Opinion
Nos. 01-56380, 01-56556.
December 3, 2003.
D.C. No. CV-00-09590-SVW.
Howard S. Fredman, Los Angeles, CA, M. Joey Lynch, Law Offices of M. Joey Lynch, Beverly Hills, CA, for Plaintiff-Appellant.
Annemarie Vels Heijn, Director, Netherlands Museums Assocation, Amsterdam, Netherlands, Stephen J. Newman, Latham Watkins, Robert P. Long, Kinkle, Rodiger Spriggs, Los Angeles, CA, for Defendant-Appellee.
Before CANBY, GOULD, and BERZON, Circuit Judges.
Order; Partial Concurrence and Partial Dissent by Judge Gould.
ORDER
The majority of the panel has voted to deny appellee's petition for rehearing. Judge Berzon has voted to deny the petition for rehearing en banc and Judge Canby so recommends. Judge Gould has voted to grant the petition for rehearing and petition for rehearing en banc.
The full court has been advised of the petition for rehearing en banc. A judge of the court requested a vote on en banc rehearing. The majority of the active judges have voted to deny rehearing the matter en banc. Fed.R.App.P. 35(f).
The petition for rehearing and the petition for rehearing en banc are DENIED. Judge Gould's dissent from denial of en banc rehearing is filed concurrently herewith.
I dissent from the denial of en banc review of this case. I remain convinced that the panel majority's interpretation of the statutory immunity found in 47 U.S.C. § 230(c) is wrong in light of Congress's intent, and will needlessly harm persons defamed on the Internet. Section 230(c) states: "[n]o provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider." Immunity thus requires, inter alia, that the communication must have been "information provided by another information content provider." "[I]nformation content provider," in turn, is defined as "any person or entity that is responsible, in whole or in part, for the creation or development of information provided through the Internet or any other interactive computer service." 47 U.S.C. § 230(e)(3).
The panel agreed that if a person who posted defamatory material on the Internet "develop[ed]" that material, that person would become the information content provider of the material and lose § 230(c) immunity. The panel majority held that Cremers did not "develop" the information by "merely editing portions of [Smith's] e-mail and selecting [it] for publication." Batzel v. Smith, 333 F.3d 1018, 1031 (9th Cir. 2003).
The panel majority held that § 230(c) requires proof that the person who submitted the information "provided" it for purposes of republication and that a publisher falls outside of the scope of § 230(c) immunity if the publisher could reasonably perceive that the provider of the information did not intend that it was "provided" for republication.
The panel majority's conclusion that pre-publication selection and editing is not "development" is not supported by the text, history, or precedent surrounding § 230(c). Initially, any entity that "creat[es]" or "develop[s]" information is an "information content provider." 47 U.S.C. § 230(e)(3). To avoid rendering "creation" superfluous, "development" must mean adding to or improving the information initially created by another "information content provider." Development is defined as "a change in the course of action or events or in conditions; a state of advancement; an addition; an elaboration." The New Shorter Oxford English Dictionary 654 (Thumb Index ed. 1993). See also www.m-w.com (definition of "develop" includes "1b: to make visible or manifest" and "3b(1): to make available or usable"). The ordinary usage of a "development" of information suggests change in, addition to, or novel presentation of, the information.
There should be little doubt, given ordinary usage that Congress presumably intended, that a publisher's affirmative choice to select certain information for publication for the first time on the Internet "develops" that information. To put the point more concretely, imagine a defamatory e-mail sent to both an on-line bulletin board for appellate litigation and to a popular appellate litigation blog. Let us say, for example, that the e-mail falsely stated that Judge X of the Y Circuit was paid by Z to render decisions in Z's favor. If the blogger decides to publish the e-mail, there is something qualitatively different about the e-mail as published on the appellate blog, as contrasted with the one posted on the bulletin board. The blogger's conscious decision to publish an e-mail would add, by virtue of his or her reputation and that of the blog, a layer of credibility and endorsement that would be lacking from the e-mail merely posted to the bulletin board. And being the first person to post the defamatory material on the Internet would be a novel presentation of the defamatory material.
An on-line bulletin board allows users to post messages on a central server and it allows any other person with access to the bulletin board to read that post. Batzel, 333 F.3d at 1027, n. 9.
To mention a few popular and respected legal blogs, see, for example, How Appealing, www.appellateblog.blogspot.com, SCOTUSBlog, www.goldsteinhowe.com/blog/index.cfm, The Volokh Conspiracy, volokh.com and Lessig Blog, www.lessig.org/blog/. The development argument is likely to hold true in other industries as well, including politics, www.instapundit.com, and software architecture, www.corfield.org/blog.
