How Internet Filters Work
Filters operate on a system of categories. Websites, or sometimes individual web pages, are categorized by filter companies. The library's filter administrator utilizes the categories to build filter profiles. For example, the adult filter profile might allow all categories of content to pass through except items categorized as "sexually explicit." The children's filter profile would undoubtedly block the "sexually explicit" content as well as other categories deemed inappropriate for children such as "hate," "firearms," and "violence." The filter company decides how each site will be categorized.
Filter companies fiercely protect their process for categorizing websites and equally fiercely protect the websites identified within each category. Part of the value of the filter is in the number of websites categorized, because sites that have yet not been categorized will not necessarily be blocked. Ironically, librarians - professionals trained to catalog and evaluate content - subcontract their cataloging job to Internet filter companies when they install a filter. Unlike librarians, the subcontractors are not information professionals, they typically use automated methods to classify the 3 billion web pages on the Internet.
The features available in state-of-the-art filters are too numerous to recount here. For a thorough summary of filter features and to compare filters, feature-by-feature, visit libraryfiltering.org. However, certain features are particularly important for libraries, such as the ability to control what is blocked, how overriding blocked pages is handled, how granular the blocking is (page, site, domain, IP address), and what information is presented to the end users when a blocked page is encountered.
Most library filters leave the choice of what to block in the hands of the filter administrator who selects the categories to block. The filters that don't allow the administrator to set up filter profiles and the categories that will be blocked for each profile are generally not used in the library. Such products might be suitable for home use but are not appropriate in a library setting.
Most filters provide some mechanism for overriding blocked pages either on-the-fly using an administrator password, or by adding sites to an "always allow" list which supersedes the block on a page caused by its categorization. This override capability provides the local administrator the ability to fix errors the filter company has made in their classification process, or to modify the filter company's classification system to more closely match the library's policies.
Because most filters do not disclose the websites contained within each category, these adjustments to the filter categories must be made by the filter administrator as they are discovered. Some filters provide more granularity in their blocking behavior than others. For example, a small number of filters allow the administrator to block certain file types (such as GIF, JPG, BMP, TIFF) within a category. Such granularity enables the filter administrator to block images within a category without blocking the text on the page.
Other filters are more gross in their blocking behavior and block the entire page, or even the entire domain. Some filters convert the domain to an IP address and block any websites sharing that IP address. Blocking shared IP addresses always results in over blocking.
Some filters rely solely on lists of URLs within categories. Other filters use a system of dynamic filtering. Dynamic filters analyze websites as they are accessed by the end-user and based on the analysis, categorize the page. As filters become more sophisticated, companies are using a dynamic filtering process to supplement their URL lists, if not replace them entirely.
When the end users encounter a blocked page, they are usually presented with a message advising them so. The default block page is often customizable and can be used to provide useful information to the patron about why the page has been blocked and what to do if they'd like the page unblocked. Filters that block pages without advising the patron that they've been blocked by the filter should be avoided.