One of the minor inconveniences of managing a blog is comment spam management. If you don’t quarantine incoming comments and review them before making them “live”, you blog will very quickly disappear under a tsunami of bizarre unrelated pitches for unsavory products. It’s mostly an annoying chore. But there is one aspect of this that I find fascinating—fun even.
Behind the scenes the would-be spammers and the white hat enemies of spam are waging a technological and linguistic war for dominance. The good guys use what are called Bayesian Filters to distinguish spam from email of value. This approach goes beyond the simpler approach of comparing email to known lists of “spam” words (such as “Viagra”, or even “enhancement”) and rejecting messages that use those words. Instead it analyzes the linguistic characteristics of a body (corpus, if you want to be all fancy and Latin about it) of messages that you indicate are real email and another that you identify as known spam. It then learns to distinguish the two based on the characteristics of your unique email. If, for example, you are a scientist that is studying Viagra your Bayesian filter will need to work overtime to distinguish the good from the bad, but if you train it well that should be possible.
Anyway, here is where the fun begins for someone who enjoys the outer fringes of language and meaning. Spammers know all about Bayesian filters and they also know that the best shot they have at defeating them is to create random, machine-generated language that looks very much like real communication. One of the last courses I took when finishing up my BA in 1995 was Tan Lin’s Language Poetry course at UVa. This type of writing downplays what we usually think of as the essential desire of a poet to express him- or herself and instead focuses on language as language, the mechanisms of constructing it, and the reader’s participation in bringing meaning out of it.
Over the years I’ve noticed that spammer’s attempts to defeat Bayesian filters occasionally sidle right up to language poetry. A particularly enjoyable example showed up this morning and I decided to share it with you …
Come about to be crocheted vests.
These cozy vest’s bodily fluid
so that you preserve your chocolate
if you necessity to draw near
an achievement bill of fare,
you can buy or be jewelry.
Pauperization accepts minor moment single.
So, ratify up for your way prize.
a mate of faddy—your odor
with component embellishments.
If you would be someone salaried
a wash soprano for an endeavor
somebody with the boundary friendship.
This is easier to effort this off;
nonetheless, you can lay aside
your approximately monetary system.
Lone example coupons
about that region of skillfulness
and what it does.
I have to wonder what spammers would think if they knew they were entertaining me in this way. I mean, really, you can ponder a thought like “you can buy or be jewelry” for hours!