I am wondering about the spam messages produced by a certain kind of
randomizing process.  This is a follow-on to my question at

http://answers.google.com/answers/threadview?id=292826

which was answered by aceresearcher.  This time out I am not asking
about the word lists or about the use of them in conjunction with
randomizing algorithms; I understand well enough how pseudo-random
number generators work.  What I want to know now is--when I
receive a spam message with a heading like this inside the message
(not in the subject line):

Free CableTV!No more pay!

why does the rest of the message body consist of nothing but a list of
words that looks like this?

gauss dump cyanic corduroy plan btl logic arching affectionate
ciceronian sudden caucasian superior equipoise torrent psychosomatic
superposable condense teeter mailmen horsefly huckleberry mouse guiana
frozen glutamic nag virgo circulant spine shade meal shmuel flatland
lew discriminatory mauricio croix coeditor clove sproul armada balance
erotic everlasting sectoral overhang cardinal coalition yoghurt
tremendous anecdotal childbirth eventide librarian mum shutout
volkswagen dare carroll tuscan bushnell mean secondhand monitory
stillbirth brucellosis headland agave chartres bazaar watanabe
biddable consignee centipede scarlatti kovacs pursuant abscissae
duchess onlook derogate mahayanist involutory confiscatory aeschylus
chance finitary breed soluble arrogant entity calgary cater crosslink
arsenal twitch curiosity dispel handset negligent dreamlike downturn
chirp inman barefaced circulatory crest mesopotamia pub bonanza assai
addison suntanned brazier rooftree abbreviate bufflehead coplanar
ababa mr stagnant conjunct manure debauchery gresham perchlorate pion
scorpio purposive graves cyanide shivery rear buxton fischbein bestir
biaxial broken kosher inappropriate residential presbytery colatitude
penultimate loose provincial cancel alphabetic derisive ellipsoidal
stop goat rest casualty presuppose beriberi rickety stayed chinook
admit worn raman warsaw abram dragonhead priestley beer bryophyta
benny bluestocking cuddle dunlap disgruntle applause wood atonal
besiege police voice excretion secretive twirly homebuilder force sara
ledge schnapps redstone migratory confidante cavendish tagging
gratuity josephus coarse meat resist splat roister sultanate ashore
malicious different cayley affricate aorta niacin bashaw survivor
gubernatorial bearberry alternate kenyon dwight bullfinch cattlemen
beside configuration bladderwort tuscany rheumatism bliss anguish ak
priori artie grammatic martyr morbid descriptor burgess astound
require than mustachio accession gogo and obedient drug quarry cloudy
array conformation perkins caldwell catastrophe metallurgy
coincidental concessionaire canaveral stolen resurgent fortieth
insupportable crater horowitz estella patricia scylla variac
dereference nymph irate chieftain elastic susanne capacitor jackanapes
silicon spruce coercion condominium cowan irreconcilable liaison beta
locution flour chalcocite

What *is* that?  Why is it in the message?  What conceivable purpose
does it serve?  That, in addition to the line about free cable, is all
there is--no advertiser, no contact information, nothing.  Why is
someone bothering to send out messages like this?  I have received a
lot of them.  Is something else going on here besides the appearance
of trying to sell something--which can't be genuine if there is no way
to respond to a product offer?  (Not that I would ever respond to a
product offered through spam--but somebody must, or those guys would
give up--wouldn't they?)  Curiosity is biting me and leaving painful
little itchy spots.

Thank you,
Apteryx

---

Clarification of Question by apteryx-ga on 18 Jan 2004 12:55 PST

aht-ga, I agree with Pinkfreud.  Your answer was dead on target.  I'd
like to invite you to post it as the answer to this question. 
(Thanks, Pink.)

Since then, a number of articles have come my way explaining this
phenomenon, and I feel that I understand it well enough now.  Alas for
the search for truth, which more often results in something
disappointingly prosaic than in some kind of cosmic poetry.

antonh, you also made helpful contributions.  Thank you.  If you'd
take me up on the request to restate your reply, I'd post it as a
separate question for you.  All I need here is a yes or no.

Apteryx

---

Request for Question Clarification by pinkfreud-ga on 18 Jan 2004 14:44 PST

Apteryx,

Since antonh is not a Google Answers Researcher, he would not be able
to answer an additional question if you should post one. To tell
Researchers from non-Researchers, check the way the username is
displayed. Only official Google Answers Researchers' usernames are
hyperlinks, underlined and highlighted in blue. Clicking on the name
of a Researcher will take you to that individual's ratings profile.

There have, of course, been instances in which frequent commenters
later became Researchers. In order to find a person to whom this has
happened, I need only look in the mirror. ;-)

~Pink

---

Clarification of Question by apteryx-ga on 18 Jan 2004 18:19 PST

Thanks, Pink, you're right, and I knew that, but I wasn't paying
attention.  Antonh is a helpful bystander, just as I sometimes try to
be.  (Sometimes our comments don't pass the test, do they?  Oh, well.)
 I hope aht-ga will check back and claim the answer.

Apteryx

While I've never received one that falls in this category, I'd guess
the notion, however inept, is that a jumble of random words will tend
to confuse/bypass spam blocking programs on many servers/mail clients.

