I.
I concur in the affirmance of the conviction and in the majority’s discussion of the special circumstance. I join wholeheartedly in the majority’s conclusion that a penalty judgment must be reversed if there is a reasonable possibility that, absent error, the jury would have rendered a different verdict.1 But I am concerned about the majority’s use, in applying that standard, of the concepts of “a reasonable jury’s verdict” and “reasonable *472jurors.” (Ante, p. 448.) The question before us in each case is the impact of the error on the jury that tried the case 2 If we have no other information about the jury’s reasoning we may have to conjecture how a “reasonable jury” would proceed. But if, as is often the case, we have some reason to know how this jury is proceeding, we could not disregard that knowledge merely because a “reasonable jury” would approach the case differently. If, for example, a question from the jury revealed that the jurors were particularly interested in certain inadmissible evidence or in a particular erroneous argument, we could not uphold a verdict on the theory that a reasonable jury would ignore that evidence or disregard that argument.
II.
1 respectfully dissent from the majority’s affirmation-of the penalty judgment. There were simply too many errors at the penalty trial for me to conclude that there is no reasonable possibility that, absent those errors, the jury would have returned a verdict more favorable to defendant.
1. The trial court erred under People v. Robertson (1983) 33 Cal.3d 21, 53-55 [188 Cal.Rptr. 77, 655 P.2d 279], in failing to instruct the jury that it could not consider past criminal activity as an aggravating factor unless it found beyond a reasonable doubt that defendant committed the crimes in question. The majority acknowledge this error. (Ante, p. 446.)
2. The trial court erred twice under People v. Boyd (1985) 38 Cal.3d 762, 772-776 [215 Cal.Rptr. 1, 700 P.2d 782], by admitting evidence of noncriminal conduct.
(a) The court admitted evidence that defendant threatened to sodomize another prisoner. The admissibility of such evidence under Penal Code section 190.3, factor (c), depends upon proof that defendant committed a criminal act involving personal violence or the threat of such violence. (See People v. Boyd, supra, 38 Cal.3d 762, 772-776.) In arguing for admissibility, the Attorney General mistakenly assumes that the threat to commit a violent criminal act is enough to establish admissibility under factor (c). It is not. The evidence must show a criminal act, and threatening words without more do not constitute a criminal act. Accordingly the evidence in question was inadmissible under Boyd.
(b) The court also admitted evidence of a “food riot” during which defendant and other inmates threw food from their cells into the hallway to *473protest the allegedly inadequate quantity served. While some inmates threatened the guards, there was no proof that defendant did so, and in any case the utterance of a threat by an inmate in no position to carry out the threat is not a crime (see Boyd, supra, 38 Cal.3d at pp. 777-778). The only violence defendant could or did perpetrate was to the food, and violent injury to property is insufficient to justify admissibility. {Boyd, 38 Cal.3d at p. 776.)
3. The trial court failed to instruct the jury that Penal Code section 190.3, factor (k) includes as a mitigating consideration not only those matters which extenuate the gravity of the crime, but also any other aspect of the defendant’s character or record that he proffers as a basis for a sentence of less than death. (See People v. Easley (1983) 34 Cal.3d 858, 878, fn. 10 [196 Cal.Rptr. 309, 671 P.2d 813].)
4. The prosecutor erroneously told the jury, contrary to People v. Davenport (1985) 41 Cal.3d 247, 288-289 [221 Cal.Rptr. 794, 710 P.2d 861] and People v. Rodriguez (1986) 42 Cal.3d 730, 789-790 [230 Cal.Rptr. 667, 726 P.2d 113], that the absence of evidence showing the existence of various statutory mitigating considerations requires the jury to weigh those considerations as aggravating factors in determining penalty.
This is the most egregious case of Davenport error we have yet encountered. The prosecutor went down the statutory list, and characterized every factor as aggravating. With respect to factor (d)—mental or emotional disturbance—he said, “This defendant . . . has no mental or emotional disturbance other than the fact that he is just a sociopath, and antisocial and likes to hurt people. So in this particular case that’s an aggravating factor.” With respect to factor (e)—victim participation—he found the absence of such participation made it an aggravating factor. He said the same as to factor (f)—belief in moral justification. For factor (g)—duress—he noted that no one dominated the defendant and compelled him to kill the victim, so “[t]his is an aggravating factor.” Defendant presented substantial evidence of factor (h)—mental disease or defect which impairs defendant’s capacity to conform to legal requirements—but the prosecutor argued that the evidence did not establish “diminished capacity,” and hence “that is a aggravating factor.” He characterized defendant’s age (factor (i)) as aggravating. He said that factor (j)—whether defendant was an accomplice—was aggravating because defendant acted alone. And in summation he asserted that factor (k)—any other circumstance which extenuates the gravity of the crime—was aggravating even though in People v. Boyd, supra, 38 Cal.3d 762, 775-776, we explained that factor (k) can only be mitigating.
