![The Map Room: A Weblog About Maps [logo]](/maproom/images/title_inverse.jpg)
Correcting Directions
Dave Winer discovers a better route than the directions suggested by Google Maps, and wonders: “So — when does mapping become a two-way app? I’d be willing to tell their software that I have a better route, it’s one that comes from living here, and being a cab driver here.”
User-submitted directions look like one of those ideas that are brilliant in theory, but you can’t figure out a way to make them work in practice. There are, I think, two problems with it. One is mediating competing directions: how do you adjudicate between two alternate routes, each provided by locals who swear by them? (I can usually figure out several ways of getting from A to B; figuring out which is best is a bit harder.) The other is that there are simply too many different A and B points to navigate between: it’s one thing to know the best route between towns, or even between neighbourhoods, but how granular would user submissions be allowed to get?
It wouldn’t surprise me if driving directions were procedurally generated rather than stored in a database, in which case most of the errors would occur when real roads don’t match up with the algorithms’ expectations. But that’s just a wild guess on my part.
Previously: Tele Atlas Introduces Map Feedback.
The route-planning algorithms have a LOT of room for improvement. Often you can tell by eyeballing a map that the computer-recommended route is poor.
For instance, in going from Champaign, IL, to Kansas City, the crucial determinant is, Where are you going to cross the Mississippi? Or, even simpler, “What Interstate highways cross the Mississippi at that latitude?” A human would immediately know to go via Hannibal, MO. Web-based map software would lead you via St. Louis instead.
Note that this is an entire CLASS of endpoints — It’s true of all endpoints in central Illinois and all endpoints in western Missouri.
Bill Strutz | 01/20/2007 at 3:30 PM | #
You have to consider factors such as weather and time of day as well. During rush hours a lot of people take roads that mean driving a longer distance but save time by avoiding crowded intersections. Sometimes an older road that paralells an expressway moves faster during rush hours. In hilly areas there are roads to avoid during snow or ice storms because, inevitably, someone unprepared for such conditions will get stuck and block all traffic. Yes, DOT could do a better job of clearing snow, but even snow plows get stuck in traffic.
James J Linnane | 01/21/2007 at 6:57 AM | #
Your starting to see some of it with mapquest that lets you create a route with 10 custom stops. I don’t think it lets you select the additional stops off a map but thats the next logical step.
http://www.mapquest.com/features/main.adp?page=whatsnew_multipoint&cid=4_learnmore_mp
AL | 01/21/2007 at 9:30 AM | #
You also get the usual problems with prank submissions and more importantly freshness. With each update about street structure, how do you determine whether a user-submitted route is still optimal or whether it should be discarded?
HD | 01/22/2007 at 12:44 PM | #