Semple v. Williams

290 F. Supp. 3d 1187

• Court: District Court, D. Colorado
• Decided: February 14, 2018
• Docket: Civil Action No. 17–cv–1007–WJM
• Status: Published

Opinion

William J. Martinez, United States District Judge

ORDER DENYING MOTION TO DISMISS & ORDER TO SHOW CAUSE

William Semple, the Coalition for Colorado Universal Health Care, ColoradoCareYes, and Daniel Hayes (together, "Plaintiffs") bring this action for declaratory and injunctive relief against Wayne W. Williams in his official capacity as Colorado's secretary of state. The Court will refer to Defendant simply as "Colorado" or "the state."

Plaintiffs claim that recent changes to the process by which the Colorado Constitution may be amended violate the First and Fourteenth Amendments to the U.S. Constitution. Currently before the Court is Colorado's Motion to Dismiss Under Fed. R. Civ. P 12(b)(6). (ECF No. 13.) Although, procedurally speaking, the specific question presented by this motion is whether Plaintiffs have pleaded enough facts to state a viable claim for relief, the parties have framed their briefs as if the outcome of the motion will decide the case. That appears to be true-there seems to be no dispute over the relevant facts, and the question is how the law applies to those facts.

Having carefully considered the matter, the Court concludes that Plaintiffs' have demonstrated a Fourteenth Amendment violation to the extent that Colorado's new amendment process requires ballot initiative proponents to gather signatures from districts with widely varying registered voter populations. Thus, part of the new amendment process is constitutionally infirm-it is, however, severable from the remainder of the new requirements.

Because there is no pending cross-motion from Plaintiffs (e.g. , for summary judgment), the Court will order Colorado to show cause why final judgment and a permanent injunction should not enter.

I. BACKGROUND

The Colorado Constitution grants Colorado citizens the power to enact legislation and amend the Constitution by initiative. See Colo. Const. art. V, § 1 (2) ("The first power hereby reserved by the people is the initiative...."). In November 2016, Colorado voters approved "Amendment 71," which altered the initiative process with respect to constitutional amendments (although not with respect to legislation).

Before Amendment 71, one could place a constitutional amendment initiative on the ballot by gathering supporting "signatures by registered electors in an amount equal to at least five percent of the total number of votes cast for all candidates for the office of secretary of state at the previous general election." Id. Amendment 71 did not change this requirement, but instead added another layer:

In order to make it more difficult to amend this constitution, a petition for an initiated constitutional amendment shall be signed by registered electors who *1189reside in each state senate district in Colorado in an amount equal to at least two percent of the total registered electors in the senate district provided that the total number of signatures of registered electors on the petition shall at least equal the number of signatures required by subsection (2) of this section [referring to the pre-existing 5% requirement].

Id. § 1(2.5) ("subsection 2.5"). In other words, any person or group wishing to place a constitutional amendment on the ballot must gather signatures from at least 2% of registered voters in each state senate district and signatures from registered voters in an amount equal to at least 5% of the votes cast for secretary of state in the previous general election.

Amendment 71 also added a supermajority requirement for ultimate approval of the proposed amendment:

In order to make it more difficult to amend this constitution, an initiated constitutional amendment shall not become part of this constitution unless the amendment is approved by at least fifty-five percent of the votes cast thereon; except that this paragraph (b) shall not apply to an initiated constitutional amendment that is limited to repealing, in whole or in part, any provision of this constitution.

Id. § 1(4)(b); see also id. , art. XIX, § 2(1)(b) (adding the same requirement to amendments originating in the state legislature).

II. FACTS

The Court presumes the following facts to be true for purposes of this motion. See Ridge at Red Hawk, LLC v. Schneider , 493 F.3d 1174, 1177 (10th Cir. 2007).

A. Plaintiffs' Interests

Plaintiff Daniel Hayes is a "designated representative" for an initiative proposing an amendment to the Colorado Constitution. (ECF No. 1 ¶ 12.) He does not describe the purpose or subject matter of his proposed amendment. However, his proposal is working its way through Colorado's process for setting the approved title and description. (Id. ) Once that process is complete, Hayes intends to begin collecting signatures. (Id. ¶ 13.) Hayes understands that subsection 2.5 "greatly increases the cost and difficulty of collecting sufficient signatures." (Id. ¶ 14.)

Plaintiff William Semple was the "designated representative" for an unsuccessful initiative on the 2016 Colorado ballot known as "Amendment 69." (Id. ¶ 4.) Plaintiffs Coalition for Colorado Universal Health Care and ColoradoCareYes were entities created to promote Amendment 69. (Id. ¶¶ 5-7.) Amendment 69, had it succeeded, would have created a statewide universal single-payer healthcare program known as "ColoradoCare." (Id. ¶ 5.) These plaintiffs intend to place a similar proposal on the Colorado ballot either in 2018 or 2020. (Id. ¶ 8.) They understand that subsection 2.5 will make it much more difficult and costly to gather the required signatures, as compared to their previous efforts. (Id. ¶ 10.)

B. Colorado's Senate Districts

Colorado's thirty-five senate districts are roughly equal in total population. However,

[t]here is a huge variation in the population of registered voters in the various state senate districts. For example, as of January 1, 2017, district 11 had 86,181 voters, district 25 had 85,051 voters, district 21 had 80,499 voters, and five other districts (1, 12, 13, 29 and 35) had between 91,728 and 96,463 voters. By way of comparison, district 4 had 121,093 voters, district 16 had 119,920 voters, district 18 had 120,222 voters, district 20 *1190had 126,844 voters, and district 23 had 132,222 voters. Thus, district 23 has 51,723 more voters than district 21, and that variance is slightly more than 60%.

(Id. ¶ 40.)

III. EQUAL PROTECTION ANALYSIS

Plaintiffs claim that subsection 2.5 violates both their First Amendment rights of political association and the "one person, one vote" principle safeguarded by the Equal Protection Clause of the Fourteenth Amendment. The Court finds Plaintiffs' Equal Protection arguments dispositive, and therefore does not reach the First Amendment arguments. Nonetheless, to understand the relevant case law, the discussion below necessarily includes some description of potential First Amendment bases for challenging ballot-access restrictions.

A. Supreme Court Guideposts

The Court begins by summarizing relevant Supreme Court authority on Equal Protection as it relates to the right to vote.

