Re: [OSM-legal-talk] Implementing the licence change
Hi, On 20.01.2012 00:09, Frederik Ramm wrote: And changing a way's references from ABC to ACB is not a modification at all, because no reference is created and no reference is removed. We cannot say that there was a modification in regard to any of the references. No, the (relative) place of the reference in the list of references also counts. Changing the node list from 1,2,3 to 3,2,1 is a meaningful change. The point is that changing the list is a modification to the node *list*, not to a single reference. In other words, you can't actually distinguish between individual nodes without looking at the list at the same time. ... I know this is kind of a quibble, but I think it helps understand the criteria for rebuilding measures. So I think we can agree on this: A change to a way's node list is a meaningful change and therefore modifications to it by non-agreers must be rolled back. Above all, we must not forget to consider whether the creation or modification of a single reference, a single role - i.e. anything we say to be atomic - can possibly constitute a creative work. Some people have called for summarily force-relicensing the contribution of anyone who has added less than a certain amount of data. That's out of the question for me. Problem is, we're starting to get into the database realm. If you take the latest Harry Potter novel then no single word in it is copyrightable. But the combination of a significant portion of words is. Our fine-grained approach (i.e. let's simply try not to use *any* word from Harry Potter, that way we're sure that we won't infringe copyright) might be erring on the side of caution, but I'd prefer that over non-agreers raising a fuss after the license change because they spot something in there that isn't clean. I understand. I was not intending to reason like a single reference is not protectable, so the node list isn't either. Rather I'm trying to consolidate a basis for argument - e.g., as I have pointed out above, I prefer not to argue based on single references. cheers ant ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
Re: [OSM-legal-talk] Implementing the licence change
Hi, On 18.01.2012 23:49, Frederik Ramm wrote: They are not known. A mailing list has been created (the rebuild list) to discuss how exactly the database rebuild is going to happen, and in I didn't know about that list - I'll join it. terms of policy, LWG will have the ultimate decision. And they are asking for out input via the What is clean page. That page is not, and was not intended to be, a binding document - it might become one later. I assume that LWG will certainly value your help in improving that document. Thanks for clarifying the purpose of the What is clean page, because I wasn't sure whether I was entitled to edit it, not being an LWG member. IANAL. But I like to approach problems in a systematical manner. For example, I recently asked myself the question, What is a copyrightable object in OSM?. I think this is a fundamental question to answer if you discuss licence topics. It has often been said that computer geeks, of which I presume you are one, are not well suited to perform legal analyis. The lawyer's answer to Is a node copyrightable? will almost certainly be it depends. (On country, circumstances, ...) Sure. In OSM, our current answers are: Yes, we treat a node as copyrightable; If yes, what's copyrightable about it? Its position and tags, unless the tags have been created automatically. What's copyrightable about a way? The sequence of its nodes and its tags. Is the list of references to nodes copyrightable separately from the way's tags? Every single tag and every single node reference are a treated as copyrightable by us. Are references to nodes atomic? (I.e. Is a single reference copyrightable? Or is only the list as a whole?) Atomic. So moving a way is not considered a modification of the way, but of the individual nodes. And changing a way's references from ABC to ACB is not a modification at all, because no reference is created and no reference is removed. We cannot say that there was a modification in regard to any of the references. Next question, since according to your answers the approach is rather fine-grained, one might ask if single words within tags are copyrightable. What about roles of relation members, are they separated from the members' references? Above all, we must not forget to consider whether the creation or modification of a single reference, a single role - i.e. anything we say to be atomic - can possibly constitute a creative work. Considering that neither the definitions of what is clean and what is tainted nor the technical details of the implementation have yet been finalized, it seems unreasonable for me to remap. Thankfully, few other people think like you do. There may be edge cases, but I guess that whichever way these edge cases are decided, a significant portion of what is now considered tainted will always be tainted. And that stuff should be remapped *now*. I will certainly start remapping at some point. It's just that I don't feel confident about it at the moment, because there are so many unanswered questions. It's ok to discuss these things, but the approach I won't move a finger until I am told *exactly* what the rules are is not helpful. The rules might *never* be final - even when we do the rebuild according to the then-believed-final rules, it could happen that someone later points out an oversight, or a court decides something, forcing us to remove things we thought we could keep or vice versa. You can only ever go up to 80% certainty in these matters. Demanding more is not realistic. I'm not demanding. I just want to help raising the bar of certainty, in order to prevent us from overseeing something. cheers ant ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
Re: [OSM-legal-talk] Implementing the licence change
