If you look at people who have a large number of followers which their activity or identity wouldn't normally warrant, you will find that the synthetic accounts who follow them exhibit a lot of random retweeting and favoriting behavior. Many of the synthetic accounts also do double duty by posting spam links should anyone visit them.
All speculation, but:
There are people who offer paid follower, favoriting, and retweeting services. I suspect the operators include more bogus activity than paid activity on the synthetic accounts. The additional activity would muddy any third-party analysis. For example, it would make it harder for Twitter to spot bogus accounts. It would also be harder to know who's really paying for artificial activity versus who's been randomly selected (if Twitter even cares about such things).
If the crap favoriting gets these accounts some non-synthetic followers or visits, they also become useful vehicles for those spam links. It's doubtful that Twitter worries as much about accounts that spam if people visit the account after a favorite and see a spam link, rather than the spammer revealing the spam to others with @mentions.