regarding their credibility, as she observed their testimony at the reference hearing and was in the best position to assess their credibility.

Masters also points to other evidence that demonstrates his innocence and thereby suggests that false evidence of his guilt was presented at his trial. Masters's theory of the case is that he was not involved with Burchfield's murder and that Harold Richardson was Woodward and Johnson's other main coconspirator. In support of this contention, he relies in part on statements made by Richardson during his debriefing. At Masters's trial, however, the court did not abuse its discretion in excluding Richardson's statements implicating himself because they were unreliable hearsay. ( Masters , supra , 62 Cal.4th at pp. 1054-1058, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) We are not now persuaded by Masters's contention that Richardson's unreliable hearsay statements indicate that false evidence was presented at trial; indeed, the unreliability of Richardson's statements is the reason we upheld the trial court's exclusion of them.

In sum, we accept the referee's findings with respect to the first question because they are supported by substantial evidence.

B. Question Two

Our second question to the referee asked, "Is there newly discovered, credible evidence indicative of [Masters]'s not having been a participant in the charged offenses? If so, what is that evidence?"

Newly discovered, credible evidence, in the context of our second question, means "evidence that has been discovered after trial, that could not have been discovered prior to trial by the exercise of due diligence, and is admissible and not merely cumulative, corroborative, collateral, or impeaching." (§ 1473, subd. (b)(3)(B), as amended by Stats. 2016, ch. 785, § 1, eff. Jan. 1, 2017.)

**249 1. Referee's findings

The referee answered the first part of this question in the negative. The prosecutors at trial argued that the murder weapon was delivered from the fourth tier in the prison (where Masters was housed) to the second tier (where Johnson was housed). Several BGF members testified at the reference hearing that it would have been less dangerous to fabricate the murder weapon on the same tier as Johnson and then pass it to him. The referee found that even if it would have been less dangerous to fabricate the weapon on the same tier, it does not follow that this is what actually happened.

*1070 With respect to Dr. Leonard, the linguistics expert, the referee found him to be a convincing witness, but she also found that his testimony was not "new." The referee ultimately concluded that Dr. Leonard's testimony did not exonerate Masters because the notes in Masters's handwriting indicated his involvement in the conspiracy to attack Burchfield regardless of whether he was the source of the notes' contents.

Both parties take exception to the referee's findings.

2. The Attorney General's exceptions

As mentioned above, Willis had stated posttrial that Masters did not author two of the notes that were admitted into evidence at trial. To bolster this contention, Masters offered the testimony of Dr. Leonard at the reference hearing. The Attorney General, pursuant to *738 People v. Kelly (1976) 17 Cal.3d 24 , 130 Cal.Rptr. 144 , 549 P.2d 1240 , moved to exclude Dr. Leonard's testimony on the ground that his analytical technique of determining a document's author had not been sufficiently accepted by the scientific community or the courts. The referee denied the motion, ruling that Kelly did not apply to this type of evidence. The Attorney General takes exception to the referee's ruling.

The referee accepted Dr. Leonard's testimony as credible but nonetheless found that Masters had participated in the conspiracy. As explained further below, because the referee's consideration of this evidence did not benefit Masters, we need not decide whether the referee erred by not conducting a Kelly hearing.

3. Masters's exceptions

Preliminarily, Masters takes exception to the referee's exclusion of evidence indicating that Willis "framed" him. In a 2005 letter from Willis to Masters's habeas corpus counsel, Willis wrote, "Why [Masters's trial] lawyers never looked through my property in San Quentin - Hint Hint." Masters sought to have his habeas corpus counsel testify about the conversation that counsel had with Willis about this topic, but the referee did not permit the testimony. Masters then made an offer of proof that Willis had told Masters's habeas corpus counsel that he hid some notes in his prison-issued portable television. Willis also told Masters's habeas corpus counsel that the hidden notes were the original notes that he later directed Masters to copy into his own handwriting. And Willis told Masters's habeas corpus counsel that the television set and the notes it contained later disappeared.

The proffered evidence indicated Willis had instructed Masters to copy notes that Willis had authored, which was consistent with Dr. Leonard's *1071 testimony. But even if we were to assume that habeas corpus counsel had testified about what Willis told him, Willis's credibility remains the ultimate issue. We agree with the referee that Willis was not credible, and his statements do not become more credible when conveyed through a layer of hearsay. Thus, even if the referee abused her discretion by excluding this evidence, it was harmless because we agree with the referee's determination that Willis was not credible.

a. Notes' authorship

Masters takes exception to the referee's characterization that Dr. Leonard's testimony was not new evidence. Dr. Leonard's testimony about the authorship of the two notes was presented for the first time at the reference hearing, but the notes themselves were introduced at Masters's trial. The Attorney General observes that "authorship analysis," that is, a stylistic comparison of questioned writings and utterances to known **250 exemplars, predates Masters's 1989 trial. (E.g., United States v. Clifford (3d Cir. 1983) 704 F.2d 86 , 90 ; United States v. Hearst (9th Cir. 1977) 563 F.2d 1331 , 1349-1350 ; but cf. Coulthard, Author Identification, Idiolect, and Linguistic Uniqueness (2004) 25 Applied Linguistics 431 [noting little growth in forensic linguistics as a discipline until the 1990s].) The Attorney General contends that Masters was not diligent in presenting this linguistic evidence in these habeas corpus proceedings: Although Masters in 1998 had received, from a different expert, a preliminary linguistic analysis regarding the two notes, he neither included this expert's analysis as an exhibit to, nor did he discuss its contents in, his petition for writ of habeas corpus filed in 2005.

As we explain further below, regardless of whether Dr. Leonard's testimony constituted new evidence, and regardless of whether Masters was diligent in discovering *739 and presenting it, it ultimately does not benefit Masters. Masters specifically takes exception to a portion of the referee's report in which she wrote, "Dr. Leonard's testimony does not exonerate [Masters]. It may suggest that Masters was not a planner or leader of the conspiracy, but Masters was not tried as the planner or leader of the conspiracy; he was tried as the knife-sharpener and messenger." The evidence presented at Masters's trial arguably suggests he was a planner and leader. But even if we were to assign no weight to the referee's suggestion that Masters only relayed messages and helped fabricate the weapon, we conclude that substantial evidence supports the referee's ultimate conclusion that the notes' authorship is not material because Masters's purported lack of authorship does not exonerate him. ( Post , 250 Cal.Rptr.3d at pp. 747-748, 446 P.3d at pp. 257-258.) *1072 b. Weapon's fabrication

Masters also takes exception to the referee's finding regarding who fabricated the weapon used to stab Burchfield. Preliminarily, whether this evidence is new is debatable, as Masters does not explain why these witnesses could not have testified at his trial. We nonetheless will assume these witnesses would have invoked their constitutional privilege against self-incrimination at Masters's trial and therefore could not have been compelled to testify. Regardless, newly discovered evidence must be credible (§ 1473, subd. (b)(3)(A)), and the referee found these witnesses were not credible.

Even if we were to assume the testimony about the possibility that the weapon was fabricated completely on the second tier might be newly discovered evidence, there was no credible evidence that it actually was sharpened there. Rhinehart, for example, testified that the weapon never left the second tier, that Johnson and the source of metal for the weapon were housed on one side of him, and that the person who sharpened the metal was housed on the other side of him. Yet Rhinehart also testified that he did not observe its fabrication or participate in passing it to Johnson. And, as the referee noted, Masters's witnesses named numerous possible fabricators of the weapon. Because the evidence at the reference hearing amounted only to speculation pointing in many different directions, we accept the referee's ultimate finding that Masters has not provided new, credible evidence that someone else fabricated the weapon used to stab Burchfield.

c. Other evidence

Masters notes the referee made other findings that were indicative of new evidence having been presented, such as Evans's having lied at trial and Willis's posttrial declaration and statements, and he therefore contends this evidence should have been included in the answer to this question. Again, newly discovered evidence must be credible (§ 1473, subd. (b)(3)(A)), and the referee found Willis and Evans were not credible. The same is true for the testimony of the other BGF witnesses who testified at the reference hearing.

Masters correctly notes that the referee found Evans's trial testimony about his relationship with Hahn was false. Because the referee found Evans credible on this discrete topic, we agree with Masters that this portion of the Evans's deposition can constitute newly discovered evidence. As we explain **251 more fully below, however, it is ultimately unavailing. ( Post , 250 Cal.Rptr.3d at pp. 746-747, 446 P.3d at pp. 256-257.)

Similarly, Masters contends Richardson was the actual mastermind behind the conspiracy to murder Burchfield, and in support *740 he cites correctional *1073 officers' reports concerning statements made by Richardson that previously had been redacted and thus were unknown and unavailable to Masters at time of his trial. Masters also notes the new evidence about Evans's possible involvement in Beasley's killing. As we explain more fully below, none of this evidence, even if new, entitles Masters to relief.

C. Question Three

Our third question to the referee asked, "What, if any, promises or threats were made to guilt phase prosecution witness Rufus Willis by District Attorney Investigator Charles Numark or Deputy District Attorneys Edward Berberian or Paula Kamena? Was Willis's trial testimony affected by any such promises or threats, and, if so, how?"

1. Referee's findings

The referee found that Numark, Berberian, and Kamena did not make any undisclosed promises or threats to Willis. As noted, Willis did not testify at the reference hearing, but Masters introduced statements Willis had made previously. The referee found that Willis was manipulative and untrustworthy, and that any new claims of undisclosed prosecutorial promises or threats were unsubstantiated.

2. Masters's exceptions

Masters takes exception to the referee's findings with respect to Berberian and Numark. Masters notes that Willis in 2010 stated that Berberian threatened to not honor a prior agreement to release him from prison if he did not testify. Willis also stated that after he had been released on parole, Berberian threatened to return him to San Quentin. And Masters observes that Berberian testified at the reference hearing but did not affirmatively refute Willis's accusations. Masters further notes that Willis stated that Numark promised him he would take steps to ensure no one would oppose Willis's parole. Masters does not dispute that there was no evidence of Kamena's having made any promises or threats to Willis.

Masters's exceptions lack merit. With respect to Numark's efforts to shorten Willis's incarceration, it was disclosed at Masters's trial that Numark had offered to help Willis secure his release from prison, but this offer was retracted before Masters's trial. Thus, to the extent Numark's offer to help affected Willis's trial testimony, the jury was aware of the investigator's efforts. Masters is correct that our question was not limited to undisclosed threats or promises, but the jury was able to assess the effect, if any, of Numark's offers on Willis's decision to implicate Masters.

*1074 In support of his claim that Berberian had threatened to return Willis to prison, Masters presented Willis's prior statements of what Berberian purportedly said to him. The referee found Willis's claims to be unbelievable, and we agree with this finding.

Moreover, the evidence of the purported threats is a web of hearsay. ( Evid. Code, § 1200.) Masters contends that Berberian's failure to refute Willis's claims about the threats constitutes an adoptive admission, which is an exception to hearsay rule. ( Id. , § 1221.) According to Masters, because there was no affirmative refutation of Willis's claims by Berberian, who did testify at the reference hearing, Berberian has adopted the statements attributed to him by Willis. But even if we were to assume Willis's statements concerning Berberian's purported threats were admissible under section 1221, that section governs only the admissibility of certain out-of-court statements, not the weight the factfinder may *741 accord to them. (See People v. Turner (1994) 8 Cal.4th 137 , 190-191, 32 Cal.Rptr.2d 762 , 878 P.2d 521 ; CACI No. 213 [the jury " may consider that statement as evidence against" the party against whom the statement was offered under specified conditions (italics added) ]; see also CALCRIM No. 357.) The referee found that Willis was untrustworthy, and by extension, she found the claims of prosecutorial **252 threats to be unsubstantiated. We agree with these findings.

In sum, the jury was aware of the discussions between Numark and Willis, as well as the terms of the immunity agreement between Willis and Berberian. To the extent Masters contends the totality of the circumstances nonetheless indicates Numark, Berberian, and Kamena made some sort of undisclosed threat or promise to Willis, substantial evidence supports the referee's contrary finding.

