Ann Arbor Business to Business
Small Business and the Internet

App Killer: Content Filtering

August 2002

By Mike Gould

There is a GeekSpeak phrase called "Killer App". (From 70's teenspeak: "Killer - like, totally righteous, bro", and "App" short for application, from 80's techtalk: "Why use more syllables than ya need to?"). This is an application (or program, as we used to call them) that jumpstarts a new technology, defines a genre, or starts an industry. Examples of this include Aldus (now Adobe) Pagemaker, which started the desktop publishing industry, Mosaic, which was the first WWW browser, and, arguably, iMovie, which is inspiring countless camcorder-wielders to make and edit their own movies in digital format.

There now seems to be some apps which do the opposite: kill the industry that spawned them. I'm talking about spam and its nemesis, server-side content filters. Folks, we are in for a bumpy ride.

Awash in Waste
As anyone with functioning email knows, spam has reached waist height on the body politic, and threatens to inundate us completely in a year or two. Junk email is cheap to ship, problematic to police, and difficult to filter at an individual level. For more on this, see my previous articles online at:
("Filtered Email")
("Spam Redux")

This flood of digital dreck seems destined to make the Internet a victim of its own success. Spam clogs the arteries of the networks, eats up user time in dealing with it, and causes valid communication to be lost in the jumble of junk.

One response from network administrators has been to filter email at a server level, where email lands in your mailbox before you download it. Here's where things get interesting.

Talking in Public
Remember, email is about as private as postcards. Anything you write can be scanned for content by your ISP, the government, or that strange person next door with all the antennas on his roof who shares your local Comcast loop. If security is important to you in regards to business email, you need to invest in some encryption software and be prepared to exert a lot more effort in your online communication.

In the above referenced article, I described how I deal with spam: I filter it after it hits my mailbox, using Eudora to sift the good stuff from the bad stuff. This is called client-side filtering, as I am a client to the server serving my email. (A server is a big computer somewhere that does something for you: stores files, performs back-ups, or in this case, manages email). Client-side filtering works by looking at the sender and subject of each email, and deciding into which local (on my computer) mailbox to file the message. Most of it gets filed in the trash, where I examine it before discarding it, in case a filter misinterprets something.

Filters in Action
Server-side filtering works like this: the part of the email server handling incoming messages looks at the sender and subject, but also examines the actual content of each message, looking for Bad Words. The administrator (admin) for this server has written a set of rules for this filtering, much the same as I have when I set up my client-side system. The problem is that you have no control over what the admin decides is bad, and may in fact have no way of knowing if such a system is even in place. And if a message has a Bad Word in it and triggers a filter, the message will be trashed and you will receive no notification that this has occurred.

Keep in mind all the filtering is being done by a computer; no one is actually reading your email using these systems. But computers are still very, very, dumb, and only as skillful as their admins in separating the good from the bad. If you are a member of a breast cancer recovery email group, say, and your admin decides that "breast" is a Bad Word, you have a problem.

This is a new phenomenon, and is just starting to be implemented by some email systems, usually at big corporations and major ISPs. Some of the first victims of this are eZines, online publications that send out thousands of copies of newsletters to subscribers. As they are sending multiple copies of the same content, their email behavior is identical to spammers in the eyes of email admins, except that they are sending content to people who have requested it. I subscribe to an eZine called TidBits, and a recent article there:

described how hundreds of email systems have rejected their postings because they contained one of a long list of fairly common words.

Serving Insidious
If people start complaining to you about messages they sent but you never received, you may be the victim of this sort of thing. Ask your ISP if they are doing content filtering, and if so, what their criteria is for Bad Words. As far as I know, this has not been an issue here locally, but it may only be a matter of time before this starts to affect everyone.

My take would be to have email filtered server-side based on sender and subject, the same as I have on my local Eudora system. If every email that contained the word "Viagra" in its subject was trashed, the world would be a much better place. There are companies that maintain lists of known spammers, and filter out email from them before subscribers see it. One such company is Brightmail:

They are kept very busy by the fact that spammers are constantly setting up new email addresses from which to do their dirty deeds.

On the one hand, it is good that email admins are starting to take action against the flood of spam. On the other hand, content filtering starts to look a lot like poorly-implemented censorship. On the third hand, the war goes on, signal vs. noise, with no end in sight.

(I just got around to updating my website, and noticed that I passed the 50th article mark a coupla months ago. A milestone of sorts. Thanks for sticking with me these last 4 years and I hope to see you back for the next 50 articles...Mike)

MonodoDyne <M> The Sound of One Hand Clicking...
734 904 0659
Entire Site © 2018, Mike Gould - All Rights Reserved