1. Reynolds v. Sims (1964): "One Person, One Vote"

The first relevant decision is Reynolds v. Sims , 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964), where the Supreme Court held that the Equal Protection Clause requires apportionment of representatives in state legislatures by population, and does not permit apportionment by geography (e.g. , one state senator per county). Id. at 568, 84 S.Ct. 1362.¹ This is so because drawing legislative districts without accounting for population can have dilutive effects from multiple perspectives. If one district has, say, 100,000 voters and the other has only 10,000 voters, each vote in the larger district has less overall impact on the outcome of a legislative election, even though both districts will be sending a single representative to the legislature. Moreover, if one district has 100,000 total inhabitants (as opposed to voters -a distinction that will become important below) and the other has 10,000 total inhabitants, the smaller district has, in effect, ten times the representation in the legislature, because each representative's vote in the legislature is equal to all other representative's votes. As the Supreme Court put it,

Overweighting and overvaluation of the votes of those living here has the certain effect of dilution and undervaluation of the votes of those living there. The resulting discrimination against those individual voters living in disfavored areas is easily demonstrable mathematically. Their right to vote is simply not the same right to vote as that of those living in a favored part of the State. Two, five, or 10 of them must vote before the effect of their voting is equivalent to that of their favored neighbor. Weighting the votes of citizens differently, by any method or means, merely because of where they happen to reside, hardly seems justifiable.

* * *

...Full and effective participation by all citizens in state government requires, therefore, that each citizen have an equally effective voice in the election of members of his state legislature. Modern *1191and viable state government needs, and the Constitution demands, no less.

* * *

...Diluting the weight of votes because of place of residence impairs basic constitutional rights under the Fourteenth Amendment just as much as invidious discriminations based upon factors such as race or economic status.

Id. at 563, 565, 566, 84 S.Ct. 1362 (citations omitted).²

The Supreme Court's formal holding in Reynolds was as follows:

We hold that, as a basic constitutional standard, the Equal Protection Clause requires that the seats in both houses of a bicameral state legislature must be apportioned on a population basis. Simply stated, an individual's right to vote for state legislators is unconstitutionally impaired when its weight is in a substantial fashion diluted when compared with votes of citizens living in other parts of the State.

Id. at 568. Reynolds thus embodies the ideal of equal voting power that is often referred to by the phrase "one person, one vote"-although that phrase does not actually appear in Reynolds . Cf . Gray v. Sanders , 372 U.S. 368, 381, 83 S.Ct. 801, 9 L.Ed.2d 821 (1963) ("The conception of political equality from the Declaration of Independence, to Lincoln's Gettysburg Address, to the Fifteenth, Seventeenth, and Nineteenth Amendments can mean only one thing-one person, one vote.").

2. Williams v. Rhodes (1968): Introduction of First Amendment Considerations

The Supreme Court began extending "one person, one vote" to ballot-access restrictions in Williams v. Rhodes , 393 U.S. 23, 89 S.Ct. 5, 21 L.Ed.2d 24 (1968), which struck down Ohio statutes that made it "virtually impossible" for third parties "to be placed on the state ballot to choose electors pledged to particular candidates for the Presidency and Vice Presidency of the United States." Id. at 24, 89 S.Ct. 5. Notably, the Court found in that case a blend of First Amendment and Equal Protection concerns: "In the present situation the state laws place burdens on two different, although overlapping, kinds of rights-the [First Amendment] right of individuals to associate for the advancement of political beliefs, and the [Equal Protection] right of qualified voters, regardless of their political persuasion, to cast their votes effectively." Id. at 30, 89 S.Ct. 5.

3. Moore v. Ogilvie (1969): Application to Geography-Based Signature-Gathering Requirements

The first time the Supreme Court applied "one person, one vote" to geography-based signature-gathering requirements was Moore v. Ogilvie , 394 U.S. 814, 89 S.Ct. 1493, 23 L.Ed.2d 1 (1969). At issue was an Illinois statute governing an independent candidate's ability to appear on the ballot. Id. at 815, 89 S.Ct. 1493. The statute required prospective candidates to obtain 25,000 signatures from "qualified voters," including 200 signatures "from each of at least 50 counties." Id. (internal quotation marks omitted). At the time, 93.4% of Illinois's registered voters resided in 49 counties, with the remaining 6.6% spread over 53 counties. Id. at 816, 89 S.Ct. 1493. The Court held that the law *1192violated the Equal Protection Clause (with no First Amendment discussion) because

the electorate in 49 of the counties which contain 93.4% of the registered voters may not form a new political party and place its candidates on the ballot. Yet 25,000 of the remaining 6.6% of registered voters properly distributed among the 53 remaining counties may form a new party to elect candidates to office. This law thus discriminates against the residents of the populous counties of the State in favor of rural sections.

Id. at 819.

4. Jenness v. Fortson (1971): "Modicum of Support"

The Supreme Court soon held, however, that the basic requirement of limiting the ballot to those independent candidates who obtain signatures of a certain percentage of registered voters was constitutionally permissible. See Jenness v. Fortson , 403 U.S. 431, 91 S.Ct. 1970, 29 L.Ed.2d 554 (1971). The Jenness case addressed Georgia's 5% requirement. Id. at 433, 91 S.Ct. 1970. The Supreme Court upheld that requirement against both a First Amendment argument that the 5% requirement "abridge[d] the rights of free speech and association" and against an Equal Protection challenge that the law made impermissible distinctions between party-sponsored candidates and independent candidates (because it was allegedly more difficult to gather the required number of signatures than to win a party primary). Id. at 439-42, 91 S.Ct. 1970. In that context, the Supreme Court announced a "modicum of support" principle:

There is surely an important state interest in requiring some preliminary showing of a significant modicum of support before printing the name of a political organization's candidate on the ballot-the interest, if no other, in avoiding confusion, deception, and even frustration of the democratic process at the general election.

Id. at 442, 91 S.Ct. 1970. The Supreme Court found that this interest justified Georgia's 5% requirement, which was not an unduly high number under the circumstances. Id.

