New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

OBO Foundry Principle 1 misleadingly states that CC-BY is recommended license #103

Closed
GoogleCodeExporter opened this Issue Aug 12, 2015 · 14 comments

Comments

Projects
None yet
4 participants
@GoogleCodeExporter

GoogleCodeExporter commented Aug 12, 2015

The documentation for the Foundry Principle 001 states matter-of-factly that 
CC-BY is the recommended license for OBO Foundry ontologies:
http://www.obofoundry.org/wiki/index.php/FP_001_open

People not versed in the legal and copyright issues have taken this in blind 
faith as guidance [1,2], yet it is misleading at best. One, it ignores a number 
of arguments as to why CC-BY (or even more restrictive copyright-based 
licenses) may not be applicable to ontologies, that theere are alternatives, 
and what problems, such as attribution stacking, may arise from using legal 
instruments to try to enforce scholarly norms [3]. Two, it omits the fact that 
CC0, a public domain waiver, is already successfully employed by a number of 
Foundry ontologies, such as TAO, VSAO, and TTO. Notably, SO is also in the 
public domain, yet does not exempt itself from scholarly norms being followed 
[4].

I suggest that it is high time to update this recommendation before more people 
in the community take it as unquestioned guidance, which would result in poor 
advice.

[1] http://obo-discuss.2851485.n2.nabble.com/Ontology-licensing-td5914012.html
[2] https://github.com/JervenBolleman/FALDO/issues/7
[3] 
http://sciencecommons.org/resources/readingroom/ontology-copyright-licensing-con
siderations/
[4] http://www.sequenceontology.org/resources/faq.html#lic

Original issue reported on code.google.com by hl...@drycafe.net on 8 Sep 2013 at 6:57

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

I should add that the questionability of applying copyright-based licenses 
arguably pertains particularly to OBO Foundry ontologies due to the realism 
paradigm, In this paradigm, ontology terms should not only be based on, but 
represent facts of nature.

Original comment by hl...@drycafe.net on 8 Sep 2013 at 7:08

GoogleCodeExporter commented Aug 12, 2015

I should add that the questionability of applying copyright-based licenses 
arguably pertains particularly to OBO Foundry ontologies due to the realism 
paradigm, In this paradigm, ontology terms should not only be based on, but 
represent facts of nature.

Original comment by hl...@drycafe.net on 8 Sep 2013 at 7:08

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

Original comment by rlwalls2...@gmail.com on 11 Sep 2013 at 10:51

  • Changed state: Accepted
  • Added labels: Type-Editorial
  • Removed labels: Type-Outreach

GoogleCodeExporter commented Aug 12, 2015

Original comment by rlwalls2...@gmail.com on 11 Sep 2013 at 10:51

  • Changed state: Accepted
  • Added labels: Type-Editorial
  • Removed labels: Type-Outreach
@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

1) There is case law that indicate that ontologies are indeed covered by 
copyright law. This was investigated (with a non trivial amount of effort) by 
the Science Commons lawyer Thinh Nguyen.

2) It is not misleading that CC-BY is recommended. That was a decision that was 
taken at a OBO Foundry meeting. But I believe that we discussed at that or a 
subsequent meeting that CC0 was also a reasonable option. 

I'm not sure public domain is a good idea as law about this differs around the 
world and the CC licenses have been crafted to address this in a useful way. 

Attribution stacking doesn't need to be a byproduct of use of CC-BY. At 
meetings I have suggested that it is explicitly stated that the attribution 
requirement is satisfied by use of our purls. Remember the CC-BY does not 
specify how attribution needs to be made. By linking attribution to the use of 
the purls, and making the purls resolvable (sustainably so), and documenting 
MIREOT we both encourage interoperability and have effective attribution.

What I do agree is that the documentation we have is inadequate and should give 
clearer direction on how and when to use licenses. The discussed choices are 
CC0 and CC-BY. Both are reasonable and an explanation should be made of how to 
use each, and how to do so if they want to.

The policy on CC-BY was made in recognition that many authors are not yet 
willing to surrender copyright, however symbolic that may be. I think that will 
remain the case, and as I don't agree on the disadvantages of CC-BY, at least 
if used in the way I suggest, I would recommend we keep both those as options.

