Part 6 - Automated Filters and Obscenity

Another perplexing question about obscenity is the process you will use to keep it from your service. With any active community, the person-power required to remove "offensive" postings in a timely manner could be prohibitive. Because of this, many service providers have opted for automated software filters (that look for unacceptable words and remove them). However, this requires you to define absolute standards for obscenity, and when you try to do this you could be surprised at some problems.

Here are some examples we've encountered. On a BBS for one client, Daytona USA, they had elected to use the twit filter with the usual range of profanities selected. One of the standard words in that list is "dick". Of course, racing fans are all familiar with Dick Trickle -- when a fan attempted to write about his racing hero, the twit filter produced, "*CENSORED* Trickle." This example is actually rather funny, and we've all had no end to good laughs over it, but there are other more potentially destructive instances.

In the "webumentary", Hong Kong '97: Lives in Transition, we had created an open forum focused around discussion of the transition of Hong Kong back to mainland China and related issues. As this was a PBS project, the twit filter was again employed. In addition to profane language, PBS was also concerned with racist language, leading them to select the word "Jew" as one that needed to be censored. Unfortunately, at one point a man describing himself as a Jew joined in the conversation, producing the same *CENSORED* result. Once again, context changed the meaning of a word from racist to merely identifying.