5. Anderson v. Celebrezze (1983): Announcing a Balancing Test

After deciding several other voting-rights cases not relevant here, the Supreme Court synthesized an analytical approach to such cases in Anderson v. Celebrezze , 460 U.S. 780, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983). At issue in Anderson was Ohio's ballot-access requirements for independent presidential candidates, which required submission of a certain number of supporters' signatures by March 20 of the election year. Id. at 782-83, 103 S.Ct. 1564. The district court held that a March 20 deadline was unconstitutional under both the First Amendment (limiting the candidate and his supporters' right to seek political change) and the Equal Protection Clause (because the same deadline did not apply to a political party's nominee). Id. at 783, 103 S.Ct. 1564.

The Supreme Court affirmed the district court's outcome, with emphasis on the First Amendment aspect. The Supreme Court described its "primary concern" as "the tendency of ballot access restrictions to limit the field of candidates from which voters might choose." Id. at 786, 103 S.Ct. 1564 (internal quotation marks omitted). Such restrictions, said the Court, potentially impinge on the First Amendment "freedom to engage in association for the advancement of beliefs and ideas." Id. (internal quotation marks omitted). But, as a practical matter, states must regulate elections "if they are to be fair and honest and if some sort of order, rather than chaos, is to accompany the democratic processes." Id. at 788, 103 S.Ct. 1564 (internal quotation *1193marks omitted). Every ballot-access restriction "inevitably affects-at least to some degree-the individual's right to vote and his right to associate with others for political ends. Nevertheless, the State's important regulatory interests are generally sufficient to justify reasonable, nondiscriminatory restrictions." Id.

Having set forth these competing interests, the Supreme Court described a court's task when facing "[c]onstitutional challenges to specific provisions of a State's election laws." Id. at 789, 103 S.Ct. 1564. The court

must first consider the character and magnitude of the asserted injury to the rights protected by the First and Fourteenth Amendments that the plaintiff seeks to vindicate. It then must identify and evaluate the precise interests put forward by the State as justifications for the burden imposed by its rule. In passing judgment, the Court must not only determine the legitimacy and strength of each of those interests; it also must consider the extent to which those interests make it necessary to burden the plaintiff's rights. Only after weighing all these factors is the reviewing court in a position to decide whether the challenged provision is unconstitutional. The results of this evaluation will not be automatic;...there is no substitute for the hard judgments that must be made.

Id. at 789-90, 103 S.Ct. 1564 (citations and internal quotation marks omitted). This Court will refer to the foregoing as the " Anderson test" or the " Anderson balancing test." Although the Supreme Court in Anderson acknowledged that it was applying this test with emphasis on the plaintiffs' First Amendment interests, it characterized the test as derived from and consistent with its previous Equal Protection cases regarding "one person, one vote." Id. at 786, 103 S.Ct. 1564 n.7.

Having set forth the test, the Court held that Ohio's early deadline for independent candidate qualification imposed substantial burdens on the First Amendment rights of a candidate and his or her supporters, and that those burdens were not outweighed by the state's interests in voter education, treating independent candidates similarly to primary-election candidates, and political stability. Id. at 790-806, 103 S.Ct. 1564.

B. Lower Court Cases Regarding Geography-Based Signature-Gathering Requirements

A number of other courts have addressed ballot-access restrictions similar to those at issue here, i.e. , requirements for a certain number of signatures not only statewide, but within designated geographic subdivisions as well.

1. Cases Striking Down Geographic-Signature Gathering Requirements

Given Moore 's invalidation of Illinois's county-based signature-gathering requirement (see Part III.A.3, above), it is not surprising that lower courts have uniformly struck down geography-based signature-gathering requirements when the relevant geographic subdivision was the county. See ACLU of Nevada v. Lomax , 471 F.3d 1010, 1018-21 (9th Cir. 2006) (striking down Nevada's initiative ballot-access requirement that proponents obtain signatures of at least 10% of eligible voters, including 10% of eligible voters in 13 of Nevada's 17 counties) (" Lomax "); Idaho Coal. United for Bears v. Cenarrusa , 342 F.3d 1073, 1076-79 (9th Cir. 2003) (striking down Idaho's initiative ballot-access requirement that proponents gather signatures of 6% of qualified voters statewide, including 6% of qualified voters in half of Idaho's 44 counties) (" Idaho CUBS "); Blomquist v. Thomson , 739 F.2d 525, 527-28 (10th Cir. 1984) (striking down Wyoming's *1194third-party ballot-access requirement that supporters gather signatures from 8,000 registered voters, "a majority of whom may not reside in the same county"); see also Gallivan v. Walker , 54 P.3d 1069, 1093-97 (Utah 2002) (striking down Utah's initiative ballot-access requirement that proponents gather signatures from registered voters in at least 20 of Utah's 29 counties equaling 10% of all votes cast in that county for governor in the last gubernatorial election, with two justices joining the lead opinion agreeing that this is a federal Equal Protection violation).

Some of these same decisions suggest that the state could remedy the defect by designating legislative districts as the relevant geographic unit. Lomax , 471 F.3d at 1021 ("[A]ssuming that ensuring statewide support of a ballot initiative is a compelling state interest...Nevada could base the 13 Counties Rule on legislative districts...."); Idaho CUBS , 342 F.3d at 1078 ("Idaho could [ensure a 'modicum of statewide support'] through a geographic distribution requirement that does not violate equal protection, for example, by basing any such requirement on existing state legislative districts.").

2. Cases Upholding Geographic-Signature Gathering Requirements

Consistent with this suggestion, courts have uniformly upheld geography-based signature-gathering requirements when the relevant geographic subdivision is a congressional district or state legislative district, given that such districts must (per Supreme Court precedent) be of approximately equal population. See Angle v. Miller , 673 F.3d 1122, 1127-36 (9th Cir. 2012) (upholding Nevada's post- Lomax initiative ballot-access requirement that supporters gather signatures from registered voters in each of the state's congressional districts equal to 10% of the votes cast in the previous general election); Libertarian Party of Virginia v. Davis , 766 F.2d 865, 868 (4th Cir. 1985) (upholding Virginia's third-party ballot-access requirement that supporters gather signatures from 0.5% of all registered voters, including at least 200 voters from each congressional district) (" Davis "); Libertarian Party v. Bond , 764 F.2d 538, 543-45 (8th Cir. 1985) (upholding Missouri's third party ballot-access requirement that supporters gather signatures from 1% of registered voters in each congressional district or 2% of registered voters in half of the congressional districts, as compared to number of votes cast in the previous gubernatorial election) (" Bond "); Udall v. Bowen , 419 F.Supp. 746, 749 (S.D. Ind. 1976) (three-judge panel) (upholding Indiana requirement that those wishing to be on the presidential primary ballot obtain at least 500 signatures from registered voters in each of Indiana's congressional districts), aff'd , 425 U.S. 947, 96 S.Ct. 1720, 48 L.Ed.2d 191 (1976) (mem.).

