The author forgot about the fact that high quality/quantity data comes from good sensors. The cost of the LIDAR sensors (the one google using and this is the one which rotates on top of the car) is pretty high (around $50K AFAIK). This is one of the reasons Google is holding back in the hope that after few years the cost of such sensors would drop dramatically.
Self driving car by just using CV just wouldn't work (atleast with the current state of the art).
Gates doesn't do charity in the traditional sense. He focuses on very specific areas and measures the success well. Page was probably talking about traditional charities.
The problem with OSM (as compared to wikipedia) is that there is less awareness about the project and contributing to OSM has to be made a lot easier. Also it needs to have enough eyeballs per sq. mile to verify that the changes made are actually good. In some places like Melbourne (and someone mentioned this earlier) OSM is horrendous.