Re: [OSM-talk] License plan - what data would need deleting
You have discussed some elaborate plans about what data from a non-relicensing contributor would have to be deleted and what would have to be kept. In the worst case, in the event of a dispute, do you really fancy trying to convince a court of law that the elaborate heuristics you applied are sufficient to make the map completely independent of the work of the users who said 'no'? The only sound rule that can be sure to stand up in court is to delete all data from the contributors who didn't give explicit permission, and all data that depends on it. Period. You may think this is unnecessarily paranoid. Indeed it is: but if the relicensing exercise doesn't put the project on a legally unassailable footing, it is not worth doing. At the moment we can say with certainty that 100% of the contributors have clicked 'yes' to an agreement to distribute their changes under CC-BY-SA. Any legal niceties tidied up by a move to a different licence are good to have, all other things being equal, but are hugely outweighed if the data becomes a questionable mishmash of contributions that have agreement, and those that don't have agreement but pass some odd set of rules we invented ourselves to convince ourselves that we didn't need to get permission. -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
On Wed, Mar 4, 2009 at 11:18 AM, Ed Avis wrote: > You have discussed some elaborate plans about what data from a > non-relicensing > contributor would have to be deleted and what would have to be kept. > > In the worst case, in the event of a dispute, do you really fancy trying to > convince a court of law that the elaborate heuristics you applied are > sufficient > to make the map completely independent of the work of the users who said > 'no'? > > The only sound rule that can be sure to stand up in court is to delete all > data > from the contributors who didn't give explicit permission, and all data > that > depends on it. Period. > > You may think this is unnecessarily paranoid. Indeed it is: but if the > relicensing exercise doesn't put the project on a legally unassailable > footing, > it is not worth doing. At the moment we can say with certainty that 100% > of the > contributors have clicked 'yes' to an agreement to distribute their changes > under > CC-BY-SA. Any legal niceties tidied up by a move to a different licence > are good > to have, all other things being equal, but are hugely outweighed if the > data > becomes a questionable mishmash of contributions that have agreement, and > those > that don't have agreement but pass some odd set of rules we invented > ourselves to > convince ourselves that we didn't need to get permission. > I believe this is a wise approach. OSM is traditionally very conservative about using any data not from a know clean source. On the grand scale its relatively easy to capture map data, the value of a clean database far outweighs the risks associated with infringing anyone's copyright. We should apply the same degree of conserativism to our CC-BY-SA licensed data as we would to any other copyrighted data. Perhaps we are thinking about this all wrong. If we considered the ODbL to be a license fork of the project (albeit a friendly from the inside fork) then it makes it much easier to think about how all this should happen. 80n > > -- > Ed Avis > > > ___ > talk mailing list > talk@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk > ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
On Wednesday 04 March 2009, Ed Avis wrote: > The only sound rule that can be sure to stand up in court is to > delete all data from the contributors who didn't give explicit > permission, and all data that depends on it. Period. I agree that the only legal sound way to do it is by removing all dependent data. But we can't even tell what data depends on data from contributors who didn't give permission... Suppose I split a way into two parts. The second part now gets uploaded as a completely new object, with nothing in its history pointing towards its origin. Or another example: I can align the outline of a forest to the road which I know is its boundary. So the forest is also a dependency, but not a single clue in the database the two might be related to each other. That's all becoming quite a minefield really. Ben ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
2009/3/4 80n <80n...@gmail.com>: > On Wed, Mar 4, 2009 at 11:18 AM, Ed Avis wrote: >> >> You have discussed some elaborate plans about what data from a >> non-relicensing >> contributor would have to be deleted and what would have to be kept. >> >> In the worst case, in the event of a dispute, do you really fancy trying >> to >> convince a court of law that the elaborate heuristics you applied are >> sufficient >> to make the map completely independent of the work of the users who said >> 'no'? >> >> The only sound rule that can be sure to stand up in court is to delete all >> data >> from the contributors who didn't give explicit permission, and all data >> that >> depends on it. Period. >> >> You may think this is unnecessarily paranoid. Indeed it is: but if the >> relicensing exercise doesn't put the project on a legally unassailable >> footing, >> it is not worth doing. At the moment we can say with certainty that 100% >> of the >> contributors have clicked 'yes' to