Opinion
Civil Action 21-cv-2380 (BAH)
03-31-2023
PEOPLE FOR THE ETHICAL TREATMENT OF ANIMALS, MADELINE KRASNO, and RYAN HARTKOPF, Plaintiffs, v. LAWRENCE A. TABAK, in his official capacity as Acting Director of the National Institutes of Health, and XAVIER BECERRA, in his official capacity as Secretary of the U.S. Department of Health and Human Services, Defendants.
MEMORANDUM OPINION
BERYL A. HOWELL, DISTRICT COURT JUDGE
Pending before this Court is a question at the frontier of courts' application of the First Amendment to the internet-an environment that is challenging to navigate under the strict categorization and spatial reasoning of First Amendment doctrine. The foundational case law of the First Amendment has developed in the quaint setting of sidewalks and parks, schools and public transit, when, by comparison, cyberspace is analogizable to the shifting sands of the Sahara. Here, the architectural guideposts in the form of physical “materials, design, and demarcation from the surrounding area,” Hodge v. Talkin, 799 F.3d 1145, 1150 (D.C. Cir. 2015), that steer traditional permissible expressive use and access analysis must be found in alternative ways. Courts confronting such issues must tread these sands with caution.
The corner of cyberspace at issue in the instant matter is the official Facebook and Instagram pages maintained by the National Institutes of Health (NIH), where animal-rights activists, including plaintiffs, regularly post comments criticizing the practice of animal testing.
See Parties' Joint Stipulation of Facts (“Jt. Stip”) ¶¶ 62-86, ECF No. 28. Indeed, NIH maintains that animal-rights activists' comments comprise “the lion's share of the comment thread[s]” responding to many of the agency's posts. Defs.' Opp'n Pls.' Mot. Summ. J. & Mem. Supp. Cross-Mot. Summ. J. (“Defs.' Opp'n”) at 12, ECF No. 31-1.In response to these posts, NIH uses custom keyword filters to hide comments containing a range of keywords, including words related to animal testing, such as “animals” and “torture.” Jt. Stip. ¶ 58. Plaintiffs initiated the instant litigation challenging NIH's keyword filters as a “patently unconstitutional practice of surreptitiously censoring speech it does not like.” Pls.' Mem. Supp. Mot. Summ. J. (“Pls.' Mem.”) at 7, ECF No. 30-1. For the reasons set forth below, the Court agrees with defendants that NIH's social media pages' comment threads are limited public fora, and NIH's enforcement of its commenting guidelines through keyword filters is viewpoint-neutral and reasonable. Accordingly, defendants' cross-motion for summary judgment is granted, and plaintiffs' motion for the same is denied.
Two memoranda filed in support of the parties' respective motions are filed twice and, to simplify citation, only one of the duplicate memoranda are referenced. Defendants' memorandum in support of their Cross-Motion for Summary Judgment and in Opposition to Plaintiffs' Motion for Summary Judgment is docketed at ECF Nos. 31 and 32; only the memorandum at ECF No. 31 is referenced. Similarly, plaintiffs' Memorandum of Law in Opposition to Defendants' Cross-Motion for Summary Judgment and Reply Memorandum in Further Support of Plaintiffs' Motion for Summary Judgment is docketed at ECF No. 34 and 35; only the memorandum at ECF No. 34 is referenced.
I. BACKGROUND
The factual background and procedural history relevant to the pending motions are described below.
A. Factual Background
NIH, the primary federal agency charged with performing and supporting biomedical and behavior research, maintains verified Facebook and Instagram accounts that are viewable by the public. See Jt. Stip. ¶¶ 4, 36-38, 41-43, ECF No. 28. Users of both social media platforms are invited to react to NIH's posts on these platforms-in the case of Facebook, through a spectrum of “reaction” options, and on Instagram, via the humble “like” reaction-and to post comments. Id. ¶¶ 14, 29, 39, 45. Both platforms are used by NIH to “‘communicate and interact with citizens' about agency-related updates and public health news,” id. ¶ 36 (quoting Web Policies and Notices, NATIONAL INSTITUTES OF HEALTH (Jan. 14, 2021), https://perma.cc/BC7R-ZFME), such as, for example, providing updates on research studies related to COVID-19 vaccines and sharing a woman's “Alzheimer's Caregiver story” to promote caregiving resources, see id. ¶ 44; id., Ex. 1, NIH Facebook Post & Viewable Comments (June 4, 2021), ECF No. 28-1; id., Ex. 15, NIH Instagram Post & Viewable Comments (Nov. 12, 2021), ECF No. 28-15.
1. NIH's Social Media Moderation Policy
Comments on NIH's Facebook and Instagram pages are governed by NIH's publicly available guidelines designed to “encourage respectful and constructive dialogue.” Id., Ex. 19, NIH Comment Guidelines, ECF No. 28-19. The guidelines, first posted online in March 2015, See Jt. stip. ¶ 52, prohibit, inter alia, “[v]ulgar, obscene, profane, threatening, or abusive language,” “[r]epetitive posts,” and-most relevant to the instant matter-“[o]ff-topic posts,” NIH Comment Guidelines at 2. At the time that the parties filed the Joint stipulation, in February 2022, NIH's Facebook and Instagram accounts contained direct links to the comment guidelines. Jt. stip. ¶ 53. The accounts also contained direct links to NIH's separate “Web Policies and Notices” page, which adds that “as a practice, comment moderator policy requires the removal from NIH Facebook pages of any comments that contain spam or are improper, inflammatory, off-topic, or offensive.” Id. ¶ 54 (not mentioning NIH's policy vis-a-vis Instagram).
NIH's own comment guidelines supplement content policies maintained by the platforms themselves. Facebook and Instagram prohibit certain content pursuant to their community standards, including content inciting or facilitating serious violence, hate speech, and spam. Id. ¶¶ 20 & n.2, 35 & n.5. Content falling within these prohibited categories are removed by the platforms with notice to users. Id. ¶¶ 20, 35. Instagram also uses an artificial intelligence algorithm to hide automatically comments that are similar to those reported as violating the platform's policies; hidden comments are not deleted, but rather accessible at the bottom of a post by clicking a “view hidden comments” option. Id. ¶ 35.
Both platforms give account-holders tools to moderate the comments on their account pages. On Facebook, account administrators, like NIH, have the option to “ban” individual users-effectively muting the targeted users by allowing them to access the page but removing their commenting or reacting privileges-and to “block” users, by preventing them from accessing the page altogether. Id. ¶ 19. Administrators can also moderate their pages comment-by-comment by manually deleting or hiding certain comments, with the latter function resulting in the comment remaining visible only to the comment's creator and that user's friends. Id. ¶ 17. On Instagram, similarly, account holders can block other users from commenting on or viewing their posts, such that blocked users' comments are rendered invisible to all but the blocked users, who are not notified of the hidden nature of their comments. Id. ¶¶ 33-34. Instagram also maintains a “restrict comments” feature, allowing account holders to pre-screen particular users' comments before they appear on the account holder's posts. Id. ¶ 32.
