PAUL BARBADORO, District Judge.

William E. Christie, Shaheen & Gordon, Gilles R. Bissonnette, American Civil Liberties Union of New Hampshire, Concord, NH, for Leon H. Rideout.

Stephen G. Labonte, Anne M. Edwards, NH Attorney General's Office, Concord, NH, for William M. Gardner.

MEMORANDUM AND ORDER

PAUL BARBADORO, District Judge.

New Hampshire recently adopted a law that makes it unlawful for voters to take and disclose digital or photographic copies of their completed ballots in an effort to let others know how they have voted. Three voters, who are under investigation because they posted images of their ballots on social media sites, have challenged the new law on First Amendment grounds. As I explain in this Memorandum and Order, the new law is invalid because it is a content-based restriction on speech that cannot survive strict scrutiny.

I. BACKGROUND

It has been unlawful since at least 1979 for a New Hampshire voter to show his ballot to someone else with an intention to disclose how he plans to vote. See N.H.Rev.Stat. Ann. § 659:35, I (2008). In 2014, the legislature amended section 659:35, I of the New Hampshire Revised Statutes (“RSA 659:35, I”) to provide that:

No voter shall allow his
or her
ballot to be seen by any person with the intention of letting it be known how he

or she

is about to vote

or how he or she has voted

except as provided in RSA 659:20.

This prohibition shall include taking a digital image or photograph of his or her marked ballot and distributing or sharing the image via social media or by any other means.

N.H.Rev.Stat. Ann. § 659:35, I (Supp.2014) (emphasis added to identify the modifications that became effective September 1, 2014). At the same time, the legislature reduced the penalty for a violation of RSA 659:35, I from a misdemeanor to a violation. 2014 N.H. Legis. Serv. 80 (codified as amended at N.H.Rev.Stat. Ann. § 659:35, IV). Thus, anyone who violates the new law faces a possible fine of up to $1,000 for each violation. N.H.Rev.Stat. Ann. § 651:2, IV(a) (establishing maximum penalty for a violation).

A. Legislative History

State Representative Timothy Horrigan introduced a bill to amend RSA 659:35, I on January 3, 2013. See Exhibit G to the Declaration of Gilles Bissonnette, Esq. in Support of Plaintiffs' Motion for Summary Judgment (“Legislative History”) at 000048, 000140, Rideout v. Gardner, No. 14–cv–489–PB (filed Mar. 27, 2015). As initially proposed, the bill simply stated that “[n]o voter shall take a photograph or a digital image of his or her marked ballot.” Id. at 000144. In testimony in favor of the bill, Representative Horrigan explained why he was proposing his amendment:

Last fall, in late October 2012, one of the workers at my local Democratic campaign office received her absentee ballot. After she filled it out, she was about to have a photo of her ballot taken to be posted to her social media accounts. We began to worry taking such a photo might be a violation of federal and state election laws. It turns out that this may not necessarily have been a violation of the letter of the law—but it would definitely be a violation of the spirit of RSA 659:35 “Showing or Specially Marking a Ballot.”

Id. at 000142. He also stated, “The main reason this bill is necessary is to prevent situations where a voter could be coerced into posting proof that he or she voted a particular way.” Id.

The bill first went to the House Committee on Election Law (the “Election Committee”), which recommended its passage with only a slight organizational change and the requirement that posters be placed in polling places informing voters of the new law. See Legislative History at 000110, 000114. Members of the Election Committee noted that “showing your ballot on social media could cause und[ue] influence from employers or parents” and that the bill “protects privacy of voter[s] and stops coercion.” Id. at 000130. Representative Mary Till wrote the statement of intent for the Election Committee, noting, “RSA 659:35 was put in place to protect voters from being intimidated or coerced into proving they voted a particular way by showing their completed ballot or an image of their completed ballot.” Id. at 000114.

The bill was then referred to the House Committee on Criminal Justice and Public Safety (the “Criminal Justice Committee”), a majority of which recommended approval of the bill with the penalty reduced from a misdemeanor to a violation. See Legislative History at 000076, 000078. Notes from the Criminal Justice Committee's hearing indicate that some committee members were concerned with whether the bill and its penalties were necessary. See id. at 000099–000100. Representative Horrigan defended the law during the hearing, explaining that it “tightens up” existing law governing election fraud. Id. at 000099. Deputy Secretary of State David Scanlan also spoke in support of the bill, providing a “history of voting irregularities, including votes being bought.” Id. at 000100. When asked whether the bill was necessary, Deputy Secretary Scanlan responded that the “privacy of [the] ballot must be preserved.” Id. Ultimately, a majority of the Criminal Justice Committee recommended passing the bill so long as the penalty was decreased to a violation. Id. at 000076, 000078.

A minority of the Criminal Justice Committee, however, filed a report concluding that it would be “inexpedient to legislate” the bill. See Legislative History at 000083. The minority wrote:

Although the Minority agrees that the Criminal Justice Committee acted wisely in reducing the penalty from a misdemeanor to a violation, we believe this remains a very bad bill.... [I]t is not needed because we already have laws which prohibit people from selling their votes for financial gain, and that was the only reason supporters gave for passing the bill.... [T]his bill as drafted is overly broad. As such, it represents an intrusion on free speech. It fights a bogey man, which does not exist, at the expense of yielding even more of our freedoms.

Id. The minority suggested further amendment of the final sentence of paragraph I as follows:

This prohibition shall include taking a digital image or photograph of his or her marked ballot and distributing or sharing the image via social media or by any other means only if the distribution or sharing is for the purpose of receiving pecuniary benefit, as defined in RSA 640:2, II(c), or avoiding harm, as defined in
RSA

640:3
.

