[OSM-talk] [OSM-legal-talk] Are we strict enough with imports ?

Frederik Ramm frederik at remote.org
Thu Feb 11 12:24:12 GMT 2010


(I'm hijacking this thread which Nic started about legalities of imports 
on legal-talk, and moving over to talk)

Nic Roets wrote:
> My suggestion is that we should have a fixed, but simple procedure for
> users who import data:

I think that every import should start with a deliberation on whether to 
import *at all*.

Currently, I have the impression that many people are very trigger-happy 
when it comes to importing data. I believe that is running the risk of 
making OSM into one giant data rubbish dump.

The old-style GIS community is currently working on several projects 
that collect what they call "metadata" - basically, because they know 
that there are so many different people with so many different data 
sets, they are working on ways to describe these datasets in a way that 
hopefully enables intelligent clients to present data retrieved from all 
of them as one coherent data set.

This is of course extremely difficult and introduces many problems that 
one does not have when using just one huge database instead of thousands 
of different databases. But since many datasets are not static, you 
cannot simply grab them and pour them into one large database and be happy.

What does this mean for our data imports?

Data that is externally "owned" and maintained should not be imported, 
with the following exceptions:

* if the data is so important for us (usu. as the foundation for other 
crowdsourced stuff) that we'd rather have and outdated version of it in 
OSM than nothing at all;
* if we are confident that we, the OSM community, will do a better, more 
reliable, more thorough, and more timely job in updating the information 
than the original owner (this includes cases where the original owner 
has ceased maintenance);
* if he are confident that we can easily synchronize our database with 
any updates made by the original owner to his data set.

In all other cases it would be *much* more desirable to establish better 
mechanisms of merging OSM data with that other data in preparation for 
map drawing etc., rather than pulling it all in and having it rot.

I would very much like to develop a kind of "litmus test" for imports, 
and get the message across that not every import is a good import (even 
if legally spotless). Today, even newcomers to OSM sometimes seem 
hell-bent on importing large quantities of data just because they can. I 
would like to remind people that OSM has a very lively culture of 
surveying data - and I'd rather have 1 sq km surveyed by a newbie than 
100 sq km imported.


More information about the talk mailing list