Bambibot: Evolution

Wow, twammers. I'm almost impressed. Nearly a 1:1 following:followed ratio, and some 1700+ followers at that. Believable Twitter updates, AND a blog backing the profile! Though bad grammar, capitalization and syntax are found throughout the text, it's sadly believable as the product of a child of the SMS generation.

Of course, turning to the blog, it's a spamblog/link farm. The supposed owner of the blog's pictures link to an adult dating site. And yet again, the screen name, "MichelleLoli," doesn't exactly match the alleged actual name, "Stefi Tossie."

Here's an account I followed for a while, where the person/people behind the account actually seem to interact with the other people the account follows. No actual conversations, but occasional, relevant, one-liner @ replies to something you said. Maybe the woman behind it really does Tweet, and she's just a really busy pornstar/dominatrix ... or maybe it's a half-hearted effort to pump traffic to the pornstar's site by some cheapo marketing firm or seospammers. Bambibot, or no?

New Twist on Twam

Noticing a scary twitterspam trend these past couple weeks: bambibots and other spambots on Twitter are getting a lot more subtle. A few months back, you saw them spewing out random bits of text obviously taken from other sources, but those bits were often fragments, and the following:followed ratio was hundreds:1 or worse. Account names were still pretty obviously simple enumerations or iterations.

This week, I'm seeing spambot accounts with reasonable following ratios, believable names, and bits of text that would almost make sense as entire thoughts or sentences, if you didn't understand that a word like "uni" for "school" or "college" wouldn't be used by a girl in Louisiana, particularly not in the middle of August.

http://twitter.com/clairemcdaniel2

Big giveaway: name reads "Kayla," Ms. 'Claire.' Ooops!

The linkspam volume has died down. You could almost believe it's a real account, not just a broadcast mouthpiece for porn, SEO and MLM links.

One giveaway the nasty spammers seem to fail to be dealing with: the source parameter. Without an officially registered app, the source parameter describing the app that a tweet was sent from is going to be displayed as "From API." Not "From Web." Not "From Tweetdeck" or "From Seesmic" -- "From API."

With OAuth, there's no reason for legitimate users to be calling in over Basic Auth anymore -- no reason a legit user should be displaying "From API" -- certainly not with any regularity.

Developers: register your app, use the freely available OAuth libraries that are proliferating, and deal with the occasional OAuth downtime.

The Bambibots Cometh

OK, so what exactly is the deal with bambibots? You know, that subset of spambots that haunt social networks, post a sexy photo and spew out lascivious crap like "I hate my [boy/girl]friend you need to cum f*ck me now www.obviouspornlink.com/ghg43p993p4" ???

Do people really click on this crap? I guess they must, because like the slew of email spam that clogs my pipes, someone puts an awful lot of effort into creating these spambots to drive these Bambi/Amber/Monik/Jezabelle/Irinia53530/so on and so forth accounts, and if there weren't a financial incentive, it simply wouldn't make sense. I wish there were a better way of blocking them.

A few weeks back I'd posted a suggestion on the Twitter development list about tunable anti-spam measures -- like the kind of utility that GMail gives you. Once I start blocking accounts, there ought to be some intelligent algorithms working behind the scenes to understand stuff like:

1. if someone is following hundreds or thousands and is followed by a handful, ignore,
2. if someone has had no conversational interaction with any other account, ignore,
3. if the "person's" photo is of them in a bikini, ignore,
4. if there's a link to any known porn site, off a list, or as recongized by other users, ignore!
5. if the words "SEO" or "empowerment" or "money" or "cash" or "prizes" or "smoke up my butt" is found anywhere on their page, IGNORE!!

I mean c'mon ... Twitter's made some strides in spam control, but they're far from where they need to be. I'm tried of waking to find my inbox filled with followspam. Aren't you?

Is GMail filtering spam Twitter account notifications?

This would be incredibly slick, and so totally Google, if true. Overnight some, but not all, Twitter notifications to my GMail account started going into the Spam folder, which has never happened before, not in the 2.5+ years I've been using Twitter. Each of the notifications that got marked as spam turned out to be for bambibot/spambot accounts with few or no followers, following a ton of people, sending out lots of spammy links.

That would be a pretty freaking cool process, if Google were in fact able to distinguish those accounts ... and if so, maybe they could/should share their algorithms with Twitter and help defeat these spammers at the source.

Edit: may have to scratch that GMail theory, I'm getting more spam follow notifications in my Inbox ...