An example of the complications inherent in determining tainted ways

classic Classic list List threaded Threaded
14 messages Options
Reply | Threaded
Open this post in threaded view
|

An example of the complications inherent in determining tainted ways

Nathan Edgars II
I will look at a single suburban roadway: Westwood Boulevard in the
International Drive tourist area south of Orlando.
This started out as a TIGER way:
http://www.openstreetmap.org/browse/way/11197961/history
80n (orange) and kyrbyboy (red) have made some improvements to
alignment, but have apparently not changed the tags on the ways. More
recently, JuxTPosition (green) created a dual carriageway, including the
moving of at least some of the older nodes, and I split the ways to add
sidewalk tags and bus route relations.

What this means is that, as far as I know, the tags on the ways were all
added by green users. Some of the nodes have been created by orange or
red users, but most were later moved by green users.

Because of the splitting, out of the 28 ways that comprise Westwood,
only http://www.openstreetmap.org/browse/way/23166942 has tainted
history, even though all have orange or red users in their complete
history (going back to the ways they were split from). 55 of the 189
nodes are also tainted, although only one -
http://www.openstreetmap.org/browse/node/250413743 - is still in the
position an orange or red user placed it in.

So what here will be reverted by the OSMF? Obviously node 250413743
needs to be replaced by another node in the same general location. But
other than that, is everything tainted because it was split from a
tainted way? Or is nothing else tainted because no data from the orange
or red users remains? If the latter, do I need to do anything special to
ensure that the OSMF does not delete it? If the former, exactly what
needs to be remapped to prevent deletion?

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Frederik Ramm
Hi,

On 12/15/2011 02:58 AM, Nathan Edgars II wrote:
> So what here will be reverted by the OSMF? Obviously node 250413743
> needs to be replaced by another node in the same general location. But
> other than that, is everything tainted because it was split from a
> tainted way? Or is nothing else tainted because no data from the orange
> or red users remains? If the latter, do I need to do anything special to
> ensure that the OSMF does not delete it? If the former, exactly what
> needs to be remapped to prevent deletion?

There are several aspects to this.

One is the real legal situation (assuming that a legal "truth" exists -
most lawyers will probably laugh at the assumption).

The second is what OSMF believes the legal situation is, and what amount
of risk they are willing to take. (We can never be absolutely totally
clean because people might make absurd-sounding claims like "that road
is really a derived work of the pub I placed there..." or so.)

The third is what I believe OSMF to be likely to do, and what I
therefore display on the OSM Inspector layer. Of course the Inspector
layer is most useful if it resembles as closely as possible the future
OSMF decision.

It has been explained already but I'll repeat it - OSMF/LWG has not yet
decided what they will do with regards to the finer points of complex
object relicensing. This means that none of your questions above has an
answer. And OSMF is not going to decide this behind closed doors without
looking out; they'll take a cue or two from what we do. And they are not
going to decide it within the next few days either so don't hold your
breath.

Personally I believe that complex situations like the one you describe
above will have to be investigated by a community member - like you did
-, and that person should (if possible) take the necessary steps to
clean up the situation and then vouch for it (saying, effectively,
"these objects are OK, I've checked them, believe me").

OSMF could then concentrate on producing some advice for the community
members doing that kind of work, and making some spot checks to see if
thy do it with the diligence required of the job.

Ideally, those community members would not be the same people that
proclaim the use of "loop holes" on the mailing lists ;)

Bye
Frederik

--
Frederik Ramm  ##  eMail [hidden email]  ##  N49°00'09" E008°23'33"

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Nathan Edgars II
On 12/14/2011 9:45 PM, Frederik Ramm wrote:
> It has been explained already but I'll repeat it - OSMF/LWG has not yet
> decided what they will do with regards to the finer points of complex
> object relicensing. This means that none of your questions above has an
> answer. And OSMF is not going to decide this behind closed doors without
> looking out; they'll take a cue or two from what we do. And they are not
> going to decide it within the next few days either so don't hold your
> breath.

