with whom McMILLIAN and MURPHY, Circuit Judges, join, dissenting.
The Court today strikes down the partisan activities clauses and the solicitation restriction as a matter of law, by summary judgment, ruling that the interests at stake are not compelling and that the clauses of Canon 5 are either too broad, or not broad enough, to justify their own existence. Preserving the integrity of a state’s courts and those courts’ reputation for integrity is an interest that lies at the very heart of a state’s ability to provide an effective government for its people. The word “compelling” is hardly vivid enough to convey its importance. The questions of whether that interest is threatened by partisan judicial election campaigns and personal solicitation of campaign contributions, and whether the measures Minnesota has adopted were crafted to address only the most virulent threats to that interest, are in part factual questions, which we should not decide on summary judg*767ment. Finally, the Court today adopts an approach to strict scrutiny that would deny the states the ability to defend their compelling interests, no matter how urgent the threat. For these reasons, I respectfully dissent.
I.
The partisan activities clauses and the solicitation restriction each serve an interest that is and has been recognized as compelling-protecting the judicial process from extraneous coercion.
A.
In the district court, the Minnesota Boards argued that the state’s compelling interest was in protecting judicial independence and impartiality, concepts that were not further defined, perhaps because the Boards considered their meaning apparent. When the announce clause was before the Supreme Court, the opinion authored by Justice Scalia determined that further definition and analysis were essential in order to determine whether impartiality was a compelling state interest and whether the announce clause was narrowly tailored to serve that interest. Republican Party of Minn. v. White, 536 U.S. 765, 775, 122 S.Ct. 2528, 153 L.Ed.2d 694 (2002). Justice Scalia divined three possible meanings for judicial “impartiality.” The last meaning was “open-mindedness.” Id. at 778, 122 S.Ct. 2528.
This quality in a judge demands, not that he have no preconceptions on legal issues, but that he be willing to consider views that oppose his preconceptions, and remain open to persuasion, when the issues arise in a pending case. This sort of impartiality seeks to guarantee each litigant, not an equal chance to win the legal points in the case, but at least some chance of doing so.
Id. (emphasis in original). Because the announce clause was “woefully underinclu-sive” to serve any interest in judicial open-mindedness, Justice Scalia concluded that Minnesota had not adopted the announce clause in order to further such an interest; he therefore found it unnecessary to consider whether preserving “judicial open-mindedness” was a compelling state interest. Id. at 778-80, 122 S.Ct. 2528. Since White, the New York Court of Appeals has held that judicial open-mindedness is a compelling interest because “it ensures that each litigant appearing in court has a genuine-as opposed to illusory-opportunity to be heard.” In re Watson, 100 N.Y.2d 290, 763 N.Y.S.2d 219, 794 N.E.2d 1, 7 (2003).
After White, by order of December 9, 2003, the Minnesota Supreme Court created an Advisory Committee to review its Canons 3 and 5 in light of White. In re Amendment of the Code of Judicial Conduct, No. C4-85-697, Slip op. at 1 (Minn. Sept. 14, 2004) (recounting history). The Committee received public comment and held a hearing. Report of the Advisory Committee to Review the Minnesota Code of Judicial Conduct and the Rules of the Board on Judicial Standards, Acknowl-edgements (April 15, 2004). Following the Committee’s report, the Minnesota Supreme Court held its own hearing and received public comment. In re Amendment of the Code of Judicial Conduct, No. C4-85-697, Slip op. at 1 (Minn. Sept. 14, 2004). In September 2004, the Minnesota Supreme Court amended Canon 5 to add a definition of impartiality that explicitly includes open-mindedness:
“Impartiality” or “impartial” denotes absence of bias or prejudice in favor of, or against, particular parties or classes of parties, as well as maintaining an open mind in considering issues that may come before the judge.
Canon 5E (as amended Sept. 14, 2004).
The Court today discusses open-mindedness as if the concern were to protect *768judicial candidates from experiences that would affect their subjective frame of mind. Thus, the Court holds that the state’s interest cannot be served by measures that only limit the candidate’s conduct during a campaign, not before: “The few months a candidate is ostensibly purged of his association with a political party can hardly be expected to suddenly open the mind of a candidate who has engaged in years of prior political activity.” Op. at 757-58.
This answers the easy question but ignores the hard one. The threat to open-mindedness at which the partisan activities and solicitation clauses aim comes not from within the candidates, but from without and consists of the candidates placing themselves in debt to powerful and wide-reaching political organizations that can make or break them in each election. This is a fundamental distinction between the partisan activities and solicitation clauses, on the one hand, and the announce clause, which was at issue in White. A central tenet of Justice Scalia’s opinion in White was that the announce clause regulated a candidate’s relation to issues, not people. See 536 U.S. at 776, 122 S.Ct. 2528 (The announce clause “does not restrict speech for or against particular parties, but rather speech for or against particular issues.”).17 The partisan activities and solicitation clauses regulate how certain speech affects a judicial candidate’s relations with people, and organizations of people, not the candidate’s relations with issues.
Our Court’s concern with temporal un-derinclusiveness, Op. at 757-58, is largely a result of its failure to address the threat to open-mindedness from external pressure. The threat to open-mindedness results from allowing the candidates to incur obligations during a campaign that can affect their performance in office. Once a person becomes a candidate, the significance of his relations with the party changes radically, as the party becomes empowered to play the role of judge-maker. Any regulation that governed relations with the party before a person had become a candidate would be overly broad, but a regulation that focuses on the campaign period is tailored to address the threat in the time-frame in which the threat is most overt.
Similarly, the Court today dismisses the danger of bias from partisan involvement in judicial campaigns even in cases in which the political parties are litigants, on the ground that the only relevant link between the judge and the party is that both have espoused similar positions on “particular issues embraced by the political party.” Op. at 755. To the contrary, once the partisan activities clauses are gone, having espoused similar positions on issues will be the least significant aspect of the party’s relationship to its successful candidate; the truly significant point is that the candidate may owe his or her accession to the bench to the litigant before the bar and may be similarly dependent on that litigant for any hope of success in future elections. Once the partisan activities clauses are gone, one may expect that party involvement will Become the norm, so that recusal *769would be pointless, since all judges would be similarly compromised.
B.
“Open-mindedness,” in Justice Scalia’s terminology, is in reality simply a facet of the anti-corruption interest that was recognized in Buckley v. Valeo, 424 U.S. 1, 26-27, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976), and subsequent campaign finance cases. See FEC v. Nat’l Conservative PAC, 470 U.S. 480, 496-97, 105 S.Ct. 1459, 84 L.Ed.2d 455 (1985) (referring to “preventing corruption or the appearance of corruption” as “legitimate and compelling government interests”); FEC v. Nat’l Right to Work Comm., 459 U.S. 197, 208, 103 S.Ct. 552, 74 L.Ed.2d 364 (1982) (importance of avoiding corruption has never been doubted and is linked to “the integrity of our electoral process” and the responsibility of the citizen for the successful functioning of that process). Allowing those responsible for the various branches of government to contract obligations inconsistent with the discharge of their public duties threatens the republic created by the Constitution; a republic designed so that it could not protect itself from such a threat would contain the seeds of its own destruction. See Nixon v. Shrink Missouri Gov’t PAC, 528 U.S. 377, 390, 120 S.Ct. 897, 145 L.Ed.2d 886 (2000) (“Leave the perception of impropriety unanswered, and the cynical assumption that large donors call the tune could jeopardize the willingness of voters to take part in democratic governance.”).
Corruption is a sufficiently serious threat to our institutions that the government may (1) seek to prevent it before it happens and (2) act against it in intermediate forms that are more subtle than bribery and explicit agreements. See McConnell v. FEC, 540 U.S. 93, 144, 150-54, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003) (corruption and appearance of corruption extend beyond bribery to other arrangements which create “sense of obligation” in or “undue influence” over officeholders); Shrink Missouri Gov’t PAC, 528 U.S. at 389, 120 S.Ct. 897 (“In speaking of ‘improper influence’ and ‘opportunities for abuse’ in addition to ‘quid pro quo arrangements,’ we recognized a concern not confined to bribery of public officials, but extending to the broader threat from politicians too compliant with the wishes of large contributors. These were the obvious points behind our recognition that the Congress could constitutionally address the power of money to ‘influence governmental action’ in ways less ‘blatant and specific’ than bribery.”); Nat’l Right to Work Comm., 459 U.S. at 210, 103 S.Ct. 552 (“Nor will we second-guess a legislative determination as to the need for prophylactic measures where corruption is the evil feared.”).
