I dissent.
In my view, the majority opinion is indefensible in two respects. First, in People v. Gonzalez the majority improperly uphold the death penalty judgment in the face of a record demonstrating perhaps the clearest violation of the principles set out in People v. Brown (1985) 40 Cal.3d 512 [220 Cal.Rptr. 637, 709 P.2d 440] that has ever come before this court. Second, the majority opinion in the consolidated matters of People v. Superior Court (Gonzalez) and In re Gonzalez effectively frustrate defendant’s attempt to discover documents in the possession of the County of Los Angeles which may demonstrate that a key prosecution witness committed perjury.
I.
Defendant in People v. Brown, supra, 40 Cal.3d 512, challenged the constitutionality of the jury instructions under the 1978 law. Although we ultimately rejected his contention, we observed that an instruction tracking the language of the statute, such as that given here, was ambiguous, and “leave[s] room for some confusion as to the jury’s role” in determining the appropriate penalty. (Brown, supra, 40 Cal.3d at p. 544, fn. 17.) In view of the dangers posed by this instructional ambiguity, we said that we would examine each case tried prior to Brown “on its own merits to determine whether, in context, the sentencer may have been misled to defendant’s prejudice about the scope of its sentencing discretion under the 1978 law.” (Ibid.) We have undertaken this review in numerous cases. In most, the only clue as to whether the jury may have been misled is the arguments of counsel, and our analysis, as it must in this case, has focused on whether counsel put before the jury an incorrect interpretation of the instruction. While prosecutorial arguments “are not to be judged as having the same force as an instruction from the court” (Boyde v. California (1990) 494 U.S. 370,_ [108 L.Ed.2d 316, 331, 110 S.Ct. 1190, 1200]), when the instruction from the court is ambiguous we may have no choice but to look to the argument to discern how that instruction was likely perceived by the jury.1
In People v. Allen (1986) 42 Cal.3d 1222, 1276-1277 [232 Cal.Rptr. 849, 729 P.2d 115], we explained that: “Our concern in Brown was that the *1271unadorned statutory instruction might in two interrelated ways lead the jury to misapprehend its discretion and responsibility, [fl] First, we pointed out that the jury might be confused about the nature of the weighing process. As we observed: ‘[T]he word “weighing” is a metaphor for a process which by nature is incapable of precise description. The word connotes a mental balancing process, but certainly not one which calls for a mere mechanical counting of factors on each side of the imaginary “scale,” or the arbitrary assignment of “weights” to any of them. Each juror is free to assign whatever moral or sympathetic value he deems appropriate to each and all of the various factors he is permitted to consider.’ [Citation.] [1|] Second, we were concerned in Brown that the unadorned instruction’s phrase, ‘the trier of fact . . . shall impose a sentence of death if [it] concludes that the aggravating circumstances outweigh the mitigating circumstances,’ could mislead the jury as to the ultimate question it was called on to answer in determining which sentence to impose. Although the quoted phrase could be understood to require a juror (i) to determine whether ‘the aggravating circumstances outweigh the mitigating circumstances’ without regard to the juror’s personal view as to the appropriate sentence, and then (ii) to impose a sentence of death if aggravation outweighs mitigation even if the juror does not personally believe death is the appropriate sentence under all the circumstances, we concluded in Brown that the statute was not intended to, and should not, be interpreted in that fashion.”2
Two recent five-to-four decisions of the United States Supreme Court, Boyde v. California, supra, 494 U.S. 370, and Blystone v. Pennsylvania (1990) 494 U.S. 299 [108 L.Ed.2d 255, 110 S.Ct. 1078], suggest that brown’s interpretation of the 1978 law was not compelled by the United States Constitution. In light of those decisions, we requested the parties to brief and argue the question whether we should reconsider our decision in People v. Brown, supra, 40 Cal.3d 512. Upon further reflection, however, we decided it was not necessary to reconsider Brown, and both the majority and this dissenting opinion follow the Brown analysis.
Brown continues to serve an essential function in cases arising under the 1978 death penalty law. The ambiguous wording of that law permits prosecutors to weave a web of false logic designed to lead jurors to believe that the structure of the law requires a death verdict whether or not that punishment is appropriate. Brown’s interpretation of the 1978 law guards against such mistaken reasoning and helps to ensure that a penalty verdict under the law is the result of a considered moral determination, rather than a mechanical or arithmetical process devoid of moral considerations. Thus *1272the present jury instructions (CALJIC No. 8.88), based on Brown, continue to govern the trial of capital cases in California.
I do not believe that we can adhere to the principles established in Brown, supra, 40 Cal.3d 512, without reversing the penalty judgment in the case before us. In my opinion, this is one of the most flagrant examples of Brown error to come before this court. Interpreting the ambiguous instructions given by the trial court, the prosecutor told the jury: (1) they could determine the penalty by the arithmetical process of assigning arbitrary weights to aggravating and mitigating factors and then adding the weights in each column; (2) inapplicable mitigating factors must be placed in the column for aggravation and assigned weight; and (3) in weighing the factors, the jurors should not consider whether death is the appropriate penalty for defendant, or how his case compares to other cases in which death may have been more appropriate. If we are to preserve Brown as living doctrine instead of a museum exhibit, we must reverse this verdict.
Brown first warned of an interpretation which would call “for a mere mechanical counting of factors on each side of the imaginary ‘scale,’ or the arbitrary assignment of ‘weights' to any of them.”3 (Brown, supra, 40 Cal.3d at p. 541, italics added.) We have not yet encountered a case in which anyone suggested that the jury decide the penalty by “counting” the factors on each side, although the prosecutor in People v. Hamilton (1989) 48 Cal.3d 1142, 1181-1182 [259 Cal.Rptr. 701, 774 P.2d 730], came close. We have encountered one case in which the prosecutor suggested that a verdict could be reached by the arbitrary assignment of numerical weights to the factors on each side, and we unanimously condemned that argument. In People v. Bittaker (1989) 48 Cal.3d 1046, 1108 [259 Cal.Rptr. 630, 774 P.2d 659], the prosecutor told the jury “if we were to give actual weight in pounds and ounces to the aggravating circumstances and the mitigating circumstances, if the aggravating circumstances weighed 10 pounds and one ounce and the mitigating circumstances weighed 10 pounds, then you would be duty bound to impose a death penalty .... [^f] If you were to give a percentage to it, if you said 50.1 percent of the evidence pointed to aggravating circumstances and 49.9 pointed to mitigating circumstances, then you’d still have to impose a sentence of death.” We found this argument erroneous, and inconsistent with precedent, “because it depicts the weighing process as one involving the application of an arithmetical formula involving the assignment of weights to each of the factors, followed *1273by an addition of the entries in each column to determine the balance.” (Ibid.)
