Ninth Circuit Rules May Be Unlawful Host

The Ninth Circuit has just ruled (en banc) that the Web site is not entitled to immunity under Section 230 of the Communications Decency Act. (Props to Eric Goldman for the link!) The opinion, by Chief Judge Alex Kozinski, is typically lucid and holds, essentially, that Roommates falls outside the safe harbor because it creates (or at least induces the creation of) some of the content on the site. I see two fun aspects to the ruling on first read: its implications for Web developers (and business models!) and its implications for statutory interpretation.

First, gets in trouble because it asks users to classify themselves via drop-down boxes: you’re forced to describe yourself by gender, sexual orientation, and whether you have children. Moreover, users offering housing space must disclose who else is living there by gender and orientation, and users seeking housing must disclose who they’re willing to live with on those criteria. As a former Web developer, I’m quite fond of drop-down boxes; they limit users’ choices so you can more readily perform sorting and classification on the back end. (It takes extra code to distinguish “Detroit” from “detroit”, and don’t get me started on spelling mistakes.) The Ninth Circuit says – correctly, in my view – that Roommates falls outside the CDA 230 safe harbor by forcing users to select among a series of choices that arguably violate the Fair Housing Act.

My only worry here is that the import of this finding may be tempered: more efficient coders than I could easily create a housing-match site that works around this. (Consider “checkboxes”: you let users pick from a set of non-mandatory, non-mutually-exclusive choices that you supply. Then, you allow other users to search on those choices. Thus, instead of forcing you to pick from the options “Gay,” “Lesbian,” “Straight,” or “Bisexual,” I give you the opportunity to select one or more of those to identify yourself. My site displays that choice(s) on your profile page. It’s a bit less efficient in coding, but it does sidestep the 9th Circuit’s objection (at least, I think it does). It’s fun to see legal code shape computer code, but the latter is far more malleable and responsive – it can adapt to work around legal restrictions in many cases. If one thinks that how matches applicants with housing is problematic, then the availability of an easy technical workaround seems to be worrisome.

Second, there’s a neat fight over statutory interpretation – specifically, the word “develop” in 230(f)(3). (In short, you’re an “information content provider” if you are entirely or partly responsible for the “creation or development” of information on a site; providers of an “interactive computer service” can’t be treated as the publishers or speakers of information provided by a different ICP.) The majority defines “development,” which would cause Roommates or other search providers to fall out of the safe harbor, as “materially contributing to [the content's] alleged unlawfulness.” (at 3462) The dissent, advancing an alternative definition, looks to the dictionary, and finds the meaning as “gradual advance or growth through progressive changes.” (at 3491)

Not so, says the majority! (see FN22) Indeed, the same dictionary gives another, alternative definition: “making usable or available.” (at 3463) The effect of this fencing – dueling dictionaries – is to define the scope not only of a word, but of the Section 230 immunity from liability. Non-lawyers may think it’s a bit ridiculous to figure out how much Congress wanted to change the scope of, say, defamation law on-line (see dissent at 3477) by referring to a standard dictionary, but it’s totally standard practice for lawyers. The degree to which this invites strategic behavior by a reviewing court is manifest in this particular scuffle.

“The question is,” said Alice, “whether you can make words mean so many different things.”

“The question is,” said Humpty Dumpty, “which is to be master — that’s all.” (Source)

I’m not troubled by the decision at all on initial read – it seems to narrow CDA 230 a bit, towards passive sites – but I want to think a little about how it’ll affect sites that try to shape or tailor user content, or to provide metadata. More soon.

6 Responses to “Ninth Circuit Rules May Be Unlawful Host”

  1. As a current coder, interpretations like this really get under my skin. Can you really say that you are less of a “content provider” because the choices provided are not unitary? What about a multi-select dropdown? I mean really…

  2. [...] Volokh | Online Liability Blog | Internet Cases | Susan Crawford | Eric Goldman | Wired | Info/Law. [...]

  3. [...] check out, ABA Journal, Decision of the Day, and Info/Law in addition to the Volokh and Professor Goldman links I posted [...]

  4. Hey Chris – I think the opinion is a little odd from a coder’s perspective, too. But my other thought is that the site is really shaping the informational content of the posts. If the site were to go one step further and induce people into creating content that violated the FHA (imagine, as an exercise, if the form had a drop-box that indicated which races of potential roommates would be acceptable), that would seem far different legally than a site with an open comments field that people used for the same purpose. This might possibly be analogous to 512(a) of the Digital Millennium Copyright Act: if you’re just a carrier or host, and you don’t shape or alter the content, then you’re off the hook. What do you think?

  5. I agree that it could well be that the data that is solicited could create violations of other statutes such as FHA. However, to my view, in that case its not the fact that the site shaped the data that causes the problem, rather it is the manner that the data was shaped.

    To me, this seems different than the 512(a) case, where, taken to an extreme, the manner in which the data is shaped is irrelevant, it is the mere act of shaping that takes the site out of the safe harbor. I hate slippery slope arguments, the lazy intellectual’s fallback, but in the end, any manner or selection or data entry ends up shaping the data to a degree. If I place a textbox on a page, that indicates that the form of the data I am expecting is text. Is that enough to take my site outside of the safe harbor?

    However, from the another perspective, the text of the statute seems to indicate that it applies to “online service providers” and does that really just mean ISPs? If the statute were worded “internet service providers”, I would have little quibble with the data shaping distinction. You aren’t really just an ISP if you are shaping the content of the data *cough* Comcast *cough*. But, the phraseology “online” service providers was used, somewhat complicating the issue.

    if what the safe harbor is really intended to protect is the mere carriers of data, rather than other types of online data services, it all makes sense. However, I think that the statute is somewhat broader than that, and carried to an extreme would exempt virtually any site which solicits user data from the safe harbors.

    The real extreme is the Google example — Google solicits user input to tailor its search results and advertising to the requested data. This solicitation of text “shapes” what the user sees. However, I would posit that these types of search sites were intended to be covered by the safe harbors, otherwise they really have little teeth except for honest to goodness mere providers of the actual network transmission media and associated services.

  6. [...] week the venerable Roommate Connection, also an online service, failed to make a similar caseĀ in the 9th Circuit. Unlike Craigslist, which provides comment fields in which writers may publish [...]