Id. at 000097 (emphasis added to denote minority's suggestions). Such an amendment, they argued, would make it illegal only to post a photo for financial gain or to avoid harm. Id. at 000083. They noted that this was the original intent of the bill according to the Secretary of State. Id. Nevertheless, the amendment was not supported by the majority of the Criminal Justice Committee and accordingly was not added to the bill that was presented to the House of Representatives. Id. at 000076, 000078.

The bill, as amended by the Election Committee and the majority of the Criminal Justice Committee, passed the full House by a veto-proof 198–96 majority. See Legislative History at 000063. On April 9, 2014, the Senate Public and Municipal Affairs Committee held a hearing, at which Representatives Horrigan and Till and Deputy Secretary Scanlan testified in support of the bill. Representative Horrigan stated that the practice of posting images of ballots on social media accounts “compromises the security of the polling place and the secrecy of the ballot.” Id. at 000063. He also cautioned that “[t]he new high-tech methods of showing a ballot absolutely could be used to further a serious vote-buying scheme.” Id. Similarly, Representative Till explained that “the seemingly innocent bragging about how one voted by posting a photo of one's completed ballot on Facebook, could undermine efforts to [e]nsure that no one is coerced into voting a particular way.” Id. at 000064. On April 17, 2014, the Senate Committee on Public and Municipal Affairs recommended that the bill “ought to pass,” and the Senate then passed the bill. Id. at 000057. On June 11, 2014, Governor Maggie Hassan signed the bill into law, effective September 1, 2014.

The new law's legislative history reveals that its opponents were concerned that the proposed law would infringe freedom of speech. In response, Representative Horrigan stated:

The bill's opponents framed this as a free speech issue, but political speech is in fact prohibited at the polling place. You absolutely have the right to engage in as much free speech as you want to beyond the boundary marked by the “No Electioneering” signs. However, the space inside that boundary is a secure space where the debate stops and the secret balloting begins.

Legislative History at 000063. Representative Till also addressed the opponents' concern, stating:

[E]very voter is free to tell as many people as they desire, in whatever forum they choose, how they voted. What is not allowed is to show one's completed ballot since, once cast, the ballot is the property of the state and in order to protect the secrecy of the ballot cannot be publicly identified with a particular voter.

Id. at 000064.

B. Vote Buying and Voter Coercion

Secretary of State William Gardner, the defendant in this action, defends the new law on the grounds that it is needed to prevent vote buying and voter coercion.

1. Evidence of Vote Buying and Voter Coercion in New Hampshire

The legislative history of the 2014 amendment to RSA 659:35 contains only a single reference to an actual alleged instance of vote buying in New Hampshire. As Representative Till described the incident:

I was told by a Goffstown resident that he knew for a fact that one of the major parties paid students from St Anselm's $50 to vote in the 2012 election. I don't know whether that is true or not, but I do know that if I were going to pay someone to vote a particular way, I would want proof that they actually voted that way.

Legislative History at 000064. She did not provide any other details about the incident, and it is not discussed elsewhere in the legislative history.

The summary judgment record does not include any evidence that either vote buying or voter coercion has occurred in New Hampshire since the late 1800s. See Doc. No. 18–1 at 2. Moreover, the state has received no complaints that images of marked ballots have been used to buy or coerce other votes. See Exhibit B to the Declaration of Gilles Bissonnette, Esq. in Support of Plaintiffs' Motion for Summary Judgment (“Exhibit B”) at 11, Rideout v. Gardner, No. 14–cv–489–PB (filed Mar. 27, 2015).

2. Vote Buying and Voter Coercion in the United States

There is no doubt that vote buying and voter coercion were at one time significant problems in the United States. See Doe v. Reed, 561 U.S. 186, 226, 130 S.Ct. 2811, 177 L.Ed.2d 493 (2010) (Scalia, J., concurring) (citing Burson v. Freeman, 504 U.S. 191, 202, 112 S.Ct. 1846, 119 L.Ed.2d 5 (1992) (plurality opinion)); Susan C. Stokes, et al., Brokers, Voters, and Clientelism: The Puzzle of Distributive Politics 200 (2013); Richard Hasen, Vote Buying, 88 Cal. L.Rev. 1323, 1327 (2000); Jill Lepore, Rock, Paper, Scissors: How We Used To Vote, New Yorker, Oct. 13, 2008, http://www.newyorker.com/magazine/2008/10/13/rock-paper-scissors.

Initially, the United States followed the viva voce system of voting used in England, in which voting “was not a private affair, but an open, public decision, witnessed by all and improperly influenced by some.” Burson, 504 U.S. at 200, 112 S.Ct. 1846. Gradually, states repealed the viva voce system in favor of written ballots. Id. At first, voters were expected to provide their own pen and paper, but when that became too complex, parties provided voters with printed ballot paper with a “ready-made slate of candidates.” L.E. Fredman, The Australian Ballot: The Story of an American Reform 21 (1968).

Because early written ballots were not secret ballots, they provided an opportunity for parties to buy votes. The parties used ballot paper that “was colored or otherwise recognizable” from a distance to ensure that the voter used the ballot he was given. Id. at 22; see Burson, 504 U.S. at 200, 112 S.Ct. 1846. Ballot peddlers or district captains then paid voters as they emerged from the polling place. Fredman, supra, at 22. For instance, in 1892, 16% of Connecticut voters were “up for sale” at prices ranging from $2 to $20. Id. at 23. Similarly, in 1887, a “study of New York City politics estimated that one-fifth of voters were bribed.” Stokes, supra, at 227.

