[OSM-talk] [Osmf-talk] Attribution guideline status update

Christoph Hormann osm at imagico.de
Sun Sep 8 18:37:11 UTC 2019

On Sunday 08 September 2019, Simon Poole wrote:
> I think you are confusing potentially extractable information with
> actual data. For example satellite imagery may have a potentially
> high information content that could be with appropriate processing be
> turned in to data, but each image in itself is at most one datum.

I see - so you want to quantify by counting 'data objects' of some sort.  
I assume for the OSM side you want to go with the quantification of one 
OSM feature equals one countable object and a large lake multipolygon 
for example can count a few thousand?

You'd still loose by a huge margin in a map with contour line relief 
rendering of course.

And i would still hold the bet that i would be able to get the OSM 
fraction of any map below 50 percent without too much effort.

> Now waiting for the every image is a pixel database argument.

You are aware that most satellite image layers used in visualizations 
are produced from hundreds of thousands or even millions of individual 
images, assembled pretty much in the same way as a map rendering is 
assembled from multiple features.  It therefore seems your 'one datum' 
concept is somewhat fragile.

I see exactly one possible quantification of data fractions in a map 
that could not be easily circumvented.  That would be based on the 
number of human work hours that went into producing the data.  This is 
a rule i could support:  If more human work hours went into producing 
the non-OSM source data used in a map than in the OSM data used 
attribution that is hidden by default is acceptable.

Christoph Hormann

More information about the talk mailing list