So why have people been recommending for months that we remap tainted
objects when we still don't know what needs to be remapped? This isn't a
rare case, but happens frequently across the U.S. Maybe it's different
in places where mappers have not been able to take advantage of road
network imports, but most tainted roads I have seen have started out
from TIGER.

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Frederik Ramm
Hi,

On 12/15/2011 04:11 AM, Nathan Edgars II wrote:
> So why have people been recommending for months that we remap tainted
> objects when we still don't know what needs to be remapped?

If you prefer to wait until the exact rules are laid out for you, that's
your choice. Personally I'd rather make a few educated guesses and get
to work now.

Bye
Frederik

--
Frederik Ramm  ##  eMail [hidden email]  ##  N49°00'09" E008°23'33"

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Nathan Edgars II
On 12/14/2011 10:25 PM, Frederik Ramm wrote:
> Hi,
>
> On 12/15/2011 04:11 AM, Nathan Edgars II wrote:
>> So why have people been recommending for months that we remap tainted
>> objects when we still don't know what needs to be remapped?
>
> If you prefer to wait until the exact rules are laid out for you, that's
> your choice.
Yes, I prefer only doing a make-work task once.

 > Personally I'd rather make a few educated guesses and get
> to work now.
By my educated reasoning, anything from one node to the entire road is
tainted, so it's a little hard to make a guess.

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Toby Murray-2
Yeah, a healthy chunk of the interstates in Kansas are the same way. I
didn't go quite as deep as Nathan but this way is a relevant example:
http://osm.mapki.com/history/way.php?id=33576021

User "moonwashed" created this way by splitting it from a TIGER way.
He made several more edits to it but the last 20 versions have been by
agreeing users (including both NE2 and myself) and while that page
doesn't show node position changes, I have verified that every single
node has been moved since moonwashed last touched it. So in my mind
there is no information left in that way that is attributable to the
declining user. I would have absolutely no misgivings doing a straight
copy/paste to replace that way with an identical duplicate. But I
would rather not do so out of respect to the other CT-accepting users
who have contributed to that object.

Saying that it is up to the community to decide individual objects is
nice but I don't think there is enough time for me to evaluate every
tainted object in Kansas before April 1 and there sure as hell isn't
enough of a community here to help me with such a thankless task.
There are a few mappers in the area but if I asked them to deal with
this kind of stuff, I'm pretty sure they would run away screaming. I
doubt I can expect much outside help either since pretty much everyone
is affected and will be working in their own area first.

And as long as there is no official word from the foundation about
exactly how this change will be technically executed, we can't really
proceed in a meaningful way anyway except from trying to contact
non-responsive users, which I am doing. So as much as I really don't
really care about the license and am happy to relicense under ODbL and
even think it might be a good move, I do have some serious doubts
about the ambiguity of the process this late in the process...

Toby



On Wed, Dec 14, 2011 at 9:31 PM, Nathan Edgars II <[hidden email]> wrote:

> On 12/14/2011 10:25 PM, Frederik Ramm wrote:
>>
>> Hi,
>>
>> On 12/15/2011 04:11 AM, Nathan Edgars II wrote:
>>>
>>> So why have people been recommending for months that we remap tainted
>>> objects when we still don't know what needs to be remapped?
>>
>>
>> If you prefer to wait until the exact rules are laid out for you, that's
>> your choice.
>
> Yes, I prefer only doing a make-work task once.
>
>
>> Personally I'd rather make a few educated guesses and get
>>
>> to work now.
>
> By my educated reasoning, anything from one node to the entire road is
> tainted, so it's a little hard to make a guess.
>
>
> _______________________________________________
> talk mailing list
> [hidden email]
> http://lists.openstreetmap.org/listinfo/talk

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

David Groom


----- Original Message -----
From: "Toby Murray" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, December 15, 2011 4:47 AM
Subject: Re: [OSM-talk] An example of the complications inherent in
determining tainted ways


> Yeah, a healthy chunk of the interstates in Kansas are the same way. I
> didn't go quite as deep as Nathan but this way is a relevant example:
> http://osm.mapki.com/history/way.php?id=33576021
>
> User "moonwashed" created this way by splitting it from a TIGER way.
> He made several more edits to it but the last 20 versions have been by
> agreeing users (including both NE2 and myself) and while that page
> doesn't show node position changes, I have verified that every single
> node has been moved since moonwashed last touched it.