Most relevant to the instant litigation are the platforms' keyword filtering functions. On Facebook, the platform offers comment filtering tools such as a “profanity filter,” which account administrators can simply enable to filter all comments containing profanity. Id. ¶ 16. Account administrators can add specified words to the filter. Comments containing filtered words are removed from public view but still accessible to the user who posted them and the user's friends. Id. On Instagram, too, account holders can automatically hide all comments containing specified words or phrases, resulting in the hidden comment only being accessible to the user who posted it and to the account holder upon choosing to “view hidden comments.” Id. ¶ 30. In addition to this function permitting customized lists of blocked keywords, Instagram maintains a “Hide Comments” function that by default identifies comments containing offensive words or phrases reported to violate the platform's terms of service. Those comments, unlike those blocked via the custom keywords list, are still visible to all Instagram users, but they are deprioritized: to view them, users must scroll to the bottom of a post's comments section and choose to view the hidden comments. Id. ¶ 31.
NIH primarily enforces its comment guidelines on Facebook and Instagram via the keyword filtering function, using Facebook and Instagram's default filters, which are supplemented with NIH's custom keyword lists, discussed infra in Part I.A.3. Id. ¶¶ 56-58. Before plaintiffs initiated this litigation, NIH endeavored manually to hide comments that violated the guidelines, but those efforts were “limited” by resource constraints, and “especially limited” in the wake of the further strains created by the COVID-19 pandemic. Id. ¶ 61. NIH has advised that the practice of manually hiding comments violating the agency's guidelines will resume upon the resolution of the instant motions. Id.
2. Plaintiffs' Anti-Animal Testing Social Media Advocacy
Plaintiff People for the Ethical Treatment of Animals (PETA) is a non-profit organization that has launched a social media campaign focusing on NIH's funding of primate testing, and as part of this campaign, its supporters and employees engage on NIH's social media channels to protest animal testing. Jt. Stip. ¶¶ 1, 88. Individual plaintiffs Madeline Krasno and Ryan Hartkopf are animal rights advocates, who use social media as a tool to “raise awareness” about animal testing, NIH's funding of research involving the practice, and “mental health issues often experienced by animal lab workers”; to pressure institutions including NIH to curtail animal testing; and to “signal to potential whistleblowers that they have allies.” Id. ¶¶ 89-90. To achieve these advocacy goals, plaintiffs frequently comment about animal testing on social media, including NIH's Instagram and Facebook pages. Pls.' Mem. at 20-21.
3. NIH's Keyword Filters
In addition to the default filters offered by Facebook and Instagram, as noted, NIH maintains customized lists of keywords that, when applied, result in comments containing those keywords being hidden from public view. A number of those keywords relate to the topic of animal testing. Jt. Stip. ¶ 58. In April 2021, NIH blocked comments on its Facebook page containing approximately 100 specified keywords, including the following fifteen keywords (and variations) related to animal testing:
1. PETA, PETALatino
Soumi and Harlow are the last names of scientists who have conducted primate experiments. Stephen Soumi is the Chief of the Laboratory of Comparative Ethology at the Eunice Kennedy Shriver National Institute of Child Health and Human Development, and Harry Harlow is a deceased psychologist who conducted maternal deprivation experiments on monkeys. Jt. Stip. ¶ 58 nn.9-10.
3. Animal(s), animales, animalitos
4. Cats, gatos
5. Chimpanzee(s), chimp(s)
6. Hamster(s)
7. Marmoset(s)
8. Monkey(s), “monkies”
9. Mouse, mice
10. Primate(s)
11. Sex experiments
12. Cruel, cruelty
13. Revolting
14. Torment(ing)
15. Torture(s), torturing
Compl., Attachment 1, Dep't of Health & Human Servs. Final FOIA Response Regarding Facebook Page (April 29, 2021) at 4, ECF No. 1-1. Other blocked keywords include words that indicated external links, such as “.org” and “www”; profanity; strings of numbers; the names of illicit drugs such as “cannabis” and “marijuana”; and a hashtag possibly associated with antivaccine activists, “#believemothers.” Id. On December 3, 2021, NIH removed “PETA” and “PETALatino” from this list but the remainder of the above-listed words are still blocked. Jt. Stip. ¶ 60.
NIH's custom keyword filter on Instagram contains fewer than thirty blocked keywords, nearly all related to animal testing. As of April 2021, NIH blocked comments including the following thirteen keywords (and variations) on Instagram:
1. #stopanimaltesting
2. #stoptesting
3. #stoptestingonanimals
4. Animal, animals
5. Chimpanzees, chimps
6. Experiment
7. Hurt, Hurting
8. Kill
9. Monkey, Monkeys
10. PETA
11. Stop
12. Test, Testing, Testing Facility
13. Torture, Torturing
Compl., Attachment 2, Dep't of Health & Human Servs. Final FOIA Response Regarding Instagram Account (April 29, 2021) at 4, ECF No. 1-2. Other blocked keywords include profanity and monkey emojis. Id. On December 3 and 7, 2021, NIH removed “PETA,” “#stopanimaltesting,” “#stoptesting,” and “#stoptestingonanimals” from its keyword list. Jt. Stip. ¶ 60.
As a result of these keyword filters, many comments that plaintiffs have attempted to post on NIH's Facebook and Instagram accounts have been hidden from public view. For example, on May 3, 2021, plaintiff Krasno commented on an NIH Instagram post of a cell infected with COVID-19, writing that “It's time we had an open conversation about all the animal testing you fund. What a waste of life and resources.” Id. ¶ 73. This comment was hidden from public view. Compare id., Ex. 33, Krasno View of Instagram Post, ECF No. 28-33 (showing plaintiff's comment), with id., Ex. 34, Public View of Instagram Post, ECF No. 28-34 (plaintiff's comment not visible). The comments that do appear on this post include reactions like “Fascinating,” questions about the post's contents, and the comment “Para de matar animals [Stop killing animals].” Id., Ex. 34, Public View of Instagram Post, ECF No. 28-34. In another example, Krasno responded to an NIH Facebook post about PTSD with the comment that she was diagnosed with PTSD after working at an NIH-funded laboratory that conducted primate testing; this post was also hidden from public view-but when Krasno retyped the same message with spaces between each letter in the blocked words (typing, for example, “t o r t u r e a n i m a l s”), the post was publicly viewable. Compare id., Ex. 41, Krasno View of Facebook Post, ECF No. 29-41 (showing both of Krasno's posts in standard formatting and with the spacing) with id., Ex. 42, Public View of Facebook Post, ECF No. 29-42 (showing only Krasno's post in which the blocked words were broken up with spaces). Employees of PETA and individual plaintiff Hartkopf similarly experienced the blocking of their comments containing keywords on NIH's social media pages, but they too were able to elude the keyword filters by modifying the spelling of blocked words. See Jt. Stip. ¶¶ 62-68, 80-85.
B. Procedural Background
In September 2021, plaintiffs initiated the instant lawsuit, claiming that defendants' practice of blocking comments on NIH's Facebook and Instagram accounts based on the use of certain words related to animal testing violated plaintiffs' First Amendment rights. See generally Compl., ECF No. 1. The parties subsequently submitted, on February 11, 2022, a joint stipulation of facts setting out an extensive narrative of agreed-upon facts in the matter. See generally Jt. stip. Plaintiffs then moved for summary judgment, See Pls.' Mot. Summ. J., ECF No. 30, and defendants have cross-moved for summary judgment, See Defs.' Cross-Mot. Summ. J., ECF No. 31. Both motions are now ripe for resolution.
