License approval request: "Open Use of Data Agreement 1.0"

Major Projects Open Definition
#1

This is a follow-up to the pre-submission. There have been only a few changes to the agreement since that draft was posted: changing ‘output’ to ‘result’, setting version to 1.0, removing draft notice, changing a period to a colon, and whitespace in the annotated version.

Colleagues at Microsoft have created the Open Use of Data Agreement 1.0 (O-UDA-1.0) and I’m submitting it for approval as conforming to the Open Definition. Note I am conflicted so my opinion will not count when it comes to the approval process.

That said, Microsoft believes that the O-UDA is conformant with the Open Definition because the O-UDA permits everything described in Section 2.1 of the Open Definition and only imposes conditions approved by Section 2.2: (1) retention of existing notices and (2) a preservation of warranty.

1. Link to the full license Text

Here is a link to the full text of the license in English.

2. License Rationale

The O-UDA is a simple agreement used to permit anyone to use data under the agreement for any purpose, with only minimal obligations. In short:

  • Use of data under the agreement carries no obligations.
  • Redistribution of the Output from use of the data under the agreement—including results of analysis of the data or ML models trained with the data—carries no obligations.
  • Redistribution of data under the agreement—modified or unmodified—carries two minor obligations. First, the redistributor must pass on all attribution information. Second, the redistributor must pass on the warranty and liability disclaimers from the data provider.
  • The redistribution obligations are designed to encourage sharing by limiting the liability of the data provider and ensuring that those downstream can identify where the data came from.

3. Explain whether the license may be used by any licensor, or is specific to an organization/place/jurisdiction.

The O-UDA may be used in any licensor, and is designed to be used across jurisdictions.

4. Compare and contrast to the most similar approved as OD-conformant licenses 5. Explain the benefit the new license brings over already approved OD-conformant licenses which would outweigh the costs of license proliferation?

The O-UDA is a short and simple agreement that is focused on contractual rights. CC-BY 4.0 focuses on copyright or sui generis database rights that may or may not exist, depending on the circumstances and jurisdiction. Although ODC-BY and CDLA-Permissive include contractual rights, O-UDA has simplified language in comparison to both, and relative to the latter, has taken a data provider-friendly approach by not including representations about the data.

5. Identify which recommended conformant licenses the new license is compatible with, and how – by alignment (permissions identical or a superset of existing license, conditions identical or a subset) and/or express permission to license the original and/or adaptations of the licensed work under an existing license.

The O-UDA is compatible with all recommended conformant licenses.

  • CC0: Data released under CC0 can be released under the O-UDA. Data released under the O-UDA cannot be released under CC0 because CC0 does not contain a limitation of liability required by section 4.3 of the O-UDA. Data sets containing data released under O-UDA and other data released under CC0 can be created and distributed under O-UDA.
  • PDDL: Data released under the O-UDA can be released under the PDDL. We believe that the PDDL use of “Rightsholder” encompasses “Data Provider” and “Upstream Data Provider” in O-UDA and therefore meets the O-UDA obligations. Data released under the PDDL may be released under the O-UDA. Data sets containing data released under O-UDA and other data released under PDDL can be created and distributed under either the O-UDA and the PDDL.
  • CC-BY 4.0: Data released under the O-UDA can be released under CC-BY 4.0. We believe the Creative Commons’ use of “Licensor” encompasses “Data Provider” and “Upstream Data Provider” in O-UDA and therefore meets the O-UDA obligations. Because CC-BY 4.0 does not permit relicensing, data released under CC-BY 4.0 cannot be released under O-UDA. Data sets containing data released under O-UDA and other data released under CC-BY 4.0 can be created and distributed under CC-BY 4.0.
  • ODC-BY: Data released under the O-UDA can be released under ODC-BY. We believe the Open Data Commons’ use of “Licensor” encompasses “Data Provider” and “Upstream Data Provider” in O-UDA and therefore meets the O-UDA obligations. Since the ODC-BY does not permit relicensing, data released under ODC-BY cannot be released under O-UDA. Data sets containing data released under O-UDA and other data released under ODC-BY can be created and distributed under ODC-BY.
  • CC-BY-SA 4.0: Data released under the O-UDA can be released under CC-BY-SA 4.0. We believe the Creative Commons’ use of “Licensor” encompasses “Data Provider” and “Upstream Data Provider” in O-UDA and therefore meets the O-UDA obligations. Because CC-BY 4.0 does not permit relicensing, data released under CC-BY 4.0 cannot be released under O-UDA. Data sets containing data released under O-UDA and other data released under CC-BY 4.0 can be created and distributed under CC-BY 4.0.
  • ODbL: Data released under the O-UDA can be released under ODbL. We believe the Open Data Commons’ use of “Licensor” encompasses “Data Provider” and “Upstream Data Provider” in O-UDA and therefore meets the O-UDA obligations. Since the ODbL does not permit relicensing, data released under ODbL cannot be released under O-UDA. Data sets containing data released under O-UDA and other data released under ODbL can be created and distributed under ODbL.
  • CDLA-Permissive: Data released under the O-UDA can be released under CDLA-Permissive. CDLA-Permissive has representations on exercising reasonable care to assure that “(a) the Data it Publishes was created or generated by it or was obtained from others with the right to Publish the Data under this Agreement; and (b) Publication of such Data does not violate any privacy or confidentiality obligation undertaken by the Data Provider.” Although the O-UDA does not have a similar representation, if a data provider has taken reasonable care to make the representation for even the O-UDA released data, data sets containing data released under O-UDA and other data released under CDLA-Permissive can be created and distributed under CDLA-Permissive.
  • GPLv3: Data released under the O-UDA can be released under GPLv3. Since GPLv3 does not permit relicensing, data released under GPLv3 cannot be released under O-UDA. Data sets containing data released under O-UDA and other data released under GPLv3 can be created and distributed under GPLv3.