an agreement to distribute their >> changes under >> CC-BY-SA. Any legal niceties tidied up by a move to a different licence >> are good >> to have, all other things being equal, but are hugely outweighed if the >> data >> becomes a questionable mishmash of contributions that have agreement, and >> those >> that don't have agreement but pass some odd set of rules we invented >> ourselves to >> convince ourselves that we didn't need to get permission. > > > I believe this is a wise approach. OSM is traditionally very conservative > about using any data not from a know clean source. On the grand scale its > relatively easy to capture map data, the value of a clean database far > outweighs the risks associated with infringing anyone's copyright. We > should apply the same degree of conserativism to our CC-BY-SA licensed data > as we would to any other copyrighted data. > > Perhaps we are thinking about this all wrong. If we considered the ODbL to > be a license fork of the project (albeit a friendly from the inside fork) > then it makes it much easier to think about how all this should happen. > I think the problem here is the statement, "delete all data from the contributors who didn't give explicit permission, and all data that depends on it. Period." If only it was that simple. There's two options: 1) Start again from the first point of time at which someone not agreeing to the switch contributed data. 2) Draw a pragmatic line somewhere to determine what constitutes a copyrightable derivation from CC-BY-SA data. Option 2 is what just about everybody is talking about. They're just putting the line in different places. So the question isn't ever really going to end in a Period. We're going to have to make a call, and that can be extremely conservative with large zones of reversion around every contaminated edit, or extremely aggressive with complex heuristics to determine "significant" edits, or any point in between. Most people seem to be aiming for middle ground with object based reversion only and extremely few heuristics (ie: a zero change edit doesn't count). Which makes some sense. But don't kid yourselves it's a simple A or B choice. Dave ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
On Wed, Mar 4, 2009 at 12:20 PM, Dave Stubbs wrote: > 2009/3/4 80n <80n...@gmail.com>: > > On Wed, Mar 4, 2009 at 11:18 AM, Ed Avis wrote: > >> > >> You have discussed some elaborate plans about what data from a > >> non-relicensing > >> contributor would have to be deleted and what would have to be kept. > >> > >> In the worst case, in the event of a dispute, do you really fancy trying > >> to > >> convince a court of law that the elaborate heuristics you applied are > >> sufficient > >> to make the map completely independent of the work of the users who said > >> 'no'? > >> > >> The only sound rule that can be sure to stand up in court is to delete > all > >> data > >> from the contributors who didn't give explicit permission, and all data > >> that > >> depends on it. Period. > >> > >> You may think this is unnecessarily paranoid. Indeed it is: but if the > >> relicensing exercise doesn't put the project on a legally unassailable > >> footing, > >> it is not worth doing. At the moment we can say with certainty that > 100% > >> of the > >> contributors have clicked 'yes' to an agreement to distribute their > >> changes under > >> CC-BY-SA. Any legal niceties tidied up by a move to a different licence > >> are good > >> to have, all other things being equal, but are hugely outweighed if the > >> data > >> becomes a questionable mishmash of contributions that have agreement, > and > >> those > >> that don't have agreement but pass some odd set of rules we invented > >> ourselves to > >> convince ourselves that we didn't need to get permission. > > > > > > I believe this is a wise approach. OSM is traditionally very > conservative > > about using any data not from a know clean source. On the grand scale > its > > relatively easy to capture map data, the value of a clean database far > > outweighs the risks associated with infringing anyone's copyright. We > > should apply the same degree of conserativism to our CC-BY-SA licensed > data > > as we would to any other copyrighted data. > > > > Perhaps we are thinking about this all wrong. If we considered the ODbL > to > > be a license fork of the project (albeit a friendly from the inside fork) > > then it makes it much easier to think about how all this should happen. > > > > I think the problem here is the statement, "delete all data > from the contributors who didn't give explicit permission, and all data > that > depends on it. Period." > > If only it was that simple. > > There's two options: > > 1) Start again from the first point of time at which someone not > agreeing to the switch contributed data. > > 2) Draw a pragmatic line somewhere to determine what constitutes a > copyrightable derivation from CC-BY-SA data. > > Option 2 is what just about everybody is talking about. They're just > putting the line in different places. > > So the question isn't ever really going to end in a Period. We're > going to have to make a call, and that can be extremely conservative > with large zones of reversion around every contaminated edit, or > extremely aggressive with complex heuristics to determine > "significant" edits, or any point in between. > Most people seem to be aiming for middle ground with object based > reversion only and extremely few heuristics (ie: a zero change edit > doesn't count). Which makes some sense. > But don't kid yourselves it's a simple A or B choice. > If someone were to fork the OSM database and try to make a PD version that only contained contribution from people who had self-declared their content as PD then we'd be right in demanding that they err on the side of caution. It's the same situation for ODbL. The fact that it's us doing this and not "them" is immaterial. 