C. Equal Population vs. Voter Population

Plaintiffs question whether these decisions properly held that districts of roughly equal total population rescue a signature-gathering requirement based on registered voter population from an Equal Protection challenge. (ECF No. 16 at 12 & n.3.) Plaintiffs emphasize that subsection 2.5 requires a percentage of signatures from each senate district's registered voter population, and that the registered voter population varies widely from district to district, sometimes more than 60%. (Id. )

1. Udall and Bond

Of the six above-cited cases holding or suggesting that it is constitutionally permissible to impose a geography-based signature requirement grounded in districts of equal total population, only two of them display any consideration of the possible *1195difference between total population and voter population. The first is the Southern District of Indiana's three-judge decision in Udall , where the court explicitly averaged Indiana's statewide registered voter count across the state's eleven congressional districts, with no inquiry into whether the average generally obtained in each district:

As the Court knows judicially, each of the eleven congressional districts contains approximately 471,000 persons, as per the 1971 redistricting, and that approximately 2,937,000 voters were registered, statewide, for the 1974 election an average of 267,000 per district. Thus to require the signatures of five hundred (500) voters per district amounts to a requirement for slightly over one-tenth of 1% of the persons or slightly less than two-tenths of 1% of the registered voters to sign.

Udall , 419 F.Supp. at 748. In other words, Udall proceeded under an unexamined assumption about the ratio of voting population to total population in the various districts. Here, Plaintiffs have alleged facts (which are probably judicially noticeable in any event) showing significant disparity between registered voter population from senate district to senate district in the Colorado senate. Udall is therefore unhelpful.³

The second case to acknowledge a potential difference between voting population and total population is the Eighth Circuit's Bond decision. Bond addressed Missouri's third-party ballot-access requirement that was not based on a percentage of registered voters' signatures, but which raised a similar problem. Missouri required a third-party candidate's supporters to gather from each of the state's congressional districts signatures equaling at least 1% of the total number of votes cast in that district for governor in the last gubernatorial election. 764 F.2d at 539. Alternatively, supporters could go to only half of the congressional districts if they could gather from those districts signatures equal to at least 2% of the relevant gubernatorial votes. Id.

The plaintiffs argued

that the State's use of a formula based on a percentage of votes cast in each district in the preceding gubernatorial election, rather than a percentage of the population of each district, creates an impermissible discrimination amongst voters. The number of votes cast in each district in the gubernatorial elections are not equal. Thus the number of signatures required from each congressional district under the State's percentage formula varies somewhat, despite the fact that the populations of Missouri's congressional districts are virtually equal.

Id. at 544 (emphasis in original). But the Eighth Circuit had before it the data on the actual number of signatures required per congressional district during the relevant election cycle, ranging from a minimum of 4,266 to a maximum of 5,348. See id. at 540, 544 n.4. The Eighth Circuit deemed this to be a "minimal variance" that "[did] not reflect an impermissible discrimination among voters." Id. at 544.

*1196"In fact," the court continued, "the State's formula measures the number of potential petition signers in each district more accurately than a 'percentage of population' formula would, since the latter formula fails to reflect the fact that not all residents of a district are registered to vote." Id.

The Eighth Circuit did not consider the possibility-likely because the plaintiffs did not raise it-that measuring the interest of registered voters directly (as opposed to through the supposed proxy of votes cast for governor) could itself raise the same problem of variance from district to district, perhaps showing impermissible discrimination. Thus, Bond has nothing to say about that particular problem. Bond , moreover, implicitly affirms Plaintiffs' proposition that a signature-gathering requirement which creates more than "minimal variance" from district to district is voter discrimination.

2. Evenwel

Colorado's primary response to Plaintiffs' voter population disparity theory is that "[t]he Supreme Court recently made clear that states may properly draw their state legislative districts based on total population, rather than the number of voter-eligible persons, without offending the Equal Protection clause's one-person, one-vote principle," citing Evenwel v. Abbott , --- U.S. ----, 136 S.Ct. 1120, 1132-33, 194 L.Ed.2d 291 (2016). (ECF No. 13 at 6.) Colorado correctly describes the Evenwel decision, but Evenwel ultimately provides no support to Colorado's position.

The Evenwel lawsuit exposed a problem lurking in the phrase "one person, one vote," namely, although every person counts when drawing legislative districts, not every person is both qualified and registered to vote. Emphasizing this disconnect, the Evenwel plaintiffs sued the state of Texas, claiming that drawing state legislative districts "on the basis of total population...produces unequal districts when measured by voter-eligible population." 136 S.Ct. at 1123. The plaintiffs urged that such districts must be drawn based on voter-eligible population "to ensure that their votes will not be devalued in relation to citizens' votes in other districts." Id.

The Supreme Court ruled against the plaintiffs, but, notably, it never disagreed with their basic premise that a disparity in voter population among legislative districts dilutes the voting power of eligible voters in voter-rich districts as compared to districts with a lower ratio of voting-eligible population to total population. This, of course, is undeniable, and it is precisely the problem the Supreme Court thought it was addressing in the original "one person, one vote" cases such as Reynolds : "Their right to vote is simply not the same right to vote as that of those living in a favored part of the State. Two, five, or 10 of them must vote before the effect of their voting is equivalent to that of their favored neighbor." 377 U.S. at 563, 84 S.Ct. 1362. But Evenwel forefronted the potential non sequitur between the problem (vote dilution) and the Supreme Court's long-prescribed solution (redistricting based on total population).

