concurring.
I concur in the per curiam opinion, but as to parts I and IV, I do so for somewhat different reasons.1
I. THE INNOCENCE INSTR UCTION
The Idaho trial court gave numerous instructions on the state’s burden of persuasion,2 and, while it was at it, gave an instruction on the presumption of innocence, which Leavitt neither objected to nor presented as a separate issue to the Idaho Supreme Court.3 That instruction read as follows:
The rule of law which clothes every person accused of a crime with the presumption of innocence and imposes upon the State the burden of proving his guilt beyond a reasonable doubt is not intended to aid anyone who is in fact guilty to escape, but is a humane provision of the law intended, so far as human agencies can, to guard against the danger of an innocent person being unjustly punished.
The State essentially argues that even if the instruction injected a note of ambiguity into the trial performance, issuance of a writ of habeas corpus on that account is barred by Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). When a state argues “that the defendant seeks the benefit of a new rule of constitutional law, the court must apply Teague before considering the merits of the claim.” Caspari v. Bohlen, 510 U.S. 383, 389, 114 S.Ct. 948, 953, 127 L.Ed.2d 236 (1994). Thus, I shall do that.
As is well known, the Teague rule provides that: “Unless they fall within an exception to the general rule, new constitutional rules of criminal procedure will not be applicable to those cases which have become final before the new rules are an*842nounced.” Teague, 489 U.S. at 310, 109 S.Ct. at 1075. And, “a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final.” Id. at 301, 109 S.Ct. at 1070.
There are generally four steps to the analysis. At the threshold, a court must decide if it is dealing with a criminal procedure rule. See Bousley v. United States, 523 U.S. 614, 619-21, 118 S.Ct. 1604, 1609-10, 140 L.Ed.2d 828 (1998). If it is, it must then move on to the classic Teague considerations:
To apply Teague, a federal court engages in a three-step process. First, it determines the date upon which the defendant’s conviction became final. Second, it must [s]urve[y] the legal landscape as it then existed, and determine whether a state court considering[the defendant’s] claim at the time his conviction became final would have felt compelled by existing precedent to conclude that the rule [he] seeks was required by the Constitution.... Finally, if the court determines that the habeas petitioner seeks the benefit of a new rule, the court must consider whether the relief sought falls within one of the two narrow exceptions to nonretroactivity.
Lambrix v. Singletary, 520 U.S. 518, 527, 117 S.Ct. 1517, 1524-25, 137 L.Ed.2d 771 (1997) (internal citations and quotation marks omitted). In this case, there is an even earlier step which must be taken because Leavitt insists that the state has waived the Teaxjue claim by not raising it in the district court.
I agree with Leavitt that, in general, an issue not raised at the district court is, and should be, deemed waived. See, e.g., Taniguchi v. Schultz, 303 F.3d 950, 958-59 (9th Cir.2002); see also Crawford v. Lungren, 96 F.3d 380, 389 n. 6 (9th Cir.1996). But that rule does not apply here. At the district court, the state did assert that Teague applied to Leavitt’s claims, albeit in a very general way and without detail as to any particular claim. While I would ordinarily look askance at that cavalier approach to the difficult task of issue identification that district judges face, we have applied a softened rule to Teague issues when we discern that the dignity of the state and its judicial processes are involved.4 Thus, we have announced that we can consider Teague issues, even when they have not been raised at all in the district court. See Boardman v. Estelle, 957 F.2d 1523, 1536-37 (9th Cir.1992). We have refused to do so when the state has explicitly declined to raise the issue in the district court and on appeal,5 but that is not this case. Here the state did, at least, allude to the issue in the district court, and it has explicitly raised it before us. I will, therefore, move on to the consideration of the usual four steps.
A. Procedural Rule
The threshold step is mounted successfully. It is rather apparent that Leavitt’s attack on the innocence instruction is based upon the thought that it had an effect on the burden of persuasion. That does not go to the substance of the crime; it is a matter of procedure. See, e.g., In re Winship, 397 U.S. 358, 363-64, 90 S.Ct. 1068, 1072, 25 L.Ed.2d 368 (1970); Harmon v. Marshall, 69 F.3d 963, 966-67 (9th Cir.1995); Adams v. Aiken, 41 F.3d 175, *843177-78 (4th Cir.1994); cf. Melton v. Moore, 964 F.2d 880, 882 (9th Cir.1992) (burden of persuasion change is procedural); Chow v. INS, 641 F.2d 1384, 1391 n. 4 (9th Cir.1981) (same). So, on to the big three.