The panel majority agrees that if Cremers "developed" the defamatory content, he would lose § 230(c) immunity. The question is whether Cremers' actions are a development of the information. Similar to the blog hypothetical above, Cremers' actions constituted a change in, addition to, and novel presentation of, the information. Cremers changed Smith's e-mail (by deleting the key word "some" describing the inherited artwork) and provided a novel presentation thereof by formatting it for Cremers' e-mail newsletter. Moreover, Cremers added what credibility he and his organization may have had to Smith's bare allegations when Cremers selected and published Smith's e-mail. And posting the message on the Internet, as Cremers did, is a novel presentation of the information because Smith, as original sender, had not posted it. Cremers' decision to select, edit, and publish Smith's e-mail in an edition of the Museum Security Network's newsletter was a change in, an addition to, and a novel presentation of what Smith had already written.
To bolster the incorrect theory that "development" does not include editing information or selecting that information for publication, the panel majority points to Congress's "exclusion of `publisher' liability," Batzel, 333 F.3d at 1031, and argues that "development" does not include the exercise of traditional editorial functions. Id. However, this reading omits a key qualification within § 230(c); namely that immunity is provided for claims that "treat" the defendant as a publisher. This method of using the word "publisher" does not at all imply that Congress intended to immunize all of the functions exercised by a "publisher" as asserted by the panel majority, but that the only immunity Congress intended to provide was immunity from defamation-type claims. Cf. Zeran v. America Online, Inc., 129 F.3d 327, 332 (4th Cir. 1997) ("the terms `publisher' and `distributor'" in § 230(c) "derive their legal significance from the context of defamation law."). In the defamation context, "publisher" has a broad meaning that includes any person who automatically republishes or distributes the defamatory material. See, e.g., W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 113 (5th ed. 1984). "Publisher" in the context of § 230(c) must be read to encompass the broad meaning of "publisher" in defamation law. Thus, "publisher" was meant to define the type of claim that § 230(c) would immunize — e.g., a claim alleging defamation must prove the element of "publication" — not to define the meaning of "development" as found in § 230(e)(3). All claims that require proof of publishing or speaking (traditionally, libel and slander) would be immunized because the plaintiff could not "treat" the defendant as "publisher or speaker." On the other hand, claims that do not require proof of publication are not, and should not be, immunized. Because Congress limited immunity only to claims that require proof of the defendant being a publisher or a speaker, it is incorrect to hold that Congress intended that the word "publisher" in § 230(c) be used to shed light on when information content is sufficiently "developed" by the publisher to prevent it from being "information provided by another information content provider."
This conclusion is further warranted because Congress immunized defendants who are "treated as publisher or speaker . . ." implying that all written ("publisher") and oral ("speaker") defamation-type claims were to be covered. 47 U.S.C. § 230(c) (emphasis added).
Cf. Mainstream Loudoun v. Bd. of Trs., 2 F.Supp.2d 783 (E.D.Va. 1998) (§ 230 immunity did not extend to a claim under 42 U.S.C. § 1983 that a library blocking access to adult web sites (a traditional editorial function) violated the library users' free speech rights).
If Congress wanted to use this more specialized meaning of the word "development," as opposed to the ordinary meaning as previously described, it could have done so in the § 230(3) "[d]efinitions" section. Congress's deliberate silence on the definition of "development" should lead us to conclude that Congress intended for us to employ an ordinary meaning of development. See Will v. Michigan Dep't of State Police, 491 U.S. 58, 64, 109 S.Ct. 2304, 105 L.Ed.2d 45 (1989) (courts should follow ordinary usage of terms unless Congress gives them a specified or technical meaning); Pittston Coal Group v. Sebben, 488 U.S. 105, 113, 109 S.Ct. 414, 102 L.Ed.2d 408 (1988) (courts should follow the dictionary definition of terms, unless Congress has provided a specific definition). These principles, of course, should lead us to reject the majority's explanation that (1) Congress intended a definition of "development" that excluded traditional editorial functions and (2) Congress chose to demonstrate this intent by using the word "publisher" in § 230(c) rather than, less deviously, defining "development" in § 230(e). The panel majority's theory makes sense if Congress routinely played "hide the ball," but observers of the legislative process in the United States know that legislators and Congressional committees do all that they can to explicate the meaning and extent of what they legislate. We should not tread the majority's path, because here there is no reason not to abide the plain meaning of the statutory text. See, e.g., United States v. Providence Journal Co., 485 U.S. 693, 700-01, 108 S.Ct. 1502, 99 L.Ed.2d 785 (1988).