By the end of the 19th century, most of the United States had adopted a new voting method referred to as the “Australian ballot.” Fredman, supra, at 83. The Australian ballot is a method of voting using a secret ballot that was first used in Australia in the mid–19th century. Id. at 7–9. It has four characteristics: (1) ballots are “printed and distributed at public expense”; (2) ballots contain the names of all nominated candidates; (3) ballots are distributed “only by ... election officers at the polling place”; and (4) “detailed provisions” are made for physical arrangements to ensure secrecy when casting a vote. Id. at 46. In 1888, Louisville, Kentucky became the first American city to adopt the Australian ballot, and in November 1889, Massachusetts was the first to use it statewide. Id. at 31, 36–39; Lepore, supra. New Hampshire has used the Australian ballot since 1891. Legislative History at 000062.

The Australian ballot drastically changed the utility of bribing voters because party workers could no longer monitor how voters voted. See Fredman, supra, at 47. Professor L.E. Fredman used the differences between the 1888 and 1892 presidential elections to highlight the effect. See id. at 83. Both elections featured Republican Benjamin Harrison against Democrat Grover Cleveland, but in the interim, 38 states had adopted the Australian ballot. Id. In 1888, the treasurer of the Republican National Committee instructed local officials: “Divide the floaters in blocks of five, and put a trusted man, with necessary funds, in charge of these five, and make them responsible that none get away.” Id. at 22. Although the memorandum exposed the extent of bribery during that election, Benjamin Harrison was elected. In the 1892 election, by contrast, “[t]here seemed to be more factual argument and fewer noisy processions, and the day itself was generally quiet and orderly.” Id. at 83; see also Stokes, supra, at 228 (“Historians also note the rising importance of party platforms in the late nineteenth century, another sign that vote buying was yielding to electoral strategies that, in [Theodore] Hoppen's phrase, ‘depended upon words.’ ”) (quoting Theodore K. Hoppen, The Mid–Victorian Generation: 1846–1886 (2000)).

For the most part, the Australian ballot is credited with delivering “a blow against clientelism,” Stokes, supra, at 241, and ending “direct bribery and intimidation.” Fredman, supra, at 129; see Burson, 504 U.S. at 204, 112 S.Ct. 1846 (“The success achieved through these reforms was immediately noticed and widely praised.”). Nevertheless, although the Australian ballot drastically reduced incentives to resort to vote buying, it did not eradicate the phenomenon entirely. For example, in Adams County, Ohio, vote buying was able to persist due to the “relative smallness” of the area. See Fabrice Lehoucq, When Does a Market for Votes Emerge?, in Elections for Sale: The Causes and Consequences of Vote Buying 33, 38 (Frederic C. Schaffer ed., 2007). There, in 1910, the “price of a vote oscillated between a drink of whisky and US$25, with the average price being US$8 per vote....” Id. (citing Genevieve B. Gist, Progressive Reform in a Rural Community: The Adams County Vote–Fraud Case, 48 Miss. Valley Historical Rev. 60, 62–63 (1961), http://www.jstor.org/stable/1902404). Similarly, due to rural populations with high poverty rates, “vote buying remained endemic well into the twentieth century” in many southern states. Stokes, supra, at 229.

Although “isolated and anachronistic,” there continue to be some reports of vote buying in the twenty-first century. Stokes, supra, at 231. For example, there have been recent prosecutions for violations of federal vote-buying statutes in Kentucky, North Carolina, and Illinois. See United States v. Thomas, 510 F.3d 714, 717 (7th Cir.2007); United States v. Shatley, 448 F.3d 264, 265 (4th Cir.2006); United States v. Johnson, No. 5:11–cr–143, 2012 WL 3610254, at *1 (E.D.Ky. Aug. 21, 2012); Stokes, supra, at 231. None of these cases, however, involved the use of a digital or photographic image of a marked ballot.

In addition to the introduction of the Australian ballot, anti-vote buying laws were a major cause of the decline of vote buying. See Allen Hicken, How Do Rules and Institutions Encourage Vote Buying?, in Elections for Sale: The Causes and Consequences of Vote Buying 47, 57 (Frederic C. Schaffer ed., 2007) (explaining that the strength of anti-vote buying rules “has the most direct impact on the expected utility of vote buying.”). In the United States, federal law makes it a crime to buy votes or engage in voter coercion. See 52 U.S.C. § 10307(b) (voter intimidation, threats, and coercion prohibited); 52 U.S.C. § 10307(c) (vote buying in certain federal elections prohibited). New Hampshire law also prohibits vote buying and voter coercion. N.H.Rev.Stat. Ann. § 659:40, I (“No person shall directly or indirectly bribe any person not to register to vote or any voter not to vote or to vote for or against any question submitted to voters or to vote for or against any ticket or candidate for any office at any election.”); N.H.Rev.Stat. Ann. § 659:40, II (“No person shall use or threaten force, violence, or any tactic of coercion or intimidation to knowingly induce or compel any other person to vote or refrain from voting, vote or refrain from voting for any particular candidate or ballot measure, or refrain from registering to vote.”); see also N.H.Rev.Stat. Ann. § 659:37 (voter interference prohibited); N.H.Rev.Stat. Ann. § 659:39 (giving liquor to voter to influence an election prohibited); N.H.Rev.Stat. Ann. § 659:40, III (voter suppression prohibited).

C. The Plaintiffs

The New Hampshire Attorney General's Office is currently investigating four individuals for alleged violations of RSA 659:35, I, including the three plaintiffs in this case. Doc. No. 18–1 at 9. The allegations concerning each of the plaintiffs arise from their votes in the September 9, 2014 Republican primary election, but the state does not contend that any of the plaintiffs were involved in vote buying. See Doc. No. 29 at 3.