B. Finality
When was Leavitt’s conviction final? The general rule of finality is easy enough to recite. “A state conviction and sentence become final for purposes of retroactivity analysis when the availability of direct appeal to the state courts has been exhausted and the time for filing a petition for a writ of certiorari has elapsed or a timely filed petition has been finally decided.” Caspars, 510 U.S. at 390, 114 S.Ct. at 953. Leavitt’s conviction was affirmed on May 30, 1989,6 and certiorari was denied on October 16, 1989.7 But in that initial opinion, Leavitt’s sentencing was reversed and returned to the trial court for further proceedings. See Leavitt I, 116 Idaho at 294, 775 P.2d at 608. He was resentenced, that was affirmed on November 27, 1991,8 and certiorari was denied on November 9, 1992.9 Thus, there are two possible dates, and the parties skirmish over which one should be applied here. While as relevant here I see no substantial change in the legal landscape between the two dates, existing authority already resolves the question as far as the innocence instruction is concerned. That goes to the conviction itself, rather than to sentencing, and the conviction must be deemed final at the end of the appellate proceedings on the guilt phase of the trial. See Brady v. Maryland, 373 U.S. 83, 85 n. 1, 83 S.Ct. 1194, 1195 n. 1, 10 L.Ed.2d 215 (1963); Gretzler v. Stewart, 112 F.3d 992, 1004 (9th Cir.1997).
C. Landscape
Turning to the next element, a court must examine the legal landscape on October 16,1989, to see if the rule it is asked to adopt was then so clearly established that “a reasonable jurist” would have thought that it was “dictated by [Supreme Court] precedents.” Caspari, 510 U.S. at 393, 114 S.Ct. at 955. That is, could it be said that “no other interpretation was reasonable,” 10 so that when Idaho chose not to overturn Leavitt’s conviction on account of the innocence instruction, it erred. In that respect, I remain mindful of the fact that a result is not compelled or dictated by Supreme Court precedent simply because it is “within the ‘logical compass’ of an earlier decision.” Jones v. Gomez, 66 F.3d 199, 203 (9th Cir.1995).
When I survey the landscape through those binoculars, it is drear indeed as far as Leavitt’s hopes are concerned. As of that date, the Supreme Court had not said that an instruction like the one at hand violates the United States Constitution. What it had said was that no presumption of innocence instruction need be given. See Kentucky v. Whorton, 441 U.S. 786, 789, 99 S.Ct. 2088, 2090, 60 L.Ed.2d 640 (1979) (per curiam); Taylor v. Kentucky, 436 U.S. 478, 485-86, 98 S.Ct. 1930, 1935, 56 L.Ed.2d 468 (1978). At that time, the circuits were actually split on the issue of whether the instruction was even improper at all. Compare Moffitt v. United States, 154 F.2d 402, 404-05 (1946) with Gomila v. United States, 146 F.2d 372, 373 (5th Cir. *8441944). The split is not too surprising. The instruction itself can be said to be a harsh or confusing one which suggests that if a person is really guilty, it is not intended that he have the benefit of the presumption. That seems like a somewhat strange suggestion, since guilt, itself, can only be found if the person has been shown to be guilty beyond a reasonable doubt. In any event, the instruction can also be seen as a benign attempt to turn away any thought that there is no reason to place a special burden upon the state in order to protect criminals — a thought that is often popularly expressed — by assuring the jury that the presumption is actually intended to protect the innocent from being improperly convicted, which necessitates that it be applied to everyone.11 Be that as it may, the division between the circuits still existed some years later. See United States v. Doyle, 130 F.3d 523, 536-37 (2d Cir.1997).12 The very fact of the intercircuit split militates in favor of a determination that we would be establishing a new rule were we to decide that the instruction in this case violated Leavitt’s constitutional rights. See Turner v. Marshall, 63 F.3d 807, 819 (9th Cir.1995), overruled on other grounds by Tolbert v. Page, 182 F.3d 677, 685 (9th Cir.1999). So, likewise, does the existence of a significant split among state courts.13
