concurring in part; dissenting in part.
I concur with the legal principles expressed in the opinion. Under the facts before us, however, I cannot agree with the result. In my view, counsel was not ineffective and the conviction should be affirmed.
Under State v. Watson, 134 Ariz. 1, 653 P.2d 351 (1982), counsel is not held to. a comparative standard of performance. Rather than adopting a comparative test, we adopted a standard which requires counsel to demonstrate minimal competence. Id. at 4, 653 P.2d at 354. In the final analysis, of course, this means that it is this court which sets the standard, presumably giving some consideration to comparative criteria and other factors as yet unspecified. In the case at bench, I do not believe that trial counsel was faced with so clear a choice that we can say he fell below minimal standards of competency in calling the two witnesses and later foregoing final argument.
As the majority indicates (maj. op. at 158), trial counsel has the power to control strategy and tactics and to determine which witnesses will be called, what arguments will be made, what exhibits offered and the like. However, that power is not unfettered. In exercising his power to control strategy and tactics, the attorney nevertheless remains circumscribed by his duty; he always “ ‘acts on behalf of his client,’ ” and his actions must be designed to give the client both “ ‘the result which his client seeks’ ” and the rights which the law allows. State v. Rodriguez, 126 Ariz. 28, 34, 612 P.2d 484, 490 (1980), quoting Lanier v. State, 486 P.2d 981, 986 (Alaska 1971). Among the rights of the defendant is the Sixth Amendment right to call favorable witnesses, Chambers v. Mississippi, 410 U.S. 284, 302, 93 S.Ct. 1038, 1049, 35 L.Ed.2d 297 (1973); Washington v. Texas, 388 U.S. 14, 18, 87 S.Ct. 1920, 1923, 18 L.Ed.2d 1019 (1967). While this right is not absolute, and does not extend to cumulative or irrelevant testimony (nor, certainly, to perjured testimony), it does extend to the testimony of competent witnesses who have personal knowledge of the events and whose testimony would be relevant, material and exculpatory. United States v. Valenzuela-Bernal, 458 U.S. 858, 102 S.Ct. 3440, 3449-50, 73 L.Ed.2d 1193 (1982).
Thus, we have a divergence between counsel’s power to control tactics and strategy, the duty owed the system which requires that perjured testimony not be offered, and the defendant’s rights to present a defense and to have the assistance of counsel in so doing. As the majority correctly holds (maj. op. at 157), defendant’s Sixth Amendment right and counsel’s duty to his client must give way if counsel knows or should know that the proposed *222witnesses will perjure themselves. The tension lies in the word “know.” It is at this point in the case at bench where the bright line of legal principle becomes clouded by actual fact. It is not clear that counsel knew that the two witnesses would perjure themselves. The transcript of the in-chamber proceedings on day of trial shows counsel’s comments as follows:
Based on all of the investigation that I have done in this case, there is no question in my mind that both of the witnesses just gave perjured testimony. But I do not know that by virtue of my client telling me it’s so. I do not know it by virtue of them telling me it’s so. It is my own speculation that their testimony is perjured.
Transcript of In-Chamber Proceedings at 4 (emphasis supplied).
In my view, trial counsel here was caught in a dilemma to which there was no clear, obvious answer. On the one hand, his client demanded his constitutional right to present witnesses to the essential facts; on the other hand, the lawyer believed that the testimony might be perjured. We may assume also that the lawyer was aware of his duty to assist his client, not as a judge of credibility, but as an advocate of the client’s case. Further, we can and should take note that if the lawyer had not called the witnesses, there is little doubt but that the client would have claimed, with some degree of plausibility, that counsel had been ineffective because he had failed to present important exculpatory evidence. The decision as to the proper course of conduct in these situations must often be made — as it was here — in the midst of trial, with little time to consult law reviews in order to obtain scholarly views on an issue on which there is essentially no case law. We have no way of knowing, but I strongly suspect that most lawyers, having no actual knowledge of perjury, would have called the witnesses, questioned them, and argued their testimony even though they were concerned about the tactical or ethical problems. Certainly, if this lawyer had pursued such a course of conduct, only speculating that there might be perjury, his conduct would not have been questioned. This lawyer did try his best to discharge both of his duties — that owed to the system and that owed to his client. In my view, while the lawyer’s decision to call the witnesses and allow them to give narrative testimony was not the best course of conduct it was not so unreasonable as to fall below minimal standards. He could not act as a judge of their credibility but chose not to assist them because he doubted their veracity.
I disagree with the suggestion that withdrawal is a viable solution. I have yet to meet the trial judge who would entertain, much less grant, a motion to withdraw made during trial on the grounds that irreconcilable differences had developed between counsel and client. The same may be said for such motions filed shortly before trial. Further, withdrawal only leaves the problem, unsolved, in the lap of the next lawyer; it also educates the defendant in methods which may be useful in befuddling the entire system.
We come, then, to the question of the decision to forego final argument. In most cases, I would agree that final argument is so important that waiver is below the minimal standard. I also agree with the majority that this case could have been argued and perhaps should have been argued. Nevertheless, the decision not to argue here was within the realm of tactics and strategy. We have indicated that such decisions will not constitute ineffective assistance. State v. Rodriguez, supra. In Rodriguez, however, counsel had discussed the waiver of argument with the client while in the instant case he did not. Nevertheless, the strategic or tactical decision here was forced upon counsel by the client’s insistence on his constitutional right to present the witnesses despite the lawyer’s warning that it would weaken the defense. In my view, the client got what he wanted and is not now in a position to complain of the consequent tactical — and ethical — decision to waive argument. The lawyer did not waive argument to “punish” the client, but because it was his judgment that it would harm the case to argue the “mere pres*223ence” defense without mentioning the suspect testimony, which was the only direct evidence on the question. This was a matter of strategy and tactics, a judgment call forced by the client’s refusal to take his lawyer’s advice, and should not be second-guessed. The lawyer did not handle this problem as well as, with the benefit of hindsight, five justices of the Supreme Court now perceive that it should have been handled. The lawyer, however, did see the problem, was honest with the court and attempted to reconcile the competing interests while giving the client what he demanded. Without the luxury of briefing by counsel, oral argument and time to mount a legal search, the lawyer chose a course of action less than ideal. However, the Watson standard requires only minimal competency, not perfection.
I would affirm the conviction.