Spam Catchers Block Latest Crypto-Gram

An anonymous reader writes "Bruce Schneier sent out a note about SpamAssassin and possibly other spam filters blocking his excellent Crypto-Gram newsletter. Fortunately you can get it here (early no less!)." Schneier's email reads, in part "Tomorrow I will be sending out the February CRYPTO-GRAM, as I do on the 15th of every month. In the process of creating this month's Crypto-Gram, I discovered that SpamAssassin thinks that this issue is spam, probably because of certain links and descriptions of scams in the text. I have anecdotal evidence that other spam filters block Crypto-Gram as well. ... I'd apologize for the inconvenience, but I'm not sure what I could do to make it less so -- I don't intend to alter my content to accommodate spam filters."

Comments

White List

[SealBeater]

That's easy to fix, add the crytogram address to a whitelist. Every spam
filtering software I've ever run, including spamassasin (which I like a great
deal) has a whitelist option. If you're running some kind of filtering
software, it behooves you to keep an eye on what it's blocking, hence, I am
sure that people are aware of it and have adjusted their software accordingly.

SealBeater

The problem with filters

[markfletcher]

This illustrates one of the big problems with filters. They will never be perfect, spammers are always adjusting to them (even the Bayesian ones), and the way many are implemented, they make email unreliable (by deleting suspected spam messages and not bouncing them). Blocking untrusted servers by IP address avoids these issues.

obPlug: This is why I created Trustic.

The problem with content filtering

[Leeji]

This is exactly the problem with most content filtering approaches.

It is very hard to discern the difference between talk about sex, spam, viruses, etc and talk from sex, spam, viruses, etc. Newsletter authors go as far as writing "v*rus" and "sl*mmer" so that pitiful content filtering blocks don't trash them.

It gets even worse for email lists that use inline text ads. The ads alone would constitute spam, but they're nestled within several paragraphs of high-quality discussion.

The problem is that content filtering approaches usually only analyze the "spamminess" of a piece. They usually don't analyze the "goodness" of a piece. So if I put "hot teens go crazy for debt-free viagra while earning $$$ from home" in the middle of some fine Shakespeare, that will get flagged as spam.

The new "bayesian" approaches are finally dealing with this problem -- something can look an awful lot like spam, but it will be saved if it looks even more like legitimate email.

In this case, spam doesn't generally run for 21 pages with words like "cryptography," and "full disclosure."

Re:This is a non-issue....

[Elwood P Dowd]

Thank you. Also, if all the bayesian filtering advocates are right, then the users should be able to mark the Cryptogram as non-spam, and the filter should adapt. More to your point, though, is that lack of spam-filtering software can cause false-positives in your own personal, analog, spam filtering algorithm. Many of my users have deleted important, non-spam, automated emails manually because they thought it was spam. Sometimes, the machine might have less false positives than they would.

Huh. It occurs to me that it seems like some spam filters might pass a turing test if the only output is their spam judgment. Wow. The future is now, dude.