[Talk-GB] Re-mapping : Are these two ways both valid ?

[Bogus Zaba]

On 20/01/12 15:58, woll wrote:
> I don't really have any LEGAL opinion/knowledge to answer your question with,
> but here is my experience:
>
> When I first started remapping, I took your option 1 (delete the feature and
> re-create it) because I felt that that option would ensure that no non-odbl
> tainting could happen. I only added tags that I could get independently from
> odbl-compatible sources, or from existing knowledge (so some tags may have
> not been 'transferred' onto the new version). To 'transfer' the maximum
> number of tags, you need to look at the history in detail, to see which tags
> are from users who have agreed to the CTs (if you can't 'transfer' the tags
> from your own knowledge/odbl-compatible sources.
>
> To find non-odbl tainted data, I used:
> http://cleanmap.poole.ch/
> the licence layer on the OSMI at tools.geofabrik.de/osmi
> and the licence highlighting of Potlatch.
>
> As I progressed, I started do a stricter version of your option 2, by
> investigating the features in more detail, mainly using the info provided
> by:
> http://osm.mapki.com/history/
> that shows which tags/features are non-odbl tainted.
>
> My stricter criteria for option 2 is  "NO data remains from non-agreers"
> (not your "enough has been done by other licence-accepting contributors that
> the feature no longer belongs to the original contributor").
>
> In a lot of the cases where I was working, I found various situations like:
> a) All the original nodes of the feature (from users not agreeing to the
> CTs) had been deleted (and new ones added)
> b) The only tags remaining from users who have not agreed to the CTs were
> factual tags like the name of the road or the number of the road, that I
> could easily get from odbl-compatible sources/knowledge.
> c) All the original nodes had been moved by agreeing users (presumably by
> aligning to aerial imagery/more accurate GPS tracks)
> d) The only thing remaining from non-agreers was a small percentage of the
> nodes in a way, and those nodes were not very accurately positioned.
>
> When I found that ALL the non-odbl data had actually been replaced, then I
> added "odbl=clean" (this last step might not be necessary in some cases,
> because the tools could probably be extended to work this out for themselves
> if they looked at the history in detail, but I did it to be safe).
>
> When I found that the only non-odbl data remaining, could be got from
> odbl-compatible sources/knowledge, I re-added it, and added "odbl=clean"
> (the last bit is necessary, because even though I deleted and re-added the
> tags, the tools can't tell that I actually did that (because the tags are
> exactly the same).
>
> If there were any non-odbl tags remaining that I couldn't get from
> odbl-compatible sources/knowledge then I deleted them. This was very rare
> for the features I was looking at, and was only 'irrelevant' things (e.g.
> not very useful notes). I probably didn't need to do this, as for most
> situations, I expect that the final process that will transfer the data
> would automatically delete these tags (it wouldn't delete a way just because
> there was just 1 non-odbl tag remaining, it would just delete that tag and
> keep everything else. however adding the "odbl=clean" tag means that the
> existing visualisation tools will then display things correctly).
>
> If the only non-odbl data remaining was a small percentage of the nodes in a
> way, I deleted/replaced them with more accurate ones based on more accurate
> GPS tracks or imagery (and normally took the opportunity to add more nodes
> for more accuracy).
>
> So, option 2 is not really "less work" - it actually takes quite a long time
> to look at the history in detail (and then drilling down to look at all the
> nodes of a way).
> Option 1 is not necessarily "more work" either: Unless you can easily get
> all of the tags from odbl-compatible sources/knowledge (or you don't care if
> you don't 'transfer' all of the existing tags) then you also need to look at
> the history in detail.
>
> To specifically answer your questions:
> Your option 1 is obviously "legal/ethical" (you're deleting everything and
> recreating it)
> I personally would not be happy doing your option 2 (which would vary on
> everyone's different interpretation of "enough"). The re-mapping I have done
> is stricter than that and ensures that there is no tainting remaining (but
> that means that option 2 is not necessarily "less work" than option 1).
>
> --
> View this message in context: http://gis.638310.n2.nabble.com/Re-mapping-Are-these-two-ways-both-valid-tp7205346p7208034.html
> Sent from the Great Britain mailing list archive at Nabble.com.
>
> _______________________________________________
> Talk-GB mailing list
> Talk-GB at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/talk-gb
>
Thanks very much for this comprehensive answer. I can see that I was 
being a bit naive in my approach to option (2), although locally some of 
the history does not appear to be all that complex so checking it 
thoroughly might not be as big a job as it would be for data that has 
been touched by the hands of many, many contributors.

One remaining question is regarding the "strength" of any odbl=clean 
tags which are applied. Is the existence of such a tag going to be taken 
as evidence that the data so tagged is indeed clean and usable, or will 
further tests be applied?

Thanks for your help.