problems faced by the manufacturers and vendors of filtering software are legion.  The Web is extremely dynamic, with an estimated 1.5 million new pages added every day and the contents of existing Web pages changing very rapidly.  The category lists maintained by the blocking programs are considered to be proprietary information, and hence are unavailable to customers or the general public for review, so that public libraries that select categories when implementing filtering software do not really know what they are blocking.

There are many reasons why filtering software suffers from extensive over- and underblocking, which we will explain below in great detail.  They center on the limitations on filtering companies’ ability to:  (1) accurately collect Web pages that potentially fall into a blocked category (e.g., pornography); (2) review and categorize Web pages that they have collected; and (3) engage in regular re-review of Web pages that they have previously reviewed.  These failures spring from constraints on the technology of automated classification systems, and the limitations inherent in human review, including error, misjudgment, and scarce resources, which we describe in detail infra at 58-74.  One failure of critical importance is that the automated systems that filtering companies use to collect Web pages for classification are able to search only text, not images.  This is crippling to filtering companies’ ability to collect pages containing “visual depictions” that are obscene, child pornography, or harmful to minors, as CIPA requires.  As will appear, we find that it is currently impossible, given the Internet’s size, rate of growth, rate of change, and architecture, and given the state of the art of automated classification systems, to develop a filter that neither underblocks nor overblocks a substantial amount of speech.

The government, while acknowledging that the filtering software is imperfect, maintains that it is nonetheless quite effective, and that it successfully blocks the vast majority of the Web pages that meet filtering companies’ category definitions (e.g., pornography).  The government contends that no more is required.  In its view, so long as the filtering software selected by the libraries screens out the bulk of the Web pages proscribed by CIPA, the libraries have made a reasonable choice which suffices, under the applicable legal principles, to pass constitutional muster in the context of a facial challenge.  Central to the government’s position is the analogy it advances between Internet filtering and the initial decision of a library to determine which materials to purchase for its print collection.  Public libraries have finite budgets and must make choices as to whether to purchase, for example, books on gardening or books on golf.  Such content-based decisions, even the plaintiffs concede, are subject to rational basis review and not a stricter form of First Amendment scrutiny.