[OSM-dev] [OSM-talk] area topology

[bvh]

On Wed, May 14, 2008 at 04:41:34PM +0200, Marcus Wolschon wrote:
> GPS-devices are very, very good at providing correct coordinates
> relative to the rest of the track. They are getting good at giving
> the same coordinates twice, when walking the same road again a few
> days later.
> Thus small topological features can be mapped very well. They may
> not be perfectly at the right place but they will be very accurately
> located to nearby features. (Given a modern GPS-device. These things
> improved quite a lot in the last years.)

This relative correctness will quickly break down in reality
The problem is osm is a collaborative undertaking. So you'll
find one user mapping the roads with a GPS receiver of chipset X
while another wants to map the forest whatever with the
yahoo satellite and yet another one is doing a building at the
opposite side of the street with GPS receiver with chipset Y
half a year later.  All with different biases on their absolute
positions... 

cu bart