6. Provide a link to any public drafting process (e.g., conducted on a public communication forum of some sort; multiple drafts presented to that forum) for the license.

The public drafting process was conducted on GitHub here.

2 Likes
#2

Thanks for the heads up, @mlinksva!

#3

Happy new decade…bumping to try to move https://opendefinition.org/licenses/process/#what-will-happen along:

After submission, the Open Definition community will discuss on the forum and reach consensus.

So far on the pre-submission we have:

It’s a pretty straightforward, but I suspect a bit more looking by more people is needed for reaching a consensus. :slight_smile:

Let me try pinging the long-acting chair @herb as well as @rufuspollock for the eventuality of:

The Open Definition Advisory Council chair will summarize the consensus to the Advisory Council on the forum and to other relevant venues, okfn-discuss at a minimum.

Hmm, that’s out of date since the mailing list hasn’t been used in years! But, maybe it should be?

Added: actually, okfn-discuss doesn’t seem to exist anymore. However, od-discuss still does. Maybe notices should be sent to it in addition to the forum?

1 Like
#4

Ni! Actually, @mlinksva, OKFN is dismantling all mailing lists by the end of this month and not keeping archives, so I think in terms of procedure it is mute to post to them. For all I know, you are already posting in the correct, definitive, place. Of course, you can still use any lists, OKFN or not, to publicize this forum thread.

Happy new decade to you too S2, despite where its first year seems to be headed : P

1 Like
#5

Thanks for the ping on this. I had missed it.

We’re not currently using od-discuss at all since moving to the forum. I think we should leave at that unless we want to bring up the topic as point of discussion in itself.

We used to send a notice to okfn-discuss whenever we were essentially about to call for approval for a license, just to cast as wide a net as possible for any objections. I think that’s still a good and relevant practice but we’d want to send it to the corresponding forum instead. I would mention it here but I actually don’t know which forum would be the corresponding one. If anyone knows please let us know.

#6

@solstag i believe they are keeping the archives.

Re process: yes, let’s use this forum going forward.

#7

Hi @solstag,

We at OKF have kept the mbox and HTML archive files for all the mailing lists which used to live at https://lists.okfn.org. We can and have made these files available to all list admins over the past couple of months and can provide them on demand.

But we will not be keeping the archives live at their https://lists.okfn.org URLs.

cc @rufuspollock

#8

@stephenabbottpugh i would suggest that we should be archiving the urls and keeping them alive - it’s really easy to do and would mean that web links would not break. Is there a reason we couldn’t do that - it’s just a question of “wget-ing” the sites and spitting out the output to an s3 bucket and implement redirects i suspect …

We seem to have done that already for some lists e.g. https://lists-archive.okfn.org/pipermail/okfn-labs/2013-July/002690.html

PS: i suggest we start a separate thread about this before we take this one off-topic.

#9

Hi,

@mlinksva posted this back in November with a nudge in January. Can I suggest that we try to move this forward @herb?

I’ve reviewed the original submission, the overview and the annotated version of the licence.

I don’t see anything that specifically contradicts either the letter or the broader intent of the open definition. So I’d be happy to indicate that this is a conformant licence.

For the purposes of decision making, if anyone has dissenting opinions they could they share them in a reply?

If you’re happy with the licence, then a quick +1 reply to this email would help us get a sense of consensus.