-Alan


Original comment by alanruttenberg@gmail.com on 3 Oct 2013 at 4:31

GoogleCodeExporter commented Aug 12, 2015

1) There is case law that indicate that ontologies are indeed covered by 
copyright law. This was investigated (with a non trivial amount of effort) by 
the Science Commons lawyer Thinh Nguyen.

2) It is not misleading that CC-BY is recommended. That was a decision that was 
taken at a OBO Foundry meeting. But I believe that we discussed at that or a 
subsequent meeting that CC0 was also a reasonable option. 

I'm not sure public domain is a good idea as law about this differs around the 
world and the CC licenses have been crafted to address this in a useful way. 

Attribution stacking doesn't need to be a byproduct of use of CC-BY. At 
meetings I have suggested that it is explicitly stated that the attribution 
requirement is satisfied by use of our purls. Remember the CC-BY does not 
specify how attribution needs to be made. By linking attribution to the use of 
the purls, and making the purls resolvable (sustainably so), and documenting 
MIREOT we both encourage interoperability and have effective attribution.

What I do agree is that the documentation we have is inadequate and should give 
clearer direction on how and when to use licenses. The discussed choices are 
CC0 and CC-BY. Both are reasonable and an explanation should be made of how to 
use each, and how to do so if they want to.

The policy on CC-BY was made in recognition that many authors are not yet 
willing to surrender copyright, however symbolic that may be. I think that will 
remain the case, and as I don't agree on the disadvantages of CC-BY, at least 
if used in the way I suggest, I would recommend we keep both those as options.

-Alan


Original comment by alanruttenberg@gmail.com on 3 Oct 2013 at 4:31

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

Alan -

Re #2, I was at that meeting (in Hinxton) and I don't recall that decision. I 
also can't find it in the minutes from that meeting. I do recall that the two 
of us were asked to work out a more balanced recommendation.

I agree that if the attribution requirement is satisfied by using the PURLs of 
reused terms (as, presumably, compared to minting new URIs for them), then 
attribution stacking should not arise as a problem from CC-BY. However, and I 
think this is part of the misleading character of the recommendation as 
currently stated, this isn't documented anywhere, neither on the OBO Foundry 
site, nor in the header of ontologies that do use CC-BY. What's more, this may 
well be in conflict with what most scientists had or have in mind when they 
insist on using CC-BY - to virtually all scientists I know who are of this 
mind, they want CC-BY to express that they require to be *cited*. That the OBO 
Foundry recommendation isn't meant to guarantee them that under all 
circumstances may come as a surprise to many. As it is now, I could reuse a 
CC-BY licensed OBO-Foundry ontology, mint new URIs for all terms, invent an 
annotation property of my own that indicates the terms' origin buried in free 
text language, provide no further citation, and still be in full compliance 
with the law and OBO Foundry recommendation (albeit probably not its spirit). 
This is neither what most scientists would have expected, nor what can be in 
the best interest of the OBO Foundry.

Bottom line, the great majority of domain scientists, and even many information 
scientists who should know better, are rather naive about the issues of using 
legal instruments versus social norms to request and ensure attribution, and I 
would argue that it behooves the OBO Foundry to document the issues and 
resulting recommendations well. 

There are some well relevant thoughts on these issues in this position:
Response to GBIF request for consultation on data licenses. Karen Cranston, 
Todd Vision, Hilmar Lapp, Jonathan Rees. figshare.
http://dx.doi.org/10.6084/m9.figshare.799766

It's about data, for which copyright doesn't apply to start with. But many of 
the considerations are along the lines of "let's assume for the sake of the 
argument that copyright does apply", and thus are in part still applicable to 
the discussion here. I think it's important, in particular for an organization 
such as the OBO Foundry, to not just give scientists some recommendation that 
may sound like what they want but in reality has little teeth in the way of 
what scientists are truly interested in, without also educating the community 
on what the issues are and the alternative ways of addressing them. 

Original comment by hl...@drycafe.net on 3 Oct 2013 at 12:57

GoogleCodeExporter commented Aug 12, 2015

Alan -

Re #2, I was at that meeting (in Hinxton) and I don't recall that decision. I 
also can't find it in the minutes from that meeting. I do recall that the two 
of us were asked to work out a more balanced recommendation.

