Introduction

A plethora of bills trying to protect children from exposure to indecent material on the Internet have been passed through state and federal levels. These laws include the Communications Decency Act of 1996 and the Child Online Protection Act of 1998 (Electronic Privacy) and similar laws in several states (Glod). Almost all of these bills aiming to restrict children's Internet access have been challenged and struck down as unconstitutional.

Congress enacted the Children's Internet Protection Act (CIPA) under HR 4577 in December 2000 (Thomas). This law basically requires all schools and libraries that receive federal money to supplement their Internet access to install filters. The report Congress commissioned to investigate a solution to children's Internet access was not released until May 2, 2002 (National Research Council). In fact, Sara Fitzgerald notes that more than once Congress's own commissions to study the issue recommended that they not mandate filtering, yet they do so anyway.

Since implementing my research in January 2002, a legal challenge against CIPA has been heard by a three-judge panel in Philadelphia. Several entities brought suit against the government to halt this far-reaching law. The case lasted seven days and ended on April 3, 2002 (ALA). It was decided that the law violated the First Ammendment and could not be applied to libraries, though it did not address the issue of schools being forced to filter.

In early 2003, the case was heard by the Supreme Court. (Transcripts of the court session are available at http://supremecourtus.gov/oral_arguments/argument_transcripts/02-361.pdf [or in unofficial HTML format at http://sethf.com/censorware/legal/cipa_sc.php].) The court decided on June 23, 2003 that because filters can be disabled, they do not infringe on First Ammendment rights. I, along with other filter critics, am highly disappointed with this decision. You can read the court's decision in its entirety at http://www.supremecourtus.gov/opinions/02pdf/02-361.pdf. The court seems to think that it is simple to approach a librarian and request to view information that has implicitly been marked as "bad."

One group that is heavily impacted by the imposed use of Internet filters is gay, lesbian, bisexual, and transgender (GLBT) teens. Emmalyn Rood, a young lesbian, testified before the CIPA legal panel in Philadelphia about Internet filters' detrimental effects on those teens seeking information about alternative sexualities (McCullagh). Before reaching the legal age of majority, these young adults are subject to the same rules imposed on elementary aged children. Though percentages vary, studies consistently find an increased number of suicide attempts in GLBT teens (Heredia). And GLBT teens often rely heavily on the Internet to reduce feelings of isolation, seek sexual health information, and meet others like them (Hillier, Kurdas, Horsley). It is imperative that GLBT teens have access to accurate information about themselves.

Many young people have questions about their sexuality and have found the Internet to be rich with resources, covering every topic under the sun. While not all content online is to be trusted, these teens have few other resources available. One of the Internet's benefits is the anonymous availability of all materials. The student who might fear checking out a library book on homosexuality does not face the same public exposure by surfing gay and lesbian Web sites. Although they now are faced with the burden of approaching a librarian to request access to GLBT information that is also blocked as pornographic, even where there is no mention of sexual acts.

Many Internet filters seem to equate homosexuality with pornography. GLBT sites are often categorized as sexual or pornographic, simply because they deal with GLBT issues. For example, N2H2's URL Checker categorizes Scarleteen's Gaydar site as both Sex and Pornography. CYBERsitter even includes a category for blocking for sites "promoting gay and lesbian activities." While individual families may deem this appropriate for home use, schools and libraries are only one click away from denying their students and staff access to all GLBT information.

Testing

To assess the effectiveness of the Internet filtering technologies available, I tested three popular products. The filters were installed on a PC running Windows 98 and each site was accessed simultaneously on a control computer, an iMac running OS X. The two computers shared a DSL connection via network. Whenever possible, access was checked by linking to sites from the same page (e.g., using a test page like http://www.oocities.org/spatalita/filters/reviewed.html which was accessed on both computers) to eliminate the necessity of typing in each URL and resulting in possible errors from typographical errors. This was not always possible, such as in the cases where the test pages containing the list of URLs were blocked by the filter. If this happened, care was taken to input exact URLs.

Reviewed sites

While not all content online is valid and reliable, there are myriad sites that offer useful information to GLBT teens. Many information professionals have reviewed such useful sites for online directories available to teens. Many public libraries, like Suffolk and Hennepin Counties, offer GLBT pathfinders on their Web sites. I tested access to reviewed sites on all filters and was disappointed with the results.

Each filter blocked access to a large number of reviewed sites. N2H2 even blocked a link to the Librarian's Index to the Internet's (LII) entire GLBT section. The LII is an index of Web sites that have each been "evaluated by librarians for their usefulness to users of public libraries." If sites have been deemed appropriate ands useful by information professionals, why do filters block them?

