Computer-generated directions sucks

The NYT has today a terrible piece about online mapping services failes to be a godsend for travelers.

Roughly 1 in 50 computer-generated directions is a dud, according to Doug Richardson, the executive director for the Association of American Geographers. He blames inaccurate road information for most of the failures. (...) "You have to have the latest data about road characteristics - things like one-way streets, turns and exits in your system in order for it to generate accurate directions," he said.

Even if the streets remained static, online mapping would be an inexact science. Most of the major Web sites draw their data from a small group of competing suppliers and update their maps quarterly. They use a process called geocoding, which assigns a latitude-longitude coordinate to an address, to find a destination. Then their systems calculate the most efficient route. Each site handles the data in a slightly different way, which is why search results vary from mapping site to mapping site.

Then, the article mentions few troubles:

"On Long Island, two nearby towns may have roads with the same name," she said. The sites do not draw a clear distinction, she added, "So it's easy to make a turn on the wrong one." (...) Her frustration recently boiled over when she queried three mapping sites for directions to LaGuardia Airport, and received three different sets of instructions. "Mapping sites give me a false sense of security," she said. "I don't trust them anymore." (...) Thanks to MapQuest, I've missed numerous business meetings in Washington. Recently, I clicked on MSN Maps and Directions to find the quickest way to an Italian restaurant in Orlando, Fla. It pointed the way to a quiet residential neighborhood but, alas, not to the pizza parlor I had in mind.

And when it comes to solutions, here is hwo it goes:

That doesn't mean you have to get lost. To improve your chances of making your next business meeting, consider buying a navigational computer that uses G.P.S. technology. Those systems constantly monitor your position and calculate the most efficient course. An old-fashioned atlas would help, too. Or you could do what Ms. Taub, the computer tutor, recently started doing. You could ask for directions.

Why do I blog this? well it's fantastic, again we have a technology complex to deploy (it's difficult to give a direction to the user when each city has same street names) and people now don't trust the system. What I find great here is the conclusion: the most efficient way to find your way is to ask someone. This is social navigation, the most common way to navigate through different space: physical space, information space... SocialNav starts from the idea that “when people need information, they will often turn to other people rather than use more formalised information artefacts” (e.g. asking people for advice when lost in a city instead of studying a map). Studies on how people followed crowd are also an influence (Follow the leader to bagage claim...). Would it be more relevant to design a navigation tool based on social navigation?