This case involves a deep and seemingly intractable conflict between principles at the heart of our representative government: the right of the people to choose whom they please to govern them, and the authority of a state to determine the structure of its political system. It also raises the novel question whether, when the people vote on an initiative measure that, if enacted, would severely limit a fundamental, constitutional right, they must be given adequate notice of the measure’s effect.
In 1990, by a narrow margin, California voters passed Proposition 140, which, as construed by the California Supreme Court, amends the California Constitution to impose a lifetime limit on the number of terms that an individual may serve in state legislative office. The plaintiffs in this case — who con*844sist of former state legislators who are permanently barred by Proposition 140 from ever seeking re-eleetion to the house in which they served, and residents of their respective districts who can never again vote to re-elect state legislators barred by Proposition 140 — together allege that Proposition 140’s lifetime term limits impose an unconstitutional burden on their fundamental rights of voting and association under the First and Fourteenth Amendments. The district court agreed and invalidated the provision.
We affirm the judgment, but not the rationale, of the district court. We do not decide whether a state may adopt lifetime term limits for its legislators without violating the Constitution. We recognize that the state has a compelling interest as a sovereign in choosing the structure of its institutions and the qualifications of its officials — an interest that may suffice to render lifetime term limits constitutional, although we do not decide that question here. Instead, we invalidate the challenged lifetime ban because we hold as a matter of federal constitutional law that a state initiative measure cannot impose a severe limitation on the people’s fundamental rights when the issue of whether to impose such a limitation on those rights is put to the voters in a measure that is ambiguous on its face and that fails to mention in its text, the proponents’ ballot arguments, or the state’s official descriptions, the severe limitation to be imposed. Neither Proposition 140, the proponents’ ballot arguments, nor the material prepared by the state mentioned “lifetime” term limits; thus, the voters were not afforded adequate notice of the severity of the limitation involved.
BACKGROUND
On November 6, 1990, California voters narrowly passed Proposition 140, which amended the California Constitution to limit the number of terms that individuals may serve in certain state offices, to curtail the retirement benefits of state legislators, and to reduce or eliminate legislative staff and support services. Cal. Const, art. IV, § 2(a) (term limits for state legislators); id. § 4.5 (retirement benefits for state legislators); id. § 7.5 (operating expenses of legislature and compensation of legislature members and staff); see also id. art. V, § 2 (gubernatorial term limits); id. art. V, § 11, art. IX, § 2, art. XIII, § 17 (term limits for other state executive officials). Although other state offices were affected, the thrust of the measure was clearly aimed at adopting term limits for the state legislature.1
There are two kinds of term limits: consecutive term limits and lifetime term limits. Consecutive term limits restrict the number of terms that a representative may serve in a row. Under this approach, which is employed by a majority of the states that have adopted legislative term limits,2 legislators generally may return to office after a reasonable hiatus. While consecutive term limits force successful incumbents to step down from office after some fixed period of service (usually after two or three terms), they allow voters to re-elect those same persons at a future time, when they no longer have the advantage of incumbency. Lifetime term limits, on the other hand, are far more restrictive. Once a legislator has served a given number of terms as a senator or assemblyman, voters are barred for life from ever again electing that person to such office.
California proposed and adopted its term limits by means of the state’s initiative process. See Cal. Elec.Code § 9000 et seq.; § 13280-82. Because Proposition. 140 was a proposed constitutional amendment, the proponents were required to present to the Secretary of State a petition signed by eight *845percent of the state’s registered voters. See id. § 9035. After doing so, the state prepared a “ballot pamphlet,” which was sent to all registered voters. See id. § 9094. The ballot pamphlet contained a complete copy of the measure; an official title and summary of the Proposition prepared by the Attorney General; an official analysis prepared by the Legislative Analyst; and “ballot arguments” by the proponents and opponents for and against the measure. See id. §§ 9084-86.3
The official title of Proposition 140 as it appeared in the ballot pamphlet read: “Limits of Terms of Office, Legislators’ Retirement, Legislative Operating Costs. Initiative Constitutional Amendment.” The Proposition itself commenced with the following preamble, which is part of the constitutional amendment:
The people find and declare that the Founding Fathers established a system of representative government based upon free, fair, and competitive elections. The increased concentration of political power in the hands of incumbent representatives has made our electoral system less free, less competitive, and less representative.
The ability of legislators to serve unlimited number of terms [sic], to establish their own retirement system, and to pay for staff and support services at state expense contribute heavily to the extremely high number of incumbents who are reelected. These unfair incumbent advantages discourage qualified candidates from seeking public office and create a class of career politicians, instead of the citizen representatives envisioned by the Founding Fathers. These career politicians become representatives of the bureaucracy, rather than of the people whom they are elected to represent.
To restore a free and democratic system of fair elections, and to encourage qualified candidates to seek public office, the people find and declare that the powers of incumbency must be limited. Retirement benefits must be restricted, state-financed incumbent staff and support services limited, and limitations placed upon the number of terms which may be served.
Cal. Const., art. IV, § 1.5 (as amended). The Proposition, in relevant part, proposed to amend the California constitutional provision governing legislators by adding the following language:
No Senator may serve more than 2 terms. ... No member of the Assembly may serve more than 3 terms.
Cal. Const., art. IV, § 2(a) (as amended) (emphasis added). Nowhere in the initiative, the title and summary, the Legislative Analyst’s statement, or the proponents’ ballot arguments, was there any mention of lifetime limits.4
On the November 6, 1990 general election, the measure was put to a statewide vote. The following language appeared on the ballot: 5
TERMS OF OFFICE. LEGISLATURE. INITIATIVE CONSTITUTIONAL AMENDMENT. Limits: terms for specified state elected officials, legislators’ retirement, pensions. Legislature’s operating costs.... Yes__No__
The measure passed by a statewide margin of 52.17% to 47.83%.
In 1991, a group of petitioners consisting of the California Senate and Assembly, a number of individual legislators, and individual constituents challenged the validity of Proposition 140 directly in the California Supreme Court by way of a petition for writ of mandate. Legislature v. Eu, 54 Cal.3d 492, 286 Cal.Rptr. 283, 816 P.2d 1309 (1991). Neither Tom Bates nor any of his constitu*846ents was a party to the Eu action. Because the California Supreme Court chose to exercise its original jurisdiction, no trial was conducted, and no lower court examined the merits of the parties’ legal and factual contentions. Id. at 500, 286 Cal.Rptr. 283, 816 P.2d 1309.
Petitioners contended that the Proposition was unconstitutional because it imposed lifetime term limits. The state of California6 argued that Proposition 140 did not contain a lifetime ban and that it merely prohibited a “Senator” or “member of the Assembly,” rather than a “person,” from seeking reelection; the state contended that once an incumbent-legislator has left office “he or she can no longer be described as a ‘Senator’ or ‘member of the Assembly’ to which the term limitation provision would apply.” Id. at 504, 286 Cal.Rptr. 283, 816 P.2d 1309. The state further urged that “a lifetime ban on candidacy is not essential to accomplish the voters’ purpose,” and that a narrow interpretation of Proposition 140 would “avoid[ ] conflict with the Constitution” and be “less intrusive of constitutionally protected rights.” Applying a highly deferential state-law standard that requires it to adopt “all presumptions [in] favor [of] the validity of initiative measures” and to uphold such measures “unless their unconstitutionality clearly, positively, and unmistakably appears,” id. at 501, 286 Cal.Rptr. 283, 816 P.2d 1309, the California Supreme Court upheld, against petitioners’ state and federal constitutional challenges, the term limits provisions of Proposition 140. Rejecting any need to adopt a narrow construction to avoid unnecessarily deciding difficult constitutional questions, it further concluded that “the measure contemplated a lifetime ban.” Id. at 505, 286 Cal.Rptr. 283, 816 P.2d 1309.
The court first expressed its agreement with one of the arguments advanced by the state and specifically found that “the language of Proposition U-0 is ambiguous as to its intent to impose a lifetime ban.” Id. at 504, 286 Cal.Rptr. 283, 816 P.2d 1309 (emphasis added). However, after examining the other official material presented to the voters in the ballot pamphlet, it concluded that Proposition 140’s framers and supporters had contemplated such a ban. Id. In doing so, it implicitly acknowledged the fact that the drafters and supporters not only failed to mention a lifetime ban in the text of the initiative itself, but also failed to do so in their official ballot arguments, both direct and in rebuttal. Ignoring these compelling circumstances, the state court, after specifically conceding that a ballot measure’s opponents are likely to overstate the effects of a measure and that their ballot arguments are not “highly authoritative” as to the measure’s actual meaning, resolved the crucial ambiguity in the ballot measure in reliance principally on the ballot arguments advanced by Proposition 140’s opponents.7 The California Supreme Court, in terms that can only be described as hesitant, then said that “[w]e think it likely the average voter ... would conclude [that Proposition 140] contemplated a lifetime ban.... ” Id.
The California Supreme Court, after setting forth its rule that requires all doubts regarding an initiative measure to be resolved in favor of constitutionality, id. at 501, 286 Cal.Rptr. 283, 816 P.2d 1309, balanced the state interests in support of lifetime term limits against the burden on the constitutional rights of voters and candidates and ruled in favor of the state. The Supreme Court denied certiorari. Californians for Citizen Gov’t v. Legislature of the State of Cal., 503 *847U.S. 919, 112 S.Ct. 1293, 117 L.Ed.2d 516 (1992) (mem.).
In 1995, plaintiff Tom Bates and a number of his constituents filed suit in district court seeking declaratory and injunctive relief against enforcement of Proposition 140’s term limits provisions. Bates represented the 14th Assembly District continuously from his initial election in 1977 until 1996. As a result of Proposition 140, Bates became ineligible to run for re-election in November of 1996, and is forever barred from seeking any other seat in the Assembly as well. Plaintiffs Browning, Buehalter, Graham, Lewis, Lyman, Sherman, and Sterling are voters in Bates’ former district. The district court found that they would have voted for Bates in 1996 had he been a candidate. Bates v. Jones, 958 F.Supp. 1446, 1456 (N.D.Cal.1997). These voters found Bates to be “an exceptional representative, who served his constituents well and was unusually concerned with and effective at representing the needs of low-income and disabled citizens and protecting the environment.” Id. In the view of these voter-plaintiffs, other candidates are unlikely to represent them as effectively or as well as Bates.
