dissenting:

As I read the record in this case, defendant Smith never invoked his right to counsel after having been given unrestricted opportunity to do so. At most he was “thinking out loud” while making up his mind in a clearly non-coercive situation. In the Miranda-Edwards-Fouche I-Fouche-II context, I would summarize the record as comprising an initial Mi randa waiver as the interrogation begins, a later ambiguous request for counsel 1 followed by clarifying questions, after which it becomes clear that defendant never did invoke his right to counsel. Therefore, I must dissent from the majority’s decision to require suppression of the confession. I would concur in the four preceding court decisions (two Alaska state courts, a federal magistrate, and a federal district judge), which found no constitutional bar to the admissibility of Smith’s confession.

The key issue is, of course, whether Smith’s statements were sufficient, as a matter of constitutional law, to invoke his right to counsel. If Smith had stopped after “Can I talk to a lawyer?,” he would have invoked unambiguously his fifth and fourteenth amendment rights and no doubt the Alaska troopers, following Miranda and Edwards, would have answered affirmatively and shown him the telephone as indeed they did in this case. In any event, “can I talk to a lawyer?” standing alone, would have been an unambiguous request for counsel.

The problem is that Smith did not stop there. He went on to say, without interruption, that he was only interested in talking to a lawyer if the police were looking at him as a suspect, whereupon he asked the question, “Are you looking at me as a suspect?” Although the majority feels otherwise, I find the trooper’s answer to that question to be responsive and comprehensible: “Well, it ... it wouldn’t be fair to you to say that we weren’t Mike.” Of course Smith was a murder suspect and Trooper Stearns told him so.

The troopers then began to clarify whether Smith, having received confirmation he was indeed a suspect, wanted to go *1535 on to invoke his right to counsel, as the following excerpt from the interrogation transcript makes clear.

These police statements were well within the clarification process because they were in response to Smith’s own questions. 2 The majority attaches great significance to the troopers’ explaining to Smith why he was a suspect, which it deems impermissible because it found a purpose to elicit incriminating statements in violation of the standard we established in Fouche I, 776 F.2d at 1405 (“Questions aimed at clarifying the desire for counsel must be strictly limited to that purpose; they may not be used to elicit incriminating information”). 3 While the Fouche I standard is strict by its very terms, I do not believe it was intended to be used as a mechanism to pick apart every word that a police officer might utter during the course of an interrogation. Indeed, the actions of the troopers are at least as solicitous concerning the suspect’s right to counsel, if not more so, as those of the FBI agent which we held to be proper in Fouche II. See also United States v. Nordling, 804 F.2d 1466, 1470-71 (9th Cir.1986). Even the majority concedes that the interrogation was conducted without the slightest trace of coercion. 4 Majority opinion at 1581.

In the clarification process, it became obvious that Smith did not intend to invoke his right to an attorney. Having been offered the telephone to call his lawyer, Smith chose instead to renew the dialogue. Indeed, Smith stated shortly before his confession, and long after having been told he was a suspect: “I don’t know if I need [an attorney] or not. That’s why I’m trying to make my mind up, if I need to go that route for myself, you know.”

Certainly, if Smith’s initial statements had been unequivocal, then I would not look beyond them in determining that he *1536 had properly invoked his right to counsel. Smith v. Illinois, 469 U.S. at 92, 105 S.Ct. at 491. But Smith’s initial statements do not rise to the level of an unambiguous request, and therefore the Smith v. Illinois rule does not apply. See Grooms v. Keeney, 826 F.2d 883, 887 (9th Cir.1987) (court stating that “[t]he Smith rule does not apply here, because it concerned an unequivocal request for counsel, and there is no persuasive reason to extend or invert its rule to fit the facts of this case”) (citation omitted).

Smith’s latter statements merely show that he had no intention of unambiguously invoking his right to counsel before continuing to answer questions. In Fouche II, we did not require in the clarification process that the suspect affirmatively state that he would continue without a lawyer present. Rather, simply enough, we evaluated whether Fouche invoked his right to counsel. See Fouche II, 833 F.2d at 1287 (“Fouche voluntarily declined to invoke his right to counsel, choosing instead to answer [the agent’s] questions”).

