Best of the ISP Lists

Filtering or Censorship?

An ISP asks for tips on efficiently filtering content. But does doing this constitute a violation of free speech?

[November 30, 1999]

On the ISP-Tech list in November 1999, TM posted a question about filtering porn:

"I run a small ISP in the remote North Fork Valley of Colorado and smut filtering is in high demand. Can someone tell me how to filter web pages by words instead of going out and finding all the smut sites I can and making a text file list to use for filtering?"

The question spawned a wide-ranging discussion that touched on many of the tricky issues surrounding filtering:

[AR wrote] "It sounds simple enough, at least in theory. Unfortunately, such an approach would not be effective, as many 'dirty' words can be perfectly kosher, depending upon the context in which they're used. Therefore, a URL-based filtering approach would be best."

[JL placed responsibility for filtering on customers] "As a network admin, I don't want the responsibility of censorship. The consumer can customize his filter to his own notion of whatever he considers objectionable. You (the ISP) don't get involved in thorny (and possibly legally expensive) censorship issues. We don't touch censorship—we leave that to the intelligence and sensibilities of our subscribers."

This evoked an opposing tide of opinion. Many respondents clearly believe that filtering does not constitute censorship:

[BW wrote] "You're not censoring if you're selling selective filtering to your clients.

[DS fired off these thoughts] "Censorship is when the government mandates that something can't be said. Filtering at the ISP level isn't censorship; it's an ISP's right. Service providers' networks are privately owned, and as owners ISPs have the right to choose what services they will sell and therefore what data they will allow on their LAN. You must have a good Terms of Service contract that clearly spells out that the service is filtered."

[KN added] "There's also a business case for filtering: Many potential customers view the Internet as loaded with useless and offensive material. Without filtering, we can't get these new users to step in. ISPs can charge extra for the service of filtering, and ISP marketing can include phrases like "The pure and clean, quality Internet Service. No reason to be anywhere else."

[JL guided the discussion back to the practicalities of the original post] "There's no question that any private individual or organization has every right to censor anything they wish. However, from a practical point of view, trying to automate the censorship process for widely differing subscriber sensibilities via currently available "broad brush" censoring software simply doesn't work. Such attempts tend also to censor information that some subscribers consider crucially important (such as medical issues dealing with body parts and/or sexual dysfunction).

"A better solution than messing with system-wide filtering would be to become a distributor of whatever consumer-level "child protection" censoring software the ISP feels best fits the demographic of his target subscriber and let it go at that."

[JM offered concrete advice] "Check out Squid (http://squid.nlanr.net), and the "plug in" SquidGuard. It lets you filter on words in the URL."

[DS took up the theme] "We use WebSense (http://www.websense.com) for filtering on NT—there is also a Solaris version. We also use a Squid proxy in front of that, which caches only pages that the filter will allow.

We offer both filtered and unfiltered service; the customer chooses at sign-on. We get a signed Terms of Service contract from all customers before we turn them on, therefore only adults are customers . . . and they knowingly sign up for filtered or unfiltered.We are focused on business customers, most of which choose our filtered system.

—End

Want to get in on the discussion? Join the ISP-Tech list and wade right in.