The state argued that Bates’ action is barred by collateral estoppel. In two rulings, one published, Bates v. Jones, 904 F.Supp. 1080 (N.D.Cal.1995) (“Bates I”), and one unpublished, the district court stated that Bates and the Eu petitioners shared the same interests but concluded as a matter of state law that neither Bates nor the voter-plaintiffs had been in privity with the Eu petitioners and therefore were not barred from bringing their action. Id. at 1088-91.8
The case proceeded to trial in October of 1996. Together, the parties presented evidence on an array of issues, including the history of term limits in the United States, the actual strength of the interests asserted by the state, the effectiveness of lifetime term limits as a means of furthering those interests, and the availability of less drastic alternatives.
*848In its final decision, Bates v. Jones, 958 F.Supp. 1446 (N.D.Cal.1997), the district court concluded in a careful and thoughtfully reasoned opinion that Proposition 140’s lifetime term limits for state legislators were unconstitutional, and granted declaratory and injunctive relief against their enforcement. Id. at 1471. The district court applied the balancing test set forth in Anderson v. Celebrezze, 460 U.S. 780, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983), and Burdick v. Takushi, 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992), for determining the constitutionality of state laws that regulate the election process and burden the rights of voters and candidates.
The district court stayed its injunction pending appeal. The state and intervenors timely appeal' the district court’s decision on the merits, and plaintiffs timely appeal the stay of injunctive relief pending appeal.9
DISCUSSION
I. Res Judicata
We first consider the state’s procedural challenge to plaintiffs’ right to pursue this action. The state contends that the res judicata effect of the California Supreme Court’s decision in Legislature v. Eu bars Bates’ claims, if not those of his former constituents as well. We reject the state’s argument.
We apply state law in determining the preclusive effect of a prior state court adjudication, 28 U.S.C. § 1738, and we review questions of res judicata and collateral estoppel de novo. In re Russell, 76 F.3d 242, 244 (9th Cir.1996). Under California law, collateral estoppel bars parties from relitigating issues of fact or law that have already been fully and fairly litigated in prior proceedings. Lumpkin v. Jordan, 49 Cal.App.4th 1223, 57 Cal.Rptr.2d 303, 307 (1996). A party asserting collateral estoppel must demonstrate that the factual or legal issue is identical to the issue that was previously litigated; that the issue was actually litigated in a prior proceeding and there was a final judgment on the merits; and that the party to be estopped in the current action was in privity with a party in the former proceeding. Lucido v. Superior Ct., 51 Cal.3d 335, 341, 272 Cal.Rptr. 767, 795 P.2d 1223 (1990); Lumpkin, 57 Cal.Rptr.2d at 307. Here, the district court found that the first two requirements had been met, but concluded that the state had failed to demonstrate privity between Bates and the Eu petitioners. We will assume, arguendo, that the district court’s finding regarding the first two requirements is correct because we agree with its decision as to the third.
Although “privity” is not a clearly defined concept under California law, it generally refers to persons who are so identified in interest with the parties to the original action that it is proper to bind them along with the parties to the original litigation. Dyson v. California State Personnel Bd., 213 Cal.App.3d 711, 262 Cal.Rptr. 112, 119 (1989). California courts look to whether the relationship between the party in the earlier suit and the nonparty bringing the current suit is “sufficiently close” to justify preclusion of the latter. Id.; People v. Drinkhouse, 4 Cal.App.3d 931, 84 Cal.Rptr. 773, 776 (1970). Due process requires at a minimum that the party to be estopped “must have had an identity or community of interest with, and adequate representation by, the losing party in the first action,” and “should reasonably have expected to be bound by the prior adjudication.” White Motor Corp. v. Teresinski, 214 Cal.App.3d 754, 263 Cal.Rptr. 26, 29 (1989).
With respect to the first inquiry — whether the nonparty had an identity of interest with, and adequate representation by, the losing party in the first action' — due process requires both that the prior litigation of the issue have been motivated by the same underlying purposes, and that the original party have had an incentive and opportunity to litigate the issue in the manner best suited to furthering those common underlying purposes. See, e.g., Clemmer v. Hartford Ins. Co., 22 Cal.3d 865, 151 Cal.Rptr. 285, 290, 587 P.2d 1098, 1103 (1978) (finding no privity where issue of “wilfullness” of homicide was litigated first as criminal defense and later to *849recover defendant’s liability insurance). With respect to the second inquiry — whether the nonparty should reasonably have expected to be bound — this notice requirement is satisfied if: (1) the nonparty had received actual notice at the time of the litigation that his interests were being litigated, see Columbus Line, Inc. v. Gray Line Sight-Seeing Companies Associated, Inc., 120 Cal.App.3d 622, 174 Cal.Rptr. 527 (1981); (2) the nonparty “had in reality contested the prior action even if he did not make a formal appearance,” such as where the nonparty had a financial interest in and “power to control” the litigation of the prior action, Lynch v. Glass, 44 Cal.App.3d 943, 949, 119 Cal.Rptr. 139 (Cal.App. 1 Dist.1975); or (3) the unsuccessful party in the first action could “fairly be treated as acting in a representative capacity” for the nonparty now being precluded. Id.
The district court concluded that of Bates’ constituents, plaintiffs Sterling, Sherman, Graham, and Browning had no identity of interest with the Eu petitioners. Sterling and Sherman were not California residente at the time of the Eu litigation, while Graham and Browning were not yet old enough to vote. Bates I, 904 F.Supp. at 1090. As to plaintiffs Lyman, Lewis, and Bates himself, it found “no indication” that they were “on notice that their First Amendment rights would be foreclosed” by Eu, given that Eu was not a class action and neither Bates nor any voters from his district participated. Id. The district court also concluded that the Eu petitioners had not adequately represented the interests of the plaintiffs in this case. It reasoned that the legislature lacked statutory authority to sue on their behalf and had been “representing its own institutional interests rather than the interests of its members or constituents.” Id. It also noted that plaintiffs had enjoyed “no control” over the Eu petitioners’ unsuccessful litigation strategies, including their “tactical” decisions to bring their action “directly to the California Supreme Court on a petition for mandate, rather than pursuing the more usual course of litigation beginning with fact-based proceedings in a trial court,” and to argue for the broadest possible interpretation of Proposition 140 — the interpretation that would most drastically limit the constitutional rights of Bates and his constituents. Id. Finally, it observed that the plaintiffs in this case had neither requested that the legislature bring the prior action nor participated in it, such as by providing litigation materials or testifying. Id. The subsequent discovery that Bates’ campaign committee had contributed $ 1,000 toward the costs of the Eu litigation, albeit without his knowledge, did not alter the district court’s privity analysis. The district court reasoned that Bates’ contribution of “approximately 0.2%” of the funds that supported the Eu litigation neither gave him “any control of the litigation” nor caused his interests to coincide with those of the Eu petitioners. Id. at Order Denying Defendant Jones’ Motion for Summary Judgment and Granting Plaintiffs’ Motion for Leave to Amend at 9. It also observed that there would be no “prudential reasons” to apply res judicata against Bates if the complaint were amended to add as plaintiffs Escutia and Friedman, who were not members of the legislature at the time of the Eu litigation and hence would not be subject to the same res judicata arguments. Id. at 11-12.
We agree with the district court’s holding in general and particularly with its conclusion that the Eu petitioners did not adequately represent the interests of Bates and his former constituents. Whatever their reasons, the Eu petitioners urged the court to construe the initiative in the manner that would most severely limit plaintiffs’ fundamental rights. That the present plaintiffs and the Eu petitioners are represented by the same law firm is irrelevant; it cannot be inferred that different clients share the same interests or would choose the same strategies simply because they are represented by the same attorneys. The fact that Bates’ campaign committee contributed a minute fraction of the costs of the Eu litigation fails to establish privity. The state does not even dispute the district court’s factual findings that the contribution was unwitting, and that no plaintiff in the present ease enjoyed control over any aspect of the Eu litigation as a result of the contribution or for any other reason.10
*850We hold, in the alternative, that even assuming that Bates and his constituents were in privity with the Eu petitioners, the district court’s res judicata ruling is correct. Under California law, even when the formal prerequisites for collateral estoppel are present, the public interest may require that relitigation of certain questions of law not be foreclosed. It is the “established rule” in California that a “prior legal determination is not conclusive ... if the public interest requires that relitigation not before-closed.” Kopp v. Fair Political Practices Comm’n, 11 Cal.4th 607, 47 Cal.Rptr.2d 108, 116 n. 16, 905 P.2d 1248, 1257 n. 16 (1995) (emphasis in original). The initial determination of an issue need not be erroneous to warrant reexamination under the “public interest” exception to res judicata. Id. Instead, reexamination of a previously decided legal question is appropriate where the potential adverse impact on the public or persons who were not parties to the previous action “demonstrates a convincing need for a new determination of the issue.” Rutherford v. State, 188 Cal.App.3d 1267, 233 Cal.Rptr. 781, 789 (1987) (applying the public interest exception to reexamine the constitutionality of a state environmental statute and the manner of its enforcement); Arcadia Unified Sch. Dist. v. State Dep’t of Educ., 2 Cal.4th 251, 5 Cal.Rptr.2d 545, 547-49, 825 P.2d 438, 440-42 (1992) (applying the public interest exception to permit reexamination of the constitutionality of a statute affecting “the public in general”); City of Sacramento v. State, 50 Cal.3d 51, 266 Cal.Rptr. 139, 145-46, 785 P.2d 522, 528-29 (1990) (applying the public interest exception where large numbers of citizens might otherwise be adversely affected by unjust tax impositions); California Optometric Ass’n v. Lackner, 60 Cal.App.3d 500, 131 Cal.Rptr. 744, 747 (1976) (invoking the public interest exception to review an administrafive agency’s violation of its ongoing obligation to administer statutes enacted for the public benefit).
This case is highly appropriate for application of the public interest exception. As we explain below, Proposition 140 imposes a severe burden on fundamental constitutional rights of voting and association that are “preservative of other basic civil and political rights.” Reynolds v. Sims, 377 U.S. 533, 562, 84 S.Ct. 1362, 1381, 12 L.Ed.2d 506 (1964). Moreover, the most serious effects of Proposition 140 are suffered by voters, who, with only a few exceptions, did not participate in, or even have notice of, the action in Eu. Finally, since the California Supreme Court’s decision, the Supreme Court has decided two cases clarifying issues about which the California Supreme Court expressed uncertainty and indicating that the state’s lifetime ban must be more carefully scrutinized as a matter of federal law than some, including the California Supreme Court, had previously believed necessary.11 Cf. Minnis v. United States Dep’t of Agriculture, 737 F.2d 784, 786 n. 1 (9th Cir.1984) (noting that presence of intervening Supreme Court decisions that clarify issue defeats collateral estoppel claim), cert. denied, 471 U.S. 1053, 105 S.Ct. 2112, 85 L.Ed.2d 477 (1985).