Admittedly, the concern with corruption in the campaign finance cases focuses on payment of money. While the solicitation clause also deals with money-raising, the partisan activities clauses do not, which distinguishes them from the campaign finance cases. Nevertheless, the Supreme Court’s decision in United States Civil Serv. Comm’n v. Nat’l Ass’n of Letter Carriers, 413 U.S. 548, 93 S.Ct. 2880, 37 L.Ed.2d 796 (1973), demonstrates that the concern with corruption and undue influence is not limited to obligations resulting from payments of money. Letter Carriers recognized the danger partisan allegiances posed to neutral administration of justice. That case upheld restraints imposed by the Hatch Act on executive branch employees’ political activities, in part because of the effect partisanship could have on the performance of their duties:
It seems fundamental in the first place that employees in the Executive Branch of the Government, or those working for any of its agencies, should administer the law in accordance, with the will of *770Congress, rather than in accordance with their own or the will of a political party. They are expected to enforce the law and execute the programs of the Government without bias or favoritism for or against any political party or group or the members thereof. A major thesis of the Hatch Act is that to serve this great end of Government-the impartial execution of the laws-it is essential that federal employees, for example, not take formal positions in political parties, not undertake to play substantial roles in partisan political campaigns, and not run for office on partisan political tickets. Forbidding activities like these will reduce the hazards to fair and effective government.
413 U.S. at 564-65, 93 S.Ct. 2880. Accord In re Raab, 100 N.Y.2d 305, 763 N.Y.S.2d 213, 793 N.E.2d 1287, 1290-91 (2003) (per curiam). Letter Carriers shows that what kind of obligations may be considered inconsistent with government office depends on the nature of the office in question. Where the office requires “impartial execution of the laws,” partisan entanglements can be inconsistent with the demands of the office. Letter Carriers and the campaign finance cases are not separate lines of authority, but are closely connected, since the Supreme Court relied heavily on Letter Carriers in identifying and defining the anti-corruption interest in Buckley v. Valeo, 424 U.S. at 27, 96 S.Ct. 612.
The Republican Party of Minnesota argues that the holding of Letter Carriers is irrelevant here because “the role of judges is closer to the role of legislators than [the] executive branch bureaucrats” affected by Letter Carriers. Supplemental brief of Oct. 22, 2002 at 6 n. 6. The Supreme Court in White specifically avoided equating the judicial office with the legislative: “[W]e neither assert nor imply that the First Amendment requires campaigns for judicial office to sound the same as those for legislative office.” 536 U.S. at 783, 122 S.Ct. 2528.
The need for “neutrality” identified in Letter Carriers is even more important for the judicial branch than the executive. A long line of cases stresses the right of litigants to a neutral adjudicator. In Marshall v. Jerrico, Inc., 446 U.S. 238, 242, 100 S.Ct. 1610, 64 L.Ed.2d 182 (1980), Justice Marshall wrote:
The Due Process Clause entitles a person to an impartial and disinterested tribunal in both civil and criminal cases. This requirement of neutrality in adjudicative proceedings safeguards the two central concerns of procedural due process, the prevention of unjustified or mistaken deprivations and the promotion of participation and dialogue by affected individuals in the decisionmak-ing process. The neutrality requirement helps to guarantee that life, liberty, or property will not be taken on the basis of an erroneous or distorted conception of the facts or the law. At the same time, it preserves both the appearance and reality of fairness, “generating the feeling, so important to a popular government, that justice has been done,” by ensuring that no person will be deprived of his interests in the absence of a proceeding in which he may present his case with assurance that the arbiter is not predisposed to find against him.
(citations omitted).
The circumstances that can result in violation of the due process right to a neutral judge are not limited to situations in which a judge has a pecuniary interest at stake in the litigation. In Ward v. Village of Monroeville, 409 U.S. 57, 93 S.Ct. 80, 34 L.Ed.2d 267 (1972), due process was violated when the defendant was convicted by the mayor of a village that depended on fines collected in the mayor’s court. The *771mayor himself did not share in the revenues, but the Supreme Court held:
[T]he test is whether the mayor’s situation is one “which would offer a possible temptation to the average man as a judge to forget the burden of proof required to convict the defendant, or which might lead him not to hold the balance nice, clear and true between the State and the accused .... ” [quoting Tumey v. Ohio, 273 U.S. 510, 532, 47 S.Ct. 437, 71 L.Ed. 749 (1927).] Plainly that “possible temptation” may also exist when the mayor’s executive responsibilities for village finances may make him partisan to maintain the high level of contribution from the mayor’s court. This, too, is a “situation in which an official perforce occupies two practically and seriously inconsistent positions, one partisan and the other judicial,” [and] necessarily involves a lack of due process of law in the trial of defendants charged with crimes before him.
Id. at 60, 93 S.Ct. 80. Other cases in which a judge’s neutrality was intolerably compromised by non-pecuniary considerations include Johnson v. Mississippi, 403 U.S. 212, 215-16, 91 S.Ct. 1778, 29 L.Ed.2d 423 (1971) (judge who had previously lost a civil rights suit to defendant could not try defendant for contempt); In re Murchison, 349 U.S. 133, 75 S.Ct. 623, 99 L.Ed. 942 (1955) (judge who acts as “one-man judge-grand jury” cannot then try indicted defendant); and Offutt v. United States, 348 U.S. 11, 17, 75 S.Ct. 11, 99 L.Ed. 11 (1954) (judge who had become “personally embroiled” with lawyer could not try the lawyer for contempt). There is no easily applied formula for deciding when a judge has an impermissible interest in litigation; “Circumstances and relationships must be considered.” Murchison, 349 U.S. at 136, 75 S.Ct. 623. But even at a point where the judge cannot be said to be actually biased, “[0]ur system of law has always endeavored to prevent even the probability of unfairness.” Id. (emphasis added).
We do not have to conclude that adjudication by judges who were selected without the protections of the partisan activities clauses violates the due process rights of the litigants. Indeed, we could not well do so, since, as the Supreme Court remarked in White, partisan judicial elections were common in the mid-nineteenth century, when the Fourteenth Amendment was adopted. See White, 536 U.S. at 785, 122 S.Ct. 2528 (“[Jjudicial elections were generally partisan during this period [nineteenth and early twentieth centuries], the movement toward nonpartisan judicial elections not even beginning until the 1870’s.”). Currently, some fifteen states maintain partisan judicial elections for at least some of their judges, American Judicature Society, Judicial Selection in the States: Appellate and General Jurisdiction Courts (2004); see Suessmann v. Lamone, 383 Md. 697, 862 A.2d 1, 19 (2004) (Maryland circuit court primaries are held to be partisan), and no one contends that the states may not choose this method of selection.
Nevertheless, the participation of judges who have been allowed or forced to make themselves dependent on party largesse for their continued tenure affects the state’s ability to provide neutral judges and the public’s perception of such neutrality. The state has a compelling interest in keeping its judges free from the odor of self-interest or partisanship. In Cox v. Louisiana, 379 U.S. 559, 85 S.Ct. 476, 13 L.Ed.2d 487 (1965), Cox was convicted of picketing near a courthouse “with the intent of influencing any judge, juror, witness, or court officer, in the discharge of his duty” after he demonstrated in protest against the arrest of students who would be tried by judges present at the courthouse during the demonstration. Id. at *772560, 564-65, 85 S.Ct. 476. The law did not prohibit all picketing, but picketing done with the intent to influence the administration of justice. Cf. Carey v. Brown, 447 U.S. 455, 460-63, 100 S.Ct. 2286, 65 L.Ed.2d 263 (1980) (prohibiting picketing for one purpose while allowing it for another is content discrimination). The Supreme Court stated that it was unlikely that the picketing would actually affect the judges’ decisions in the students’ cases. Cox, 379 U.S. at 565, 85 S.Ct. 476. The Court did not suggest that if the state had allowed the picketing, it would have caused a due process violation for any particular litigant. However, the Court said, “A state may protect against the possibility of a conclusion by the public under these circumstances that the judge’s action was in part a product of intimidation and did not flow only from the fair and orderly working of the judicial process.” Id. Thus, the state’s interest in preserving the appearance of neutrality justified the restriction on expressive conduct. Accord In re Chmura, 461 Mich. 517, 608 N.W.2d 31, 40 (2000) (“state’s interest ... extends to preserving public confidence in the judiciary”). Cf. Letter Carriers, 413 U.S. at 565, 93 S.Ct. 2880 (“[I]t is not only important that the Government and its employees in fact avoid practicing political justice, but it is also critical that they appear to the public to be avoiding it, if confidence in the system of representative Government is not to be eroded to a disastrous extent.”).