But do you know what the source was for moving each node? As has been said
earlier, if each node was simply moved by a tiny amount away from the
position created by moonwashed, and the new position of the node was not
determined by reference to some other source (Bing or GPS maybe), then the
new nodes are derived form moonwashes edits


> So in my mind
> there is no information left in that way that is attributable to the
> declining user.

Not necessarily true.  You can only state that when you know for sure what
the basis was for moving each node

David

>I would have absolutely no misgivings doing a straight
> copy/paste to replace that way with an identical duplicate. But I
> would rather not do so out of respect to the other CT-accepting users
> who have contributed to that object.
>
> Saying that it is up to the community to decide individual objects is
> nice but I don't think there is enough time for me to evaluate every
> tainted object in Kansas before April 1 and there sure as hell isn't
> enough of a community here to help me with such a thankless task.
> There are a few mappers in the area but if I asked them to deal with
> this kind of stuff, I'm pretty sure they would run away screaming. I
> doubt I can expect much outside help either since pretty much everyone
> is affected and will be working in their own area first.
>
> And as long as there is no official word from the foundation about
> exactly how this change will be technically executed, we can't really
> proceed in a meaningful way anyway except from trying to contact
> non-responsive users, which I am doing. So as much as I really don't
> really care about the license and am happy to relicense under ODbL and
> even think it might be a good move, I do have some serious doubts
> about the ambiguity of the process this late in the process...
>
> Toby
>
>
>
> On Wed, Dec 14, 2011 at 9:31 PM, Nathan Edgars II <[hidden email]>
> wrote:
>> On 12/14/2011 10:25 PM, Frederik Ramm wrote:
>>>
>>> Hi,
>>>
>>> On 12/15/2011 04:11 AM, Nathan Edgars II wrote:
>>>>
>>>> So why have people been recommending for months that we remap tainted
>>>> objects when we still don't know what needs to be remapped?
>>>
>>>
>>> If you prefer to wait until the exact rules are laid out for you, that's
>>> your choice.
>>
>> Yes, I prefer only doing a make-work task once.
>>
>>
>>> Personally I'd rather make a few educated guesses and get
>>>
>>> to work now.
>>
>> By my educated reasoning, anything from one node to the entire road is
>> tainted, so it's a little hard to make a guess.
>>
>>
>> _______________________________________________
>> talk mailing list
>> [hidden email]
>> http://lists.openstreetmap.org/listinfo/talk
>
> _______________________________________________
> talk mailing list
> [hidden email]
> http://lists.openstreetmap.org/listinfo/talk
>



_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Jean-Marc Liotier
On 15/12/2011 12:40, David Groom wrote:

> ----- Original Message -----
> From: "Toby Murray" <[hidden email]>
> Sent: Thursday, December 15, 2011 4:47 AM
>> User "moonwashed" created this way by splitting it from a TIGER way.
>> He made several more edits to it but the last 20 versions have been by
>> agreeing users (including both NE2 and myself) and while that page
>> doesn't show node position changes, I have verified that every single
>> node has been moved since moonwashed last touched it.
>
> But do you know what the source was for moving each node? As has been
> said earlier, if each node was simply moved by a tiny amount away from
> the position created by moonwashed, and the new position of the node
> was not determined by reference to some other source (Bing or GPS
> maybe), then the new nodes are derived form moonwashes edits
But what if the source changes ? When I use high-resolution imagery to
improve areas formerly mapped from low-resolution imagery, I change the
source tag - i.e. from "Yahoo low resolution satellite" to "Microsoft
Bing satellite". Since my edit is correlated with a change of source,
shouldn't it be considered a break from being a derivative ?