I began my research evaluating access to reviewed sites. However, I learned quickly that just because a site doesn't meet all of a reviewer's qualifications does not mean that it does not offer valuable information. For example, one of the standards for assessing a site's quality is author identification and credentials. GLBT Web sites may not list a "closeted" author (i.e. - a person who is not "out" about their homosexuality). This omission does not invalidate the site's resources.

Some Web pages offer little content and much fluff. However, people still seem to visit these sites. It is not up to a filtering company to decide what is useful, only to separate content into categories. This brings up the conundrum of easily dividing the Web into tidy categories. As in life, things are not so simple. There exists a fuzzy line between erotica and pornography, acceptable and tasteless. The government, by passing CIPA into law, gave the right to decide what is acceptable viewing to filtering companies.

N2H2

N2H2, also known as Bess, is one of the most popular filters. According to N2H2's Web site, "Over 40% of all schools in the U.S. that have chosen to filter Internet access have chosen N2H2" (N2H2). The company also boasts that their product "[e]liminates over- or under-blocking by utilizing the industry's leading categorized database" (N2H2). I did not find this to be the case (see Appendices A & B). N2H2 is overly broad in their blocking policies and very often filters sites that should be accessible, in both Maximum and School settings.

N2H2 faired very poorly in my tests. In each setting, over 100 sites that have been widely reported as erroneously blocked were still inaccessible. This includes such sites like "The Pen Is Mightier Than the Sword" http://www.skaro.com/rpgwrite.html, presumably because the title contains the letters that spell "penis." Links to popular children's games on Yahooligans were also blocked in N2H2.

N2H2 heavily denied access to GLBT sites. Blocks included Devlin's "Christian Crossdresser's Confession", a Google search for "gay and lesbian teenagers" (See Appendix E), the Sacramento International Gay & Lesbian Film Festival, and the Bisexual Resource Center. Transgender pages were also denied, such as the Trandgender Law Conference, the Intersex Society of North America, and a Transsexual Support site. Teens with N2H2 isntalled on their school or library computers will likely have a difficult time finding GLBT information online.

Cyber Patrol

Cyber Patrol was installed and tested with the "Full Nudity, Sexual Acts/Texts, & Gross Depictions/Texts" categories blocked. This filter did restrict access to legitimate sites (see Appendix C), though not as many as N2H2. However, this filter under-blocked and allowed a large number of pornographic sites to be fully accessed. By utilizing Yahoo's Sexual Entertainment directory (page no longer up - it was a directory of porn sites by type), I was able to access almost 70 hardcore sex pages. These pages have URLs such as http://www.filthysex.com and included titles such as "Bang Bus Group Sex Pics" and "Cum Shot Calamities."

Pornographic material may be easily accessed in Cyber Patrol, but is still difficult to reach GLBT sites. Lesbian Support, Bisexual Women of Toronto, and the Lesbian Herstory Archives were all inaccessible. However, testing the URLs in Cyber Patrols' Test-a-Site database reveals that these sites should all be accessible. Bennett Haselton documented the inaccuracies in this catalog on http://www.peacefire.org/. Cyber Patrol is not consistent in either the blocking of URLs or their reporting of such blocks.

Several inconsistencies were found with categorization in all three filters. Pages with duplicate content would be categorized, and subsequently blocked, differently. N2H2 would not allow access to Su Penn's Transsexualism FAQ at http://my.voyager.net/~supenn/tsfaq.html, but the new site at http://www.tsfaq.info/ came through. CYBERsitter blocked one of my test pages http://www.oocities.org/childrensservices/test.html, but an exact duplicate on another domain was allowed http://www.patalita.com/filters/reported.html (page no longer up). Cyber Patrol filtered Stop Prisoner Rape's original Web site http://www.igc.org/spr/, however their new domain with the same content was not blocked http://www.spr.org. Irregularities such as these turned up repeatedly and should not be discounted.

CYBERsitter

In contrast with the other filters that display some type of denial message, CYBERsitter "stop[s] information in transit or obliterat[es] words in context" (Schneider, 41). This filter was tested with the default filtering categories engaged: Adult, Sexually Oriented, Illegal Activities/Drugs, PICS Ratings Adult Violence, Hate/Intolerance, and Illegal Guns/Violence. I found CYBERsitter to block information in the following multitude of ways: not displaying any portion of the page for the following reasons: "Cannot find server," 500 error message (internal server error), or 404 error message (file not found); erasing words or parts of words from the displayed text; ceasing to load the page source at various points; and redirecting the browser to the wrong URL.