Plaintiff Leon Rideout, who represents District 7 in Coos Country in the New Hampshire House of Representatives, voted in Lancaster, New Hampshire where he was on the ballot. Prior to casting his marked ballot, he took photographs of it with his phone. The ballot reflected that he voted for himself as well as other Republican candidates. Hours after he cast his ballot, he posted the photograph to Twitter with the text, “# COOS7 vote in primary 2014 # nhpolitics.” Doc. No. 18–1 at 9. He also posted the photograph to his House of Representatives Facebook page. In a September 11, 2014 article in the Nashua Telegraph, Rideout explained, “I did it to make a statement.... I think [RSA 659:35, I is] unconstitutional.... It's really just an overreach of the government trying to control something that, in my opinion, doesn't need to be regulated.” David Brooks, You Didn't Take a Picture of Your Ballot Tuesday, Did You? (It's Illegal), Nashua Telegraph, Sept. 11, 2014, http://www.nashuatelegraph.com/news/1046026–469/you–didnt–take–a–picture–of–your.html. After Rideout posted the image, Paul Brodeur, an investigator from the Attorney General's Office, called him and requested an interview, which was conducted on September 16, 2014. The Attorney General's Office threatened to prosecute Rideout under RSA 659:35, I, but no complaint was served because the plaintiffs entered into agreements with the state to toll the statute of limitations period. Doc. No. 18–1 at 11.

The Attorney General's Office is also investigating Andrew Langlois, who voted in Berlin, New Hampshire. Because Langlois did not approve of his Republican choices for U.S. Senate, he wrote the name of his recently-deceased dog, “Akira,” as a write-in candidate. He took a photograph of his ballot on his phone while in the ballot booth. He later posted the photograph on Facebook, writing in part, “Because all of the candidates SUCK, I did a write-in of Akira....” Doc. No. 19–20 at 2. Brodeur called Langlois after the election and explained that he was being investigated for posting his ballot on social media. Because Langlois was unaware of RSA 659:35, I, he initially thought Brodeur's call was a “joke.” Doc. No. 18–1 at 12.

Brandon Ross, the third plaintiff, voted in Manchester, where he was a candidate for the New Hampshire House of Representatives. With his phone, Ross took a photograph of his marked ballot, which reflected his vote for himself and other Republican candidates. He took the picture to keep a record of his vote and to preserve the opportunity to show his marked ballot to friends. He was aware of RSA 659:35, I when he took the photograph, and he did not immediately publish it because of the law's penalties. After learning that the Attorney General's Office was investigating voters for violating RSA 659:35, I, on September 19, 2014, Ross posted the photograph of his marked ballot on Facebook with the text “Come at me, bro.” Doc. No. 19–22 at 2. Representative Horrigan, the sponsor of the bill to amend RSA 659:35, filed an election law complaint, which triggered an investigation of Ross by the Attorney General's Office.

D. Procedural History

On October 31, 2014, Rideout, Langlois, and Ross filed a complaint pursuant to 42 U.S.C. § 1983 challenging the constitutionality of RSA 659:35. They requested declarations that the new law is facially unconstitutional and unconstitutional as applied to the plaintiffs. Doc. No. 1 at 20–21. They also sought an injunction to prohibit the state from enforcing RSA 659:35, I. Id. at 21.

On November 11, 2014, the plaintiffs filed a motion for a preliminary injunction. Ten days later, the parties agreed to an expedited discovery schedule in order to allow the issue to be decided on the merits rather than on a motion for a preliminary injunction. See Fed.R.Civ.P. 65(a)(2) (authorizing court to consolidate preliminary injunction hearing and trial).

The parties have filed cross motions for summary judgment. See Doc. Nos. 18, 22. Both parties agree that there is no need for a trial because none of the material facts are in dispute. Doc. No. 29 at 2.

II. STANDARD OF REVIEW

This case will be resolved on cross motions for summary judgment.

Summary judgment is appropriate when the record reveals “no genuine dispute as to any material fact and [that] the movant is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(a). The evidence submitted in support of the motion must be considered in the light most favorable to the nonmoving party, drawing all reasonable inferences in its favor. See Navarro v. Pfizer Corp., 261 F.3d 90, 94 (1st Cir.2001).

A party seeking summary judgment must first identify the absence of any genuine dispute of material fact. Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). A material fact “is one ‘that might affect the outcome of the suit under the governing law.’ ” United States v. One Parcel of Real Prop. with Bldgs., 960 F.2d 200, 204 (1st Cir.1992) (quoting Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986)). If the moving party satisfies this burden, the nonmoving party must then “produce evidence on which a reasonable finder of fact, under the appropriate proof burden, could base a verdict for it; if that party cannot produce such evidence, the motion must be granted.” Ayala–Gerena v. Bristol Myers–Squibb Co., 95 F.3d 86, 94 (1st Cir.1996); see Celotex, 477 U.S. at 323, 106 S.Ct. 2548.

On cross motions for summary judgment, the standard of review is applied to each motion separately. See Am. Home Assurance Co. v. AGM Marine Contractors, Inc., 467 F.3d 810, 812 (1st Cir.2006); see also Mandel v. Boston Phoenix, Inc., 456 F.3d 198, 205 (1st Cir.2006) (“The presence of cross-motions for summary judgment neither dilutes nor distorts this standard of review.”). Hence, I must determine “whether either of the parties deserves judgment as a matter of law on facts that are not disputed.” Adria Int'l Group, Inc. v. Ferré Dev., Inc., 241 F.3d 103, 107 (1st Cir.2001).

III. ANALYSIS

Plaintiffs challenge only the portion of RSA 659:35, I that makes it unlawful for a voter to take and disclose an image of his or her marked ballot. As they see it, this act of disclosure, which ordinarily occurs far from the polling place and will generally be accomplished through the use of social media, is an important and effective means of political expression that is protected by the First Amendment. In contrast, Secretary Gardner defends the law primarily by arguing that it is a necessary restraint on speech that is required to prevent vote buying and voter coercion.

