[thelist] Re: random text generator spam?

Kevin Martin evolt at brasscannon.net
Mon Oct 25 07:10:49 CDT 2004


Quoth "ANDREA STREIGHT",
> [text] expected fell out nature blew outside cheat car, seems object wind 
> news wife, very familiar recognize patient should fathers office, there 
> pride hurry sir air hot less worth [etc., I think you get the idea]
> At any rate, is this a spammer using a Random Text Generator just to fill in 
> the message area?

[Snip tale of drunken attempts to generate simlar text in one's 
youth.]  Basically a "Markov chain" text generator.

> Could there be "hidden malicious code" in such text? Er, this is plain text, 
> I turned off the "read HTML" in my email client.
> 
> Any comments? Is it safe to preview panel such crap?

It's primarily targeting Bayes-based spam detectors, ones that look
for patterns of spammy vs non-spammy words and save the recurring
patterns to learn over time.

The spammer's hope is that:

a) enough non-spam content will put the meesage-as-a-whole
below the spam threshold and let it through; and/or

b) if you tag this as spam, your Bayes "spam words" database
will eventually be flooded with non-spam words and become
useless.

As already observed, the payload is not in the random words,
but what is attached to it is neither more nor less dangerous
than anything else.  (Loosely interpreted:  Your mail program
HAS a preview panel?!?  Uh-oh....)


More information about the thelist mailing list