In so stating, I hasten to add that I am well aware of the fact that by 1989 this circuit had already declared itself to be on the side of those who reject the instruction. See Reynolds v. United States, 238 F.2d 460, 463 (9th Cir.1956). But we did not state that the instruction violated the United States Constitution. In fact, what we did say was: “Since it is right to instruct on the presumption of innocence, it is wrong to add this self-defeating qualification.” Id. As we saw it, when the qualification is added to the presumption of innocence instruction, “the result is to leave matters about where they would have been had no instruction on the presumption been given.” Id. Of course, as already noted, as far as the Constitution is concerned, if the situation became the same as if no instruction at all had been given, that would be fine. Moreover, Reynolds was a direct appeal case, and a rule announced by us for direct appeal purposes does not translate itself into directives of the Supreme Court for Teague purposes.14
*845Also, I am aware of the fact -that the presumption of innocence instruction in this ease alluded to the beyond a reasonable doubt burden of persuasion. But, in context, that does not make the instruction more or less ambiguous, and does not change the legal landscape as of 1989 in any significant way. In a sense, the two are different sides of the same coin, and courts have not placed weight upon that difference. Rather, they have treated the instruction as one about the presumption of innocence itself. See, e.g., Doyle, 130 F.3d at 533-34 (and cases cited therein); United States v. Bifield, 702 F.2d 342, 350-51 (2d Cir.1983); Moffitt, 154 F.2d at 404-05. I will do the same because, again, I am satisfied that to determine that the offending language violates the Constitution of the United States would create a new rule. That determination is, therefore, barred by Teague. See DelValle, 306 F.3d at 1200.
D. Exceptions
Finally, a court must ask whether the new rule, were it adopted, would come within one of the two Teague exceptions. Those exceptions are: (1) determinations that place private conduct beyond the proscriptive power of the state15 and (2) those that establish “watershed rules of criminal procedure.”16 Leavitt does not assert that the first of these applies to his case; he could not do so successfully. He does assert that the second exception applies here; I think not.
As the Supreme Court has explained, it is not enough that a rule improves trial accuracy; “[a] rule that qualifies under this exception must not only improve accuracy, but also alter our understanding of the bedrock procedural elements essential to the fairness of a proceeding.” Sawyer v. Smith, 497 U.S. 227, 242, 110 S.Ct. 2822, 2831, 111 L.Ed.2d 193 (1990) (citation and internal quotation marks omitted). It is unlikely that' it could be said that the presumed-innocence-instruction rule argued for here would meet the bedrock standard. In fact, as the Court has said, it is “unlikely that many such components of basic due process have yet to emerge.” Id. at 243, 110 S.Ct. at 2832 (citation omitted).
That thought is very relevant because the Court has recently had occasion to reflect upon the retroactivity of Cage v. Louisiana, 498 U.S. 39, 111 S.Ct. 328, 112 L.Ed.2d 339 (1990). In Cage, the Court dealt directly with the definition of beyond a reasonable doubt, and decided that the instruction in question there diluted the burden of persuasion enough to amount to a due process violation. Id., at 41, 111 S.Ct. at 329-30. Then, in Tyler v. Cain, 533 U.S. 656, 121 S.Ct. 2478, 150 L.Ed.2d 632 (2001), the Court was asked to consider whether its holding in Cage was retroactive to cases on collateral review.
While Tyler did not decide the Teague question, it did have some interesting and important ruminations about it. Tyler had argued that still another Supreme Court decision,17 which said that Cage error was structural, “makes clear that retroactive application [of Cage] is warranted by the principles of Teague ....” Id. at 665, 121 S.Ct. at 2483. Not so, said the Court, because, among other things, no case has *846held “that all structural-error rules fit within the second Teague exception.” Id. at 666, 121 S.Ct. at 2484. “[A] holding that a particular error is structural does not logically dictate the conclusion that the second Teague exception has been met.” Id. at 666-67, 121 S.Ct. at 2484. The Court then further discussed Teague, and finished by stating that “it is unlikely that any of these watershed rules ‘ha[s] yet to emerge.’ ” Id. at 666-67 n. 7, 121 S.Ct. at 2484 n. 7 (citation omitted).
