Filters and the First Amendment

Minnesota Law Review

Volumn 83, Issue 3, 1999, Pages 755-812

  Wagner, R Polk ^a,b  

^a STANFORD LAW SCHOOL   (United States)

^b UNIVERSITY OF MICHIGAN   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 33750559182     PISSN: 00265535     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Review

References (250)

1. See, e.g., Philip Elmer-Dewitt, On a Screen Near You: It's Popular, Pervasive and Surprisingly Perverse, According to the First Survey of Online Erotica, TIME, July 3, 1995, at 38.
2. President's Remarks Announcing Steps to Make the Internet Family-Friendly, 33 WEEKLY COMP. PRES. DOC. 1077 (July 16, 1997) [hereinafter President's Remarks].
3. Search of LEXIS, News library, Majpap File (May 10, 1998).
4. 117 S. Ct. 2329, 2344 (1997).
5. Paul Resnik & James Miller, PICS: Internet Access Controls Without Censorship, 10 COMM. ACM, Oct. 1996, at 87.
6. See Lawrence Lessig, Tyranny in the Infrastructure, WIRED 5.07, July 1997, at 96, 96.
7. Simson Garfinkel, Good Clean PICS: The Most Effective Censorship Technology the Net Has Ever Seen May Already Be Installed on Your Desktop, HOTWIRED (Feb. 5, 1997) 〈http://www.hotwired.com/packet/garfuikel/97/05/ index2a.html〉.
8. See generally Lawrence Lessig, What Things Regulate Speech: CDA 2.0 vs. Filtering, 38 JURIMETRICS J. 629 (1998).
9. Lawrence Lessig, in particular, has persuasively made this point. See id.
10. See, e.g., Reno v. ACLU, 117 S. Ct. 2329, 2353-54 (1997) (O'Connor, J., dissenting).
11. See, e.g., The White House, Office of the Press Secretary, Internet Online Summit: Focus on Children, Mission Statement (last modified Nov. 19, 1997) 〈http://www.kidsonline.org/mission/〉 (encouraging technological "user empowerment tools" to shield children from inappropriate material).
12. See, e.g., Rowan v. United States Post Office, 397 U.S. 728, 740 (1970) (allowing the Post Office to implement mail patron requests to block mail based on sender address).
13. Cyber Patrol, SurfWatch, CyberSitter, and X-Stop are all examples of database-driven filtering software products. See, e.g., Cyber Patrol (visited Oct. 16, 1998) 〈http://www.cyberpatrol.com〉;
14. Surfwatch (visited Oct. 16, 1998) 〈http://www.surfwatch.com〉.
15. The Platform for Internet Content Selection (PICS) is the best-known example of an embedded system. See, e.g., Paul Resnik, Filtering Information on the Internet, SCI. AM., Mar. 1997, at 62.
16. This is primarily because the huge volume of content to be rated and the difficulty of keeping the database system current virtually mandates spreading the costs. See Study: Search Engines Fall Short, CNET NEWS.COM (Apr. 2, 1998) 〈http://news.com/News/Item/0,4,20728,00.html〉 (finding that even search engines index only a fraction of the web). Not all database systems are run by third parties, however. The University of Michigan is demonstrating a database ratings system that will use ratings assigned by the "community" using the ratings system.
17. See PICS Application Incubator (visited Oct. 16, 1998) 〈http://krabappel.si.umich.edu/incubator/〉 ("Such a system uses embedded labels in combination with externally created ratings.").
18. For example, The Learning Company offers Cyber Patrol 4.0, a parental control software product, for $29.95. See Cyber Patrol Fact Sheet (visited Oct. 16, 1998) 〈http://www.cyberpatrol.com/fact.htm〉. Cyber Patrol uses a CyberNOT and CyberYES list to allow users the option of blocking the content on the CyberNOT list or limiting content to that on the CyberYES list. See id. The lists are updated regularly by a "team of professionals . . . including parents and teachers." Id. Subscriptions to the updating service are an extra charge. See id. Of course, the ratings need not be assigned by humans. NetNanny is software that scans the content of materials before it is displayed on a user's screen. If the content matches words or phrases defined by the user to be disallowed, then the content will not be displayed.
19. See Netnanny FAQ (visited Dec. 10, 1998) 〈http://www.netnanny.com/support/faq.htm〉.
20. Declan B. McCullagh & Brock N. Meeks, Keys to the Kingdom, CYBERWIRE DISPATCH (July 3, 1996) 〈http://www.cyberwerks.corn/cyberwire/ cwd/cwd.96.07.03.html〉;
21. Peacefire: Youth Alliance Against Censorship, Why Peacefire Opposes Blocking Software (visited Oct. 25, 1998) 〈http://www. peacefire.org〉 [hereinafter Peacefire];
22. Electronic Privacy Information Center, Faulty Filters: How Content Filters Block Access to Kid-Friendly Information on the Internet (visited Jan. 5, 1999) 〈http://www2.epic.org/reports/filter-report.html〉.
23. For example, recent reports by the Censorware Project and Peacefire.org note that the highest-rated filtering software product, Cyber Patrol, blocks the web sites of Planned Parenthood, Envirolink (an environmental clearinghouse), the AIDS Authority, the MIT Project on Mathematics and Computation, the University of Arizona, and the U.S. Army Corps of Engineers Construction Engineering Research Laboratories. See Peacefire, supra note 17; The Censorware Project, Blacklisted by Cyber Patrol: From Ada to Yoyo (visited Dec. 23, 1997) 〈http://www.spectacle.org/cwp/ada-yoyo.html〉;
24. see generally Jonathan Weinberg, Rating the Net, 19 HASTINGS COMM. & ENT. L.J. 453 (1997).
25. A recent study of search engines concluded that (as of early 1998) there are more than 320 million individual pages on the web alone. Study: Search Engines Fall Short, supra note 15. By some accounts, the number of pages on the web has been doubling every three to six months since 1993.
26. See Matthew Gray, Web Growth Summary (visited Oct. 16, 1998) 〈http://www. mit.edu/people/mkgray/net/web-growth-summary.html〉. The database filtering systems could respond to this volume by assigning ratings more genetically - by rating an entire domain or host rather than individual web pages, for example. Of course, this only exacerbates the inaccuracy of the ratings. For example, Cyber Patrol has listed all pages at "members.tripod.com" under the categories of "violence," "nudity," "sexual acts," and "satanic/cult information." See The Censorware Project, supra note 18. That host is home to 1.4 million web pages, and while some of them probably do have offensive material, most probably do not, and are blocked by the program. See id.
27. Many of the filtering software programs denote specific criteria by which they rate pages. For example, Cyber Patrol has a list of twelve categories of rated material: violence/profanity, partial nudity, full nudity, sexual acts, gross depictions, intolerance, satanic/cult, drugs/drug culture, militant/extremist, sex education, questionable behavior or gambling, and alcohol and tobacco. See Cyber Patrol, CyberNOT List Criteria (visited Oct. 16, 1998) 〈http://www.cyberpatrol.com/cp_list.htm〉. Users of the software can choose to block any or all of these rated categories. The story of Solid Oak Software's CyberSitter is an enlightening example of the ways that these databases can be abused. After Brock Meeks and Declan McCullagh wrote Keys to the Kingdom, a Cyberwire Dispatch story exposing the overbreadth of filtering software (including the fact that CyberSitter blocked the National Organization for Women for "sexual content"), CyberSitter blocked the Cyberwire Dispatch site. The software still blocks the site for Peacefire, a young persons' organization against online censorship that contains material critical of filtering software, including CyberSitter. See McCullagh & Meeks, supra note 17; Peacefire, supra note 17. Solid Oak Software has threatened to block all 2500 web sites hosted by Peacefire's provider unless the provider removed the Peacefire site.