II. LEGAL STANDARD
Under Federal Rule of Civil Procedure 56, “[a] party is entitled to summary judgment only if there is no genuine issue of material fact and judgment in the movant's favor is proper as a matter of law.” Soundboard Ass'n v. FTC, 888 F.3d 1261, 1267 (D.C. Cir. 2018) (quoting Ctr. for Auto Safety v. Nat'l Highway Traffic Safety Admin., 452 F.3d 798, 805 (D.C. Cir. 2006)); see also FED. R. CIV. P. 56(a). When parties file cross-motions for summary judgment, each motion is viewed separately, in the light most favorable to the non-moving party, with the court determining, for each side, whether the Rule 56 standard has been met. See Baylor v. Mitchell Rubenstein & Assocs., P.C., 857 F.3d 939, 952 (D.C. Cir. 2017) (when considering “cross-motions for summary judgment, [courts] must accord both parties the solicitude owed non-movants”); see also CEI Wash. Bureau, Inc. v. DOJ, 469 F.3d 126, 129 (D.C. Cir. 2006) (per curiam) (noting that “[i]t is of no moment that the parties filed cross-motions for summary judgment and that neither party explicitly argued that there are genuine disputes about material facts” because “[a] cross-motion for summary judgment does not concede the factual assertions of the opposing motion”); Sherwood v. Wash. Post, 871 F.2d 1144, 1147 n.4 (D.C. Cir. 1989) (“The rule governing cross-motions for summary judgment . . . is that neither party waives the right to a full trial on the merits by filing its own motion; each side concedes that no material facts are at issue only for the purposes of its own motion.” (quoting McKenzie v. Sawyer, 684 F.2d 62, 68 n.3 (D.C. Cir. 1982))).
III. DISCUSSION
The First Amendment provides, in relevant part, that “Congress shall make no law . . . abridging the freedom of speech . . . or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.” First Amendment freedoms “are delicate and vulnerable, as well as supremely precious in our society.” NAACP v. Button, 371 U.S. 415, 433 (1963). Thus, the Supreme Court has warned that “[b]ecause First Amendment freedoms need breathing space to survive, government may regulate in the area only with narrow specificity.” Id. Nonetheless, “[n]othing in the Constitution requires the Government freely to grant access to all who wish to exercise their right to free speech on every type of Government property without regard to the nature of the property or to the disruption that might be caused by the speaker's activities.” Cornelius v. NAACP Legal Def. & Educ. Fund, Inc., 473 U.S. 788, 799-800 (1985). Acknowledging that the government, “no less than a private owner of property, has power to preserve the property under its control for the use to which it is lawfully dedicated,” id. at 800 (quoting Greer v. Spock, 424 U.S. 828, 836 (1976)), the Supreme Court “has adopted a forum analysis as a means of determining when the Government's interest in limiting the use of its property to its intended purpose outweighs the interest of those wishing to use the property for other purposes,” id.; see also Hodge, 799 F.3d at 1157 (explaining that “the Supreme Court's ‘forum analysis'” is used to “determine when a governmental entity, in regulating property in its charge, may place limitations on speech” (quoting Christian Legal Soc'y Chapter of the Univ. of Cal., Hastings Coll. of the Law v. Martinez, 561 U.S. 661, 679 (2010) (cleaned up))).
Here, defendants rightly do not dispute that plaintiffs' comments censored by the keyword filters were protected by the First Amendment. See generally Defs.' Opp'n.; Defs.' Reply Supp. Cross-Mot. Summ. J. (“Defs.' Reply”), ECF No. 37. Consequently, the parties urge that a forum analysis be applied to NIH's Facebook and Instagram comment threads as the threshold issue in evaluating the constitutionality of the agency's use of keyword filters. That forum analysis of NIH's Facebook and Instagram comment threads is addressed first and, upon determining that the threads constitute a limited public forum, NIH's keyword blocking practice is evaluated for whether that practice is reasonable in light of the forum's purpose, and whether it is viewpoint-neutral.
A. Forum Analysis
In conducting a forum analysis, the Court “proceed[s] in three steps: first, determining whether the First Amendment protects the speech at issue, then identifying the nature of the forum, and finally assessing whether the [government's] justifications for restricting . . . speech ‘satisfy the requisite standard.'” Mahoney v. Doe, 642 F.3d 1112, 1116 (D.C. Cir. 2011) (quoting Cornelius, 473 U.S. at 797). For purposes of this analysis, government property is divided into three categories-the traditional public forum, designated public forum, and nonpublic forum. Id. The category “determines what types of restrictions will be permissible” on that property. Initiative & Referendum Inst. v. U.S. Postal Serv., 685 F.3d 1066, 1069 (D.C. Cir. 2012). First, a traditional public forum, such as a public street or park, is government property “that has ‘by long tradition or by government fiat . . . been devoted to assembly and debate.'” Id. (quoting Perry Educ. Ass'n v. Perry Local Educators' Ass'n, 460 U.S. 37, 45 (1983)); see also Snyder v. Phelps, 562 U.S. 443, 456 (2011) (“[W]e have repeatedly referred to public streets as the archetype of a traditional public forum, noting that time out of mind public streets and sidewalks have been used for public assembly and debate.” (internal quotation marks and citations omitted)). Any restriction based on the content of speech in a public forum “must satisfy strict scrutiny, that is, the restriction must be narrowly tailored to serve a compelling government interest.” Christian Legal Soc'y Chapter of the Univ. of Cal., 561 U.S. at 679 n.11 (internal quotation marks and citations omitted); Pleasant Grove City v. Summum, 555 U.S. 460, 469 (2009) (noting that in a traditional public forum, “[r]easonable time, place, and manner restrictions are allowed . . . but any restriction based on the content of the speech must satisfy strict scrutiny, that is, the restriction must be narrowly tailored to serve a compelling government interest” (citations omitted)).
Second, a designated public forum is “property that ‘the State has opened for use by the public as a place for expressive activity.'” Initiative & Referendum Inst., 685 F.3d at 1069 (quoting Perry Educ. Ass'n, 460 U.S. at 45). “Although a state is not required to indefinitely retain the open character of the facility, as long as it does so it is bound by the same standards as apply in a traditional public forum.” Perry Educ. Ass'n, 460 U.S. at 46. Designated public fora are created only when the government “intentionally open[s] a nontraditional forum for public discourse”-not “by inaction or by permitting limited discourse.” Cornelius, 473 U.S. at 802. To determine whether the government intended to designate a place as a public forum, courts examine the “policy and practice of the government,” as well as “the nature of the property and its compatibility with expressive activity.” Id.
Finally, a nonpublic forum is “not by tradition or designation a public forum.” Archdiocese of Wash. v. Wash. Metro. Area Transit Auth., 897 F.3d 314, 322 (D.C. Cir. 2018). In a nonpublic forum, the government “may reserve the forum for its intended purposes, communicative or otherwise, as long as the regulation on speech is reasonable and not an effort to suppress expression merely because public officials oppose the speaker's view.” Id. (quoting Perry Educ. Ass'n, 460 U.S. at 46). The stepchild of this category is the limited public forum, which is defined as a forum that “is limited to use by certain groups or dedicated solely to the discussion of certain subjects,” Am. Freedom Def. Initiative v. King Cnty., Wash., 577 U.S. 1202, 1202 (2016) (Mem.) (Thomas, J., dissenting from denial of cert.) (quoting Summum, 555 U.S. at 470); accord Walker v. Texas Div., Sons of Confederate Veterans, Inc., 576 U.S. 200, 215 (2015), and is subject to the same speech protections as the nonpublic forum, Price v. Garland, 45 F.4th 1059, 1068 (D.C. Cir. 2022).