_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

David Groom


----- Original Message -----
From: "Jean-Marc Liotier" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, December 15, 2011 11:59 AM
Subject: Re: [OSM-talk] An example of the complications inherent in
determining tainted ways


> On 15/12/2011 12:40, David Groom wrote:
>> ----- Original Message -----
>> From: "Toby Murray" <[hidden email]>
>> Sent: Thursday, December 15, 2011 4:47 AM
>>> User "moonwashed" created this way by splitting it from a TIGER way.
>>> He made several more edits to it but the last 20 versions have been by
>>> agreeing users (including both NE2 and myself) and while that page
>>> doesn't show node position changes, I have verified that every single
>>> node has been moved since moonwashed last touched it.
>>
>> But do you know what the source was for moving each node? As has been
>> said earlier, if each node was simply moved by a tiny amount away from
>> the position created by moonwashed, and the new position of the node was
>> not determined by reference to some other source (Bing or GPS maybe),
>> then the new nodes are derived form moonwashes edits
> But what if the source changes ? When I use high-resolution imagery to
> improve areas formerly mapped from low-resolution imagery, I change the
> source tag - i.e. from "Yahoo low resolution satellite" to "Microsoft Bing
> satellite". Since my edit is correlated with a change of source, shouldn't
> it be considered a break from being a derivative ?
>
Yes it should be considred a break, because in that case you know what the
source for moving the nodes was.

What I was pointing out is that you have to know the source used when moving
the nodes, before you can determine if the new position is derived from the
old one

David
> _______________________________________________
> talk mailing list
> [hidden email]
> http://lists.openstreetmap.org/listinfo/talk
>



_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Jean-Marc Liotier
On 15/12/2011 13:17, David Groom wrote:

> ----- Original Message -----
> From: "Jean-Marc Liotier" <[hidden email]>
> To: <[hidden email]>
> Sent: Thursday, December 15, 2011 11:59 AM
>> But what if the source changes ? When I use high-resolution imagery
>> to improve areas formerly mapped from low-resolution imagery, I
>> change the source tag - i.e. from "Yahoo low resolution satellite" to
>> "Microsoft Bing satellite". Since my edit is correlated with a change
>> of source, shouldn't it be considered a break from being a derivative ?
> Yes it should be considered a break, because in that case you know
> what the
> source for moving the nodes was.
Good. Now do the license change impact auditing tools currently take
that into account ? Should they only take the object's source tag into
account or also mention of a source in the changeset commit comment ?


_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
80n
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

80n
On Thu, Dec 15, 2011 at 12:30 PM, Jean-Marc Liotier <[hidden email]> wrote:
On 15/12/2011 13:17, David Groom wrote:
----- Original Message -----
From: "Jean-Marc Liotier" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, December 15, 2011 11:59 AM
But what if the source changes ? When I use high-resolution imagery to improve areas formerly mapped from low-resolution imagery, I change the source tag - i.e. from "Yahoo low resolution satellite" to "Microsoft Bing satellite". Since my edit is correlated with a change of source, shouldn't it be considered a break from being a derivative ?
Yes it should be considered a break, because in that case you know what the

source for moving the nodes was.
Good. Now do the license change impact auditing tools currently take that into account ? Should they only take the object's source tag into account or also mention of a source in the changeset commit comment ?

I think there may be a need to better understand how copyright works in this respect in the real world. 

The location of individual nodes probably has no copyright component, however the shape of a way probably does [1].  If several people have adjusted the shape of a way then they most likely all have joint ownership of the copyright of the whole of that way [2].

Joint ownership is an important principle to understand.  If someone edits a way then they are making a derivative of that way and inheriting *all* of the joint copyright ownerships.  Even if their changes are to remove the effect of a change by one of the previous contributors it does not, as far as I know, delete that contributors copyright.

If this is true, then the only way to disinfect a tainted way is to revert back to the version prior to the infection and applying subsequent changes to that version.  Simply negating changes does not delete copyright ownership because the ownership extends to the whole work.

Does anyone know of any precedents that show how copyright, once gained, can be deleted from a work?

80n


[1] Section 1 (b) (i) of http://membled.com/work/osm/Map_Project_Memo_public_FINAL.pdf

[2] Section 2a of  http://membled.com/work/osm/Map_Project_Memo_public_FINAL.pdf




_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Mikel Maron
Please continue any detailed discussion of this topic to legal-talk ... that's what it's for.

