I agree with mbbrutman. Ahm posted this link on the IRC channel the other day, and it left me with more questions than I was impressed. I know practically nothing about Twitter, but suppose communication to their servers is done over TCP/IP, possibly even HTTP? I get the feeling you would browse a particular web page, but perhaps that is an optional way of reading the messages.
The page mentions Ruby on Rails and the backend being kind of heavyweight but I thought the computation hard part was the server end, that a relatively inferior client should be able to poll and pop home new messages as subscribed to. After all there exists HTML browsers for 1 MHz 6502 computers, and a mainly (entirely?) text based service like Twitter shouldn't really be more complex than a generic, low end web page.