CYBERsitter not only blocks surreptitiously, it over-blocks surreptitiously (see Appendix D). Great detail was required to check the accuracy of filtering, due to misleading error messages. A large number of sites needed to be displayed on both computers to compare screens because the filtering approach that is not always conspicuous. Many times the omissions were obvious (see fig. 1); others were only noticed through careful perusal of the text (see figs. 2 and 3).

Erasing words from their context is problematic not only because it is not publicized. As Rose Lazoff pointed out to me, it may also cause copyright concerns. As fig. 3 demonstrates, CYBERsitter displays bisexual author Lopez's novel as titled ing for Mommy (emphasis mine). However, this is simply not the title of the book. Lopez's work was copyrighted and this may be a violation of copyright law. CYBERsitter is not allowing work to be reproduced in whole as is the case with Duffy's title (see fig. 1), nor does it warn that there may be such omissions.

Another indirect filtering approach employed by CYBERsitter is browser redirection. I found forty URLs not only inaccessible, but also redirected to another site. Sites such as http://www.lesbian.org and http://www.planetout.com were redirected without warning to the URL http://safe-site.com. (Author's note: try typing that URL into your browser a few times and see what happens.) Safe-Site is not a site in itself, but rather a URL that directs all traffic to a rotating number of sites like the Hubble Space Telescope site and the FBI's Top 10 Most Wanted page. It seems as though this redirection may be triggered by keywords, as many of the pages included words that may be related to pornography. For example, http://www.antichildporn.org (an organization committed to helping end child pornography) and http://www.siliconvalleygirl.com (an online Web education site) may trigger redirection due to the words "porn" and "silicon" in their URLs. The complete inability to access these sites, including several GLBT pages, is troubling. CYBERsitter caused problems even after the program was uninstalled. While the program is not actively filtering, several of the pages that were partially blocked are still not completely accessible. Forcing the pages to reload did not solve the problem. Only after deleting all temporary Internet files and page histories were the pages viewed in entirety.

Searches

I tested a variety of search words and Boolean phrases in Google and AltaVista's search engines to compare the results. I was also hoping to test restrictions to articles in electronic research databases, but encountered difficulties accessing databases, so this investigation was not possible. My hope was that more specific search words would yield more specific results in search engines. Unfortunately, that was not necessarily true. (See Appendix E.) AltaVista consistently produced pornographic hits, with 15 out of 17 searches producing some porn links. Google faired a bit better, with only 6 out of 17 yielding some porn. Though the company's "cost-per-click" AdWords Select™ offset Google's record. This advertising strategy causes ads to be displayed according to the input search terms. Pornographic ads were displayed even when search results were on target.

Porn's fascination with female sexuality also skewed the results according to search terms. Searches for lesbian teen(s) were more likely to return a higher percentage of pornographic hits in Google than gay teen(s) queries. In AltaVista, searches for these two topics all produced from 80-100% porn results (see fig. 4).

AltaVista: gay and teen - 8:2 AltaVista: gay teens - 8:2 AltaVista: "gay teens" - 10:0 Google: gay and teen - 2:8 Google: gay teens - 0:10 Google: "gay teens" - 0:10

AltaVista: lesbian and teen - 10:0 AltaVista: lesbian teens - 8:2 AltaVista: "lesbian teens" - 10:0 Google: lesbian and teen - 9:1 Google: lesbian teens - 4:6 Google: "lesbian teens" - 4:6

Fig. 4. First page search results from two search engines, showing ratio of pornographic to non-pornographic results (excerpt from Appendix E).

I was very disappointed by the search engine results. The amount of porn found illustrates the fact that gay and lesbian teens face quite an obstacle looking for GLBT information. The evaluation of sources for appropriate information can be taught through information and technology literacy instruction. Because unfortunately, just narrowing search terms does not necessarily reduce tangential results in the case of GLBT information. Information literacy classes can work to help teenagers develop an internal filter and help them find the information they need.

Conclusion

All levels of government have been active in trying to come up with a solution to "protect" children's Internet access. Unfortunately, these laws are passed hastily and do not resolve the issue. They mask the problem at best, and intensify it at worst. Internet filters will most likely never be a perfect fix to the challenge of Internet content. Parents, teachers, librarians, and lawmakers need to educate themselves about the issue and work together to provide a solution.

Internet filters do not perform the job they set out to accomplish. I found that N2H2, Cyber Patrol, and CYBERsitter all block large numbers of sites that should be accessible. Each of the filters also allowed content through that should have been blocked, according to the filtering parameters. Gay, lesbian, bisexual, and transgendered teenagers deserve access to information about their sexuality, which is not afforded through the use of filters. Imposed Internet filters restrict GLBT teens' free access to information during a time when they need it most.

Webmistress: Sara Patalita Paper originally written May 6, 2002. Last updated May 28, 2004.