[OSM-talk] Defective GPS trace

[andrzej zaborowski]

2010/1/2 Anthony <osm at inbox.org>:
> This seems like somewhere that the wisdom of crowds actually applies.  I
> think I'd trust the average of a bunch of independent GPS traces to a single
> orthorectified aerial - especially in an area which isn't extremely flat.
> But I guess I might be convinced otherwise, if I'm actually shown the
> "sub-metre accuracy" claim, which presumably outlines the methods utilized
> to ensure such accuracy.

Average sub-metre error would still allow for a place or two with 50m
error especially over a big area such as Australia, and Steve might
have visited one of these places.  It would be best to just go there
again and see if on the second survey the error is siginificantly
lower (or there's still error but the the vector is opposite
direction).

For example based on GPS traces I can say with some confidence that
Yahoo is on average < 3m off in my city even though there is a couple
of places that are badly mangled.

Cheers