Further, it is not as if the panel reached this result by simply following the holdings of other circuits. Make no mistake: the panel majority coldly goes where no circuit court has gone before, reaching further and providing broader automatic immunity of the most callous and damaging defamation that anyone might maliciously post on the Internet. The majority cites four out-of-circuit cases for the proposition that "development" does not include the exercise of traditional editorial functions. Batzel, 333 F.3d at 1031 n. 18. These cases have no persuasive force in assessing Batzel's claim against Cremers. None of these cases dealt with similar facts, and none address the question of whether the pre-publication selecting and editing of another entity's information is a "development" of information, which would place the selector or editor outside the statutory immunity declared by Congress. Each of the four cases involved the actual or potential editing or removing of posted information after the information had already been posted on the Internet. No federal court of appeals, save the panel majority in Batzel, has ever held that statutory immunity should be given for the pre-publication selection and posting of harshly defamatory material.
Zeran v. America Online, Inc., 129 F.3d 327 (4th Cir. 1997) held that AOL was immune under § 230(c) against the claim that it had failed to remove defamatory messages posted on an AOL bulletin board by an unknown third party, refused to post retractions, and failed to screen for similar postings thereafter. In Zeran, the third party was completely responsible for the content of the bulletin board posts. AOL selected nothing to be published on the Internet.
In Ben Ezra, Weinstein and Company, Inc. v. America Online, Inc., 206 F.3d 980 (10th Cir. 2000), AOL published continuously updated stock quotation information based upon information provided by two independent third parties. AOL was determined to be immune under § 230(c), although AOL had informed the third parties of data errors and had removed erroneous data from the website after the information had been posted. Thus, AOL selected nothing to be posted, and the only "editorial function" exercised by AOL was deletion after the unadulterated information had already been posted.
In Blumenthal v. Drudge, 992 F.Supp. 44 (D.D.C. 1998), AOL was held immune under § 230(c) because plaintiffs could provide no factual support for the assertion that "AOL . . . had some role in the writing or editing the material in the Drudge Report." AOL did have the right (apparently never exercised) to remove content, direct [Drudge] to remove content, or require reasonable changes to content. Thus, "AOL was nothing more than a provider of an interactive computer service on which the Drudge Report was carried. . . ." Id. at 50. AOL conceded that parties could be joint information content providers, providing the examples of joint authors, and a lyricist and a composer. It is an apt analogy to say that the relationship of Smith and Cremers as similar to that of lyricist (Smith) and composer (Cremers).
Finally, in Schneider v. Amazon.com, Inc., 108 Wash.App. 454, 31 P.3d 37 (2001), the court held that Amazon.com was immune under § 230(c). A third party had posted defamatory comments of a forum where visitors could voice their opinions about the book. While Amazon had the right to edit or remove postings that violated published guidelines and had a "royalty-free right to use the review," Amazon was determined to not be an information content provider. Neither of these factors address the effect of pre-publication selection and editing of defamatory material.
Even more recent decisions do not address the precise question of pre-publication selection and editing, but instead focus on post-publication editing and/or screening. See Carafano v. Metrosplash.com, Inc., 339 F.3d 1119 (9th Cir. 2003); Green v. America Online, 318 F.3d 465 (3d Cir. 2003). Batzel remains the only case to address the "development" question.
It is one thing to say, as all of the cited cases have said, that Congress wanted to preserve the ability of a website operator to edit something posted by another, thereby diminishing harm to the public when defamation or harmful material are excised. It is quite another, and diametrically opposed to what Congress had in mind, to hold, as the panel majority mistakenly held, that Congress intended to immunize someone like Cremers who first places on the Internet a patently defamatory missive that the sender had not even aimed for Internet publication.
Finally, the panel majority argues that there is no functional difference between a selection and a screening approach, rendering the addition of Cremers' credibility moot. Batzel, 333 F.3d at 1032. But Congress recognized this difference when it granted immunity for good faith actions to "restrict availability" of objectionable material. Congress knew the difference between screening (or "restricting availability") and selecting (or providing availability). Moreover, the panel majority's argument is flawed by the fact that a screening approach leads to the inclusion on the Internet of more material than a selection approach, and less of an endorsement. This demonstrates the difference between selection and screening that Congress recognized: Screeners of matters posted by others get immunity; those who select what is posted do not. Section 230 demonstrates a clear Congressional intent to immunize screening and post-publication removal, but shows no corresponding Congressional intent to immunize pre-publication selection and editing. This line, drawn by Congress, is sensible, easy to defend, and easy to apply. By contrast, the line drawn by the majority is not sensible, cannot be defended, and will be difficult to define in proceedings related to what a recipient "reasonably" understood a sender to want, especially since immunity questions arise at the beginning of litigation. Congress provided immunity for web site operators who edit or remove what another has placed on the Internet. But Congress did not provide for immunity for web site operators who place, for the first time, defamatory information on the Internet.
Specifically, the statute provides immunity for "any action voluntarily taken in good faith to restrict access to or availability of material . . ." 47 U.S.C. § 230(c)(2)(A). This presupposes that the information is already posted or would be posted but for intervention by screening. Unless the offensive materials were already posted, how else could one talk about restricting access or availability? The majority asserts that by not selecting information for publication, availability is restricted. But when a person selects and posts information never posted before, that is promoting access, not restricting access, to material.