28. Rebecca Vessel, CyberSitter Goes After Teen, WIRED NEWS (Dec. 9, 1996) 〈http://www.wired.com. news/story/901.html〉. In 1997, it was revealed that CyberSitter's installation software scanned the user's hard drive to determine whether the user had visited the Peacefire web site, and would not install the software if it had.
29. See Brian McWilliams, CYBERsitter Filters Out Privacy, Says Anticensorship Group, PC WORLD NEWSRADIO (July 2, 1997) 〈http://www.pcworld.com/ pcwtoday/article/0,1510,5006,00.html〉.
30. Unless, of course, the producer inaccurately classifies the content.
31. See Resnik & Miller, supra note 5.
32. See id. at 87.
33. See id. at 88.
34. Id. at 93.
35. See, e.g., A Family Friendly Internet (visited Oct. 16, 1998) 〈http://www.whitehouse.gov/WH/New/Ratings/〉 ("On Wednesday, July 16, President Clinton and Vice President Gore announced a strategy for making the Internet 'family friendly.'").
36. Through a very scientific observation process, I assure you.
37. See Recreational Software Advisory Council on the Internet (visited Oct. 16, 1998) 〈http://www.rsac.org/fra_content.asp?onIndex=21〉 (instructing parents how to activate the filter). The factory default setting is to allow unrated sites to be seen.
38. Microsoft rates their pages according to the Recreational Software Advisory Council (RSACi) system. See msn.com (visited Dec. 10, 1998) 〈http: //home.microsoft.com/〉.
39. I suspect that this pressure will not be limited to commercial producers, but will extend to anyone who wishes to achieve the largest possible audience size. In order to "raise your voice" in cyberspace, it may become necessary to label it.
40. For example, Yahoo.com, the second most popular site on the net, is presently unrated, as are Netscape.com, the third most popular site, the New York Times, and others. See Media Metrix, Top New Media Results (visited March 1998) 〈http://www.npd.com/interact_mmnewmedia398.htm〉 (ranking the popularity of web sites among users connecting from their homes).
41. See Resnik & Miller, supra note 5, at 91.
42. Stanford University, for example, routes most Internet traffic through a computer named "sunet-gateway.stanford.edu," an obvious place for a filter. Most corporations have similar set-ups.
43. See Lessig, supra note 6, at 96. Control over access points is not a trivial matter. New interconnections are springing up all the time; democratic governments at least would be hard pressed to keep up with this growth.
44. Reno v. ACLU, 117 S. Ct. 2329, 2354 (1997) (O'Connor, J., dissenting).
45. See Garfinkel, supra note 7 ("Resnick and Miller have done a great job designing a framework for censorship. I don't think I could have done it better myself.").
46. See Whitney v. California, 274 U.S. 357, 377 (1927) (Brandeis, J., concurring) ("If there be tune . . . to avert the evil [flowing from speech] by the process of education, the remedy to be applied is more free speech, not enforced silence."), overruled in part by Brandenburg v. Ohio, 395 U.S. 444 (1969) (per curiam).
47. See Paul Resnick, Platform for Internet Content Selection (PICS) (last modified Jan. 3, 1998) 〈http://www.w3.org/PICS/〉. PICS allows the embedded code to refer to another Internet site for the label rather than containing the label itself. See id. This essentially turns the PICS system into a database system, as the external labels are typically assigned by a third party. SurfWatch, Cyber Patrol, and NetSheperd are all examples of PICS-enabled third-party rating systems. See id.
48. See About RSACi (visited Oct. 18, 1998) 〈http://www.rsac.org/fra_ Content.asp?onIndex=1〉.
49. See Recreational Software Advisory Council Rates Over 50,000 Web Sites Free: Voluntary System Leads the Industry in Safeguarding the Internet for Children (visited Dec. 1, 1997) 〈http://www.rsac.org/fra_Content.asp?onIndex=73〉.
50. See Recreational Software Advisory Council on the Internet, supra note 28.
51. See Parents - How to Use RSACi (visited Oct. 18, 1998) 〈http://www. rsac.org/fra_content.asp〉.
52. See RSACi Terms & Conditions (visited Oct. 18, 1998) 〈http://www. rsac.org/content/register/terms.html〉. This at least raises the possibility that RSACi will supplant a producer's judgment for its own, although I am unaware of any such circumstances where this has arisen.
53. See About RSACi, supra note 39.
54. See id.
55. Id. The terms used to describe the categories are further defined. Thus, "mild expletives" are further described as: "The words hell and damn, ass and horse's ass, BUT NOT asshole, assface, asswipe; butthead and buttface BUT NOT butthole and buttwipe." Definitions for RSACi Language Questions (visited Oct. 18, 1998) 〈http://www.rsac.org/content/register/def/ language.html〉. The system's definition of "innocent kissing" (suitable for a zero rating on the sex scale) is: Any portrayal of humans or human-like creatures which a reasonable person would consider as just kissing on lips (without touching of tongues), head, shoulder, hands or arms, but not any other areas including but not limited to neck, breasts, torso, or legs. Innocent kissing shows affection and/or love, but creates no reasonable perception of stronger sexual activity.
56. Definitions for RSACi Sex Questions (visited Oct. 18, 1998) 〈http://www.rsac. org/content/register/def/sex.html〉.
57. See Playboy Online (visited Oct. 21, 1998) 〈http://www.playboy.com/〉. While the PICS ratings are normally invisible to the user, anyone can see the ratings code (if any) of any page by selecting the "view source" function of the web browser.
58. See About RSACi, supra note 39.
59. See id.
60. See, e.g., ROBERT S. PINDYCK & DANIEL L. RUBINFELD, MICROECONOMICS 118-20 (3d ed. 1995) ("The Bandwagon Effect"). The argument, as made above, is that as more users enable the PICS rating systems on their computers (and some percentage of these choose to opt-out of unrated sites), the benefit to the producer of self-rating will increase. As the number of rated sites increase, the value of PICS - to both the user and the producer - increases rapidly, encouraging even more use.
61. This applies to either producer/raters or users/choosers, though the costs are especially low for the users.
62. Wendy Grossman, Signs of Insecurity in Cyberspace, DAILY TELEGRAPH (London), July 2, 1998, at 13.
63. The U.S. Government advanced a version of this argument in Reno v. ACLU, 117 S. Ct. 2329 (1997), as an additional and "equally significant" interest to that of protecting children. Id. at 2351. The Court found the argument to be "singularly unpersuasive" as a countervailing interest to free speech. Id.
64. Reno v. ACLU left intact part of the Communications Decency Act, 47 U.S.C. § 223(a) (Supp. 1997), prohibiting the knowing transmission of obscene messages to any recipient under 18 years of age. See Reno v. ACLU, 117 S. Ct. at 2340; see also A Family Friendly Internet, supra note 26 (noting that the decision in Reno v. ACLU "did not affect U.S. laws against obscenity, child pornography, and on-line stalking").
65. At this juncture, all indications are that the RSACi is well-positioned in this regard. See Mark A. Lemley & David McGowan, Legal Implications of Network Economic Effects, 86 CAL. L. REV. 479, 597-99 (1998).
66. Backbone carriers are primarily large telecommunications concerns that carry the bulk of Internet traffic between major interconnection points.
67. 117 S. Ct. at 2343-44.
68. See id. at 2329; Simon & Schuster, Inc. v. Members of N.Y. State Crime Victims Bd., 502 U.S. 105, 116 (1991).
69. See Simon & Schuster, 502 U.S. at 118.
70. See Ginsberg v. New York, 390 U.S. 629 (1968); Prince v. Massachusetts, 321 U.S. 158 (1944).
71. Reno v. ACLU, 117 S. Ct. at 2346.
72. See Miller v. California, 413 U.S. 15, 24-25 (1973). The Miller test looks at three factors to determine whether or not material is obscene. The factors are "(a) whether the average person, applying contemporary community standards would find that the work, taken as a whole, appeals to the prurient interest; (b) whether the work depicts or describes, in a patently offensive way, sexual conduct specifically defined by the applicable state law, and (c) whether the work, taken as a whole, lacks serious literary, artistic, political, or scientific value." Id. (citations omitted).