I agree that if the attribution requirement is satisfied by using the PURLs of 
reused terms (as, presumably, compared to minting new URIs for them), then 
attribution stacking should not arise as a problem from CC-BY. However, and I 
think this is part of the misleading character of the recommendation as 
currently stated, this isn't documented anywhere, neither on the OBO Foundry 
site, nor in the header of ontologies that do use CC-BY. What's more, this may 
well be in conflict with what most scientists had or have in mind when they 
insist on using CC-BY - to virtually all scientists I know who are of this 
mind, they want CC-BY to express that they require to be *cited*. That the OBO 
Foundry recommendation isn't meant to guarantee them that under all 
circumstances may come as a surprise to many. As it is now, I could reuse a 
CC-BY licensed OBO-Foundry ontology, mint new URIs for all terms, invent an 
annotation property of my own that indicates the terms' origin buried in free 
text language, provide no further citation, and still be in full compliance 
with the law and OBO Foundry recommendation (albeit probably not its spirit). 
This is neither what most scientists would have expected, nor what can be in 
the best interest of the OBO Foundry.

Bottom line, the great majority of domain scientists, and even many information 
scientists who should know better, are rather naive about the issues of using 
legal instruments versus social norms to request and ensure attribution, and I 
would argue that it behooves the OBO Foundry to document the issues and 
resulting recommendations well. 

There are some well relevant thoughts on these issues in this position:
Response to GBIF request for consultation on data licenses. Karen Cranston, 
Todd Vision, Hilmar Lapp, Jonathan Rees. figshare.
http://dx.doi.org/10.6084/m9.figshare.799766

It's about data, for which copyright doesn't apply to start with. But many of 
the considerations are along the lines of "let's assume for the sake of the 
argument that copyright does apply", and thus are in part still applicable to 
the discussion here. I think it's important, in particular for an organization 
such as the OBO Foundry, to not just give scientists some recommendation that 
may sound like what they want but in reality has little teeth in the way of 
what scientists are truly interested in, without also educating the community 
on what the issues are and the alternative ways of addressing them. 

Original comment by hl...@drycafe.net on 3 Oct 2013 at 12:57

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

I'm pretty sure that the decision was taken in the meeting previous to the one 
you attended. I can dig through the notes but I won't at the moment, as we need 
to resolve this in any case.

Re: paragraph 2, yes you could. You would be even more free to do so with a 
public domain or CC0. 

We are in agreement about the need to document

We can try to educate better and give clearer instruction on how to license but 
we still have no leverage other than when someone applies to be Foundry. 

Ontologies are different from data and there is reason to believe we might want 
tighter controls, than CC-BY gives. However trying to figure out exactly what, 
in a way that doesn't do damage, is a piece of work that hasn't been done.

As for CC0 versus CC-BY, I think that debate isn't terribly helpful, and I 
think we would be better off all advocating for a single solution. As I say, in 
my judgement that's CC-BY. And then help is needed to do work of documenting. 
Are you interested in contributing to that? 

Original comment by alanruttenberg@gmail.com on 3 Oct 2013 at 4:29

GoogleCodeExporter commented Aug 12, 2015

I'm pretty sure that the decision was taken in the meeting previous to the one 
you attended. I can dig through the notes but I won't at the moment, as we need 
to resolve this in any case.

Re: paragraph 2, yes you could. You would be even more free to do so with a 
public domain or CC0. 

We are in agreement about the need to document

We can try to educate better and give clearer instruction on how to license but 
we still have no leverage other than when someone applies to be Foundry. 

Ontologies are different from data and there is reason to believe we might want 
tighter controls, than CC-BY gives. However trying to figure out exactly what, 
in a way that doesn't do damage, is a piece of work that hasn't been done.

As for CC0 versus CC-BY, I think that debate isn't terribly helpful, and I 
think we would be better off all advocating for a single solution. As I say, in 
my judgement that's CC-BY. And then help is needed to do work of documenting. 
Are you interested in contributing to that? 