On 19 January 2012 21:48, ant antof...@gmail.com wrote: Hi, On 18.01.2012 23:49, Frederik Ramm wrote: They are not known. A mailing list has been created (the rebuild list) to discuss how exactly the database rebuild is going to happen, and in I didn't know about that list - I'll join it. terms of policy, LWG will have the ultimate decision. And they are asking for out input via the What is clean page. That page is not, and was not intended to be, a binding document - it might become one later. I assume that LWG will certainly value your help in improving that document. Thanks for clarifying the purpose of the What is clean page, because I wasn't sure whether I was entitled to edit it, not being an LWG member. IANAL. But I like to approach problems in a systematical manner. For example, I recently asked myself the question, „What is a copyrightable object in OSM?“. I think this is a fundamental question to answer if you discuss licence topics. It has often been said that computer geeks, of which I presume you are one, are not well suited to perform legal analyis. The lawyer's answer to Is a node copyrightable? will almost certainly be it depends. (On country, circumstances, ...) Sure. In OSM, our current answers are: Yes, we treat a node as copyrightable; If yes, what's copyrightable about it? Its position and tags, unless the tags have been created automatically. What's copyrightable about a way? The sequence of its nodes and its tags. Is the list of references to nodes copyrightable separately from the way's tags? Every single tag and every single node reference are a treated as copyrightable by us. Are references to nodes atomic? (I.e. Is a single reference copyrightable? Or is only the list as a whole?) Atomic. So moving a way is not considered a modification of the way, but of the individual nodes. And changing a way's references from ABC to ACB is not a modification at all, because no reference is created and no reference is removed. We cannot say that there was a modification in regard to any of the references. Next question, since according to your answers the approach is rather fine-grained, one might ask if single words within tags are copyrightable. What about roles of relation members, are they separated from the members' references? Above all, we must not forget to consider whether the creation or modification of a single reference, a single role - i.e. anything we say to be atomic - can possibly constitute a creative work. To be safe you cannot make any decision based on what rights a *single* instance of anything would have because the criteria will be applied in bulk. In practice you always have to consider what rights a database of such references could be protected by. Secondly it's known that in some instances in some countries creativity is not required for copyright to work. Thirdly in many countries there are other intellectual property rights that could play some roles. So the criteria have to be based on where you can say there's no content left from an incompatible edit, not whether it's uncreative. Or where something is part of a bulk edit that will be trivial to recreate if it's lost. Cheers ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
Re: [OSM-legal-talk] Implementing the licence change
Hi, On 01/19/2012 09:48 PM, ant wrote: So moving a way is not considered a modification of the way, but of the individual nodes. Yes. And changing a way's references from ABC to ACB is not a modification at all, because no reference is created and no reference is removed. We cannot say that there was a modification in regard to any of the references. No, the (relative) place of the reference in the list of references also counts. Changing the node list from 1,2,3 to 3,2,1 is a meaningful change. Next question, since according to your answers the approach is rather fine-grained, one might ask if single words within tags are copyrightable. Our current approach is to take a tag value as a whole. This is certainly not always correct. Also, let me remind you that we don't judge what is copyrightable and what isn't; we're trying to do something that is *reasonable* with regard to copyright. This involves a lot of judgment calls. What about roles of relation members, are they separated from the members' references? I'd treat them like a tag, so yes. Above all, we must not forget to consider whether the creation or modification of a single reference, a single role - i.e. anything we say to be atomic - can possibly constitute a creative work. Some people have called for summarily force-relicensing the contribution of anyone who has added less than a certain amount of data. Problem is, we're starting to get into the database realm. If you take the latest Harry Potter novel then no single word in it is copyrightable. But the combination of a significant portion of words is. Our fine-grained approach (i.e. let's simply try not to use *any* word from Harry Potter, that way we're sure that we won't infringe copyright) might be erring on the side of caution, but I'd prefer that over non-agreers raising a fuss after the license change because they spot something in there that isn't clean. I'm not demanding. I just want to help raising the bar of certainty, in order to prevent us from overseeing something. Well if you find certainty, be sure to inform us since we'll be very interested ;) Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
[OSM-legal-talk] Implementing the licence change