73. See Roth v. United States, 354 U.S. 476 (1957).
74. See supra note 18 and accompanying text.
75. New York Times Co. v. United States, 403 U.S. 713, 714 (1971) (per curiam); see also Bantam Books, Inc. v. Sullivan, 372 U.S. 58 (1963).
76. See Freedman v. Maryland, 380 U.S. 51, 58-59 (1965).
77. In one sense, the technology places the burden upon the speaker, as she will have to conform the speech to the rubric of the filtering system. However, in Freedman, the Court noted the tendency for bureaucratic overreach as the animating purpose behind the burden shifting. See id. at 57-58. Here the technology, of course, is perfectly consistent. And finally, the fact that the filtering system is fixed and preexisting means that the speaker has an immobile target at which to aim; she will be able to perfectly tailor her speech to meet the filter.
78. The volume of content and speed at which it gets filtered would themselves be significant impediments to appeal.
79. See, e.g., Jeri Clausing, In Rejecting Dismissal of Filtering Case, Judge Sets High Standard for Libraries, N.Y. TIMES (CYBERTIMES) (April 9, 1998) 〈http://www.nytimes.com/library/tech/98/04/cyber/articles/09library. html〉.
80. 2 F. Supp. 2d 783 (E.D. Va. 1998).
81. See Loudoun, 2 F. Supp. 2d at 793; see also Board of Educ. v. Pico, 457 U.S. 853 (1982).
82. Pico, 457 U.S. at 867.
83. Id. at 871.
84. Loudoun, 2 F. Supp. 2d at 792.
85. Id. at 793 (citation omitted). First Amendment advocates will be wise to avoid hanging too much on this factual distinction. Certainly the Internet can be viewed as an undifferentiated whole - a system of computer networks using the same sets of protocols and allowing ready access to any part of the system from any other. But the net can just as easily be viewed with slightly more granularity, as a "vast library including millions of readily available and indexed publications," according to the Supreme Court. Reno v. ACLU, 117 S. Ct. 2329, 2335 (1997). This line of factual argument is also vulnerable to what Justice O'Connor described in her dissent in Reno v. ACLU as "gateway" technology - which "construct[s] barriers in cyberspace" between content. Id. at 2353-54. The technology of the net is changing so rapidly that viewing it as an undifferentiated whole is likely to become increasingly inapt as filtering and other "gateway" technologies become more widespread and effective.
86. See Loudoun, 2 F. Supp. 2d at 795. The Supreme Court has long held that the state has special interests in controlling the operation and curriculum of schools and even in "inculcating fundamental values" in schoolchildren. Pico, 457 U.S. at 864; see also Pierce v. Society of Sisters, 268 U.S. 510 (1925); Meyer v. Nebraska, 262 U.S. 390 (1923). The Court has also found that those under the age of majority have less First Amendment rights than adults. See Ginsberg v. New York, 390 U.S. 629 (1968).
87. 475 U.S. 41 (1986). The Court in Renton found the government interest in zoning regulations to be "substantial," and the zoning itself to be effective. See id at 50-51.
88. See id.
89. In Renton, Justice Rehnquist, writing for the Court, noted the prevention of crime, the protection of the city's retail trade, and the quality of urban life as the secondary effects. See id. at 48.
90. See Boos v. Barry, 485 U.S. 312, 321 (1988) (noting that the desire to prevent "psychological damage" was targeted at the direct effects of the speech, not secondary effects).
91. Professor Tribe argues that laws seen by the courts as related to "communicative impact" will receive strict scrutiny, while those that relate to "non-communicative impact" will not. LAURENCE TRIBE, AMERICAN CONSTITUTIONAL LAW § 12-2, at 790 (2d ed. 1988). Of course, differentiating between the two is not trivial.
92. See John Hart Ely, Comment, Flag Desecration: A Case Study in the Roles of Categorization and Balancing in First Amendment Analysis, 88 HARV. L. REV. 1482 (1975) (arguing that the distinction between communication-based restrictions and non-communication-based restrictions is difficult to discern). Given the Supreme Court's strong support for the Internet as a medium for free speech, one suspects that courts will look askance at arguments that content regulation on this medium are directed solely at "non-communicative impact." See Reno v. ACLU, 117 S. Ct. 2329, 2544 (1997) ("[O]ur cases provide no basis for qualifying the level of First Amendment scrutiny that should be applied to this medium.").
93. Though the required "fit" under the time, place and manner intermediate scrutiny test is not stringent, it is non-zero. The physical "screening-off" of the Internet access computers from other patrons could be viewed as a far more effective (and speech-neutral) method for reducing secondary effects in the library.
94. See New York v. Ferber, 458 U.S. 747, 765-66 (1982) (holding that obscene material is not entitled to First Amendment protection); Miller v. California, 413 U.S. 15, 36 (1973) (holding that restrictions on obscene materials receive rational basis review). Of course, if the filtering technology in fact sweeps broader than the library policy indicates, then courts should read the parameters of the technology as the de facto library policy.
95. See Reno v. ACLU, 117 S. Ct. 2329, 2346 (1997) (applying strict scrutiny review to broad restrictions on adult speech designed to protect minors); Mainstream Loudoun v. Board of Trustees, 2 F. Supp. 2d. 783, 793 (1988) (statutes restricting speech harmful to minors receive strict scrutiny review).
96. 117 S. Ct. at 2346.
97. Id. (quoting Bolger v. Youngs Drug Prod. Corp., 463 U.S. 60, 74-75 (1983)).
98. 390 U.S. 629, 636 (1968) ("[T]he concept of obscenity or of unprotected matter may vary according to the group to whom the questionable material is directed or from whom it is quarantined."). Importantly, a public school library may have even more latitude to filter, as content selection within school libraries has been upheld to be permissible when grounded upon legitimate pedagogical purposes. See Board of Educ. v. Pico, 457 U.S. 853, 871 (1982).
99. See supra note 18 and accompanying text; see also Reno v. ACLU, 117 S. Ct. at 2344-45; Gooding v. Wilson, 405 U.S. 518, 521 (1972) (holding that overbroad statutes are unconstitutional); NAACP v. Alabama, 357 U.S. 449, 462 (1958) (holding that the government regulation must fit the intended ends). The current scope of the overbreadth doctrine is uncertain. In Broadrick v. Oklahoma, 413 U.S. 601 (1973), the Court noted that regulations will be invalidated on overbreadth grounds only if "substantially overbroad in relation to the statute's plainly legitimate purpose." Id. at 622. And in Brockett v. Spokane Arcades, 472 U.S. 491 (1985), the Court held that a statute susceptible to a narrowing construction will not be held overbroad on the basis of a broader construction. See id. at 509. In the filtering and blocking case, courts will have ample evidentiary material to consider the quantitative approach of Broadrick, though the line between substantial and insubstantial overbreadth is not at all clear. Brockett, on the other hand, seems inapplicable, since the technological filter at issue is resistant to alternative constructions, broader or narrower.
100. This suggests that the government could anticipate the positive network externalities associated with labeling by requiring less than all of the net content to be rated. For example, all commercial content could be required to be rated, creating less of a regulatory burden and decreasing some First Amendment impact.
101. Senator Patty Murphy announced in late summer 1997 that she would introduce legislation making the misrating of sites criminally punishable. See Danny Westneat, Senator Murray's Internet Controls Draw Fire, SEATTLE TIMES, July 15, 1997, at A12. To date, however, no such legislation has been introduced on a national level. Because the fraud penalties are contingent upon the underlying question of whether the government may directly coerce labeling of content, they are beyond the scope of this paper. I mention them simply to denote some enforcement mechanism that would get at a crucial problem with self-rating systems: misrating.