We conclude that application of the public interest exception is warranted by the overwhelming public importance of the issues involved; the numbers of persons whose fundamental rights were severely limited as a result of a writ proceeding that was conducted primarily between governmental entities and that bypassed the ordinary judicial procedures; and the existence of intervening, clarifying authority. In concluding that the public interest exception permits a federal court to reexamine federal questions previously decided in state court, we do not, of *851course, fashion a new doctrine in tension with principles of comity and federalism. Rather, we apply the law of California as fashioned by the California Supreme Court.12
II. Proposition 140 and the Burden it Imposes
Proposition 140 presents a novel and difficult question of constitutional law. Neither the Supreme Court,13 nor any circuit court,14 has ever considered the constitution*852ality of lifetime term limits for state legislators or, indeed, any state officials. A lifetime ban is, for two primary reasons, different in kind from all other state laws affecting the election process that the Supreme Court and circuit courts have previously considered in a First Amendment context.
First, Proposition 140 affects a change in the state’s governmental structure that is far more basic than any provision dealt with in prior eases examining state election laws for First Amendment violations. The state interest underpinning lifetime term limits is not merely the customary one of regulating elections, but, rather, it lies in the state’s power to determine the fundamental nature of its legislature.15 Here, it does this by attempting to ensure that future lawmakers will all be “citizen-legislators.” In contrast, as the Court said recently in U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995), the state election laws at issue in prior cases (other than equal protection cases), related to the states’ adoption of “election procedures,” and were designed to ensure that elections would be conducted in an “orderly, fair, and honest” manner. Id. at 834-35, 115 S.Ct. at 1869-70.16
Second, the severity of the burden that Proposition 140 imposes on the people’s right to vote for the candidates of their choice far exceeds any limitation imposed by laws that simply regulate the state electoral process.17 In the past, the burdens imposed by state election laws have required candidates for office to comply only with “procedural hurdles” in order to earn a place on the ballot. Thus, as the Court explained in U.S. Term *853Limits, none of its decisions upholding state laws as legitimate exercises of the power to regulate elections addressed a law that limited voting rights by “exclud[ing] candidates from the ballot without reference to the candidates’ support in the electoral process.” Id. at 835, 115 S.Ct. at 1870.18 None severely limited the voters’ right to vote for the candidate of their choice.
We do not in any way minimize the state’s interest in adopting measures like Proposition 140. It cannot be doubted that the state has a particularly strong interest in its right to determine the structure of its political institutions. “Through the structure of its government, and the character of those who exercise government authority, a State defines itself as a sovereign.” Gregory v. Ashcroft, 501 U.S. 452, 460, 111 S.Ct. 2395, 2400, 115 L.Ed.2d 410 (1991); see also Dunn v. Blumstein, 405 U.S. 330, 344, 92 S.Ct. 995, 1004, 31 L.Ed.2d 274 (1972) (stating that a state enjoys inherent authority to define itself by “preserv[ing] the basic conception of a political community”); Sugarman v. Dougall, 413 U.S. 634, 648, 93 S.Ct. 2842, 2850-51, 37 L.Ed.2d 853 (1973). Indeed, the manner in which a state designs its own political structure and sets qualifications for its officials “lie[ ] at the heart of representative government.” Gregory, 501 U.S. at 463, 111 S.Ct. at 2402 (internal quotation marks and citation omitted).19
On the other hand, the Supreme Court has left no doubt that the burden created by the state’s lifetime ban is exceptionally severe. While we have no need in this opinion to balance the state’s interest in adopting lifetime term limits against the burden they impose on fundamental voting rights, determining the extent of this burden is important to our analysis.
The Supreme Court has emphatically stated that “[t]he right to vote freely for the candidate of one’s choice is of the essence of a democratic society, and any restrictions on that right strike at the heart of representative government.” Reynolds, 377 U.S. at 555, 84 S.Ct. at 1378. Because “[s]tate legis*854latures are, historically, the fountainhead of representative government in this country,” this right applies with equal force “in state as well as federal elections.” Id. at 554, 564, 84 S.Ct. at 1377-78, 1382-83. “Other rights,” the Supreme Court has explained, “even the most basic, are illusory if the right to vote is undermined.” Williams v. Rhodes, 393 U.S. 23, 31, 89 S.Ct. 5, 10, 21 L.Ed.2d 24 (1968) (quoting Wesberry v. Sanders, 376 U.S. 1, 17, 84 S.Ct. 526, 534-35, 11 L.Ed.2d 481 (1964)). Hence, Proposition 140’s lifetime ban on the people’s right to vote for legislators of then-choice does not simply handicap the rights of individuals; it violates a “fundamental principle of our representative democracy” that “the people should choose whom they please to govern them.” U.S. Term Limits, 514 U.S. at 795, 115 S.Ct. at 1851 (quoting Powell v. McCormack, 395 U.S. 486, 547, 89 S.Ct. 1944, 1977-78, 23 L.Ed.2d 491 (1969) (quoting 2 Debates on the Adoption of the Federal Constitution 257 (J. Elliott ed. 1863) (Alexander Hamilton, New York))) (internal quotation marks omitted).
As Madison pointed out at the Convention, this principle is undermined as much by limiting whom the people can select as by limiting the franchise itself.... “The people are the best judges who ought to represent them. To dictate and control them, to tell them who they shall not elect, is to abridge their natural rights.”
Powell, 395 U.S. at 547, 89 S.Ct. at 1977; id. at 541 n. 76, 89 S.Ct. at 1974 n. 76 (quoting 2 Elliot’s Debates 292-93). In Powell, the Supreme Court “agreed with the sentiment expressed on behalf of Wilkes’ admission to Parliament: ‘That the right of electors to be represented by men of their own choice, was so essential for the preservation of all their other rights, that it ought to be considered as one of the most sacred parts of our constitution.’ ” U.S. Term, Limits, 514 U.S. at 795, 115 S.Ct. at 1851 (quoting Powell, 395 U.S. at 534 n. 65, 89 S.Ct. at 1971 n. 65 (quoting 16 Pari. Hist. Eng. 589-90 (1769))).
Lifetime term limits impose a penalty on the voters that is both severe and paradoxical. Unlike consecutive term limits, which are designed to obviate any unfair advantages incumbents may have, lifetime term limits do not simply provide a hiatus from office, but, rather, impose a permanent ban. The effect of lifetime term limits, as Justice Kennedy explained in his concurring opinion in U.S. Term Limits, is “that [a state] can burden the rights of resident voters in federal elections by reason of the manner in which they earlier had exercised [those rights]. If a majority of the voters ... successfully] seleet[s] a candidate, they [are] penalized from exercising that same right in the future.” 514 U.S. at 844, 115 S.Ct. at 1874 (Kennedy, J., concurring). The logic of this statement is, of course, equally applicable to state elections; the penalty imposed on the voters is identical.
Despite the strength of the Supreme Court’s statements, the state contends that term limits do not impose a severe burden on individual rights, in part because they neither “discriminate” against a particular political group or viewpoint nor constitute a “content-based restriction.” In addition, the state emphasizes that voters may still vote for candidates with “similar views and ability” to those incumbents barred by term limits and that individuals barred from the legislature by term limits remain eligible to hold state executive office, elective or appointed, or to seek federal office. The burden Proposition 140 imposes on the rights of voters and candidates is truly severe for the reasons already explained. We need not consider whether, in addition, the measure can be described as “discriminatory” or “content-based.” That there may be other candidates whose views the voters may find palatable, and that “termed out” candidates are free to seek non-legislative offices simply does not address the constitutional injury inflicted: the Proposition permanently deprives the voters of the right to vote for the person they wish to represent them in the state legislature}20
*855The state offers two additional arguments in mitigation of the burden imposed by lifetime term limits — namely, that the burden is lessened by the possibility of future repeal, and that term limits reduce existing burdens on voting rights. These arguments are without merit. First, the ability of a majority of voters to repeal term limits in no way mitigates the burden that the measure imposes while it is in effect. Were laws to be held not to limit fundamental rights because they can be repealed, the Constitution would become a meaningless document. Every law, no matter how violative of individual rights, would pass the state’s test. That the law was passed by initiative makes it even harder to rid society of the measure because of the extreme financial costs involved in putting-initiatives on the ballot. See infra note 29. Second, the state’s argument that term limits reduce existing burdens on voting rights by enlarging the field of eligible candidates is pure speculation. More important, the assertion that Proposition 140 constitutes an effort to enhance a fundamental right by infringing on it is a woefully misguided argument that the Court has emphatically rejected in the past. See Buckley v. Valeo, 424 U.S. 1, 48-50, 96 S.Ct. 612, 649, 46 L.Ed.2d 659 (1976) (per curiam) (stressing that “the concept that government may restrict the speech of some ... to enhance the relative voice of others is wholly foreign to the First Amendment”).
III. Proposition 140’s Enactment and Its Lack of Notice
In this case of first impression, we are faced with directly conflicting rights and interests of the upmost importance. The state has attempted to resolve this historic conflict by means of a direct ballot initiative that would severely limit the right of the people to elect whomever they choose. Whether or not a state may lawfully adopt lifetime legislative term limits as a general matter, the techniques it employed here raise a serious issue of constitutional dimension: whether the voters are entitled to adequate notice that the measure, if enacted, would severely limit their fundamental rights. Proposition 140 on its face contained no reference to lifetime limits, and the ballot arguments submitted by the initiative’s proponents failed to mention that the measure contemplated such a ban; so, too, the materials prepared by the state were wholly silent on the point. The initiative barely passed with an affirmative vote of 52%; a switch of approximately 2% of the votes would have resulted in its defeat.
