It's a Mad, Mad, Mad, Mad Twitter

Colleagues, I would like to ask for your advice.
I'm experimenting with Twitter API in a paradigm rails-app. Scheme of the application:

1. get all the followers in array
2. get all friends in array
3. compare the two arrays and get lists of accounts for Follow and Unfollow

Much is unclear to me. For example.
Take a look, please. At the level of Twitter, these requests, probably, require the same resources:


Unfollowing works without problems:



But for Follow I have to take a rescue for the application (Twitter::Error::TooManyRequests) every 10 to 15 minutes:




Sometimes Follow does not work at all (this can last a day or more), and the log looks like this:


I do not understand the algorithms of limits of Twitter. I think the rules are more complicated than described . Does Twitter try to prohibit following? I'm sorry, I do not understand. Any ideas?
А по-русски - здесь .