The panel majority's argument that the dissent's position cannot logically co-exist with immunity for those who edit data posted by others on a website virtually ignores Congress's intent and doesn't follow Congress's language. The majority would have to say that there are only two categories of information available to an Internet publisher: publishable information and not publishable information. If that were true, the information published will be the same regardless of whether one selects or screens. However, in the real world there are several types of information available to a publisher, including: (1) information that the publisher wants to publish; (2) information that the publisher does not affirmatively want published, but is willing to allow others to publish on the publisher's web site; and (3) information that the publisher does not want published. It is with respect to the second category that the difference between selection and screening is a difference that counts: A selection approach yields only the information in category (1) while a screening approach yields the information in both categories (1) and (2). The result of using a different approach toward the information provided by another generates different sets of information that are ultimately published. Thus, it is illogical for the majority to say that there is no difference between an inclusion (selection) and exclusion (screening) approach to publishing third-party information on the Internet. And, of course, when we look at Congress's language and intent, it is clear that Congress recognized this difference.
The majority argues that § 230 does not specify a preference for screening and post-publication removal, but that argument is incorrect. See supra note 9; 47 U.S.C. § 230(b)(3) (noting Congressional policy to encourage the development of user based blocking software); id. at § 230(b)(4) (stating the policy of removing disincentives to the development and utilization of blocking and filtering technologies). By contrast, the statutory text does not show any preference for or acknowledgment of selection. The legislative history points to the same conclusion. See 141 Cong. Rec. H8471 (daily ed. Aug. 4, 1995) (remarks of Rep. Goodlatte that there is a disincentive to "detect and remov[e] objectionable content."); Batzel, 333 F.3d at 1029-30 (noting that § 230(c) was intended to overrule Stratton Oakmont, Inc. v. Prodigy Services Co., 1995 WL 323710 (N.Y.Sup.Ct. May 24, 1995), where Prodigy used filing software and post-publication removal and was held liable for defamatory material posted on a bulletin board that Prodigy was monitoring).
Not only is the majority's interpretation unfaithful to the statutory text, it opens the door for any Internet publisher to amplify the defamatory words of any person who communicates, or reasonably seems to communicate, a desire that their defamatory missive be published. It can only be imagined that a malicious sender of defamatory material will want the publisher to post the information. Thus, in the worst cases of malicious defamation, the defamer will only be too happy to ask expressly for publication to spread damaging lies and accomplish an illicit purpose. In such a case, under the majority's non-test, the publisher surely would reasonably think that publication was intended, and immunity would surely and automatically follow, no matter how false and defamatory the content.
The most blatant and malicious defamers doubtless would request publication of their poisoned words on a blog or other publication with a large audience. History, as well as this case, tells us that such persons exist and that they will seek an audience for defamation. And if immunity automatically follows, why would any Internet publisher refuse to publish anything that might draw eyes to the website? Doubtless some would decline out of virtue, but others will almost certainly publish what brings profit. None of this would do any good for the person defamed within the lawless territory that the majority would make of the Internet.
Although defamation might be bad enough amplified by blogs and newsletters, imagine the damage if it were amplified by the Internet portal of a major news organization, e.g., www.cnn.com, www.msnbc.com, or www.fox.news.com. We might hope that these organizations would be responsible and not amplify the defamatory message to a world-wide audience. But, apart from the protections mandated by the First Amendment, see, e.g., New York Times v. Sullivan, 376 U.S. 254, 84 S.Ct. 710, 11 L.Ed.2d 686 (1964), the panel majority has demolished any remaining legal and financial incentive for these major news outlets to be responsible with what they publish on their Internet portals.
I do not believe that Congress intended to make, or ever would consciously make, the policy choice made by the panel majority. Human reputations, built on good conduct over decades, should be not so easily tarnished and lost in a second of global Internet defamation. Under the panel majority's rule, there might be a remedy against the initial sender, but there is no remedy against the person who willingly chooses, with no exercise of care, to amplify a malicious defamation by lodging it on the Internet for all persons and for all time. Unless this result were commanded by Congress, we should not create such a system.
In light of these concerns, a well-known and respected commentator has critiqued the panel majority's holding, see, e.g., John Dean, Defamation Immunity on the Internet: An Evolving Body of Law Has Been Stretched Beyond Its Limits (July 4, 2003), available at http://writ.findlaw.com/dean/20030704.html.
The panel majority's decision is not faithful to Congressional intent, will have a broad and potentially harmful impact in cases of misuse of the Internet, and should have been reconsidered by our entire court. I respectfully dissent from our decision not to do so.