We are faced, in short, with two constitutional issues in this case: one procedural, one substantive. Given the magnitude of the substantive constitutional question presented by lifetime legislative term limits and the narrowness of the procedural question regarding the need for fair notice in this type of initiative measure, it seems prudent — as well as logical — to decide the procedural question first. Such an approach is consistent with Justice Powell’s admonition in Communist Party of Ind. v. Whitcomb, 414 U.S. 441, 452, 94 S.Ct. 656, 663, 38 L.Ed.2d 635 (1974), that “the appropriate exercise of judicial power requires that important constitutional issues not be decided unnecessarily where narrower grounds exist for according relief.” Id. at 452, 94 S.Ct. at 663 (Powell, J., concurring) (opting, joined by three other Justices, to rest concurring vote on narrow equal protection ground instead of broad First Amendment freedom-of-speech ground); see also Thompson v. Oklahoma, 487 U.S. 815, 858, 108 S.Ct. 2687, 2711, 101 L.Ed.2d 702 (1988) (O’Connor, J. concurring) (reiterating long-standing doctrine that federal courts “avoid unnecessary, or unnecessarily broad, constitutional adjudication”); City of Mesquite v. Aladdin’s Castle, 455 U.S. 283, 297, 102 S.Ct. 1070, 1078-79, 71 L.Ed.2d 152 (1982) (White, J., concurring in part and dissenting in part) (espousing adherence to principle that, when presented with two separate constitutional questions, Court decide narrow constitutional question). That approach, as Justice O’Connor stated, finds its genesis in Justice Brandéis’ eoncur*856rence in Ashwander v. TVA, 297 U.S. 288, 341-56, 56 S.Ct. 466, 480-87, 80 L.Ed. 688 (1936) (Brandéis, J., concurring). Its basic teaching, which we follow today, is the need to abstain from deciding fundamental constitutional questions prematurely, and perhaps unnecessarily.
A careful analysis of Proposition 140 and the California Supreme Court decision construing it reveals that the Proposition and the relevant ballot materials failed adequately to advise the voters of its principal and most drastic effect. The people were required to vote on an initiative measure without adequate notice of the severity of the burden it imposed on their fundamental voting rights — a measure that did not clearly inform them that they were being asked to adopt a lifetime ban on future legislative service by the persons they choose to be their representatives. Moreover, it is likely that the number of voters who were not aware of the initiative’s principal effect was sufficiently large that it could have affected the outcome of the election. We conclude that under these circumstances, even if a majority of the voters of the state of California may, under other circumstances, lawfully decide to adopt lifetime legislative term limits, the passage of Proposition 140 did not constitutionally achieve that objective.
The California Supreme Court expressly determined that “the language of Proposition 140 is ambiguous as to its intent to impose a lifetime ban.” Legislature of the State of California v. Eu, 54 Cal.3d 492, 286 Cal.Rptr. 283, 289, 816 P.2d 1309, 1315 (1991), cert. denied sub nom. Californians for Citizen Gov’t v. Legislature of the State of Cal., 503 U.S. 919, 112 S.Ct. 1293, 117 L.Ed.2d 516 (1992). In this respect, it accepted the argument advanced by the Attorney General of California on behalf of the Secretary of State. We agree and consider the California court’s conclusion authoritative. See, e.g., Johnson v. Fankell, — U.S. -, -, 117 S.Ct. 1800, 1804, 138 L.Ed.2d 108 (1997) (state supreme court constructions of state laws are binding on federal courts); Easyriders Freedom F.I.G.H.T. v. Hannigan, 92 F.3d 1486, 1494 n. 4 (9th Cir.1996) (requiring deference to state supreme court constructions of state laws). The preamble of Proposition 140 focuses almost exclusively on curbing unfair “incumbent” advantages, an objective that underlies the imposition of both consecutive and lifetime term limits. The Proposition goes on to state that “[n]o Senator” and “[n]o member of the Assembly” may serve more than two or three terms respectively. Cal. Const., art. IY, § 2, subd. (a).21 The state Attorney General contended that this language applies only to senators and members of the assembly and not to former officeholders. Read in this manner, the measure would impose consecutive term limits, not a lifetime ban. The phrase “lifetime limits,” which the state uses so forcefully now, appears nowhere in the text of the amendment, the proponents’ ballot arguments, or the official statements prepared by the state. As the California Supreme Court recognized, therefore, the Proposition could easily be read as applying only to incumbent office holders. Eu, 286 Cal.Rptr. at 288-89, 816 P.2d at 1314-15. Nonetheless, the court, after conceding that opponents tend to overstate the effects of measures in their ballot arguments, relied primarily on the opponents’ ballot arguments in reaching the equivocal conclusion that, despite the initiative’s facial ambiguity, “We think it likely [that] the average voter ... would conclude the measure contemplated a lifetime ban.” Eu, 286 Cal.Rptr. at 290, 816 P.2d at 1316 (emphasis added). As we read the court’s opinion, that is by no means a finding (or a conclusion) that all of the voters who voted in favor of the measure so understood it, or even that a sufficient number of those voting “yes” to constitute a majority did so. Given the 52% affirmative vote, the court’s statement falls far short of supporting any such inference. In fact, it appears *857to us to acknowledge implicitly that a number of voters who do not fall within the classification “average”22 are likely to have suffered the inevitable consequence of a lack of adequate notice — i.e., that a fair number may (because of the lack of such notice) have failed to understand that a lifetime ban was contemplated.23
Although the proper interpretation of Proposition 140 is purely a matter of state law, the issue whether the people’s fundamental political rights may be severely burdened by means of an “ambiguous” initiative that “the average voter” would only “likely” understand is, contrary to the state’s assertion in its supplemental brief, a question of federal law. Under the California rule for construing initiatives, which requires courts to indulge “all presumptions [in] favor [of] the validity of initiative measures,” id. at 501, 286 Cal.Rptr. 283, 816 P.2d 1309, it is enough for purposes of interpreting the meaning of the measure that the “average” voter “likely” understood the measure. But the federal standard for adequate notice when fundamental rights are at stake requires a different and more stringent inquiry. Thus, while we accept the California Supreme Court’s determination that the measure contains a lifetime ban, how and in what manner a state may limit rights guaranteed by the Constitution is a quintessential federal law issue. Here, that issue directly implicates due process concerns.
It is well established that when a restriction “severely” limits fundamental voting rights, courts must assess the state’s interests in doing so with great care. See Burdick, 504 U.S. at 434, 112 S.Ct. at 2063-64; Norman v. Reed, 502 U.S. 279, 289, 112 S.Ct. 698, 705-06, 116 L.Ed.2d 711 (1992); Reynolds, 377 U.S. at 562, 84 S.Ct. at 1381—82. Our exercise, although derived from the principles of Burdick, is one step removed. Before even imposing such burdens on fundamental rights, we must be assured that the process permits the imposition.
We believe that Burdick, along with the Supreme Court doctrine discussed below, requires that we carefully scrutinize the process by which initiatives, such as Proposition 140 — initiatives that have the effect of severely limiting fundamental voting rights— are enacted. We must do so in order to ensure that sufficient notice has been given that fundamental rights will be so burdened. Simply put, we believe that the careful scrutiny required by Burdick must apply to the process as well as the effect of such laws. Indeed, the Supreme Court has, on a number of occasions, refused to permit the imposition of severe limitations on fundamental rights in the absence of adequate notice and a carefully considered decision to do so. Years ago, for instance, the Court explained that decisions to restrain “long-accepted notions of fair procedures ... must be made explicitly not only to assure that individuals are not deprived of cherished rights and procedures not actually authorized, but also because explicit action, especially in areas of doubtful constitutionality, requires careful and purposeful consideration by those responsible for enacting ... our laws.” Greene v. McElroy, 360 U.S. 474, 507, 79 S.Ct. 1400, 1419, 3 L.Ed.2d 1377 (1959) (internal citations omitted); see also United States v. Bass, 404 U.S. 336, 349, 92 S.Ct. 515, 523, 30 L.Ed.2d 488 (1971) (“In traditionally sensitive areas ... the requirement of clear statement assures that the [enacting body] has in fact faced, and intended to bring into issue, the critical matters involved in the judicial decision.”).
While this principle has most often been employed in order to construe narrowly stat*858utes of questionable constitutionality,24 the Court has, on occasion, invoked it to strike down laws when a narrow construction was unavailable. In Hampton v. Mow Sun Wong, 426 U.S. 88, 96 S.Ct. 1895, 48 L.Ed.2d 495 (1976), for example, federal regulations deprived aliens of the right to work in federal service positions. In support of the regulations, the federal government asserted an “overriding” national interest for its discriminatory rule. Recognizing that if it accepted the government’s interest it could not reasonably apply any balancing test, the Court struck down the regulations because nothing demonstrated that the parties responsible for enacting the regulations had given the matter “a considered evaluation.” 426 U.S. at 115, 96 S.Ct. at 1911. Even assuming that the government’s overriding interest “would adequately support an explicit determination by Congress or the President to exclude all noncitizens from public service,” the Court held that due process required more thorough and definitive action by the entities responsible for enacting the' drastic regulation. 426 U.S. at 116, 96 S.Ct. at 1911-12.25
Perhaps the case that is most comparable to the one before us is Thompson v. Oklahoma, 487 U.S. 815, 857, 108 S.Ct. 2687, 2710-11, 101 L.Ed.2d 702 (1988). In that case, the Supreme Court faced the question whether states could constitutionally impose the death penalty on children younger than 16. Because Oklahoma law, like the law of 19 other states, sometimes channeled 15-year-olds into the adult criminal justice system, it allowed Thompson to be executed. An Oklahoma jury sentenced Thompson to death, and Oklahoma’s Court of Criminal Appeals affirmed both the constitutionality of its law and the sentence. 487 U.S. at 820, 108 S.Ct. at 2690-91. Presented with such a powerful interpretation of state law, the plurality and the dissent attempted to discern whether states like Oklahoma actually intended their laws to apply this way and disagreed sharply over whether there was a national consensus forbidding such executions. Justice O’Connor, however, concluded that it was unnecessary to take a position on the issue that divided the Court because of the basic flaw inherent in the Oklahoma law. While not expressly mentioning due process, she nonetheless cast the deciding vote to hold the law unenforceable because “there [was] a considerable risk that the Oklahoma Legislature either did not realize that its actions would have the effect of rendering 15-year-old defendants death eligible or did not give the question the serious consideration that would have been reflected in the explicit choice of some minimum age for death eligibility.” Id. at 857, 108 S.Ct. at 2711 (O’Con-nor, J., concurring). Because those voting for the law severely burdened the fundamen*859tal constitutional right to be free from cruel and unusual punishment “without the earmarks of careful consideration that [the Court] has required for other decisions” in this fundamental, constitutional area, Justice O’Connor cast the deciding vote not to enforce the Oklahoma law. Id.26
Justice O’Connor repeatedly emphasized that Thompson was a death penalty case, and that death penalty cases are “different.” She also stressed that Thompson was an “unusual,” even “unique,” case. Id. at 858, 108 S.Ct. at 2711 (O’Connor, J., concurring).27 The case before us, however, is equally unusual and, in many ways, as compelling. First, as we have previously emphasized, the right at issue here is also fundamental. The passage of the measure directly affects every voter in California in the most fundamental way; in our representative democracy, the right to elect whomever one chooses is the most central of all rights. See Reynolds, 377 U.S. at 562, 84 S.Ct. at 1381-82 (explaining that “the right to exercise the franchise in a free and unimpaired manner is preservative of other basic civil and political rights”); supra at 853-54. Second, the absence of due process is even more apparent in the California initiative because it was explicitly found “ambiguous” on its face, not just possibly unintended. The measure is not one that might have required the voters simply to predict certain potential applications or fill in gaps in the initiative; they had to guess as to the very meaning and effect of the measure’s principal provision. The ambiguity lies at the heart of the language of the measure, at the heart of the provision that implements its principal objective — a provision that has since been construed as imposing a lifetime ban, a severe limitation on the people’s most fundamental right.