The Fourth Circuit recently observed that “there is considerable confusion over whether there are three or four types of free-speech forums.” White Coat Waste Project v. Greater Richmond Transit Co., 35 F.4th 179, 196 n.13 (4th Cir. 2022). Courts of Appeals have variously treated the limited public forum as a sub-category of the designated public forum, see, e.g., Hopper v. City of Pasco, 241 F.3d 1067, 1074-75 (9th Cir. 2001); Johnson v. Perry, 859 F.3d 156, 171 (2d Cir. 2017); as synonymous with the nonpublic forum, see Victory Through Jesus Sports Ministry Found. v. Lee's Summit R-7 Sch. Dist., 640 F.3d 329, 334-35 (8th Cir. 2011); or simply as a distinct fourth category, see, e.g., OSU Student All. v. Ray, 699 F.3d 1053, 1062 (9th Cir. 2012); see also Walker, 576 U.S. at 215-16 (addressing the limited public forum as a distinct category from designated public fora and nonpublic fora). Whatever the proper categorization, the First Amendment analysis is the same for the limited public and nonpublic fora. See Price, 45 F.4th at 1068. For clarity, this memorandum opinion uses the term “limited public forum.”
Thus, if the comment threads are determined to be a nonpublic forum or limited public forum, as defendants urge, “it is . . . black-letter law that . . . the government . . . can exclude speakers on the basis of their subject matter, so long as the distinctions drawn are viewpoint neutral and reasonable in light of the purpose served by the forum.” Davenport v. Wash. Educ. Ass'n, 551 U.S. 177, 189 (2007). By administering a limited public forum, NIH would have the most leeway in limiting the scope of the comments on its social media pages. Indeed, “[l]imitations on expressive activity conducted on this last category of property must survive only a much more limited review” than would be the case for public or designated public fora. Int'l Soc'y for Krishna Consciousness v. Lee, 505 U.S. 672, 678-79 (1992). On the other hand, if the comment threads are determined to be a “designated public forum,” as plaintiffs urge, any content-based restrictions on expressive activity must satisfy strict scrutiny, Archdiocese, 897 F.3d at 322, requiring that “the restriction must be narrowly tailored to serve a compelling government interest,” Summum, 555 U.S. at 469. In either case, defendants' restrictions would be unconstitutional if they are not viewpoint-neutral, that is, if they reveal “an effort to suppress expression merely because public officials oppose the speaker's view.” Minnesota Voters All. v. Mansky, 138 S.Ct. 1876, 1885 (2018) (quoting Perry Educ. Ass'n, 460 U.S. at 46).
The comment threads at issue are limited public fora: virtual spaces opened by the government to the public for the purpose of the discussion of only certain subjects. To distinguish between a limited public forum and a designated public forum, the “touchstone” is the government's “‘intent in establishing and maintaining' that forum.” Bryant v. Gates, 532 F.3d 888, 895-96 (D.C. Cir. 2008) (quoting Stewart v. District of Columbia Armory Bd., 863 F.2d 1013, 1016 (D.C. Cir. 1988)). “[O]bjective indicia of intent” include the nature of the property and its compatibility with expressive activity, as well as the “consistent policy and practice of the government.” Id. (emphasis in original) (quoting Stewart, 863 F.2d at 1016-17). As to this second indicia of intent, to find a limited public forum, courts look for “[s]tandards for inclusion and exclusion” that are “unambiguous and definite.” Garnier v. O'Connor-Ratcliff, 41 F.4th 1158, 1178 (9th Cir. 2022) (internal quotations omitted); see also Christ's Bride Ministries, Inc. v. Se. Penn. Transp. Auth., 148 F.3d 242, 248-49 (3d Cir. 1998) (noting that a designated public forum is not created merely because the government allows “limited discourse,” requiring examination of the restrictions on the forum).
Thus, to create a limited public forum, the government must “set prospective, categorical, subject-matter rules by which to evaluate private speech” in order to provide “public notice of what speech is permissible and constrain[] the discretion of government actors to pick favorites on an ad hoc basis.” Archdiocese, 897 F.3d at 337. In the physical environment, limited public fora are exemplified by school board meetings and town halls, in which citizens are encouraged to engage in dialogue-but only to the extent that the dialogue is on-topic to the meeting's agenda, lest the meetings inevitably devolve into chaos. See, e.g., Steinburg v. Chesterfield Cnty. Plan. Comm'n, 527 F.3d 377, 385 (4th Cir. 2008) (county planning commission was limited public forum); Barrett v. Walker Cnty. Sch. Dist., 872 F.3d 1209, 1225 (11th Cir. 2017) (same for school board meeting); Galena v. Leone, 638 F.3d 186, 198-99 (3d Cir. 2011) (same for public commentary portion of county council meeting). The comment threads at issue are close analogies to such government-hosted gatherings in the virtual environment.
As to the nature of the forum, social media is undoubtedly the site of “a wide array of protected First Amendment activity on topics ‘as diverse as human thought,'” constituting “vast democratic forums.” Packingham v. North Carolina, 582 U.S. 98, 104-105 (2017) (quoting Reno v. ACLU, 521 US. 844, 852, 868 (1997)). Public comment threads on Facebook and Instagram are “compatible with expressive activity,” as plaintiffs argue, Pls.' Mem. at 27-30, but this is far from the end of the inquiry. “[T]he mere fact that an instrumentality is used for the communication of ideas does not make a public forum.” Perry Educ. Ass'n, 460 U.S. at 49 n.9; see also Defs.' Opp'n at 26-27.
Plaintiffs urge that the “comment threads' undisputed compatibility with expressive activity cannot be discounted,” Pls.' Reply Supp. Mot. Summ. J. & Opp'n Defs.' Cross-Mot. Summ. J. (“Pls.' Opp'n”) at 17, ECF No. 34, in reliance on the following quoted language from One Wisconsin Now v. Kremer that “[i]f defendants truly had no intention to create a space for public interaction and discourse, they would not have created public [social media] accounts in the first place.” Id. (quoting 354 F.Supp.3d 940, 954 (W.D. Wisc. 2019)). This is an overly facile view of social media. The fact that the government created a space for public interaction does not mean that this platform was intended for unbridled public interaction about any topic imaginable.
