[Openstreetmap] Talking about Licences

[Jo Walsh]

On Thu, May 26, 2005 at 11:35:05AM +0100, Tom Carden wrote:
> When you have enough streets in the map, you need a way for people to find
> their own street and then tell OSM what their postcode is.  Would that work?

Using web-facing sources of postcode geocoding, i often get really
terrible results; the Town Hall, for example, consistently
postcode-geocodes 200M from its real location via google, streetmap, etc.

In the US where postal zones are much larger, address-based geocoding
works pretty well, based on address numbers for the top and bottom of
the left and right sides of the street segment; this is how
geocoder.us works, extrapolating between the numbers. I'd be
interested in pursuing this in a very constrained area of London and
see how it works out.

Our architectural topology makes this rather harder than in the US;
walk round a classic 60/70s social housing estate and get a sense of
the weird walks and courts and alternanumbering schemes for different
blocks. :/ but given that, we might still get on-average closer than
PAF?
  
> So, suggestions please, for attracting "eyeballs and footsteps" and
> developing "doorways".  And there I was dissing marketing-speak.

For me, instant gratification after the upload of a GPX track is a key
missing piece. Unless i'm really, really committed, then i'll upload
one track, look at all that debugging information, and think, "now
what?" An overlay of what i've just uploaded on the existing data set
would really make a difference to the experience.

Saul and i had talked about having 'bounties' for specific areas -
defined by a set of streets or just an arbitrary shape; you'd sign up
for a competition, and at the end of the week, the bounty would be
split between the participants, relative to the amount of time they'd
spent, or a more complex metric that wasn't too weighted in favour of
car drivers ;) this involves fundraising, though :/

Generating our own Garmin basemaps - so dedicated people could load up
a morning dump and seek out places that hadn't been covered - would be
sweet, but i don't know a free and commandline way of doing that.

But this is just to get over the hump with the line segments; creating
and gnoming metadata is the much bigger and less
pleasurably-actionable commitment. When i was messing around with the
Distributed Proofreading side of gutenberg, i found the 'league table'
aspect pretty compelling; enough to drive me to kick out a few pages a
day, while the craze lasted with me, so as to not see my puny rating
drop. 

Finding interest groups with a big use case would help this grow.
Whether that's the wireless network scene - e.g. bristolwireless are
interested in GPS mapping east bristol in order to have a base map for
their network - http://www.eastlondonpostcard.co.uk/ are interested in
publishing walks to go along with their amazing collection of
comparative photographs, modern exact recreations of locations and
angles of old photographs - really angst-inducing. A lot of
semi-retired people belong to local history interest groups in this
way; they care deeply for the local and could feel proud of data
gardening their backyard. 'adopt-a-tile'? ;)


zx