But what really interests me is a set of bizarre tweets like these, all of which mention "Spinuzzi" in some way:
These all appear autogenerated from a relatively small set of templates:
- (firstname) (lastname) is (adjective). (pronoun) (verb) the (noun).
- Going to watch (movie name), (firstname) (lastname) r u ready?
- (firstname) (lastname) thinks I'm (noun) and (noun).
They've been going on for a while now, at least a few months, and they've increased dramatically.
Let's click through to one of these accounts and see what it looks like:
Again, we see a relatively small number of templates cranking out a huge number of variations. Why? I'm guessing it has something to do with some of the other tweets interspersed between these ones:
If an account exclusively sent out links to products, perhaps it would be caught in Twitter's spam filters. But if the account also sends out a wide range of legitimate sentences - sentences that have nouns, verbs, and adjectives - it might trick the spam filter into seeing the account as genuine. I suspect that's what's happening here. Bots lying to bots.