D. Question Four

Our fourth question to the referee asked, "Were there promises, threats or facts concerning guilt phase prosecution witness Bobby Evans's relationship with law enforcement agencies of which Deputy District Attorneys Berberian and Kamena were, or should have been, aware, but that were not disclosed to the defense? If so, what are those promises, threats or facts?"

1. Referee's findings

The referee answered the first part of this question in the negative. The referee found that Evans had more extensive contacts with Hahn than were disclosed at trial, but that Berberian and Kamena did not know about these contacts. The referee also found that the evidence of the contact between Evans and Hahn was not material because Masters was aware of Evans's lack of credibility and had extensively urged the jury not to believe him.

*1075 2. Masters's exceptions

Masters again takes exception to the referee's finding regarding Evans's contacts with Hahn. Masters first contends that Evans was, in essence, a professional informant who worked regularly with Hahn. Although the referee did not expressly characterize Evans as such, she agreed that Hahn and Evans's working relationship was underplayed at Masters's trial. Masters also contends that Hahn was a member of the prosecution team and, for that reason, Berberian's and Kamena's actual knowledge is immaterial because Hahn's knowledge is imputed to them.

Masters similarly contends the prosecution team should have known about Evans's possible involvement in James Beasley's killing. As the testimony of Hahn and others indicates, Hahn knew Evans was a suspect in Beasley's killing. In addition, one of the prosecutors' investigators, David Gasser, recalled some discussion about Evans's possible involvement in Beasley's killing. Masters therefore again contends that Hahn's knowledge is imputed to Berberian and Kamena.

As Masters does not dispute that Berberian and Kamena did not actually know the extent of Hahn's working relationship with Evans or Evans's status as a suspect in Beasley's killing, we accept the referee's finding that the prosecuting attorneys lacked actual knowledge of any promises, threats, or facts concerning Evans that were not disclosed to Masters or otherwise discovered during his trial. With respect to facts that can be imputed to Berberian and Kamena, we address those contentions below.

E. Question Five

Our fifth question to the referee asked, "Did Deputy District Attorneys Berberian and Kamena knowingly present false testimony *742 by Bobby Evans? If so, what was that testimony?"

The referee found that Berberian and Kamena did not knowingly present false testimony by Evans at Masters's trial. At the reference hearing, Masters conceded there was no evidence to support this contention.

Neither party makes an exception to the referee's finding, and we accept it.

F. Question Six

Our sixth question to the referee asked, "What, if any, promises or threats were made to Bobby Evans by District Attorney Investigator Numark, Department of Corrections Investigator James Hahn, or Deputy District *1076 Attorneys Berberian and Kamena? Was Evans's trial testimony affected by any such promises or threats, and, if so, how?" **253 1. Referee's findings

The referee found that there was no evidence that Berberian, Kamena, or Numark had made any threats or promises to Evans, and Masters conceded there was no evidence to support these contentions. At the reference hearing, Masters contended that Hahn promised Evans that if he testified against Masters, then Hahn would ensure that Evans would not be implicated in Beasley's killing. The referee found that the evidence presented did not support this contention.

2. Masters's exceptions

Masters takes exception to the referee's finding only with respect to Hahn. Preliminarily, Masters contends that the referee erred by not considering prior statements made by Evans in other proceedings that were consistent with his deposition testimony here. We have reviewed these statements, and they generally are consistent with Evans's deposition testimony. But even if we were to assume they should have been admitted during the reference hearing, it is doubtful they would have altered the referee's finding that Evans had not been truthful in his deposition testimony, which the referee described as "spectacularly unreliable."

Masters concedes there is "no direct evidence" that Hahn promised Evans that he would not be implicated in Beasley's killing if he testified against Masters. Rather, Masters contends that Evans was the primary suspect in Beasley's killing but was no longer considered a suspect after he implicated Masters. As such, "one of the most plausible explanations" for this shift, according to Masters, was Hahn's involvement. The officers investigating Beasley's killing, however, expressly denied that they stopped investigating Evans in exchange for his testimony at Masters's trial, and the referee found this testimony credible. We accept the referee's credibility findings as well as the referee's ultimate finding that Hahn did not influence the investigation of Beasley's killing.

Masters also contends that Hahn's threats or promises to Evans were not limited to the Beasley investigation. Rather, Masters contends that Evans and Hahn had a pre-existing, ongoing relationship in which Evans supplied Hahn and other law enforcement agencies with information. Masters further contends any threats or promises made to Evans ought to be viewed in the context of the expectation *1077 of their working relationship continuing past Evans's testifying against Masters. But besides this generalized expectation of their working relationship continuing past Masters's trial, he presented no evidence of any specific promises made by Hahn to Evans that were not otherwise disclosed or discovered during the course of Masters's trial. *743 Masters further contends that the jury had incomplete information about the threats made against Evans. Although Evans testified at Masters's trial that he had pleaded guilty to attempted robbery in exchange for serving no more than 16 months, in his deposition he said he was facing 18 years in prison if he did not testify against Masters. But the only source of evidence indicating that Evans was threatened with a much longer sentence was Evans. Because the referee found Evans to be an unbelievable witness, we accept the referee's finding that Evans received no threat or promise that was not otherwise disclosed or discovered during Masters's trial.

G. Question Seven

Our seventh and final question to the referee asked, "Did penalty phase prosecution witness [Johnnie] Hoze provide false testimony regarding [Masters]'s involvement in the murder of inmate David Jackson? If so, what was that false testimony?"

The referee found no basis to believe that Hoze testified falsely at Masters's trial. Although Hoze did not testify at the evidentiary hearing, the referee reviewed his numerous previous inconsistent statements about Masters's involvement in Jackson's murder. The referee's review of Hoze's statements led her to conclude that "Hoze recant[s] and [unrecants] with alarming frequency" and "admits he has a motive to do so .... His recantations [of his trial testimony] are not believable."

Neither party makes an exception to the referee's finding, and we agree with it.

