Bots, scrapers, harvesters etc. It seems they adapt to every new platform from Facebook and Digg through to Q&A sites and IM – and of course microblogging services like Twitter.

Following on from Archibold Teriyaki’s Twitter email spam test, I figured he had a few more uses before becoming too cross-contaminated for me to consider the tests objective.

Objective

If I send an @reply to a new Twitter account from an existing account, will the auto-follow bots be clever enough to follow the new account?

• Set up a new Twitter account - @ArchiboldT
• Do not tweet anything with this new account
• Tweet a message to the public timeline from my own Twitter account @lindop containing ArchiboldT’s username
• Monitor anyone that chooses to follow ArchiboldT

The Test

Tweeted 19:23 10th July 2009 from @Lindop

@ArchiboldT – this is a test

Results

12:35 1st August 2009

ArchiboldT’s Twitter account shows no followers, bots or otherwise. ArchiboldT’s Twitter alerts via email says otherwise!

• qdouble_tree (double_tre427) – 12th July
  0 followers, 1 update, following 245 people
• Karla Marrufo (KarlaVenez460) – 14th July
  44 followers, 2 updates, following 1986 people
• Georgia Martelino (georgiasdf4256) – 15th July
  1 follower, 2 updates, following 759 people

All three accounts have been suspended by Twitter “due to strange activity”. This would explain why Archibold Teriyaki’s Twitter account shows no followers.

Conclusion

Yep. Bots can and will use replies from other people to discover new accounts to follow. So it makes sense that the more people send tweets to you via @username, the more auto-follow bots will find you. That’ll teach you to be so popular!

What can we take away from all this? Well, it looks like the team at Twitter are doing a fairly good job of identifying and cleaning things up. The very nature of Twitter means your usernames will always be visible to any basic script, and having fake profiles following you isn’t going to hurt you so just sit back and accept it as inevitable.