First of all I must say that I highly respect the work of everyone who has been actively involved in the licence change, including the LWG members, the writers of licence change inspection programs and everyone involved in discussions. I have been watching the process for more than two years and have ever since been a supporter of the change. However, especially since the switchover date has been announced and the phase of remapping has started, I have become more and more skeptical about the way things are going on. I want to discuss a couple of concerns I have. 1. The black box As far as I can see the details of the implementation of the licence change, i.e. of what is actually going to happen on April 1st, are not known - or at least not revealed. Correct me if I am wrong. Particularly, the wiki page „What is clean?“[1], which has been said to be the binding document, is in its current state not sufficient to serve as a reference for any measures regarding the cleaning of data: * The considerations in the section „Edge cases“ are only a random selection of cases that have been discussed. Neither conditional stetements like „if it can be seen not to influence the current version“ nor questions like „Can you copyright the state of something not being there?“ (rhetorical?) are helpful. The list somewhat lacks a systematic approach. * The „deletion paradox“ is, as it has been pointed out on the discussion page, no paradox at all (rather it depends on the strategy of cleaning). * The section „What taints data?“ repeats the above-mentioned list, but is differently (better) structured and different in content. Statements in this list, however, contradict, or supersede, previous statements („A tag modified by a non-agreeing mapper is tainted“, whereas: correcting a tagging typo is not tainted). Furthermore the list contains instructions, which should not be the case in a mere specification of what is clean. The clause saying that intermediate versions should be created during remapping (a) does not belong here and (b) is questionable, as it is based on assumptions regarding the implementation of switchover, which has not yet been decided upon. * There should be rationales explaining for each statement why it is so and not different. Basically I think that this document needs a rewrite that shall contain unambiguous statements preceded by precise definitions. In order to get there, however, we must of course have a discussion. 2. Getting clear about taintedness IANAL. But I like to approach problems in a systematical manner. For example, I recently asked myself the question, „What is a copyrightable object in OSM?“. I think this is a fundamental question to answer if you discuss licence topics. Is a node copyrightable? If yes, what's copyrightable about it? What's copyrightable about a way? Is the list of references to nodes copyrightable separately from the way's tags? Are references to nodes atomic? (I.e. Is a single reference copyrightable? Or is only the list as a whole?) Sorry for the rhetoric, but these questions do bother me. I believe they have to be answered prior to discussing which kinds of modifications to what object have what effect (- taintedness). And when that has been settled, we can talk about measures. All in all I think that the approach to the whole thing so far has been too pragmatic, just like identifying edge cases and modeling something around it. Of course, this might somehow work and the result might even be satisfying, but to me it doesn't seem appropriate in a legally significant matter like this. 3. Remapping Considering that neither the definitions of what is clean and what is tainted nor the technical details of the implementation have yet been finalized, it seems unreasonable for me to remap. I don't want to discover later that I have done unnecessary work. Besides, current remapping practice is completely based on the available inspection tools that implement - more or less precisely - a taintedness policy that is still in draft status. For this reason I also refuse to use the odbl=clean tag. Now I could elaborate a lot more. But the purpose of my post actually is to start a discussion, and I am asking you. Me too wants the licence change to be a success. So let's go. Cheers ant [1] http://wiki.openstreetmap.org/wiki/Open_Data_License/What_is_clean%3F ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
Re: [OSM-legal-talk] Implementing the licence change
Knowing it does not really start the discussion: I totally agree. Lukas (LM_1) 2012/1/18 ant antof...@gmail.com: First of all I must say that I highly respect the work of everyone who has been actively involved in the licence change, including the LWG members, the writers of licence change inspection programs and everyone involved in discussions. I have been watching the process for more than two years and have ever since been a supporter of the change. However, especially since the switchover date has been announced and the phase of remapping has started, I have become more and more skeptical about the way things are going on. I want to discuss a couple of concerns I have. 1. The black box As far as I can see the details of the implementation of the licence change, i.e. of what is actually going to happen on April 1st, are not known - or at least not revealed. Correct me if I am wrong. Particularly, the wiki page „What is clean?“[1], which has been said to be the binding document, is in its current state not sufficient to serve as a reference for any measures regarding the cleaning of data: * The considerations in the section „Edge cases“ are only a random selection of cases that have been discussed. Neither conditional stetements like „if it can be seen not to influence the current version“ nor questions like „Can you copyright the state of something not being there?“ (rhetorical?) are helpful. The list somewhat lacks a systematic approach. * The „deletion paradox“ is, as it has been pointed out on the discussion page, no paradox at all (rather it depends on the strategy of cleaning). * The section „What taints data?“ repeats the above-mentioned list, but is differently (better) structured and different in content. Statements in this list, however, contradict, or supersede, previous statements („A tag modified by a non-agreeing mapper is tainted“, whereas: correcting a tagging typo is not tainted). Furthermore the list contains instructions, which should not be the case in a mere specification of what is clean. The clause saying that intermediate versions should be created during remapping (a) does not belong here and (b) is questionable, as it is based on assumptions regarding the implementation of switchover, which has not yet been decided upon. * There should be rationales explaining for each statement why it is so and not different. Basically I think that this document needs a rewrite that shall contain unambiguous statements preceded by precise definitions. In order to get there, however, we must of course have a discussion. 