My friend Liz forwarded an email to me awhile ago, in response to my SPAM blogging, saying "You get free verse, I get magnetic poetry." The missive in question was from "Bree Harris" and featured the subject line "hotter." It was unlike the SPAM poesy I was getting, and I must admit, I felt left out.
Good news, however. I am now apparently one of the cool kids, since I received this:
Date: Wed, 6 Aug 2003 03:41:19 GMT
From: "Linda Gasser"
X-Spam-Level: Spam-Level SSSS
poverty exquisiteness exchequers accuracy humbled scallops brant polishes porcupine pools admixture bethlehem plumbago theaters matinee searing hospitality adagio even blushes mention tanners bolivia talismanic scribbler ampex mescaline pleases acreage tapper $RANDO MIZE anabel territories scantier tad posse hurling scrim polisher country sandpile microcodes tangential scoffing cot addressers tangled exemption maturate executors boulders bombs excommunicating pollen corroborate scholastics targeting sating auckland accordance blurting $RANDOM IZE posse exhaustion exploration crinkle courthouse bordered poured acidulous experiences ethology hurray bonnet activate illustrates memorial adjudications etc exclaims cringed alaska sarcasms
actinolite expressively humblest tags bonnets potentiometers technologies hunches boniface scotland excommunicate exemption illustrates hunks scorning anglicanism hydroxide blythe crowder ploy mealtime postlude hymns husks hostile exponentiation brass beltsville ethology idyll $RANDOM IZE meteorology technicality crabmeat county hydrostatic crawling atlanta experienced taft idealization euphemisms illustratively counting covertly scripture advance tallest braces tectonic hurl adamson accreditation cow hypothalamus mightier portray hopscotch adores breakoff adulthood $RANDOMI ZE bolshevik poultry cozen measuring brazilian sauces cowslips taunter that plungers everlasting acculturates RANDOMIZE coweringly annette household microscopes brat playhouse breadfruit accolade
As for the reason behind the waterfall of nonsense, well, according to Liz (or rather, a friend of hers), a type of Spam heuristic known as a Bayesian classifier is used to differentiate between Spam and Not Spam based both partly on the proportion of text to HTML, as suggested by this notice from SPAM assassin:
* 3.0 -- BODY: HTML has images with 0-200 bytes of words
and partly on the actual words themselves. Or, as Liz put it, "the spam I forwarded you was attempting really hard to look like Not-spam by having lots of words commonly found in Not-spam."
I went looking for more about Bayesian and found this:
Given training, a spam heuristics engine can take the most "spammy" and "hammy" words and apply probabilistic analysis. Furthermore, once given a basis for the analysis, the engine can continue to learn iteratively by applying both it's non-Bayesian and Bayesian ruleset together to create evolving "intelligence."
So there you have it. Also, while I was snooping around, I found some wonderful, wry commentary at the heuristics portion of SPAM assassin that I urge you to investigate. Favorites include:
How dear can you be if you don't know my name? DEAR_FRIEND
Gives a lame excuse about why you were sent this spam EXCUSE_1
A dodgy mortgage testimonial HELPED_FINANCE
Apparently, you'll be amazed BE_AMAZED
Risk free. Suuurreeee.... RISK_FREE
Talks about Oprah with an exclamation! BANG_OPRAH
And so, with all that info mind, I give you:
Date: Mon, 18 Aug 2003 23:44:33 GMT
From: "Eloisa Allen"
exploding exigent scrawl appleby creeper expositions terrors crabmeat bowlines membrane hydroxide mellowed exaltation accepts expedient scheduler becker tailgate scoffs plebiscite medieval potentiality alden booky breakaway criminals braes alcestis houses tetanus $RANDO MIZE plotters plots boston acquiesce plop hustling terraced experimenter evicted menhaden bombarding bread terminate crescent portraiture annette polygynous methionine administrations craving maximally memoirs poorest maturing bothering belgium tetrahedra advisees antoine boastful $RANDOM IZE branches terminates telephony scrambling popularizes screamed exclaims adsorb tamarind experimenters tam adumbrates brandywine cowslips scraper poisoning executional coward atlantic bolometer testaments
hypothesis creditable polite humaneness bates acknowledge tempter bottleneck tasting countersink their bravo housetop tapering maximal hundredfold medley adhered adages tempering imbibe potable excerpted explorations medicinal explicit cradled activations postcondition teaming $RANDOM IZE potentates acrobatics adopted bottlers megaton taffy immaterial adriatic adjusting sawtooth crock exposing idiots croak cretin talked metamathematical acknowledging identifiers excerpt ansi acknowledged boomtown bones meriting matting scarface bodybuilders tautly cower $RANDOMI ZE saps sears mellowness scans scribe accomplisher teachable admittance explore bonnets arthur tektite RANDOMIZE botching imitates barnes cotillion mentions hugely crate scale