Strange Blog Spam
January 21, 2006
I pulled down my e-mail this morning to observe that about four or five comments to posts had gotten through my RDOS filter. The strange thing was that the comments themselves could almost be legitimate.
In the end I deleted them because:
- They were posted to very old posts (the older a post is the less likely it is that a comment on it is legitimate since it falls out of RSS and someone would have had to have gone digging for it).
- The posts themselves probably weren’t comment worthy.
- The comments looked vaguely related to the post, but it was almost as if they had skimmed for a couple of keywords and then concocted a sentence to suit.
- Four of the comments return to the same URL.
It is possible that the spam bot was automated but I have to wonder. One of the URLs was actually mistyped which programs tend not to do unless they are specifically programmed to – if thats the case then the stacks in the spam war may be getting raised where simple keyword filters won’t do the trick.
January 21, 2006 at 12:00 am
Sounds like a human. Porn-addicted humans apparently work as decent ‘bots’ from what I’m told.
January 23, 2006 at 12:00 am
Mitch,
I hate to tell you this but I actually pioneered this type of spam. I like to call it Blam. I wrote it as a proof of concept for a dirty form of ‘legitimate’ marketing and never really did anything with it.
Here is the way it works (quite simple actually and it let me fiddle with the neat text mining components of SQL05). Oh first though. I’m not the guy blamming your blog.
Start at a place like Syndic8.com. Pull thier primary RSS feed and start a ‘tree’ walk aggregating the blog content into a local data store. Have some peson write a dozen or so ‘trainer’ blurbs of content about a product topic you are trying to sell. Make them from the first person view and make them generic, but ‘on topic’. Next leverage the comment API to post the stock comments to posts that ‘match’ with a relatively high degree of confidence (95%+).
I originally did the proof of concept to see how people respond to ’spam’. My hypothesis was that most people don’t have a problem with an unsolicated message they have problems with unsolicated messages that don’t appeal to them. When I ran my tests I was successful and getting ‘advertisements’ into peoples blogs without them deleting them and without them assuming they were ’spam’.
I didn’t follow up with the project though. I went through this kick about a year ago and it was basically a diversion back into technology from my extended diversion into the World of Warcraft (9 straight months of WoW is not a good thing… go figure). I might have done more with it at the time but just didn’t see it as a viable way to make money.
Again my point in the project was to get a feel for the consumer response and what I learned was very telling (well not to me, I validated what I already assumed).
People are not offended by the unsoliceted offer when it is relevant to them.
January 23, 2006 at 12:00 am
Hi Paul,
Heh – in this instance it was a link to some kind of Destiny’s Child fan site. In made me nauseous, I’d say that was unwanted
Now if it was a Pink Floyd site on the other hand . . .