Again, Tyler did not actually decide the issue, but its tenor cannot help but give one pause despite (or, perhaps, because of) the fact that before it was decided, a number of courts of appeals had stated that Cage was, indeed, within the second Teague exception. See Williams v. Cain, 229 F.3d 468, 473 (5th Cir.2000); Tillman v. Cook, 215 F.3d 1116, 1122 (10th Cir.2000); West v. Vaughn, 204 F.3d 53, 55 (3d Cir.2000); Gaines v. Kelly, 202 F.3d 598, 605 (2d Cir.2000); Adams v. Aiken, 41 F.3d 175, 178-79 (4th Cir.1994); Nutter v. White, 39 F.3d 1154, 1157-58 (11th Cir.1994). Tyler does seem to etiolate those holdings. Whether the courts that made them will stay on the same course now that they can see the brume emanating from Tyler is anyone’s guess, but, at the very least, a court should be wary and should shy away from the temptation to dub any disapproval of the ambiguous instruction at hand a bedrock rule, even were it to decide that the instruction violates the Constitution.18
Especially is that true when I reflect upon the fact that many cases have refused to find that instructions which lower the prosecution’s burden of persuasion in some respects are within the second Teag-ue exception. That has been notable in the area of so-called Sandstrom19 error. In Sandstrom, the Supreme Court declared that a presumption instruction was unconstitutional because it lifted from the state’s shoulders the burden that it “prove ‘beyond a reasonable doubt ... every fact necessary to constitute the crime ... charged.’ ” Id. at 523, 99 S.Ct. at 2459 (citation omitted). Despite that instruction’s ultimately baleful effect on the burden of persuasion, most of the federal courts of appeals that have considered the issue have held that Sandstrom error is not within the second Teague exception. See Johnson v. McKune, 288 F.3d 1187, 1200 (10th Cir.2002); Lyons v. Stovall, 188 F.3d 327, 341 (6th Cir.1999); Cain v. Redman, 947 F.2d 817, 822 (6th Cir.1991); Prihoda v. McCaughtry, 910 F.2d 1379, 1382 (7th Cir.1990); contra Hall v. Kelso, 892 F.2d 1541, 1543 n. 1 (11th Cir.1990). Moreover, the same view has been taken of other instructions that allegedly undermine, without entirely overturning, the beyond-a-reasonable-doubt burden that is placed upon the prosecution. See United States v. Mandanici, 205 F.3d 519, 529-30 (2d Cir.2000); Thompson v. Dixon, 987 F.2d 1038, 1043 (4th Cir.1993); see also Willis v. Aiken, 8 F.3d 556, 558, 568 (7th Cir.1993) (confusing instruction).
On a continuum, the innocence instruction at hand is closer in effect to the Sandstrom case than it is to the Cage case. That is, assuming that the instruction does inject some ambiguity into consideration of the role of the presumption of innocence, it certainly does not inform the jury that it need not find the defendant guilty beyond a reasonable doubt. Considering the Ty*847ler cloud that Cage retroactivity itself is under, I cannot say that the innocence instruction comes within the second Teag-ue exception.
In fine, were I to declare that the ambiguous presumed innocence instruction was unconstitutional and apply it to this case, I would be adopting a new rule in violation of Teague. The district' court did just that and, for that reason, I agree that we are required to reverse its decision to issue a writ of habeas corpus on that basis.
II. ALIBI INSTRUCTION
Leavitt asked the trial judge to give an alibi instruction in a particular form. The trial judge complied with his request and, not surprisingly, Leavitt did not object or suggest any changes. Nor did he raise the question on appeal to the Idaho Supreme Court. Yet, by the time he reached the district court he was arguing that the instruction had some defect.. The district court said that if the instruction contained an error,20 that error was invited and, therefore, the Idaho courts would have rejected an appeal based upon it.21 On appeal Leavitt argues that the district court could not so decide because there is no Idaho doctrine of invited error when it comes to instructions.22 That claim invites us to commit error.