Because the Supreme Court could not deny that the Evenwel plaintiffs alleged a classic vote dilution problem, the court fell back on "constitutional history, [its own prior] decisions, and long-standing practice" to reject their claim. 136 S.Ct. at 1123. Given these sources of authority, the Court held that drawing districts based on total population "complies with the requirements of the one-person, one-vote principle." Id. at 1132. The Court chose not to address the United States's contention (as amicus curiae ) "that reapportionment by total population is the only permissible *1197standard," id. at 1141 (Thomas, J., concurring in judgment); see also id. at 1143 (Alito, J., concurring in judgment), or Texas's argument that reapportionment based on voter-eligible population would be permissible, even if Texas does not currently do it, id. at 1133.

Evenwel nonetheless acknowledges the tension between total population and voter population when discussing the "one person, one vote" principle: "For every sentence [the plaintiffs quoted from previous 'one person, one vote' opinions regarding dilution of actual voting power], one could respond with a line casting the one-person, one-vote guarantee in terms of equality of representation, not voter equality." Id. at 1131. The Court went on to say that its prior decisions had "suggested, repeatedly, that districting based on total population serves both the State's interest in preventing vote dilution and its interest in ensuring equality of representation," id. (emphasis in original), but the Court did not explain how these "suggestions" could be accurate, empirically speaking.

Regardless, this is where the inapplicability of Evenwel to the present dispute becomes most apparent. In Evenwel , as in nearly every previous "one person, one vote" case, there were two potentially competing interests involved: (1) "preventing vote dilution" and (2) "ensuring equality of representation." Id. (emphasis added). Avoiding vote dilution, "demonstrable mathematically," is supposedly the hallmark of "one person, one vote." Reynolds , 377 U.S. at 563, 84 S.Ct. 1362. But there is also a deeply rooted constitutional commitment to the idea that elected representatives represent all people within their legislative districts, not just those who have the power to put them into or remove them from office (i.e. , eligible voters). Evenwel , 136 S.Ct. at 1127-30. The fact that those two interests cannot always be reconciled is the basic problem with which Evenwel struggled. The Supreme Court chose to resolve the problem on the narrowest ground possible, namely, Texas had not violated the Equal Protection Clause by favoring equality of representation over equality of voting power. Id. at 1132-33.

In the context of direct democracy, however, the tension between preventing vote dilution and ensuring equality of representation falls away because, with no "representation" in the ballot petition form of direct democratic rule, there is no representative equality component of the equation to balance against the integrity of the vote. In other words, there is no representation; there is only voting. To be sure, in common speech we are accustomed to referring to an election outcome as "the will of the people," even though it is strictly speaking only the will of the voters. But "the will of the people" is meant as an expression of commitment to the democratic process-that we agree to abide by the outcome of an election. It is not meant as an expression that each voter has a duty to account for the interests of the general population within his or her voting district. One who votes in favor of a candidate or proposition surely does not represent anyone else in the same district (voter or non-voter) who opposes the candidate or proposition. A signatory to a ballot petition initiative surely does not represent anyone else in the same district who refused to sign the petition, much less any person who never learned about it in the first place.

There is a social assumption that parents-to the extent they are voters- represent the interests of their minor children at the ballot box, and in some districts it may be that minor children comprise the majority of the nonvoting population. But it is easy to imagine a district where many nonvoters are ineligible to vote because they are noncitizens or have been convicted *1198of a felony, and it is equally (and unfortunately) easy to imagine a resulting wide gulf between the political preferences of the voting and nonvoting populations in such a district.

In sum, the Court finds that Evenwel 's endorsement of legislative districts of roughly equal total population does not answer the question of whether a direct democracy mechanism violates the Equal Protection Clause when it calls for a percentage of registered voters' signatures from geographic districts where there is a significant variation of registered voter population in those districts.

3. The Anderson Balancing Test: "Character and Magnitude of the Asserted Injury"

Colorado contends that "[i]f Evenwell 's [sic ] logic is sufficient to protect the sacrosanct right to vote against unlawful vote dilution, it is equally sufficient to protect the lesser state-created right of initiative." (ECF No. 17 at 2.) This raises a number of questions about whether "one person, one vote" applies with equal force in the context of petition signatures (as compared to actual votes), or in the context of petition signatures for ballot initiatives (as compared to signatures for candidates). These questions fall within the first part of the Anderson balancing test, i.e. , assessment of "the character and magnitude of the asserted injury." 460 U.S. at 789, 103 S.Ct. 1564 ; see also Blomquist , 739 F.2d at 527.

Moore forecloses any argument that signature-gathering provisions cannot run afoul of the vote dilution problem simply because petition signatures are not "votes" in the traditional sense of that word. See 394 U.S. at 818, 89 S.Ct. 1493 ("The use of nominating petitions by independents to obtain a place on the Illinois ballot is an integral part of her elective system. All procedures used by a State as an integral part of the election process must pass muster against the charges of discrimination or of abridgment of the right to vote." (citation omitted) ). The Moore majority came to this conclusion over a dissent from Justice Stewart on that point, among others. See id. at 819, 89 S.Ct. 1493 ("I cannot join in the Court's casual extension of the 'one voter, one vote' slogan to a case that involves neither voters, votes, nor even an ongoing dispute.").

But Moore was about gathering signatures to place a candidate on the ballot. Arguably the right to vote for state representatives is a federal constitutional right under the "Guarantee Clause." See U.S. Const. art. IV, § 4 ("[t]he United States shall guarantee to every State in this Union a Republican Form of Government"); The Federalist No. 43, at 271 (J. Madison) (C. Rossiter ed., 1961) (explaining the Guarantee Clause: "In a confederacy founded on republican principles, and composed of republican members, the superintending government ought clearly to possess authority to defend the system against aristocratic or monarchical innovations."). There is no corresponding federal constitutional guarantee of direct democracy procedures such as voter-initiated legislation. Grant v. Meyer , 828 F.2d 1446, 1455 (10th Cir. 1987) (en banc), aff'd , 486 U.S. 414, 108 S.Ct. 1886, 100 L.Ed.2d 425 (1988). Accordingly, should signatures in favor of placing an initiative on the ballot receive the same protection as signatures in favor of placing a candidate on the ballot?

