The Twitter Turing Test

There are a number of interesting scams on Twitter. “Scams” is maybe too harsh a word, because there’s not much harm being done; but there’s deception all the same.

Supposedly, the reason why social media is such an effective marketing tool is because of its “intimacy” and “authenticity”. But such things take a long time to cultivate, so people will naturally want a shortcut to the pay-off. And thus are launched a thousand Twitter scams, which are all variations on one key technique:

Make a bunch of fake Twitter accounts, but try to make them look real.

You can pay people to make these fake accounts for you and then have them all “follow” you on Twitter and tweet about how awesome you are. And if you did this, you’d be in good company: Mitt Romney’s campaign has been accused of doing it.

The only problem is, it’s hard to make a lot of fake accounts that actually look real. In fact, they’re usually quite inept. You know some hacker has written a program to automate the creation of the accounts and their tweets, and I’ll give the hackers some credit: usually, the accounts at least have profile pictures and plausible names, instead of having no pictures and names like “X11J9”.

But still, it’s not hard to detect the fraud. Just today, I saw I’d gotten two tweets from different people with the same exact wording — which, moreover, consisted almost entirely of a regurgitation of one of my own recent tweets:

All they did was add, bizarrely, the word “Lotta” to my tweet, along with a link — which happened to be to a site trying to sell you Kindle Fires for less than Amazon sells them.

Here’s a look at the tweets of “June Paterson”. Again, pretty transparent, but notice there’s at least some attempt at sophistication. First, “June” has more than zero followers, which would have been a dead giveaway that she was just a spambot. (Of course, June’s followers are almost certainly other spambots created by the same hacker.)

 

Second, the hacker has maintained a plausible ratio of, say, three to one, of June’s “personal tweets” to the tweets they really want to send (the ads for the Kindle Fire).

What I love is how nonsensical these “personal tweets” are. The next-t0-last one in the screenshot above (“vS”) surely takes the cake for the least convincing tweet ever. Here’s another sampling:

 

What I want to know is, what the hell is up with that next-to-last tweet? Do these auto-generated tweets draw upon a corpus that contains rap lyrics, or is this spambot just plain racist?

More Tales of Twitter here.

No scams here — just pure wonk rock!

This entry was posted in Tales of Twitter. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *