[OSM-talk] OpenStreetMap License Change Phase 3 Pre-Announcement

[Frederik Ramm]

Hi,

On 04/16/2011 07:47 PM, Kevin Peat wrote:
> Such as is it the LWG's intention to make the
> license/ct's compatible with OS Opendata? If it isn't then all those
> people currently tracing thousands of roads a week in the UK might as
> well take a break and get some fresh air.

If people are indeed doing that then I would *definitely* suggest the 
fresh air option, no matter what we intend to do license-wise; see 
recent imports discussion on talk-gb ("Adding a further 250,000 roads 
quickly using a Bot").

> After so many years, someone must surely have given at least a bit of
> thought to how removing incompatible/un-relicensed contributions might
> be handled?

Yes, certainly. As a non-LWG-member but avid reader of legal talk, and 
someone who can add two and two together, I can say the following with 
some certainty:

* Data that has been created/modified only by people who agree to the 
license change will be kept unchanged. If someone should later complain 
nonetheless (e.g. "I have created and not relicensed A road and B road, 
and this pub which is at the intersection is definitely a derived work, 
I demand it must be removed") then such data might be removed.

* A mechanism will be devised to determine the splitting and merging of 
ways and take that into account, i.e. if a way has been created by 
someone who doesn't agree, but later split in two by someone else, then 
methods will be found to make sure the partial way newly created by the 
split will not count as "created/modified only by people who agree".

* Data that has been created/modified only by people who do not agree to 
the license change will not be kept in the new data base. If other data 
depends on this then that other data must be modified accordingly (e.g. 
a node might have to be removed from a way).

* There will be some mechanism to "remind" us that something is missing. 
It is totally unclear what form this will take; it might be a separate 
database that says "whithin this rectangle, 10 streets have been 
removed" (without saying what they are and where they were so as not to 
infringe the copyright of whoever did not relicense those streets), or 
it might be a "note" tag on a route relation saying "as part of the 
license change, 5 members of this route had to be removed" or so.

* There will be mechanism that show us these things *long before* the 
license change actually happens so that we have a chance to go there and 
resurvey the area, manually deleting the not-relicensed objects in that 
area. (In fact this is already starting to happen in some places in 
Germany, where the work of known "objectors" is being replaced by new 
data - premature action, I think, because people can still change their 
minds.)

* Data that has been created/modified by some people who do agree and 
others wo don't will have to be scrutinized. Some decisions can perhaps 
be made in an automated or semi-automated process; for example, assume 
that someone has removed the "created_by" tags of 100.000 nodes in one 
go - this is certainly not an act that warrants any copyright, so if 
that person is the only one in the history of the object to have not 
agreed then that will simply be ignored. Other cases will probably be 
more complex, and it is very likely that we will want to involve the 
community, i.e. there will be ways to "flag" objects with an unclear 
licensing status, there will be general guidelines issued by OSMF, and 
mappers will then be asked to decide for themselves wheter something can 
be copied or not (just like mappers today need to decide whether a 
source can be used or not). There will be likely some form of mechanism 
for recording such decisions, e.g. "mapper <name> decided on <date> that 
this object can be relicensed in spite of its unclear licensing status 
for the following reason: ..."

* It is likely that data from people who have explicitly said "no I 
don't relicense" will be treated differently from those who simply don't 
say anything. I could imagine - pure speculation on my part though! - a 
scene like this: The community in city X comes together and discusses 
prolific mapper Y, who hasn't been to the pub meet for the last year and 
hasn't made a decision regarding the license change. They don't know 
what happened to him but some mappers remember that he always said that 
he doesn't give a damn for intellectual property rights and he would 
prefer OSM to be PD. The mappers then decide to continue using his data 
in the relicensed database even though they don't have an explicit OK, 
knowing that if mapper Y should show up and say no they would have to 
remove his data later. This is a gamble that OSMF will certainly not 
make centrally but it might be possible locally.

* In general, just because an object has once in its history been 
touched by someone who doesn't agree to the new license doesn't mean it 
cannot be kept; just the information added in that one step cannot be kept.

When it comes not to individual mappers who decide against relicensing 
but to organisations who have made data available, and where we cannot 
secure their agreement to the CT, then when it is a case where data has 
been imported wholesale under one account, we *might* make an exception 
and say "ok we keep this data although this account hasn't signed the 
CT; we know that company X is ok with ODbL and if we ever change again 
we'll have to ask them". On the other hand, when it is a case where many 
individuals have added data using that source, then we will have to 
treat the data contributed by those individuals just as we would treat 
data contributed by people who don't relicense. If the individuals in 
question have either used proper source tags (PLEASE, PLEASE, everyone, 
do...) or they can pinpoint the changesets in question, then we can 
separate those changes (we might, technically, re-assign these changes 
to a different account that we create specially for that) and treat only 
those changes as un-relicensable; if on the other hand the user says 
"I've been using <source> for as long as I can think and it could be in 
all of my changesets" then we'll have to treat all contributions by that 
person as un-relicensable.


Most of what I wrote here hasn't been formally said by anyone in OSMF, 
but OSMF haven't fully thought this through either, and the above is 
simply the logical course to take given all the conditions and plans 
that *have* been discussed.

Bye
Frederik