Maintenance and future of the Open Data Commons licences

Hi Mor,

Sorry for the slow response. Yes, I think a blog post to spur some wider discussion is a great idea. Let me know if I can help!

Cheers,

L.

My understanding is that your paper isn’t an adapted work. You’re not republishing the ODbL licenced dataset, or a substantial portion of it, just using it in your analysis. Therefore you only need to attribute the data you’re using. It’s the essentially the same situation as including a CC licenced image in a document. I just attribute it, I don’t have to licence my whole document openly. Otherwise all open licences would effectively have a share-a-like provision.

These issues around how you comply with licences, and the different scenarios that apply are important to highlight. OSM have produced some great guidance that helps users understand what is permitted.

I guess what I’m looking for in the context of this discussion, is how ODbL supports “Produced Works” and delivers necessarily features for both publishers and reusers in a way that the CC licences don’t?

Are there concrete legal provisions or flexibility that the licence provides that would make it a better choice in some situations? If there are then I’m not sure they’re well communicated at the moment.

Cheers,

L.

2 Likes

This is a healthy sounding arrangement. I see there exists an ODC Advisory Council. Has its role in maintaining the ODC licenses and related documentation been formalized like the Open Definition AC’s maintenance of the OD?

Also, should there be an ODC category in this forum?

Unluckily things are not quite that simple, with different flavours, depending on CC variant and version. For example the CC BY * anti-DRM provision would clearly effect distribution of a collective work (up to version 3.0). The 4.0 versions do not define or reference that construct any more, but CC does indicate in their FAQ Frequently Asked Questions - Creative Commons that it is intended to work the same.

As to the guidance we have produced, you do need to put it in to a context of a continuously changing database of geo-information, some of our statements would likely be different if it was say a slowly revised catalogue of art work or whatever.

Just saying I’ve just checked in this thread - in the midst of travelling atm.

:thumbsup: on @mlinksva comments. There is an advisory council - though it could definitely get more active again. The main thing there would be to have concrete issues to consider: e.g. we should change X or Y in the license. (For example, there was some discussion initiated by Mapbox a few years ago within OSM and ODC re the form of the share-alike clauses).

I think @Mor and @ldodds suggestion of a summary post is also a good idea. I’ll be a bit stacked until end of Feb - and then more available.

As a thought, that specific point sounds a bit weird. If some group independently makes a license and calls it (say) “ODbL v2.0”, wouldn’t they be be putting themselves in a bad position re: trademark law?

@justinclift It would be bad rep but the term ‘ODbl’ is not registered for trademark protection and nor should it be. (please @rufuspollock correct me if I’m wrong)

This is indeed a problem how do you protect the quality and avoid confusion (like ‘green’ and ‘open’ washing) . What we can do as a community is to make sure this does not happen and take action when it does happen. So please when in doubt point it out !

Looking at the Advisory Council, it looks like Mike Collinson should be changed to Simon Poole. Mike was chair of the OSMF LWG, and that’s now Simon.

For public sector information (PSI), it seems that the OKI has now resolved this discussion in favor of CC‑BY‑4.0 if attribution is required and CC0‑1.0 otherwise. Indeed, ODC‑By‑1.0 and ODbL‑1.0 are now implicitly deprecated in this domain. Provocative, I know, so here is my source: Lämmerhirt (2017).

There are nonetheless quite a few problems with Lämmerhirt (2017) which the OKI might like to consider fixing and reissuing. Major shortcomings:

  • deprecation of Open Data Commons licenses in the context of government licensing policy should be stated explicitly
  • no detailed treatment on common license compatibilities, particularly between the ODC and CC families (leading one to suspect no such analysis exists)
  • the paragraph on permissive and copyleft licensing is simply incorrect

Minor issues:

  • no such thing as “civic law” (should be civil law?)
  • graphic 1 shows a previously unknown CC‑SA license, moreover license versions should be stated as version-agnostic interoperabilities do not exist
  • license annotation on graphic 2 is confusing and potentially ambiguous (I did eventually work out what “CC BY/SA 4.0” meant)
  • section on misuse of data provisions is useful but some remarks are wayward, for instance US fair use is intentionally framed to provide discretion by courts
  • full references should be given and shortened URLs especially should be avoided for reasons of information and linkrot (as per Wikipedia guidelines)
  • the term public sector information should be preferred over government information as the former is broader in scope

I suspect that problems in determining cross-license compatibilities — should these even exist — have ultimately led to the Creative Commons family becoming dominant outside of legacy usage. That, in my view, is a good outcome for open data.

