Spambot spams about spambots
A while ago, I posted about "spambots" -- little robots that troll around the Net looking for blogs, then leaving advertising messages in the comment fields. As public rancor about this problem has grown, the bots themselves have become oddly solicitous. For example, when I recently wrote a blog entry entitled "Automatic Butt Kicker," a bot wrote the following comment:
Automatic butt kicker?! That's classic! :)
Posted by: Online Pharmacy on October 31, 2003 01:05 PM
You can see the algorithm at work here. The bot's been programmed to take the title to the blog entry -- "Automatic Butt Kicker" -- and simply regurgitate it as a statement, in classic Eliza fashion. The ad, such as it is, is very discreet: It's just the URL for "Online Pharmacy." The spambot even renders the echoed phrase in lower case, to make it look more realistic yet.
All of which might make you wonder: Why are spammers going to such incredible lengths? Because bloggers are starting to wise up and mass delete all these spambot postings. They can recognize the obvious pitchmanship from a mile away. Thus, the more human-like the spambot appears to be -- the more its written comments seem to be that of a genuinely real Collision Detection visitor -- the more likely I am to simply not notice that its comments are ads, and to leave them in place.
Consider just how Darwinian this is. We are witnessing, in essence, a multistage evolutionary fight. In stage 1), the spambots come online, filling blogs with product-shilling so brazen that we bloggers immediately recognize the bots are, well, bots. In stage 2), the spammers realize they have to pass the Turing Test -- they have to create better artificial intelligence so that the bloggers won't even notice that bots are posting to their boards. The hostile environment online is forcing spambots to evolve better and more lifelike A.I. At this rate, in six months they'll grow opposable thumbs and kill us all.
It gets even weirder. Two weeks ago, I wrote about spambots for the first time, in a posting entitled "Spambots: The new scourge of blogs". In the discussion area, several Collision Detection readers posted intelligent comments about different ways to block spambots, which was very cool. But then today I discovered a new comment:
I agree..
Posted by: Canadian Pharmacy on November 2, 2003 11:46 AM
It was my old friend, the Canadian Pharmacy spambot! The situation couldn't be more dementedly self-referential: A spambot had posted its earnest agreement to a posting that was itself about how to get rid of spambots.
If that isn't the nine billionth name of God, I don't know what is.
Posted by Clive Thompson at November 02, 2003 07:18 PM
Trackback Pings
TrackBack URL for this entry: http://www.collisiondetection.net/mt3/mt-tb.cgi/639
Listed below are links to weblogs that reference Spambot spams about spambots:
I got one once that was actively insulting, "this is an awful idea" or something of the sort. Not cool, but at the same time surprisingly effective, held my attention more then any other spam. The psychological tactics of spammers are getting quite interesting I must say.
Posted by: Abe at November 2, 2003 8:27 PM
The depressing thing here is how easy it is for a mindless spambot to approximate what passes for "discussion" these days.
If a spambot that simply regurgitates basic, common phrases into comments is indistinguishable from a real human, surely real humans that simply regurgitate basic, common phrases into comments are indistinguishable from mindless spambots!
Posted by: marc at November 3, 2003 8:49 AM
You're right on the money Marc! I can think of at least 100 co-workers who would serve to support your postulation.
Posted by: Robert at November 3, 2003 10:46 AM
Heh. Yes, Marc, you've hit upon the epiphany that led Richard Wallace to program his quite-cool chatbot ALICE. Most A.I. people have assumed that to model human conversation, they needed to make the bot able to reason, to have basic memory, and general commonsense knowledge about the world. But Wallace realized that the vast majority of humans don't have those attributes. Indeed, the vast majority of humans spend almost all their time just wasting oxygen every time they open their mouths. Thus, a bot that automatically says "your just a hata" and "why you gotta be that way??" as a response to, well, almost any utterance, not matter how illogicaly connected it might be, would come off as more real than a bot that actually attempted to engage with your conversation.
Posted by: Clive at November 3, 2003 12:08 PM
Posted by: Allan at November 29, 2003 6:46 AM
Posted by: Online Casino at January 16, 2004 4:27 PM
Posted by: julia at January 24, 2004 7:06 PM
Posted by: order cialis at February 22, 2005 3:00 PM
Post a comment
I got one once that was actively insulting, "this is an awful idea" or something of the sort. Not cool, but at the same time surprisingly effective, held my attention more then any other spam. The psychological tactics of spammers are getting quite interesting I must say.
Posted by: Abe at November 2, 2003 8:27 PM
The depressing thing here is how easy it is for a mindless spambot to approximate what passes for "discussion" these days.
If a spambot that simply regurgitates basic, common phrases into comments is indistinguishable from a real human, surely real humans that simply regurgitate basic, common phrases into comments are indistinguishable from mindless spambots!
Posted by: marc at November 3, 2003 8:49 AM
You're right on the money Marc! I can think of at least 100 co-workers who would serve to support your postulation.
Posted by: Robert at November 3, 2003 10:46 AM
Heh. Yes, Marc, you've hit upon the epiphany that led Richard Wallace to program his quite-cool chatbot ALICE. Most A.I. people have assumed that to model human conversation, they needed to make the bot able to reason, to have basic memory, and general commonsense knowledge about the world. But Wallace realized that the vast majority of humans don't have those attributes. Indeed, the vast majority of humans spend almost all their time just wasting oxygen every time they open their mouths. Thus, a bot that automatically says "your just a hata" and "why you gotta be that way??" as a response to, well, almost any utterance, not matter how illogicaly connected it might be, would come off as more real than a bot that actually attempted to engage with your conversation.
Posted by: Clive at November 3, 2003 12:08 PM
I watch big brother
Posted by: Allan at November 29, 2003 6:46 AM
Nice site. thx.
Posted by: Online Casino at January 16, 2004 4:27 PM
Posted by: julia at January 24, 2004 7:06 PM
order cialis
Posted by: order cialis at February 22, 2005 3:00 PM