Cases involving the adoption of state laws or constitutional amendments by means of the initiative process raise concerns that are not present when ordinary legislative lawmaking is involved. Though the initiative process was virtually unknown to the Framers, the Federalist papers — and, to a lesser extent, the Constitution — reflect a wariness of the products of simple majoritarianism. In Federalist No. 10, Madison stated that majority factions ought to be contained because they are apt to be willing to sacrifice the “public good and the rights of citizens” to their “ruling passion or interest.” Federalist No. 10, at 60-61 (James Madison) (J. Cooke ed.1961). He was even more direct in Federalist No. 51: “If the majority be united by a common interest, the rights of the minority will be insecure.” Id. No. 51, at 351 (James Madison). Of particular relevance to the case at hand, Madison continued, “There are particular moments in public affairs when the people stimulated by some irregular passion, or some illicit advantage, or misled by the artful representations of interested men, may call for measures which they themselves will afterwards be the most ready to lament.” Id. No. 63, at 425 (James Madison) (emphasis added). It is government’s duty under our constitutional design, therefore, to “safeguard against the tyranny of [such] passions.” Id. We agree with Hamilton that the need for judicial independence is the greatest when constitutional impairments are “instigated by the major voice of the community.” Id. No. 78, at 528 (Alexander Hamilton).
Although it has long been settled that, notwithstanding these concerns, states may pass laws by initiative, see Pacific States Tel. & Tel. Co. v. Oregon, 223 U.S. 118, 32 S.Ct. 224, 56 L.Ed. 377 (1912), direct ballot measures lack the kinds of critical, deliberative filters that the Framers contemplated and that, to some extent, the Constitution created as prerequisites to the passing of legisla*860tion.28 Before an initiative becomes law, no committee meetings are held; no legislative analysts study the law; no floor debates occur; no separate representative bodies vote on the bill; no reconciliation conferences are held; no amendments are drafted; no executive official wields a veto power and reviews the law under that authority; and it is far more difficult for the people to “reconvene” to amend or clarify the law if a court interprets it contrary to the voters’ intent.29 The public also generally lacks legal or legislative expertise — or even a duty (as legislators have under Article VI) to support the Constitution. It lacks the ability to collect and to study information that is utilized routinely by legislative bodies.30 Thus, at least in instances like this, in which the measure raises serious constitutional questions, our usual assumption that laws passed represent careful drafting and consideration does not obtain.31
Moreover, the search for the people’s intent in passing initiatives is far different from the attempt to discern legislative intent; there are no legislative hearing transcripts, committee reports, or other legislative history.32 See Jane S. Schachter, The Pursuit of “Popular Intent’’: Interpretive Dilemmas in Direct Democracy, 105 Yale L.J. 107 (1995). Proposition 140 suffers from precisely this problem. There is nothing, other than the facially ambiguous initiative, the official ballot arguments and the state-prepared materials, to look to in order to discern the people’s intent in passing the measure. And Proposition 140 simply presents us with no evidence, in its language, in the proponents’ ballot argument, or in the state-prepared material that the people had adequate notice of the measure’s principal objective and most drastic effect: the severe burden, in the form of lifetime term limits, on fundamental voting rights. While the California court relied primarily on the opponents’ ballot arguments and secondarily on the Legislative Analyst’s report describing the measure, those writings fall far short of what is necessary to meet the due process requirements of fair notice to the voters and a concomitant showing of careful consideration of the burden to be imposed. The Legislative Analyst’s report is almost as ambiguous as the initiative itself, and the opponents’ arguments are entitled to little weight even under California law. See Eu, 54 Cal.3d at 505, 286 Cal.Rptr. 283, 816 P.2d 1309 (stating that an initiative’s opponents “frequently overstate the adverse *861effects of [ ] challenged measure[s],” and, therefore, that their claims “are not highly authoritative in construing the measure”);33 cf. DeBartolo Corp. v. Fla. Gulf Coast Trades Council, 485 U.S. 568, 585, 108 S.Ct. 1392, 1402-03, 99 L.Ed.2d 645 (1988) (“The views of opponents of a bill with respect to its meaning ... are not persuasive.”); NLRB v. Fruit Packers, 377 U.S. 58, 66, 84 S.Ct. 1063, 1068, 12 L.Ed.2d 129 (1964) (“In then-zeal to defeat a bill, [opponents] understandably tend to overstate its reach. ‘The fears and doubts of the opposition are no authoritative guide to the construction of legislation. It is the sponsors that we look to when the meaning of the statutory words is in doubt.’ ”) (quoting Schwegmann Bros. v. Calvert Distillers Corp., 341 U.S. 384, 394-95, 71 S.Ct. 745, 750-51, 95 L.Ed. 1035 (1951)). Thus there is insufficient indication that the voters had adequate notice that the measure would burden so severely the people’s fundamental constitutional right to choose then-legislative representatives.34
Furthermore, when fundamental, constitutional rights, such as voting rights, are at stake,35 basic tenets of procedural due process require states to adopt minimally burdensome safeguards if there is a significant risk of an erroneous deprivation of such a right. See, e.g., Mathews v. Eldridge, 424 U.S. 319, 334-35, 96 S.Ct. 893, 902-03, 47 L.Ed.2d 18 (1976) (establishing balancing test with respect to those factors). The Supreme Court stated in Burdick that “[c]ommon sense, as well as constitutional law, compels the conclusion that [state] government must play an active role in structuring elections; ‘as a practical matter, there must be a substantial regulation of elections if they are to be fair and honest and if some sort of order, rather than chaos, is to accompany the democratic processes.’ ” 504 U.S. at 433, 112 S.Ct. at 2063 (quoting Storer v. Brown, 415 U.S. 724, 730, 94 S.Ct. 1274, 1279, 39 L.Ed.2d 714 (1974)) (emphasis added). Thus, the state of California has a constitutional obligation to ensure the fairness of its elections, including those involving the initiative process. California, indeed, has adopted several statutory provisions designed to ensure that voters will be afforded fair notice of the contents of ballot measures. See, e.g., Cal. Elec.Code §§ 9051-52 (investing Attorney General with authority to provide a ballot title for initiative measures and to “give a true and impartial statement of the purpose *862of the measure”); id. § 9085 (requiring Legislative Analyst to provide in the ballot pamphlet a summary of the measure); id. § 9087 (requiring Legislative Analyst to “generally set forth in an impartial manner the information the average voter needs to adequately understand the measure”).36 In this case, the state could have ensured with only a minimal effort that adequate notice was afforded the voters of the severe burden imposed by the measure. In fact, it appears that all that was necessary to provide adequate notice to the voters that their fundamental voting rights would be so burdened was compliance with the state’s own statutes. At no point in the process, however, did either the Attorney General or the Legislative Analyst include the word “lifetime” or “for life,” or even a word such as “permanent,” “total,” “prohibition,” or “ban” in any of the ballot materials that it was their duty to prepare. To the contrary, they perpetuated at every turn the ambiguity created by the measure’s proponents. Not only would performing their duties in the manner contemplated by state law have imposed no burden on the state officials, but it would have afforded the voters the notice of the measure’s principal effect that due process of law requires.
Similarly, if the proponents of Proposition 140 truly intended to put to the voters the proposal of lifetime term limits, they could have simply put that phrase into the initiative. The phrase is perfectly clear and existed long before the election. This safeguard would have imposed no burden at all on the measure’s sponsors, yet would have ensured that voters would have received adequate notice of the initiative’s sweeping effect. Cf. Mathews, 424 U.S. at 334-35, 96 S.Ct. at 902-03.
Counsel for the state, Professor Elhauge, conceded at oral argument that the voters must be clearly informed of the initiative’s effect, and, specifically, that the voters should be advised that “they are adopting term limits for life.”37 We agree. In the state’s supplemental brief, however, Professor Elhauge argues that it is immaterial whether the voters understood that Proposition 140 imposed lifetime limits. That is so only in the narrowest sense. If the voters were not adequately informed of the lifetime ban, voter ignorance because of lack of notice is highly material. The state supreme court, the state now urges, “is a lawmaker” for federal constitutional purposes. Supp. Br. at 8. According to the state, we may invalidate the initiative measure only if its effect, as subsequently construed by the California Supreme Court, violates the Constitution, and not because the measure itself as put to the voters concealed its true objective and precluded a fair expression of the voter’s will.
We of course acknowledge that the interpretation of a state statute by the state’s highest court “puts ... words into the statute as definitely as if it had been so amended by the legislature.” Winters v. New York, 333 U.S. 507, 514, 68 S.Ct. 665, 669-70, 92 L.Ed. 840 (1948); see also, e.g., Poulos v. New Hampshire, 345 U.S. 395, 402, 73 S.Ct. 760, 764-65, 97 L.Ed. 1105 (1953). This line of authority, however, is essentially inapplicable. The issue we face and answer here is not how to construe a state statute that has already been construed by the state supreme court, but whether a state may impose severe burdens on people’s most fundamental consti*863tutional rights through an initiative measure that is ambiguous when presented to the voters and that fails adequately to inform the people at the time they cast their ballots that they are being asked to limit their most fundamental rights. It is the measure as presented to the people, and the extent to which that measure informs the voters of the burden that it would impose on their fundamental rights, with which we are concerned—not the “meaning” of the measure as it is subsequently construed or limited by the state supreme court. The meaning afforded a state law by a state’s highest court determines the meaning we give the law; it does not resolve the question whether the measure sufficiently advised the voters of the limitations to be imposed. Given that federal courts are charged with protecting the people’s fundamental constitutional rights against undue intrusion by the states, the latter question surely is of federal concern.