In the context of First Amendment challenges to signature-gathering requirements (e.g. , that certain requirements inhibit the right of political association), some courts have held that signature-gathering does not receive as much protection as voting itself. See Taxpayers United for Assessment Cuts v. Austin , 994 F.2d 291, 296- 97 (6th Cir. 1993) ;

*1199Gibson v. Firestone , 741 F.2d 1268, 1273 & n.8 (11th Cir. 1984). The Tenth Circuit, however, has spoken in strong language suggesting otherwise. See Grant , 828 F.2d at 1455-56 ("[I]t is said that the Colorado statute's interference with First Amendment rights is minimal since the Constitution does not require states to provide their citizens with an initiative procedure. We disagree. * * * [W]e do not think that Colorado's constitutional choice to reserve the initiative for the people leaves the State free to condition its use by impermissible restraints on First Amendment activity."). But again, these cases involve the First Amendment implications of signature-gathering requirements for ballot measures. In other words, the "the character...of the asserted injury," Anderson , 460 U.S. at 789, 103 S.Ct. 1564, and potentially its "magnitude," id. , is different from a vote-dilution injury under the Equal Protection Clause.⁴

As for the difference between candidate signatures and initiative signatures in the Equal Protection context, the Court is aware of only two cases-both from the Ninth Circuit-making any explicit comment on the subject. In Idaho CUBS , the Ninth Circuit declared that "[n]ominating petitions for candidates and for initiatives both implicate the fundamental right to vote, for the same reasons and in the same manner, and the burdens on both are subject to the same analysis under the Equal Protection Clause." 342 F.3d at 1077. The Ninth Circuit's later Angle decision, however, casts some indirect doubt on this pronouncement. Understanding how Angle may have limited Idaho CUBS requires a certain amount of detail regarding Angle 's approach to the arguments before it.⁵

Angle was a challenge to Nevada's requirement that those wishing to place an initiative on the ballot gather "signatures from a number of registered voters equal to 10 percent of the votes cast in the previous general election...in each of the state's congressional districts," a.k.a. the "All Districts Rule." 673 F.3d at 1126-27. In its Equal Protection analysis, Angle first concluded that

the All Districts Rule grants equal political power to congressional districts having equal populations. It thus does not trigger strict scrutiny under the principle announced in Moore , and it survives rational basis review because it serves the state's legitimate interest in ensuring a minimum of statewide support for an initiative as a prerequisite to placement on the ballot.

Id. at 1129. This analysis displays the very assumption Plaintiffs challenge here, i.e. , that equal total population among districts means equal political power among districts, regardless of voter population. Apparently the plaintiffs in Angle did not assert a voter-population argument. Regardless, Angle 's reasoning is clear: because each congressional district had equal political power, there was no voter discrimination based on geography, and so whatever discrimination might nonetheless exist need only satisfy rational basis review. Moreover, said Angle , ensuring a statewide modicum of interest in ballot initiatives was a legitimate state interest sufficient for rational basis review.

Angle then addressed a further argument from the plaintiffs based on "another *1200set of Supreme Court cases" (i.e. , cases other than Moore and similar decisions). Id. at 1129. The plaintiffs specifically cited the court to Gray v. Sanders , 372 U.S. 368, 83 S.Ct. 801, 9 L.Ed.2d 821 (1963) and Gordon v. Lance , 403 U.S. 1, 91 S.Ct. 1889, 29 L.Ed.2d 273 (1971), which established that statewide elections based on systems similar to the Electoral College violate the Equal Protection Clause. See Angle , 673 F.3d at 1129-30. According to the plaintiffs, those cases "suggest[ed] that, with respect to a statewide election, equal protection requires votes to be counted on a statewide, rather than a district-by-district, basis." Id. at 1129. The plaintiffs' point was that "a ballot initiative may obtain the total number of signatures required statewide, but fail to qualify for the ballot solely based on where signers live," which seems to discriminate based on residence in violation of Gray and Gordon , as well as Reynolds . Id. at 1130.

The Ninth Circuit agreed that Gray , Gordon , and Reynolds "suggest[ed]" at least that "a district-by-district system of counting votes in a statewide election would violate equal protection, [but] none of the decisions suggests that district-by-district counting of signatures obtained to qualify an initiative for the ballot presents the same problem." Id. at 1130 (emphasis in original). Citing Idaho CUBS -although not for the precise quotation, above, about the equivalence between petitions for candidates and petitions for signatures- Angle reasoned that Equal Protection guarantees apply both to votes and signatures "as a general matter." Id. However, they

serve different purposes. A ballot access requirement determines whether there is a minimum level of grassroots support for an initiative to warrant its inclusion on the ballot. An election, by contrast, measures the collective, aggregate will of the electorate. These differences suggest that the bar on district-by-district counting apparently embodied in Gray , Gordon and Reynolds does not apply to the counting of petition signatures to qualify initiatives for the ballot.

Id.

The Ninth Circuit does not go on to explain why the "differences suggest" that district-by-district counting is permissible for ballot signatures as compared to votes, but this Court need not address that question. The import of Angle to the current discussion lies elsewhere. To begin, Angle does not state that there is a difference between signature-gathering for candidates and signature-gathering for initiatives. Rather, Angle claims there is a difference between signature-gathering for initiatives and actual voting. How far this principle goes, assuming it is correct, is unclear. But more importantly, Angle settled on this principle only after previously concluding that no vote dilution was at stake . See id. at 1129. All of Nevada's congressional districts, said Angle , had "equal political power," id. , and so the court treated the plaintiffs' argument as one asserting pure geographic discrimination despite equal voting power.

We cannot know what Angle would have done had it found that voter population substantially differed from district to district, and had it accepted that voter population was the relevant metric.⁶ Under Moore , that is "vote dilution," even though the state is counting signatures rather than marks on a ballot. This Court is unaware of any authority on which the Ninth Circuit could have drawn to classify such acknowledged dilution as deserving of less protection. To the Court's knowledge, *1201there has only been one case since the beginning of the "one person, one vote" era that has stared mathematically significant vote dilution square in the face and chosen not to provide a remedy. That case is Evenwel . And, as explained (Part III.C.2), the only way Evenwel could reach that conclusion on a sound, principled basis was by emphasizing the long-cherished competing value of representational equality. Again, no such competing value exists in a direct democracy context.