Long ago, the Idaho Supreme Court stated that a defendant who induces the trial court to instruct in a particular manner cannot then argue error on appeal. See State v. Lopez, 100 Idaho 99, 593 P.2d 1003 (1979). In that case, the defendant argued against the giving of an instruction, so the trial court did not give it. Id. at 102, 593 P.2d at 1006. He then tried to argue error on appeal and elicited this response: “The failure of the trial court to instruct on assault with a deadly weapon was caused by defendant’s objection and therefore was invited error and will not be considered on appeal.” Id. As the court explained in a later case: “The purpose of the invited error doctrine is to prevent a party who caused or played an important role in prompting a trial court to give or not give an instruction from later challenging that decision on appeal.” State v. Blake, 133 Idaho 237, 240, 985 P.2d 117, 120 (1999); see also State v. Carlson, 134 Idaho 389, 402, 3 P.3d 67, 80 (Idaho Ct.App.2000). And similarly in a civil case, the court opined that when error was invited by a request for an erroneous instruction, that could not be used to obtain a reversal on appeal. See Laidlaw v. Barker, 78 Idaho 67, 75, 297 P.2d 287, 291 (1956), overruled on other grounds by Crane v. Banner, 93 Idaho 69, 455 P.2d 313 (1969).
That authority is the broadside which sinks Leavitt’s argument here. He did not merely fail to object to an instruction; he asked that it be given. He cannot now be heard to say that he is entitled to habeas corpus relief because his wishes were ac*848ceded to by the trial court.23 The district court did not err when it rejected this claim.24
Thus, I respectfully concur in the per curiam opinion.
. In addition, I do not join in footnote 28 of the per curiam opinion because the issues discussed there were not raised at the district court. Had they been, I would agree with their resolution in the per curiam.
. The innocence instruction was given with a group of others that explicated the state's burden to prove Leavitt's guilt beyond a reasonable doubt. No serious claim is made that the others, alone or together, violated Leav-itt's constitutional rights. Thus, the attack here comes down to an assault on the innocence instruction itself within its overall contextual setting, and it is to that instruction that I, therefore, direct my attention. I take up Leavitt’s separate attack on an alibi instruction in part II of this concurring opinion.
.However, taking the Idaho Supreme Court at its word, as I must, I presume that it reviewed this claim because it said: "Since the instant case involves a conviction of first degree murder and the imposition of the death penalty, we have carefully reviewed the record for any indication of prejudicial error occurring at trial, regardless of whether or not error has been specifically asserted by the defendant.” Leavitt I, 116 Idaho at 288, 775 P.2d at 602.
. We have done the same regarding procedural default issues. See Windham v. Merkle, 163 F.3d 1092, 1100-01 (9th Cir.1998); cf. Vang v. Nevada, 329 F.3d 1069, 1073 (9th Cir.2003) (state on notice, but did not raise claim).
. Garceau v. Woodford, 281 F.3d 919, 920 (9th Cir.2002), rev'd on other grounds by 538 U.S. 202, 123 S.Ct. 1398, 155 L.Ed.2d 363 (2003).
. Leavitt I, 116 Idaho at 285, 775 P.2d at 599.
. Idaho v. Leavitt, 493 U.S. 923, 110 S.Ct. 290, 291, 107 L.Ed.2d 270 (1989).
. Leavitt II, 121 Idaho at 4, 822 P.2d at 523.
. Leavitt v. Idaho, 506 U.S. 972, 113 S.Ct. 460, 121 L.Ed.2d 368 (1992).
. Lambrix, 520 U.S. at 538, 117 S.Ct. at 1530.
. See, e.g., Turner v. State, 102 Ind. 425, 1 N.E. 869, 870 (1885); State v. Cugliata, 372 A.2d 1019, 1032 (Me.1977), overruled on other grounds by State v. Brewer, 505 A.2d 774, 777 n. 5 (Me.1985); State v. Hanlon, 38 Mont. 557, 100 P. 1035, 1044 (1909).
. Incidentally, while in Doyle the court came down on the side of those who disapprove of the instruction, it expressly declined to decide that its decision would apply to state proceedings on habeas corpus review. Id. at 540 n. 14. Later on, the court declared that the rule was limited to direct review cases. See DelValle v. Armstrong, 306 F.3d 1197, 1200 (2d Cir.2002).