**254 IV. DISCUSSION

"Because a petition for a writ of habeas corpus is a collateral attack on a presumptively final criminal judgment, 'the petitioner bears a heavy burden initially to plead sufficient grounds for relief, and then later to prove them.' [Citation.] To obtain relief, the petitioner must prove by a preponderance of the evidence the facts that establish entitlement to relief." ( Cowan , supra , 5 Cal.5th at p. 243 , 234 Cal.Rptr.3d 75 , 419 P.3d 535 .)

A. False Evidence

In his habeas corpus petition, Masters contends false evidence was presented at his trial. We deny relief on this claim.

*1078 Habeas corpus relief is available if "[f]alse evidence that is substantially material or probative on the issue of guilt or punishment was introduced against a person at a hearing or trial relating to his or incarceration." (§ 1473, subd. (b)(1).) "Determining that the evidence was false clears the first hurdle to relief. 'The statute and the prior decisions applying section 1473 make clear that once a defendant shows that false evidence was admitted at trial, relief is available under section 1473 as long as the false evidence was "material." ' [Citation.] Materiality is shown if there is a reasonable probability the result would have been different without the false evidence." ( In re Figueroa (2018) 4 Cal.5th 576 , 588-589, 229 Cal.Rptr.3d 673 , 412 P.3d 356 .) "This required showing of prejudice is the same as the reasonably probable test for state law error established under People v. Watson (1956) 46 Cal.2d 818 , 836, 299 P.2d 243 . [Citation.] We make such a determination based on the totality of the relevant circumstances." ( In re Richards (2016) 63 Cal.4th 291 , 312-313, 202 Cal.Rptr.3d 678 , 371 P.3d 195 .)

1. Willis, Evans, and other BGF members

Masters contends that Willis and Evans testified falsely at his trial. As noted, the referee found that Willis and Evans *744 had generally recanted their trial testimony but that both were chronic liars. As such, the referee did not believe their recantations. The referee presided over Evans's deposition, and we have accepted her findings regarding his credibility. We independently reviewed Willis's statements and have agreed with her finding regarding his credibility. Habeas corpus is an attack on a presumptively final judgment, and neither Masters's generalized arguments about Willis's and Evans's credibility nor his generalized arguments that he was falsely accused are sufficient to warrant relief. Those arguments do not show the falsity of any specific evidence of his involvement in the conspiracy. The referee similarly rejected the testimony of the other BGF members who testified at the reference hearing in support of Masters, and we have accepted her findings on these matters as well.

The referee did find that the trial testimony concerning Evans and Hahn's pre-existing, ongoing working relationship was false because it failed to adequately describe the nature and extent of their relationship, and we have accepted this finding. As an initial matter, we note the referee found that Evans had more extensive contacts with law enforcement than was disclosed at trial, but she did not expressly find that Hahn's trial testimony was false. And it does not appear that Hahn, at any time during Masters's trial, was questioned about the extent of his relationship with Evans before June 1989.

The jury knew that Hahn tracked parolees such as Evans, which suggests that Hahn was aware of Evans's criminal history and his propensity to not tell *1079 the truth. The jury also knew that Evans was a violent felon, had contacted Hahn and offered to disclose information about Burchfield's murder, and was testifying to receive a benefit for his cooperation, that is, a possible reduction in his own sentence to be facilitated by Hahn's intervention.

With respect to Evans's trial testimony regarding the nature and extent of his previous contacts with Hahn, Evans testified that before he met with "the District Attorney," he had met Hahn only "[o]ne time prior." But Evans also testified, before Masters's trial, that he (or, in one instance, his wife) had provided Hahn with incriminating information **255 about a correctional officer and, in separate incidents, about two BGF members other than Johnson, Masters, and Woodard. The jury further knew that Evans wanted to avoid going to prison because his "life had been threatened" by the BGF and that Hahn was aware that Evans feared retaliation by the BGF. Evans also testified that Hahn had put some money on his books in a correctional facility and had given him money for cigarettes. In sum, the jury heard that Evans, before testifying at Masters's trial, had provided Hahn with information on multiple occasions in addition to the information about Burchfield's murder, and that Hahn had given Evans money.

The evidence at the reference hearing showed that Evans, before coming forward about Burchfield's murder, had often been paid by various law enforcement agencies for his informant work, including work that had been arranged by Hahn. The evidence also showed that Hahn knew Evans previously had provided false information, but nonetheless made efforts to ensure Evans would testify at Masters's trial. Masters further suggests that Evans had an expectation to continue his work as an informant after Masters's trial. Although this additional information sheds further light on Evans's credibility, it is not so different from the evidence adduced at trial, which provided the jury ample knowledge *745 about Evans for purposes of assessing his credibility.

Masters contends that the jury did not know that Evans had served as an informant before Evans disclosed information about Burchfield's murder to Hahn. But the jury was aware that Evans had provided information, including information about members of his own gang, to law enforcement. And, as the referee noted, Masters "staged an unsparing attack on Evans" and his credibility at trial. Indeed, Masters's counsel specifically argued to the jury that Evans had "snitched" on others besides Johnson, Masters, and Woodard.

Moreover, Willis was the primary witness against Masters, as he had personal knowledge about the conspiracy; "Evans's testimony served only to confirm Willis's testimony." ( Masters , supra , 62 Cal.4th at p. 1068 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) The notes in Masters's handwriting further implicated him. In addition, there was *1080 no material falsity regarding the benefits Evans actually received in exchange for his testimony at Masters's trial. (See id. at pp. 1036-1037, 1064-1068, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .)

In sum, the false evidence at Masters's trial provided the jury with an incomplete account of Evans and Hahn's relationship. But there was ample evidence at trial that provided the jury with strong reasons to question Evans's credibility and to view his testimony with caution or suspicion. The totality of the circumstances does not make it reasonably probable that the jury would have returned a different verdict had it known the full scope of the relationship between Evans and Hahn, including any possible expectation regarding future informant assignments.