The cases plaintiffs cite in arguing that “nearly every court . . . has concluded that the interactive components of government-run social media accounts are public forums,” Pls.' Mem. at 33, underscore the key role of the next consideration-the government's policy and practice with respect to the forum. Where courts have found interactive spaces on social media to constitute publica fora, the government actors “made [their] interactive features accessible to the public without limitation.” Knight First Amendment Inst. v. Trump, 928 F.3d 226, 237 (2d Cir. 2019), vacated as moot, 141 S.Ct. 1220 (2021) (Mem.) (regarding former President Trump's Twitter account); see also Garnier, 41 F.4th at 1178-79 (school board trustees' social media pages with “[n]o policy statement” restricting the form or content of comments); Pls.' Mem. at 33 (citing Davison v. Randall, 912 F.3d 666, 687 (4th Cir. 2019) (Facebook page administered by county board of supervisors chair expressly open to “ANY” user to post on “ANY issues”); Faison v. Jones, 440 F.Supp.3d 1123, 1128, 1134-35 (E.D. Cal. 2020) (county sheriff's Facebook page that used the profanity filter and had banned users and blocked comments, but appeared to have no stated engagement policy); & One Wisconsin Now v. Kremer, 354 F.Supp.3d 940, 954 (W.D. Wisc. 2019) (Twitter accounts for which administrating Wisconsin state officials took no steps to “limit[] access by the general public” or otherwise control the interactive space)). Cf. Krasno v. Mnookin, Case No. 21-cv-99, 2022 WL 16635246, *12 (W.D. Wisc. Nov. 2, 2022) (holding that social media pages' comment threads were limited public fora or nonpublic fora due to policy prohibiting “off-topic” comments); Charudattan v. Darnell, 2019 WL 12043587, *6 (N.D. Fla. Feb. 7, 2019) (same); Davison v. Plowman, 247 F.Supp.3d 767, 776 (E.D. Va. 2017), aff'd by 715 Fed.Appx. 298 (4th Cir. 2018) (same). But see Kimsey v. City of Sammamish, 574 F.Supp.3d 911, 918-20 (W.D. Wash. 2021) (concluding Facebook page prohibiting “off topic” comments but with inconsistent enforcement was designated, rather than limited, public forum).
Few of these cases proceeded to determining what type of public forum the social media pages constituted, instead finding that viewpoint discrimination rendered the censorship at issue unconstitutional regardless. See Knight, 928 F.3d at 237-38; Davison, 912 F.3d at 687; Faison, 440 F.Supp.3d at 1135-36 (in preliminary injunction context, holding that plaintiffs were likely to succeed in showing viewpoint discrimination).
Here, by contrast, NIH did not throw open its Instagram and Facebook pages to indiscriminate public engagement. Instead, the agency has maintained a publicly viewable commenting policy with “prospective, categorical, subject-matter rules,” Archdiocese, 897 F.3d at 337, requiring that comments remain on-topic-a key difference between the social media comment threads at issue and those considered in plaintiffs' proffered above-cited cases. See supra Part I.A.1 (describing the relevant guidelines). Garnier provides an illustrative contrast. See Pls.' Notice of Suppl. Authority, ECF No. 38 (notifying Court of Ninth Circuit decision); Defs.' Resp. to Pls.' Notice of Suppl. Authority at 3-4, ECF No. 39 (distinguishing Garnier's social media pages for lacking commenting rules). In this recent Ninth Circuit case, two school board trustees who maintained social media pages on Twitter and Facebook to interact with their constituents “never adopted any formal rules of decorum or etiquette for their pages,” 41 F.4th at 1178; see also id. at 1165. When the Garniers, two especially vocal parents critical of the Board, began posting long and repetitive comments about the school district to the trustees' social media pages-in one instance, posting nearly identical comments on 42 separate Facebook posts made by a trustee-the trustees began deleting or hiding the Garniers' comments on Facebook. Id. at 1166. Eventually, the trustees blocked the Garnier parents altogether from their pages. The trustees weakly urged that their actions were pursuant to “an unspoken policy” against repetitive comments, which the Ninth Circuit dismissed as failing to “satisfy the requirement that standards for inclusion and exclusion must be unambiguous and definite.” Id. at 1178 (cleaned up). The nature of the government actors' control over the space proved dispositive, however: when the trustees later effectively blocked all comments on Facebook by using keyword filters for common words, they “‘exercise[d] the clear and consistent control' over the interactive portions of their Facebook pages [necessary] ‘to maintain a limited public forum.'” Id. at 1179 (quoting Hopper v. City of Pasco, 241 F.3d 1067, 1080 (9th Cir. 2001)).So too here: NIH's commenting policy is public-not a mere “unspoken policy”-and clearly prohibits off-topic comments.
Defendants urge that Garnier supports their arguments because NIH uses keyword filters, the use of which the Ninth Circuit found to have rendered the trustees' Facebook pages limited public fora. This argument is unconvincing: the trustees used keyword filters effectively to block all comments on their pages-one of the trustees added around 2,000 commonly used words to his filter, 41 F.4th at 1166. NIH's keyword filter list, by contrast, is far more targeted and does not gag public discourse entirely. Defendants' discounting of the expansiveness of the trustees' keywords as “irrelevant” is blind to the practical realities of the trustees' use of such an extensive keyword filter to block all comments. Defs.' Resp. to Suppl. Notice at 3, ECF No. 39. Nor does defendants' argument that NIH's keyword filters are similarly broad go the distance, where NIH's keyword filters are plainly tailored to block primarily animal-testing-related commentary, external websites, and references to marijuana. Id.
Plaintiffs point out that, prior to the filing of the instant lawsuit, links to NIH's commenting guidelines were not provided on the agency's social media pages. Pls.' Mem. at 32 n.13; Pls.' Opp'n at 18. True, but the comment guidelines were available on NIH's website, under a page titled “Social Media and Outreach.” Defs.' Reply at 11-12. Moreover, the pages did link to a privacy policy that also prohibited, inter alia, spam and “off-topic” speech. See Jt. Stip. ¶ 54.
Plaintiffs contend that, whatever NIH's stated commenting guidelines, those guidelines were “essentially unenforced,” rendering them a mere “paper policy.” Pls.' Reply Supp. Mot. Summ. J. & Opp'n Defs.' Cross-Mot. Summ. J. (“Pls.' Opp'n”) at 20, ECF No. 34. Certainly, “[m]ere statements of policy, if consistently contradicted by practice, are not dispositive.” Stewart, 863 F.2d at 1021. At the same time, courts' examination of the government's actual practice in enforcing its policy is not a gotcha game; rather it is a tool to smoke out the government's “intent” with regard to the forum's purpose-the “touchstone” of the forum inquiry. Bryant, 532 F.3d at 895-96. As a result, the government is not held to the standard of perfection in administering its policy. Consequently, even if enforcement is “erratic,” so long as defendants “consistently limit[ed] [speech] it saw as in violation of its policy . . . [it] evidenced its intent not to create a designated public forum.” Ridley v. Mass. Bay Transp. Auth., 390 F.3d 65, 78 (1st Cir. 2004). See also Hodge, 799 F.3d at 1162 (holding that Supreme Court Police's failure to enforce the relevant policy perfectly did not transform the Supreme Court plaza into a public forum, given the “usual practice of strict enforcement”).
Where courts have held that inconsistent enforcement of a governing policy would favor a finding of a designated public forum, rather than a limited one, they often have done so in the context of ambiguous, contradictory, or non-existent ex ante guidelines. The D.C. Circuit's decision in Stewart illustrates such a determination. There, the administrators of RFK Stadium in Washington, D.C. had removed banners proclaiming “John 3:16,” purportedly pursuant to the stadium's policy prohibiting signs unrelated to the stadium's scheduled events. 863 F.2d at 1014. At the same time, contradictory evidence was presented as to the prevailing sign policy, since the stadium had apparently also expressed a policy of permitting all banners except the obscene and commercial. Id. at 1019. Confronted with conflicting evidence of the stadium's policy-none publicly stated before the censorship of the religious banners-the D.C. Circuit advised the trial court, on remand, that “[r]esolving the dimensions of the policy . . . is crucial to the threshold question of whether or not RFK Stadium is a public forum.” Id. at 1021. There, the key concern was that the stadium may have used an on-paper-only policy as “ex post justification” for the exclusion of the religious banners. Id. at 1020; see also Christ's Bride Ministries, Inc., 148 F.3d at 250-53 (finding a designated public forum where public transit authority retained authority to exclude advertisements for any reason, but then permitted “at least 99% of all ads” without objection, including those on the same topic as the excluded antiabortion ad at issue); Hopper, 241 F.3d at 1078 (finding a city hall art gallery a designated public forum where city broadly prohibited “controversial” art, but “established no specific criteria for exclusion of art from the program” and did not always exclude controversial art).