Compare what the prosecutor said in the case at bar. After reviewing the facts of the case, he told the jury:
“I—I built a little scale. That’s something I had in my garage. It’s not very professional, as you can see, but the idea is to give you a concept, I’m giving you an idea . . . . [fl] You’ve got the first job, to take the evidence and decide which side does it go on? Is it aggravating or is it mitigating? [fl] After you decide that, then you have the next job, to decide how much weight am I going to give to it? . . . [fl] That is your job. You have got to give the weight to them. A little weight, a lot of weight. That’s your decision. So schematically I couldn’t do it without giving it equal weight.”
After classifying the various factors as aggravating or mitigating—mislabeling many of them as will be seen later—the prosecutor in his closing argument turned to the weight to be assigned each factor:
“You have got one column for aggravation; you have got another column for mitigation and let’s say you—you are doing your job as a juror. You are trying to decide where they fit and how much weight to give to them .... [fl] Let’s do it this way. Let’s say that you can give a number of 1 to 10, 1 to 10 for each one of these. 1 to 10.
“How about the killing? The nature of the crime? What are we going to consider that? I mean, that was cold. That was planned .... On a scale of 1 to 10, that would be, let’s say, a number 9. Couldn’t get much more aggravating than that, the manner in which it was done.
“Next, prior violence. As I have indicated, a conviction for great bodily injury, Mr. Bencangey [defense counsel] says, ‘Well, but there’s no evidence that he actually did great bodily injury.’ Well, that’s—that was not the point. He was convicted of the crime of. He was convicted of the crime of battery. I mean, how many of you or how many of your friends, how many of your relatives have been convicted of both those things? I mean, that does show that a person is a violent person and has had violent criminal activity. Now, on a scale of 1 to 10, maybe that would be a 5 ... .
“Okay. Now, we have no evidence of mental disturbance. Let’s give that a 1.
“No victim participation. Give it a 1.
*1274“No moral justification. Now, Mr. Bencangey wants to argue that one and say, ‘Well, there might be some extenuating circumstances,’ but until he gives you some hard evidence of that, let’s give that a 1.
“There is no distress. [Sic —He probably meant “duress.”] Give that a 1.
“There is no evidence of intoxication. Give that a 1.
“The defendant is not of tender age, certainly doesn’t fit on mitigation, it fits on aggravation. Give it a 1.
“All of these are the benefit of the doubt. I’m giving them the lowest possible.
“The defendant was a principal in this. He was not the accessory. Let’s give that a 1.
“Now, this one [(apparently Pen. Code, § 190.3, factor (k))], we are still waiting on and maybe Mr. Bencangey can change all of this by indicating that that should be on this side, but there was no extenuating circumstances. We will give that a 1.
“And then we have, yeah, on the mitigation side, we have got prior felony. There is no prior felony. Let’s say give that a 3.
“Okay, now, Mr. Bencangey is going to say afterward, ‘Wait just a minute. How did Mr. Bowers [the prosecutor] arrive at these?’ I’m giving you an example. I’m not telling you that these are numerical, that this is the way you’re going to do it. I’m just trying to get across concepts and the only way I know how to do it is to give illustrations. Your numbers may be entirely different. You may approach it differently, but you have got to think of it in a logical, and I know that you will think of it in a logical manner, of what is aggravating and what is mitigating.
“When you total it up—”
Several jurors: “22.”
Mr. Bowers: “Oops. Okay. I didn’t say that I excelled in math. 22 versus 3. You change the numbers however you want to change them and you figure out how you can get it to come out so that the mitigation is greater than the aggravation. You will not. You will not.”
I submit that it is clear that the prosecutor put before the jury an interpretation of the instructions which we have repeatedly said was incorrect. *1275He told them that the process was essentially an arithmetical process; that a logical, legitimate way to reach a verdict is to determine whether each factor is aggravating or mitigating, assign it a numerical value, total those values, and see which column is the larger. We should not hesitate to condemn this as a misleading description of the jury’s role.4
The majority nevertheless assert that the jury could not have been misled. But they provide no explanation how they know that the jury did not, or could not, have decided the case by the means the prosecutor suggested. Instead, they advance arguments which miss the point.
First, the majority say, the jury was told that an attorney’s argument is not evidence. Of course. Argument is never evidence, and if that fact precludes reversal, we could never reverse for Brown error, and the case-by-case examination we promised in Brown would be meaningless. Juries do not decide cases on the evidence alone; they decide them by applying the law to the evidence. It has never before been thought that a misunderstanding as to the applicable law is less important than an error in the admission or exclusion of evidence.
Next, they assert, the jury was told to “weigh,” not “count,” the factors. Brown spoke of two ways in which the instruction on weighing could be misunderstood: that the instruction could be read to call “for a mere mechanical counting of factors on each side of the imaginary ‘scale,’ or the arbitrary assignment of ‘weights’ to any of them.” (Brown, supra, 40 Cal.3d at p. 541, italics added.) That the prosecutor avoided the first mistake does not mean that he avoided the second.
Finally, although the gist of the prosecutor’s argument was misleading, the majority say that he did refer to normative judgment when, in the last sentence of his rebuttal argument, he told the jury that death was the “just and fair” penalty defendant had earned. Do the majority believe that by this single phrase, after hours of argument, the prosecutor rejected the idea which he had just explained at great length and illustrated with scales and a chart—that a verdict could be reached by assigning numerical values to the various factors and adding them? If we followed the prosecutor’s methodology and placed on one side of a scale those remarks which support a *1276mechanical, arithmetical process of reaching a verdict, and on the other those remarks supporting a normative approach, the first would far outweigh the second. Or if we simply listen for the essence of the prosecutor’s message, that too is clear: if you put the factors in the columns, assign them relative numerical values, and add the columns—and do not concern yourself with such matters as sympathy and appropriateness—you cannot avoid arriving at a verdict of death. Whether defendant deserves that verdict or not, he does not deserve a verdict arrived at in that way.
