I'm sure the anecdotes are fascinating, but hard data is always more useful. Slightly over a year ago I did a quick analysis of the accident patterns regarding bikes in my own state, Queensland Australia, using the data that the Transport dept. released.
http://www.unorthodox.com.au/map/what_hits_bikes/
The data shows that there are particular 'hotspots' where bike accidents occur, usually were flows of heavy vehicles and multiple (complex) lanes meet. That suggests that we don't need bike lanes everywhere, but we do need clear bikeways through heavy traffic spots. Incidents are _not_ correlated with population density, but with heavy traffic density. There are also odd hotspots around places like public parks, though they seem to be of the more low-level parking-style incidents.
The original data included all road accidents, but was winnowed to just the bike events. The incident reports don't reveal individual details like which vehicle's driver was injured, but in the fatality incidents between, eg: a pushbike and a road-train, I'm pretty sure it wasn't the three-ton truck driver who came off worse for wear.