Where NIH has attempted to enforce its policy, even if off-topic comments still appear on its pages' comment threads, this failure of perfect enforcement does not fatally undermine its intent for the comment threads to pertain to only relevant subjects. Plaintiffs have pointed out a number of comments in the record that are off-topic but not removed by NIH's administrators, such as a political post urging “BRING BACK CHRIS CUOMO” in response to NIH's post about brain activity, See Jt. Stip., Ex. 5, NIH Facebook Post (Dec. 2, 2021), ECF No. 28-5, and nonsensical comments in response to a post about “Public Service Recognition Week,” see id., Ex. 38, NIH Facebook Post (May 5, 2021), ECF No. 28-38 (comments saying “hi,” “resist devil really love no one your self,” and “You are so friendly, Prety”).
Here, the particular context of cyberspace is particularly relevant: “The interactive comment threads on social media pages are a unique type of fora, able to host [] an almost unlimited amount of expression by an unlimited number of unknown users.” Mnookin, 2022 WL 16635246 at *12. This infinite supply of commentary from individuals anywhere in the world makes “100% consistency” an impossible task, id., such that the failure of perfect policing of non-compliant content on the agency's social media pages does not evince the government's intent. In addition, NIH has exercised control over the comment threads via manual moderation-which though stopped in response to this lawsuit, the agency intends to resume- and keyword filtering, the latter of which targets off-topic posts regarding animal testing as well as comments containing profanity or linking to external websites in violation of its policy. See Defs.' Reply at 10.
Finally, requiring perfect enforcement of commenting guidelines for the government to maintain social media pages' status as limited public fora would result in perverse outcomes for free expression. As the Supreme Court has explained, the distinction between the designated public forum and the limited public forum (or nonpublic forum) “furthers First Amendment interests” by “encourag[ing] the government to open its property to some expressive activity in cases where, if faced with an all-or-nothing choice, it might not open the property at all.”
Arkansas Educ. Television Comm'n v. Forbes, 523 U.S. 666, 680 (1998). Holding that a forum which the government intends to limit to only certain subjects is in fact a public forum would “result in less speech, not more,” because “faced with the prospect of cacophony, on the one hand, and First Amendment liability, on the other, a [platform] might choose not to” permit speech at all. Id. at 680-81. The outcome in Garnier illustrates this risk: the Ninth Circuit held that the school board trustees had created a limited public forum only once the trustees effectively closed their social media pages to public discourse by precluding all comments. 41 F.4th at 1179.
Defendants go so far as to argue in favor of a “presumption against finding government-run forums to be public forums” based on this language in Arkansas Educational Television Commission, but they but cite no cases that actually espouse such a broad rule. Defs.' Opp'n at 17-18. Plaintiffs vigorously criticize this argument, urging that Arkansas Educational Television Commission merely explains “why courts scrutinize restrictions in limited or nonpublic forums less skeptically,” Pls.' Opp'n at 15, and insisting that courts engage in forum analysis “based on objective factors, not pro-government presumptions.” Id. Plaintiffs' criticism is well-placed: objective factors must guide the forum analysis, with a focus on the First Amendment principles undergirding the forum doctrine. Thus, defendants' proffered presumption to avoid altogether a forum analysis is a shortcut that is legally unsupportable.
Resultantly, the comment threads in NIH's Facebook and Instagram pages are limited public fora.
B. Reasonableness
Restrictions on speech in limited public fora need only be “reasonable in light of the purpose which the forum at issue serves,” Perry Educ. Ass'n, 460 U.S. at 49-a significantly lighter burden than the strict scrutiny applied to traditional or designated public fora. “Implicit in the concept of the nonpublic forum is the right to make distinctions in access on the basis of subject matter and speaker identity,” which “distinctions may be impermissible in a public forum but are inherent and inescapable in the process of limiting a nonpublic forum to activities compatible with the intended purpose of the property.” Id. In making such distinctions, “[t]he [g]overnment's decision to restrict access to a nonpublic forum need only be reasonable; it need not be the most reasonable or the only reasonable limitation.” Cornelius, 473 U.S. at 808 (emphasis in original). The reasonableness standard is a “relatively low bar,” but it still “requires something more than the toothless ‘rational basis' test used to review the typical exercise of a state's police power.” Price, 45 F.4th at 1072 (citations omitted). In order to show that a speech restriction is “reasonable,” the government must show that its restraint: (1) furthers a “permissible objective;” and (2) contains “objective, workable standards” that are “capable of reasoned application.” Mansky, 138 S.Ct. at 1886, 1891-92.
NIH argues that the keyword filtering furthers the governmental interest of preventing “disruptive” comments that “may discourage interested citizens from viewing” the pages at issue. Defs.' Opp'n at 35. Filtering comments containing keywords related to animal testing, defendants urge, provides an “efficient, though imperfect, way of targeting impermissible comments,” Defs.' Reply at 25, particularly, they contend, because plaintiffs have alternative channels by which to make their opposition to animal testing heard, id. at 26; Defs.' Opp'n at 36-37. Plaintiffs counter that the keyword filtering is unreasonable by “target[ing] animal advocacy, and animal advocacy alone, in a way that is completely ‘unmoored' from the interests allegedly motivating its speech restrictions,” reflecting inconsistent enforcement of the commenting guidelines. Pls.' Opp'n at 25 (quoting Manksy, 138 S.Ct. at 1888); see also id. at 24-27.
NIH has strong legitimate interests in maintaining on-topic discussions in the comment threads of the agency's social media posts. These social media accounts are used to “‘communicate and interact with citizens' about agency-related updates and public health news,” Jt. Stip. ¶ 36 (quoting Web Policies and Notices, NATIONAL INSTITUTES OF HEALTH (Jan. 14, 2021), https://perma.cc/BC7R-ZFME). The clutter of off-topic comments detracts from the comment threads' purpose of fostering productive dialogue; after all, the “failure to effectively moderate a public discussion may be as deleterious to dialogue in [a limited public] forum as censorship.” Davison, 247 F.Supp.3d at 778. Examples of NIH's posts and their attached comment threads in the record illustrate the problem: for example, in response to an NIH post about caregiving for an Alzheimer's patient, the vast majority of comments advocated against animal testing, most prominently urging NIH to “FREE THE BEAGLES.” Jt. Stip., Ex. 15, NIH Instagram Post & Viewable Comments (Nov. 12, 2021), ECF No. 28-15. The few relevant comments interacting with the post-such as a commenter applauding caregivers as “the unsung heroes of medicine!”-are so buried in the repetitive animal advocacy comments as to be drowned out. See also id., Ex. 43, NIH Instagram Post & Viewable Comments (April 23, 2021), ECF No. 28-43 (NIH post about National DNA Day flooded with comments encouraging the closure of National Primate Research Centers, burying relevant comments such as those encouraging NIH to “provide more information about Henrietta Lacks” when discussing HeLa cells).