The New York Court of Appeals recently confirmed that a state has a compelling interest in presenting the appearance as well as the fact of due process:
[LJitigants have a right guaranteed under the Due Process Clause to a fair and impartial magistrate and the State, as the steward of the judicial system, has the obligation to create such a forum and prevent corruption and the appearance of corruption, including political bias or favoritism.
In re Raab, 100 N.Y.2d 305, 763 N.Y.S.2d 213, 793 N.E.2d 1287, 1290-91 (2003) (per curiam).
In its September 2004 deliberations about whether to amend the partisan activities clauses of Canon 5, the Minnesota Supreme Court articulated just those concerns outlined above. The court stated:
[T]he goal of an impartial judiciary is compelled by the due process rights of litigants. Due process requires deci-sionmakers who are fair, unbiased, and impartial, and importantly, decisionmak-ers who are perceived as such by the litigants who appear before them. See Aetna Life Ins. Co. v. Lavoie, 475 U.S. 813, 825, 106 S.Ct. 1580, 89 L.Ed.2d 823 (1986); In re Raab, 100 N.Y.2d 305, 763 N.Y.S.2d 213, 793 N.E.2d 1287, 1290-91 (2003) (per curiam). Moreover, we cannot underestimate the importance of the public’s perception that judges are fair, unbiased, and impartial to the continued respect for and legitimacy of the judicial branch. See Mistretta v. United States, 488 U.S. 361, 407, 109 S.Ct. 647, 102 L.Ed.2d 714 (1989). Without this perception, the public’s confidence and support cannot be maintained and the very independence of the judicial branch mandated by the Constitution will be threatened.
In re Amendment of the Code of Judicial Conduct, No. C4-85-697, slip op. at 4-5 (Minn. Sept.14, 2004). These concerns fit within the concept of judicial open-mindedness, and they are a compelling state interest.
C.
Although in White Justice Scalia observed that the parties and this Court appeared to make no distinction between the concepts of judicial “independence” and “impartiality,” 536 U.S. at 775, 122 S.Ct. 2528 n.6, in its September 14, 2004 *773order, the Minnesota Supreme Court explained its decision not to amend the partisan activities clauses partly by relying on the need for separation of powers:
[T]he separation of powers inherent in the creation of three distinct branches of government, one of which is the judicial branch, in article III, section 1 of the Minnesota Constitution provides the constitutional underpinning for the independence of the Minnesota judiciary. As the executive and legislative branches are inextricably intertwined with partisan politics, maintenance of an independent judicial branch is reliant on the freedom of its officials from the control of partisan politics.
In re Amendment of the Code of Judicial Conduct, No. C4-85-697, slip op. at 4-5 (Minn. Sept.14, 2004). The separation of powers interest is a concern for institutional independence that is distinct from concern for impartiality in any of the senses identified by Justice Scalia.
Separation of powers is a concept basic to the states’ constitutions as well as the federal Constitution. A state’s choice of how to organize its government is “a decision of the most fundamental sort for a sovereign entity.” See Gregory v. Ashcroft, 501 U.S. 452, 460, 111 S.Ct. 2395, 115 L.Ed.2d 410 (1991) (authority of the people of the states to determine qualifications of important state officials is such a decision). Even the narrowest notion of federalism requires us to recognize a state’s interest in preserving the separation of powers within its own government as a compelling interest.
D.
The extent and severity of the threat to the state’s interests are factual questions that must be proven empirically. See Shrink Mo. Gov’t PAC, 528 U.S. at 390-94, 120 S.Ct. 897. In the proceedings in the district court, the Boards adduced sufficient evidence of that threat so that summary judgment for the plaintiffs would not have been appropriate. But recent events make it far less appropriate that our Court should enter judgment as a matter of law on questions of fact as to which there is no record before us.
The record below contained the affidavit of a former governor of Minnesota who stated that he had a lifetime of experience in understanding how Minnesota citizens “think and feel” and that partisan judicial campaigns would lessen Minnesotans’ confidence “in the independence of the judiciary.” A former Chief Justice of the Minnesota Supreme Court stated that partisan judicial campaigns would “put pressure on judges to decide cases in ways that would impress the judge’s supporters favorably.”
But far more important to our holding today is the fact that the Minnesota Supreme Court has recently reconsidered the provisions of Canon 5 at issue here, held hearings, and received public comment. It is a matter of interest that the parties in this case, in briefing and argument, made no méntion of this development.- As the canon was reconsidered, amended in part and reiterated in part while this case was pending on rehearing, failure to consider the effect of these developments may well cause this Court’s opinion to be moot from its inception.
This suit seeks a permanent injunction; the relief the Court orders today does not merely look to the past, when the district court record was made, but operates in the present and into the future. See Stuart Minor Benjamin, Stepping Into the Same River Twice: Rapidly Changing Facts and the Appellate Process, 78 Tex. L.Rev. 269, 276-77 (1999) (where relief is prospective, appellate court should respond to changes in facts that affect validity of legal ruling). The Supreme Court in a number of cases has made clear that we must *774consider on appeal any change, either in fact or in law, which has supervened since the judgment was entered. The Court in Ashcraft v. Tennessee, 322 U.S. 143, 156, 64 S.Ct. 921, 88 L.Ed. 1192 (1944), stated: “In disposing of cases before us it is our responsibility to make such disposition as justice may require. ‘And in determining what justice does require, the Court is bound to consider any change, either in fact or in law, which has supervened since the judgment was entered.’ Patterson v. Alabama, 294 U.S. 600, 607, 55 S.Ct. 575, 79 L.Ed. 1082 [1935]; State Tax Commission v. Van Cott, 306 U.S. 511, 515-16, 59 S.Ct. 605, 83 L.Ed. 950 [1939].”
The Advisory Committee appointed by the Minnesota Supreme Court to study the issue concluded that there was a threat to ■ the state’s interest that required regulation of partisanship in judicial campaigns: “In considering the need for restrictions on the political activity of judicial election candidates, the Advisory Committee is also cognizant of the experience of actual or perceived corruption of the judiciary in states that permit partisan judicial elections.” Report of the Advisory Committee to Review the Minnesota Code of Judicial Conduct and the Rules of the Board on Judicial Standards, Comments-Canon 5, No. C4-85-697 (Minn. April 15, 2004). While we do not have access to the evidence before the Committee, widely available and publicized evidence substantiates the fear that the majority of the public believes that partisanship does influence the decisions of state courts. For instance, a poll conducted in 1999 showed that 81% of the respondents agreed that “politics influences court decisions.” National Center for State Courts, How the Public Views the State Courts 41 (1999). The Advisory Committee recommended deleting the party identification and the attend and speak clauses on narrow tailoring grounds. Advisory Committee Report at Comments-Canon 5. After receiving the Committee report and conducting a hearing and receiving public comment, the Minnesota Supreme Court decided to retain all three partisan activities clauses. In re Amendment of the Code of Judicial Conduct, No. c4-85-697 (Minn. Sept. 14, 2004).
The Advisory Committee unanimously recommended against changing the ban on the judicial candidate’s personal solicitation of campaign contributions. Advisory Committee Report at Canon 5B(2)-Person-al Solicitation of Campaign Contributions. Again, widely available poll numbers support the Committee’s conclusion that solicitation of campaign contributions carries with it a significant threat to the state’s interest in freedom from external coercion of judges. For example, “a recent Wisconsin poll found that more than three-quarters of those surveyed believe that campaign contributions from lawyers and plaintiffs in high-profile cases influence the decisions of these judges in court,” and a study in Texas “found that 83 percent of the public and 79 percent of lawyers believe that campaign contributions have a significant influence on a judge’s decision.” Alexander Wohl, Justice for Rent, The American Prospect (May 22, 2000). A summary of poll results compiled by the National Center for State Courts reported on fifteen recent polls, showing not only that the public believes campaign contributions affect judicial decisions, but also that lawyers and even judges agree. For instance, a poll from Texas showed that 48% of state appellate and trial judges surveyed believed that campaign contributions had a fairly significant or very significant degree of influence over judicial decisionmaking. National Center for State Courts, Summary of Responses to Public Opinion Survey Questions Concerning Judicial Campaign Fund-raising (July 28, 2004). A Pennsylvania survey of registered voters showed that *77595% of those surveyed believed that judges’ decisions were influenced by large contributions to their election campaigns at least some of the time. Id. This is the hind of evidence that would substantiate the threat to judicial open-mindedness (and the appearance of it) from partisan obligations and from judicial campaign fund-raising.
