[OSM-talk] Critical Mass for license change-over

andrzej zaborowski balrogg at gmail.com
Sat Jan 28 06:16:43 GMT 2012


Hi,

On 27 January 2012 21:52, Toby Murray <toby.murray en gmail.com> wrote:
> Specific questions that need an answer:
> 1) As Nathan pointed out, nothing currently looks at way
> splitting/combining.

While I think it is important to implement the detection of
split/combined ways in the final algorithm, I don't think it changes
the statistics Mike Collinson quoted by a lot because splits and
merges by CT-accepting users will even out with those by non-accepting
users.

> ...
> 3) will the odbl=clean tag be respected? We are closing in on 15,000
> uses of this tag.

The biggest issue I see with this tag is that people are setting it
based the CT-acceptance of users in the object history.  But really
(and this is confirmed in various LWG minutes) CT-acceptance doesn't
imply ODbL compatibility.  The "odbl=clean" tag on the other hand does
seem to imply ODbL-compatibility.  So really you're only entitled to
set odbl=clean on an object if all the information that remains is
either:

 * your own work, or,
 * from sources that you know are ODbL compatible, like survey by
someone you know or import from a Public Domain source.  There's no
legal basis to assume every user showing green in the license tools is
ODbL "clean" and if you make that assumption you're putting OSMF in an
unclean situation.

I think this is the biggest issue before LWG.  To make the switch to
ODbL real, they need to come up with a corrected version of
Contributor Terms and start asking mappers to accept it ASAP.  (the
other reason the majority of contracts firmed between contributors and
OSMF may be considered invalid is that the users who clicked "Accept"
until December were shown an incomplete text of the CT due to a bug)

---
BTW here's my answer to Mike Collinson's question I wrote in March 2011:

"I think here you have to use a sort of a logarithmic scale
and I hope the license change working group is going to use that scale
when/if they're deciding whether the moment is right to remove data
from the editable database.

7% or 45% or 62% are all insignificantly small if you think of the
amount of map data that remains incompatible.  Even 98% is
insignificantly small if this means that 2% of the userbase's data is
going to be dropped.  For the license change to not leave the project
dead, the number needs to be really close to 100% of the user
collected (i.e. not imported) data.  1% of a couple hunderds GB is
really a lot of data, 0.1% is still probably more than some of the
individual country extracts, and then 0.01% is probably an amount that
we could afford losing if everyone put a lot of effort in fixing the
breakage.
"

Cheers



More information about the talk mailing list