The Garnier Court held that the disruption threatened by online comments pales in comparison to the effect of irrelevant remarks in real-world forums, such as “physical city hall meetings, where there is limited time and space available for public remarks.” Garnier, 41 F.4th at 1181. Viewers of the social media pages, the Court said, “can, with the flick of a finger, simply scroll past repetitive or irrelevant comments.” Id. This is certainly correct, up to a point. Social media pages' seemingly infinite amount of space for public commentary does not mean, however, that administrators of those pages and the users who hope to view and engage in meaningful dialogue have the infinite time, resources and patience to sift through the cacophony to find relevant information.Confronted with a forum overwhelmed with irrelevant commentary, many users will simply choose not to engage or even view at all. NIH could “reasonably conclude” that the distracting comments detracted from the purpose of the fora, Cornelius, 473 U.S. at 809.
The record demonstrates that on-topic engagement was also stymied by animal advocates' commentary on NIH's Twitter. In one instance, NIH hosted a “chat” on Twitter on the topic of brain health, to which members of the public replied that they could not “follow chat among distraction,” which another person described as a “bombardment” of the chat that was “disrespectful to other participants.” Jt. Stip. ¶ 87. Another member of the public noted that, as a result of the flood of animal advocacy commentary, she “missed” NIH's tweets about “retina issues and Alzheimer's” that were “important for [her] family.” Id. This data point provides evidence that off-topic and repetitive comments can be disruptive in cyberspace, even where the physical restraints of in-person engagement do not apply. Although NIH's Twitter page is not at issue in the present case, this evidence is useful for that point alone. See Defs.' Resp. Suppl. Auth. at 5, ECF No. 39.
Plaintiffs correctly note that the Garnier Court concluded that curtailing off-topic comments was not a “significant governmental interest,” 41 F.4th at 1181-82, in the specific context of a designated public forum analysis, in which the far more demanding standard of strict scrutiny applied. See Pls.' Notice of Suppl. Authority, ECF No. 38. The Garnier Court suggested that, while the page administrators could not ban the disruptive parents entirely from the social media pages, the administrators could hide or delete the comments, or at least establish clear rules of engagement. 41 F.4th at 1182.
Plaintiffs' main contention is that the keyword filtering is unreasonable due to its results, allowing inconsistent enforcement of NIH's stated commenting guidelines, particularly the prohibition of off-topic comments. See Pls.' Mem. at 46; Pls.' Opp'n at 24-27. As plaintiffs correctly observe, NIH's keyword filtering mechanism is both overinclusive, by excluding comments about animals even when those comments may be relevant to a specific post by NIH, and underinclusive, by hiding off-topic comments protesting animal testing but leaving visible off-topic comments about other subjects. Pls.' Opp'n at 26-27. Plaintiffs cite Mansky for the principle that “unfair or inconsistent enforcement,” 138 S.Ct. at 1891, renders speech restrictions unreasonable. Pls.' Mem. at 46; Pls.' Opp'n at 25, 27. This argument merely reflects a repackaging of the same inconsistent-enforcement justification already rejected in the forum analysis, however, and the argument fails in this stage for parallel reasons. In Mansky, the Supreme Court held that a Minnesota regulation prohibiting all “political” apparel in polling places on Election Day was unconstitutional as an unreasonable restriction on speech in a nonpublic forum. Mansky, 138 S.Ct. at 1888. “The crux of the Court's decision was that the State's discretion in enforcing the statute had to be ‘guided by objective, workable standards.' Because the unqualified ban on ‘political' apparel did not provide those standards, it was unreasonable.” Am. Freedom Def. Initiative v. Wash. Metro. Transit Auth. (“AFDI”), 901 F.3d 356, 372 (D.C. Cir. 2018) (quoting Mansky, 138 S.Ct. at 1891). Much like the D.C. Circuit's skepticism of the seemingly contradictory banner policy in Stewart, Mansky's “indeterminate prohibition” presented an invidious “opportunity for abuse.” Mansky, 138 S.Ct. at 1891. See also Zukerman v. U.S. Postal Serv., 961 F.3d 431, 448-49 (D.C. Cir. 2020) (applying this distillation of Mansky to hold that ban of “political” content in stamps was unreasonable); see also People for the Ethical Treatment of Animals v. Gittens, 215 F.Supp.2d 120, 131 (D.D.C. 2002) (holding that arts commission's exclusion of PETA submission from art show, a limited public forum, was “inherently unreasonable” where commission's articulated standard was the amorphous question of whether submissions were “art”-a standard then inconsistently applied).
Plaintiffs cite Gittens in urging that NIH's imperfect enforcement of its commenting policy revealed that the agency had engaged in unconstitutional viewpoint discrimination, though this case mainly dealt with the reasonableness inquiry. See Pls.' Mem. at 39; Pls.' Opp'n at 12. In arguing that NIH's “dramatic underbreadth” in enforcement “supports the inference that the NIH is suppressing animal testing criticism rather than enforcing a viewpoint-neutral rule,” Pls.' Opp'n at 12, plaintiffs repackage the inconsistent-enforcement argument for a third time. It fails for all the same reasons. The Court addresses plaintiffs' other arguments that the keyword filtering policy was viewpoint-motived infra in Part III.C.
Unlike in Mansky and Zukerman, the enforcers of NIH's policy against off-topic commentary-here, NIH's social media managers-have a clear mandate to apply the “objective, workable standard[]” of topicality. Mansky, 138 S.Ct. at 1881. As the Mnookin Court explained, “although reasonable people may have different degrees of tolerance for when something is ‘not germane' or ‘off topic,' the terms as commonly understood are sufficiently objective to preclude wildly divergent applications.” 2022 WL 16635246, at *17. See also Davison, 247 F.Supp.3d at 778 (holding that prohibition on off-topic comments was not unconstitutionally vague). As to the keyword filtering and manual deletion practice used to enforce this policy, plaintiffs have not demonstrated that NIH's enforcement is conducted in “arbitrary and unreasonable ways.” Archdiocese, 897 F.3d at 330 (quoting brief of appellant). To the contrary, NIH's keyword filter is a reasonable approach to maintaining orderly, on-topic comment threads. This is so for two reasons.
First, the prevalence of animal testing-related keywords in its filters is not the result of arbitrariness. As defendants have demonstrated-and plaintiffs have not meaningfully disputed, See Pls.' Opp'n at 11-12 -comments about animal testing comprised large shares of off-topic commentary on many of NIH's social media posts, becoming a highly prevalent irrelevant topic in violation of the commenting guidelines. Defs.' Opp'n at 12-14. Contrary to plaintiffs' assertion that “there is nothing ‘inherently disruptive' about animal advocacy that makes it ripe for categorical treatment,” Pls.' Opp'n at 27 (quoting United States v. Kokinda, 497 U.S. 720, 732 (1990)), this category of comments stood out for NIH's focused enforcement because this was a flagrant offending subject matter in the comment threads.