Due process and the Supreme Court’s established practice of requiring considered decision-making in highly sensitive areas of constitutional law compel our holding. Although we fully accept the California Supreme Court’s construction of Proposition 140, the Constitution requires us to invalidate an initiative if it fails to provide adequate notice to the voters that it would severely burden the people’s fundamental rights. Here, we hold as a matter of federal law that the state may not enforce Proposition 140’s lifetime legislative term limits because the provision imposes a severe limitation on the people’s fundamental right to elect whomever they choose and the voters were not provided with adequate notice of that limitation.38
Although we recognize and appreciate that states have an important constitutional interest in designing their own political institutions, we conclude that when a state uses an initiative to adopt a form of governmental structure that severely burdens the fundamental rights of voting and association, due process requires that the state give the voters notice of the limitation that the measure would impose. Absent adequate notice, we cannot hold that the people intended severely to burden their most fundamental right, the right to vote. In matters this important, the state simply must tell its citizens what they are voting on.
IV. Conclusion
This conclusion leaves us no choice but to invalidate the legislative-term-limits provision of Proposition 140. As we stated above, and as the state correctly reminds us, the California Supreme Court’s determination that the measure imposes a lifetime ban, Eu, 286 Cal.Rptr. at 290, 816 P.2d at 1316, prevents us from narrowly construing the initiative, potentially to save its constitutionality; “[w]e must take the statute as though it read precisely as the highest court of the State has interpreted it.” Minnesota ex rel. Pearson v. Probate Court, 309 U.S. 270, 273, 60 S.Ct. 523, 525, 84 L.Ed. 744 (1940); accord Easyriders Freedom F.I.G.H.T., 92 F.3d at 1494 n. 4.
Nor may we sever the lifetime limits requirement. Severability is a question of state law. Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 506, 105 S.Ct. 2794, *8642803, 86 L.Ed.2d 394 (1985). California law prescribes three criteria for severability: (1) the invalid provision must be mechanically and grammatically separable; (2) it must be functionally separable; and (3) the remainder of the law would have been adopted by the legislative body if it had foreseen the partial invalidity of the law. Calfarm Ins. Co. v. Deukmejian, 48 Cal.3d 805, 258 Cal.Rptr. 161, 771 P.2d 1247 (1989). Putting aside the difficulties discussed, supra at 860, surrounding any attempt to divine the “intent” of the people, it is apparent that the first criterion is not met here. There is no lifetime limits “provision” that is “mechanically and grammatically separable.” Id.
Hence, we have no choice but to invalidate the provision for legislative term limits. In doing so, however, we reiterate that we express no view as to the right of the state to adopt lifetime term limits for legislative office. That is an important and difficult constitutional question, but it is one that we defer to another day.
The judgment of the district court is
AFFIRMED.
. Although the measure contains a number of other provisions, it is obvious that the measure's primary focus is the legislative-term-limits provision. The term-limits provision is the first item mentioned in the measure’s title, the measure itself, and in the ballot arguments. In fact, the measure's proponents dedicated seven and one-half of its ten-paragraph ballot argument to debating the merits of legislative term limits (without mentioning that the limits to be imposed by the measure were lifetime). Moreover, as the proponents emphasized, legislators constitute 120 of the 132 elected officials affected by the term limits.
. Excluding California, 14 of the 20 states that have term limits employ consecutive term limits. See Bates v. Jones, 958 F.Supp. 1446, 1454 n. 5 (N.D.Cal.1997) (collecting state laws).
. Ballot arguments are arguments written by the proponents and opponents of an initiative, advocating the pros and cons of the measure. See id. §§ 9041-42. They cannot exceed 500 words. Id.
. By contrast, the state of Oregon's constitutional provision imposing a lifetime ban reads: "No person shall serve more than six years in the Oregon House of Representatives, eight years in the Oregon Senate, and twelve years in the Oregon Legislative Assembly in his or her lifetime." Ore. Const., art. II, § 19, cl. 1 (emphasis added).
.The remaining ballot language described the predicted financial impact of the measure, as required by California law. The actual text of the measure appeared nowhere on the ballot.
. For simplicity, we refer throughout this opinion to the respondent in the state court proceeding as the slate of California, or "the state." The actual respondent was the Secretary of State, March Fong Eu, whose duty it is to implement the principal provisions of the initiative. The respondent was represented by the state Attorney General, Daniel E. Lungren.
. The court also relied secondarily on the fact that, in his analysis, the Legislative Analyst described Proposition 140 as limiting " 'the number of terms that an elected state official can serve in the same office.' ” Id. at 505, 286 Cal.Rptr. 283, 816 P.2d 1309 (emphasis and citation omitted). Although this statement used the term "elected state official" instead of “person” and omitted any mention of a “lifetime” or "permanent” ban, and thus contained the same ambiguity as the Proposition itself, the court deemed it relevant that the Legislative Analyst "made [no suggestion] that only a consecutive term limitation was contemplated.” Id.
. The district court also granted Bates leave to amend his complaint to add as plaintiffs Assembly members Martha Escutia and Barbara Friedman, as well as voters from their respective districts. The voter-plaintiffs, Hernandez, Navar, Pena, Sculley, and Zarakov, voted for Escutia and Friedman in the past and wish to be able to do so in the future. Neither Escutia nor Friedman was a member of the Assembly, however, at the time of the Eu litigation. Escutia was first elected in 1992, and Friedman joined the Assembly following a special election in mid-1991.
The state argues that the district court abused its discretion in granting the original plaintiffs leave to join these additional plaintiffs. We disagree. Federal Rule of Civil Procedure 15(a) directs a court freely to grant a party leave to amend its complaint "when justice so requires” and "should be applied with 'extreme liberality.’ ” DCD Programs, Ltd. v. Leighton, 833 F.2d 183, 186 (9th Cir.1987) (internal citation and quotation marks omitted). Delay alone is insufficient to justify denial of leave to amend; the party opposing amendment must also show that the amendment sought is futile, in bad faith.or will cause undue prejudice to the opposing party. Id. at 186-87; Webb, 655 F.2d at 980. First, based upon the evidence, the district court found that neither Bates nor his fellow plaintiffs had acted in bad faith. This finding is not clearly erroneous. See Sorosky v. Burroughs Corp., 826 F.2d 794, 805 (9th Cir.1987). Plaintiffs did not attempt to prolong meritless litigation by making repeated and baseless amendments to their complaint; nor is there any evidence of wrongful motive. To the contrary, the district court found that Bates moved to join the additional plaintiffs promptly after discovering his unwitting financial contribution to the Eu litigation. The motion to amend was not untimely, repetitious, or for the purpose of keeping meritless claims in court, but to ensure that the claims already presented would be decided on their merits, in furtherance of the very purpose of Rule 15. See Foman v. Davis, 371 U.S. 178, 182, 83 S.Ct. 227, 230, 9 L.Ed.2d 222 (1962); Breier v. Northern Cal. Bowling Proprietors’ Ass’n, 316 F.2d 787, 789 (9th Cir.1963). Second, we reject the state’s argument that the district court should have denied the motion to amend because Escutia and Friedman lacked standing. There can no longer be any question as to whether Escutia and Friedman have standing. Escutia was re-elected to a third term, and the district court properly allowed Friedman to remedy the defect in her pleadings by alleging a specific intent to return to the Assembly. Even viewing the question of standing through the eyes of the district court at the time it granted leave to amend, however, Escutia and Friedman did not lack standing. The district court correctly concluded that Escutia and Friedman demonstrated an adequate likelihood of future injury for justiciability purposes merely by alleging their desire to run in a particular election for a fourth term in the Assembly. See Thorsted v. Munro, 75 F.3d 454, 456 (9th Cir.1996) (holding that congressman’s challenges to federal term limits were not rendered moot by his defeat in an intervening election, in light of his desire to seek re-election in the future).
. We dispose of the appeal from the stay by separate order.
. Compare, e.g., Montana v. United States, 440 U.S. 147, 154, 99 S.Ct. 970, 974, 59 L.Ed.2d 210 *850(1979) (though not formally a party to prior state court litigation, government was estopped from relitigating in federal court where it required the lawsuit to be filed; reviewed and approved the complaint; paid the attorneys' fees and costs; directed a state court appeal; appeared and submitted an amicus brief in the state supreme court; directed the filing of a notice of appeal to the Supreme Court; and directed the plaintiff to abandon that appeal).
. See discussion of U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) and Burdick v. Takushi, 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992), infra at 852.
. The state also contends that the voter-plaintiffs from Bates' former district lack standing because they have alleged no injury that would be redressed by a decision invalidating the term limit provisions of Proposition 140. It suggests that because Bates is barred from running not by Proposition 140, but by the res judicata effect of Eli, a judicial ruling in favor of the voter-plaintiffs would not redress their injury. The argument is utterly without merit. Bates is not in fact barred by res judicata, and even if he were, it cannot be questioned that invalidation of Proposition 140's term limit provisions would enable him to run again and thereby redress the injury alleged by the voter-plaintiffs. It is Proposition 140, not the California Supreme Court’s interpretation and application of Proposition 140 in Eu, that prevents Bales from seeking re-election.