102. Compare West Virginia State Bd. of Educ. v. Barnette, 319 U.S. 624, 633-34 (1943) (striking down mandatory flag salutes as promoting a government orthodoxy) with Buckley v. Valeo, 424 U.S. 1, 58 (1976) (upholding disclosure requirements as the "least restrictive means" of supporting the government's significant - but politically neutral - anti-corruption interests).
103. As I describe more fully below, this is primarily because "avoiding the details" appears to exploit the crucial public-private distinction. See infra notes 159-69 and accompanying text.
104. See Lessig, supra note 8, at 665-69.
105. See, e.g., McIntyre v. Ohio Elections Comm'n, 514 U.S. 334, 348 (1995) (state cannot require individuals to identify themselves in campaign literature); Riley v. National Fed'n of the Blind, 487 U.S. 781, 795 (1988) (striking down requirement that charities reveal the percentage of their funds spent on charitable activities); Wooley v. Maynard, 430 U.S. 705, 714 (1977) (state cannot require an individual to participate in the dissemination of an ideologically hateful message); Miami Herald Co. v. Tornillo, 418 U.S. 241, 256 (1974) (state cannot force a newspaper editor to publish editorial responses); Barnette, 319 U.S. at 633 (compulsory flag salutes are unconstitutional).
106. See Wooley, 430 U.S. at 714; Barnette, 319 U.S. at 633.
107. See Riley, 487 U.S. at 798.
108. See Pacific Gas & Elec. Co. v. Public Util. Comm'n, 475 U.S. 1, 17-18 (1986) (striking down requirement that public utility place third-party newsletter in mailings to customers).
109. See McIntyre, 514 U.S. at 379.
110. See Hurley v. Irish-American Gay, Lesbian and Bisexual Group, 515 U.S. 557, 559 (1995) (holding that parade organizers had no obligation to allow an ideologically opposed organization in their parade).
111. See Meese v. Keene, 481 U.S. 465, 480 (1987) (holding that the state can compel the placement of a "political propaganda" label on foreign films).
112. See id. (holding forced speech constitutional where it places no burden on speaker).
113. See Buckley v. Valeo, 424 U.S. 1, 29 (1986) (holding that the government's interest in appearance of a proper political process is a substantial interest that outweighs certain First Amendment values). This is generally considered to be strict scrutiny.
114. While I am unaware of any challenge to these compelled speech schemes on First Amendment grounds, these regulations may be less constitutionally questionable because they deal with commercial speech, and - at least in the case of the FDA regulations - the labels are affixed to products rather than to speech. In this sense, Glickman v. Wileman Bros., 117 S. Ct. 2130 (1997), is instructive. In Wileman Bros., the Court found that compelled subsidies for generic fruit advertising did not burden First Amendment rights, and instead evaluated the program "under the standard appropriate for the review of economic regulation." Id. at 2138.
115. 481 U.S. at 468. That the underlying speech in Keene was of foreign origin (foreign films) does not make the case any less analogous. In the compelled speech context, the First Amendment right is that of the editor (i.e., the individual wishing to not speak), not necessarily the original producer of the material. See, e.g., Miami Herald Co. v. Tornillo, 418 U.S. 241, 258 (1974) (discussing First Amendment rights of editors). So in Keene, the asserted First Amendment right was that of the distributor, an American citizen. 481 U.S. at 467, 474 (describing Keene as an attorney, member of the California State Legislature, and citizen). Indeed, the Court in Keene explicitly compared the rights at issue to those presented in Lament v. Postmaster General, 381 U.S. 301 (1965), and Virginia Pharmacy Bd. v. Virginia Citizens Consumer Council, Inc., 425 U.S. 748 (1976), cases involving both domestic "editors" and domestic speech. See Keene, 481 U.S. at 475-76, 481-82. Finally, there is no hint in the language or reasoning of the Keene opinion that the origin of the underlying speech was given any weight whatsoever.
116. Keene, 481 U.S. at 468-69.
117. Id. at 480. At least one commentator has suggested that Keene is irreconcilable with Riley's "broad language" supporting strict scrutiny for compelled speech. See Chris Kelly, The Spectre of a 'Wired' Nation: Denver Area Educational Telecommunications Consortium v. FCC and First Amendment Analysis in Cyberspace, 10 HARV. J.L. & TECH. 559, 636 (1997); see also Lessig, supra note 8, at 662 n.95. But the cases can simply be harmonized by recognizing that the Court - in all such cases - first determines whether the compelled speech "burdens protected expression." Keene, 481 U.S. at 480. While the Court has noted that compelled speech is as constitutionally questionable as speech regulation, the Court reaches this point after determining that the compulsion of speech was as burdensome as direct speech regulation. See Tornillo, 418 U.S. at 256. In Wooley and Barnette, the burden on protected speech was the attempt to promote government orthodoxy (and perhaps thereby drown out other views). See Wooley v. Maynard, 430 U.S. 705, 715 (1977); West Virginia State Bd. of Educ. v. Barnette, 319 U.S. 624, 641-42 (1943). In Pacific Gas and Tornillo, the burden was much more direct: that the right of response was triggered on the basis of earlier speech meant that the speaker would be chilled from uttering some types of speech. See Pacific Gas & Elec. Co. v. Public Util. Comm'n, 475 U.S 1, 18 (1986); Tornillo, 418 U.S. at 257. And in McIntyre, the chilling effect from the possibility of reprisal was an unquestionable burden. See McIntyre v. Ohio Elections Comm'n, 514 U.S. 334, 379-80 (1995). In contrast, the Court in Wileman Bros. found that the plaintiffs' "criticisms of generic advertising provide no basis for concluding that the factually accurate advertising constitutes an abridgement of anybody's right to speak freely." 117 S. Ct. at 2139-40. And while it is quite possible to argue against the Wileman Bros. and Keene decisions on this basis, the Court's unifying theme appears to be that the compelled speech must burden in some sense the concomitant rights of speech. Id. at 2138. Therefore, in order to raise a First Amendment compelled speech claim, a challenger must show a causal link between the compelled speech and a burden on protected expression.
118. Meese v. Keene, 481 U.S. 465, 480-81 (1987).
119. Id. at 483. The Court also noted the "respect" for Congress's power to define the terms that it uses hi legislation. Id. at 484. Because the term was defined neutrally in the statute, and had been widely used without challenge, the Court would not attribute a negative meaning. See id.
120. See supra notes 83-84 and accompanying text.
121. Ideally, there would be empirical evidence to demonstrate how many potential audience members were lost due to a particular rating, such as for "sexual activity."
122. See Keene, 481 U.S. at 481.
123. In American Libraries Ass'n v. Pataki, 969 F. Supp. 160 (S.D.N.Y. 1997), one long time Internet expert testified that he tried to imbed PICS-compliant labels in his online newsletter site but finally gave up after several hours. See Testimony cited in ACLU, Fahrenheit 451.2: Is Cyberspace Burning? (visited Oct. 20, 1998) 〈http://www.aclu.org/issues/cyber/burning.html〉. This is distinct from the author's experience, where a web page was labeled within fifteen minutes and with minimal effort.
124. The only possible question is whether the PICS statute is likely to be at all effective. Given the low level of rationality review, however, it seems highly likely that Congress would be given the benefit of the doubt. See Glickman v. Wileman Bros, 117 S. Ct. 2130, 2152 (1997) (holding that regulation of economic activity "is plainly permissible short of something so arbitrary as to fail the rational basis test").
125. 512 U.S. 622, 663-64 (1994). At issue in Turner were a series of "must carry" provisions that forced cable companies to carry local commercial and public broadcast stations. The Court upheld these provisions. See id.