In his habeas corpus petition, Masters contends that Hoze testified falsely during the penalty phase. The referee found no basis for believing Hoze testified falsely at Masters's trial, and Masters does not dispute this finding. Because we agree with the referee's finding, Masters is not entitled to relief on this basis.

2. Authorship of the notes

Masters also contends that the evidence at trial that he authored the two notes attributed to him was false. Again, Willis testified at trial that he sent Masters a note requesting written details about the murder. Willis then received a note in Masters's handwriting but signed with an ambiguous code name. This note discussed some aspects of the murder. Willis then sent Masters another note requesting another report, and Willis received a second note, again in Masters's handwriting, which contained more details about the attack. The second note was signed with a code name assigned to Masters. Willis's testimony, when coupled with the testimony of the handwriting expert, suggests that Masters physically wrote the incriminating notes, even though Willis did not observe Masters having done so. The timing of Willis's requests and Masters's responses, moreover, also supports the inference that Masters had enough time to compose the **256 contents of the notes, that is, he was not merely copying notes that someone else had composed. (See Masters , supra , 62 Cal.4th at p. 1030 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) And Willis testified at Masters's trial that Masters had written other letters about Burchfield's murder that Willis later destroyed. But Willis also testified that the BGF sometimes had its members copy notes that originated from another author. At the reference hearing, Masters produced Willis's prior statements in which he said Masters did not author these two notes.

Masters does not now show that any of the trial testimony or evidence concerning *746 these two notes was false. Notably, Willis never testified at trial that Masters authored the notes. Although the jury could have inferred from *1081 Willis's trial testimony that Masters authored the contents of the notes, Masters does not show that Willis's testimony about his receiving the two notes, in Masters's handwriting, was false. Nor does Masters now dispute the trial evidence that showed that the notes were in his handwriting. Rather, Masters's claim is that the evidence adduced at the reference hearing, including Dr. Leonard's expert opinion, casts doubt on any inference that Masters authored the contents of the notes.

Ultimately, even assuming this evidence showed the falsity of any suggestion in Willis's trial testimony that Masters authored the notes' contents, the fact that Masters did not author the notes, if accurate, does not lead us to conclude there was a reasonable probability of a different outcome. The jury knew that BGF members sometimes copied notes authored by others. Even if Masters only copied notes composed by others, this would not mean their contents were false, and Masters has not provided any credible evidence to show that their contents were false. Further, as the referee observed, even if Masters merely copied the notes, his doing so could reasonably be viewed as evidence that he participated in the conspiracy to murder Burchfield.

B. Newly Discovered Evidence

In his habeas corpus petition, Masters contends that newly discovered evidence casts doubt on the verdicts at both the guilt and penalty phases of his trial. Habeas corpus relief based on newly discovered evidence may be granted when "[n]ew evidence exists that is credible, material, presented without substantial delay, and of such decisive force and value that it would have more likely than not changed the outcome at trial." (§ 1473, subd. (b)(3)(A).)

In support of this claim, Masters presented the following evidence: (1) Willis's, Evans's, and Hoze's posttrial recantations of their trial testimony; (2) testimony that Masters did not fabricate the weapon used to kill Sergeant Burchfield; (3) Dr. Leonard's testimony that Masters did not author the notes introduced at his trial; (4) Evans's possible involvement in Beasley's killing; and (5) statements concerning Harold Richardson, which previously were unknown to Masters. We conclude that none of this evidence, individually or collectively, would have more likely than not changed the outcome at trial.

Preliminarily, we note that Masters contends some of the evidence that he presented at the reference hearing supports his claims for both false evidence and newly discovered evidence. Newly discovered evidence does not necessarily mean that false evidence was presented at trial. But under the circumstances before us, we accept Masters's contention that at least some of his evidence supports both types of claims.

*1082 1. Witnesses' recantations

At the reference hearing, Masters presented evidence of Willis's, Evans's, and Hoze's posttrial recantation of their trial testimony. With respect to Willis and Evans, the referee observed that she "cannot find that there is any 'new evidence' now - there is only 'different evidence' from the same witnesses, in the form of their recantations and unreliable memories." Willis's and Evans's posttrial recantations are, strictly speaking, new. But the referee's comments make clear that she was finding these witnesses not credible; she also found Hoze not credible.

Regardless of how the referee characterized the posttrial statements of these *747 witnesses, she did not believe their recantations. **257 We have accepted the referee's findings regarding Evans's credibility, and we agree with her findings regarding Willis's and Hoze's credibility. (§ 1473, subd. (b)(3)(B) [newly discoverable evidence must be credible].) At best, Masters has demonstrated that these witnesses generally are liars, but he does not offer any persuasive reason to credit their recantations over their trial testimony.

As we have accepted Masters's contention that Evans's testimony about his relationship with Hahn was false, we accept his contention that Evans's recantation on this topic also constitutes newly discovered evidence. However, just as we have found no reasonable probability of a different result at trial absent this false evidence, we also conclude that Masters has not shown that this newly discovered evidence would have more likely than not changed the outcome at his trial. Again, the trial evidence already gave the jury ample reason to doubt Evans's credibility. Further, the inconsistency in Evans's statements is notable. In the context of a motion for a new trial based on newly discovered evidence, for example, we have said " 'the trial court may consider the credibility as well as materiality of the evidence in its determination [of] whether introduction of the evidence in a new trial would render a different result reasonably probable.' " ( People v. Delgado (1993) 5 Cal.4th 312 , 329, 19 Cal.Rptr.2d 529 , 851 P.2d 811 ; see Evid. Code § 780, subd. (h) [when assessing a witness's credibility, the trier of fact may consider a witness's prior inconsistent statements].) Similarly here, Evans's new statements about his preexisting relationship with Hahn need not be considered in a vacuum but may be considered in light of his prior inconsistent statements on this topic.

2. Weapon's fabrication

At the reference hearing, several witnesses testified about the fabrication of the weapon used to kill Burchfield. But the referee found none of them credible, we have accepted the referee's finding in this matter, and therefore their testimony cannot constitute newly discovered evidence.