The Court today errs grievously in issuing a ruling that strikes the provisions based on the 1997 factual record without considering the September 2004 record before the Minnesota Supreme Court. Cf. Ashcroft v. ACLU, 542 U.S. 656, 124 S.Ct. 2783, 2794, 159 L.Ed.2d 690 (2004) (taking into account technological changes in internet filtering since time of district court’s findings). Since the holding is based on a factual record that antedates the most recent version of Canon 5, one must question whether the Court’s holding today even applies to the current version of Canon 5, based as it is, on a 2004 factual determination which the Court does not take into account.
E.
The Court today holds that Minnesota’s interest in judicial open-mindedness is not a compelling interest because the solicitation and partisan activities clauses are “un-derinclusive,” meaning that they do not address all “significant threats” to the state’s asserted interest. Op. at 759. The Court today says that underinclusiveness of a regulation will establish that the state’s purported interest is not compelling:
A clear indicator of the degree to which an interest is ‘compelling’ is the tightness of the 'fit between the regulation and the purported interest: where the regulation fails to address significant influences that impact the purported inter-e$t, it usually flushes out the fact that the interest does not rise to the level of being ‘compelling.’
Op. at 750. This is only partly correct and the part that is correct is not relevant to our case.
Where the governmental actor asserts an interest that has not been previously recognized as compelling • and either the importance of the interest or the urgency of the threat to the interest is debatable, adoption of a regulation that is only partly effective may show that the governmental actor can, indeed, live with the problem. Where the state’s asserted interest is previously unrecognized as compelling and is not self-evidently vital to the state’s ability to function as intended in the Constitution, then the state’s own lack of zeal or care in protecting the interest does argue that the asserted interest should not be considered paramount to individuals’ First Amendment rights.
However valid that reasoning may be in cases where the asserted interest is novel or questionable, it is not valid here because the interests at stake in this case have already been recognized as compelling. Compelling interests cannot be negated simply because a particular measure adopted in their name is deemed ineffective. The Court today acknowledges that avoiding judicial bias that denies litigants due process is a compelling interest, whether or not a particular measure furthers it effectively.18 Likewise, protecting the integrity of the states’ courts has long been recognized as compelling, and by the same reasoning, that interest cannot be negated simply because a particular measure may not protect it fully. As Justice Kennedy wrote in his concurrence in White:
Here, Minnesota has sought to justify its speech restriction as one necessary to *776maintain the integrity of its judiciary. Nothing in the Court’s opinion should be read to cast doubt on the vital importance of this state interest. Courts, in our system, elaborate principles of law in the course of resolving disputes. The power and the prerogative of a court to perform this function rest, in the end, upon the respect accorded to its judgments. The citizen’s respect for judgments depends in turn upon the issuing court’s absolute probity. Judicial integrity is, in consequence, a state interest of the highest order.
536 U.S. at 793, 122 S.Ct. 2528. It is a misreading of the Supreme Court’s under-inclusiveness discussions, and, most significantly, a nonsequitur as well, to say that the interest in judicial integrity could be reduced to insignificance because Canon 5. does not go far enough to protect it.
F.
Preserving judicial open-mindedness, and the appearance of it, should be recognized as the same compelling state interest in avoiding corruption interest that was identified in Buckley v. Valeo and the campaign finance cases. Though it is the same anti-corruption interest, the need to protect that interest is more urgent and vital in the context of the judiciary because in that context outside influences threaten litigants’ due process interest in adjudication in accord with the law and the facts of their case. A further state interest in preserving the separation of powers between state branches of government should also be recognized as compelling. The Minnesota Supreme Court has recently re-examined Canon 5 and clarified that the Canon is meant to protect those state interests. Judicial integrity and separation of powers are interests of the highest importance in guaranteeing the proper functioning of state government and we have no warrant to deny their importance.
II.
A.
Though the Court today errs in holding that underinclusiveness of a regulation can negate the importance of the state’s interest in the integrity of its judiciary, underinclusiveness does indeed point to a different problem-it raises an inference of pretext. Even where an asserted governmental interest is undeniably compelling, a failure to fully address threats to that compelling interest can be evidence of pretext. The governmental actor may have missed the target because it was not aiming at it, but was actually seeking to accomplish some other, impermissible goal, such as viewpoint discrimination. In such a case, underinclusiveness of a regulation does not cast doubt on whether the asserted interest is compelling, but whether it is genuine. See Johnson v. California, — U.S. —, —, 125 S.Ct. 1141, 1156, 160 L.Ed.2d 949 (2005) (Stevens, J., dissenting) (failure to take measures that would be more effective at addressing prison violence than challenged regulation “undercuts the sincerity” of the state’s concern about violence). This is the same point Justice Scalia made in White when he quoted his earlier words, “[A] law cannot be regarded as protecting an interest of the highest order, and thus as justifying a restriction upon truthful speech, when it leaves appreciable damage to that supposedly vital interest unprohibited,” Republican Party of Minnesota v. White, 536 U.S. 765, 780, 122 S.Ct. 2528, 153 L.Ed.2d 694 (2002) (quoting Florida Star v. B.J.F., 491 U.S. 524, 541-42, 109 S.Ct. 2603, 105 L.Ed.2d 443 (1989) (Scalia, J., concurring in judgment)), and spoke of a “challenge to the credulous.” White, 536 U.S. at 780, 122 S.Ct. 2528.
The Supreme Court has twice upheld speech restrictions on strict scrutiny review where the measure was tailored to *777address only the most critical threat to the governmental interest, even where some threat to the asserted interest remained unaddressed. See Austin v. Michigan Chamber of Commerce, 494 U.S. 652, 110 S.Ct. 1391, 108 L.Ed.2d 652 (1990), and McConnell v. FEC, 540 U.S. 93, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003). In Austin, Michigan restricted the political expenditures of corporations in support of or opposition to candidates for state office. The compelling interest for such regulation was the concern that entities that amassed wealth in the economic marketplace would parlay that wealth into “unfair advantage in the political marketplace,” id. at 659, 110 S.Ct. 1391, because their ability to spend corporate money bore no relation to political support for the ideas the Corporations spent their money to promote, id. at 659-60, 110 S.Ct. 1391 (opinion of Court) and 670, 672 (Brennan, J., concurring). The Chamber of Commerce attacked the Michigan law as underinclusive because it did not regulate expenditures by unincorporated labor unions, which also amassed political war chests. Id. at 665, 110 S.Ct. 1391.
Austin rejected the underinclusiveness challenge, reasoning that the corporations enjoyed greater government-conferred legal advantages enhancing their ability to accumulate wealth. 494 U.S. at 665, 110 S.Ct. 1391. These legal advantages of corporate form made a crucial distinction between corporations and unions. Id. at 666, 110 S.Ct. 1391 (“Michigan’s decision to exclude unincorporated labor unions from the scope of '§ 54(1) is therefore justified by the crucial differences between unions and corporations.”). Additionally, case law permitted union members to opt out of contributing to the union’s political activities, which meant that “the funds available for a union’s political activities more accurately reflects members’ support for the organization’s political views than does a corporation’s general treasury.” Id. The Michigan law therefore passed strict scrutiny.
