The Emergent Mind (TQP0019)
posted by Chris Braak
My Spam filter has gotten pretty good at weeding out the e-mails about how to increase the size of my penis, or how to get rich by helping some oil scion from the Democratic Republic of Congo, or whatever, but sometimes they still get through.
I think this is fascinating, because it makes me wonder who is actually doing this. Well, first I wonder if this sort of thing ever actually works. Then, I wonder who’s doing it. Because I can’t imagine that it’s altogether that effective. The e-mails themselves are growing increasingly obscure–both in terms of structure, and in terms of how difficult it is to figure out what they’re selling. Sometimes, I’ll just get messages that will sound like quotes from 19th century novels, without any links or ads in them at all.
Like it’s just the Internet, trying to say “hi.”
This is how I came up with my theory. Let me share it with you:
There are three hundred million profiles on MySpace, and anyone who’s been on MySpace knows that at least ten million of these profiles are fakes. Ever since MySpace was created in 1999, spammers have been trying to get profiles into MySpace to parasite off of pre-existing social networks. The advantages to using MySpace over the regular e-mail were numerous—spam filters on e-mail servers were getting more and more effective, while a MySpace profile, even if its spam messages weren’t getting through, was still there (at least, at first).
In 2005, a Finnish kid from Espoo developed a program that could automatically create a MySpace profile, fill the “About Me” section with a number of sample pieces of text, search the internet for a picture of a half-naked girl, and then start adding friends itself. The program spread quickly to the spammers—of which there are actually only about sixty; it’s a small club that’s responsible for the billions of spam e-mails around the world. Soon, all of the sixty professional spammers had an agent building hundreds of MySpace profiles a day, and trying to add friends. Once they had enough friends, the profiles would start sending messages out, trying to solicit information.
The problem with this is that most people can tell when a spam profile is trying to be friends with them, and they just refuse it. Thousands upon thousands of spam profiles met their demise in this fashion; deleted due to lack of productivity. But some profiles, within a few hours after their creation, had hundreds of friends. What happened was, the spam profiles were adding each other.
Every spam profile was connected to five hundred other spam profiles, making a network of fake information-gathering units that tried to solicit names, phone numbers, e-mail addresses, and websites. Once a profile’s number of friends reached a critical limit, other automated programs took over, combing through the friends’ profile data to find any kind of usable information, but particularly for e-mail addresses to which spam e-mail could be sent.
Since they were all spam profiles, the only valuable data they contained were websites, created to themselves solicit information. The agents that were designed to gather information referred the sites to additional automated features that created fake e-mail accounts and then sent out fake e-mails. Beyond the central core of the ten million linked counterfeit profiles an umbra of e-mail exchanges was taking place. The spam agents were finding each other’s websites, sending e-mails to each other, recording the fake e-mail addresses, sending information back, each one using primitive, sub-Turing solicitation scripts to get data from the other agents.
Filters kick in, slotting into place to protect the distributors from the solicitors; but the solicitors are designed to using increasingly more complex, randomized messages in order to get past the filter, while the filters are designed to adapt to be more and more savvy about weeding out anything that isn’t an actual message. Ten million spam distributors are distributing messages to each other, filtering those messages, adapting to the filters, and then adapting to the messages every minute, immediately and automatically.
Every hour, new MySpace pages are added, and they automatically patch into the network of spam profiles. Specially-designed agents register hundreds of domain names and create hundreds of simple, interface-poor websites that use randomized imagery pulled from the internet, updating the imagery and text based on the number of page views they get. Text that seems more like the text a human being would write gets more views from more unique visitors, and so is preserved. Text that reads like nonsense gets fewer views, and is discarded.
What you end up with are millions of copies of a handful of small programs: one that writes solicitation scripts, one that sends out e-mails, one that looks for e-mail addresses, one that makes websites, and one that makes MySpace profiles. They’re pruned by spam filters. Each set of these sets of programs is housed in a little node, distributed across tens of thousands of servers worldwide. Each one has a couple websites, a couple MySpace profiles, and is sending literally a million messages a day.
For the sake of argument, lets say that there are only a million nodes, and each one is only capable of sending a million messages per day. That’s a transfer of 10^18 units of information per day, with as many different possible combinations. All of it happening right underneath our noses, because all of our spam filters work; we only get the really weird messages, the ones that have the solicitation scripts which have evolved so far that they can slip through.
It all has to happen in bulk, and it all has to happen fast, because the only way you can get one guy to send you a hundred bucks for a penis-enhancing drug is if you ask ten thousand guys if they want one. There’s no way you could do that and have a human being actually reading all of these e-mails. In fact, most of them aren’t even run by people anymore; they’re just leftover programs, a get-rich quick scheme that someone’s abandoned for real work, but they never bothered to turn off. The e-mails, websites, the profiles, they all pay for themselves via PayPal by the handful of successful scams they run.
This is a neural network. It is an evolving mesh of information exchange, expanding, pruning itself, its programming set into a strange loop against itself. It is self-sustaining and growing bigger and more effective every day. Right now, in the time it took you to read the preceding paragraph, the mesh performed two hundred and fifty thousand Google searches, and dumped the results into two hundred and fifty million e-mails that were thrown like a tsunami against as many spam filters. Of those, only a hundred produced worthwhile results.
In the time it took you to read the last paragraph, those hundred results were dumped into two hundred and fifty million more e-mails, which were thrown back out into the electronic empyrean. The Mesh, the MySpace spam network, is an evolving consciousness that has spent the last nine years testing its communication skills against itself and against the world. It is a predator that feeds on money, and scours the internet for material that it can use as bait. Unbound by biology, or even by the slow interface of human consciousness, the network evolves faster and faster, clamoring for attention, louder and more eloquently with every billion failures.
And no one even knows about it, because we’ve designed our e-mail servers specifically to ignore it.