80n > > Dave > ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Dave Stubbs wrote: > But don't kid yourselves it's a simple A or B choice. Absolutely. Steve actually answers this in his (very good IMO) "Licence to kill" post. You can theoretically work out a complicated Boolean system of "is this derived from an ODbL refusenik's work?". You can read every bit of discussion about what "substantial" might mean in different jurisdictions, and write some clever fuzzy-matching software to reflect that. I think that's what people are talking about here. But as Steve points out, that's a programmer's answer, all very black-and-white. It doesn't actually work like that. What really makes the difference, in my very limited understanding (but hey, I'm a journalist not a programmer, limited understanding is a speciality :) ), is intent. Intent, and acting in good faith at all times. If we can demonstrate that we've taken reasonable precautions; that we have removed people's data on request (which, of course, we can do at any time); And for those who say "well, let's stick with the clean dataset we have now": We don't actually have a clean dataset. Nowhere near. We have material from Google Maps in there. We have material from the Ordnance Survey. We may even have entire countries which have been taken from a source not compatible with our current licence - see the discussion about some of the ex-USSR states. The reason we haven't been sued is exactly the same. Intent and good faith. Things like community pressure, the stuff in the FAQ, and the warning you get when you start Potlatch. The efforts we go to to gather our own data. That is real, hard proof. And that won't change - we should make real efforts, and we will, but clinical boolean precision is a distraction. (Ed asked how we'd "convince a court of law" - that's how. At the very least, if Paul The Disaffected Mapper doesn't want to go to ODbL, some of his stuff somehow remains in, and he says "ha, I'm going to sue", that _very_ instant a crowd of OSMers would go and survey the place in question to replace his data. You know what we're like. We like a challenge. :) ) cheers Richard -- View this message in context: http://www.nabble.com/License-plan-tp22245532p22329361.html Sent from the OpenStreetMap - General mailing list archive at Nabble.com. ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Hi, Richard Fairhurst wrote: > What really makes the difference, [...] > is intent. Intent, and acting in good faith at all times. Could we perhaps shred all this legalese then, be done with the license (which is, in effect, an attempt at codifying things in a manner you and Steve have just discounted), and instead write an one-page statement of intent that says how we'd like our data to be used and how not, and that's it? Bye Frederik ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Richard Fairhurst systemed.net> writes: >We don't actually have a clean dataset. Nowhere near. >The reason we haven't been sued is exactly the same. Intent and good faith. You are right. So what is the way of dealing with a relicensing that preserves the intent of the contributors and is done in good faith? I would argue that only a conservative approach is good enough - removing all the 'no' contributions and things that depend on them. A complex set of rules and hand-waving to see what we can get away with isn't really acting in good faith and doesn't really respect the intent of people who uploaded data expecting it to have a CC-BY-SA licence. Again I would suggest the 'Google Maps Test' for any proposed scheme. Suppose back in 2005 somebody made substantial changes that were blatantly copied from Google Maps. They have lain undetected until now and others have made improvements to that area of the map. But now the copying has been detected, how much data must be deleted? Another way to consider it is that one OSM contributor, having done most but not all of the work in a particular region, wants to make a proprietary, copyrighted map of the region based on OSM data. How much of the data contributed by other OSM mappers does he need to remove? What about if another contributor originally added a feature and then he improved it? And so on. It is very tempting to give ourselves special indulgences when considering all this. Surely it won't matter if we just do X; we can mix together different licences for a transition period; it would be unkind and inconvenient to throw away large chunks of data just because the person who happened to map the region first couldn't be contacted. After all, we're all conscientious people acting in good faith and we want to do the right thing - that's the main thing, surely? Acting in good faith means taking the most cautious approach, seeing things from the other side, and not treating the OSM project any differently