Original comment by alanruttenberg@gmail.com on 3 Oct 2013 at 4:29

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

Hilmar and all: As a result of the recent reviews of OBI and PO, the editorial 
working group has been reviewing all accepted OBO Foundry principles (including 
principle 1) and working on recommendations for making them easier to comply 
with. We also had a discussion of licensing issues at yesterday's Operations 
Committee meeting. The short summary is that everyone agrees that better 
documentation and advice for licensing OBO Foundry ontologies are sorely 
needed. To that end, we decided to schedule a special meeting to hammer out the 
recommendations. I somehow volunteered to organize that call. I will be setting 
up an agenda and sending out invitations in the next few days. I hope you will 
be able to attend (for just an hour, I hope), as we would really love your 
input here.

Original comment by rlwalls2...@gmail.com on 10 Oct 2013 at 10:08

GoogleCodeExporter commented Aug 12, 2015

Hilmar and all: As a result of the recent reviews of OBI and PO, the editorial 
working group has been reviewing all accepted OBO Foundry principles (including 
principle 1) and working on recommendations for making them easier to comply 
with. We also had a discussion of licensing issues at yesterday's Operations 
Committee meeting. The short summary is that everyone agrees that better 
documentation and advice for licensing OBO Foundry ontologies are sorely 
needed. To that end, we decided to schedule a special meeting to hammer out the 
recommendations. I somehow volunteered to organize that call. I will be setting 
up an agenda and sending out invitations in the next few days. I hope you will 
be able to attend (for just an hour, I hope), as we would really love your 
input here.

Original comment by rlwalls2...@gmail.com on 10 Oct 2013 at 10:08

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

A doodle poll is at http://doodle.com/hnpg6ddvdamh7sm2

Add your availability if you want to attend the meeting.

Original comment by mcour...@gmail.com on 22 Oct 2013 at 11:10

GoogleCodeExporter commented Aug 12, 2015

A doodle poll is at http://doodle.com/hnpg6ddvdamh7sm2

Add your availability if you want to attend the meeting.

Original comment by mcour...@gmail.com on 22 Oct 2013 at 11:10

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

An additional requirement (From Peter D'Eustachio) that any license be 
consistent with NIH rules for sponsored research resources. (I don't know what 
that policy is - this requires research)

Original comment by alanruttenberg@gmail.com on 28 Oct 2013 at 3:41

GoogleCodeExporter commented Aug 12, 2015

An additional requirement (From Peter D'Eustachio) that any license be 
consistent with NIH rules for sponsored research resources. (I don't know what 
that policy is - this requires research)

Original comment by alanruttenberg@gmail.com on 28 Oct 2013 at 3:41

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

The NIH guideline (http://grants.nih.gov/grants/sharing.htm) state that 
"results and accomplishments of the activities that it funds should be made 
available to the public.  PIs and funding recipient institutions are expected 
to make the results and accomplishments of their activities available to the 
research community and to the public at large." but they are pretty vague about 
how to do that. Regarding publications, the public access policy "ensures that 
the public has access to the published results of NIH funded research at the 
NIH NLM PMC, a free digital archive of full-text biomedical and life sciences 
journal literature (http://www.pubmedcentral.nih.gov/)."

I think that either either CC-BY or CC-0 should satisfy the NIH requirements. 
Any publication about an ontology that is funded by NIH research should follow 
their guidelines for publishing in Pubmed.

Original comment by rlwalls2...@gmail.com on 1 Nov 2013 at 12:41

GoogleCodeExporter commented Aug 12, 2015

The NIH guideline (http://grants.nih.gov/grants/sharing.htm) state that 
"results and accomplishments of the activities that it funds should be made 
available to the public.  PIs and funding recipient institutions are expected 
to make the results and accomplishments of their activities available to the 
research community and to the public at large." but they are pretty vague about 
how to do that. Regarding publications, the public access policy "ensures that 
the public has access to the published results of NIH funded research at the 
NIH NLM PMC, a free digital archive of full-text biomedical and life sciences 
journal literature (http://www.pubmedcentral.nih.gov/)."

I think that either either CC-BY or CC-0 should satisfy the NIH requirements. 
Any publication about an ontology that is funded by NIH research should follow 
their guidelines for publishing in Pubmed.

