What NEED does this meet?
Everybody who's had to rely on online map/directions sites know that often they create utterly useless directions such as "at the roundabout turn onto the A123", when there are two A123 exits or "on the B1234 take a left turn onto Fox Street" (so you end up driving slowly peering at every street sign hoping it'll be your turn)
What is the APPROACH?
Instead, use directions written by people. You know how to get from the motorway to your street, so you can write directions that a human can follow
"At the roundabout take the 3rd exit (there's a chipshop just before it) onto the A123 You should pass a big mobile phone warehouse then go under a railway bridge soon after the bridge, opposite the post office, take a left turn onto fox street"
Once the database is filled up with snippets of directions, the code would be able to string these together into longer journeys (and of course there'd be some kind of moderation system to make sure directions make sense)
What are the BENEFITS to people?
Nobody likes getting lost in a town they don't know, and often when we rely on machine-generated directions they can just get us more and more lost
What is the COMPETITION?
All the other mapping services I know of just let a computer generate the route and give a disclaimer that you should check it for validity yourself (which is fine if you know the town your driving in)
Giving nicely written understanable directions would make everyone's lives easier
What BUDGET & LOGISTICS are required?
I biggest hurdle would be actually joining the snippets of directions together into single routes. Access to existing map databases (or postcode/location databases) would be beneficial
One potential problem here is cost. As a GIS developer I can tell you that map data to use as a base for routing is _incredibly_ expensive.
There is also the question of maintenance and authority. Road numbers change relatively infrequently, unlike the type of landmarks mentioned in the proposal. Today's 'chip shop' might be a 'wine bar' tomorrow, the 'Red Lion' could become 'The Furry Monkey'. If you're expecting people to rely on landmarks they need to be kept up-to-date (a _big_ job) otherwise they become less useful than not having them at all.
Posted by: andrew at Oct 31, 2003 11:11:04 AM
I've addressed the cost and update issues in another proposal:
Posted by: Chris at Oct 31, 2003 11:54:22 AM