This is the story from the guy, who worked as an admin at school.
Most large scale companies and well established school systems use a filter service that uses the filter definitions. At that school they use Lightspeed. Lightspeed installes a server that filters all traffic from the internet into the school's network. Every day Lightspeed downloads the sites addresses their customers have blocked by hand, process that definitions by their proprietary algorithms and serve out a new list of definitions. Additionally, Lightspeed spiders are also crawling the web and assigning categories based on content, keywords, and server owner, as well as other factors. Lightspeed (the service) also uploads the defintions to the local filtering server which are then joined to our local filter. Most proxy websites are categorized under proxy/security. That ones that the filter hasn't caught, admins get from the students by walking around and looking at the screens. :)
Even more, filtering servers can also filter web sites through keywords and redirects. If anyone try to search the term "proxy" the filter blocks that search query. When students type the myspace.com, the filter automatically redirect to yahoo.com. I suppose most enterprise-level filters act in the same way.