Google started with a simple assumption and built routing software on that assumption. Then they found a special case. They talk to people and realize that, for many humans, directions rely on their relationship to landmarks and streets don't always have names. Adjust the mental model and your software accordingly.<p>This is an excellent case study in the importance of data modeling as part of the design process. Google started their mapping software with Western assumptions ("All streets have names, and directions rely on transitions between named streets"). If Google didn't adjust their approach, they would be irrelevant to billions of people who navigate primarily by landmarks.
Great article. I noticed how bad google maps were back in India. I am very excited to try out the latest iteration when I get back!!<p>I wonder how Waze compares to google maps now.