All that said, perhaps it is still true that signatures in favor of a ballot initiative simply deserve less protection than signatures in favor of a candidate, or actual votes. If so, Colorado has not explained why, other than dismissing such signatures as part of "lesser state-created right of initiative." (ECF No. 17 at 2.)⁷ Judging from the authorities Colorado propounds, no court, much less the Supreme Court or Tenth Circuit, has ever suggested that the signature-based "voting" rights associated with a state-created ballot-access procedure deserve lesser solicitude and protection under the Equal Protection Clause if mathematically significant dilution is, in fact, occurring. Cf . Lemons v. Bradbury , 538 F.3d 1098, 1102-04 (9th Cir. 2008) (emphasizing Moore 's applicability to state ballot-initiative procedures and choosing not to apply strict scrutiny only after finding that the challenged procedure had no dilutive effect based on a voters' residence); Green v. City of Tucson , 340 F.3d 891, 899-900 (9th Cir. 2003) (same); see also Pub. Integrity All., Inc. v. City of Tucson , 836 F.3d 1019, 1027 & n.4 (9th Cir. 2016) (in the context of procedures for city council elections, find that strict scrutiny was not required because "no geographically based vote dilution allegation is before us"), cert. denied , --- U.S. ----, 137 S.Ct. 1331, 197 L.Ed.2d 518 (2017).

Compounding this weakness in the state's argument is the fact that Colorado nowhere articulates a principled explanation for why voter dilution should be tolerated to a greater degree when it arises in the context of petition signatures. Given this lack both of authority and argument why dilution should be considered more tolerable as to ballot-initiative signatures-"an integral part of [Colorado's] elective system," Moore , 394 U.S. at 818, 89 S.Ct. 1493 -the Court holds that the "character and magnitude" of the injury Plaintiffs allege here, Anderson , 460 U.S. at 789, 103 S.Ct. 1564, is not, from a constitutional perspective, any different than the electoral injuries at issue in Moore and Reynolds . See also Harper v. Va. State Bd. of Elections , 383 U.S. 663, 665, 86 S.Ct. 1079, 16 L.Ed.2d 169 (1966) ("once the franchise is granted to the electorate, lines may not be drawn which are inconsistent with the Equal Protection Clause of the Fourteenth Amendment").

Consider, for example, Colorado's senate districts 21 and 23, which have 80,499 voters and 132,222 voters, respectively. (ECF No. 1 ¶ 40.) Under subsection 2.5, it takes only 1,610 signatures to meet the 2% threshold in district 21, whereas it requires 2,644 signatures in district 23. Thus, each registered voter in district 23 has only about 60% of the ability to influence the outcome of a signature-gathering drive as compared to each registered voter in district 21. Cf . Idaho CUBS , 342 F.3d at 1078 ("Here, in the smallest county a 'vote' may count where 61 others sign, whereas in the largest county it may require up to 18,054 other signatures before the individual's 'vote' will count.").

*1202Or, from a somewhat different perspective, one could characterize subsection 2.5 as granting to each legislative district one "vote" in favor of or against placing a proposed initiative on the ballot. That vote is "yea" if 2% or more of the district's registered voters sign the petition, and otherwise "nay." District 21 needs only 1,610 signatures to cast a "yea" vote, whereas district 23 needs 2,644 signatures-yet each district casts, or may withhold, one equally weighted vote.

In sum, to the extent that the registered voter population varies significantly within Colorado's senate districts, subsection 2.5 creates a classic vote-dilution problem, demanding strict scrutiny under the Equal Protection Clause.

4. The Anderson Balancing Test: "The Precise Interests Put Forward by the State as Justifications for the Burden Imposed"

The Court must now examine Colorado's interests in setting up a system that requires a percentage of signatures from districts where the relevant population is unequal. See Anderson , 460 U.S. at 789, 103 S.Ct. 1564 ; Blomquist , 739 F.2d at 528.

The Equal Protection portions of Colorado's briefs (as opposed to the First Amendment portions) do not contain any argument in this regard. (See ECF Nos. 13 at 4-6; ECF No. 17 at 1-3.) Colorado instead argues that there is no Equal Protection problem at all, relying on the decisions cited above that sustain requirements similar to subsection 2.5. As already discussed, none of those decisions seriously grapples with the problem of substantially differing voter population from district to district. Most of them simply assume that districts of roughly equal total population solve any vote dilution problem. As Evenwel highlights, this assumption is a non sequitur absent a showing that the ratio of registered voters to total population is approximately the same from district to district.

Colorado's reliance on case law that ignores or avoids the issue presented here leaves the state with no argument that it has an interest compelling enough to outweigh registered voters' right not to have the value of their petition signatures diluted. Moreover, to the extent Colorado might assert that subsection 2.5 serves the interest of ensuring statewide support for ballot measures, the Supreme Court and the Tenth Circuit have already characterized such an interest as insufficiently compelling to justify infringement on the political rights guaranteed by the Equal Protection Clause. See Moore , 394 U.S. at 818-19, 89 S.Ct. 1493 ("It is no answer to the argument under the Equal Protection Clause that this law was designed to require statewide support for launching a new political party rather than support from a few localities. This law applies a rigid, arbitrary formula to sparsely settled counties and populous counties alike, contrary to the constitutional theme of equality among citizens in the exercise of their political rights."); Blomquist , 739 F.2d at 528 (citing Moore and announcing, "We are not persuaded that the State has a compelling interest in requiring that supporters of a new political party be scattered across the state.").⁸

*1203* * *

In short, to the extent that there exists a material difference in the registered voter population from senate district to senate district, subsection 2.5 violates the Equal Protection Clause.⁹

IV. SEVERABILITY ANALYSIS

Colorado argues that subsection 2.5 is severable from the remainder of Amendment 71-the remainder being the supermajority (55%) requirement now codified in the Colorado Constitution at article V, § 1 (4)(b) and article XIX, § 2 (1)(b). (ECF No. 13 at 12-13.) Colorado correctly points out that Plaintiffs' complaint contains no allegation that the supermajority requirement itself violates the U.S. Constitution. (Id. ) Plaintiffs instead assert that Amendment 71 must be treated as an inseparable whole, meaning that the supermajority requirement must fall if subsection 2.5 falls. (ECF No. 16 at 14-15.)

