I got a piece of junk mail the other day from Sprint PCS, my cell phone provider. Typical stuff, but what caught my eye was that the postcard had a little graphic that said where I could find my "Neighborhood Sprint Store". Presumably this was done by hooking into some GIS database and figuring out that the closest store to me, as the crow flies, is somewhere else in Brooklyn. But if I were to go shopping for a new Sprint phone I wouldn't go there: It may be the closest to me, but that doesn't mean it's easiest to get to.
Other subway-dependent Brooklyn residents know what I'm talking about: The subway is Manhattan-centric by design. So traveling to another part of Brooklyn often involves going into Manhattan and then going back out by another line. I'm far more likely to go phone shopping at a store in Manhattan, especially considering that I work in Manhattan as well.
Getting from point A to point B doesn't just involve geographic distance. It involves a lot of other questions: How much will this travel cost? How congested is the route? What are the chances I'll get lost? If you own a car and live in a city with only moderate traffic you don't think about these questions as much. But to extend my example: Taking the city bus to another part of Brooklyn is not an option, since the bus lines just seem too confusing, and I never became familiar with their routes. Yes, this might mean I'm lazy, but I know a lot of other New Yorkers who feel the same way.
Human engineering reworks space by making it easier to travel along some routes than others. Take flying as another example. The cost of a plane ticket is roughly related to the distance, but if you search for tickets on a site like Expedia you'll be offered choices such as paying less money to fly a longer route, through an intermediate hub. And why does a one-way ticket often cost more than a round-trip ticket? Because once the air travel network has been built out, miles flown is not as important to the airlines as the complicated task of balancing cost, capacity, and demand.
The importance of distance breaks down most visibly in a place like New York, where dense transit networks and intense crowding conspire to create a physical environment in which normal strategies are irrelevant. Mapping software can give me the nearest Sprint store by geographical distance, but not by subway commuting time. Even if I rent a car for the weekend, Mapquest can still fail me: Ask it how to drive to Philadelphia and it will give you no sensible advice as to whether you should take the George Washington Bridge or the Holland Tunnel out of Manhattan. To do that it would need intimate knowledge of Manhattan traffic patterns, and it would need to know what time of day you're planning on leaving, and even then it wouldn't be enough because traffic varies heavily from day to day. Mapquest can recommend the shortest route by geographical distance, but if you hit a bad stretch of highway coming out of Manhattan on a Friday evening you could waste an hour in stop-and-go traffic, at which point driving a quarter-mile less in distance would be scant consolation.
In fact, you could say that a service like Mapquest proffers a conception of space that's profoundly suburban: Space is flat and undifferentiated, and it's assumed that everybody owns a car. Not that the city kids don't have distortions of their own. The New York City subway map, probably the single most commonly viewed map of the city, is wildly distorted to account for the fact that subway lines and stations are much more dense in Manhattan than in the outer boroughs. Of course, this means that Manhattan is proportionally bigger in the subway map than in real life, which could affect our psychological impression of Manhattan's importance, which could affect City Hall's funding priorities regarding the development of new subway lines ... We make our tools, and our tools make us.