I understand I'm not in the cool camp here..but you are completely wrong.
Sigh. You're correct with your examples, but there's other issues at work here.
First, there are perfectly valid unsecured wireless networks out there. Ones that want you to use them. Some people even run them out of their houses. So it's rather difficult to tell whether or not you're supposed to use the network or not (and by use, I mean just connect to and surf normally; not try and break into any machines, or use it as a dupe network for spam/pirating/child porn/etc).
Optimally you'd contact the owner of the network beforehand, or there would be some flag on the network indicating that it was for free use -- but there really isn't. You could accomplish the former with sufficient triangulation, but it's rather difficult to do. As for the latter, well, that's what broadcasting the SSID is supposed to be for, but it's enabled by default on virtually every router out there, so it's useless as well. And there's no convention for SSID naming to indicate open use.
On top of that, there's some FCC regulations on broadcasts, even low power ones such as wireless networks, which make it difficult to claim theft-of-service if you don't take even rudimentary procedures to secure the broadcast station (just turning off SSID broadcast would be sufficient; adding a key, even if crackable, is better).
This is a different situation from leaving your door unlocked, or money laying around, etc. It's like putting up an "Open House" sign on Sunday and then complaining about people wandering through your house the next week because you forgot to take the sign down.