It's interesting how so many of the early tools were designed to create "communities" (mesg, talk etc.). The semi open nature of the platform really encouraged it too. It's nice to be able to cd into someone else's home directory and look at their files.
The passwords were only if you were connecting over the network. If you were using a directly attached terminal, you didn't need one.
RMS insisted that everyone use their UNAME as their password, but he wasn't widely listened to because the whole reason PWORD came into effect was because turists were getting increasingly destructive. People weren't happy when their mail got marked read (or worse, deleted) because some random from the network had logged in as them simply because they could and did not understand what their automatic login script was doing.
That was only true in very early systems. By the time of the PDP-10, HACTRN will nag you to log in if you run most commands and the gunner would kill off your job after a relatively short interval (the exact interval differed from machine to machine).
ITS had no file permissions, but even before PWORD was installed to keep randoms from the network away there were means of keeping the turists out when the system was to be reserved for Real Work. Other parts of the system that were considered sensitive were hidden behind undocumented commands or program-level passwords - For example, the innards of INQUIR, since the INQUIR database determined who was to be excluded and who was not.
There may have been no file permissions, but there was a definite hierarchy of users that was enforced by other (generally more subtle) means.