The majority acknowledge that the prosecutor’s argument “seems contrary to Davenport, ” but state that “we do not consider the prosecutor’s *474argument to be reversible prosecutorial misconduct, because a timely objection and admonition could have cured any harm.” (Ante, p. 456.) This statement sloughs over the question whether defense counsel can reasonably be expected to object to such argument in a trial preceding Davenport, supra, 41 Cal.3d 247, or Rodriguez, supra, 42 Cal.3d 703, and the question whether the trial court would have sustained such an objection. (Since the judge, ruling on the motion to modify the judgment, said that various neutral or mitigating factors were aggravating it is likely that he would have overruled any Davenport objection.)
I agree, however, that there was no prosecutorial misconduct. At the time of argument the prosecutor could reasonably believe that he could properly argue the absence of evidence of a mitigating factor rendered the factor aggravating. But the question of misconduct is irrelevant. As we explained in People v. Lucero (1988) 44 Cal.3d 1006, 1031, footnote 15 [245 Cal.Rptr. 185, 750 P.2d 1342], “[o]ur concern is not with the ethics of the prosecutor or the performance of the defense, but with the impact of the erroneous interpretation of the law on the jury.” The majority tacitly concede the point, for in this, as in every other case involving Davenport error, they do not stop with noting the failure to object, but go on to treat the issue on its merits.
The majority further argue that the jury was instructed to consider the listed factors only “if relevant,” and that a reasonable jury “would have understood that it was allowed to give to those factors whatever little weight they deserved.” (Ante, p. 456.) But the jury does not know what factors the law considers relevant. The vice of Davenport error is that it tells the jury that factors are relevant when in fact they are not, and that factors are aggravating when in fact they are neutral or mitigating. Of course a jury which followed the prosecutor’s advice would still be “allowed” to assign the factors what little weight they deserved. It is also allowed to assign them a great weight which they do not deserve. The one thing it cannot do is the thing it is supposed to do: assign them no aggravating weight at all.
5. In his argument to the jury, as in his prior voir dire of the jurors, the prosecutor misinformed the jury concerning the process by which they should determine the penalty, in violation of People v. Brown (1985) 40 Cal.3d 512, 541 [220 Cal.Rptr. 637, 709 P.2d 440], reversed on other grounds, sub nomine, California v. Brown (1987) 479 U.S. 538 [93 L.Ed.2d 934, 107 S.Ct. 837],
The prosecutor began his closing argument by reminding the jurors of the voir dire, and said that “we all agreed that if in weighing these factors the aggravating factors outweighed, are heavier than the mitigating factors, the *475penalty to be imposed is the death penalty.” We must therefore examine the voir dire to find the “agreement” to which the prosecutor referred.
The prosecutor told one prospective juror, who actually served on the jury, that “if you determine that the aggravating factors are just a little bit heavier than the mitigating factors, then the law requires you to return a death penalty verdict.” The dialogue continued: Prosecutor: “Do you have any quarrel with that?” Juror: “No.” Prosecutor: “And, can you look at yourself and say, ‘okay,’ ... I can feel these aggravating factors outweigh the mitigating factors, but I am still kind of on the fence about the death penalty. I am not sure about the death penalty, but in my mind I feel the aggravation outweighs the mitigation. You will comply with the law and return a death penalty verdict?” Juror: “Yes. If that’s how I feel. If I feel that one outweighs the other.”
The prosecutor’s statements were clearly error, because they left the juror in question with the mistaken impression that the jury “was ‘required’ to return a sentence of death if ‘aggravation outweighed mitigation’ without, or even despite, each juror’s personal conclusion from the evidence, about whether a sentence of death was appropriate under the circumstances for the offense and offender.” (People v. Allen (1986) 42 Cal.3d 1222, 1278 [232 Cal.Rptr. 849, 729 P.2d 115], italics in original.) While other aspects of the voir dire of this juror were not necessarily erroneous—she said she would examine the evidence and impose the death penalty only when “necessary,” whatever that meant—they did not make it clear that the quoted remarks of the prosecutor were in error, and that it was her duty not to impose the death penalty, regardless of the balance of aggravation and mitigation, unless she believed death to be the appropriate penalty. Thus at least one juror who tried defendant probably did so under an erroneous view of her responsibility. The voir dire of other jurors was not as obviously erroneous, but nonetheless proceeded in much the same vein.3
The prosecutor then reviewed the statutory factors, concluding, erroneously, that all were aggravating. Instead of informing the jurors that they *476should consider these factors in deciding whether death was the appropriate penalty, he told them that the law makes that determination for them: “[T]he law says if you add those factors up and the aggravating outweigh the mitigating, that’s a case where the death penalty is the only appropriate penalty.”4
III.