Brown, supra, 40 Cal.3d 512, also warned of any interpretation that might mislead a juror “to impose a sentence of death if aggravation outweighs mitigation even if the juror does not personally believe death is the appropriate sentence under all the circumstances.” (People v. Allen, supra, 42 Cal.3d 1222, 1277.) The majority appear to agree that if the jury is to decide penalty by simply balancing aggravation against mitigation, it is essential that they consider the appropriateness of the penalty in striking that balance. (Maj. opn., ante, p. 1230.)
Here is what the prosecutor had to say on this subject. In his closing argument, he proposed an arithmetical model for determining penalty, then said, “When all of these [factors] are added up, . . . you weigh them. If the aggravating factors outweigh the mitigating factors, the sentence is death. There’s no—alternative for you. If they outweigh, it is automatically death. [If] On the other hand, if mitigation outweighs aggravation, it is automatically life imprisonment without possibility of parole. []f] So there is no option, once you do your calculations.”
There is nothing here to suggest that the jury considered the appropriateness of the death penalty at any stage of the analysis.5 And in rebuttal, the prosecutor made that clear: “[T]he defense claims that circumstances [were] not aggravated. []f] What did I tell you? He uses the example of other brutal, vicious crimes, talks about the death penalty, when is the death penalty appropriate? [fl] You didn’t hear His Honor say anything about *1277when it is appropriate, [fl] His Honor said your job is a weighing process. It’s not for you to think, ‘Well, I think this is an appropriate-type case; no, this is not an appropriate-type case.’ []f] That is not for you to decide. Not really. It’s kind of misleading. That’s misleading because if it’s an appropriate-type of situation then it becomes a comparison issue, [^j] Are you here to compare this with the Manson case, the Bittaker case, to other atrocious crimes? [fl] No, you are not.”
The majority generously suggest that the prosecutor’s rebuttal remarks were addressing only a question whether comparison to other well-publicized murders was relevant.6 But while that may be the context of his remarks, their scope was not so limited: “You didn’t hear His Honor say anything about when it is appropriate.” “It’s not for you to think, ‘Well, I think this is an appropriate-type case; no, this is not an appropriate-type case.’ That’s not for you to decide.” The prosecutor never told the jury that his objection to considerations of appropriateness involved only the comparing of this case to other cases. He never said that in the case at hand, the jury should determine whether death was the appropriate penalty. Had he intended his remarks to be so construed, he would never have said without qualification that the jury is not supposed to decide whether this is “an appropriate-type case.”
Having told the jury that they could decide the case by placing the factors in the aggravating and mitigating columns and assigning arbitrary weights, without any consideration of appropriateness, the prosecutor compounded the error by explaining erroneously how factors are classified and weighed. The majority recognize this error, noting that “[i]t does appear that the prosecutor placed on the aggravating side of his ‘scale’ and ‘balance sheet’ the absence of such factors as ‘extreme’ mental or emotional disturbance ([Pen. Code,] § 190.3, factor (d)); victim participation or consent (id., factor (e)); belief in moral justification (id., factor (f)); extreme duress (id., factor (g)); mental disease or intoxication (id., factor (h)); mere minor participation (id., factor (j)); and other mitigating evidence (id., factor (k)). In his rebuttal argument, the prosecutor’s ‘score’ of 22 aggravating points to 3 mitigating points was based on the assumption that the absence of prior *1278felony convictions (id., factor (c)) was the only mitigating factor; he deemed all other statutory factors to be aggravating, [f ] The prosecutor thus misrepresented the sentencing formula. The mere absence of extenuating circumstances in the case cannot weigh in favor of death.” (Maj. opn., ante, pp. 1233-1234.)7
But even though the prosecutor misrepresented the sentencing formula, the majority assert that the jury was not misled. The jury, they say, was instructed to consider each sentencing factor only “if applicable”; they ignore that the prosecutor told the jury, incorrectly, that all factors were applicable and must be given weight. The majority also claim the prosecutor assigned only nominal weights to the seven absent extenuating factors. He actually assigned each a weight of “1.” He assigned a weight of “5” to the prior violent misdemeanors, and a weight of “3” to the absence of prior felony convictions. Thus, under this arithmetical analysis the combined weight of the absent extenuating factors was more than the weight assigned to prior violent crimes, and more than twice the weight given the absence of prior felony convictions.
I cannot share the confidence of the majority that a jury, given ambiguous instructions and presented with an argument as completely and thoroughly misleading as the prosecutor’s argument here, escaped unscathed. Not only is there a reasonable likelihood that the jury misunderstood the way in which it was to determine penalty, there is also a reasonable likelihood that this misunderstanding prejudiced defendant. While it is clear that defendant killed Deputy Williams, the appropriate punishment was a close and difficult decision. The first jury deadlocked at the penalty trial. Defense counsel at the habeas corpus hearing testified that jurors told him the deadlock resulted from doubts concerning the extent of defendant’s culpability.
The record justifies such doubts. Defendant claimed that he thought the intruders were not police but members of a rival Hispanic gang. This story has some plausibility defendant was a gang leader and a likely target; the police were not in uniform and arrived in unmarked cars. It also has some weaknesses: the police were older than typical gang members, mostly White (although the first to enter was Hispanic), and announced their presence. *1279There was disputed testimony whether forcible entry into a residence was a common or conceivable gang tactic. But the prosecutor’s theory that defendant was tipped off by someone in the police department and knew in advance of the police raid, that he hid the narcotics, and then waited for his chance to “bag” a police officer—also has weaknesses. We have no corroborating evidence that anyone in the police department has been giving suspects advance warnings of police raids. It also seems improbable that a drug dealer who had successfully concealed the incriminating evidence would still court a violent confrontation with the police, and incredible that he would do so armed with only a single-shot weapon.
In deciding whether to believe the prosecution’s theory or the defendant’s testimony, the critical evidence was the testimony of defendant’s cellmate, William Acker. He said that defendant told him of defendant’s plan to kill a policeman and get off by claiming he thought the man was a member of a rival gang. But as the prosecutor explained, Acker’s testimony was credible not because of Acker’s character, which was despicable, but because Acker knew details of the crime which he could learn only from defendant or police reports. As explained in part II of this opinion, however, we now know how easy it was for inmates of the Los Angeles County jail to obtain such detailed information about their cellmates’ crimes from the police. Acker’s testimony, suspect when defendant was tried, would be even more suspect today.