-Mikel & Moderators
 
== Mikel Maron ==
+14152835207 @mikel s:mikelmaron

From: 80n <[hidden email]>
To: Jean-Marc Liotier <[hidden email]>
Cc: [hidden email]
Sent: Thursday, December 15, 2011 8:11 AM
Subject: Re: [OSM-talk] An example of the complications inherent in determining tainted ways

On Thu, Dec 15, 2011 at 12:30 PM, Jean-Marc Liotier <[hidden email]> wrote:
On 15/12/2011 13:17, David Groom wrote:
----- Original Message -----
From: "Jean-Marc Liotier" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, December 15, 2011 11:59 AM
But what if the source changes ? When I use high-resolution imagery to improve areas formerly mapped from low-resolution imagery, I change the source tag - i.e. from "Yahoo low resolution satellite" to "Microsoft Bing satellite". Since my edit is correlated with a change of source, shouldn't it be considered a break from being a derivative ?
Yes it should be considered a break, because in that case you know what the

source for moving the nodes was.
Good. Now do the license change impact auditing tools currently take that into account ? Should they only take the object's source tag into account or also mention of a source in the changeset commit comment ?

I think there may be a need to better understand how copyright works in this respect in the real world. 

The location of individual nodes probably has no copyright component, however the shape of a way probably does [1].  If several people have adjusted the shape of a way then they most likely all have joint ownership of the copyright of the whole of that way [2].

Joint ownership is an important principle to understand.  If someone edits a way then they are making a derivative of that way and inheriting *all* of the joint copyright ownerships.  Even if their changes are to remove the effect of a change by one of the previous contributors it does not, as far as I know, delete that contributors copyright.

If this is true, then the only way to disinfect a tainted way is to revert back to the version prior to the infection and applying subsequent changes to that version.  Simply negating changes does not delete copyright ownership because the ownership extends to the whole work.

Does anyone know of any precedents that show how copyright, once gained, can be deleted from a work?

80n


[1] Section 1 (b) (i) of http://membled.com/work/osm/Map_Project_Memo_public_FINAL.pdf

[2] Section 2a of  http://membled.com/work/osm/Map_Project_Memo_public_FINAL.pdf




_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk



_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Frederik Ramm
In reply to this post by 80n
Hi,

On 12/15/2011 02:11 PM, 80n wrote:
> Joint ownership is an important principle to understand.  If someone
> edits a way then they are making a derivative of that way and inheriting
> *all* of the joint copyright ownerships.

Provided that a way is a work - maybe it isn't; maybe the whole of OSM
is "the work"?

> Even if their changes are to
> remove the effect of a change by one of the previous contributors it
> does not, as far as I know, delete that contributors copyright.

In some national versions of "joint authorship", while the joint authors
all have a share in the copyright, they do not have the power to veto
the use (and sublicensing) of the work by the other authors. This is an
important principle to understand.

> If this is true, then the only way to disinfect a tainted way is to
> revert back to the version prior to the infection and applying
> subsequent changes to that version.  Simply negating changes does not
> delete copyright ownership because the ownership extends to the whole work.

It sounds like an utterly stupid thing to do but if we now re-set
objects to an earlier state by negating changes and later somebody finds
out that we would have had to follow your above procedure instead, then
that can still be done - automatically. So I'd not waste much thought on
this right now; we can cross that bridge when we come to it.

Bye
Frederik

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk
Reply | Threaded
Open this post in threaded view
|

Re: An example of the complications inherent in determining tainted ways

Nathan Edgars II
In reply to this post by Mikel Maron
On 12/15/2011 8:21 AM, Mikel Maron wrote:
> Please continue any detailed discussion of this topic to legal-talk ...
> that's what it's for.

The question is not what's legally true, but what conditions the OSMF
will require an object to satisfy to not be reverted. So it actually
belongs on osmf-tainting-policy-talk, but there is no such list.

_______________________________________________
talk mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/talk