Whether a state statutory or constitutional provision is severable "is of course a matter of state law." Leavitt v. Jane L. , 518 U.S. 137, 139, 116 S.Ct. 2068, 135 L.Ed.2d 443 (1996) (per curiam). Plaintiffs' only argument that Amendment 71 must stand or fall as a package relies on the Colorado Constitution's mandate that "[n]o measure shall be proposed by petition containing more than one subject." Colo. Const. art. V, § 1 (5.5). But the Colorado Constitution places a similar requirement on the legislature's enactments, see id . § 21 ("No bill, except general appropriation bills, shall be passed containing more than one subject...."), yet Colorado has a robust law of severability. Colorado presumes statutes to be severable, see Colo. Rev. Stat. § 2-4-204, and applies this assumption to portions of statutes much more closely related to each other than the various portions of Amendment 71, see, e.g. , Rodriguez v. Schutt , 914 P.2d 921, 929 (Colo. 1996).

Plaintiffs have offered no reason why Colorado would treat provisions of its own constitution differently. Indeed, Plaintiffs have cited no case law establishing that the single-subject rule has any bearing whatsoever on severability. The Court accordingly holds that subsection 2.5 is severable from the remainder of Amendment 71.

V. FURTHER PROCEEDINGS

On the arguments presented by the parties and assuming the Plaintiffs' allegations to be true, the Court has determined that subsection 2.5 violates the Fourteenth Amendment's Equal Protection Clause. But if Colorado has a good faith basis for believing it can develop empirical data showing that vote dilution is not actually occurring as between the various state senate districts, the Court will not foreclose that opportunity. The Court's order to show cause (below) will give Colorado an opportunity to request such discovery, or to state any other reason why it would *1204be premature to enter a permanent injunction and final judgment.

VI. CONCLUSION

For the reasons set forth above, the Court ORDERS as follows:

1. Colorado's Motion to Dismiss (ECF No. 13) is DENIED;

2. Colorado is ORDERED TO SHOW CAUSE, on or before March 9, 2018 , why the Court should not enter final judgment against it and a permanent injunction against enforcing subsection 2.5 to the extent there exists a material difference in voter population between state senate districts. In its response to this order to show cause, Colorado shall set forth any dates the Court should be aware of (including relevant past and future deadlines) with respect to the 2018 election cycle as it relates to the ballot initiative process; and

3. Plaintiffs may, but are not required to, file a reply to Colorado's response to the Court's order to show cause no later than March 16, 2018 .

1 A few months before Reynolds , the Supreme Court had reached a similar conclusion with respect to federal congressional districts, although with emphasis on the U.S. Constitution's structural requirements for the House of Representatives. See Wesberry v. Sanders , 376 U.S. 1, 7-8, 84 S.Ct. 526, 11 L.Ed.2d 481 (1964) ("We hold that, construed in its historical context, the command of Art. I, [§] 2, that Representatives be chosen 'by the People of the several States' means that as nearly as is practicable one man's vote in a congressional election is to be worth as much as another's." (footnote omitted) ).

2 Obviously, the fact that every state gets two senators and at least one representative in Congress, regardless of population, creates precisely this sort of dilution. The Supreme Court dismissed this as a "compromise...[a]rising from unique historical circumstances," and not intended as an endorsement of similar arrangements for state legislatures. Id. at 574, 84 S.Ct. 1362.

3 The Supreme Court's affirmance of Udall by memorandum disposition, see 425 U.S. 947, 96 S.Ct. 1720, 48 L.Ed.2d 191 (1976), holds no weight in the present circumstances. "[T]he precedential effect of a summary affirmance can extend no farther than the precise issues presented and necessarily decided by those actions." Ill. State Bd. of Elections v. Socialist Workers Party , 440 U.S. 173, 182, 99 S.Ct. 983, 59 L.Ed.2d 230 (1979) (internal quotation marks omitted). The "precise issue[ ] presented" in Udall was the constitutionality of a geography-based signature-gathering requirement under the assumption that the ratio of voting population to total population remained constant across legislative districts. Plaintiffs have pleaded facts undermining any such assumption in Colorado.

4 A First Amendment challenge generally focuses on restrictions that affect a proponent's ability to distribute his or her message in the process of seeking signatures, see, e.g. , Grant , 828 F.2d at 1452-55, or (less successfully) on the potential chilling effect created by the difficulty of the ballot-qualification procedure or a supermajority adoption standard, see, e.g. , Initiative & Referendum Inst. v. Walker , 450 F.3d 1082, 1099-1105 (10th Cir. 2006) (en banc ).

5 Angle also deserves extended discussion because, although not controlling, Colorado relies on it heavily in its briefing. (See ECF No. 13 at 2, 5; ECF No. 17 at 2-3, 6-7.)

6 Given that Nevada's signature-gathering requirement measured the 10% threshold based on votes cast in the previous general election, there might also have been an argument- as in Bond -that there existed an allegedly significant votes-cast disparity between districts.

7 The Court suspects that Colorado is actually encouraging backwards reasoning, where the Court first decides that ensuring geographically distributed support for ballot measures is a worthy goal, and then the Court looks for a reasonable-sounding way to devalue the right to vote so that the state's goal is not thwarted.

8 Again, Jenness held that ensuring a "modicum of support" was a valid state interest (see Part III.A.4, above), but the question is whether a state may insist on a modicum of statewide support when the process used to gauge that support dilutes the value of certain voters' signatures based on where they live. Because the Court concludes that the answer is "no," the Court need not address Plaintiffs' argument that ensuring geographically distributed support is not a valid state interest at all, even assuming that each senate district contains about the same number of registered voters. (See ECF No. 1 ¶ 38; ECF No. 16 at 5-6.)

9 Colorado informs the Court that "[a]t least nine other states have geographic distribution requirements" similar to Colorado's. (ECF No. 13 at 6 & n.4 (citing National Conference of State Legislatures, Signature Requirements for Initiative Proposals (July 2014), available at http://www.ncsl.org/Portals/1/Documents/Elections/2014_Sig_Reqs.pdf).) This Court's ruling naturally does not control as to other states' ballot-access requirements, but the Court understands that this decision may cast doubt on them. Even so, the fact that this Court may be the first in the nation to analyze the issue of voter dilution from the perspective of registered voters vs. total population is no reason not to resolve the present case, or to defer to Colorado simply because it can point to nine sister states that potentially dilute the value of registered voters' signatures in the same manner.