There was virtually no penalty phase evidence apart from that relating to the circumstances of the crime. The defense proved that defendant had no prior felony convictions. The prosecution proved two misdemeanor convictions involving violence, but did not introduce evidence of the conduct on which they were based. This was, in short, not a case in which a death verdict was obvious, but the kind of case in which the jury might be persuaded to vote for either life or death, and the instructions and argument could make the difference. In my judgment, the grossly misleading explantion of the penalty determination process which the prosecutor argued to the jury should compel us to reverse the penalty judgment.
II.
In October 1988, after defendant was convicted and after his original and supplemental petition for habeas corpus were filed with this court, a newspaper article in the Los Angeles Times revealed a widespread practice of perjury by inmate informers held in the Los Angeles County jail. The inmates in question would manufacture false confessions by fellow inmates, report these confessions to law enforcement personnel, and in some cases testify to those confessions at trial. In return, they would receive some sort *1280of benefit in terms of reduced prosecution, reduced sentence, or better jail conditions. Acker was among those persons who on numerous occasions reported and testified to confessions by fellow prisoners, and received benefits in return.
Defendant has maintained from the date of trial that Acker’s testimony was peijured, but he has not been able to present conclusive evidence to corroborate his claim. He now seeks discovery of Acker’s law enforcement files. This is not what is sometimes referred to pejoratively as a “fishing expedition”; we know the files exist, and probably contain material evidence.8 What defendant does not know is the nature of that evidence—that is why he needs to see the files.
The majority never dispute that defendant is entitled to the information. To the contrary, they suggest that the prosecutor has an ethical duty to disclose it. Neither do they dispute the contention that aggressive legal intervention is essential to redress a widespread fraud on the courts perpetrated by inmate informers with police and prosecutor complicity. But when it comes down to the nitty-gritty of finding a way to enforce the ethical obligation of disclosure or to conduct the inquiry into possible perjury, whatever defendant tries, the majority find a technical barrier.
Specifically, the majority conclude that (a) defendant cannot get discovery in connection with his automatic appeal because the appeal is limited to the appellate record; (b) he cannot get discovery in an independent trial court action because discovery must be ancillary to a pending proceeding; (c) he cannot get discovery in connection with his pending habeas corpus petition because the issue of Acker’s perjury is not within the scope of the order to show cause; and, (d) even taking into account new allegations of Acker’s perjury and new evidence which gives plausibility to those allegations, he cannot obtain discovery by filing a new habeas corpus proceeding alleging Acker’s perjury because, without discovery, he cannot allege sufficient facts proving Acker’s perjury to state a prima facie case. In short, the majority pose a perfect “Catch-22” logical conundrum, under which defendant cannot obtain discovery of the law enforcement records because without the information contained in those records he cannot file an action that would support discovery.
A. Discovery in the Trial Court.
The superior court granted defendant’s motion for discovery of the records in question. The district attorney petitioned for mandate from this *1281court. The majority hold that mandate should issue to nullify the superior court’s order. To understand the blindness which afflicts the majority’s view of this proceeding, one must appreciate the extraordinary circumstances which impelled the trial court to act as it did.
On October 27, 1988, a newspaper article appeared in which inmate Leslie White described how inmate-informers concocted false but convincing confessions implicating other prisoners, testified falsely against those prisoners, and in return received special benefits. As described in that article and on subsequent occasions, the inmate would arrange to be confined or transported with the suspect, so he could show he had the opportunity to hear a confession. (Sometimes the jail officials would facilitate this by deliberately housing the suspect in the portion of the facility where known informers are confined for their own protection.) The inmate informant would phone law enforcement personnel, pose as a fellow investigating officer, and learn details of the crime. Since those details were known only to law enforcement personnel (who presumably would not disclose them to a prisoner) and to the criminal himself, the informant, by including such details in a false confession, could give his story a spurious air of authenticity. The inmate would then report the false confession to the police and, if requested, testify to it in court. In some cases inmates bargained for specific benefits, but this was not an essential part of the system; the inmates knew that if they regularly came forward with useful information they would be rewarded.
When this system of perjury was exposed, the district attorney began an investigation of all cases in which a jailhouse informant was used. At the district attorney’s request, the deputy who tried this case submitted a memorandum in which he noted that before the Gonzalez trial Acker had “testified previously once or twice in other cases,” and since that trial he had testified in other cases.
On November 17, 1988, the district attorney issued special directive 88-14, which said: “We must also address those cases where jailhouse informants have been used in the past; the most objective way to do this is on a case by case basis before the court. This will insure an independent review on the merits of each case. Accordingly, we will begin immediately to notify the attorney of record in each case in which a jailhouse informant testified. The attorney will be advised of the information we have received and encouraged, where appropriate, to make a motion to bring the matter before the court . . . .”
Pursuant to this directive, the district attorney sent defense counsel in this case a letter which said that his office would undertake an investigation *1282of all jailhouse informant cases, and that the information would be “made available to defense counsel on individual cases” and “in appropriate cases ... be thoroughly aired in open court.” The letter went on to state: “We anticipate that review will be conducted by several judges listening to testimony elicited by the prosecution and the defense. That forum provides opportunity for the most effective method of getting at the facts: examination of witness under oath and the compulsory production through subpoena of documents . . . .”
Information revealed in the trial of this case, and subsequently, shows that Acker was an informant-witness in three cases before defendant’s trial, that he was engaged in providing information to prosecutors in two other cases at the time of the trial, and thereafter continued regularly to provide testimony against fellow inmates. It also shows that Acker received benefits for providing information and testimony. But it does not show whether his testimony in this or any other case was false.
In January of 1989 the Honorable Otto M. Kaus, a former justice of this court, was appointed to head a grand jury investigation. The grand jury heard testimony from six informants, and investigators talked to nineteen others. Its report is replete with detail about the informant practices in question, but omits all names. We do not know, for example, whether Acker was among the informants who testified or were interviewed.
In the meantime, recognizing that trial counsel were under no duty to bring actions on behalf of defendants whose cases were on appeal, the district attorney and the defense bar requested the court to appoint counsel for that task. The trial court thereupon appointed defendant’s appellate counsel to represent him in trial court proceedings to discover whether his conviction or sentence was the result of perjured informer testimony.