126. Id. at 645.
127. Id. at 662. Content-neutral regulations that impose an "incidental" burden on speech are analyzed according to intermediate scrutiny. See United States v. O'Brien, 391 U.S. 367, 376 (1968); see also Ward v. Rock Against Racism, 491 U.S. 781, 789 (1989). So-called "time, place, and manner" regulations are also analyzed in this mode. See, e.g., Clark v. Community for Creative Non-Violence, 468 U.S. 288, 293 (1984). Under O'Brien, a content-neutral regulation will be upheld if: "[1] it furthers an important or substantial governmental interest; [2] if the governmental interest is unrelated to the suppression of free expression; and [3] if the incidental restriction on alleged First Amendment freedoms is no greater than is essential to the furtherance of that interest." 391 U.S. at 377 (numbering added). The significant analysis of the O'Brien test is primarily part [2]; the Court has not closely followed the "no greater than essential" restriction in part [3]. See Clark, 468 U.S. at 299.
128. "It would be error to conclude . . . that the First Amendment mandates strict scrutiny for any speech regulation that applies to one medium (or a subset thereof) but not others." Turner, 512 U.S. at 660. Instead, the Court has held that such media discrimination will only raise serious constitutional questions in certain circumstances, such as when it targets a small number of speakers, and is justified when the differential treatment is based on the special characteristics of the medium being regulated. See id. (citing Arkansas Writers' Project, Inc. v. Ragland, 481 U.S. 221 (1987); Minneapolis Star & Tribune Co. v. Minnesota Comm'r of Revenue, 460 U.S. 575 (1983)). Here, the Internet contains a very large number of speakers and the pervasiveness, ease of access, and difficulty of public regulation probably avoids the discriminatory concern.
129. Professor Lessig, drawing support from some of the Court's language in Riley, makes the more direct claim that compelled speech is inherently content-based. See Lessig, supra note 8, at 662 n.95. This conclusion, if it was ever applicable to the facts of Riley - and I doubt it was - certainly did not survive Turner, where the Court not only described Riley as content-based in application - noting that the basis of the Riley holding was that "solicitation of funds trigger[ed a] requirement to express [a] government-favored message" - but also proceeded to analyze the "compelled speech" must-carry provisions as content-neutral regulations meeting intermediate scrutiny. 512 U.S. at 655.
130. To be fair to Professor Lessig, he appears to be analyzing a statute that would require rating only of what he calls "Ginsberg speech" - speech from which the government has a legitimate interest in protecting children. See Lessig, supra note 8, at 657-69. A statute only requiring labeling of pages with a significant degree of sexual content is plainly content-based; my point is that a relatively minor change in the statute leads to a very different result and is probably as likely to occur, given the advantages.
131. The Court simply did not accept that the application of a "propaganda" label would "have actually had any adverse impact on the distribution" of the materials. Meese v. Keene, 481 U.S. 465, 483-84 (1987).
132. The Court noted that in the context of fundraising, the compelled disclosure of financial overhead percentages "could encourage or discourage the listener from making a . . . donation [and] would clearly and substantially burden the protected speech." Riley v. National Fed'n of the Blind, 487 U.S. 781, 798 (1988). Admittedly, on this point, Riley and Keene are difficult to square.
133. Cf. Miami Herald Co. v. Tornillo, 418 U.S. 241 (1974); Pacific Gas & Elec. Co. v. Public Util. Comm'n, 475 U.S 1 (1986).
134. Cf. Riley, 487 U.S. at 798.
135. Turner Broad. Sys. v. FCC, 512 U.S. 622, 663 (1994). I must admit that it seems a bit odd to talk of interests in "diversity" as content-neutral. Certainly diversity interests are not content-specific, but they seem to imply an interest in the different, an interest in anything not currently available, which seems less than wholly agnostic.
136. Relatedly, Turner's finding that the importance of local broadcasting outlets themselves "can scarcely be exaggerated," may support a claim that the growth of the Internet itself would be a sufficiently important content-neutral basis for imposing self-rating. See id. at 663 (citing United States v. Southwestern Cable Co., 392 U.S. 157, 177 (1968); see also Reno v. ACLU, 117 S. Ct. 2329, 2344 (1997) (noting the great communicative virtues of the Internet). However, in Reno v. ACLU, the Court noted that this argument assumes that "the unregulated availability of [unwanted content] is driving countless citizens away from the medium because of the risk of exposing themselves or their children to harmful material." 117 S. Ct. at 2351. Certainly the continued growth of the Internet seems to repudiate the idea that any stifling is occurring. There are debates about the most appropriate measure of the "size" of the Internet (and therefore the size), but nobody seems to think that the growth rate is slowing. See, e.g., Jamie Murphy, It's Not the Size that Counts, But How You Measure It, N.Y. TIMES (CYBERTIMES) (visited June 5, 1998) 〈http://www.nytimes.com/library/tech/98/07/cyber/articles/05big.html〉. Note, however, that time could change this conclusion: it is plausible that the Internet might not reach its potential as a widespread commercial and communicative medium without some way to allay the fears of those concerned about the availability of inappropriate materials. For example, the manipulation of popular Internet "search engine" results by pornographic web sites is an issue likely to resonate with those considering whether to connect to the Internet.
137. See, e.g., www.oneplace.sex? Internet Porn Would Be Easy to Filter (or Find), SAN JOSE MERCURY NEWS, June 5, 1998, at 6C (editorial noting that the Internet is turning into a minefield for parents because "their children are . . . inadvertently being exposed to . . . the profane" and that "[e]ven word searches for seemingly innocuous terms can trigger ads for erotica").
138. See America Online (visited Oct. 16, 1998) 〈http://www.aol.com〉.
139. See 1997 Annual Report - America Online, Inc. (visited Oct. 16, 1998) 〈http://www.aol.com/corp/inv/reports/1997/6_financials/6_pg03a.html〉.
140. See Turner Broad. Sys. v. FCC, 117 S. Ct. 1174, 1190 (1997).
141. An obvious rejoinder is that users will not put up with this, and will either pressure AOL to change its policies, or switch to another ISP. In this way, AOL competes in a much more robust market than cable operators, who might have a local monopoly. But the cable operators in Turner had, and lost, this argument as well - that users could simply use over-the-air signals to tune into the local broadcasters. See Turner, 512 U.S. at 661.
142. Id. at 665.
143. One possible route of exploration is through a market-based analysis. That is, a speaker-producer of protected speech could claim that the existence of low-cost private choices regarding censorship would result in the constructive suppression of particular views. Although this case is hard to make - it will depend in large part upon a court's acceptance of network externality theory and perhaps economic modeling - it is more likely to be a successful approach to this public-private discontinuity than direct or indirect attacks on the doctrine itself.
144. Denver Area Educ. Telecomm. Consortium, Inc. v. FCC, 518. U.S. 727, 737 (1996).
145. The relative burdens analysis is more properly located in the "fit" or narrow tailoring inquiry discussed below. See infra notes 142-44, 149-52 and accompanying text.
146. See United States v. O'Brien, 391 U.S. 367, 377 (1968); see also supra note 116 and accompanying text (discussing the O'Brien test and related Supreme Court holdings).
147. Since the early 1980s, the Court has approached the "state action" requirement from two perspectives, neither of which seems to weigh in favor of equating a private ratings system with state action. The first is that used in Blum v. Yaretsky and Rendell-Baker v. Kohn: that "constitutional standards are invoked only when it can be said that the State is responsible for the specific conduct of which the plaintiff complains." Blum v. Yaretsky, 457 U.S. 991, 1004 (1982); see also Rendell-Baker v. Kohn, 457 U.S. 830, 840 (1982). The second was used in Lugar v. Edmondson Oil Co. and Edmonson v. Leesville Concrete, where the Court looked to two factors: (1) whether state authority undergirded the alleged unconstitutional act, and (2) whether the litigant must be fairly considered a state actor. See Edmonson v. Leesville Concrete Co., 500 U.S. 614 (1991); Lugar v. Edmondson Oil Co., 457 U.S. 922 (1982). The privately-developed PICS-compatible ratings system is not supported by the PICS-enforcing statute in a Lugar and Edmonson way - the government's requirement that labels be applied does not bear a causal link to a particular private ratings system - for in the hypothetical PICS-world, a (private) market for ratings systems develops. See, e.g., Lessig, supra note 8, at 663. And the private creator of a private ratings system can hardly be fairly described as a state actor; indeed, there is no certainty that the market-developed ratings systems would even square well with Congressional interests (a risk that Congress may have to take when conceding the details of ratings to private parties).