. The state contends that the constitutionality of Proposition 140 is controlled by Moore v. McCartney, 425 U.S. 946, 96 S.Ct. 1689, 48 L.Ed.2d 190 (1976) (mem.), in which the Supreme Court summarily dismissed an appeal from State ex rel. Maloney v. McCartney, 159 W.Va. 513, 223 S.E.2d 607 (1976), rejecting an incumbent-governor's federal constitutional challenge to lifetime term limits for state executive officials for want of a substantial federal question. A summary dismissal by the Supreme Court for want of a substantial federal question is a decision on the merits, but is controlling only on "the precise issues presented and necessarily decided.” Mandel v. Bradley, 432 U.S. 173, 176, 97 S.Ct. 2238, 2240, 53 L.Ed.2d 199 (1977); see also Illinois Elections Bd. v. Socialist Workers Party, 440 U.S. 173, 183, 99 S.Ct. 983, 989-90, 59 L.Ed.2d 230 (1979); Mandel, 432 U.S. at 180, 97 S.Ct. at 2242-43 (Brennan, J., concurring). As a threshold matter, even if we were to assume that Moore concerned precisely the same issues involved here, extensive intervening doctrinal developments — including Burdick v. Takushi, 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992), Anderson v. Celebrezze, 460 U.S. 780, 788, 103 S.Ct. 1564, 1569-70, 75 L.Ed.2d 547 (1983), and U.S. Term Limits v. Thornton, 514 U.S. 779, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) — strongly suggest that continued reliance on Moore is unwarranted. See Hicks v. Miranda, 422 U.S. 332, 344, 95 S.Ct. 2281, 2289, 45 L.Ed.2d 223 (1975) (reliance on the fact that the Supreme Court has branded a particular question unsubstantial ceases to be appropriate "when doctrinal developments indicate otherwise”). More important, it is unclear whether the Supreme Court had jurisdiction in Moore to consider the rights of voters, as opposed to those of candidates. No voters were parties either to the state court action or to the subsequent federal appeal. Maloney, 223 S.E.2d at 610-13. There is also, of course, the question whether executive and legislative term limits present the identical issue. Finally, Moore is not controlling here because the plaintiff in Moore, unlike plaintiffs in the present case, did not raise any First Amendment challenge to term limits. The West Virginia Supreme Court considered only the argument that lifetime term limits for state executive officials violate equal protection. Maloney, 223 S.E.2d at 610-13.
.The only circuit court decision that even considers the general subject of state term limits, League of Women Voters v. Diamond, 82 F.3d 546 (1st Cir.1996), involves the constitutionality of consecutive term limits, and does not address the merits of the issue owing to the procedural posture of the case on appeal. The district court had denied relief on the ground that the statute was "likely” to survive constitutional scrutiny, but had done so expressly "[wjithout determining th[e] issue finally on its merits.” League of Women Voters v. Diamond, 923 F.Supp. 266, 272 (D.Me.1996). In a one-page, per curiam opinion, the First Circuit affirmed "substantially for the reasons recited by the district court,” agreeing both that the plaintiffs had "established something less than a probability of success on the merits” and that the equities favored denial of preliminary injunctive relief as well. Diamond, 82 F.3d at 547.
The state and the Schabarum inlervenors nevertheless cite a number of state and federal district court decisions for the proposition that the constitutionality of term limits for state officials is well established. The decisions establish no such proposition. Some of these decisions concerned executive or judicial officials, rather than legislators, while others addressed only the constitutionality of consecutive term limits. See Dutmer v. City of San Antonio, 937 F.Supp. 587, 591-95 (W.D.Tex.1996) (upholding lifetime term limits for city officials); Nevada Judges Ass’n v. Lau, 112 Nev. 51, 910 P.2d 898, 901-03 (1996) (upholding lifetime term limits for judges); Roth v. Cuevas, 158 Misc.2d 238, 603 N.Y.S.2d 962 (Sup.Ct.1993) (upholding consecutive term limits for city officials), aff'd, 197 A.D.2d 369, 603 N.Y.S.2d 736 (App.Div.), aff'd, 82 N.Y.2d 791, 604 N.Y.S.2d 551, 624 N.E.2d 689 (1993); State ex rel. Maloney v. McCartney, 159 W.Va. 513, 223 S.E.2d 607, 611-13 (1976) (upholding lifetime gubernatorial term limits), discussed infra. In fact, only one state supreme court decision has upheld lifetime term limits for legislators. See U.S. Term Limits, Inc. v. Hill, 316 Ark. 251, 872 S.W.2d 349, 359-60 (1994) (upholding lifetime term limits for legislative and executive officials), aff d in other respects sub nom. U.S. Term Limits *852v. Thornton, 514 U.S. 779, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) Most recently, the Supreme Judicial Court of Massachusetts struck down a law imposing consecutive term limits on state executive and legislative officials, but did so on state law grounds. League of Women Voters v. Secretary of the Commonwealth, 425 Mass. 424, 681 N.E.2d 842 (1997).
. The Supreme Court has dealt with cases involving similar state interests in the equal protection context. The state’s interest in designing its own legislative structure was essentially the interest involved in Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964) and Lucas v. Forty-Fourth Gen. Assembly of Colo., 377 U.S. 713, 84 S.Ct. 1459, 12 L.Ed.2d 632 (1964). In Lucas, the Court invalidated an initiative passed by the citizens of Colorado which apportioned the House of Representatives on the basis of population, but apportioned the Senate according to population and other factors, including geographical, historical, topographic, and economic considerations, 377 U.S. at 738, 84 S.Ct. at 1474-75. Reynolds involved the same basic issue, except the state law there was a decades-old act passed by the state legislature. In deciding the cases, the Court gave little deference to the states' interest in defining its political institutions as it saw fit, concluding that the full and unfettered right to vote in state elections is of the upmost importance. It then invalidated the state laws under the Equal Protection Clause on the ground that one person's vote could not count more than another’s.
. The Court cited as examples of such regulatory laws Burdick v. Takushi, 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992) (upholding Hawaii statute requiring state legislative candidates to comply with state procedures in order to qualify for ballot and prohibiting write-in voting); Anderson v. Celebrezze, 460 U.S. 780, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983) (invalidating Ohio's early filing deadline for independent candidates for President); and Storer v. Brown, 415 U.S. 724, 94 S.Ct. 1274, 39 L.Ed.2d 714 (1974) (upholding California’s timing and signature requirements for appearing on ballot).
. U.S. Term Limits v. Thornton, 514 U.S. 779, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) dealt expressly with the people's constitutional right to vote for whomever they wish and analyzed the burden that lifetime term limits imposes on that right. The Court made it clear that it believed the burden to be severe. However, because the term limits in that case applied to federal, as opposed to state, legislators, the Court ultimately based its holding principally on the Qualifications Clause of the U.S. Constitution. See 514 U.S. at 787-826, 115 S.Ct. at 1847-66.
Justice Kennedy, who joined the Court’s opinion in U.S. Term Limits, wrote a separate concurrence explaining the differences between the majority and dissenting opinions regarding the Qualifications Clause. He did not, however, express any disagreement with the Court's view that lifetime term limits impose a severe limitation on the exercise of fundamental rights, nor did he disavow the historical analysis of that question that runs throughout the Court's opinion. Whatever one’s view of the outcome of any balancing test that may be used to determine whether the state's interest outweighs the burden inherent in limiting voters' rights, there should be little doubt that precluding all of the voters of the state from again voting for the candidates they have chosen to be their choice as their representatives severely limits their right to vote for whom they choose.
. The cases that the Court distinguished include all but one of the cases on which the state principally relies. While the Court did not mention Gregory in its decision, the reasons are obvious. First, the plaintiffs in Gregory did not assert any voting rights claim. Rather, the plaintiff-judges alleged only an equal protection violation, and no voter was a party to the proceeding. Moreover, there are three key differences between Missouri’s law in Gregory and the type of law to which the Court was referring. First, the judges did not run against other candidates in the elections; at the end of a fixed term, rather, a question appeared on the ballot as to whether they should be returned to office. See Gregory, 501 U.S. at 472, 111 S.Ct. at 2407 (citing Mo. Const., art. V, § 25(a)). Second, Missouri’s age-restriction, like minimum age restrictions, operated to exclude candidates for a very limited time period (the period of life expectancy after the age of 70). Third, electing judges invokes significantly different considerations than that of electing legislators. Legislators are supposed to draw public support by representing the popular will and passing laws in the people's interest; thus, under the democratic process, legislative elections, unlike unopposed judicial elections, are a crucial method of ensuring that public officials implement the people's desired state policies. See id.; cf. Reynolds, 377 U.S. at 562, 84 S.Ct. at 1381-82 (highlighting the paramount importance to our democracy of state legislatures). Gregory differs from the case before us in these same respects.
. States, however, are not free to create their legislative structures in violation of constitutional restraints. See Reynolds, 377 U.S. at 566, 84 S.Ct. at 1383-84; accord Lucas, 377 U.S. at 736, 84 S.Ct. at 1473-74. As the Court very recently emphasized, ”[s]tates ... enjoy wide latitude to establish conditions of candidacy for state office, but in setting such conditions, they may not disregard basic constitutional protections.” Chandler v. Miller, - U.S. -, -, 117 S.Ct. 1295, 1302, 137 L.Ed.2d 513 (1997) (citing Gregory, 501 U.S. at 463, 111 S.Ct. at 2402).
Nor, contrary to the state’s suggestion — which the dissent apparently adopts at least to some degree — may states insulate from judicial review their decisions regarding their political structures by claiming that they present nonjusticiable "political questions.” The Supreme Court has, for decades, soundly rejected this argument. See, e.g., Baker v. Carr, 369 U.S. 186, 209-37, 82 S.Ct. 691, 705-21, 7 L.Ed.2d 663 (1962). It is abundantly clear since that historic decision that constitutional claims are not rendered nonjusticiable merely "because they touch matters of stale governmental organization.” Id. at 229, 82 S.Ct. at 716-17. Clements v. Fashing, 457 U.S. 957, 102 S.Ct. 2836, 73 L.Ed.2d 508, upon which the dissent relies, merely states the obvious under that rule: that courts should not disturb "the manner in which a State has decided to govern itself" if that manner does not violate the First or Fourteenth Amendment. Id. at 972, 102 S.Ct. at 2848.
. Contrary to the suggestion of the Schabarum intervenors, the ability of states to disenfranchise convicted felons does not demonstrate that permanent restrictions on the franchise are reasonable. The Court has held that the permanent disenfranchisement of convicted felons does not violate equal protection for the specific reason that it has "an affirmative sanction in § 2 of the *855Fourteenth Amendment” demonstrated by both the text and history of that provision. Richardson v. Ramirez, 418 U.S. 24, 54, 94 S.Ct. 2655, 2671, 41 L.Ed.2d 551 (1974). The Schabarum intervenors do not suggest, and we see no basis for concluding, that term limits enjoy similar support in the text and history of the Constitution.
. This language stands in contrast not only to other stales' lifetime bans, see supra note 4, but also to the lifetime-presidential-term-limits provision of the U.S. Constitution. See U.S. Const., amend. XXII ("No person shall be elected to the office of the President more than twice ....") (emphasis added). Proposition 140’s drafters, therefore, easily could have presented to the voters an unambiguous lifetime-ban provision by simply tracking the language of the U.S. Constitution.