Counsel accordingly moved for discovery of law enforcement files relating to Acker. Although the district attorney had previously offered to make such information available to defense counsel, he opposed the motion. Noting the critical importance of Acker’s testimony to the finding of special circumstance and the penalty judgment, Superior Court Judge Cianchetti granted the motion in part.
The district attorney then brought the present mandate petition, claiming that the trial court had no jurisdiction to grant discovery when no action was pending before that court. The majority endorse this position.
In choosing to analyze this proceeding by addressing only the narrow question whether the trial court ordinarily has jurisdiction in a case pending *1283on appeal, the majority miss the broad, extraordinary question posed by the facts of this case. That question is the authority of a trial court to inquire into charges that it has been the victim of widespread fraud, that an entire class of cases may have been tainted by a pattern of perjury, and to determine which cases have been so affected. The limiting rules cited by the majority are inadequate to this situation. They would compel the court to fragment its inquiry, depending upon whether a case was awaiting trial, on appeal, or already final, and in many instances bar inquiry at the threshold because the defendant does not already know the facts the inquiry might reveal.
I firmly believe that it is within the inherent power of any court to conduct an inquiry into well-founded charges that its proceedings in a large number of cases have been tainted by a pattern of perjury, and, to facilitate that inquiry, to order the prosecutor to disclose material information to likely victims of the perjury. “A court set up by the [California] Constitution has within it the power of self-preservation, indeed, the power to remove all obstructions to its successful and convenient operation. This arises from the fact that it is part of and belongs to one of the three independent departments set up by the Constitution.” (Millholen v. Riley (1930) 211 Cal. 29, 33-34 [293 P. 69].) That power includes the power to “cure abuses or overreaching involving confidential information . . . .” (Peat, Marwick, Mitchell & Co. v. Superior Court (1988) 200 Cal.App.3d 272, 286-287 [245 Cal.Rptr. 873].) It includes the power “to prevent abuse of its process, and to create a remedy for a wrong even in the absence of specific statutory remedies.” (Western Steel & Ship Repair, Inc. v. RMI, Inc. (1986) 176 Cal.App.3d 1108, 1116 [222 Cal.Rptr. 556].)
The decision of the Tenth Circuit in Hopkinson v. Shillinger (10th Cir. 1989) 866 F.2d 1185, 1220-1221, demonstrates the power of a court in somewhat similar circumstances to grant discovery despite the absence of a pending action. In that case the defendant had been convicted of murder; his conviction had been affirmed by the Wyoming Supreme Court, and was final. A grand jury, however, had continued to investigate the murder. The defendant asserted that it had heard evidence tending to exonerate him, but he could not point to any specific evidence. The circuit court ruled that since the defendant had shown a “particularized need” for the grand jury transcript (see Dennis v. United States (1966) 384 U.S. 855, 870 [16 L.Ed.2d 973, 984, 86 S.Ct. 1840]), he was entitled to an order directing the federal district court judge to review the transcript in camera and to reveal any exculpatory evidence—even though the defendant had no independent *1284proceeding pending.9 This holding indicates that defendant Gonzalez is entitled to discover the evidence presented to the grand jury—which may be of some use—but it goes further: it suggests that if a defendant is entitled to discovery of evidence in the hands of the state, he can bring an independent action to enforce that right even if no appeal or habeas corpus proceeding is pending.
People v. Ainsworth (1990) 217 Cal.App.3d 247 [266 Cal.Rptr. 175], the case on which the majority rely, actually supports only the proposition that the trial court loses its jurisdiction to entertain a postjudgment discovery motion when remittitur has issued following an unsuccessful appeal and the defendant has not filed a petition for habeas corpus, so there is no pending proceeding which could result in a new trial. The majority approve of the Ainsworth court’s reasoning that “a discovery motion is not an independent right or remedy. It is ancillary to an ongoing action or proceeding. After the judgment has become final, there is nothing pending in the trial court to which discovery may attach.” (217 Cal.App.3d at p. 251.) But what the majority overlook is that the Ainsworth court stressed that it was the finality of the judgment and the lack of any pending “collateral attack” which robbed the trial court of its jurisdiction to entertain the discovery motion at issue. Here, no remittitur had issued; defendant’s appeal and habeas corpus petition were both pending at the time he made his discovery request.
As the court held in Wisely v. Superior Court (1985) 175 Cal.App.3d 267, 269 [220 Cal. Rptr 893], the fact that his appeal was pending in this court at the time defendant filed his discovery request would not deprive the trial court of its jurisdiction to hear the merits of the request. Under California Code of Civil Procedure section 916, subdivision (a), “the perfecting of an appeal stays proceedings in the trial court upon the judgment or order appealed from or upon the matters embraced therein or affected thereby, including enforcement of the judgment or order, but the trial court may proceed upon any other matter embraced in the action and not affected by the judgment or order.” (Italics added.) The purpose of section 916 is to “protect the jurisdiction of the appellate court; the rule prevents the trial court from rendering the appeal futile by changing the judgment into something different. Accordingly, whether a matter is ‘embraced’ in or ‘affected’ by a judgment. . . depends on whether postjudgment proceedings on the matter would have any effect on the ‘effectiveness’ of the appeal.” (In re Marriage of Horowitz (1984) 159 Cal.App.3d 377, 381 [205 Cal.Rptr. 880], italics added and citations omitted; see also 9 Witkin, Cal. Procedure (3d ed. 1985) Appeal, §9, p.40.)
*1285The granting or denial of a discovery request in itself will have no impact on the effectiveness of the appeal. The best that can be said for the majority’s position is that the granting of the request might lead to the discovery of evidence which might lead to the amending of a habeas corpus petition pending in this court, which in turn might lead to a hearing on the habeas corpus petition which might moot all or part of the appeal. This is far too remote a danger to divest the trial court of jurisdiction under Code of Civil Procedure section 916, subdivision (a).
Thus the trial court’s order should be sustained under Code of Civil Procedure section 916, subdivision (a). But I would not rest our decision on so narrow a basis. Even if defendant’s appeal were final, and no collateral action were pending, the court should still have jurisdiction to inquire into well-founded charges that its decisions have been tainted by a pattern of inmate perjury and government complicity, and to grant discovery orders incident to that inquiry.