The Supreme Court also upheld a statute on strict scrutiny review despite an underinclusiveness attack in its recent campaign finance case, McConnell v. FEC. In McConnell, the Court considered a challenge to section 203 of the Bipartisan Campaign Reform Act of 2002, which in turn amended section 316(b)(2) of the Federal Election Campaign Act of 1971, to prohibit use of corporations’ and unions’ treasury funds to pay for certain kinds of election advertising. 540 U.S. at 204 & n. 87, 124 S.Ct. 619. Because the provision - restricted expenditures rather than contributions, it had to be tested by strict scrutiny. Id. at 205, 124 S.Ct. 619, (asking whether “compelling governmental interest justifies” measure); id. at 330, 124 S.Ct. 619 (Kennedy, J., dissenting) (“All parties agree strict scrutiny applies [to section 203].”). The plaintiffs contended that the section was underinclusive because it did not apply to election advertising in the print media or on the Internet. Id. at 207, 124 S.Ct. 619. The Court held that the evidence in the case supported the conclusion that television advertising posed the greater threat, and therefore, “The record amply justifie[d] Congress’ line drawing.” Id. at 208, 124 S.Ct. 619. The Court said that “reform may take one step at a time, addressing itself to the phase of the problem which seems most acute to the legislative mind.” Id. at 207-08, 124 S.Ct. 619 (quoting Buckley v. Valeo, 424 U.S. 1, 105, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976)). McConnell confirms that the sort of underinclusiveness that is fatal in strict scrutiny is arbitrary underinclusiveness, not underinclusiveness that results from attempting to focus the restriction on only the severest form of the threat to a compelling governmental interest.
A content-based restriction was also upheld on strict scrutiny review despite an *778underinclusiveness challenge in Burson v. Freeman, 504 U.S. 191, 207, 112 S.Ct. 1846, 119 L.Ed.2d 5 (1992). Freeman contended that a statute restricting campaign activity within 100 feet of the polls on election day was underinclusive because it did not restrict other types of speech, including commercial solicitation and exit polling. The Court rejected that attack because there was “simply no evidence that political candidates have used other forms of solicitation or exit polling to commit ... electoral abuses.” Id. Thus, whether the statute was impermissibly un-.derinelusive depended on the evidence about the extent and severity of the threat to the state’s asserted interest.
B.
The question at issue in our consideration of the partisan activities clauses, as in Austin, is whether there is a “crucial difference,” 494 U.S. at 666, 110 S.Ct. 1391, in the threat posed by some entities that justified regulating them while leaving others unregulated. To rebut the inference of pretext, the government must show that the speech it has burdened poses a different, more serious threat to its asserted interest than the speech it chose not to regulate. See Erznoznik v. City of Jacksonville, 422 U.S. 205, 215, 95 S.Ct. 2268, 45 L.Ed.2d 125 (1975) (“[E]ven a traffic regulation cannot discriminate on the basis of content unless there are clear reasons for the distinctions.”); Eugene Volokh, Freedom of Speech, Permissible Tailoring and Transcending Strict Scrutiny, 144 U. Penn. L.Rev. 2417, 2423 (1966) (“[A] law is not narrowly tailored if it fails to restrict a significant amount of speech that harms the government interest to about the same degree as does the restricted speech.”) (emphasis added); In re Dunleavy, 838 A.2d 338, 351 (Me.2003) (“Canon 5(A)(1)(e) is narrowly tailored to meet [state’s interest in avoiding bias] because it applies only to conduct which presents the greatest risk , to that interest ....”), cert. denied, 541 U.S. 960, 124 S.Ct. 1722, 158 L.Ed.2d 401 (2004).
Recently, the Supreme Court has held that the differences between political parties and other interest groups could warrant differential regulation of the two kinds of groups. This distinction between political parties and other interest groups was at issue in McConnell, where the Court considered Title I of the Bipartisan Campaign Reform Act, which imposed restrictions on political parties’ fund-raising activities that were not imposed on interest groups, such as the National Rifle Association, the American Civil Liberties' Union or the Sierra Club. The plaintiffs contended that the distinction violated Equal Protection. The Court held the distinction was permissible, because
Congress is fully entitled to consider the real-world differences between political parties and interest groups when crafting a system of campaign finance regulation. Interest groups do not select slates of candidates for elections. Interest groups do not determine who will serve on legislative committees, elect congressional leadership, or organize legislative caucuses. Political parties have influence and power in the legislature that vastly exceeds that of any interest group .... Congress’ efforts at campaign finance regulation may account for these salient differences.
540 U.S. at 188, 124 S.Ct. 619 (citation omitted).19
Before the district court, the Boards contended that special restrictions on judi*779cial candidates’ reliance on political parties were necessary to protect Minnesota’s tradition of non-partisan judicial elections, which dates from the enactment in 1912 of the statute making Minnesota judicial elections non-partisan. See 1912 Minn. Laws, Spec. Sess., eh. 2; Peterson v. Stafford, 490 N.W.2d 418, 422 (Minn.1992) (discussing history of Minnesota’s judicial elections).
The Minnesota Supreme Court greatly amplified that explanation when it decided to reject the Advisory Committee’s proposed revisions to the partisan activities clauses in September 2004. The supreme court order stated, “We conclude that the restrictions on partisan political activity contained in our Code of Judicial Conduct are too important to undermine based on the possibility that they may be vulnerable to constitutional attack, particularly as we are convinced that there are sound bases for their constitutional validity.” In re Amendment of the Code of Judicial Conduct, No. C4-85-697, slip op. at 2-3 (Minn. Sept.14, 2004). The court then reviewed the history of Minnesota’s commitment to non-partisan judicial elections. Id. at 639 N.W.2d 55.
The movement-towards non-partisan judicial elections was a reform movement meant to insulate judges from the- party machines that had captured the state courts during the late nineteenth and early twentieth centuries. F. Andrew Hanssen, Learning About Judicial Independence: Institutional Change in the State Courts, 33 J. Legal Studies 431, 448-50 (2004). Between 1910 and 1958, eighteen states adopted non-partisan judicial elections. Id. at 436. Among states that elect their judges, the majority use nonpartisan elections; currently, twenty states have nonpartisan elections for at least some of their judgeships, as opposed to fifteen who have at least some partisan elections.20 American Judicature Society, Judicial Selection in the States: Appellate and General Jurisdiction Courts (2004); see Suessmann v. Lamone, 383 Md. 697, 862 A.2d 1, 19 (Md.2004) (Maryland circuit court primaries are held to be partisan). Among the states with non-partisan judicial elections, there is a wide variety of measures to enforce the non-partisan character of the election; some states have few such measures, but many have measures similar to those at issue here.21 Thus, the idea that *780non-partisan campaigns might protect the judiciary from improper external pressures is hardly a novel idea, but must be placed within a broad national reform movement that still has significant sway within the states.
The partisan activities clauses at issue here were adopted in 1997 as part of an effort to clarify and formalize Minnesota’s tradition of non-partisan elections and to supplement the guidance given by the nonpartisan election statute. The mere omission of party names from the ballot apparently does little to make campaigns nonpartisan, as shown by the situations in Ohio and Michigan, where ostensibly nonpartisan general elections are preceded by vigorous partisan campaigns. See American Bar Ass’n Commission on the 21st Century Judiciary, Justice in Jeopardy 17 (2003) (“Even states with ostensibly nonpartisan general elections for judges, such as Michigan and Ohio, have experienced highly politicized races when two-party competition is fierce and the party affiliations of the candidates are widely known.”); id. at 77 (“States with true nonpartisan elections, such as those in Wisconsin, must therefore be distinguished from states such as Michigan, where the political parties nominate candidates to run in ostensibly nonpartisan general elections, which in fact have been very partisan indeed.”); Anthony Champagne, Political Parties and Judicial Elections, 34 Loy. L.A.L.Rev. 1411, 1415-16 n. 16 (Michigan, Ohio, and Idaho have non-partisan ballots but partisan campaigns). M. DePaul Wil-lette, Executive Secretary of the Board on Judicial Standards, stated to the Minnesota Supreme Court, “The nonpartisan nature of Minnesota’s judicial elections have been taken for granted. But a search of the statutes and rules showed there is, in fact, very little, if any, specific language describing what nonpartisan elections are.” Comments to the Minnesota Supreme Court, No. C7-81-300 at 2 (Nov. 14, 1997). Willette’s commentary on the proposed amendments to Canon 5, which contained the three partisan activities clauses at issue here, shows that the Board was animated by a desire to effectuate more completely the non-partisan election statute and formalize the tradition of non-partisan elections that the Board believed was not adequately protected by existing Minnesota law. One of the changes made to Canon 5 in 1997 was to include the statement, “Each justice of the supreme court and *781each court of appeals and district court judge is deemed to hold ■ a separate nonpartisan- office. MS 204B.06 Subd. 6.” Canon 5A (1997). The Board on Judicial Standards announced in proposing - the amendments that they were meant to protect Minnesota’s “long tradition of nonpartisan judicial elections.” Comments to the Minnesota Supreme Court, No. C7-81-300 at 4 (Nov. 14,1997).
