[Talk-us] tiff, dwg and nad83

Serge Wroclawski emacsen at gmail.com
Wed Apr 18 01:18:04 BST 2012

On Tue, Apr 17, 2012 at 7:51 PM, Dale Puch <dale.puch at gmail.com> wrote:

>> I think this point is controversial, so let's stick with some points
>> that aren't:
>> 1. In our history of imports, a very small percentage have been good.
> Compare a users early imports to their early mapping

They aren't nearly comparable in my experience. In all my mapping, I
see tons of bad imported data, but I see very few errors due to newbie
mappers. Sometimes I'll find a mistagged feature, or an unclosed way,
but overall I find users to be fairly careful, and when they're not,
the mistakes they make are small and confined.

Imports, on the other hand, are widespread, pervasive and are often
hard to detect. I see features which are hard to verify without
surveying, and I find odd OSM features, like ways that share the same
geometry (even the same nodes), or duplicated features close to each
other, but from different sources.

Your argument that the problems are similar don't bear out in my
experience mapping with OSM, the problems are neither similar in type
nor scale.

Nonetheless I've personally taken steps to address both. For new
mappers I am very involved in helping the community. I've run mapping
parties, I've made a video tutorial series, I'm active on the newbies
list and (less) active on help.osm.org, and can often be found helping
users on our IRC channel.

Importers, sadly, rarely ask as many questions as new mappers.

>> 2. We have no current technology to keep an import synced with another
>> data source, so the data goes stale quickly.
> Also true for manual edits.  Why is this considered just an import issue?

For imports, this is often touted as a benefit- that's why the lack of
an actual implementation is important to point out.

>> 3. Many imports are of datasets we don't want in OSM and we have
>> little/no mechanism outside community vetting to discover that until
>> it's too late.
> Any user can decide that they want to include "X" into the dataset imported
> or manually mapped.

If a user manually surveys data, there is an assumption of timeliness
and accuracy of that survey. That's not the case with imported data,
despite oftentimes being stamped "official".

>> 4. Most imports are bad, and most imports are left in OSM, which means
>> we have tons of garbage in OSM. Garbage in OSM makes the map harder to
>> work with for mappers, for tool-makers and for users.
> see all above

I've spent certainly well over a hundred hours cleaning up imported
data in OSM. I've never spent close to that dealing with new user

- Serge

More information about the Talk-us mailing list