B. Discovery in Connection With Habeas Corpus.
Defendant filed a petition for habeas corpus with this court in 1984. We issued an order to show cause. Defendant filed a supplemental petition in 1986, which served also as a traverse to the return to the 1984 petition. We issued an amended order to show cause on the supplemental petition, but that order did not include the question whether Acker acted as a government agent in eliciting incriminating statements from defendant, rendering Acker’s testimony inadmissible under Massiah v. United States (1964) 377 U.S. 201 [12 L.Ed.2d 246, 84 S.Ct. 1199] and United States v. Henry (1980) 447 U.S. 264 [65 L.Ed.2d 115, 100 S.Ct. 2183]. The majority therefore conclude that defendant cannot obtain discovery of Acker’s file in connection with the habeas corpus proceeding pending here.
In the amended habeas corpus petition he filed in 1986, defendant alleged that Acker was a state agent, rewarded by the law enforcement system for eliciting information from fellow inmates; that he had a history of informing prior to the time he informed on and testified against defendant; that he had prior relationships with law enforcement officials and that they encouraged him to elicit information; that he was ostensibly no more than a fellow inmate; that he may have been deliberately placed in a cell next to defendant; and that he was not merely a passive listening post but, as Acker admitted, deliberately initiated conversations with defendant. (Compare Kuhlmann v. Wilson (1986) 477 U.S. 436 [91 L.Ed.2d 364, 364-365, 106 S.Ct. 2616]; People v. Whitt (1984) 36 Cal.3d 724 [205 Cal.Rptr. 810, 685 P.2d 1161].) The habeas corpus petition alleged that although Acker did not admit to being an informant when he testified in defendant’s guilt phase *1286trial, he had informed in at least three or four other cases before he testified against defendant.10 By the time of defendant’s first penalty trial some four months later, the petition alleged, Acker had given information in at least three more cases.11 Additionally, the petition alleged, Acker sought, was promised and received numerous benefits for informing, despite his testimony at the guilt phase and first penalty phase trials that he had received no promises in exchange for his testimony and expected no consideration. Those benefits allegedly included, inter alia, immunity on murder, kidnapping, robbery and grand theft charges pending against him, removal of tattoos and possibly other plastic surgery to remove identifying marks, and protective transfer to a facility where he would be safe from inmates who might attack him for being a “snitch.” Finally, the petition alleged that Acker had testified falsely, and that representatives of the state were aware of the perjured character of Acker’s testimony.
To state a prima facie basis for relief under Massiah v. United States, supra, 377 U.S. 201, and United States v. Henry, supra, 447 U.S. 264, a petitioner must show that a government agent—a prisoner acting under state direction or inducement—deliberately elicited incriminating statements from the defendant, and that such statements were prejudicial. (See People v. Hovey (1988) 44 Cal.3d 543, 559-561 [244 Cal.Rptr. 121, 749 P.2d 776] and cases there cited.) To state a prima facie case for relief based on the knowing use of perjured testimony, a petitioner must show that “[fjalse evidence that is substantially material or probative on the issue of guilt or punishment was introduced” (Pen. Code, § 1473, subd. (b)(1)) and that the perjury was prejudicial.12 (In re Wright (1978) 78 Cal.App.3d 788, 808 [144 Cal.Rptr. 535].) We could have found defendant’s 1986 petition stated a prima facie case on both grounds, but were led astray by our ignorance of the practices of informers in the Los Angeles County jail, and by the manner in which the allegations were set out in the petition. The petition put the Massiah/Henry claim in terms of a failure of counsel to investigate, but the allegations showed no significant deficiency. In hindsight, it is now clear that the pattern of informer perjury in the jail system was clever enough to deceive even a competent investigator. The petition also alleged *1287that Acker had informed on many inmates and received rewards, but it failed to allege specifically that any state representative had asked or directed Acker to elicit incriminating information from defendant, or that he had been promised or received a reward for his testimony in that case. Again in hindsight we see that such specific requests and promises are unnecessary; the state set up a system in which a savvy inmate knows that if he comes forward with a confession from a notorious prisoner, true or false, he will be rewarded. Finally, the petition’s allegations of Acker’s perjury concentrated not on the alleged false testimony describing defendant’s confession—testimony that was clearly prejudicial—but on false statements concerning Acker’s informing in other cases.
But even if the 1984 and 1986 petitions were not suificient to state a prima facie case for relief, I find incredible the majority statement that, even treating the allegations in the mandate proceeding as a supplement to the petition, no prima facie case is stated. Defendant’s response to the mandate petition makes clear that defendant now asserts that Acker testified falsely as to matters material and probative on the issues of special circumstance and penalty, which is grounds for habeas corpus under Penal Code section 1473, subdivision (b)(1).
Defendant has maintained from the initial trial that although he did converse with Acker, he did not share with Acker the details of his case, nor did he make the incriminating statements to which Acker testified. But the evidence to support his claim that Acker invented the confession is no longer limited to defendant’s word. We have taken judicial notice of the report by the 1989-1990 Los Angeles County Grand Jury detailing how various inmates at the Los Angeles County jail successfully fabricated confessions by other inmates and how authorities encouraged or ignored the fabrications. It is true, as the majority point out, that nothing in that report names Acker as a participant in the fabrication scheme, and that there is no specific indication that the prosecution’s files will yield information that will identify Acker as a participant. That, of course, is why defendant seeks discovery. But to characterize defendant’s request as a attempt to “fish” through files to discover new grounds for relief, or to confirm “mere speculation” (maj. opn., ante, p. 1259) is grossly unfair; defendant states that he knows that Acker committed perjury and that Acker was a regular informant at a time when many informants regularly perjured themselves. His case is no more speculative than any case which must depend upon investigation and discovery to substantiate the allegations of specific fact.