The Supreme Court has developed a template for resolving conflicts between speech rights and governmental interests. Speech restrictions are first sorted by whether they are content based or content neutral. Content-based restrictions are subject to strict scrutiny, “ ‘which requires the Government to prove that the restriction furthers a compelling interest and is narrowly tailored to achieve that interest.’ ” Reed v. Town of Gilbert, ––– U.S. ––––, 135 S.Ct. 2218, 2231, 192 L.Ed.2d 236 (2015) (quoting Ariz. Free Enter. Club's Freedom Club PAC v. Bennett, 564 U.S. 721, 131 S.Ct. 2806, 2817, 180 L.Ed.2d 664 (2011)). Content-neutral restrictions, however, are subject only to intermediate scrutiny, meaning “the government may impose reasonable restrictions on the time, place, or manner of protected speech,” so long as “ ‘they are narrowly tailored to serve a significant governmental interest, and that they leave open ample alternative channels for communication of the information.’ ” Ward v. Rock Against Racism, 491 U.S. 781, 791, 109 S.Ct. 2746, 105 L.Ed.2d 661 (1989) (quoting Clark v. Cmty. for Creative Non–Violence, 468 U.S. 288, 293, 104 S.Ct. 3065, 82 L.Ed.2d 221 (1984)).

I begin by determining whether the 2014 amendment to RSA 659:35, I is a content-based or content-neutral restriction on speech.

A. Content Neutrality

As the Supreme Court recently explained in Reed v. Town of Gilbert, “[g]overnment regulation of speech is content based if a law applies to particular speech because of the topic discussed or the idea or message expressed.” 135 S.Ct. at 2227. A law that distinguishes between permitted and prohibited speech based on the subject matter, function, or purpose of the speech is content based on its face. Id. Additionally, even a facially-neutral law will be deemed to be content based if it either cannot be justified without reference to the content of the speech or discriminates based on the speaker's point of view. Id.

A law that is content based on its face will be subject to strict scrutiny even though it does not favor one viewpoint over another and regardless of whether the legislature acted with benign motivations when it adopted the law. See id. at 2229–30. As the Reed court explained, “[i]nnocent motives do not eliminate the danger of censorship presented by a facially content-based statute, as future government officials may one day wield such statutes to suppress disfavored speech.” Id. at 2229; see also Turner Broad. Syst., Inc. v. FCC, 512 U.S. 622, 642–43, 114 S.Ct. 2445, 129 L.Ed.2d 497 (1994) (“Nor will the mere assertion of a content-neutral purpose be enough to save a law which, on its face, discriminates based on content.”).

In Reed, the Court applied these principles to invalidate a sign code that governed the manner in which people could display outdoor signs in Gilbert, Arizona. Reed, 135 S.Ct. at 2226. The sign code generally prohibited the display of outdoor signs anywhere within the town without a permit. It exempted twenty-three categories of signs from that requirement, but placed various lesser requirements on each of those twenty-three categories. For example, a political sign could be larger than a temporary directional sign and could be displayed for a longer amount of time. The Court held that the sign code was content based on its face because it treated each sign category differently dependent upon the type of content conveyed. Id. at 2227. Because the sign code was facially content based, the Court subjected it to strict scrutiny without attempting to identify the legislature's purpose or justification. Id.

In the present case, as in Reed, the law under review is content based on its face because it restricts speech on the basis of its subject matter. The only digital or photographic images that are barred by RSA 659:35, I are images of marked ballots that are intended to disclose how a voter has voted. Images of unmarked ballots and facsimile ballots may be shared with others without restriction. In fact, the law does not restrict any person from sharing any other kinds of images with anyone. In short, the law is plainly a content-based restriction on speech because it requires regulators to examine the content of the speech to determine whether it includes impermissible subject matter. Accordingly, like the sign code at issue in Reed, the law under review here is subject to strict scrutiny even though it does not discriminate based on viewpoint and regardless of whether the legislature acted with good intentions when it adopted the law.

The Secretary nevertheless contends that the new law should be exempt from strict scrutiny even if it is a content-based restriction on speech because it is only a partial ban on speech about how a voter has voted. In other words, because the new law leaves voters free to use other means to inform others about how they have voted, the Secretary argues that the law is merely a time, place, or manner restriction on speech that is subject only to intermediate scrutiny. This argument is a nonstarter. As the Supreme Court explained in United States v. Playboy Entertainment Group, Inc., “[t]he distinction between laws burdening and laws banning speech is but a matter of degree. The Government's content-based burdens must satisfy the same rigorous scrutiny as its content-based bans.” 529 U.S. 803, 812, 120 S.Ct. 1878, 146 L.Ed.2d 865 (2000). Here, the law at issue is a content-based restriction on speech that deprives voters of one of their most powerful means of letting the world know how they voted. The legislature cannot avoid strict scrutiny when it adopts such a law merely by leaving voters with other arguably less effective means of speaking on the subject.

The Secretary also argues that the law should not be considered a content-based restriction on speech because paragraph II of RSA 659:35 additionally prohibits a voter from placing “a distinguishing mark upon his or her ballot.” See N.H.Rev.Stat. Ann. § 659:35, II. That is, because paragraph II prohibits another type of marking on ballots, the new law barring a voter from disclosing an image of a marked ballot is content neutral. This argument fails. The two paragraphs simply regulate two different categories of speech: paragraph I regulates a certain type of speech that ordinarily occurs outside the polling place and paragraph II regulates what types of markings a voter can make on a ballot while in the polling place. Because paragraph I regulates speech based on the content conveyed, paragraph II cannot save it from being a content-based restriction on speech.