These errors do not exist in isolation; instead, each adds to the impact of the others. The prosecution strategy follows a consistent pattern. At voir dire, the prosecutor told the jurors that if aggravating factors outweighed mitigating factors a death verdict was required. With attention thus focused on the relative number of aggravating and mitigating factors, and the evidence presented to support such factors, the prosecutor introduced inadmissible evidence of aggravating factors, and other evidence admissible only under a reasonable-doubt instruction, which was not given.
The prosecutor’s closing argument put the package together. He told the jury that five irrelevant factors were aggravating. He told them that two others which could only be mitigating—factors (h) and (k)—were aggravating. He concluded by explaining that all eleven of the factors were aggravating, when in fact there were three aggravating, two mitigating, and the rest irrelevant. Having thus left the jury with a highly exaggerated view of the number of aggravating factors, he reminded them that on voir dire they had agreed to vote for a death verdict if aggravation outweighed mitigation and concluded that the law required them to return a death verdict. In short, the combined effect of the prosecution strategy and the numerous penalty phase mistakes in this case was to leave the jurors with the erroneous impression that, if they were faithfully to perform their legal duty, they must return a death verdict. I conclude that it is reasonably possible that, but for the cited errors] the jury would have returned a more favorable verdict.
IV.
In addition to the errors during the penalty trial, the court also erred in ruling on the motion to modify the judgment (Pen. Code, § 190.4). Its *477ruling erroneously stated that the absence of evidence of mitigating factors rendered those factors aggravating, and unconstitutionally limited the scope of mitigating evidence under Penal Code section 190.3, factor (k). The majority opinion acknowledges those errors, but states since it found similar errors did not prejudice the jury’s penalty determination, by analogy such errors did not prejudice the judge’s ruling. (Ante, p. 462.)
The analogy is faulty. In finding the Davenport and factor (k) errors nonprejudicial, the majority speculate that the jury may have rejected the prosecutor’s mistaken arguments. That reasoning does not apply to our review of the judge’s ruling on the motion for modification. We cannot speculate that the judge may have rejected the prosecutor’s theory and decided the case according to a correct view of the law; we know, from his written order, that he shared the prosecutor’s misconception.
In effect, the majority are saying that even though the judge had a totally skewed view of the case—he thought there were eleven aggravating factors and no mitigating factors, when in fact there were three aggravating and two mitigating—he still would have reached the same result under a correct view of the law. He might have. But to claim that it is not reasonably possible that he would have reached a different result, and on that basis to condemn a man to die, goes further than I am willing to go.
Appellant’s petition for a rehearing was denied October 13, 1988. Broussard, J., was of the opinion that the petition should be granted.
A different verdict, under our test of error, does not necessarily mean a verdict of life imprisonment without possibility of parole. A hung jury is a more favorable verdict, since it not only avoids the death penalty but also, upon a second occurrence, gives the trial judge discretion to deny a retrial and impose a penalty of life without possibility of parole. (Pen. Code, § 190.4, subd. (d).) Thus we need only inquire whether it is reasonably possible that a single juror might have changed his or her vote in the absence of error.
Similarly in reviewing motions to modify a verdict under Penal Code section 190.4, we consider whether the judge who tried the case ruled properly on the motion. We do not discuss how a hypothetical reasonable judge would rule.
The prosecutor did not tell the other jurors that they had a duty to return a death verdict whenever aggravation outweighed mitigation in spite of doubts whether death was appropriate. He did, however, describe the process of determining penalty as an arithmetical process devoid of the moral and normative considerations implicit in any attempt to determine the appropriate penalty. He told one juror, for example, that “You add this up, and if the aggravating evidence outweighs the mitigating evidence, then the judge will instruct you that you have to return the death penalty.” He told another: “If the aggravating factors are heavier than the mitigating factors or if the aggravating factors outweigh the—or outnumber the mitigating factors, then the law requires you to vote for the death penalty.” To another he said, “[I]f the aggravating factors, the factors that aggravate the defendant’s past conduct and this crime here, if that outweighs the mitigating evidence, the law requires that you vote for the death penalty. . . . There are actual factors that the judge will tell you about. And, you can just add those up and weigh them.”
Inventing a kind of inverse Gresham’s Law (Gresham’s Law is that bad money drives out good money), the majority assume that correct argument dispels the effect of incorrect argument. Thus when the prosecutor states the law incorrectly, and defense counsel states it correctly, the majority assume without further inquiry that the jury followed the defense version. It is not a matter of roles; doubtless if the prosecutor stated the law correctly and defense counsel misinformed the jury, the majority would be equally willing to assume the jury followed the prosecutor’s version.