[OSM-legal-talk] viral attribution and ODbL

[TimSC]

Hi all,

I am back to trying to get my head around ODbL. I am now wondering about 
attribution and the viral nature of it. Apologies if this has been 
raised before. Many licenses have a term stating the copyright notice 
must be preserved (ignoring for a moment that copyright is probably not 
approprate for databases). Examples include the X11 license and the 
CC-BY license (term 4b in 
http://creativecommons.org/licenses/by/3.0/legalcode). From memory, 
CC-BY-SA also has this condition, but I am not certain. Ok, so we might 
create a produced work and release it under the public domain. I could 
foresee a scenario:

1) Create a produced work under ODbL term 4.3 with proper attribution
2) Release produced work as public domain with proper attribution
3) Strip off legal notices and attribution (which I think is allowed, 
almost by definition, for public domain works)
4) Republish as public domain or any other license, without attribution

My question: where is the term that copyright notices must be preserved 
done the chain of derived works? ODbL term 4.3 only protects us as far 
as step 1 in the above example. And if we must insist on attribution 
being retained, are we saying we can't release ODbL produced works into 
the public domain?

The use case touches on this issue but mainly with respect to trying to 
reverse engineer the database. I think attribution is a separate issue. 
The comment in the use case document pretty much implies that this could 
be an issue.  
http://wiki.openstreetmap.org/wiki/Open_Data_License/Use_Cases#Using_OSM_data_in_a_raster_map_for_a_book.2C_newsletter.2C_website.2C_blog_or_similar_work

Second issue, which is probably the flip side of the same coin: people 
might be inclined to use works that use some sort of attribution license 
and incorperate them into OSM (this almost certainly has already 
happened, OS opendata, etc). The attribution must be included in any 
derived works. Now this seems incompatible with the contributor terms, 
which grants OSMF an unlimited license. So, I can't add any "viral 
attribution" data via the contributor terms, as OSMF might one day try 
to change its attribution terms, since it is not bound to only use ODbL. 
It would seem to be that the contributor terms would at least put the 
viral attribution condition on the OSMF. The worst case scenario is the 
contributor terms cannot accept any data with an attribution condition. 
Hopefully that is not the case! Is that interpretation any way valid, 
interesting, cross eyed? If the answer is already out there, just link 
to it. Thanks!

TimSC