Conformance aside, I’m always wary about seeing proliferation of new licences but that shouldn’t hold us up from indicating that new licences are conformant. In fact we need to do that as soon as possible to build confidence.

The licence is compatible with many existing licences, so there is plenty of freedom for combining together new and existing data and relicensing as appropriate. I really like that this is clearly stated in the supporting documentation.

There are some slightly different design choices here, e.g. freedom to re-licence, no requirement to attribute use, distinction between Results and Data, which I think are interesting (but again, not relevant to the review).

1 Like
#10

Hello All, I would like to move forward with this. Thanks to all who have kept the ball rolling. I am going to assume everyone has had a chance to review the license and the thoughtful sumbission provided by @mlinksva.

I would note that the actual point 5 is partially combined with point 4 in the submission, so that point 5 in the list would correspond to point 6 in our “Submitting The License” guide in the process. In any case, it seems like we have all of the information we need to proceed.

I don’t see anyone raising any issues thus far. Does anyone see any issues that they feel need some discussion before we proceed? If so, please reply on this thread by May 11, 2020. If any issues are raised we can discuss, if not, we’ll proceed with proceed with step 2 of the process.

Since it’s been a while, I would appreciate it if the council members could please check in with some sort of acknowedgement that they have seen this note so I can reach out to those that don’t respond.

Thank you,
Herb

2 Likes
#11

As it wasn’t explicitly mentioned, writing to say that I would have just marked “like” to show a response that I saw the post. But posting this reply to be clearer this time.

1 Like
#12

HI Herb,

Thanks for your email. I don’t have any particular issues associated with this agreement.

Apologies too as I have been MIA for some time. But Im back on deck now.

Kind regards to you and all.

b

1 Like
#13

Thanks all for following up on this. FYI O-UDA-1.0 has been merged in the SPDX license list repository and will be included in the next version of the license list.

1 Like
#14

Thank you to those who responded. I’m going to reach out by direct email now to those who have not yet responded.

#15

Sorry for the late reply. I didn’t realize I was in the loop until recently. I agree with the other comments and see no problem in calling the new license conformant.

#16

As per others I approve.

#17

I find some of the drafting exceedingly odd (weird mix of what the Data Provider does/promises not to do, v. what the O-UDA does/not does, when the O-UDA is not an actor) but I think that’s not material, so I’m a :+1:t3:

#18

I like how brief this document is. There are a number of strange aspects to it though that make me wonder why.

The open definition requires an open license or status. In the first line this document says “… you agree as follows.”… not “we agree as follows.” Is this a license? If so why is it called an agreement and not a license? And, why does the word license not appear anywhere? Further, why is only one party agreeing, the consumer.

1.1 - no issues
1.2 - this seems like a strange thing to have in a license but I don’t think it does harm
1.3 - no issues

2, 2.1, 2.2, 2.3 - no issues - like @ldodds I find the distinction of results interesting

3, 3.1 - no issues
3.1.1 - no issues, clause 2.2.1 of the definition permits this restriction
3.1.2 - This is the one clause that I am having trouble with. While I don’t see it as particularly onerous, I also don’t see where the definition permits this condition. Can anyone can help me see where this would fit?

4.* and 5.* no issues

So, in summary, 1) I question whether or not this is actually a license, 2) assuming it is, then I wonder about 3.1.2 being an acceptable condition.

#19

The open definition requires an open license or status. In the first line this document says “… you agree as follows.”… not “we agree as follows.” Is this a license? If so why is it called an agreement and not a license? And, why does the word license not appear anywhere? Further, why is only one party agreeing, the consumer.

Often there are few, if any, licensable rights in data, and so this is framed as an agreement to ensure that all required permissions are granted subject to the conditions. To the extent that there are any licensable rights, they are licensed by section 1.

O-UDA-1.0 might stick out for not containing the word license but the approach doesn’t seem fundamentally different to me from that of other open data license/agreements such as those from ODC or CDLA (latter of which we should review at some point).

The OD (recent versions anyway) was written to generalize accordingly: The term license refers to the legal conditions under which the work is provided.

3.1.2 - This is the one clause that I am having trouble with. While I don’t see it as particularly onerous, I also don’t see where the definition permits this condition. Can anyone can help me see where this would fit?

I think a couple places:

  • 2.2.3 ShareAlike: if it’s OK to require the same terms, surely it’s OK to require only an equivalent portion the same terms
  • 2.2.1 Attribution and 2.2.4 Notice: often proxies for passing along those sorts of limitations of liability. For example, in CC-BY-4.0 Sec. 3 is labeled “attribution” and Sec. 3.a.1.A.iv requires “a notice that refers to the disclaimer of warranties”. O-UDA-1.0 is similar, but more permissive.