The majority are concerned that if we accept defendant’s allegations as suificient to permit discovery of law enforcement records, we will have to allow it in every case challenging the use of testimony from an informant *1288housed at the Los Angeles County jail during the years 1979 through 1988. (See maj. opn., ante, p. 1260, fn. 56.) That, of course, is exactly what the district attorney proposed to the superior court, and what the superior court planned to do until this court granted review in the mandate action. However, the number of cases affected is relatively small. Although there are 8,600 persons confined in the jail at any one time, the grand jury report notes that the district attorney’s jailhouse informant litigation team has identified just 153 cases in which jailhouse informants were called to testify in the 10 years prior to October 1988. (Rep. of the 1989-90 L.A. County Grand Jury, p. 4.) A list of jailhouse informant cases released by the Los Angeles County District Attorney’s office and published in the Los Angeles Daily Journal on January 3, 1989, numbered 130; defendant’s case was among them. Moreover, the majority gloss over a crucial fact alleged by defendant and supported by documentation: defendant was invited by the district attorney to make “any appropriate motion” concerning the possibility that Acker was involved in the confession fabrication scheme.13 It is true that the district attorney did not “specifically indicate” that the prosecution’s files would yield information that would undermine Acker’s testimony, but neither did he deny that fact.
What must defendant allege to state a prima facie case under Penal Code section 1473, and thus acquire the right to discovery? First, he must allege that when Acker described what petitioner supposedly told him about the crime, Acker committed perjury. This, I believe, he has done. Defendant, of course, was present when the conversation occurred, and his statement under penalty of perjury would be direct evidence of what he actually told Acker.
We cannot, however, issue orders to show cause whenever a prisoner alleges that a prosecution witness has lied. The credibility of prosecution witnesses is generally resolved at trial, and a defendant’s disagreement with the jury’s assessment is not grounds for habeas corpus. Thus to state a prima facie case, defendant must also allege additional facts that have come to light subsequent to trial, which lend such additional weight to his claim that a new inquiry is warranted. Defendant here has met this burden, setting out both inconsistencies between that testimony and later discovered evidence and, more importantly, new evidence of a pattern of perjury by informers in the Los Angeles County jail at the time Acker was an informer in residence. The majority assert that habeas corpus “is not a device for investigating possible claims, but a means for vindicating actual claims.” (Maj. opn., ante, p. 1260.) Defendant has actually claimed that Acker *1289testified falsely as to material and probative facts bearing upon the finding of special circumstances and penalty—which is grounds for relief under Penal Code section 1473—and has set forth facts showing this to be a credible claim, thus meeting the majority’s criteria.
Once an order to show cause issues, the issuing court should have discretion to grant discovery. Although there are apparently no California cases on point, the rule is clear in the federal courts. In the leading case, Harris v. Nelson (1969) 394 U.S. 286 [22 L.Ed.2d 281, 89 S.Ct. 1082], the United States Supreme Court upheld a federal district court order granting discovery in a habeas corpus proceeding, stating that “where specific allegations before the court show reason to believe that the petitioner may, if the facts are fully developed, be able to demonstrate that he is confined illegally and is therefore entitled to relief, it is the duty of the court to provide the necessary facilities and procedures for an adequate inquiry. Obviously, in exercising this power, the court may utilize familiar procedures, as appropriate . . . .” (Id. at p. 300 [22 L.Ed.2d at p. 291].) The American Bar Association Standards for Criminal Justice also endorse the use of discovery in habeas corpus proceedings (4 ABA Standards for Criminal Justice, std. 22-4.5 (2d ed. 1988) p. 22.46-22.49); the commentary to this standard explains that “One recurrent problem . . . arises when a prisoner alleges in the application an apparently meritorious claim but the evidence that can be adduced to support the claim is unknown to the court.... Controlled use of discovery devices, if only on a limited scale, will demonstrate what applications are baseless while marshaling the evidentiary basis for applications with merit.” I have no doubt that we should follow this guidance and permit limited discovery in habeas corpus proceedings.
The majority, however, find no prima facie case. Their blithe perpetuation of the Catch-22 is particularly disturbing in a case in which it is clear that if defendant could gain access to the information he needs to satisfy the majority’s stringent requirements for a prima facie showing that Acker’s testimony was perjured, he could easily show prejudice. Acker’s testimony, that defendant told him that he had been “tipped” to expect a police raid, that he wanted to “bag a cop,” and that he planned to claim, falsely, that he believed the officers were members of a street gang at war with his neighborhood, is the principal evidence to support the special circumstance allegation that defendant intentionally killed a police officer engaged in the performance of duty. In fact, the original trial judge in this case recalled that *1290Acker’s testimony was “critical to the issue of whether or not the special circumstance which was alleged was involved.”14
The majority, however, declare that even though the state is withholding potentially exculpatory evidence, the limited scope and pleading requirements of habeas corpus prevent defendant from stating a prima facie case and discovering the concealed evidence. In light of their explanation of how the pleading requirements of habeas corpus can stifle discovery of critical evidence at every turn, it seems appropriate to quote a court with a different perspective on the scope of habeas corpus; “The writ of habeas corpus is the fundamental instrument for safeguarding individual freedom against arbitrary and lawless state action .... The scope and flexibility of the writ— its capacity to reach all manner of illegal detention—its ability to cut through barriers of form and procedural mazes—have always been emphasized and jealously guarded by courts and lawmakers. The very nature of the writ demands that it be administered with the initiative and flexibility essential to insure that miscarriages of justice within its reach are surfaced and corrected ....[][]... And this Court has emphasized, taking into account the office of the writ and the fact that the petitioner, being in custody, is usually handicapped in developing the evidence needed to support in necessary detail the facts alleged in his petition, that a habeas corpus proceeding must not be allowed to founder in a ‘procedural morass' ” (Harris v. Nelson, supra, 394 U.S. 286, 290-292 [22 L.Ed.2d 281, 286], italics added.)
It is plain that the majority are uncomfortable with the conclusion that defendant has no direct remedy, either here or in the trial court, to obtain discovery of the Acker file. They suggest that defendant try other, more circuitous, legal remedies, but without any confidence that he will succeed in getting discovery.15 They note that defendant retains extrajudicial means of investigation,16 although I know of none that would compel access to *1291confidential law enforcement records. Finally, they express confidence that the People’s lawyers will discharge their ethical obligation to disclose any material evidence they have. (Maj. opn., ante, pp. 1260-1261.) The confidence appears misplaced in light of the abrupt about-face witnessed here: the People’s attorneys have moved to block discovery after inviting defendant to seek it, and continue to maintain that this is not a case in which they have any obligation of disclosure.
