[OSM-talk] Announce: Beginning of Phase 4 of license change process
Eric Marsden
eric.marsden at free.fr
Thu Jun 16 08:17:30 BST 2011
>>>>> "rf" == Richard Fairhurst <richard at systemed.net> writes:
rf> Sorry, you've puzzled me a bit here.
rf> You state that it's better to cite "how much data would be deleted".
rf> However, that directly contradicts your previous paragraph, in which you
rf> quote, um, the number of users, not the amount of data.
rf>
rf> Reading odbl.de, although "60% of users" in Europe have accepted the new
rf> contributor terms, that actually equates to between 80% and 92% of nodes,
rf> and between 70% and 93% of ways. In North America, your "40% of users" is
rf> 54%-94% of nodes, and 66-85% of ways.[1]
Hello Richard,
It's quite simple: I object to the OSMF using what I consider to be
very misleading statistics in communication on the ODBL process.
Michael Collinson's message can be interpreted as saying that 0.2% of
users haven't accepted the new contributor terms. I point out that a
more reasonable way of presenting the data is that between 40 and 55%
(depending on the region) of users haven't accepted the new
contributor terms.
I then argue that the most important statistic in deciding whether to
go ahead with the big delete is how much data would be removed.
odbl.de indicates (for Europe) 80% of nodes, 70% of ways, 50% of
relations -- much lower for other areas such as Australia or the USA
--- are at version 1 with a user having accepted the CT. As you point
out, this is a lower bound on the amount of data that would be
retained, since objects with a version > 1 and only CT-accepting users
would also be retained.
There used to be a map at http://osm.informatik.uni-leipzig.de/map/
highlighting how much data would be retained, but it seems to have
disappeared.
On a related note concerning the process, I find it unreasonable for
OSMF to ask people to accept the new CT without having first decided
on a tolerability threshold on loss of data.
--
Eric Marsden
More information about the talk
mailing list