Second, the Supreme Court's guidance that reasonableness “must be assessed in the light of the purpose of the forum and all the surrounding circumstances,” AFDI, 901 F.3d at 370 (quoting Cornelius, 473 U.S. at 809), is particularly apt in the context of social media comment threads. Here, defendants' arguments regarding the impracticability of comment-by-comment review hold sway. See Defs.' Opp'n at 36. Each of NIH's posts on Facebook or Instagram could garner hundreds of comments. See, e.g., Jt. Stip., Ex. 25, Facebook Post & Viewable Comments (May 27, 2021) (576 comments); id., Ex. 4, Facebook Post & Viewable Comments (Nov. 15, 2021) (429 comments); id., Ex. 1, Facebook Post & Viewable Comments (June 4, 2021) (334 comments); id., Ex. 21, Facebook Post & Viewable Comments (Sept. 24, 2020) (328 comments). The accessibility and ease of internet engagement presents commenting policy enforcers with a Sisyphean task-one for which the standards of consistency to which the government is held cannot be so exacting as might be possible on a smaller scale, such as the enforcement of a public transit's advertisement policy, as in Archdiocese and AFDI. Otherwise, if the standard is perfectly consistent enforcement, it is hard to imagine any social media commenting policy that would survive the test of reasonableness without severely throttling the public's ex ante access to the forum.
Finally, plaintiffs' access to alternative channels weighs in favor of the reasonableness of the restriction. See Cornelius, 473 U.S. at 809 (“Rarely will a nonpublic forum provide the only means of contact with a particular audience.”). Plaintiffs defend the importance of their ability to comment directly on NIH's social media pages by arguing that such comments (1) “demonstrate to the NIH the size and passion of the animal rights movement,” Jt. Stip. ¶ 88, (2) reach “a large number of people who have a mutual interest in scientific research and the NIH,” id. ¶ 90, and (3) conveys to “NIH's supporters and followers . . . how animals . . . are treated in laboratories that are directly funded by the NIH,” id. ¶ 89. Baldly violating NIH's commenting guidelines by filling its social media pages with animal-testing-related comments may be the most efficient way of achieving those goals, but defendants rightly note that the “First Amendment does not demand unrestricted access to a nonpublic forum merely because use of that forum may be the most efficient means of delivering the speaker's message.” Defs.' Opp'n at 37 (quoting Cornelius, 473 U.S. at 809). Plaintiffs can achieve the same goals by creating other Facebook or Instagram accounts dedicated to their advocacy.
C. Viewpoint Neutrality
The final requirement of speech restrictions in nonpublic fora is viewpoint neutrality. “The government may not discriminate against speech based on the ideas or opinions it conveys.” Iancu v. Brunetti, 139 S.Ct. 2294, 2299 (2019). “[I]n determining whether the State is acting to preserve the limits of the forum it has created so that the exclusion of a class of speech is legitimate, [courts] have observed a distinction between, on the one hand, content discrimination, which may be permissible if it preserves the purposes of that limited forum, and, on the other hand, viewpoint discrimination, which is presumed impermissible when directed against speech otherwise within the forum's limitations.” Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 829-830 (1995); see also Cornelius, 473 U.S. at 806; Archdiocese, 897 F.3d at 339 (Wilkins, J., concurring) (“Forum doctrine's boundary between permissible subject-matter restrictions and impermissible viewpoint discrimination is a load-bearing wall in the First Amendment's structure.”). Indeed, “[g]overnment discrimination among viewpoints- or the regulation of speech based on ‘the specific motivating ideology or the opinion or perspective of the speaker'-is a ‘more blatant' and ‘egregious form of content discrimination.'” Reed v. Gilbert, 576 U.S. 155, 168 (2015) (quoting Rosenberger, 515 U.S. at 829). No matter the forum, viewpoint-based discrimination is never permitted.
Plaintiffs urge that NIH's keyword filtering is “axiomatic viewpoint-based censorship,” designed to “suppress animal advocates' viewpoints by blocking words closely associated with their efforts to urge NIH to stop animal testing.” Pls.' Mem. at 34-35.The keyword filter's targeting of words associated with animal testing, however, does not rise to the level of a viewpoint-based, rather than content-based, restriction. The Supreme Court's decisions in Lamb's Chapel v. Ctr. Moriches Union Free Sch. Dist., 508 U.S. 384 (1993), Rosenberger, and Good News Club v. Milford Cent. Sch., 533 U.S. 98 (2001), all of which involved prohibitions on the use of limited public fora for religious purposes, illustrate the distinction. In those cases, the Supreme Court identified the restrictions as singling out a “perspective, not [a] general subject matter.” Rosenberger, 515 U.S. at 831. Rather than prohibit the discussion of the subject matter of religion, those restrictions targeted “a specific premise, a perspective, a standpoint from which a variety of subjects may be discussed and considered,” id. at 828-31; see also Lamb's Chapel, 508 U.S. at 392-96; Good News Club, 533 U.S. at 107-112. By contrast, the filtering of words such as “animal” and “torture” merely targets a subject matter-animal testing-that defendants, as administrators of a limited public forum, are entitled to exclude.
Plaintiffs' argument that the keyword filters “suppress[] animal advocacy and not animal-related content generally” because keywords like “scorpion” and “guinea pig” are not filtered presents a strawman. Pls.' Mem. at 36. The subject matter targeted by the filter is animal testing, and the particular animals designated by the keyword filter are those that have been the subject of multiple repetitive and off-topic comments.
Plaintiffs' contention that “PETA,” “PETALatino,” “#stopanimaltesting,” “#stoptesting,” and “#stoptestingonanimals” are viewpoint-based keyword restrictions is more persuasive, but defendants have removed those keywords from their filter mechanisms. Pls.' Opp'n at 8-9. Contrary to plaintiffs' argument that these keywords provide “evidentiary support for the inference that the NIH has chosen to single out an anti-animal-testing viewpoint,” id. at 9, however, the former inclusion of these keywords is better understood as an overzealous attempt by a NIH social media manager to tamp down irrelevant posts-many of which happened to include these hashtags. See Defs.' Opp'n at 31-32.
The fact that only certain perspectives are represented in the subject matter being restricted-here, animal advocates like plaintiffs-is hardly novel or problematic. Defendants contend that a restriction “serv[ing] purposes unrelated to the content [expressed] is deemed neutral, even if it has an incidental effect on some speakers or messages but not others.” Defs.' Opp'n at 32-33 (quoting Ward v. Rock Against Racism, 491 U.S. 781, 791 (1989), and citing Madsen v. Women's Health Ctr., Inc., 512 U.S. 753, 758, 762-63 (1994)). Plaintiffs' attempt to distinguish Madsen and Ward by accurately identifying the restriction at issue here-unlike in Madsen and Ward-as content-based, Pls.' Opp'n at 11 n.2, but even a content-based restriction is not necessarily viewpoint-discriminatory merely because it disproportionately burdens one viewpoint. In Perry Education Association, for example, the Supreme Court upheld an access policy that had the effect of favoring one viewpoint over another in a nonpublic forum, where there was “no indication that the [government] intended to discourage one viewpoint and advance another.” 460 U.S. at 49. “Implicit in the concept of the nonpublic forum is the right to make distinctions in access on the basis of subject matter and speaker identity.” Id. “[A] restriction limiting a forum to discussion of selected topics-a common restriction among limited public forums-is self-evidently viewpoint neutral.” Davison, 247 F.Supp.3d at 777.
As a result, NIH's keyword filtering policy is permissibly based on the comments' subject matter, rather than viewpoint.
IV. CONCLUSION
The comment threads on NIH's Facebook and Instagram pages are limited public fora subject only to the standards that speech restrictions must be viewpoint-neutral and reasonable in light of the forum's purposes and circumstances. Defendants have cleared both of those hurdles. Accordingly, defendants' Cross-Motion for Summary Judgment is granted; plaintiffs' Motion for Summary Judgment is denied.
An order consistent with this Memorandum Opinion will be entered contemporaneously.