Protecting against knowledge

By Scott Bradner
Network World, 10/25/99

If you have a sensitive nature, you should not read this column. According to the Internet gateway manager at a large international company, this column contains a word that is "abusive."

I'm managing an IETF mailing list that's looking into what the IETF should do if we are faced with a request to add legal intercept (wiretapping) functions to IETF protocols. As the list's manager, I get sent copies of any messages concerning delivery problems the mailing list finds.

The other day I received the following: "This message was rejected and nondelivered by our Internet Mail Gateway [this scans all incoming and outgoing Internet messages]. The message was rejected because of abusive or offensive content. Please reword the message and resend it."

I exchanged mail with the gateway's manager, and after searching the mailing list archives I determined that the rejection was because of this paragraph from someone who did not like the idea that the IETF might develop protocols with wiretapping functions:

"If the Federales [sic] want to develop such an application, let them do it. I'm certainly not about to commit time and money developing an application and/or hardware interface that benefits THEM! Isn't that why they take 40% of my paycheck each week, so that they have money to piss away on stupid stuff like that?"

The gateway manager said: "We do a general check for abusive words, not a context search. If any abusive words are found, the e-mail is stopped either going out or incoming. Our users accept this limitation and are happy to remain within [sic] it." Due to the outbound filter, the gateway manager had to type the offending word on four lines with one character each.

Because the IETF mailing list software automatically unsubscribes any entry that causes a bounce, the subscriber at this company is no longer receiving postings from the list. The company policy has protected him from what the company sees as abuse - as well as from the important discussion taking place on the mailing list.

I can see why a company might want to filter language in outgoing mail that in one way or another might harm the company or its image. Simple word scans will not catch misstatements of fact that are the most likely to harm the company, but flagging deeply offensive language is an understandable protection. But extending this to blocking incoming mail containing a word that some 5-year-olds use comfortably is more depressing than anything else.

This company demonstrated a mistrust of its employees that is impressive indeed. The image of happy employees, protected at work from the evils of the world by a paternalistic management that treats them like children is a very sad one to me.

Disclaimer: Harvard's mission is to get people to think, so treating them like children would be counterproductive. But the above lament is mine.