patrons from accessing obscenity and child pornography.  Unless software filters are themselves perfectly effective at preventing patrons from accessing obscenity and child pornography, “[i]t is no response that [a less restrictive alternative] . . . may not go perfectly every time.”  Playboy, 529 U.S. at 824; cf.  Denver Area Educ.  Telecomm.  Consortium, Inc. v.  FCC, 518 U.S. 727, 759 (1996) ("No provision . . . short of an absolute ban, can offer certain protection against assault by a determined child.").

The government has not offered any data comparing the frequency with which obscenity and child pornography is accessed at libraries that enforce their Internet use policies through software filters with the frequency with which obscenity and child pornography is accessed at public libraries that enforce their Internet use policies through methods other than software filters.  Although the government’s library witnesses offered anecdotal accounts of a reduction in the use of library computers to access sexually explicit speech when filtering software was mandated, these anecdotal accounts are not a substitute for more robust analyses comparing the use of library computers to access child pornography and material that meets the legal definition of obscenity in libraries that use blocking software and in libraries that use alternative methods.  Cf.  Playboy, 529 U.S. at 822 ("[T]he Government must present more than anecdote and supposition.").

We acknowledge that some library staff will be uncomfortable using the “tap-on-the-shoulder” method of enforcing the library’s policy against using Internet terminals to access obscenity and child pornography.  The Greenville County Library, for example, experienced high turnover among library staff when staff were required to enforce the library’s Internet use policy through the tap-on-the-shoulder technique.  Given filters’ inevitable underblocking, however, even a library that uses filtering will have to resort to a tap-on-the-shoulder method of enforcement, where library staff observes a patron openly violating the library’s Internet use policy, by, for example, accessing material that is obviously child pornography but that the filtering software failed to block.  Moreover, a library employee’s degree of comfort in using the tap-on-the-shoulder method will vary from employee to employee, and there is no evidence that it is impossible or prohibitively costly for public libraries to hire at least some employees who are comfortable enforcing the library’s Internet use policy.  We also acknowledge that use of a tap on the shoulder delegates to librarians substantial discretion to determine which Web sites a patron may view.  Nonetheless, we do not believe that this putative “prior restraint” problem can be avoided through the use of software filters, for they effectively delegate to the filtering company the same unfettered discretion to determine which Web sites a patron may view.  Moreover, as noted above, violations of a public library’s Internet use policy may be detected not only by direct observation, but also by reviewing the library’s Internet use logs after the fact, which alleviates the need for library staff to directly confront patrons while they are viewing obscenity or child pornography.