The hearing the Minnesota Supreme Court held before the 1997 amendments to Canon 5 included consideration of whether partisan activities restrictions should be limited to political parties a? defined in Canon 5 or whether they should apply to other advocacy groups. There was testimony on both sides of that issue. In addition to the testimony of Judge Meyer (which the Court quotes at slip op. 36-37 n. 13) and others against the definition adopted, DePaul Willette testified:
Let’s assume that the rule is not in place and two candidates in a race; one is endorsed by the republican party, one is endorsed by the democratic party. What do we have? We have a party race. It’s not a nonpartisan contest. We have a party contest which will lead us, in my judgment, to the kind of fund-raising and the problems that Illinois and Texas are facing today with multimillion dollar budgets for people who want to retain or gain judicial positions.
Hearing before the Minnesota Supreme Court on Amendment to Canon 5 of the Code of Judicial Conduct, at 20-21 (Nov. 17,1997).
Willette’s testimony also refutes the idea that the Minnesota Supreme Court intentionally failed to address the threat from partisan activity by single-issue interest groups. Willette testified that one reason single-issue interest groups were not 'included in the partisan activities clauses is that single-issue groups would require a commitment that would have been banned under the announce clause at the time. See id. at 12-13. Obviously, the announce clause can no longer play any role in the regulatory scheme; however, the Minnesota Supreme Court’s expectation that the announce clause would serve to moderate a candidate’s relation with interest groups was reasonable at the time and therefore tends to show that the partisan activities clauses were effective at the time adopted. Moreover, the invalidation of the announce clause has apparently had a profound effect on the pressures on judicial candidates in that it is apparently now common for organizations to send judicial candidates questionnaires asking them to state then-positions on an array of disputed legal issues. See, e.g., North Dakota Family Alliance, Inc. v. Bader, 361 F.Supp.2d 1021, 1027 (D.N.D.2005) (example of “voter’s guide” questionnaire submitted to judicial candidates in North Dakota, including items asking candidate to agree or disagree with statements such as: “I believe that the North Dakota Constitution does not recognize a right .to homosexual sexual relationships” and “I believe that the North Dakota Constitution does not recognize a right to abortion.”). In light of the invalidation of the announce clause, I believe a remand for further evidence on the issue of pretext would be more appropriate than for us to order summary judgment on a record with evidence supporting both sides of the question.
Once again, the most pertinent evidence about the thinking behind the current Can-, on 5 is evidence that has not yet been presented to the district court. The Advisory Committee in 2004 reviewed McConnell and other Supreme Court case law and concluded:
In the Advisory Committee’s view, there is ample support for Canon 5’s current limitation of the political activity restrictions to political party activities, while leaving unregulated candidate activities relating to special interest or other *782groups that do not rise to the level of a political party. As noted above, McConnell itself clearly supports the validity of this limitation in order to promote the compelling interests in judicial impartiality, independence, and appearance of impartiality and independence.
Advisory Committee Report, Comments-Canon 5. The Committee reasoned that parties differ from other interest groups in the degree of symbiosis that exists between candidate and party and in the unique role that political parties play in the workings of the other branches of government. Id. Finally, the Committee concluded that recusal was not an adequate remedy for the problems posed by partisan involvement in judicial elections because recusal requests depend on the parties’ ability to know all relevant facts, because recusal involves significant administrative costs, and because it may simply be inadequate to counteract the damage to the reputation of the judiciary from the appearance of institutional partiality. Id. The Committee later noted that a minority within the committee had been concerned that the restrictions were underinclusive in failing to address special interest and other political groups, but that “the Committee as a whole acknowledges that it would be difficult to draft a workable rule to limit involvement by special interest or other political groups for a number of reasons.” Id. at Application of the Code of Judicial Conduct.
McConnell demonstrates that the distinction between political parties and other interest groups could be defended as a valid response to “salient differences” between the kind of threat each sort of organization poses to the state’s interests. In addition to its institutional experience with non-partisan judicial elections since 1912, in 1997 the Minnesota Supreme Court had before it some evidence validating the distinction between political parties and other interest groups, and some challenging that distinction. It resolved that conflict, concluding that political parties posed the greater threat. The conclusion was reaffirmed in 2004 by a committee of lawyers and scholars charged with the task of scrutinizing Canon 5 for constitutional problems, and later by the Minnesota Supreme Court. Our Court errs in concluding as a matter of law that the distinction between political parties and other interest groups is pretextual. The evidence as to this distinction is best considered by the district court on remand.
III.
Our Court’s underinclusiveness analysis goes astray by failing to recognize a compelling interest and by failing to allow the Boards to rebut the inference of pretext. Sections I & II, supra. But the signal failing of the Court’s underinclusiveness analysis is that it envisions a kind of strict scrutiny22 that simply cannot work when *783applied to real cases because it does not take into account the need for limited deference to the state’s attempt to solve the problems that besiege it.
“Deference” is not a word we associate with strict scrutiny review, but there is indeed a place for limited deference, as shown in the recent case of Grutter v. Bollinger, 539 U.S. 306, 328, 123 S.Ct. 2325, 156 L.Ed.2d 304 (2003) (“The Law School’s educational judgment that such diversity is essential to its educational mission is one to which we defer.”). There are three reasons why we should employ some limited deference to the judgment of the state of Minnesota in this case, if after remand, we were satisfied that the judgment was well-supported by cogent evidence and the possibility of pretext had been rebutted.
The Court’s primary reason for striking the partisan activities clauses today is that the provisions are underinclusive. The main thrust of the narrow-tailoring requirement is directly to protect speech rights by avoiding an infringement broader than the need to protect the government’s interest: “The purpose of the test is to ensure that speech is restricted no further than necessary to achieve the goal, for it is important to assure that legitimate speech is not chilled or punished.” Ashcroft v. ACLU, 542 U.S. 656, 124 S.Ct. 2783, 2791, 159 L.Ed.2d 690 (2004). Exacting, de novo review by the courts to assure that the government has chosen the least restrictive alternative directly protects the individual’s speech right. The objection that a measure is underinclusive, on the other hand, cuts in the opposite direction; it being the command of the First Amendment not to abridge the freedom of speech, one is at first surprised to learn that a law can offend the First Amendment because the law does not forbid enough speech. The vice in an underinclusive law is not that the underinclusiveness directly suppresses speech but that it raises a suspicion of pretext-which is just an inference, and which can be rebutted by sufficient evidence. Even in questions subject to strict scrutiny, there simply has to be some room for judgment about how wide to cast the net, and it should be apparent that it is more offensive to the First Amendment for a measure to be too broad than to be too narrow. The problem with applying the same kind of exacting, de novo review to underinclusiveness as we do to overinclusiveness is that the two requirements form a Catch 22 situation, in which a drafter’s very effort to avoid over-inclusiveness makes the measure vulnerable to attack for underinclusiveness.23
In McConnell, even when applying strict scrutiny, the Court acknowledged the need for “legislative choice” and “Congress’ line drawing.” 540 U.S. at 207-08, 124 S.Ct. 619. Thus, when considering the argument that the restrictions on corporations’ and unions’ expenditures for “electioneering communications” were underinclusive because they did not extend to print or internet advertising, the Court quoted Buckley in stating that “reform may take one step at a time, addressing itself to the phase of the problem which seems most acute to the legislative mind.” Id. The *784Court rejected the argument that the restriction on expenditures for electioneering communications was underinclusive because it did not include print or internet ads; in order to respond to the problem at all, Congress had to be able to draw lines somewhere: “One might just as well argue that the electioneering communication definition is underinclusive because it leaves advertising 61 days in advance of an election entirely unregulated.” Id. at 208, 96 S.Ct. 612. The Court considered it obvious that someone had to decide whether the restrictions should take effect 60 days or 61 days in advance of the election; if Congress had to be able to show that the problem was utterly eradicated by the 60-day limit and no part of the problem persisted by allowing the same conduct on the 61st day, the legislation would, of course, have failed.
