Internet Filtering: Summary and Prognostications

Currently, the most widely used Internet filtering is URL black lists on proxy servers. This approach works very well for most applications, but runs into problems as you try to scale up to very large installations or try to provide per-user control of the filtering.

My predictions for the future:

• I believe that the scaling issues will be addressed primarily by using label servers in conjunction with routers and/or content-aware switches. The control issues will be addressed by providing client software which implements multi-user control of filtering, but which utilizes a label server for the actual filtering.

• Artifical intelligence will continue to be developed and will start to play a significant role in both databases and real-time filtering over the next several years.

• Human review will continue to be an important component of filtering both as a technical necessity and as a point of comfort for consumers.