Which open data license?

Policy and Research
, ,
1

Cross‑posting here on the selection of open data licenses covering non‑personal information in the domain of energy systems analysis:

Release 10 of the embedded diagram showing open data‑capable license interoperability.


license-data-code-chart.10
license-data-code-chart.103636×3125 1.39 MB

1 Like
2

I am currently extending the above compatibility graph to include various national licenses — including from the United Kingdom, France, and Germany. In that quest, does anyone have any analysis on the legal compatibility between the OGL‑UK‑3.0 and CC‑BY‑4.0 licenses?

I have asked the same question in a number of other venues, involving software lawyers and data platform builders to name two, and have not received an answer. This question is absolutely pivotal when datasets are to be mixed to produce derivative works.

1 Like
3

Hi @robbiemorrison! Looks like they should be compatible: Open Government Licence terms states “These terms are compatible with the Creative Commons Attribution License 4.0 and the Open Data Commons Attribution License, both of which license copyright and database rights. This means that when the Information is adapted and licensed under either of those licences, you automatically satisfy the conditions of the OGL when you comply with the other licence. The OGLv3.0 is Open Definition compliant.”

4

Hi @lwinfree! That is not so great for the open data world. That means the CC‑BY‑4.0 is inbound compatible with the OGL‑UK‑3.0 but not visa‑versa. So material from various sources under Creative Commons data‑capable licenses, say from the European Commission or the World Resources Institute or Zenodo or from any number of scientific papers, can be mixed in with UK government data, but not be exported back out to those sources. If I was in the United Kingdom, I would certainly push for CC‑BY‑4.0 licenses on public interest information. But that’s not my patch. Many thanks for your reply. Much appreciated. I now have an arrow to place on my chart! And another data silo to label as such, regrettably.

Can anyone confirm that the OGL‑UK‑3.0 can only be applied by United Kingdom public bodies but no one else? The wikipedia article does not contradict that view. I’ve also heard it said but I would like a reliable secondary source, either way.

My next task is trying to uncover the relationship between the newly released CDLA‑Permissive‑2.0 license from the Linux Foundation and the existing Creative Commons licenses. All I have from the Linux Foundation so far is a press release that skips over much of the detail.

I think most new licenses make CC‑BY‑4.0 inbound compatibility a design requirement. To maximize their prospects in the same way that back holes attempt to.

Regarding Open Definition compliance, I am also of the opinion that the Open Knowledge Foundation should take a strategic decision to freeze all approval applications and settle on “CC‑BY‑4.0, CC0‑1.0, or something inbound compatible” as their recommendation. Legacy licenses like the ODbL‑1.0, as used by OpenStreetMap, will need to persist of course. However I note that OSM asks new registrants if they consent to dual licensing with CC0‑1.0 these days.

It is worth stressing that “interoperability” can only ever be a one‑way street. This is a basic result of formal logic. If two‑way interoperability exists, then the two interacting licenses must be legally identical in every way.

1 Like
5

This seems like two different decisions. There aren’t a torrent of approval applications, so there’s not really anything to freeze. The currently recommended licenses are close to what you recommend, except that ShareAlike is included, and of course corresponding ODC licenses are included. Regarding ODC, haven’t heard a peep in years, but that’s probably yet another mostly separate topic.

6

I will post an extended version of the directed graph shown in the top posting in due course, hopefully within the week.

Here is a very recent OKF license approval request:

7

I unluckily just noticed this now.

However I note that OSM asks new registrants if they consent to dual licensing with CC0‑1.0 these days.

This is not correct, what the check box actually means is not just a bit of a mess, but it further definitely doesn’t have any practical consequences such as OSM data being available on such terms, see The “PD checkbox” | OpenStreetMap Blog

Further I’ve noticed that your graph contains following, rather loaded, statement:

OpenStreetMap does not accept that any Creative Commons licenses are definitely inbound compatible.

The OSMFs current position on inbound licence compatibility can be found here: Licence/Licence Compatibility - OpenStreetMap Foundation as you can see for example CC0 is considered inbound compatible.

What is true is that CC BY licences are not considered input compatible, for (at least) the reasons mentioned here Use of CC BY 4.0 licensed data in OpenStreetMap | OpenStreetMap Blog, a document you are aware of. As it clearly states, the post was the conclusion of joint CC and OSMF work. I don’t think I’m spilling too many beans here if I point out that at the time CC would have been really happy if the outcome would have been different (and even if it would have only been for positive PR), but the incompatibility was completely clear and couldn’t be talked away.

8

@SimonPoole Thanks for the feedback, much appreciated. I only relatively recently signed up for OpenStreetMap (OSM) and also opted for the public domain checkbox. And naturally presumed that that action meant dual licensing with CC0‑1.0.

This is speculation on my part, but I don’t think OSM would have been as successful under a weaker attribution license or public domain waiver. In that regard though, OSM is an outlier — most exercises benefit from the ability to share information under more permissive terms.

The whole house of legal cards that make up the various “opens” are pretty shaky. Pam Chestek (2017) reviews software practice in the context of United States law, but most of what she writes would apply equally to collaborative data projects under European law (hence her paper is recommend reading for open data advocates):

Moreover, the European Commission recently contributed to the legal confusion by defining “reuse” as mere “use” in §2.11 of its open data directive. The legal concept of use in this context does not give one the right to optionally modify and republish, which is what we so desperately need. (I guess Orwell would not have been surprised by that definition.)

I completely agree with the comment by @SimonPoole on inbound compatibility with respect to the ODbL‑1.0. This is simply a matter of semantics: some regard waivers and dedications as types of license and some do not. I usually adopt the latter idiom. In any case, I’ll fix the wording on that diagram to make it clear what is being conveyed. (And I know that the CC0‑1.0 can be both a waiver and a maximally permissive license, the latter acttive in legal jurisdictions where the concept of a public domain is largely absent.)

As an aside, the Linux Foundation also very recently added to legal confusion by describing their CDLA‑Permissive‑2.0 public license as an “agreement” despite the fact it would not fall under United States contract law.

Returning to OSM, it would certainly help if the meaning of the public domain checkbox is clarified as far as is possible — while acknowledging again that many of the underlying legal questions cannot be resolved by individual projects and ultimately require litigation, or better still, considered legislative reform.

9

I’ve actually quizzed real old timers on this (and the initial discussions are documented in the mailing list archives) and the result was inconclusive. What is rather clear though is that latest during the peak of licence change discussions the share-alike camp was clearly dominant. If it really made a difference I doubt a bit. New signups tend to not be involved in such discussions and accept whatever is presented to them.

Returning to OSM, it would certainly help if the meaning of the public domain checkbox is clarified as far as is possible …

Most likely it will be simply removed (something that I’ve been on the brink of creating a PR for, many many times).

1 Like
10

I was recently asked to peel a PDF from the blog indicated earlier and upload it to zenodo so that it could be formally cited. As follows. Just for your information.

  • Morrison, Robbie (6 February 2022). Which open data license? — Release 06. doi:10.5281/zenodo.5987672. 14 pages.