A second reason for some limited deference is that this is a case of competing constitutional interests, so that whatever protection is afforded First Amendment interests comes at the expense of due process and separation of powers interests. See In re Raab, 100 N.Y.2d 305, 763 N.Y.S.2d 213, 793 N.E.2d 1287, 1292 (N.Y.2003) (“[A] number of competing interests are at stake, almost all of a constitutional magnitude. Not only must the State respect the First Amendment rights of judicial candidates and voters but also it must simultaneously ensure that the judicial system is fair and impartial for all litigants, free of the taint of political bias or corruption, or even the appearance of such bias or corruption.”); Roy Schotland, Myth, Reality, Past and Present, and Judicial Elections, 35 Ind. L.Rev. 659, 665 (2002) (“Referring to constitutional rights [in judicial election context], without even mentioning due process, is stunning shallowness.”). In the order promulgating the September 2004 version of Canon 5, the Minnesota Supreme Court made this very point: “The goal of maintaining an independent and impartial judiciary is not merely a policy choice, but embodies constitutional principles that are properly counterbalanced against the First Amendment interests that are affected by regulation of elections.” In re Amendment of the Code of Judicial Conduct, No. C4-85-697, slip op. at 4 (Minn. Sept.14, 2004). Where “constitutionally protected interests lie on both sides of the legal equation,” Nixon v. Shrink Mo. Gov’t PAC, 528 U.S. 377, 400, 120 S.Ct. 897, 145 L.Ed.2d 886 (2000) (Breyer, J., concurring), adoption of an only partly effective regulation may simply be the result of an unavoidable necessity to accommodate conflicting constitutional mandates.
Finally, this is a case in which the parameters of the evil addressed cannot be outlined with a high degree of precision. The difficulty is that the threat to the governmental interest is not from unambiguously evil conduct, but from behavior that forms part of a continuum with desired behavior-attempts of the citizenry to make their voices heard in their government. The critical and difficult question posed by this case is that the danger to judicial neutrality comes from that sometimes salutary behavior, at the point at which participation in the democratic process becomes undue influence over judicial decisionmaking, preventing a judge from acting as the law’s representative, rather than as the representative of a political patron or donor. That point will vary from candidate to candidate, according to whether he or she is stubborn or persuadable, experienced or naive, young or old, poor or independently wealthy, ambitious or modest. No law can account for all these imponderables without restricting some candidate who would not have been swayed by temptation or leaving some candidate at liberty to compromise himself.
The Supreme Court acknowledged the same problem in the context of political *785contribution and expenditure restrictions. The use of money to influence elections and hence, government policy, is not simply either wholesome activism or influenee-buying-the same action can partake of both, which is why someone has to decide when a contribution or expenditure goes from being civic activism to a bid for undue influence. For that reason, the Supreme Court has said that regulation of political contributions and expenditures can go beyond forbidding outright bribery to regulation of more subtle forms of conduct that pose the same kind of threat, but in a lesser degree'or in a more ambiguous form. This was debated in McConnell, where the dissenters contended that bribery laws ought to be enough to protect the government’s anti-corruption interest, see 540 U.S. at 266, 124 S.Ct. 619 (Thomas, J., concurring and dissenting) and 540 U.S. at 258, 124 S.Ct. 619 (Scalia, J., concurring and dissenting) and 540 U.S. at 292-98, 124 S.Ct. 619 (Kennedy J., concurring and dissenting), but the Court’s opinion held that the anti-corruption interest “extends beyond simple eash-for-votes corruption to curbing ‘undue influence on an officeholder’s judgment,’ ” 540 U.S. at 150, 124 S.Ct. 619. When Congress grapples with such a protean concept as “undue influence on an officeholder,” the Supreme Court applies strict scrutiny in such a way as to acknowledge that Congress’ task requires exercise of some judgment. In contrast to the Supreme.Court’s approach, our Court today takes a bludgeon to a state’s attempt to solve a delicate problem.
IV.
The futility of requiring unattainable precision is illustrated by our Court’s treatment of the solicitation clause. The basic scheme of the solicitation clause is to erect the campaign committees as a barrier between the candidate and contributor. As recently as 2002, all but four of the states that had judicial elections prohibited candidates from personally soliciting campaign contributions. Roy- Schotland, Myth, Reality, Past and Present, and Judicial Elections, 35 Ind. L.Rev. 659, 666 (2002). (Since White, several states have amended their rules.24). The Court today *786seems to implicitly approve the concept of the campaign committee as a barrier between contributors and the judge or would-be judge. Yet, in effectuating the concept, there are necessarily details which could be moved an inch one way or another. It is clear that for the candidate to sign letters himself is one way to hack at the wall between the candidate and contributor-presumably, that is why Wer-sal wants to do it. It is perhaps true that the entire wall would not fall down, but it would be somewhat less effective in achieving the goal of removing personal obligation from the candidate-contributor relation. If each detail of the scheme must be proved as critical, rather than as forming a part of a scheme that works, then each detail, and therefore the scheme as a whole, is foredoomed.
Moreover, while the Court’s ruling today seems to attack only one small aspect of the solicitation-restriction scheme, the ruling contains the seeds to strike the whole scheme. Today Wersal asks only to sign solicitation letters himself and to personally ask for money from large groups. However, the Court states that any candidate can flank the campaign committee’s confidentiality obligation simply by looking up public records showing who contributed to whom. In light of the Court’s underinclu-siveness analysis, this reasoning will likely require us to condemn the entire scheme as soon as the next plaintiff asks us to.
In sum, though strict scrutiny must, of course, be strict, it must, at least in some instances, be applied with limited deference to the decisionmaker’s exercise of judgment. If we pretend that it is otherwise, we adopt a model for strict scrutiny under which no state’s attempt to deal with certain problems can survive, and so very real and dangerous problems must be left unaddressed. Every place where the line is drawn is arguably either overinclu-sive, because too much activism is restricted, or underinclusive, because too much threat to judicial open-mindedness is tolerated. The courts then occupy the enviable position of not being required to say in advance what line would be permissible, but of being privileged to veto every possible legislative attempt to draw the line because it would have "been possible to draw the line somewhere else. If strict scrutiny is simply a way to strike down laws, in which any law is doomed as soon as we invoke strict scrutiny, it is a charade. That is not how the Supreme Court has applied strict scrutiny,25 nor should we adopt this flawed methodology in our Circuit. Instead, where the states or other branches draw the line in a place which the governmental actor can defend, with convincing evidence, as the place where the threat to its interest becomes the most acute, the measure should pass strict scrutiny, though it might have been possible for another hypothetical decisionmaker to have moved the line an inch in one direction or another.
V.
There can be no question that the interests at stake here are compelling. There are questions of fact-first, as to whether the threat to those interests posed by partisan involvement in judicial elections and personal solicitation of contributions are *787severe enough to warrant the measures taken by the Minnesota Supreme Court and ^second, as to whether the particular remedy chosen was truly selected for the asserted reason. I would remand to the district court for trial of these factual questions in light of new evidence of the Minnesota Supreme Court’s most recent deliberations on the subject. If the defendants prove by convincing evidence that the threat was as they assert and that the clauses were adopted to remedy that threat, I believe the clauses should be upheld as constitutional. Today’s ruling invalidates Minnesota’s current attempts to preserve its courts’ integrity and public repute without any evidence having been heard on the most recent rule amendments. At the same time, our ruling in effect dooms any future attempt as well by adopting a form of strict scrutiny that no measure will pass. I therefore respectfully dissent.
. One of the chief differences between the opinion of the Court in White and the dissents was that the dissents considered the announce clause to affect the candidate's relationship with supporters in a significant way, whereas Justice Scalia did not. Compare 536 U.S. at 782, 122 S.Ct. 2528 (Scalia, J.) (judge who changes position stated in campaign no more vulnerable than any other judge who rules in a way disliked by the public) with id. at 800, 122 S.Ct. 2528 (Stevens, J., dissenting) (candidate who announces views is telling electorate, “Vote for me because I believe X and I will judge cases accordingly.”) and id. at 816, 122 S.Ct. 2528 (Ginsburg, J., dissenting) (judge who fails to honor campaign promises may be thought to have betrayed supporters).
. “It can hardly be argued that seeking to uphold a constitutional protection, such as due process, is not per se a compelling state interest.” Op. at 753.
