[OSM-dev] data redaction api

[Matt Amos]

unfortunately, sometimes data finds its way into OSM which might be
copyrighted. the data working group was set up to help the community
deal with these incidents when usual methods of community arbitration
fail.

the usual course of action when copyright violation is strongly
suspected is to revert those changes to the database. however, when
data is reverted using the normal API the previous versions of
reverted elements are still available in the history. while this is
fine for vandalism, it's not so great for copyrighted stuff - it'll
still be accessible via the API.

one solution would be to delete the old versions from the database by
hand; but this isn't very scalable as only DB admins can do it.

the solution wikipedia uses is to keep the (potential) copyrighted
works in its database, but flag them as hidden (or interdicted,
banned, redacted - whatever your preferred terminology). this means it
can't be accessed via their web interface by normal users. and has the
further advantage that, if the data is later shown to have been
legally imported, the revert can be reverted and the data be made
visible again.

if this functionality is suitably general-purpose, it could also be
used during the re-licensing to hide the data of users who haven't
agreed to the ODbL. leaving arguments about the suitability of the
ODbL to one side (or legal-talk, as it's otherwise known), this has
the huge benefit of leaving the door open for users to agree at a
later point in time (maybe to a later version of the license, or when
they realise the ODbL is actually good) and to "recover" their data
without needing to look through old planet files or database backups.

these are just some of the possible use cases for a "data redaction
api". the question is; what would such an API look like, and how would
it be implemented without serious performance loss?

looking forward to hearing everyone's ideas,

cheers,

matt