47 USC 230 Talk at Fordham

By Eric Goldman

A couple weeks ago, I gave a 12 minute talk at Fordham Law School as part of a day-long conference on intermediaries. My talk notes:


230(c)(1) means websites and other actors aren’t liable for third party content…PERIOD. The “period” makes lawyers think surely they can devise a way around the statute. However, 230 is an incredibly robust immunity. There have been 100+ cases interpreting it in the past 13 years, and only a very small handful have led to defense losses. Plaintiffs lawyers who think they can “outsmart” 230 are probably wrong; many clever arguments have already been tried and failed.

Nevertheless, there are two bona fide limits on the “period.”

First, the statutory exclusions:

* ECPA = null set

* Federal crimes. Examples: gambling, obscenity, child porn. But state crimes are preempted

* “Intellectual property claims.” Federal IP claims (copyright and trademark) are not preempted, but there is a split of authority on state IP claims (state copyright, state trademark, trade secret, publicity rights, hot news). In the 9th circuit: they are preempted. But in other jurisdictions, they are not preempted.

Second, a website’s marketing representations. The theory of the case is to hold a website liable for its first-party statements, not third party content. But what if the marketing representations are rendered untrue by user content/behavior? Ex: this social networking site is “safe” for kids. In Mazur v. eBay, eBay represented that a third party service was “safe.” The court says 230 doesn’t apply. A bigger problem is when plaintiffs try to turn EULA negative covenants into affirmative marketing representations

From a policy standpoint, 230’s “problem” is that it “breaks” tort law the way we learned it in law school. It’s black letter common law that a participant in tortious conduct is liable for the tort. As applied in the publishing context, exercising editorial control over content creates liability for that content. To avoid such liability, one must be a “passive conduit” of third party content.

230 breaks apart those principles: active participation/control do not create liability:

…even if they receive a C&D/takedown notice

…even if they prescreen or manage the database or edit specific entries

…even if they profit from, or take ownership of, the content

This helps explain why bright judges impressed with their analytical abilities resist 230—it doesn’t comport with standard legal reasoning.

From my perspective, 230’s breaking of traditional tort law is a feature, not a bug. 230 prevents lopsided databases, i.e., databases filled only with positive comments because all of the negative comments have been taken down. I explain more about the lopsided database problem here. Further, as I will explore in my Economics of Reputational Information project, 230 has contributed to the most robust reputation ecosystem we’ve ever seen–so many existing reputational systems are broken, but online reputational mechanisms (such as product reviews) are among the healthiest and deepest reputational systems ever. This isn’t to say that 230 is cost-free–it’s not–but every system has its costs, including a system that creates greater liability.


Blog posts about the talk:

* Fordham’s live-blogger

* Rebecca Tushnet