All of this discussion of alternative remedies is smoke, not substance. The bottom line is that the majority affirm the conviction and deny all discovery or collateral relief. The State of California may now proceed to execute the defendant without revealing information it has concealed which may show that defendant’s conviction and penalty were procured with perjured testimony. This is a miscarriage of justice which may return to haunt us.
Appellant’s petition for a rehearing was denied February 27, 1991. Broussard, J., was of the opinion that the petition should be granted.
The majority note that according to Boyde v. California, supra, 494 U.S. 370, “prosecutorial commentary should not be given undue weight in analyzing how a reasonable jury understood capital sentencing instructions” and “a ‘court should not lightly infer that a prosecutor intends an ambiguous remark to have its most damaging meaning ....”’ (Maj. opn., ante, p. 1224, fn. 21, italics added.) This kind of language is of no help in deciding cases. Courts should never give “undue” weight to anything, or, when matters of life and death are at issue, draw inferences “lightly.” But the problem with the majority opinion is that, from among the possible interpretations of the prosecutor’s statements, they consistently adopt the most innocuous one, even when in context that is the less reasonable interpretation.
This is familiar language to this court, having been quoted, paraphrased, or endorsed in over 20 cases in the 5 years since Brown was filed.
I note also Brown’s assertion that a statute would be unconstitutional if “it required jurors to render a death verdict on the basis of some arithmetical formula.” (40 Cal.3d at p. 540.) Whether or not this statement is correct as a matter of constitutional law in light of the Supreme Court’s recent decisions, it makes clear Brown’s view that the California statute should not be interpreted to base the penalty determination on an arithmetical formula.
Justice Arabian condemns the prosecutor for making this argument, stating that “[p]rosecutorial tactics such as those exhibited here push against the very limits of acceptable behavior . . . . ” (Cone, opn., ante, p. 1262.) I respectfully suggest that he is shooting at the wrong target. This case was tried before we filed our opinion in Brown, supra, 40 Cal.3d 512. The prosecutor based his argument on one reasonable interpretation of the statutory language. He should not be criticized for failing to foresee that we would adopt a different interpretation in Brown.
The majority accuse me of assuming that the “weighing” and “appropriateness” functions are separate and independent. (Maj. opn., ante, p. 1230.) I make no such assumption. I agree with the majority that the jury may consider appropriateness when they are weighing the relevant factors. Actually, it is the prosecutor the majority should be criticizing, for he treated the two functions as separate, telling the jury, “your job is a weighing process. It’s not for you to think, ‘Well, I think this is an appropriate-type case; no, this is not an appropriate-type case.’ ”
I do disagree with the statement in People v. Boyde (1988) 46 Cal.3d 212, 253 [250 Cal.Rptr. 83, 758 P.2d 25], quoted by the majority, ante, at page 1230 that “when jurors are informed that they have discretion to assign whatever value they deem appropriate to the factors listed, they necessarily understand they have discretion to determine the appropriate penalty.” This is not a statement of law, but of fact—of how jurors in fact reason in capital trials—and is unsupported by empirical evidence.
Even in the context of comparison with other cases, the prosecutor’s remarks may be misleading. I know of no authority that a juror, in deciding the weight to be given to the circumstances of the crime, or whether death is the appropriate penalty, may not mentally compare the facts of the case to those of other cases.
We recently affirmed the penalty verdict in People v. Frank, ante, page 718 [274 Cal.Rptr. 372, 798 P.2d 1215], The prosecutor in that case told the jury, “I think it’s important for you to appreciate that not all murders are the same .... And in evaluating whether or not this murder . . . [is] an appropriate case for the death penalty, you are certainly, I think, free to determine where does this fall in the spectrum of murders and kidnappings.” I think the Frank prosecutor was completely correct in this assertion.
The prosecutor’s argument was erroneous in another respect. As the majority note, in his closing argument he incorrectly argued that defendant’s lack of remorse was aggravating. (Maj. opn., ante, pp. 1231-1232.) They say that he tempered this theory in rebuttal when he only asserted that defendant had failed to show remorse as mitigation. But this is only half the story. Having asserted that defendant failed to prove remorse as mitigation under Penal Code section 190.3, factor (k), the prosecutor went on to assert that factor (k)—any circumstances which extenuate the gravity of the crime—belongs in the aggravating column. This is error under People v. Boyd (1985) 38 Cal.3d 762, 775-776 [215 Cal.Rptr. 1, 700 P.2d 782],
Defendant offered to drop his discovery request if a representative of the district attorney’s office would swear under penalty of perjury that the files contain no material evidence relating to Acker’s testimony. The district attorney did not accept the offer.
Hopkinson's habeas corpus proceeding simply contended that the state had violated his constitutional rights by not disclosing the evidence presented to the grand jury.
Acker gave information about a murder of which his wife was accused at some time before January 1, 1979. He gave information about a murder of which fellow inmate Torres was accused, starting in March 1980. He contacted the sheriif’s office to say that fellow inmate Anderson had confessed to him about a murder in early August 1980. He testified in defendant’s guilt phase for the first time on August 19, 1980. During questioning, he admitted giving information on another case involving fellow inmates Burkett and Greer, and indicated that he may have given information in that case before he began informing on defendant.
Those cases involved Acker’s fellow inmates, Davis, LaScola and Williams.
The former requirement, that the petition must allege that the prosecution knew or should have known of the falsity of the evidence, has been abolished. (See Pen. Code, § 1473, subd. (c).)
The district attorney has never been willing to commit himself to saying what motion would be “appropriate.” Neither will the majority.
The only other evidence to contradict the defense theory of the case was the testimony of the two deputies who said that they had interviewed defendant the day after the shooting and that at one point, he explained that he ran inside his house when he saw “the cops” coming. They further testified, however, that when they asked defendant to repeat what he had said about “the cops,” he corrected them, explaining that he had not said that and that they must have been confused.
The majority refer to Government Code section 6250 et seq., the California Public Records Act. But documents before this court show how the civil suit filed on behalf of another criminal defendant under that act to obtain information involving a Los Angeles County jailhouse informant has been stymied by several law enforcement agencies, who have apparently taken the position that the requested information is exempt from disclosure under the act.
One remedy that may work, however, is not mentioned by the majority: an action to discover the evidence presented before the grand jury. (See ante, pp. 1283-1284.)
I note, in contrast, the initial position of the district attorney that the courtroom was the appropriate forum in which to determine which cases were affected by perjured informer testimony.