[Spambayes] my pet theory on spam and language got proven today
Seth Goodman
sethg at goodmanassociates.com
Sun Oct 29 20:53:04 CET 2006
Amedee Van Gasse wrote on Friday, October 27, 2006 6:08 AM -0500:
> This only means spammers are getting "smarter" - or more likely: all
> kinds of spam filtering serve as darwinian pressure on the evolution
> of spam.
A very apt comparison. The behavior of this system we have all been
monitoring has much in common with biological systems. Both spam and
filters are adaptive, self-organizing systems. They both are resource
constrained, so the "smartest" solutions tend to win. The dynamics of
the process are also typical of those in biological systems.
I suggest that the most apt analogies are to pathogens and parasites,
and not just because they are nasty! These organisms have very simple
genomes, so there are severe constraints on the possible solutions.
They make up for this with a short generation time and an unusually
"labile" genome that produces large numbers of new combinations to
test. This is in direct contrast to complex organisms that could be
easily wiped out by an overly labile genome.
When a new genetic combination appears that provides a selection
advantage in a particular niche, its population surges. By decimating
their target population and providing strong selection pressure for
those hosts (spam filters) with even marginal resistance, pathogenic
pandemics tend to be self-limiting. This is likely what will occur
with the current pandemic of image spam. Either filters will catch
up, or MUA's will block remote images entirely, and eventually images
embedded in the message.
In the long run, blocking images is more sensible than trying to decode
them. We already have captcha's to make this difficult, and there is
little need to fight this battle in the first place. The justification
for images in messages is largely for the benefit of advertisers. I
hate to admit that Mr. Bill was right when he proposed, over a decade
ago, that the commercial internet will be financed by advertising.
While correct for the time being, this doesn't mean that email
messages themselves must carry it.
--
Seth Goodman
More information about the SpamBayes
mailing list