In a last-ditch effort to save the law from strict scrutiny, the Secretary argues that completed ballots are a form of government speech and thus do not trigger First Amendment protection at all. He cites Walker v. Texas Division, Sons of Confederate Veterans, which held that Texas's specialty license plate designs constituted government speech and thus Texas was entitled to refuse to issue plates featuring a group's proposed design. ––– U.S. ––––, 135 S.Ct. 2239, 2253, 192 L.Ed.2d 274 (2015). In reaching its decision, the Court in Walker relied on the facts that (1) license plates “long have communicated messages from the States,” (2) Texas license plate designs “are often closely identified in the public mind with the State,” and (3) Texas maintains direct control over the messages conveyed on its specialty plates. Id. at 2248–49 (internal quotations and alterations omitted). The problem at issue here, however, is quite different from the problem the Court resolved in Walker. First, ballots do not communicate messages from the state; they simply list slates of candidates. Second, although blank ballots may be identified with the state, there is no possibility that a voter's marking on a ballot will be misinterpreted as state speech. Third, New Hampshire does not maintain direct control over the messages that people convey on ballots, apart from the restriction that they place no distinguishing mark on their ballot. See N.H.Rev.Stat. Ann. § 659:35, II. Accordingly, any markings that voters place on their ballots clearly do not qualify as government speech.

Although the Secretary does not press the point, Representative Horrigan also suggested during debate on the new law that it could be justified because it regulates speech at the polling place where electioneering is not permitted. I disagree. RSA 659:35, I does not bar voters from taking pictures of their completed ballots before they are cast. What they may not do is disclose images of a completed ballot to others. Because disclosure will generally take place far away from the polling place, the Secretary cannot prevent the new law from being subject to strict scrutiny by claiming that it is merely a restriction on speech in a nonpublic forum, where speech rights are more limited. See e.g., Burson v. Freeman, 504 U.S. 191, 214, 112 S.Ct. 1846, 119 L.Ed.2d 5 (1992) (Scalia, J., concurring in the judgment) (arguing that viewpoint-neutral restrictions on speech in the vicinity of polling places should not be subject to strict scrutiny because they restrict speech in what is traditionally a nonpublic forum).

For similar reasons, a law that restricts a person's ability to tell others how he has voted is not exempt from strict scrutiny merely because the ballot itself is a nonpublic forum. See, e.g., Timmons v. Twin Cities Area New Party, 520 U.S. 351, 363, 117 S.Ct. 1364, 137 L.Ed.2d 589 (1997) ( “Ballots serve primarily to elect candidates, not as forums for political expression”). The law at issue here does not restrict what a voter may write on his ballot; it regulates the way in which he can disclose his vote to others. Thus, the nonpublic forum doctrine cannot be invoked to save the law from strict scrutiny because the speech that the law restricts necessarily occurs in forums that the government does not own or control. To illustrate the point, consider a law that bans public discussion of what is said at a candidate debate held by a public broadcaster. Is there any doubt that such a law would be subject to strict scrutiny even though the Supreme Court has held that the debate itself occurs in a nonpublic forum? See Ark. Educ. Television Comm'n v. Forbes, 523 U.S. 666, 680, 118 S.Ct. 1633, 140 L.Ed.2d 875 (1998) (debate conducted by a public broadcaster is a nonpublic forum). Obviously not. For the same reasons, the law at issue here is not exempt from strict scrutiny merely because the ballot itself is a nonpublic forum.

B. Strict Scrutiny

Because the 2014 amendment to RSA 659:35, I is a content-based restriction on speech, it can stand only if it survives strict scrutiny, “ ‘which requires the Government to prove that the restriction furthers a compelling interest and is narrowly tailored to achieve that interest.’ ” Reed, 135 S.Ct. at 2231 (quoting Ariz. Free Enter., 131 S.Ct. at 2817). The Secretary bears the burden of establishing both requirements. See id. As I explain below, he has failed to meet his burden on either part of the strict scrutiny test.

1. State Interests

The Secretary argues that a ban on displays of completed ballots serves the state's compelling interest in preventing vote buying and voter coercion. While both interests are plainly compelling in the abstract, the mere assertion of such interests cannot sustain a content-based speech restriction.

For an interest to be sufficiently compelling, the state must demonstrate that it addresses an actual problem. Brown v. Entm't Merchs. Ass'n, 564 U.S. 786, 131 S.Ct. 2729, 2738, 180 L.Ed.2d 708 (2011) (“The state must specifically identify an ‘actual problem’ in need of solving ....” (quoting Playboy, 529 U.S. at 822–23, 120 S.Ct. 1878)); see also Asociación de Educación Privada de Puerto Rico, Inc. v. García–Padilla, 490 F.3d 1, 18 (1st Cir.2007) (“We cannot conclude that [the Puerto Rico Department of Consumer Affairs] has a legitimate state interest in fixing a problem it has not shown to exist.”). To satisfy this requirement, the government ordinarily must point to sufficient evidence in the law's legislative history or in the record before the court to show that the problem exists. See Turner, 512 U.S. at 667, 114 S.Ct. 2445 (explaining that without evidence of an actual problem, “we cannot determine whether the threat [asserted by the government] is real enough” to survive strict scrutiny). “Anecdote and supposition” cannot substitute for evidence of a real problem. Playboy, 529 U.S. at 822, 120 S.Ct. 1878; Consol. Edison Co. of N.Y., Inc. v. Pub. Serv. Comm'n of N.Y., 447 U.S. 530, 543, 100 S.Ct. 2326, 65 L.Ed.2d 319 (1980) (“Mere speculation of harm does not constitute a compelling state interest.”).