Happy to answer questions or provide further background.

References

Lämmerhirt, Danny (December 2017). Avoiding data use silos: how governments can simplify the open licensing landscape. Open Knowledge International. Cambridge, United Kingdom.

1 Like

As an experiment I’ve taken a copy of the ODbL 1.0 text and started to annotate sections of it to highlight design decisions, key clauses, areas where I feel the licence is unclear and some clarifying statements.

I’m using hypothes.is so if anyone is interested they can reply to individual comments.

https://ldodds.github.io/odbl-1.0-annotations/#annotations:query:

Cheers,

L.

1 Like

Here is a good example of line-by-line license analysis which might serve as a template:

Peterson, Scott K. (23 March 2018). Why so little love for the patent grant in the MIT license?. Opensource.com.

1 Like

Hello @ldodds I am looking for a reliable source that confirms (or otherwise) that ODC‑By‑1.0 licensed material is inbound compatible with ODbL‑1.0 licensed databases. Can someone provide an authority for this? Thanks in advance, R.

Sigh, really, recommending CC BY 4.0 as a data license is such a bad idea, that one needs to ask the question if OKI actually ever studied the text of the licence and the consequences on downstream users of such licensed data. The rest of the points wrt governments not creating additional licences and so on are clearly OK, but CC BY 4.0 is not the answer to that problem.

Given that we (the OSMF LWG) clearly have the largest body of first hand experience with the ODbL, we’re glad to help with any effort that could lead to an ODbL revision and is not a purely academic exercise (I have an old list of issues somewhere that I can probably dig out).

Simon

PS: wrt academic exercise: there is still that small textual clarification to the ODC that hasn’t gone anywhere Fix sec 4.2 of ODC_BY to make clear it is not ShareAlike · Issue #3 · okfn/opendatacommons · GitHub

@SimonPoole It would be useful if you could explain why CC‑BY‑4.0 is unsuited for data in your view. Do you advocate CC0‑1.0 or OBbL‑1.0 or some other license instead? The reason I ask is that the open energy modeling community is debating a set of open data licensing recommendations at present. In most cases, neither do we get to choose a license, that is the clear prerogative of the data provider. And most, I would say, would wish to be credited, hence the need for BY licensing. (PS: thanks for responding to my earlier suggestion for an edit.)

1 Like

It would be useful if you could explain why CC‑BY‑4.0 is unsuited for data in your view. Do you advocate CC0‑1.0 or OBbL‑1.0 or some other license instead? The reason I ask is that the open energy modelling community is debating a set of open data licensing recommendations at present. In most cases, neither do we get to choose a license, that is the clear prerogative of the data provider. And most, I would say, would wish to be credited, hence the need for BY licensing

Definitely requiring downstream attribution is an understandable requirement by data providers and while it does create some downstream complications, I believe it should be catered for.

To understand the issue with CC BY and CC BY 4.0 it is important to realize that CC expects them to work in a classical “layers of copyright” fashion, that is, while the creator of “Adapted Material” can licence the changes how they see fit, the “layer” of CC BY licenced material continues to exist and any terms that the “Adapted Material” is distributed on must not conflict with the CC BY licence.

Now if CC BY was truly only a licence that required downstream attribution, this wouldn’t really be an issue as providing attribution in a suitable fashion should in general not interfere with downstream reuse. However CC BY and more so CC BY 4.0 add additional restrictions on how you can downstream distribute so licensed material.

2.a.5. Downstream recipients.

  A. ...

  B.  No downstream restrictions. You may not offer or impose any additional
       or different terms or conditions on, or apply any Effective Technological 
       Measures to, the Licensed Material if doing so restricts exercise of the 
       Licensed Rights by any recipient of the Licensed Material.

Obviously this disallows using DRM distribution for any Adapted Material that you’ve created and further can be read as not allowing any kind of terms that would not at least allow the rights listed in 2.a.1 to be exercised by any downstream recipient. Oddly enough CC claims that CC BY-SA 4.0 is compatible with CC BY, but if the SA terms aren’t "additional or different terms " that “restricts exercise of the Licensed Rights” I need to twist my brain in to several knots.