. California is no Lake Wobegon where all people are "above average,” see Garrison Keillor, Lake Wobegon Days (1985); nor, obviously, are 96% of the voters who cast ballots on the measure, pro or con, "average,” or above. (Ninety-six percent of the "yes" voters is the number of voters necessary to constitute a majority Of those voting.)
. We emphasize that the closeness of the vote is not necessary to our decision. Nevertheless, we note that the state now contends that "[n]o voter could have missed the point.” Supp. Br. at 2. In doing so, the state flatly contradicts its statements to the state supreme court regarding the measure's ambiguity. Ambiguity, in itself, shows that some voters, at least, are likely to have “missed the point.”
. See, e.g., Kent v. Dulles, 357 U.S. 116, 129-30, 78 S.Ct. 1113, 1119-21, 2 L.Ed.2d 1204 (1958) (declining, because no clear statement was present, to read statute delegating authority to the Secretary of Slate as granting the authority to limit people’s fundamental right to travel and, therefore, invalidating State Department directive limiting citizens' passport rights).
. Specifically, the Court stated: "When the Federal Government asserts an overriding national interest as justification for a discriminatory rule which would violate the Equal Protection Clause if adopted by a State, due process requires that there be a legitimate basis for presuming that the rule was actually intended to serve that interest." 426 U.S. at 103, 96 S.Ct. at 1905. The majority invalidated the regulations despite the dissent's reminder that "the power of the federal courts is severely limited in the areas of immigration and regulation of aliens.” Id. at 117, 96 S.Ct. at 1912 (Rehnquisl, J., dissenting) (gathering cases). The power of the federal courts, on the other hand, is perhaps at its zenith in the areas of the First and Fourteenth Amendments and voting rights. We therefore believe that the procedural due process requirements identified in Hampton apply with particular force to initiatives like Proposition 140.
The Court’s Hampton decision was very much like two decisions in which the Court had applied due process to direct democracy in the context of restraints on the use of land. See Washington ex rel. Seattle Title Trust Co. v. Roberge, 278 U.S. 116, 49 S.Ct. 50, 73 L.Ed. 210 (1928); Eubank v. City of Richmond, 226 U.S. 137, 33 S.Ct. 76, 57 L.Ed. 156 (1912). In both cases, the court struck down as violative of due process local ordinances that allowed zoning decisions to be made simply by a two-thirds popular vote of the residents in the area. In both cases, the Court concluded that the authority of the property owners "uncontrolled by any standard or rule prescribed by the legislative action,” with no provision for review, violated the people’s due process right to a considered decision by an independent body. Roberge, 278 U.S. at 121-22, 49 S.Ct. at 51-52; accord Eubank, 226 U.S. at 144, 33 S.Ct. at 77.
. It is unclear from the Court’s decision whether it struck down Oklahoma’s law or whether it somehow merely held it inapplicable. Obviously, however, the result was the same: the Court held that the state could not enforce its law.
. Although the holding' was "unusual,” it certainly did not lack precedent. For another such holding, see Califano v. Goldfarb, 430 U.S. 199, 223 n. 9, 97 S.Ct. 1021, 1035 n. 9, 51 L.Ed.2d 270 (1977) (Stevens, J., concurring) (casting deciding vote to hold that Social Security Act provision violated equal protection right because there was no clear evidence of a "considered legislative choice,” but stating that "[pjerhaps an actual, considered legislative choice would be sufficient to allow this statute to be upheld, but that is a question I would reserve until such a choice has been made”).
. See, e.g., Guido Calabresi, A Common Law for the Age of Statutes 70-71 (1982) (describing on paucity of procedural protections in initiative process); Julian Eule, Judicial Review in a Direct Democracy, 99 Yale L.J. 1503, 1525 (1990) (contrasting initiative measures with legislation).
. The financial cost of passing or defeating ballot measures, and of qualifying them for the ballot in the first place, is inordinately high. See, e.g., Coalition for Economic Equity v. Wilson, 946 F.Supp. 1480, 1498-99 (1996) (describing the costs — in the millions of dollars — of introducing and supporting initiatives in California), vacated on other grounds, 110 F.3d 1431 (9th Cir.1997); Bill Stall & Dan Morain, Prop. 209 Wins, Bars Affirmative Action Initiatives, L.A. Times, Nov. 6, 1996, at A1 (stating that $45 million spent by both sides combined on Proposition 211 and that $85 million was spent on campaigns for initiatives in election).
. As Hans Linde, a former Justice on the Oregon Supreme Court, once explained:
Whatever the private goals of the sponsors, once a measure is drafted it is past systematic factfinding, analysis, amendment or compromise. Aside from newspaper editorials or an occasional voters' pamphlet, the debate leading to discussion is left to the electioneering slogans of competing advertising firms. Yet such a measure may repeal, alter, or contradict the most carefully studied and best designed enactment of the legislature.
Hans Linde, Due Process of Lawmaking, 55 Neb. L.Rev. 197, 228 (1976).
. Compare, e.g., Ogden v. Saunders, 25 U.S. (12 Wheat.) 213, 6 L.Ed. 606 (1827):
It is but a decent respect due to the wisdom, the integrity and the patriotism of the legislative body, by which any law is passed, to presume in favour of its validity, until its violation of the constitution is proved beyond all reasonable doubt. This has always been the language of this court, when that subject has called for its decision; and I know that it expresses the honest sentiments of each and every member of this bench.
Id. at 270.
. Any materials that advocacy groups circulate describing the initiative .are not analogous sources, for they are not produced by the voters through a deliberative process. Such materials are, at bottom, only advertisements. Relying on them as indicative of the voters' intent would be tantamount to relying on political parties' campaign advertisements to interpret legislative acts.
. Also compare San Diego Coast Reg'l Comm’n v. See the Sea, Ltd., 9 Cal.3d 888, 891, 109 Cal.Rptr. 377, 513 P.2d 129 (1973) (construing initiative as not imposing construction moratorium because the initiative "[n]owhere ... expressly provide[d] for a moratorium") with id. at 897 n. 5, 109 Cal.Rptr. 377, 513 P.2d 129 (Mosk, J., dissenting) (noting that the opponents’ ballot argument “warned that the act would impose a moratorium”).
. After stating that the California Courts may, under California law, examine material that lies "outside the four corners of the initiative” in order to determine voter intent, the dissent relies extensively on material either not before the California Supreme Court or rejected by it. We are aware, for example, that when the voters adopted Proposition 140 they also rejected Proposition 131, which among other things, provided expressly for state-financed election campaigns as well as consecutive term limits. The California Supreme Court did not mention Proposition 131 in its opinion and the state also does not mention it in its briefs in this case. The state, however, did point out the Proposition in its brief to the California Supreme Court and contended that the court could not reasonably conclude that the voters rejected Proposition 131 on the basis that its term limits were not permanent because "Proposition 131 was 14 pages long (Ballot Pamphlet, pp. 101-105) and covered several, complex issues ... [including] expenditures of public funds, but was vague on the certainty of commensurate revenues.” Br. of Secretary of State, Legislature of the State of Cal. v. Eu, 54 Cal.3d 492, 286 Cal.Rptr. 283, 816 P.2d 1309 (1991), at 25 n. 13. The proponents then seized on the state’s argument and urged an opposite position in its reply brief. We agree with the state's argument and with the California Supreme Court's decision to give no weight to the differences in the term limits language in the two propositions and to decline to speculate as to why the voters adopted one measure and not the other. Although some newspaper articles and the opponents' ballot arguments contrasted the measures as “lifetime limits versus consecutive limits," such editorials and arguments hardly rise to the level of something like legislative history.
.Fundamental rights of course qualify as "liberty interests," thus triggering the protections of the Fourteenth Amendment’s Due Process Clause. See, e.g., Paul v. Davis, 424 U.S. 693, 711 n. 5, 96 S.Ct. 1155, 1165 n. 5, 47 L.Ed.2d 405 (1976) (noting that rights protected by the Bill of Rights are protected by procedural due process requirements).
. Voters may enforce all of these mechanisms in state court, but only to a limited extent. See Cal. Elec.Code § 9092 (allowing voters to seek a peremptory writ of mandate to require title, summary, or statement to be amended or deleted from ballot pamphlet). The California courts will issue such a writ, however, "only upon clear and convincing proof that the copy in question is false, misleading, or inconsistent with the requirements of [the election code].” Lungren v. Superior Court, 48 Cal.App.4th 435, 439, 55 Cal.Rptr.2d 690 (1996). When the Attorney General has already provided an "impartial statement” of the initiative, a voter may not compel a writ to require him "fully [lo] state[ ] the main purpose or chief point of the initiative.” See id. at 441, 443, 55 Cal.Rptr.2d 690 (vacating lower court order requiring such action).
. In response to the court's question, "Must it be clearly put to [the voters] so that they know what they are voting on — so that they know that they are adopting these term limits for life?,” Professor Elhauge stated, "Well, I think if the question is should it be put to them that they are adopting term limits for life, uh, yes. And I think the answer to that is yes.”
. We also note that requiring a state to provide adequate notice of an initiative’s principal effect if that effect is to severely burden the people’s fundamental rights, constitutes a reasonable and effective method of safeguarding the people’s right to a republican form of government. Although. the Court in Pacific States ruled that the Guaranty Clause, U.S. Const., art. IV, § 4, did not impose a substantive limitation on the states’ ability to enact laws through the initiative process, nothing in that holding disqualifies federal courts from utilizing other techniques of judicial review to protect the constitutional values at stake. In fact, this is a very similar method to the one that the Court utilizes to protect the constitutional values implicated by its refusal to draw substantive boundaries to cabin the non-delegation doctrine, see Mistretta v. United States, 488 U.S. 361, 371-79, 109 S.Ct. 647, 654-59, 102 L.Ed.2d 714 (1989), and to protect states' "fundamental functions” from federal intrusion under the Commerce Clause. Compare Garcia v. San Antonio Metro. Transit Auth., 469 U.S. 528, 105 S.Ct. 1005, 83 L.Ed.2d 1016 (1985) (holding such intrusion nonjusticiable) with Gregory, 501 U.S. at 457-64, 111 S.Ct. at 2399-2403 (adopting plain statement rule). Precisely because the people may enact any type of law they wish by initiative, and because most of the deliberative filters of legislation have been removed, federal courts must ensure that states do not exploit the deficiencies of direct democracy to limit the people’s rights through ambiguous initiatives that fail to disclose the true objectives of the measure.