In the present case, neither the legislative history nor the evidentiary record compiled by the Secretary in defense of this action provide any support for the view that the state has an actual or imminent problem with images of completed ballots being used to facilitate either vote buying or voter coercion. The law's legislative history contains only a single unsubstantiated third-hand report that vote buying occurred in Goffstown during the 2012 election. See Legislative History at 000064. Although the Secretary was given the opportunity to do so, he produced no evidence that either vote buying or voter coercion are current problems in New Hampshire. Plaintiffs, in contrast, have produced undisputed evidence that there have been no vote buying prosecutions and no complaints of vote buying in the state since at least 1976. Exhibit B at 11. More to the point, even though small cameras capable of taking photographic images of ballots have been available for decades and cell phones equipped with digital cameras have been in use for nearly 15 years, the Secretary has failed to identify a single instance anywhere in the United States in which a credible claim has been made that digital or photographic images of completed ballots have been used to facilitate vote buying or voter coercion. Although legislatures are entitled to deference when making predictive judgments, deference cannot be blind to the complete absence of evidence when speech restrictions are at issue. Here, the Secretary offers only anecdote and speculation to sustain the law, which is insufficient when it is applied to a content-based restriction on speech.

The Secretary invokes the Supreme Court's plurality decision in Burson v. Freeman to support his claim that content-based speech restrictions can be justified without evidence that compelling state interests are under actual threat. There, the statute under review established a buffer zone around polling places to protect voters from solicitation and the distribution of campaign materials. Burson, 504 U.S. at 193–94, 112 S.Ct. 1846 (plurality opinion). In sustaining the statute against a First Amendment challenge, the plurality relied heavily on historical evidence demonstrating that predecessor statutes to the one under review had been adopted long ago to respond to a situation in which “[a]pproaching the polling place ... was akin to entering an open auction place.” Id. at 202, 112 S.Ct. 1846. The Court concluded that it was appropriate for the state to act without evidence of a current problem in part because the “long, uninterrupted and prevalent” use of similar statutes throughout the United States made it difficult for the state to determine what would happen if the challenged law were invalidated. Id. at 208, 112 S.Ct. 1846.

Burson, however, is a very different case from the one I decide today. In contrast to the statute at issue in Burson, the 2014 amendment to RSA 659:35, I is quite new and cannot be tied to historical evidence of recent vote fraud. Although it is true that vote buying was a problem in this country before the adoption of the Australian ballot, the historical record establishes that vote buying has not been a significant factor in elections in more than 100 years. Further, because the law at issue here is new and the technology it targets has been in use for many years, it is reasonable to expect that if the problem the state fears were real, it would be able to point to some evidence that the problem currently exists. Under these circumstances, both history and common sense undermine rather than support the state's contention that vote buying and voter coercion will occur if the state is not permitted to bar voters from displaying images of their completed ballots.

Because the Secretary has failed to demonstrate that the law serves a compelling state interest, it fails to satisfy strict scrutiny.

2. Narrow Tailoring

Even if the Secretary had proved that the new law serves a compelling interest, it would still fail the strict scrutiny test because it is not narrowly tailored to address the alleged state interests.

When the government attempts to restrict speech in order to further a state interest, it ordinarily must demonstrate that the restriction “ ‘is narrowly tailored to achieve that interest.’ ” Reed, 135 S.Ct. at 2231 (quoting Ariz. Free Enter., 131 S.Ct. at 2817). Even content-neutral restrictions require narrow tailoring because “silencing speech is sometimes the path of least resistance ... [and] by demanding a close fit between ends and means, the tailoring requirement prevents the government from too readily ‘sacrific[ing] speech for efficiency.’ ” McCullen v. Coakley, –––U.S. ––––, 134 S.Ct. 2518, 2534, 189 L.Ed.2d 502 (2014) (quoting Riley v. Nat'l Fed'n of the Blind of N.C., Inc., 487 U.S. 781, 795, 108 S.Ct. 2667, 101 L.Ed.2d 669 (1988)). This tailoring requirement is even more demanding when the state elects to restrict speech based on its content. In such cases, the burden is on the state to demonstrate that the restriction it has adopted is the “least restrictive means” available to achieve the stated objective. Ashcroft v. ACLU, 542 U.S. 656, 666, 124 S.Ct. 2783, 159 L.Ed.2d 690 (2004); McCullen v. Coakley, ––– U.S. ––––, 134 S.Ct. 2518, 2530, 189 L.Ed.2d 502 (2014) (dictum); Globe Newspaper Co. v. Pokaski, 868 F.2d 497, 505 (1st Cir.1989); but cf. Williams–Yulee v. Fla. Bar, ––– U.S. ––––, 135 S.Ct. 1656, 1671, 191 L.Ed.2d 570 (2015) (narrow tailoring does not require perfect tailoring even when a content-based speech restriction is under review).

Among other reasons, a law is not narrowly tailored if it is significantly overinclusive. See Brown, 131 S.Ct. at 2741; Simon & Schuster, Inc. v. Members of N.Y. State Crime Victims Bd., 502 U.S. 105, 121, 123, 112 S.Ct. 501, 116 L.Ed.2d 476 (1991); First Nat'l Bank of Boston v. Bellotti, 435 U.S. 765, 794–95, 98 S.Ct. 1407, 55 L.Ed.2d 707 (1978). For example, in Simon & Schuster, Inc. v. Members of New York State Crime Victims Board, the law at issue required that an accused or convicted criminal's income from works describing his crime be deposited in an escrow account and made available to the victims of the crime and the criminal's other creditors. 502 U.S. at 108, 112 S.Ct. 501. The Supreme Court held that the law was a “significantly overinclusive” means of ensuring that victims are compensated from the proceeds of crime, and therefore the law was not narrowly tailored. Id. at 121, 123, 112 S.Ct. 501. Describing the reach of the statute, the Court stated:

Should a prominent figure write his autobiography at the end of his career, and include in an early chapter a brief recollection of having stolen ... a nearly worthless item as a youthful prank, the [government entity] would control his entire income from the book for five years, and would make that income available to all of the author's creditors....