Now this again wouldn’t be so bad in a typical data reuse scenario, lets say you have CC BY licensed address data that you’ve included in your geo-data and you’ve modified the locations to be better, you can argue that the terms would actually only apply to the original and modified data. Previous CC BY versions explicitly mentioned Collections of works, It seems as if the explicit mention was dropped as this is typically already the law and it wasn’t necessary to repeat it.

Enter CC BY 4.0 Section 4.b

if You include all or a substantial portion of the database contents in a database in which You have Sui Generis Database Rights, then the database in which You have Sui Generis Database Rights (but not its individual contents) is Adapted Material; and

With other words inclusion in a database in the EU that is eligible for Sui Generis Database protection forces you to apply 2.a.5 B to your complete database.

This is not just me making things up, see Use of CC BY 4.0 licensed data in OpenStreetMap | OpenStreetMap Blog for the results of our discussions with CC (note: Section 4 didn’t play a role in this as it was clear that we would need to require a waiver of 2.a.5 B because of the relaxed terms for “Produced Works” in the ODbL in any case).

My take on CC BY is that is really a share-alike licence that simply doesn’t specify the exact licence you need to utilize, but otherwise has the same advantages and disadvantages as a SA licence. Characterizing it as an “attribution-only” licence which is so popular is in any case extremely misleading.

Naturally if you are not considering a scenario in which there will be multiple downstream reuses of what you are producing and that some of them may have more restrictive usage terms, you may come to other conclusions.

1 Like

In my opinion it’s a mistake to continue spreading such alarmistic messages as “CC BY […] is really a share-alike licence”. CC BY 4.0 is simply a modern license which includes protection from certain abuses which would otherwise make the license terms meaningless and deprive users of all their rights in practice. Such abuses are especially common with databases.

People supposedly choose CC BY for their databases because they want to really ensure (as far as private contracts can go) that the spirit of the license is respected and they’re attributed, even beyond moral rights and ethical practices. If they didn’t care so much, they could just use CC-0. Completely waiving the totally pointless and broken sui generis rights remains the safest legal solution, while attepting to use it for good is extremely difficult, but as we know it’s not always possible.

1 Like

You do realize that you are making my point? I have nothing against the use of CC BY when the way it works is declared properly. But the fact is that most users of it assume that it is something like CC0 + an attribution requirement, which it clearly isn’t. I realize that there is a certain interest in not rocking the boat here and with CC, as you illustrate nicely, but, sorry to say, that doesn’t change the facts.

Coming back to this discussion thanks to your reminder after the OSMF endorsed CC-0 for inbound contributions… I think I didn’t understand your point 5 years ago, but maybe I do now.

If I understand correctly, you’re saying that people use CC BY for databases/datasets thinking that the CC BY only adds innocuous requirements, while actually the attribution and anti-DRM provisions are so onerous that OSM has to require a waiver. Others will counter that actually the requirements are easy to respect and maybe that those waivers aren’t even needed because OSMF’s attribution practices and technological measures are already sufficient for CC BY. However this won’t change anyone’s mind so we’d need a clarification at the source.

As an aside, I wonder if OSMF has had the chance to revisit some of its position about CC BY in light of new(ish) facts. For example, Portugal’s law 36/2017 changes the meaning of “technical measure” and may impact §1(d) “Effective technological measures” and I understand this interpretation is spreading. There’s also new case law on attribution requirements.

Going back to my proposal though, I still wonder what would happen if we took database rights out of the equation completely. Obviously CC-0 is the easiest way to do that, but can OSM import non-copyrightable information from a dataset released under CC BY 4.0 + a waiver of database rights? If there’s no copyright and no database rights left from upstream, how can the resulting database be considered “adapted material”? Even if it were, §8(a) would kick in and throw any requirements out of the window.

It should be clear from Use of CC BY 4.0 licensed data in OpenStreetMap | OpenStreetMap Blog that this was based on the opinion of CC. They didn’t want to make their own statement for understandable obvious reasons, but that’s where the whole thing originates from.

PS: at OSM we tend to respect the wishes of our datasources even if they don’t know what they are doing.