WHY DID THE COW WANT A DIVORCE ANSWER KEY

[Draimeionitte]

We delete as much as a couple of hundred spam posts a day around here, but once in a while the topic title is so bizarre that I have to let one go (after deleting the message itself). Since I'm home sick, today, this is another one.

What the hell goes on with those Russian spam bots that makes them create such odd titles?

Tom Allen

[celticqueens_sub]

Hope you feel better soonest.

[Kiki]

Is this a super sekrit Tom Allen account that just blew its cover? Or, a clever attempt to get us to click the (rather suspicious) link in the post's footer? Because, this is not the usual "Tom Allen" account. Do I smell a rat?!

[Tom Allen]

Is this a super sekrit Tom Allen account that just blew its cover? Or, a clever attempt to get us to click the (rather suspicious) link in the post's footer? Because, this is not the usual "Tom Allen" account. Do I smell a rat?!

Naw, it's me. But don't click the link - I have no idea what it is.

As I said - sometimes the spam titles are so ridiculous that I just delete the message and post them for entertainment value.

"The Original"

[Whizbang]

Arise from the dead, zombie post!

Waiting for chastity posts, I got desperate and I was looking through archives and, lo and behold, there's an off-topic post I could actually weigh in on.

Imagine you are a spammer. You want to sell an item and issue ten million emails "BUY SUPERWIDGET X NOW!".

This used to work back in the day. The thing is, people hate spam, especially system administrators. So pretty quickly, computer programs were written that would look for lots of duplicate messages that were sent into a given administrative domain and then they'd automatically squelch them. That was initially pretty effective.

Of course, the spammers didn't like that much. So, they responded by putting small variations into the spam messages. But then, in response, the spam blockers would look at messages and, if the differences were only very small, they'd also squelch the messages.

So spammers took a different approach. The trick with spam blockage is that you want to block the spam but not block legitimate messages. It's very, very hard for a computer to really understand if the text of a message has real meaning. Computers simply don't understand whether a string of alphabetic symbols really makes human sense.

So enter a spammer's secret weapon, Markov chaining. Markov chaining is a computerized method of generating sentences that *sort of* make human-like sentences.

The way that it works is that computers sample lots and lots of online text (hello Google!). Then they perform a bunch of statistic analysis on pairs of words. For instance, they might find that after the word "I" that it's really frequent that the next word is "am". They'll collect the probability that, given the word "I" that the next word was "am" (and "was" and "went" and "had") and then they'll get this giant table.

From there, they can generate a bunch of "human-looking" language. Say they start with the word "I". For the second word, they'll roll some dice and choose from the table whatever word comes up, say, "am". Then, given the word "am", they'll roll the dice and pick the word that the table says follows "am" for that word... and so on.

This generates "English looking" but non-sensical emails that, somewhere within, have the link that the spammers want you to click (to sell you something or, worse, to hack you). Generate a new one of these every 100 emails or so and then computerized spam filters can't tell if it is a real or fake message!

That's it in a nutshell.