On 27/01/2012 19:19, Nathan Edgars II wrote:
On 1/27/2012 12:44 PM, Michael Collinson wrote:
As the license change process evolved, concern was expressed that an
unacceptable amount of data might be lost from the current version of
the OSM database and consensus was reached that phase 5 [1] - the actual
license cut-over - should only happen when a "critical mass" was
achieved.
The question I ask you is, do you agree that we have reached critical
mass?
This cannot be answered until we know exactly what criteria will be
used for determining taintedness. The current tools (all? based on
WTFE) do not take into account splitting and merging of ways. If the
OSMF decides to be more diligent and dig into the history, we have a
big problem, since much of the remapping effort will have been based
on assumptions that are no longer true. (It's possible that this is a
much bigger issue in places like the U.S. where the initial street
data was imported, and thus splits and merges of initially-untainted
ways that contain untainted nodes are more common, so perhaps anyone
looking at this issue should concentrate on these areas.)
In short, we cannot know if the data loss is acceptable without
knowing what data will be lost. Will the OSMF follow the WTFE algorithm?
More than one person has expressed similar concern, I hope this
adequately responds to everyone.
I cannot as yet speak for the OSMF board. However, the LWG position is that:
Reasonable efforts should now be made to remove the contribution value
of non-agreeing mappers in a process that is both fair to non-agreers,
for whatever reason, and to those that are continuing to map and that
the process should be community-driven.
In other words, we (the LWG) should get out of the way unless we feel
that non-agreers contributions are not being dealt with fairly. The
resource we watch, and so far have seen no need to intervene on, is the
What is clean? wiki page [1].
"Reasonable efforts" includes something which is practically and
technically implementable. WTFE [2] is, as far as I am aware, the only
quantitative heuristic to have been developed and, again as far as I am
aware, follows the What is Clean? criteria.
I therefore we suggest that we now explicitly adopt the WTFE algorithm
as criteria for any final rebuild, provided that incremental improvement
can continue and also that folks can challenge any precepts on
legal-talk. That vastly reduces the uncertainty that several folks
express as we can use the figures it produces for the basis of this
discussion.
That leaves splits and merges as a potential uncertainty factor. My
personal opinion here, and I stress personal, is that we make no
adjustment for them and I'll be happy to discuss this further on
legal-talk I will however put the USA situation on the LWG agenda.
Mike
[1]
http://wiki.openstreetmap.org/wiki/Talk:Open_Data_License/What_is_clean%3F
[2] http://wiki.openstreetmap.org/wiki/WTFE
_______________________________________________
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk
