[OSM-dev] A new take on the "mutable" idea

Jochen Topf jochen at remote.org
Mon Jun 22 07:38:48 BST 2009


On Sun, Jun 21, 2009 at 03:15:12PM +0100, Matt Amos wrote:
> we went over this before: if it isn't editable, it shouldn't be in
> OSM. there's nothing to stop people adding 3rd party datasets into
> renderings based on OSM or using these boundaries to perform analysis
> on OSM data. i can't see a good reason why OSM has to import
> *everything* - we're not the ultimate data repository, we're the free
> *wiki* world map.

Maybe we should have everything in OSM, but we should have those valuable data
sources with the accurate data!

Currently we have a special database for GPS tracks. This is part of our source
data. Everybody can use it to check his own tracks against other peoples tracks
and both against the main OSM data. Maybe we should extend this model to other
data sources. Say, you import all the roads in one country. The first step
could be to load this into a "source" database. Like with the GPS tracks, it
would contains the original data and that would never change. Then the normal
data conversion and import into the mail OSM database is done.

Along with the data we can store information about its source, accuracy, etc.
The mapper then does what cartographers have been doing for centuries: They
collect all data about an area, compare it and distill the best view out of the
data that they can get. Everybody can still change everything, but everybody
can also always get back to the original data. If we want it to be fancy, we
can add ways of tracking where which data came from and introduce alerting
mechanisms.

Basically this idea acknowledges what has been true for a while: GPS tracks are
only one of many sources. Imports are more important now and getting more and
more so. And with more and more areas where the basic data is already complete,
the job of the mappper moves from beeing somebody who enters new data to
somebody who edits existing data. Somebody who does the hard work of integrating
the data from different sources into a coherent hole. A job that doesn't work
without the human judgment.

Of course, for the mapper this would bring new challenges. He might be able to
see 10 different layers of data in his editor with as many data sources. And he
might have to look at them all to decide who is right and whats important. But
humans are good at those things.

Jochen
-- 
Jochen Topf  jochen at remote.org  http://www.remote.org/jochen/  +49-721-388298





More information about the dev mailing list