148. Lessig argues that a PICS-enforcing statute would be unconstitutional because it is not narrowly-tailored and runs afoul of the Court's decisions in Lamont v. Postmaster General, 381 U.S. 301 (1965) and Bolger v. Youngs Drug Prod. Corp., 463 U.S. 60 (1983). See Lessig, supra note 8, at 666-67.
149. See Lessig, supra note 8, at 666.
150. This is not to criticize Professor Lessig's efforts to change the thinking about private-public distinctions. His efforts to do so are important. My intent here, however, is to note how the present doctrine might apply to these issues, to promote a full understanding of the underlying questions involved.
151. The Court has often disavowed analysis based on legislative motivation. See, e.g., United States v. O'Brien, 391 U.S. 367, 383 (1968) ("It is a familiar principle of Constitutional law that this Court will not strike down an otherwise constitutional statute on the basis of an alleged illicit legislative motive.").
152. See Glickman v. Wileman Bros., 117 S. Ct. 2130 (1997) (plaintiffs forced to pay for generic advertising;); Riley v. National Fed'n of the Blind, 487 U.S. 781, 798 (1988) (forced to disclose particular financial details); Wooley v. Maynard, 430 U.S. 705 (1977) (forced to be a "government bill-board"); Buckley v. Valeo, 424 U.S. 1 (1976) (forced to disclose contributors); Tornillo v. Miami Herald Co., 418 U.S. 241 (1974) (forced to print others' speech); West Virginia State Bd. of Educ. v. Barnette, 319 U.S. 624 (1943) (forced to salute flag).
153. See United States v. O'Brien, 391 U.S. 367, 377 (1968); see also supra note 116 and accompanying text (discussing the O'Brien test and related Supreme Court holdings).
154. See, e.g., Clark v. Community for Creative Non-Violence, 468 U.S. 288, 299 (1984) (holding that the Court of Appeals' view that there were less speech-restrictive alternatives available was no more than a disagreement with the Park Service about the proper implementation of a policy, and that O'Brien does not "assign to the judiciary the authority to replace the Park Service as the manager of the Nation's parks"). See generally Ely, supra note 81, at 1486-88 (noting the "apparent discontinuity" in the Court's approach to the less restrictive alternative analysis).
155. Turner Broad. Sys. v. FCC, 512 U.S. 622, 622 (1994) (citing Ward v. Rock Against Racism, 491 U.S. 799 (1989)). This is related to the discussion of equally effective alternatives below. See infra notes 149-53 and accompanying text.
156. Lessig correctly notes that there are differences in the "narrowly-tailored" requirement in the content-based and content-neutral cases, but argues that a PICS-enforcing statute would fail either primarily because "[s]uch a regime would result in a wider range of filtered speech than the legitimate interests of government would allow." Lessig, supra note 8, at 665 n.101. As I noted above, I view this as another approach to the private-public distinction that these issues raise. See supra text accompanying notes 123-124. Left out of Lessig's claim is the recognition that the "wider range of filtered speech" would result from a myriad of individual private choices, not from government action to suppress speech. I do not argue that the PICS-enforcing statute will not result in more filtered speech - it surely will - I simply argue that where the filtering is a private concern, it cannot be analyzed as if it were state action.
157. Buckley, 424 U.S. at 68.
158. See Sable Communications v. FCC, 492 U.S. 115, 126 (1989); Ginsberg v. New York, 390 U.S. 629, 636 (1968).
159. Prince v. Massachusetts, 321 U.S. 158, 166 (1944).
160. See FCC v. Pacifica Found., 438 U.S. 726, 749 (1978); Ginsberg, 390 U.S. at 639.
161. Buckley, 424 U.S. at 68.
162. See Simon & Schuster, Inc. v. Members of N.Y. State Crime Victims Bd., 502 U.S. 105 (1995); Burson v. Freeman, 504 U.S. 191 (1992); Boos v. Barry, 485 U.S. 312 (1988).
163. Eugene Volokh, Freedom of Speech, Shielding Children, and Transcending Balancing, 1997 SUP. CT. REV. 141, 148-49 (1997). As Volokh notes, the Reno v. ACLU test is significantly less protective of speech: the government could overcome the "narrowly tailored" requirement by developing factual findings that the asserted alternatives are not as effective. "The pregnant negative in the Court's reasoning is that, had there really been no equally effective alternatives (as in fact there are not), the CDA should have been upheld." Id. at 157. In criticizing this approach, Volokh disputes that "tagging," as a factual matter, would be as effective as the CDA's flat ban. See id. at 149.
164. Reno v. ACLU, 117 S. Ct. 2329, 2346 (1997).
165. Ginsberg v. New York, 390 U.S. 629, 636-37 (1968). In Reno v. ACLU, the Court noted that the spillover argument cannot be used to justify the "unnecessarily broad suppression of speech addressed to adults." 117 S. Ct. at 2335. As long as the Court determines that the burden on speaker-producers caused by forced labeling is small, then this "narrowly tailored" argument would apply.
166. This strict scrutiny framework - the requirement for the law to be "narrowly tailored," "equally effective," and/or "least restrictive" - appears to set up a form of sliding scale analysis. The more restrictive the law (i.e., the more burdensome on speech), the more closely aligned to the interests asserted and effective it must be. Likewise, laws with little restrictions or burdens may not be forced to be as closely tailored. See Volokh, supra note 151, at 193-94. Volokh argues that this "weighing" of the benefits and burdens is judicially inadministrable, and that the Court would be better served in developing more categorical rules regarding substantial and insubstantial burdens. See id. at 194. While I share Professor Volokh's desire for clear rules, the Court itself has made no signs that it intends to move away from the present mode of analysis. See generally Reno v. ACLU, 117 S. Ct. at 2335.
167. For example, children would still be able to access improper material from computers other than their parents', or might be able to break the code protecting the rating choices on their home computer. The PICS-enforcing statute also assumes that parents with Internet connections would be knowledgeable enough to implement the controls. Both of these questions are distinctly factual in nature, and ones I think can go either way.
168. See supra note 147 and accompanying text.
169. See supra notes 102-10 and accompanying text (discussing the criteria the Court uses for determining which standard of review applies).
170. See United States v. O'Brien, 391 U.S. 367, 377 (1968).
171. See, e.g., Lessig, supra note 8, at 665-67.
172. Id. at 665.
173. See id at 668.
174. As an initial matter, it is important to note that the architecture that would support broader filtering does so only by private parties. The public/private distinction - as discussed above - matters a great deal in this analysis. See supra notes 122-40 and accompanying text. This "architectural" approach can thus be (fairly, I think) boiled down to a contention that that the government may not enable broader private regulation if narrower public regulation is available. And while it is entirely likely that this argument is a good principle of public policy, the more significant question is whether it has support in the Constitution.
175. Lessig, supra note 8, at 668.
176. See, e.g., Reno v. ACLU, 117 S. Ct. 2329, 2342 (1997); Lessig, supra note 8, at 669.
177. Indeed, it appears to assume a statute that requires PICS labeling of only that speech which is deemed to be harmful to children. See Lessig, supra note 8, at 666-68.
178. The architecturalists' argument is thus "exposed" (though it is clear enough not to need exposure) as another version of the argument against the rigid public/private distinction established by Supreme Court doctrine: that courts should consider the enablement of private regulation on (nearly) the same footing as direct public regulation. See supra note 135 and accompanying text. Admittedly, PICS presents this issue pretty squarely - the network effects and unintended consequences of labels make powerful arguments in favor of a more flexible state action doctrine. It just doesn't work under the current technological and legal situation.
179. Bolger v. Youngs Drug Prod. Corp., 463 U.S. 60, 69 (1983) (striking down a government ban on direct-mail advertisements).