. See, e.g., State v. Schiappa, 248 Conn. 132, 171-72, 728 A.2d 466, 487-88 (1999); State v. Cari, 163 Conn. 174, 180, 303 A.2d 7, 10 (1972) (and cases cited therein); People v. Dowling, 95 Ill.App.2d 223, 228, 238 N.E.2d 131, 133 (Ill.Ct.App.1968); People v. Rees, 268 Ill. 585, 594-595, 109 N.E. 473, 476 (1915); State v. Barnes, 202 Kan. 21, 24, 446 P.2d 774, 776 (1968); State v. Medley, 54 Kan. 627, 39 P. 227, 227 (1895); State v. Gee Jon, 46 Nev. 418, 211 P. 676, 679 (1923).
.Bell v. Hill, 190 F.3d 1089, 1092-93 (9th Cir.1999), is not to the contrary. In that case, we held that a rule we had announced in a previous habeas corpus case bound us in the case then before us. That was not based on Teague; it was based on the binding effect of our own habeas corpus jurisprudence. Nor does Belmontes v. Woodford, 335 F.3d 1024, 1047-48, amended by 350 F.3d 861 (9th Cir.2003) change matters. It, too, relied on an earlier case wherein we had already applied the rule in question to a § 2254 petition, and, *845while we referred to Teague, the law of the circuit compelled that reliance.
.Lambrix, 520 U.S. at 539, 117 S.Ct. at 1530-31.
. Lambrix, 520 U.S. at 539, 117 S.Ct. at 1531.
. Sullivan v. Louisiana, 508 U.S. 275, 279-81, 113 S.Ct. 2078, 2081-82, 124 L.Ed.2d 182 (1993).
. We have recognized that, in light of Tyler, the second exception of Teague is even more stringent than structural error. See United States v. Sanchez-Cervantes, 282 F.3d 664, 670 (9th Cir.2002); see also Jarrett v. United States, 266 F.3d 789, 791 n. 1 (8th Cir.2001).
. Sandstrom v. Montana, 442 U.S. 510, 99 S.Ct. 2450, 61 L.Ed.2d 39 (1979).
. The district court thought that the instruction was erroneous, but, as I see it, we need not decide that question.
. Of course, that is a basis for deciding not to reach the merits of an issue on habeas corpus review. See Fields v. Bagley, 275 F.3d 478, 486 (6th Cir.2001); Leverett v. Spears, 877 F.2d 921, 924 (11th Cir.1989); Miller v. Oberhauser, 293 F.2d 29, 31 (9th Cir.1961).
.Aside from his argument about the propriety of the instruction itself, the only other argument made by Leavitt is that when the district court said it would not consider issues procedurally defaulted, it could not thereafter consider invited error. That is an interesting semantic argument, but it isr plain that the district court considered invited error to be a separate concept, and I am in no position to tell it that it really meant no such thing.
. Leavitt makes an ineffective attempt to save his ship by asserting that the broadside actually fired blanks due to State v. Nunez, 133 Idaho 13, 981 P.2d 738 (1999). That case does not help him. What it says is that where a defendant submits an erroneous instruction which results in his being erroneously convicted of a misdemeanor rather than a felony, the prosecutor cannot have the defendant sentenced for a felony anyway' on the basis that the defendant invited the error. Id. at 19-20 & n. 3, 981 P.2d at 744-45 & n. 3. That only suggests that the defendant gets both the benefit and the detriment of his accepted invitations.
. Even if the instruction was constitutionally defective, an issue I see no need to reach, it is plain that any error would be harmless. The only evidence of alibi was Leavitt's own statement, that he was at home watching T.V. which, that starkly put, is not really different from: "I did not do it.” Even that, by the way, was significantly impeached. In light of the evidence in this case, and the jury’s obvious rejection of Leavitt’s stories, it is almost inconceivable that the error had any injurious effect or influence upon the jury’s verdict, much less a substantial one. Brecht, 507 U.S. at 638, 113 S.Ct. at 1722; see also California v. Roy, 519 U.S. 2, 4-5, 117 S.Ct. 337, 338, 136 L.Ed.2d 266 (1996).