The only real problem I can imagine is when you are looking for an old tweet which you know is beyond x amount of pages. In the old system you could select a page number and start from there. I guess Twitter is insisting on the limited relevancy of tweets by allowing you to choose your own depth of recent tweets rather than providing x amount of pages of archives.]]>
The ajax style 'more' is interesting in twitter's case because often there is no 'back' or at least it's changed quite a bit since the user was there.
search.twitter.com is a good example of this. They still use the < newer | older > pagination there. On a popular query the 'first' page often changes while you're viewing the second. Going 'back' can potentially skip over whole 'pages' of tweets.
One option would be to populate the browser's history as you extend the page's content with 'more'. 'Back' wouldn't change the the content at all, it'd just take back to the last piece of content before you clicked 'more'.]]>