If someone runs an entry node, they get to see the IP addresses of 1/N of Tor users, how much data the users send and receive, and when. But they don't get to see what data the users are sending and receiving, or even what exit node they're using. If someone runs an exit node, they get to see (and filter) what data 1/N of Tor users are sending and receiving, and when, but they don't get to see who those users are, or even what entry node they're using. And the data they see can still be encrypted; e.g. connections to Gmail will use TLS/SSL, so even the exit node doesn't get to steal your mail.
FWIW, during the recent DigiNotar SSL shambles, there was evidence that some Tor exit nodes were using forged Google certificates to MITM "secure" connections to Googles services.
So that may be why.