*1083 In addition to their lack of credibility, Masters's witnesses named multiple possible fabricators, but they could not all have fabricated the weapon used to kill Burchfield. Given the inconsistencies in and between their testimony, it is not more likely than not that the evidence now presented about the weapon's fabrication would have changed the outcome of the trial. Moreover, Masters's participation in the conspiracy was not limited to helping fabricate the weapon.

3. Authorship of the notes

Masters contends that Dr. Leonard's testimony that Masters did not author the contents of the two incriminating notes introduced at his trial constitutes new evidence. As noted, evidence was introduced at Masters's trial that two notes about the attack were written in his handwriting, and Willis testified at trial that notes were sometimes copied by several people to obscure the identity of their authors. ( Masters , supra , 62 Cal.4th at p. 1031 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .)

Masters contends that Dr. Leonard's testimony helps show that he did not plan or carry out the plan to attack Burchfield. But evidence that Masters did not author the notes does not necessarily mean he did not plan or participate in the attack on Burchfield. Rather, a reasonable jury could have believed that although Masters copied and did not author the notes, he still participated in the planning of the attack and sharpened the weapon that was used. Because the accuracy of the contents *748 of these notes under these circumstances is not necessarily dependent on who authored them, we cannot say Dr. Leonard's testimony about the authorship of the notes more likely than not would have changed the outcome of Masters's trial.

Evidence that someone other than Masters authored the notes supports Masters's theory that Willis framed him. It is also consistent with the possibility that Masters was coerced into falsely confessing that he planned Burchfield's murder. But evidence that Masters did not author the notes does not establish that their contents were false. Moreover, the jury already had reason to doubt Willis's credibility, and Masters's trial counsel argued **258 to the jury that Willis "doctor[ed] up" the notes and specifically suggested that Willis had provided Masters with a draft for him to copy. In other words, the jury had grounds for doubting Masters's authorship of the notes but nonetheless convicted him.

At trial, the case against Masters rested to a significant degree on Willis's credibility. Through a variety of sources, Masters now attempts to further undermine the credibility of Willis's testimony. But even with these additional bases for doubting his credibility, we are not persuaded it is more likely than not that the outcome of the trial would have been different, especially since Willis's credibility was litigated extensively at trial.

*1084 4. Beasley's killing

Evans's possible involvement in Beasley's killing likely constitutes new evidence because we accept the premise that Masters, at the time of his trial, reasonably could not have been expected to know the details of an unrelated homicide or Evans's status as a possible suspect. Nonetheless, the officers investigating Beasley's killing testified that they did not forgo investigating Evans in exchange for his trial testimony, the referee credited this testimony, and we have accepted this finding. Further, in light of what the jury already knew about Evans's violent criminal past, Masters fails to show that Evans's possible involvement in Beasley's killing would have more likely than not changed the outcome at his trial.

5. Richardson's statements

Before his trial, in an effort to sever his trial from Johnson's and Woodard's, Masters unsuccessfully sought to rely on partially redacted statements made by Richardson about Burchfield's murder; at his trial, the prosecutor sought to exclude these statements. In these statements, Richardson implicated himself and did not mention Masters. The trial court excluded Richardson's statements as unreliable hearsay, and we ruled that the trial court did not abuse its discretion in doing so. ( Masters , supra , 62 Cal.4th at pp. 1054-1058, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) At the reference hearing, to bolster Richardson's credibility and otherwise cast doubt on the prosecution's case, Masters relied on an unredacted version of Richardson's statements as well as correctional officers' reports concerning statements made by Richardson. During these proceedings, Masters provided us with additional and more complete reports, authored by correctional officers, concerning Richardson's statements.

The unredacted portions of Richardson's statements and some of the related reports are not new evidence; Masters had access to them at his trial. The redacted portions and the reports not disclosed previously to him are new to Masters, however, as he did not have access to them at trial. We have reviewed these additional materials, including the materials recently supplied directly to us by Masters. Even if we were to assume that this additional evidence *749 would have been admissible at trial, it enhances only slightly Richardson's credibility, and none of it would have altered our conclusion that Richardson's statements were not sufficiently trustworthy.

At the time of his trial, Masters was aware of most of Richardson's statements to prison officials, and the trial court excluded them as unreliable hearsay. As the trial court observed, Richardson did not speak to prison officials about Burchfield's murder until more than a year after the attack had occurred. Richardson thus had ample opportunity to glean the relevant details *1085 from others and then pass them off to prison officials as his own personal knowledge. Masters now presents no new information from Richardson about Burchfield's murder; instead, he relies on additional, generalized information from others about Richardson that, at most, somewhat bolsters Richardson's credibility. In light of the totality of the evidence at trial, Masters has not shown that this additional information from the reports concerning Richardson's debriefing, even if somehow admissible at his trial, would have more likely than not changed its outcome.

C. Prosecutorial Misconduct

In his habeas corpus petition, Masters contends that various actions by the prosecutors rendered his trial unfair. We reject these contentions.

**259 "In California, the law regarding prosecutorial misconduct is settled: 'When a prosecutor's intemperate behavior is sufficiently egregious that it infects the trial with such a degree of unfairness as to render the subsequent conviction a denial of due process, the federal Constitution is violated. Prosecutorial misconduct that falls short of rendering the trial fundamentally unfair may still constitute misconduct under state law if it involves the use of deceptive or reprehensible methods to persuade the trial court or the jury.' " ( Masters , supra , 62 Cal.4th at p. 1052 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .)

1. Threats or promises of leniency

Masters contends that his trial was fundamentally unfair because Willis's testimony was unreliable due to improper coercion by the prosecution. Specifically, Masters alleges that Berberian threatened that if Willis's testimony did not implicate Masters, Willis would be returned to San Quentin, where he feared he would be killed by other inmates.

