There can also be a code win: when forced to deal with short lines, it's often easier to split a long line of code into two easier to understand lines than it is to split on an operator.
Sometimes a split line is a lot worse than a long line, which is why the rule shouldn't be applied religious -- you should only be called on it if you have many lines over the limit.
IMO, 80 is a little bit tight; I prefer 100. But I wouldn't bike shed it if there was a standard in place.
Moreover if you need lines longer than 80 chars you're probably doing it the wrong way: long lines are difficult to understand
In principle, having a line width limit is a worthwhile idea. 120 rarely gives me problems, and I can fit 240 characters wide in any case.
Of course we do. We also have serial ports that go way faster than 9600, so at last "ed" can be used to its full potential.
Just keep in mind that code will be read and shown in a variety of different places. For example reading the code on github or other source control tracking tools, reading it on different terminals with splits active (tmux and VIm splits), etc. Keeping it short just makes it more portable for readability purposes. That's definitely a good thing.
Now get off my damned lawn.