Info/Law

The Brennan Center for Justice, located at NYU’s law school, put out a report on Internet filtering in U.S. schools and libraries. The study is essentially a literature review; it collects and summarizes most analyses of the effectiveness and errors of filtering software (such as NetNanny or SmartFilter) employed in response to the Children’s Internet Protection Act. (CIPA requires schools and libraries to put a “technology protection measure” in place, and to use it to block access to material that is obscene, child pornography, or harmful to minors. Institutions that fail to comply forfeit discounted telecommunications, under the e-Rate program, and, for schools, cannot use federal funds to pay for computers or access to the Net.)
The report is an excellent collection of resources for Internet types, though it’s very light on policy analysis. I wanted to highlight three things in particular:

1. Resnick, Hansen, and Richardson’s 2004 article on filtering software’s error rates makes an important point about the “denominator problem”: calculating overblocking depends how one measures the problem. Is the correct criterion the ratio of improperly blocked sites to acceptable sites, or improperly blocked sites to all blocked sites? This relates to a point the OpenNet Initiative struggles with in its reports: how does one measure the amount of the Internet that a given state filters? What’s the denominator? Even Google doesn’t know how many sites there are.
2. Employing filtering software transfers decision-making responsibility to private corporations, who decide (for example) what qualifies as “pornography.” This occurs both in terms of classification (sex ed vs. porn vs. art) and in terms of granularity – does the software require blocking all Web pages that discuss abortion, or can one refine the decision to exclude one position (pro-life vs. pro-choice, with standard caveats about these loaded terms) from access?
3. Judicial reasoning regarding the Internet is still too frequently substandard. In the case that challenged CIPA’s provisions regarding libraries, Chief Justice Rehnquist (writing for the majority) analogized filtering to librarian’s decisions regarding which books to purchase. (See p. 11 of the linked PDF.) This is not only unhelpful, it’s asinine. Books exist in a world of significant marginal acquisition (purchase) and maintenance (storage / shelf space) costs. Web pages do not. Different information channels require considerably different legal approaches to regulation, not tired analogy-driven thinking that should embarrass a first-year law student.

A final note: the Brennan Center does a good job of highlighting filtering software’s shortcomings. I wish the Center – and we as Internet users more generally – would face the hard choices, though. Filtering overblocks and underblocks – conceded. But we need to measure those harms (to kids who stumble across porn, or a blogger whose musings are erroneously censored) against the countervailing problems that filtering seeks to address. Access to information is vitally important, but it is not an absolute.

Filed under: Filtering, Intermediaries