Recent efforts of mine to nudge the local administration to sign a waiver resulted in the answer: “We already pro-actively support community efforts by offering our data CC-BY-4.0”. Further efforts of mine to nudge them to offer their data under ODbL too stalling.
I can only second-guess the administrations point of view, but I think they might be more forthcoming if the ODbL was among the licences they could choose from:
in their INSPIRE form
in their ESRI workspace
…
From talks on issue trackers I learned that people from the other end of the world face similar problems. Me thinks, this larger than life for simple OSM contributors. Please no comments like “are you just posting here to make others do your job”.
A bit of lore: The administration provides very good data. They give that away for free on their arcgis workspace: orthophotos and terrain models of great use for mapping. They want to get attributed. Will ODbL provide that? I know from talks that other users of this data, when they put it into Autodesk Revit will see the attribution in their workspace and it will appear on printed layouts, but likely not on photorealistic 3D renderings shown to the general public. If there are differences between CC-BY and ODbL e.g. in this case or others, they should be mentioned explicitly in statements to data providers. So much of a rant.
Please do not suggest that administrations release data under ODbL when they are already publishing it under a much more generous CC-BY license.
The way to make CC-BY data available for OSM would be to get them to sign the correct waiver. (It sounds like this is what you tried first.)
If you want to encourage them to pick a “ready-made” licensing option which doesn’t require a special waiver for use in OSM, push for CC0/public domain. (Which, in my personal opinion, government-published data ought to be released as anyway.) But that obviously wouldn’t give them the attribution they seem to consider important.
That is very debatable for CC-BY 4.0, just because we focus on the DRM aspect in the waiver doesn’t mean that there are not other issues with the licence, they are just defused a bit by the waiver.
But you are right in that recommending the ODbL is problematic particularly because of the licence lock-in using ODbL licenced data creates.
From what I’ve seen, most dataset authors who insist on attribution do so because they’re worried about plagiarism. Creative Commons themselves have long emphasized that copyright/copyleft licenses are an inadequate tool for preventing plagiarism, though I don’t think they expected that LLMs would come and demonstrate the point so forcefully.
Whenever I inquire about public domain status, I always offer to reproduce an acknowledgment or disclaimer statement on our Contributors page (and possibly the import proposal). It seems to be good enough for those I’ve reached out to, but local expectations may vary. If an agency is already accustomed to their CC BY terms, replacing that with anything else may be a tough sell.
Public administrations are right to avoid CC-BY when possible.
Attribution at scale is problematic, and CC-BY often creates more friction than value.
It is important to note that the official open data guidelines of many countries explicitly recommend CC-BY and at the same time discourage any form of share-alike licences.
This makes it even more unlikely that a public administration would modify its licensing choices to match the ODbL or sign a waiver to accommodate OSM’s requirements.
From their perspective, doing so would risk compromising institutional neutrality.
A better option for open governmental data would be CC0.
I’ve personally raised this point during the EU discussions on High Value Datasets (HVDs):
not only did we advocate for expanding the list of geospatial datasets, but we specifically asked for CC0 as the recommended licence.
That was actually the hardest part of the discussion: many institutions felt uncomfortable with the idea of releasing data without any conditions.
A modern alternative like CDLA-Permissive could work well for data, but it remains little-known and therefore rarely considered by public bodies.
For these reasons, rather than “campaigning for ODbL waivers”, perhaps we should reflect on whether the ODbL itself could evolve.
A future revision - ideally in collaboration with Creative Commons and the Open Knowledge Foundation -might preserve the values of the project while improving interoperability and making it easier for governments to contribute data legally and comfortably.
CC-BY -is- to some degree a share-alike licence in that is prescribes on what kind of terms you are allowed to distribute the data downstream outside of just requiring attribution, it just doesn’t say so on the box. That is why the EU recommendations never made any sense at all and essentially no use of CC-BY 4.0 data in a commercial product is going to be licence conforming.
While a ODbL revision would make sense because of some of the wording issues in the licence, there is no scenario in which that would be more compatible with CC BY 4.0. The problem is that the ODbL 1.0 is too -permissive- and making it more restrictive to be compatible with CC BY (that would essentially boil down to disallowing parallel distribution and doing away with the concept of Produced Works) would kill off major parts of the OSM ecosystem.
PS: I don’t believe that the OSMF has any where near enough money it would have to wave in front of the OKFN to get even token interest in a revision to start with, but that is a different topic.
I completely agree with you.
Many people have underestimated what the CC-BY 4.0 attribution requirement actually implies in practice, especially the obligation to keep attribution “reasonable and visible” downstream.
When you combine that with stacked attribution and the fact that datasets may be transformed, merged or generalized in ways that make the original contributions hard to isolate, CC-BY becomes extremely difficult to comply with - particularly in commercial scenarios, as you pointed out.
That’s one of the reasons why I personally consider CDLA-Permissive a much cleaner and more modern option for data, even though it is unfortunately not widely adopted yet.
Attribution is less of an issue with CC BY 4.0, matter of fact in the talks with CC that led up to Use of CC BY 4.0 licensed data in OpenStreetMap | OpenStreetMap Blog they suggested that it was unnecessary to ask for permission for central attribution. But at the time we felt that it would be very surprising for data owners that had been told that CC BY was an “attribution-only” licence and then they wouldn’t get any direct attribution, and that would lead to avoidable conflicts.
Administrations in the EU guided by this framework - English version - German version - which hints at a licence not strictly necessary, e.g. CC0 a valid choice, and when not possible, preferably use a licence that poses no restrictions apart from attribution.
So they opt for CC-BY. I fully understand them, its great, its easy. They support an ecosystem of downstream users while getting at least attributed. The OGD (open government data) portal even solicits back-links, so they get to see how their data is used. People wishing to not attribute can always buy the data under different conditions. (This will stop under CC0 of course.)
The OSM Editor Layer Index refuses to add CC-BY-4.0 licenced maps. CC-BY certainly allows them to show in the editor, when attributed. Further though, the assumption being, that e.g. aligning a road to the aerial or a path under trees to the terrain model makes the road or path a derived work of CC-BY data, which is incompatible with ODbL data. My reasoning was, if dual licenced, might work? Therefore the title, campaigning to make ODbl more prominent.
A bit of a loss here now: What I gather from the contributions so far:
Providers of orthophoto and LIDAR scans should be able to choose a licence different from CC-BY 4.0, in case of when they want atttibution, especially one that would allow them to follow the advice of the European Commission more closely, one that does not give extra restrictions beyond attribution? Preferably no licence (CC0) but certainly not the ODbL?