Offering witnesses immunity in return for testifying or "confronting [witnesses] with the predicament" they are in is not necessarily improper coercion. ( People v. Badgett (1995) 10 Cal.4th 330 , 355, 41 Cal.Rptr.2d 635 , 895 P.2d 877 .) Such conduct is acceptable so long as it is with the understanding that the witness's gaining the benefit or avoiding the punishment is conditioned on the witness's testifying "fully and fairly." ( Ibid. ) It is unacceptably coercive, however, for an agreement to require that the witness testify consistently with a previous statement to the authorities. ( Id. at p. 358, 41 Cal.Rptr.2d 635 , 895 P.2d 877 .) " ' "[A] defendant is denied a fair trial if the prosecution's case depends substantially on accomplice testimony and the accomplice witness is placed, either by the prosecution or the court, under a strong compulsion to testify in a particular fashion." [Citation.] Thus, when the accomplice is granted immunity subject to the condition that his testimony substantially *1086 conform to an earlier statement given to police [citation], or that his testimony result in defendant's conviction [citation], the accomplice's testimony is "tainted beyond redemption" [citation] and its admission denies defendant a fair trial. On the other hand, although there is a certain degree of compulsion inherent in any plea agreement *750 or grant of immunity, it is clear that an agreement requiring only that the witness testify fully and truthfully is valid.' " ( Ibid. )

Although Willis had stated that Berberian threatened to return him to San Quentin if he did not testify at Masters's trial, the referee found that Willis's statement was unconvincing. We agree with this finding, and no other credible evidence was presented showing that Berberian made such a threat. To the extent Masters contends that Numark promised Willis his release in exchange for the incriminating notes in Masters's handwriting, Berberian had made clear that there would be no agreement involving Willis's release.

Similarly, Evans stated at his deposition that he was threatened with being prosecuted in numerous cases and charged under a recidivist statute if he did not implicate Masters. As with Willis, the referee rejected Evans's statements as unbelievable. We have accepted this finding, and no other credible evidence was presented showing that Evans ever was improperly coerced into testifying against Masters.

2. Exculpatory material

Masters contends the prosecutors failed to disclose threats or promises of leniency made to Evans or other facts that might have affected his credibility with the jury.

" ' "In Brady [ v. Maryland (1963) 373 U.S. 83 , 83 S.Ct. 1194 , 10 L.Ed.2d 215 ], the United States Supreme Court held 'that the suppression by the prosecution of evidence favorable to an accused upon request violates due process where the evidence is material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.' [Citation.] The high court has since held that the duty to disclose such evidence exists even though there has been no request by the accused [citation], that the duty encompasses impeachment evidence as well as exculpatory evidence [citation], and **260 that the duty extends even to evidence known only to police investigators and not to the prosecutor [citation]. Such evidence is material ' "if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different." ' [Citation.] In order to comply with Brady , therefore, 'the individual prosecutor has a duty to learn of any favorable evidence known to the others acting on the government's behalf in the case, including the police.' " ' [Citation.] *1087 As such, the prosecutor has a duty to 'disclose to the defense and jury any inducements made to a prosecution witness to testify and must also correct any false or misleading testimony by the witness relating to any inducements.' [Citation.]"

"For a defendant to obtain relief under Brady , ' " '[t]he evidence at issue must be favorable to the accused, either because it is exculpatory, or because it is impeaching; that evidence must have been suppressed by the State, either willfully or inadvertently; and prejudice must have ensued.' [Citation.] Prejudice, in this context, focuses on 'the materiality of the evidence to the issue of guilt and innocence.' [Citations.] Materiality, in turn, requires more than a showing that the suppressed evidence would have been admissible [citation], that the absence of the suppressed evidence made conviction 'more likely' [citation], or that using the suppressed evidence to discredit a witness's testimony 'might have changed the outcome of the trial' [citation]. A defendant instead 'must show a "reasonable probability of a different result." ' [Citation.]" [Citation.] We independently review the question whether a Brady violation has occurred, but give *751 great weight to any trial court findings of fact that are supported by substantial evidence. [Citation.]' " ( Masters , supra , 62 Cal.4th at pp. 1066-1067, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .)

Preliminarily, the parties dispute whether Hahn was a member of the prosecution team. As we noted in Masters's automatic appeal, a prosecutor has a duty to learn of any possible inducements made by law enforcement officers or other agents of the state, provided that these agents are acting on the prosecutor's behalf in the case. (See Masters , supra , 62 Cal.4th at p. 1067 , 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) And we had assumed under these circumstances that Hahn was part of the prosecution team. ( Ibid. ) Moreover, the prosecutors' investigators knew at least some information about Evans, and they undoubtedly were members of the prosecution team. Thus, for Brady purposes, we consider the information about Evans to have been in the prosecutors' possession.

Masters contends the prosecutors failed to disclose three categories of information about Evans: (1) the prosecutors threatened Evans with a lengthy incarceration if he did not implicate Masters; (2) Evans was a suspect in Beasley's killing, with the implication that he was not prosecuted for that homicide in exchange for his testimony against Masters; and (3) Evans and Hahn had a pre-existing, ongoing working relationship, which included Hahn referring Evans to other government agencies for paid informant work, and that the extent of this relationship was greater than what was described at Masters's trial.

As to the first category, even assuming the agreement between Evans and Hahn contained any threats (or otherwise was coercive), the referee found the *1088 coercion or threats were already disclosed or discovered at Masters's trial and therefore not suppressed. (See People v. Morrison (2004) 34 Cal.4th 698 , 715, 21 Cal.Rptr.3d 682 , 101 P.3d 568 ["evidence that is presented at trial is not considered suppressed, regardless of whether or not it had previously been disclosed during discovery"].) The referee found no credible evidence of other, undisclosed threats, and we have accepted the referee's finding on this issue.

Next, with respect to Evans's possible involvement in Beasley's killing, Masters speculates that the police did not investigate Evans's involvement in exchange for his testimony and that the prosecutor failed to disclose this arrangement. Masters presented no evidence in support of this speculation, and multiple witnesses denied that such an arrangement existed. To the extent Masters contends that Evans's status as a suspect in a homicide case was by itself exculpatory for Masters because it implicated Evans's credibility, we are not persuaded. Even if **261 Evans's possible involvement in Beasley's killing was favorable to Masters and therefore should have been disclosed, it was not material under Brady because the jury was aware of Evans's extensive and violent criminal history. It is not reasonably probable that information concerning Evans's possible involvement in Beasley's killing would have altered the jury's assessment of his credibility or the weight to place on his testimony to such an extent that it would have produced a different trial outcome.