from another entity which wants to take the contributed work and strip off CC-BY-SA. -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Ben Laenen writes: >Suppose I split a way into two parts. The second part now gets uploaded >as a completely new object, with nothing in its history pointing >towards its origin. Although the way is new, don't the nodes along it keep their identity? >Or another example: I can align the outline of a forest to the road >which I know is its boundary. Yes, indeed. If the road is exactly along the forest boundary then you can again look at the nodes, but if there is some separation between the two there is no clue in the database. Yet clearly, tracing a road from the boundary of a forest, when you don't have permission to distribute the forest data itself under the licence you want, is no different to tracing a road from the boundary of an object you saw on Google Maps and overlaid on your screen. So, sadly, I suppose an 'exclusion zone' would have to be imposed. -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Hi! Frederik Ramm wrote: >[..] > Could we perhaps shred all this legalese then [..] > and instead write an one-page statement of > intent that says how we'd like our data to be used and how not, and > that's it? I don't want to sound stupid or offensive, but - sarcastic or whatever - I absolutely like Frederiks idea. Sorry... - Andreas ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
On Wed, Mar 4, 2009 at 08:19, Ed Avis wrote: > Ben Laenen writes: > >>Suppose I split a way into two parts. The second part now gets uploaded >>as a completely new object, with nothing in its history pointing >>towards its origin. > > Although the way is new, don't the nodes along it keep their identity? True, but what if I create a node at the same place I split? For instance, if I take an existing way and split it in two places to add a bridge tag, you'd end up with something like this: ( 1 )--A( 2 )B( 3 )--C--( 4 ) Nodes 2 and 3 have no reference to the original user in their individual histories, so it would be impossible to know that they are derived from whoever did the original way. Similarly, neither Way B nor any of its nodes have any reference back. Or, what if I happen to move every node in a way and change every single tag, and then the original mapper opts out of the license? What appears in the database at the current time is entirely my work, even though the refusenik appears in the history. Someone upthread suggested reverting to the database as it stood just before the first contribution from someone who refuses the new license, and I'm beginning to fear that's the only way to guarantee a completely clean version. -- David J. Lynch djly...@gmail.com ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
On Wed, Mar 04, 2009 at 11:18:48AM +, Ed Avis wrote: > You have discussed some elaborate plans about what data from a non-relicensing > contributor would have to be deleted and what would have to be kept. Are you responding to my mail, or one earlier in the thread? I stated that everything should be reverted to before each incompatible change. > The only sound rule that can be sure to stand up in court is to delete all > data > from the contributors who didn't give explicit permission, and all data that > depends on it. Period. Yes, like this. If you are referring to my mail, I can only assume you are referring to the last example sequence of edits I gave, and you assume that C’s edits are dependent on B’s incompatible edits. C created a completely new way where there was none. C may or may not have seen B’s removal of A’s scribble, but regardless, his edit is a completely new work. Simon -- A complex system that works is invariably found to have evolved from a simple system that works.—John Gall signature.asc Description: Digital signature ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
OK, so lets assume that some data would have to be deleted (hopefully not lot of them, otherwise it would probably kill the project and spawn some forks with "complete" cc-by-sa data). Where there is the exact line between deleted and kept data is on another debate, but I wonder the way how the data would be deleted: - Would people know in advance what data are going to be deleted/reverted, so they can perhaps delete and redraw them themself? - If yes, how that would be done and how long before marking the data and actual deletion? - After deletion, would there be some marker like "there is something deleted from here"/"this way was reverted" for people to see, so they can improve the data again? - Would there be some "log" summarizing what data was/are going to be deleted from where after/before the deletion? Martin ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] License plan - what data would need deleting
Simon Ward bleah.co.uk> writes: >Are you responding to my mail, or one earlier in the thread? I stated >that everything should be reverted to before each incompatible change. I wanted to make the general point that while technically we can devise rules for deciding what changes are compatible and incompatible, there is a certain maximum level of complexity, and rules that generate long discussions on the mailing list are probably too complex to stand up in court. (Taking a pessimistic view.) -- Ed Avis ___ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk