Feedback on the Web Affiliation's empirical examine of Part 230 instances
Elizabeth Banker of the Internet Association has published "A review of the meaning and application of Section 230 based on more than 500 cases". This complements Prof. David Ardia's extensive empirical study of case 230 of the case a decade ago. It's great to take another look at the cases. (I will go into the methodological limitations in a moment.)
The summary of the report (in bold for emphasis):
The importance of section 230 is best demonstrated by the lesser known cases that escape the headlines. These decisions show that the law continues to function as Congress intended, and tacitly protects soccer parents from smear claims, nurse and police roundtables from harassment lawsuits, and local newspapers from liability for comment roll.
* * *
Let's take a closer look at the six key results of the report:
"Find # 1: A broad cross-section of individuals and organizations is based on Section 230."
My Comment: Google and Facebook qualify for Section 230, but anyone who takes Section 230 reform seriously should pay special attention to other services based on Section 230. The report mentions numerous beneficiaries: “Online users; Internet service providers and website hosts; Newspapers; Universities; Libraries; Search engines; Employer; Bloggers, website moderators and listserv owners; Social media company; Marketplaces; App stores; Anti-spam and anti-fraud tools; and domain name registrars. “The 230 reform will almost certainly hurt the less-announced 230 beneficiaries – and the pro-speech results they promote – while anchoring established companies.
Find # 2: Immunity under section 230 was the main basis for a court's decision in only 42 percent of the decisions examined. The courts called for the factual development of cases where the question was whether the platform played a role in the creation or development of the content in question. "
My Comment: Judges have several options for solving a case and a lot of discretion in choosing between those options. Sometimes they choose procedural techniques to avoid talking about substance at all. Sometimes they comment on any issue contested by the parties. In other cases, they have chosen to address a dispositional problem and bypass the others. I routinely see clearly doomed cases in which the court rejected or refused Section 230 just to dismiss the case for other reasons. (Here's a good example) For this reason, section 230 can be helpful for judges, since the law provides a fast lane to get the potentially obvious result, rather than requiring a time-consuming and expensive process that will clearly lead to the same result.
"Finding No. 3: A significant number of claims in the decisions failed without application of Section 230 because the courts found that they were unfounded or dismissed them for other reasons."
My comment: The amendment to section 230 has no advantage if other laws prescribe the same material result. In these cases, the reform screwed up § 230 without consideration; and the reform made everyone worse by increasing the cost of case law.
Find # 4: 43 percent of core decision-making claims related to defamation allegations, as did the Stratton Oakmont case against Prodigy Services, which spurred Section 230 to pass. "
My comment: This percentage seemed high. I wonder if older cases have distorted this. I also wonder if this could be an artifact of the case indexing method.
"Finding No. 5: Cases in which allegations of criminal activity were made by the federal or state governments were often decided in whole or in part for other reasons."
My comment: This area would definitely benefit from further studies.
"Finding # 6: Section 230 protects providers who moderate content, but typically by using subsection (c) (1) instead of (c) (2)."
My Comment: Since courts interpret section 230 (c) (1) to cover content and account removals, the use cases for section 230 (c) (2) have been limited. However, section 230 (c) (2) (B) still plays an essential role for anti-threat software vendors, at least until it is jeopardized by the Enigma v Malwarebytes judgment.
* * *
Comments on the methodology of the paper
The report said in a footnote: “This summary and the review of the impact assessment do not provide a comprehensive overview and, given the complexity of the attempt to generalize from litigation, are far from perfect. We show that a more extensive review by an institution better equipped to conduct such a study would add significant value to the conversation. "
Amen. However, I am not volunteering for this broader review. Empirical case law projects are difficult. So I hope someone else does as suggested. I would like to see the work.
The IA data set contains 516 cases. The methodology explained that the data set was originally created using a snowball methodology, followed by searches in free legal search engines (but apparently not in Lexis / Westlaw).
On July 27, I inquired about Shepard's Lexis and listed 977, citing references to 47 USC 230-756 federal decisions and 219 state court decisions. Here is the breakdown of the federal court by courts in each circuit:
The tool shows the relatively linear growth of quotations over time:
The 977 Shepard quotes are not directly comparable to the 516 cases in the IA data set. The IA record appears to have indexed a case only once, even if it made multiple decisions. while Shepards counts multiple decisions in the same case.
On the other hand, Shepard's 977 record is still incomplete. Westlaw's citation number is 824, and I would expect Westlaw to contain 100-200 citations that Lexis does not. I also routinely post opinions that none of the indexes contain. So I think the entire universe of case quotations in section 230 is more like 1,200+. In this case, the IA record was likely to contain less than half of the indexable cases.