2. Getting clear about taintedness IANAL. But I like to approach problems in a systematical manner. For example, I recently asked myself the question, „What is a copyrightable object in OSM?“. I think this is a fundamental question to answer if you discuss licence topics. Is a node copyrightable? If yes, what's copyrightable about it? What's copyrightable about a way? Is the list of references to nodes copyrightable separately from the way's tags? Are references to nodes atomic? (I.e. Is a single reference copyrightable? Or is only the list as a whole?) Sorry for the rhetoric, but these questions do bother me. I believe they have to be answered prior to discussing which kinds of modifications to what object have what effect (- taintedness). And when that has been settled, we can talk about measures. All in all I think that the approach to the whole thing so far has been too pragmatic, just like identifying edge cases and modeling something around it. Of course, this might somehow work and the result might even be satisfying, but to me it doesn't seem appropriate in a legally significant matter like this. 3. Remapping Considering that neither the definitions of what is clean and what is tainted nor the technical details of the implementation have yet been finalized, it seems unreasonable for me to remap. I don't want to discover later that I have done unnecessary work. Besides, current remapping practice is completely based on the available inspection tools that implement - more or less precisely - a taintedness policy that is still in draft status. For this reason I also refuse to use the odbl=clean tag. Now I could elaborate a lot more. But the purpose of my post actually is to start a discussion, and I am asking you. Me too wants the licence change to be a success. So let's go. Cheers ant [1] http://wiki.openstreetmap.org/wiki/Open_Data_License/What_is_clean%3F ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk
Re: [OSM-legal-talk] Implementing the licence change
Hi, On 01/18/2012 06:13 PM, ant wrote: As far as I can see the details of the implementation of the licence change, i.e. of what is actually going to happen on April 1st, are not known - or at least not revealed. Correct me if I am wrong. They are not known. A mailing list has been created (the rebuild list) to discuss how exactly the database rebuild is going to happen, and in terms of policy, LWG will have the ultimate decision. And they are asking for out input via the What is clean page. That page is not, and was not intended to be, a binding document - it might become one later. I assume that LWG will certainly value your help in improving that document. IANAL. But I like to approach problems in a systematical manner. For example, I recently asked myself the question, „What is a copyrightable object in OSM?“. I think this is a fundamental question to answer if you discuss licence topics. It has often been said that computer geeks, of which I presume you are one, are not well suited to perform legal analyis. The lawyer's answer to Is a node copyrightable? will almost certainly be it depends. (On country, circumstances, ...) In OSM, our current answers are: Yes, we treat a node as copyrightable; If yes, what's copyrightable about it? Its position and tags, unless the tags have been created automatically. What's copyrightable about a way? The sequence of its nodes and its tags. Is the list of references to nodes copyrightable separately from the way's tags? Every single tag and every single node reference are a treated as copyrightable by us. Are references to nodes atomic? (I.e. Is a single reference copyrightable? Or is only the list as a whole?) Atomic. As I said, this is just our current working assumption, not something set in stone. All in all I think that the approach to the whole thing so far has been too pragmatic, just like identifying edge cases and modeling something around it. Of course, this might somehow work and the result might even be satisfying, but to me it doesn't seem appropriate in a legally significant matter like this. It is possible that the approach only *seems* too pragmatic to you. I'm not with LWG but I would assume that they would welcome you into their weekly telephone sessions if you want. Considering that neither the definitions of what is clean and what is tainted nor the technical details of the implementation have yet been finalized, it seems unreasonable for me to remap. Thankfully, few other people think like you do. There may be edge cases, but I guess that whichever way these edge cases are decided, a significant portion of what is now considered tainted will always be tainted. And that stuff should be remapped *now*. I don't want to discover later that I have done unnecessary work. Your call. I'd rather re-map a few items too many than fix the holes in my street in April. Besides, current remapping practice is completely based on the available inspection tools that implement - more or less precisely - a taintedness policy that is still in draft status. For this reason I also refuse to use the odbl=clean tag. The odbl=clean tag is a kludge for difficult cases anyway. If you bring an object into a state that does not have any of the properties added by a decliner, that is sufficient to make it clean automatically. Now I could elaborate a lot more. But the purpose of my post actually is to start a discussion, and I am asking you. Me too wants the licence change to be a success. So let's go. It's ok to discuss these things, but the approach I won't move a finger until I am told *exactly* what the rules are is not helpful. The rules might *never* be final - even when we do the rebuild according to the then-believed-final rules, it could happen that someone later points out an oversight, or a court decides something, forcing us to remove things we thought we could keep or vice versa. You can only ever go up to 80% certainty in these matters. Demanding more is not realistic. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ legal-talk mailing list legal-talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/legal-talk