[Talk-ca] OSM New-York - Import de contours de batiments et adresses

Pierre Béland pierzenh at yahoo.fr
Mon Jan 6 18:14:47 UTC 2014


Richard

we are trying to build a community, to work with various governments to provide some data.

At the same time, we should bring a minimum of nuances in discussing in subjects like the imports.

Yes there are mappers that make errors, and yes we should help them to do better, we should take care to have good guidelines, good follow-up.

And this not only for imports. The same with field surveys and Aerial imagery.  If you use a GPS to locate a store, there is good chance that you will not place a commerce at the right place and even place it on the wrong side of the street.  And if you trace from Imagery in sandy areas dozens of paths going in all directions, this is not meaningful.

It is very difficult to discuss, when somebody comes with such a radical position as you present every time we discuss of imports. 

Let's hope that we can have  more constructive ways to discuss on this list in 2014.


 
Pierre 



________________________________
 De : Richard Weait <richard at weait.com>
À : Pierre Béland <pierzenh at yahoo.fr> 
Cc : Talk- CA Open Street Map <talk-ca at openstreetmap.org>; "diane.mercier at gmail.com" <diane.mercier at gmail.com> 
Envoyé le : Lundi 6 janvier 2014 12h49
Objet : Re: [Talk-ca] OSM New-York - Import de contours de batiments et adresses
 

On Mon, Jan 6, 2014 at 8:58 AM, Pierre Béland <pierzenh at yahoo.fr> wrote:
> Richard
>
> I dont think that we should advocate against import.

Then we differ.

I've been advocating for "better imports" with every import I've seen
since 2006.  While the tools have improved, the results for the most
part, just haven't.

>  Let's try in 2014 to
> be more positive with that and suggest ways to do it better.

You might continue to believe that "imports are just fine", I do not.
Imports are harmfull to OpenStreetMap and to the OpenStreetMap
community in all except extremely limited circumstances.

Invariably, when I add that "except in extremely limited
circumstances", the listener will presume that they are in fact the
exception.

Invariably, they aren't the exception.

They are well intentioned.  They are in love with data.  They want a
better OpenStreetMap.  And then they make an import of some sort and
cause harm to the data base and community that they then never clean
up because it is too much work.

The linked thread regarding the NYC building import discusses ways to

do it better.

The after action report on any decent effort at an import has
discussed ways to do better in the future.  Technically, essentially
every import has been better than the one before.  To date,
overwhelmingly, better is still just not good enough.

My recommendation is never to import.  "Import" should be a very dirty
word in OpenStreetMap.

By comparison, I think we should focus on doing the best mapping that
we can with our surveys, and with external resources that we have
permission to use.  Use external resources* by comparing each item
with all of the other existing resources, including imagery, existing
OpenStreetMap data, your survey, local knowledge, and curate the
external source before placing it into OpenStreetMap.

But never import.**

Yes. It's way slower.  Yes, it takes more time, and a more-experienced
mapper.  But it is what you owe to the project, the community and to
your reputation as a mapper.

To be clear, I love that external resources are becoming available to
OpenStreetMap in greater numbers.   I have every bit as much "data
love" for a new data set as the next mapper.  Dumping huge amounts of
un-curated data into the OpenStreetMap data base at one time is not
the way to use that data, or OpenStreetMap to best effect.

* Only the ones for which we have explicit permission to use.
** except in those extremely limited circumstances which don't apply here.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20140106/dfe0c53d/attachment.html>


More information about the Talk-ca mailing list