180. Lamont v. Postmaster Gen., 381 U.S. 301, 307 (1965) (striking down a requirement that postal recipients specifically request delivery of "communist propaganda").
181. Rowan v. United States Post Office, 397 U.S. 729 (1970).
182. I would argue, however, that Rowan does not necessarily compel the conclusion that the government may not be involved in the design of ratings systems themselves. While the Post Office in that case merely provided a ministerial function, the basis of the decision is clearly that user-selected choices are constitutionally permissible. See Rowan, 397 U.S. at 737 ("Nothing in the Constitution compels us to listen to or view any unwanted communication . . . . The ancient concept that 'a man's home is his castle' into which 'not even the king may enter' has lost none of its vitality . . . ."). There is no suggestion in Rowan that the government may not in some cases further assist the addressee in blocking unwanted mail, just an offhanded remark that "Congress provided this sweeping power [to block to the addressee] not only to protect privacy but to avoid possible constitutional questions that might arise from vesting the power to make any discretionary evaluation of the material in a governmental official." Id. This statement implies that the government may not itself rate or label speech, but I believe it is silent to the question of whether the government may create labels for others to use. The relevant question that we should ask whether Rowan answers is the following: may the Post Office, in support of addressee's blocking choices, create a set of categories for content sent in the mails (presumably including various categories that many addressees might find offensive) and force mailers to categorize their mail as such? I am envisioning a sort of second zip code, containing ratings. Addressees could select which categories of content they did not wish to receive, and their choices would be administered by the Post Office. Under Rowan, the addressee has the unquestionable right to block unwanted mail, and - significantly - the Court tells us that the addressee need not see individual items before they are blocked. See id. at 738 ("The continuing operative effect of a mailing ban once imposed presents no constitutional obstacles; the citizen cannot be put to the burden of determining on repeated occasions whether the offending mailer has altered its material so as to make it acceptable. Nor should the householder have to risk that offensive material come into the hands of his children before it can be stopped."). While not crystal clear, I believe this may be fairly read to support the dual propositions that the Post Office may help reduce the burdens of blocking from the addressee, and that the addressee need not actually see or know the specifics of the material blocked. And while it is probably a stretch to thus claim that Rowan allows government-imposed labeling under government-designated categories, it makes much more questionable the claim that Rowan forbids the government from designing (and requiring) rating systems. See, e.g., Lessig, supra note 8, at 668 (arguing that "Rowan limits the government's power" to design rating systems).
183. For example, she may find the "levels" of the ratings system to be wholly inadequate or offensive in some way. She may feel that the system does not provide her an opportunity to adequately express her rating of her content, or may feel offended by having to rate her page for sexual content even where it contains no such content.
184. 431 U.S. 209, 234 (1977) (striking down compelled fees from Union members insofar as they were used to support causes members found objectionable). This freedom of "non-association" has been limited by later cases to those organizations which are expressive in purpose. See, e.g., Hurley v. Irish-American Gay, Lesbian and Bisexual Group, Inc. 515 U.S. 557, 573 (1995) (upholding parade organizers' rights to exclude groups wishing to march); New York State Club Ass'n v. City of New York, 487 U.S. 1, 8 (1988) (upholding anti-discrimination law targeted at some classes of private clubs); Roberts v. United States Jaycees, 468 U.S. 609, 611 (1984) (upholding anti-discrimination law as applied to the organization).
185. Note here that I am making an assumption - one that I contend is reasonable - that the insertion of the PICS "tags" does not alter the content of the speaker's message. Because PICS tags are invisible to the viewer, they do not affect the reaction or impact of the message merely by virtue of their insertion. This assumption distinguishes this case from Hurley, where distinct ideological anathema was found unnecessary to the compelled association claim. See Hurley, 515 U.S. at 574-75.
186. Miami Herald Co. v. Tornillo, 418 U.S. 241 (1974).
187. Pacific Gas & Elec. Co. v. Public Util. Comm'n, 475 U.S. 1 (1986).
188. See Pacific Gas, 475 U.S. at 20-21; Tornillo, 418 U.S. at 258.
189. 427 U.S. 347, 371 (1976).
190. 447 U.S. 74 (1980).
191. 117 S. Ct. 2130 (1997).
192. Pruneyard, 447 U.S. at 75.
193. Id. at 87.
194. Wileman Bros., 117 S. Ct. at 2139.
195. The PICS system does not contemplate the introduction of additional information beyond the rating labels themselves, so a speaker-producer has no opportunity to explain or disavow the rating. See supra note 112 and accompanying text.
196. See supra notes 37-48 and accompanying text.
197. In Wileman Bros., the Court noted that the "mere fact that objectors believe their money is not being well-spent does not mean [that] they have a First Amendment complaint." 117 S. Ct. at 2140 (citations omitted).
198. See supra note 43 and accompanying text (describing how one ratings system, RSACi, forces self-raters using the system to affirm an "agreement" between the rater and the system developers, and noting that the current agreement contains provisions for the developers themselves to rate pages if disputes arise). To the extent that the ratings systems developers inject themselves into the rating and labeling process, arguments in favor of forced association only increase.
199. Objectivity can be increased by closely defining the content that each level can contain. See supra note 46 (discussing the great detail involved in the RSACi rating system definitions). This interest in objectivity can be as easily ascribed to a desire for accuracy as to any desire to impose ideology.
200. Professor Lessig in particular notes that a chief danger of PICS lies in what he calls "upstream" filtering rather than simply user filtering. See Lessig, supra note 8, at 660-61. I share his concern on this point - upstream filtering raises issues quite different from those discussed above. But I think that a compelled association approach to this concern may be more helpful than the attack on the public-private distinction that Lessig suggests. Under a compelled association approach, ratings systems supported by a PICS-enforcing statute would become more likely to be seen as "compelled" as they moved further upstream, because of market power and scope of end users covered. Admittedly, my approach relies upon a PICS-enforcing statute to create a nexus to the government. Then again, so do other analyses. See, e.g., Lessig, supra note 8.
201. See Internet Freedom and Child Protection Act of 1997, H.R. 774, 104th Cong. (1997); Family-Friendly Internet Access Act of 1997, H.R. 1180, 104th Cong (1997).
202. See Safe Schools Internet Act of 1998, H.R. 3177, 104th Cong (1998).
203. The Federal Government used a purchasing-type market-influencing technique to promote the development of key escrowed encryption technology. See Michael Froomkin, The Metaphor is the Key: Cryptography, the Clipper Chip, and the Constitution, 143 U. PA. L. REV. 709, 769-772 (1995).
204. For excellent doctrinal and theoretical treatments of the issue, see David Cole, Beyond Unconstitutional Conditions: Charting Spheres of Neutrality in Government-Funded Speech, 67 N.Y.U. L. REV. 675 (1992);
205. Kathleen Sullivan, Unconstitutional Conditions, 102 HARV. L. REV. 1413 (1989);
206. Richard Epstein, The Supreme Court 1987 Term - Foreword: Unconstitutional Conditions, State Power, and the Limits of Consent, 102 HARV. L. REV. 4 (1988).
207. See Rust v. Sullivan, 500 U.S. 173 (1991).
208. See Arkansas Writers Project, Inc. v. Ragland, 481 U.S. 221 (1987).
209. Sullivan, supra note 192, at 1416.
210. See id. at 1418-19; Epstein, supra note 192, at 15; Cole, supra note 192, at 681-82.
211. See, e.g., Sullivan, supra note 192, at 1419.
212. The Court has long recognized a right to receive information inherent in the First Amendment. See, e.g., Board of Educ. v. Pico, 457 U.S. 853, 867-68 (1982). Note that the end users - who of course are not privy to the bargain - would still perhaps be able to bring a First Amendment challenge against the school or library on the basis of government-imposed filtering. See supra Part III.
