I'm hearing from many Twitter users that the frustration level caused
by the "Britney Bots" is rising. I'm going to use some euphemisms to
make this message safe for work, but the particular bots in question
are certainly not work-safe.
The _modus operandi_ of these bots is as follows:
1. Get a Twitter account. These are usually of the form <small English
word><5 digit number>. The profile picture is typically not safe for
work.
2. Collect screen names somehow. They must at least be polling the
public timeline. Frequent tweeters seem to get more of them. Perhaps
they are doing searches as well, or mining the profiles of the screen
names they've collected for more screen names.
3. Send an @ reply to each name collected. These come in bursts - I
haven't done any research into the frequency at which they are sent
but a number of tweets go out in a burst. The tweets themselves are
not safe for work.
The bots do *not* appear to be following anybody - they only show up
if you do a "mentions" search. What's worse, though, is that people
are retweeting these things! There is a movement on Twitter, using the
hashtag #StopBritneyBots, to attempt to get Twitter to put some kind
of filtering in place. I'm not sure what the status of that is in
Twitter - perhaps some of the Twitter people on this list can chime
in. Meanwhile, this particular bot has an easily-detected "signature"
- you can collect the bot names via Twitter search!
1. Do a Twitter search for the following string (the double quotes are
part of the string!):
'"(Click the link at top right of my profile)"'
Note that the returned tweets from this search will mostly be not
safe for work!
2. Break each resulting tweet into space-separated tokens.
3. Scan the tokens from right to left. The first @name you encounter
will be the destination "victim". The second one you encounter will be
the bot that sent it.
At this point, you could build a bot to report the bots as spammers.
Personally, I think anyone who retweets one of these ought to be
considered a spammer as well. ;-) In any event, I've got some code
using the Net::Twitter Perl library that collects the tweets, and I
can supply a list of names to Twitter if they'd like.
I'd prefer, of course, that Twitter deal with this at the inlets to
the tweet stream. But I think there's a significant enough groundswell
in the "community" that we will see bots arise using the algorithm
I've described above. I've been asked to create one, but I'm holding
off - there are some murky legalities involved and I have more
interesting research in Twitter text mining I want to do. ;-)
Twitter, what say you? Developer community, what say you?
--
M. Edward (Ed) Borasky
http://borasky-research.net/smart-at-znmeb
"I've always regarded nature as the clothing of God." ~Alan Hovhaness