Similarly, with respect to Evans's relationship with Hahn, Masters fails to demonstrate that this information was material. In Masters's automatic appeal, we generally agreed with him that the prosecutor disclosed incomplete information about the extent of the agreement concerning the benefits Evans received to testify. ( Masters , supra , 62 Cal.4th at pp. 1064-1069, 199 Cal.Rptr.3d 85 , 365 P.3d 861 .) We held, however, that the jury *752 knew that Evans had testified in exchange for measures to protect his safety and that the undisclosed details of the arrangement were not material because there was no reasonable probability of a different result had the full extent of the agreement been disclosed. In addition, the jury was aware of Evans's negative character. And, as we have explained, the additional evidence of Evans and Hahn's pre-existing, ongoing relationship presented during the reference hearing does not materially alter the calculus.

Masters also contends that additional evidence of Evans and Hahn's relationship indicates that Evans had a motive to curry Hahn's favor for possible future benefits or consideration (and that this additional exculpatory evidence was not addressed in his automatic appeal). Initially, we doubt that Evans's expectations regarding future assignments as an informant induced him to testify against Masters; Evans appeared to have been motivated primarily by his desire to avoid being sent to state prison, as he feared the *1089 BGF would retaliate against him for providing information about its members to law enforcement. Even if we were to agree that Evans's testimony might have been motivated partially by his desire for future assignments, such additional motivation was not material under Brady because the jury already knew that Evans provided Hahn information in exchange for measures to protect his safety. It is not reasonably probable that this additional expectation of future benefits (not otherwise inferable from the evidence that was presented at trial) would have affected the jury's determination of Evans's credibility or the weight to place on his testimony to such an extent that it would have produced a different trial outcome.

3. Presentation of false testimony

Masters contends Berberian and Kamena knowingly presented Evans's false testimony. A defendant's due process rights are violated if a prosecutor knowingly presents false testimony or fails to correct such testimony after it has been elicited. (See People v. Vines (2011) 51 Cal.4th 830 , 874, 124 Cal.Rptr.3d 830 , 251 P.3d 943 .)

At the reference hearing, Masters conceded there was no evidence showing that the prosecutors knew that Evans's testimony was false. To the extent Masters contends that the prosecutors intentionally withheld exculpatory material about Evans, there was no evidence presented that the prosecutors personally knew of such information, and we previously rejected this claim based on their presumed duty to learn about such information. In any event, a prosecutor's presentation of conflicting evidence does not necessarily mean that a prosecutor has presented false evidence. So long as impeachment material is not concealed, a prosecutor may present conflicting testimony and let the jury make a determination as to the witnesses' credibility. (See People v. Letner and Tobin (2010) 50 Cal.4th 99 , 167, 112 Cal.Rptr.3d 746 , 235 P.3d 62 .)

CONCLUSION

Based on our acceptance of most of the referee's findings, we hereby discharge the order to show cause because Masters has not met the applicable standards for relief under any claim raised in his habeas corpus petition and referenced in our order to show cause.

Because our order to show cause and reference order were limited to these questions **262 and claims, we do not address any other claim raised in the petition for writ of habeas corpus. The remaining claims will be resolved by a separately *753 filed order. (See Cowan , supra , 5 Cal.5th at pp. 248-249.)

We Concur:

CANTIL-SAKAUYE, C. J.

CHIN, J.

CORRIGAN, J.

CUÉLLAR, J.

KRUGER, J.

GROBAN, J.

Concurring Opinion by Justice Liu *1090 Habeas corpus is a limited remedy against a presumptively valid final judgment. The availability of this remedy is governed by judicially articulated standards and, as relevant here, by statutes applicable to claims involving false evidence or new evidence. False evidence must be "substantially material" to warrant habeas corpus relief. ( Pen. Code, § 1473, subd. (b)(1) ; see In re Figueroa (2018) 4 Cal.5th 576 , 589, 229 Cal.Rptr.3d 673 , 412 P.3d 356 ["Materiality is shown if there is a reasonable probability the result would have been different without the false evidence."].) Statutory relief based on newly discovered evidence is available only if the evidence is "of such decisive force and value that it would have more likely than not changed the outcome at trial." ( Pen. Code, § 1473, subd. (b)(3)(A).) As today's opinion explains, petitioner Jarvis Masters has not met those standards.

Masters contends that the referee's findings cast serious doubt on the credibility of the prosecution's two main witnesses, Rufus Willis and Bobby Evans, and thereby undermine confidence in their trial testimony. As the referee put it, both are " 'liars with highly unreliable and selective memories. [¶] ... [¶] Evans and Willis are utterly lacking in credibility. Both are career criminals whose word, under oath or otherwise, means nothing. Both are well-known snitches. Both would say anything to save their own hide - and both have so admitted. Both are manipulative and unreliable.' " (Maj. opn., ante , 250 Cal.Rptr.3d at p. 734, 446 P.3d p. at 246.) But the jury heard both witnesses and made its own credibility determinations, and the judgment at trial is entitled to a presumption of finality. The standards for habeas corpus relief are crafted with this presumption in mind. Still, it is understandable why Masters finds the referee's report unsettling.

The denial of relief in this matter follows from the statutory standards of review applicable to claims of false evidence and newly discovered evidence. We have no occasion in this posture to consider whether, in light of the trial evidence as well as the reference hearing and findings, we can be confident of the verdict beyond a reasonable doubt. Nor do we have occasion here to consider whether, in light of all relevant circumstances, the fact that Masters was sentenced to death - while his codefendants Andre Johnson (the actual killer) and Lawrence Woodard (a prison gang "lieutenant" who, according to one witness, "had given the order for Burchfield's murder" ( People v. Johnson (1993) 19 Cal.App.4th 778 , 780, 783, 23 Cal.Rptr.2d 703 )) were not - may be indicative of arbitrariness in the application of the death penalty.

I Concur:

CUÉLLAR, J.