Mike, it's great that you're looking at this. I think three years ago if you'd offered an upgrade path within CC then OSM would have snapped your hands off. But the OSM situation has moved on a lot in these three years and I hope you'll take some comments from a friendly OSM licence activist.
Time is not on your side. OSM's change to ODBL is proposed to take place by 1st April. If it happens (and if it doesn't it'll be the fault of the Contributor Terms, not ODBL itself) it will, in practice, make upgrading much harder. I know the CTs have an upgrade clause, but the community will not stand for this whole argument all over again. This time round there's a sense of "we didn't know what we were doing when our founders chose a creative works licence five years ago". There won't be that excuse next time.
So you need to issue a public statement of intent as soon as possible, and it needs to cover the major areas - and cover them in a way acceptable to a data project like OSM. I'll go through them here.
Firstly, how the data is protected. You need a clear statement of how CC 4.0 will use the sui generis EU database right as well as, much harder, the Feist vs Rural problem. I accept that you won't use contract under any circumstances, and that's fine, but you do need to say how you would tackle this - whether it's SC-style Community Norms or whatever.
Please don't be confused by talk of a "computer cartography licence" in this thread; it's a distraction. No-one is seriously arguing that the cartography (the art) can't be covered by a current CC licence: it's the one thing that unequivocally is. But OSM is a data project, not a cartography project. We want to protect our data dump (planet.osm), not the map tiles which are a convenient demonstration of the project.
In particular, the challenge is the part of our data that correlates to Feist vs Rural: big networks of street geometries and their basic, factual properties (name, road type, turn restrictions etc.). It's the most commercially valuable part of OSM, the most work to gather, and the least likely to be protected.
Secondly, attribution. I believe you've gone a long way towards this in CC 2.5 and subsequent but you need to communicate this to the OSM community (and the community, as you can tell from this comment thread, is not always perfectly aligned with the Foundation). People need to understand CC's mechanisms for attributing thousands of contributors (or not!), and for preserving attribution for imported data from CC-BY sources or similar - for example, the UK Ordnance Survey OpenData.
Thirdly, the Produced Work issue. This is essentially derivative/collective (or adaptation/collection to use your newer language) as applied to data. There are two related problems here.
a) Current CC-BY-SA does not preserve access to the source data. If someone produces a map from OSM data plus significant additional data of their own, and publishes that map under CC-BY-SA as required, OSM does not get access to the additional data. Rather, we are required to 'reverse engineer' it from the produced map, which at certain scales and renderings may not be possible at all. This is obviously a severe problem for data projects.
b) Inconsistent scope. Because current CC-BY-SA is a creative works licence, its copyleft 'infects' classic creative works made with OSM data, but not (say) software made with OSM data. This leads to some bizarre situations: a highly artistic, cartographic printed map will be caught within SA, as it will if it's served as a JPEG to a web browser. But if you deliver the same map data, cartographic styling, and map display applet to a browser as three separate components, then combine them on the browser, the cartography will not be caught within SA. Bad luck, print cartographers!
Again, as OSM is a data project, this means that the stuff we don't want (artistic cartography) is caught within SA although we have no use for it. This is a serious disbenefit for OSM: above all, we want people to use our data. Much as we don't require software using OSM data to be GPL-licensed, we shouldn't require art using it to be SA-licensed.
ODBL's concept of a Produced Work is a neat solution. You have the opportunity to do something similar in CC by using the adaptation/collection distinction. In other words, adding to or augmenting CC-licensed *data* would be an adaptation. Incorporating it as part of another, non-data work is a collection.
Finally, over and above these three issues, the public statement needs to come across as "we're listening and we want to work with you". Earlier Science Commons-flavoured statements to OSM in favour of the public domain had an element of "we know what's good for you and it's nothing like what you've been doing". I accept they were well-intentioned but they had the effect of dissuading even the PD-minded people within OSM. I'm not trying to go over old ground here - simply offer a suggestion as to how you can go forward with OSM.