213. See, e.g., Ambach v. Norwick, 441 U.S. 68, 70, 81 (1979) (upholding a state statute forbidding teacher certification of non-citizens); Meyer v. Nebraska, 262 U.S. 390, 401 (1923) (noting that "the state may do much, go very far, indeed, in order to improve the quality of its citizens").
214. See Miami Herald Co. v. Tornillo, 418 U.S. 241, 256 (1974).
215. This assumes the "editorial" right is burdened enough to raise the unconstitutional conditions issue in the first place. It is plausible that the Court would not consider the "burden" placed upon the right - the alteration in local editorial control - to rise to the level required. It is well-recognized that the constitutional interest at issue must rise to the level of a right ordinarily protected by strict judicial review. See Sullivan, supra note 192, at 1427. So the operative question is whether, standing alone, the removal of some editorial control over Internet access by the Federal Government would gain strict review. In my view, it probably would, though it is not the clearest case.
216. GERALD GUNTHER & KATHLEEN SULLIVAN, CONSTITUTIONAL LAW 1318 (1997).
217. See id. at 1319.
218. See FCC v. League of Women Voters, 468 U.S. 364 (1984).
219. See Regan v. Taxation With Representation, 461 U.S. 540 (1983).
220. An alternative approach that the Court might take would be in view of the prohibition on selective subsidization on the basis of viewpoint discrimination. See Rosenberger v. Rector and Visitors of the Univ. of Virginia, 515 U.S. 819, 831-837 (1995) (striking down a public university's refusal to support financially a religious student group on the basis of viewpoint discrimination). Assuming in this case that the Federal Government enforced filtering of sexual material, the "viewpoint" that would be discriminated against would probably be "sexuality," which is not among the range of viewpoints that ordinarily leaps to mind when considering viewpoint discrimination. But it's no more odd than the Court's invocation of "religion" as a viewpoint in Rosenberger.
221. It could also more simply require that any filtering and blocking options be turned "on" by default, a less extreme case that nonetheless might have similar market-influencing effects.
222. This is probably a stretch, but I would argue not wholly implausible. If the browser manufacturers were otherwise near indifference regarding the filtering and blocking features (say because of some significant consumer demand), then the addition of even some governmental purchasing influence might make a difference.
223. For a speaker-producer, hyperlinks from other pages are among the most valuable assets. Indeed, the provision of hyperlinks (usually via advertisements) is a foundation of the current commercialization of cyberspace.
224. See, e.g., Link Law on the Internet: A Panel Discussion, 38 IDEA 197 (1998).
225. I ignore here the claims of federal employees, who may themselves have a claim under a right-to-receive information theory. See United States v. National Treasury Employees Union, 513 U.S. 454, 470 (1995) (noting the public's right to receive information); Board of Educ. v. Pico, 457 U.S. 853, 867 (1982) (describing the importance of the right to receive ideas). Obviously, this claim is not unique to this context.
226. For example, those challenging the Federal Government's restrictions on the export of encryption technology rely heavily upon the claim that computer software is speech. See Bernstein v. Department of State, 922 F. Supp. 1426, 1436 (N.D. Cal. 1996) (holding that cryptographic source code is speech). See generally R. Polk Wagner, The Medium Is the Mistake: The Law of Software for the First Amendment, 51 STAN. L. REV. 387 (1999).
227. See Pico, 457 U.S. at 867 (quoting Madison to emphasize the right's importance). For an excellent treatment of the right to receive information in the cyberspace context, see Julie E. Cohen, A Right to Read Anonymously: A Closer Look at 'Copyright Management' in Cyberspace, 28 CONN. L. REV. 981 (1996).
228. See Epstein, supra note 192, at 7.
229. See id. at 8.
230. This is obviously an area for further exploration. Without the application of unconstitutional conditions, I see no other avenue for the end users and speaker-producers to challenge the restraints on their expression - the public-private distinction is even stronger in this context (where the government is acting in its capacity as a purchaser rather than regulator) than in the compelled speech context noted above. Again, because the architecturalist view does not recognize the significance of the public-private distinction, it seems of no more help here than in the prior examples. See supra notes 165-66 and accompanying text.
231. Sullivan, supra note 192, at 1427.
232. See, e.g., Rust v. Sullivan, 500 U.S. 173 (1991).
233. Regan v. Taxation With Representation, 461 U.S. 540 (1983).
234. See id. at 547-50.
235. See Arkansas Writers' Project, Inc. v. Ragland, 481 U.S. 221, 229-32 (1987).
236. See Rust, 500 U.S. at 197; Taxation With Representation, 461 U.S. at 544.
237. See Arkansas Writers' Project, 481 U.S. at 233 (applying strict scrutiny to the selective application of a sales tax to magazines and holding the tax unconstitutional).
238. Though it would seem to apply as well to the hyperlinking issue.
239. See, e.g., Edmonson v. Leesville Concrete Co., 500 U.S. 614 (1991); TRIBE, supra note 81, § 12-4; LAURENCE H. TRIBE, CONSTITUTIONAL CHOICES 246-48 (1985) [hereinafter TRIBE, CONSTITUTIONAL CHOICES].
240. See, e.g., President's Remarks, supra note 2.
241. See Lessig, supra note 8, at 642; Matthew L. Spitzer, An Introduction to the Law and Economics of the V-Chip, 15 CARDOZO ARTS & ENT. L.J. 429, 433 n.12 (1997).
242. Pub. L. No. 104-104, § 551(c), 110 Stat. 56, 141 (1996).
243. See id. § 551(e).
244. See id. § 551(b).
245. See TRIBE, CONSTITUTIONAL CHOICES, supra note 225, at 246.
246. See Sullivan, supra note 192, at 1415.
247. See supra note 214-216 and accompanying text.
248. On this conclusion I differ with the analysis conducted by Professor Spitzer. He notes four reasons that he believes state action is present in the V-Chip case: (1) Congressional suggestions that the governmental grant of digital bandwidth be tied to a system of content rating; (2) requiring the V-Chip in the television set itself; (3) governmental threats, advocacy, and review of the system; and (4) the requirement that ratings be transmitted if they exist. See Spitzer, supra note 227, at 433-51. Factors (1), (2), and (4), however, seem more appropriately analyzed under the unconstitutional conditions doctrine than under state action. In these three cases, state action unquestionably exists (I'm assuming that (1) was actually legislated - it is simply irrelevant if it was not); the task now is to show how the state action impermissibly forces the distributor to label her show. For example, if the Congress conditioned the grant of additional spectrum on the imposition of ratings, then the question is not simply whether the government acted to burden a right, but whether this constitutes an impermissible burden on First Amendment rights - an impermissible bargain. Similarly, requiring the V-Chip in TV sets is itself a clear governmental act - the question is whether the act conditions access to those television sets upon the use of a ratings system. A very analogous analysis would exist for factor (4). The question of the relationship between an existing government act and a burdened right is one for unconstitutional conditions rather than state action. Factor (3), however, raises questions of state action, but appears to fall far short of the level of "action" required: the "entanglement" principle that Spitzer relies upon in his analysis, see id. at 439, is more properly viewed through what Professor Tribe calls the "close-up lens" form of analysis. TRIBE, CONSTITUTIONAL CHOICES, supra note 225, at 248. This lens looks at the nexus between a state activity and the person or entity who might be blamed.
249. See id. It is not a theory of coercion. Cf. Spitzer, supra note 227, at 442 ("[T]he drafting of both the present ratings and brawnier new system was as voluntary as facing a firing squad.") (quoting Howard Rosenberg, Holdout by NBC to Provide Real Test, L.A. TIMES, July 30, 1997, at F1). The nexus to an actual state regulation is a necessary element of the "close-up lens" analysis.
250. See TRIBE, CONSTITUTIONAL CHOICES, supra note 225, at 250. Reducing state action to "pressure from President Clinton, threats from congressmen," see Spitzer, supra note 227, at 444, or even the non-binding recommendations of the FCC would go a long way toward making the requirement for state action meaningless.