I think that is the heart of the matter. The most difficult place for cyclists is intersections, and if bike lanes are done intelligently there, they might help. Unfortunately, it seems that a lot of traffic engineers don't know how, or don't bother, to do that. We could probably debate all day whether having bike lanes that give up at intersections are better or worse than nothing, but let's not. My point is that if a particular bike lane doesn't help you at intersections, I don't see much return on investment from only having it on the straightaway. Even if it encourages more cyclists, which is a standard bike lane justification, I hope they don't get hurt at the intersection!

I've only been bike commuting (and bike riding at all as an adult, aside from very infrequent casual neighborhood rides) for 4 years. I still remember from before I started, noticing bike lanes while driving my car. Most of them ended before the intersection, and I remember seriously wondering what bikes were supposed to do at intersections if they wanted to go straight or turn left. I really didn't know. I don't think I was even sure when I started bike commuting. Most of what I know (which currently roughly equates to what is taught in the LAB Road I course, which I took last June) I learned from these forums, and the resources they pointed me to. I guess it is welcoming to new cyclists to see bike lanes, but my own initial cluelessness about intersections leads me to believe that unless bike lanes alone can reliably lead cyclists through intersections safely, which IMO is doubtful, there has also got to be some cyclist education somewhere.

Motorist education and punishment can't be the entire solution, IMHO. What is a motorist supposed to do when a cyclist turns left in from of him from the bike lane because the cyclist thinks he has to stay in the bike lane at all times and hasn't looked back properly?