Nothing in Edwards or its progeny mandates a reversal in this case. Smith made an ambiguous inquiry about right to counsel, to which the troopers responded properly by seeking to clarify Smith’s true intentions. In so doing, the troopers did not stray from the limits we laid down in Fouche I. I am aware of no Supreme Court or Ninth Circuit precedent which freezes police in their tracks after the defendant says “I’m trying to make my mind up” in this context. I am concerned that the majority’s analysis unnecessarily complicates this area of the law. I fear that the majority’s holding does not accomplish what in large measure it sets out to do, i.e., provide a bright line for all concerned parties to follow. In fact, I think it may likely have the opposite effect. The analysis essentially injects a new concept into the fifth-fourteenth amendment lexicon: “conditional unequivocal request for counsel.” In so doing, it confuses and diverts attention from whether right to counsel has been invoked. It also carries with it a potentially thorny new analytical requirement, i.e., whether a “condition” laid down by the suspect is satisfied. 5

Even if my fears that the majority’s “bright line” will become blurred are un *1537 justified, I am concerned that it loses sight of the policy underlying that line. Until now, we have sought to maintain a delicate balance 6 between ensuring that suspects are properly insulated against police overreaching while allowing the law enforcement community to perform its duties effectively. 7

Perhaps the majority’s new “bright line” is only an accretion from the old. But one only has to look back less than a decade to the facts of Edwards (which extended Miranda) and compare them to the present case to see how accretion becomes avulsion. The Edwards Court was presented with detectives who went to the jail where defendant was being held to badger him into talking after he had unequivocally invoked both his right to counsel and his right to remain silent; the Court quite properly drew a line beyond which the police could not go and required suppression of the confession. Here, we delegate to a knowledgeable defendant during a Miranda -waived interrogation at trooper headquarters the right to lay down to police the conditions upon which his interrogation may or may not proceed. We have come a long way since Miranda and Edwards — too far, too fast, in my view.

. Smith’s statements evince no more than an ambiguous or equivocal assertion of the right to counsel, see United States v. Fouche, 776 F.2d 1398 (9th Cir.1985) (“Fouche I"), even when read in light of the Supreme Court’s charge to construe broadly any such requests. See, e.g., Michigan v. Jackson, 475 U.S. 625, 633 , 106 S.Ct. 1404, 1409 , 89 L.Ed.2d 631 (1986).

Moreover, the simple fact that the suspect has mentioned the word “lawyer" is not dispositive of whether he has made even an ambiguous request for counsel. See United States v. Jardina, 747 F.2d 945, 949 (5th Cir.1984), cert. denied, 470 U.S. 1058 , 105 S.Ct. 1773 , 84 L.Ed.2d 833 (1985) (“The word ‘attorney’ has no talis-manic qualities. A defendant does not invoke his right to counsel any time the word falls from his lips”).

. These statements also shed light on what we have referred to as the "critical factor” in determining the validity of the government's behavior; i.e., "whether a review of the whole event discloses that the interviewing agent has impinged on the exercise of the suspect’s continuing option to cut off the interview.” United States v. Fouche, 833 F.2d 1284, 1287 (9th Cir.1987), ce rt. denied, — U.S. -, 108 S.Ct. 1756 , 100 L.Ed.2d 218 (1988) ("Fouche II") (quoting Nash v. Estelle, 597 F.2d 513, 518 (5th Cir.) (en banc), cert. denied. 444 U.S. 981 , 100 S.Ct. 485 , 62 L.Ed.2d 409 (1979)).

The majority seems to think that the clarification process should be limited to the police answering a suspect’s questions. See majority opinion at 1533-1534. Such an approach is at odds, however, with the very wording of the standard we adopted in Fouche I, wherein we stated that “[qjuestions aimed at clarifying the desire for counsel must be strictly limited to that purpose.” Fouche I. I found no constitutional requirement that a suspect and police reverse roles at this stage; the police are still in charge of the interrogation and are not barred from asking clarifying questions.

