Comment by hn_acker | Hacker News Reader

Comment by hn_acker | Hacker News Reader

Parent Full thread

For anyone making claims about what the authors of Section 230 intended or the extent to which Section 230 applies to targeted recommendations by algorithms, the authors of Section 230 (Ron Wyden and Chris Cox) wrote an amicus brief [1] for Google v. Gonzalez (2023). Here is an excerpt from the corresponding press release [2] by Wyden:

> “Section 230 protects targeted recommendations to the same extent that it protects other forms of content presentation,” the members wrote. “That interpretation enables Section 230 to fulfill Congress’s purpose of encouraging innovation in content presentation and moderation. The real-time transmission of user-generated content that Section 230 fosters has become a backbone of online activity, relied upon by innumerable Internet users and platforms alike. Section 230’s protection remains as essential today as it was when the provision was enacted.”

[1][PDF] https://www.wyden.senate.gov/download/wyden-cox-amicus-brief...

[2] https://www.wyden.senate.gov/news/press-releases/sen-wyden-a...

This statement from Wyden's press release seems to be in contrast to Chris Cox's reasoning in his journal article [1] (linked in the amicus).

  It is now firmly established in the case law that Section 230 cannot act as a shield whenever a website is in any way complicit in the creation or development of illegal content.

  ...

  In FTC v. Accusearch,[69] the Tenth Circuit Court of Appeals held that a website’s mere posting of content that it had no role whatsoever in creating — telephone records of private individuals — constituted “development” of that information, and so deprived it of Section 230 immunity. Even though the content was wholly created by others, the website knowingly transformed what had previously been private information into a publicly available commodity. Such complicity in illegality is what defines “development” of content, as distinguished from its creation.

He goes on to list multiple similar cases and how they fit the original intent of the law. Then further clarifies that it's not just about illegal content, but all legal obligations:

  In writing Section 230, Rep. Wyden and I, and ultimately the entire Congress, decided that these legal rules should continue to apply on the internet just as in the offline world. Every business, whether operating through its online facility or through a brick-and-mortar facility, would continue to be responsible for all of its own legal obligations.

Though, ultimately the original reasoning matters little in this case, as the courts are the ones to interpret the law. In fact Section 230 is one part of the larger Communications Decency Act that was mostly struck down by the Supreme Court.

EDIT: Added quote about additional legal obligations.

[1]: https://jolt.richmond.edu/2020/08/27/the-origins-and-origina...

The Accusearch case was a situation in which the very act of reselling a specific kind of private information would've been illegal under the FTC Act if you temporarily ignore Section 230. If you add Section 230 into consideration, then you have to consider knowledge, but the knowledge analysis is trivial. Accusearch should've known that reselling any 1 phone number was illegal, so it doesn't matter whether Accusearch knew the actual phone numbers it sold. Similarly, a social media site that only allows blackout challenge posts would be illegal regardless of whether the site employees know whether post #123 is actually a blackout challenge post. In contrast, most of the posts on TikTok are legal, and TikTok is designed for an indeterminate range of legal posts. Knowledge of specific posts matters.

Whether an intermediary has knowledge of specific content that is illegal to redistribute is very different from whether the intermediary has "knowledge" that the algorithm it designed to rank legally distributable content can "sometimes" produce a high ranking to "some" content that's illegal to distribute. The latter case can be split further into specific illegal content that the intermediary has knowledge of and illegal content that the intermediary lacks knowledge of. Unless a law such as KOSA passes (which it shouldn't [1]), the intermediary has no legal obligation to search for the illegal content that it isn't yet aware of. The intermediary need only respond to reports, and depending on the volume of reports the intermediary isn't obligated to respond within a "short" time period (except in "intellectual property cases", which are explicitly exempt from Section 230). "TikTok knows that TikTok has blackout challenge posts" is not knowledge of post PQR. "TikTok knows that post PQR on TikTok is a blackout challenge post" is knowledge of post PQR.

Was TikTok aware that specific users were being recommended specific "blackout challenge" posts? If so, then TikTok should've deleted those posts. Afterward, TikTok employees should've known that its algorithm was recommending some blackout challenge posts to some users. Suppose that TikTok employees are already aware of post PQR. Then TikTok has an obligation to delete PQR. If in a week blackout challenge post HIJ shows up in the recommendations for user @abc and @xyz, then TikTok shouldn't be liable for recommendations of HIJ until TikTok employees read a report about it and then confirm that HIJ is a blackout challenge post. Outwardly, @abc and @xyz will think that TikTok has done nothing or "not enough" even though TikTok removed PQR and isn't yet aware of HIJ until a second week passes. The algorithm doesn't create knowledge of HIJ no matter how high the algorithm ranks HIJ for user @abc. The algorithm may be TikTok's first-party speech, but the content that is being recommended is still third-party speech. Suppose that @abc sues TikTok for failing to prevent HIJ from being recommended to @abc during the first elapsed week. The First Amendment would prevent TikTok from being held liable for HIJ (third party speech that TikTok lacked knowledge of during the first week). As a statute that provides an immunity (as opposed to a defense) in situations involving redistribution of third-party speech, Section 230 would allow TikTok to dismiss the case early; early dismissals save time and court fees. Does the featured ruling by the Third Circuit mean that Section 230 wouldn't apply to TikTok's recommendation of HIJ to @abc in the first elapsed week? Because if so, then I really don't think that the Third Circuit is reading Section 230 correctly. At the very least, the Third Circuit's ruling will create a chilling effect on complex algorithms in violation of social media websites' First Amendment freedom of expression. And I don't believe that Ron Wyden and Chris Cox intended for websites to only sort user posts by chronological order (like multiple commenters on this post are hoping will happen as a result of the ruling) when they wrote Section 230.

[1] https://reason.com/2024/08/20/censoring-the-internet-wont-pr...

I'm skeptical that Ron Wyden anticipated algorithmic social media feeds in 1996. But I'm pretty sure he gets a decent amount of lobbying cash from interested parties.

Reply on news.ycombinator.com