I am trying to get the list of followers for a script I'm working on. When
I get the full list in JSON/XML format, all is great. When I try to use
pagination (?page=x) it acts weird.

I tried to obtain the list of TechCrunch and this is how it works for the
first few pages:
Request URL (http://twitter.com/followers/ids/techcrunch.json?page=1),
returns 5000 results
Request URL (http://twitter.com/followers/ids/techcrunch.json?page=2),
returns 15000 results!!
Request URL (http://twitter.com/followers/ids/techcrunch.json?page=3),
returns 20000 results!
(the next few seem to be consistent, somewhat! Increasing 5k per request)
Request URL (http://twitter.com/followers/ids/techcrunch.json?page=7),
returns 38464 results!!
Request URL (http://twitter.com/followers/ids/techcrunch.json?page=10),
returns 53814 results!!

It's obviously a defect!

Isn't it supposed to return 5000, 10000, 15000 and so on, or - even more
reasonably - 5000 per page without including the previous one? 

Thanks...