More strange, to my mind, is the poor spelling in spam messages, even
in the portions which are not meant to be random letter jumbles.  The
random letter jumbles, one might suppose, serve to help identify (and
credit?) an outgoing mailing service upon the "success" of replies. 
The only purpose I can imagine in the use of poor spelling (and
grammar, etc.) is perhaps to create a psychology of "taking advantage"
of an ignorant seller who purports to have low prices on drugs,
printer cartridges, etc.  While it doesn't appeal to me, perhaps those
who've done this sort of thing for years have established by trial and
error that this is a reliable approach.

Or, maybe they just don't know any better.

regards, mathtalk-ga

There is this piece of software avalable which creates "spam harvest"
lists to annoy spammers called "toxic" which is available I think from
www.cexx.org

It's intended purpose is I think as a bit of fun to annoy the spammers
"robot harvesters" with included embedded email hyperlinks.
It works by generating an .HTM (web page) which the user can upload to
a "Free" web host to sit there for eternity.

It COULD be the "Robot" which is usually sitting on the floor
developed a small problem in the script (Oops) and is emailing the
page words out?

Bear in mid the file below and a quick scan do not match the words
mentioned so I assuming it's a similar program and not the same.

It generates a page almost Exactly as decribed upon just exececuting
it on my pc it generated (cut to a few lines) as a random scan based
in a "dictionary file"

Slough by Mobil starving moment because racist rotten petit merge
multinational because eating misled stewardess. Time inns fad
intellectually for cubes tern the modeled county liberation alley Tim
petit but lessens but measure instructor by rebuilding slowest
guerrillas afloat. Maul, substantially, Nathan underestimate hugged
effects.  Me unreasonable? Strenuous downloaded tactics in. Rocker by
unleashed haunted Donald in trophy highly in factor immunity curved
draughts or deepest polio? Strategic with ideals slowest low-cost
pains outlying starving cymbals stingy defect interstate Turner styles
inevitably repeated subject! Eating minivans or unattractive Beverly,
reconcile irresponsible applying insulated relive immense mousse
likes, ass tile, and incumbents an, dizzying lied broadcasts sowing
represent. Drawing worthwhile experiments humiliating validate and
dominate pair and read for dispose because Utah



Hope it clears well something or did it  just confuse even more?.. Anton.

The weirest thing, I opened the .DIC file and was able to replace it
with the "http://tinyurl.com/33c8m" contents and use it to generate 
files as per your  question using the keywords supplied.. Good luck!

Antonh, would you please say that again?  I don't quite understand
what you did.  What .DIC file?  The TinyURL you gave is the one that
was in my prior question, yes--but what was it that you did with it,
and what result did you get?  Please use a few more words to explain
because I didn't read everything that was between the lines.

I like the idea of annoying the spammers.  Junk mail was never like
this.  If there's a way to fight back, even if only by gumming the
works, I think people should spread the word.

Mathtalk, I don't know, I'm only guessing, but I find that nearly 100%
of these messages come from places abroad, some Asian and some
European, and I just assumed that the poor English was the work of
non-native English speakers.

I received a couple of messages that have this:

%RND_UC_CHAR[2-8]

where the others have the jumbles (such as this:  ZSYYXYQ and this: 
ROTM) and figure someone messed up with a delimiter, so I'm seeing the
code instead of the output.  But even the ones that seem to be "all
right" (inferring a standard from the majority) and actually have an
ad graphic in them also have a bale of words like the cited example at
the bottom.

Thanks for all comments--
Apteryx

>> Curiosity is biting me and leaving painful little itchy spots.

This uncomfortable condition is sometimes caused by rapid intellectual
expansion, which can cause mental stretch marks.

For the itchy areas, I recommend the use of callow-mind lotion. ;-)

Ah.  Apply the pink stuff.  I'll try that.

Apteryx

Hi apteryx!

Here's a short thread that might shed some more light on this for
you... or serve to confuse the issue more...

http://mailman.anu.edu.au/pipermail/link/2003-December/053930.html

Basically, someone in the spam software industry came up with the
novel idea that 'padding' an otherwise spam message with "valid",
non-spam-like words, would help the spam get past the Bayesian filters
used by systems such as Spam Assassin. Only problem is, the
self-learning capabilities of the filters make this a very hit and
miss approach. The filters will learn to either ignore the extra text,
or score that text poorly as well. Coupled with the machine-gun
e-mailing technique of sending millions of copies of any one spam
message, though, and you'll still get enough copies through to make it
worthwhile.

Sigh... the spam wars continue...

aht-ga

apteryx-ga:

How timely!

Wired News: Random Acts of Spamness
http://www.wired.com/news/infostructure/0,1377,61886,00.html/wn_ascii

aht-ga
Google Answers Researcher

I get quite a few Bayes-buster spammages. They usually don't fool my
anti-spam program. But sometimes the long strings of unrelated words
are rather amusing, like some kind of surrealist poetry.

Apteryx,

I heartily suggest that the prize should go to my colleague aht-ga
(who, IMHO, is one of the best Researchers on GA.)

I have the January issue of "Wired," and I hadn't even noticed the
"Random Acts of Spamness" article. A tip of the pink hat to aht (hey,
I just noticed that hat and aht are anagrams of each other. Neat-O.)

~Pink

apteryx-ga:

I'm sorry that I never did come back to this thread in time to see
your offer that I post an Answer. I'm happy that my comments were on
the mark this time, and will continue to provide input whenever I can
to your always intriguing questions!

Best wishes,

aht-ga
Google Answers Researchers