One man’s spam is another’s delicate repast?

Last week, David posted, saying "One man's spam is another's delicate repast." Steve Kille from Isode commented that this might only be true very rarely.

Steve's right, in a literal sense. However, the real "spam vs. delicacy" issue is in the grey area--where the newsletters and "legitimate" direct marketing (DM) live. Far from being "very, very rare," for many users this remains a significant issue. (We concede that Steve probably isn't one of those users — he's far too careful of whom he gives his email address to. I'm not one, either.)

But for the majority of users, the situation is more complex.

There are two main vectors for grey-area spam:

  1. Users receive mail that they at one time requested, but have forgotten they asked for. Is this spam?
  2. Users receive mail sent to a purchased
    opt-in list. Their email address appears on that list quite legitimately,
    because they permitted someone to resell their address (what I call the
    "Faustian bargain checkbox"). Is this spam?

The problem comes when server-based spam filters start being told by some users that this sort of email is spam. A user reports a false negative ("spam" that arrives in the inbox), and the report is fed back to the filter, which learns from its "mistake." The simpler of those filters will begin to filter such email for all users. That's when we get false positives.

What's needed are spam filters that learn from users, but apply that learning per-user, not per-organization. A notable example of such a filter is Cloudmark's.

One Comment

  1. Posted February 7, 2005 at 11:08 PM | Permalink


    I’ll stand by my comment! There is some quite agressive opt-in marketing, but in my experience this is still quite separate from the very high volume spam. I;m sure its true that some people can’t tell the difference.

    You make the interesting point that software finds it hard to separate “ham” from “spam” (even though a human can do it quite easily).

    Separating one person’s ham from spam is easier. The commercial solution you noted, and somme free solutions have been able to do this pretty well. The difficulty is that most people cannot be bothered to train a spam engine. Nor is it reasonable that individuals should need to.

    People need solutions to “just get rid of the stuff”. Automated solutions can do pretty well, and for most people I think that this is the answer.


Post a comment

You must be logged in to post a comment. To comment, first join our community.