. As Dean Briffault has observed:
The “special relationship” between parties and officeholders that the Court recognized in McConnell can also threaten judicial independence by too closely linking judges to party leaders and the preferences of those *779leaders, even In cases in which neither the party leaders nor the party itself are participating.
Richard Briffault, Judicial Campaign Codes after Republican Party of Minnesota v. White, 153 U. Penn. L.Rev. 181, 232 (2004). This substantiates the threat to the separation of powers interest discussed in section IC, supra.
. Some states have both partisan and nonpartisan elections for different offices; in this dissent they are counted as both partisan and non-partisan election states. Some states have hybrid systems in which candidates are nominated through party primaries or conventions, but appear without party designation on the general election ballot. These hybrid-system states are here counted as partisan election states, following the classification used by the American Judicature Society. See American Judicature Society, Judicial Selection in the States: Appellate and General Jurisdiction Courts (2004).
. Of the twenty states with at least some non-partisan judicial elections, nine have restrictions of some kind on stating party affiliation, and one forbids actual party affiliation. Ark.Code of Jud. Cond., Canon 5A(l)(f) and 5C(l)(a)(iii) (candidate may privately, but not publicly, identify self as affiliated with political party); ' Fla.Code of Jud. Cond., Canon 7C(3) ("The candidate should refrain from commenting on the candidate’s affiliation with any political party .... ”); Ky. Supreme Court Rule 4.300, Canon 5A(2) ("A judge or candidate shall not identify himself or herself as a member of a political party in any form of advertising, or when speaking to a gathering. If not initiated by the judge or candidate for such office, and only in answer to a direct question, the judge or candidate may identify *780himself or herself as a member of a particular political party.”); Minn.Code of Jud. Cond., Canon 5A(l)(a); Miss.Code Ann. § 23-15-973 ("It shall be unlawful for any candidate for any of the offices mentioned in this section to align himself with ... any political faction or any political party at any time during any primary or general election campaign.”); Nev.Code of Jud. Cond., Canon 5C(l)(ii) (candidate may identify party "upon request”); Ore.Code of Jud. Cond., Canon 4-102(C) (judicial candidate shall not knowingly "[p]ub-licly identify the judicial candidate for the purpose of election, as a member of a political party other than by registering to vote or as allowed by ORS 249.015.”); S.D.Code of Jud. Cond., Canon 5C(l)(a)(ii) (judge may identify self as member of political party "for voting purposes only”); Wash.Code of Jud. Cond., Canon 7(A)(1)(e) (judicial candidates shall not "identify themselves as members of a political party, except as necessary to vote in an election.”), enforced In re Kaiser, 111 Wash.2d 275, 759 P.2d 392, 400 (1988); Wis. S.Ct. Rule 60.06(2)(b) (judicial candidates shall not be a member of any political party).
Arkansas and Florida allow speaking at party gatherings only if the candidate’s opponent is also invited to speak. Ark.Code of Jud. Cond., Canon 5 C(l); Fla.Code of Jud. Cond., Canon 7C(3).
Three other states directly or indirectly prohibit a candidate from seeking party endorsement. Ark.Code of Jud. Cond., Canon 5A(l)(d); Idaho Code of Jud. Cond., Canon 5A(l)(d); Miss.Code Ann. § 23-15-973 ("It shall be unlawful for any candidate ... to align himself ... with any political faction or any political party ....").
. Whether strict scrutiny should be applied to the solicitation clause is not entirely free from doubt. In an amicus brief filed in this en banc rehearing, the Conference of Chief Justices argues that the personal solicitation ban regulates contributions to a political candidate and therefore should not be subjected to strict scrutiny, but to the lesser standard appropriate to campaign contribution limitations. Conference of Chief Justices' brief at p. 16. This argument finds support in McConnell v. FEC, 540 U.S. 93, 138-41, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003), which applied the less rigorous scrutiny when re•viewing a law restricting national political parties' ability to solicit contributions. See also Richard Briffault, Judicial Campaign Codes after Republican Party of Minnesota v. White, 153 U. Pa. L.Rev. 181, 225-26 (2004) ("The restriction on personal solicitation by a candidate should be subject to the same less rigorous standard of review as the restriction on contributions."). Because the parties have assumed the same level of scrutiny would apply to the solicitation clause as to the other provisions of Canon 5, neither the panel opinion in Kelly, 247 F.3d 854 (8th Cir.2001), nor *783that on remand, 361 F.3d 1035, 1048-50 (8th Cir.2004), entertained the possibility that the contributions standard should apply. Though the argument is worthy of more attention, I will follow the parties and our Court in assuming the proper standard is strict scrutiny.
. The overinclusive-underinclusive trap is illustrated in the advice given to would-be drafters of revisions to state canons of judicial ethics that general language can be attacked as vague or overbroad, but specific provisions make the canon vulnerable to charges of un-derinclusiveness. See.J.J. Gass, After White: Defending and Amending Canons of Judicial Ethics 23 (Brennan Center for Justice 2004).
. Twenty-four states currently prohibit personal solicitation, while six states allow it. The states that now prohibit personal solicita-ron by the candidate are: Arizona, Ariz. Rule 81, Code of Jud. Cond., Canon 5B(2); Arkansas, Ark.Code of Jud. Cond., Canon 5C(2); Florida, Fla.Code of Jud. Cond., Canon 7C.(1); Idaho, Id.Code of Jud. Cond., Canon 5C(2) (may not solicit "in person”); Illinois, Ill.Code of Jud. Cond., Canon 7B(2); Indiana, Ind.Code of Jud. Cond., Canon 5C(2); Kansas, Kan. S.Ct. Rule 601, Code of Jud. Cond., Canon 7B(2)(c); Kentucky, Ky. Rule 4.300, Code of Jud. Cond., Canon 5B(2); Louisiana, La.Code of Jud. Cond., Canon 7D(1); Michigan, Mich.Code of Jud. Cond., Canon 7B(2); Minnesota, Minn.Code of Jud. Cond., Canon 5B(2); Mississippi, Miss.Code of Jud. Cond., Canon 5C(2); Missouri, Mo. S.Ct. Rule 2.03, Canon 5B(2); New York, N.Y. Judiciary Law, Book 29 App., Canon 5A(5); North Dakota, N.D.Code of Jud. Cond., Canon 5C(2); Ohio, Ohio Code of Jud. Cond., Canon 7(C)(2)(a); Oklahoma, Okla.Code of Jud. Cond., Canon 5C(2); Oregon, Ore.Code of Jud. Cond. R.4-102(D); Pennsylvania, 42 Pa. Cons.Stat. Ann., Code of Jud. Cond., Canon 7(B)(2); South Dakota, S.D.Code of Jud. Cond., Canon 5C(2); Tennessee, Tenn. S.Ct. Rules, Rule 10, Code of Jud. Cond., Canon 5C(2)(a); Washington, Wash.Code of Jud. Cond., Canon 7B(2); West Virginia, W. Va.Code of Jud. Cond., Canon 5C(2); and Wisconsin, Wis. S.Ct. Rule 60.06(4).
States that'permit personal solicitation are: California, Cal.Code of Jud. Ethics, Canon 5 (comment.); Georgia, Ga.Code of Jud. Cond., Canon 7B(2) (amended to comply with Weaver v. Bonner, 309 F.3d 1312 (11th Cir.2002)); Maryland, Md. Rules 16-813, Canon 5 (rescinded effective July 1, 2005); Montana, Mont. Canons of Jud. Ethics; Nevada, Nev. Code of Jud. Cond., Canon 5C(2); New Mexico, N.M.Code of Jud. Cond. R. 21-800B and F (candidates may solicit but not from attorneys or litigants with cases pending before the candidate); North Carolina, N.C.Code of Jud. *786Cond., Canon 7(B)(4) (amended in response to White, see Matthew Eisley, Election Rules Relaxed for Judges: Permission to Solicit Lawyers for Money Brings a Fear of ‘Shakedowns', 26 Nat'l L.J. No. 7 (Oct. 13, 2003)); and Texas, Tex.Code of Jud. Cond., Canon 4D(1). Alabama adopted a rule in 2004 that does not prohibit but "strongly discourages” personal solicitation. Ala. Canons of Jud. Ethics, Canon 7B(4).
. "Strict scrutiny is not 'strict in theory but fatal in fact.' ” Grutter v. Bollinger, 539 U.S. at 326, 123 S.Ct. 2325.