Id. at 123, 112 S.Ct. 501. That is, the statute applied to a wide range of literature that would not enable a criminal to profit while a victim remained uncompensated. Because the law covered far more material than necessary to accomplish its goals, the Court held that the statute was vastly overinclusive and therefore not narrowly tailored. Id.

Here, like the law at issue in Simon & Schuster, the 2014 amendment to RSA 659:35, I is vastly overinclusive and is therefore not narrowly tailored to further a compelling interest. Even if the Secretary could demonstrate that New Hampshire has an actual problem with either vote buying or voter coercion and that allowing voters to display images of their ballots would exacerbate either problem, the means that the state has chosen to address the issue will, for the most part, punish only the innocent while leaving actual participants in vote buying and voter coercion schemes unscathed. As the complaints of the voters who are now under investigation reveal, the people who are most likely to be ensnared by the new law are those who wish to use images of their completed ballots to make a political point. The few who might be drawn into efforts to buy or coerce their votes are highly unlikely to broadcast their intentions via social media given the criminal nature of the schemes in which they have become involved. As a result, investigative efforts will naturally tend to focus on the low-hanging fruit of innocent voters who simply want the world to know how they have voted for entirely legitimate reasons. When content-based speech restrictions target vast amounts of protected political speech in an effort to address a tiny subset of speech that presents a problem, the speech restriction simply cannot stand if other less restrictive alternatives exist.

Because the 2014 amendment is a content-based restriction on speech, it falls to the government to demonstrate that less speech-restrictive alternatives will not work. Playboy, 529 U.S. at 816, 120 S.Ct. 1878. In the present case, the state has an obviously less restrictive way to address any concern that images of completed ballots will be used to facilitate vote buying and voter coercion: it can simply make it unlawful to use an image of a completed ballot in connection with vote buying and voter coercion schemes. The Secretary has failed to explain why this alternative would be less effective. At most, he has offered a generalized complaint that vote buying and voter coercion are difficult to detect. This explanation, however, merely highlights the ineffectiveness of the approach to the problem that the legislature has adopted. Vote buying and voter coercion will be no less difficult to detect if the statute remains in effect because people engaged in vote buying and voter coercion will not publicly broadcast their actions via social media. Accordingly, rather than demonstrating that alternatives would be ineffective, the Secretary's response only demonstrates the ineffectiveness of the law at issue.

Because the 2014 amendment to RSA 659:35, I is vastly overinclusive and the Secretary has failed to demonstrate that less speech-restrictive alternatives will be ineffective to address the state's concerns, it cannot stand to the extent that it bars voters from disclosing images of their completed ballots.

IV. CONCLUSION

The Supreme Court requires lower courts to use a categorical approach when resolving First Amendment problems of the type at issue here. Thus, the viability of a challenged statute will turn on questions such as whether the statute is “content based,” whether it serves “compelling governmental interests,” and whether it is “narrowly tailored” to achieve those interests. I have followed this approach in concluding that the new law is a content-based restriction on speech that cannot survive strict scrutiny because it neither actually serves compelling state interests nor is it narrowly tailored to achieve those interests.

One sitting Supreme Court Justice has called for the lines between constitutional categories to be softened to permit judges to address the competing interests that underlie disputes such as the one at issue here more directly and with greater flexibility. See Reed v. Town of Gilbert, 135 S.Ct. 2218, 2234 (Breyer, J., concurring) (“The First Amendment requires greater judicial sensitivity both to the Amendment's expressive objectives and to the public's legitimate need for regulation than a simple recitation of categories, such as ‘content discrimination’ and ‘strict scrutiny,’ would permit.”) Although there are sound policy reasons to allow judges greater flexibility when analyzing First Amendment questions, I would not come to a different conclusion in this case even if I were free to more directly balance the interests that are at stake here. At its core, this dispute turns on a claim that the political speech rights of voters must be curtailed to protect the vote against those who would corrupt it with cash and coercion. If this claim could be grounded in something other than speculation, it would be more difficult to resolve because few, if any, rights are more vital to a well-functioning democracy than either the right to speak out on political issues or the right to vote free from coercion and improper influence. But the record in this case simply will not support a claim that these two interests are in irreconcilable conflict. Neither the legislative history of the new law nor the evidentiary record compiled by the parties provide support for the view that voters will be either induced to sell their votes or subjected to coercion if they are permitted to disclose images of their ballots to others. Nor is there any reason to believe that other less restrictive means could not be used to address either problem at least as effectively as the massively overinclusive law that is at issue here. Accordingly, this case does not present the type of conflict between speech rights and other governmental interests that can be used to justify a law that restricts political speech.

Although the plaintiffs have sought both declaratory and injunctive relief, I have no reason to believe that the Secretary will fail to respect this Court's ruling that the new law is unconstitutional on its face. Accordingly, I grant the plaintiffs' request for declaratory relief but determine that injunctive relief is not necessary at the present time. See Wooley v. Maynard, 430 U.S. 705, 711, 97 S.Ct. 1428, 51 L.Ed.2d 752 (1977) (injunctive relief is not required if the plaintiffs' interests will be protected by a declaratory judgment). The plaintiffs' motion for summary judgment (Doc. No. 18) is granted to the extent that it seeks a judgment for declaratory relief, and the Secretary's corresponding motion (Doc. No. 22) is denied. The clerk shall enter judgment for the plaintiffs.

SO ORDERED.