. Indeed, the troopers’ actions neither violate the Fouche I standard nor would they violate the necessarily stricter standard we apply in reviewing police conduct following a suspect’s unequivocal request for counsel. A recent case is illustrative, in which we held that police did not violate Miranda when, after a suspect had unequivocally requested an attorney, they falsely told him that he had been positively identified by the victim of the rape. Shedelbower v. Estelle, 859 F.2d 727 (9th Cir.1988). One would be hard-pressed to argue that lying to a suspect about his implication in the crime about which he is being questioned is somehow more appropriate police conduct than that in which the troopers engaged here.

.I am mindful of the Court’s admonition that Edwards v. Arizona, 451 U.S. 477 , 101 S.Ct. 1880 , 68 L.Ed.2d 378 (1981), and its progeny provide for "a prophylactic safeguard whose application does not turn on whether coercion in fact was employed.” Smith v. Illinois, 469 U.S. 91 , 99 n. 8, 105 S.Ct. 490 , 495 n. 8, 83 L.Ed.2d 488 (1984) (emphasis supplied). I am not questioning in this dissent whether Edwards should be applied; rather, I suggest that evaluating the underlying purposes of the Edwards rule offers guidance on how to apply the rule.

. This is especially bothersome here, where the alleged condition relates to a matter within the exclusive knowledge of the interrogators. Where, as here, the answer is obvious, i.e., the troopers were looking at Smith as a murder suspect (and so informed him), the problems are admittedly minimal. But what about a situation in which the suspect makes his request contingent upon a matter that is not readily apparent? Then will the focus of the inquiry have to shift to the state of mind of the interrogators to see if the alleged condition is satisfied?

Until now, right to counsel jurisprudence recognized two categories of request with a bright line rule to deal with each: “unequivocal” (all interrogation ceases until accused initiates further communication: Miranda-Edwards) or "ambiguous-equivocal" (clarifying questions only: Fouche I-Fouche II). The majority sets up a new category of “conditional” somewhere between Miranda and Fouche 1; I would place “conditional,” if such a category were to exist at all, entirely within "ambiguous-equivocal."

First, a two category rule definitely offers a much more luminous "bright line” than does the majority. Second, such a standard actually serves to protect the suspect undergoing questioning better than does the majority approach because under the latter analysis, what happens to requests based on conditions that are not satisfied? Can they be treated as nullities? Again, the analysis will unravel into a thorny bramble in which the court will be called upon to determine whether the condition is satisfied. Third, the "strictly limited” standard set forth in our Fouche I decision guards against agents using responses to an allegedly conditional request for counsel as opportunities to obtain incriminating statements. Indeed, under the standard of conduct exhibited in this case, the interrogator would be required to do more than simply provide any knowledge he may have in regard to the "condition” the suspect has attached to his purported invocation; he should also give the suspect, having been made aware of any information the agent may have had, an additional and explicit opportunity to invoke the right to counsel (as the troopers did here). Finally, such a standard comports with the fact that "Edwards focuses on the state of mind of the suspect and not of the police." Arizona v. Roberson, — U.S. -, 108 S.Ct. 2093, 2101 , 100 L.Ed.2d 704 (1988). Indeed, as Justice Kennedy stated in his Roberson dissent, "[o]ur ultimate concern in Edwards , and in the cases which follow it, is whether the suspect knows and understands his rights and is willing to waive them.” Roberson, 108 S.Ct. at 2102 .

. I advisedly speak of maintaining a balance between these competing interests in this context. As Justice Kennedy reminds us in his dissent in Roberson , the Edwards rule is just that — a rule — and not a constitutional mandate. Roberson, 108 S.Ct. at 2101-02 (Kennedy, J., dissenting). Therefore, we must be careful in applying the rule in contexts that represent an "inching away” from the moorings of the original purposes for the rule.

. The words of Justice Kennedy's dissent in Roberson seem particularly apt here as well: “The majority's rule is not necessary to protect the rights of suspects, and it will in many instances deprive our nationwide law enforcement network of a legitimate investigative technique now routinely used to resolve major crimes." Roberson, 108 S.Ct. at 2102 (Kennedy, J., dissenting).