Original comment by rlwalls2...@gmail.com on 1 Nov 2013 at 12:41

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

I would like to see some examples of cases where attribution stacking is 
thought to be an issue for use of ontologies.  If you are building an ontology 
that pulls in classes from many other ontologies, it doesn't seem particularly 
onerous to include some form of citation of each of those sources in the 
ontology you are building. 
Is the worry that publishing an analysis that uses a large set of ontologies 
will require citation of all ontologies directly or indirectly used for that 
analysis?  Is this really an implication of using CC-BY?

If a general decision is taken that CC-0 should be used, then I'd like to see a 
Foundry document with non-binding recommendations for good practice in ontology 
re-use.  This could cover important issues not covered by CC-BY.  For example, 
a recommendation that IDs and ID spaces not be re-used without permission.


Original comment by dosu...@gmail.com on 4 Nov 2013 at 2:05

GoogleCodeExporter commented Aug 12, 2015

I would like to see some examples of cases where attribution stacking is 
thought to be an issue for use of ontologies.  If you are building an ontology 
that pulls in classes from many other ontologies, it doesn't seem particularly 
onerous to include some form of citation of each of those sources in the 
ontology you are building. 
Is the worry that publishing an analysis that uses a large set of ontologies 
will require citation of all ontologies directly or indirectly used for that 
analysis?  Is this really an implication of using CC-BY?

If a general decision is taken that CC-0 should be used, then I'd like to see a 
Foundry document with non-binding recommendations for good practice in ontology 
re-use.  This could cover important issues not covered by CC-BY.  For example, 
a recommendation that IDs and ID spaces not be re-used without permission.


Original comment by dosu...@gmail.com on 4 Nov 2013 at 2:05

@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

An update that we have been working on the wording of this principle in the 
editorial group. It should be ready to move on to the main committee next month.

Original comment by rlwalls2...@gmail.com on 6 Feb 2014 at 5:43

  • Changed state: Started

GoogleCodeExporter commented Aug 12, 2015

An update that we have been working on the wording of this principle in the 
editorial group. It should be ready to move on to the main committee next month.

Original comment by rlwalls2...@gmail.com on 6 Feb 2014 at 5:43

  • Changed state: Started
@GoogleCodeExporter

This comment has been minimized.

Show comment
Hide comment
@GoogleCodeExporter

GoogleCodeExporter Aug 12, 2015

David:
The example would be the Neurocommons or similar, which at the time included 15 
or so databases and all the OBO ontologies and in principle could have included 
hundreds more. The issue comes up around the status of a query. Technically we 
would need to know which terms were touched in satisfying the query, and this 
is difficult to do. So to be safe you would have to include attribution for all 
those ontologies in the database, which is unbounded.

There are workarounds possible to address this but they need to be spelled out 
if we're going to dot all the "i"s. One way would be to include language about 
the need for attribution that would say that in cases like this a) Only the 
ontologies for terms appearing in the result set of such a query need to be 
attributed directly and b) The others can be attributed to the Foundry all at 
once if the user desires.

Original comment by alanruttenberg@gmail.com on 13 Feb 2014 at 6:02

GoogleCodeExporter commented Aug 12, 2015

David:
The example would be the Neurocommons or similar, which at the time included 15 
or so databases and all the OBO ontologies and in principle could have included 
hundreds more. The issue comes up around the status of a query. Technically we 
would need to know which terms were touched in satisfying the query, and this 
is difficult to do. So to be safe you would have to include attribution for all 
those ontologies in the database, which is unbounded.

There are workarounds possible to address this but they need to be spelled out 
if we're going to dot all the "i"s. One way would be to include language about 
the need for attribution that would say that in cases like this a) Only the 
ontologies for terms appearing in the result set of such a query need to be 
attributed directly and b) The others can be attributed to the Foundry all at 
once if the user desires.

Original comment by alanruttenberg@gmail.com on 13 Feb 2014 at 6:02

@cmungall cmungall reopened this Aug 11, 2016

@cmungall cmungall added the Editorial label Aug 11, 2016

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Aug 11, 2016

I'm reopening as there was no reason given for its closure. While we are having the CC-0 discussion over on #285, there are various things that need to be acted on in this ticket. We could do a lot on the wording of fp-001 based on useful comments above. I myself am unclear on a lot of the wording in fp-001 - for example, the part about "Ontologies may use different wording appropriate to their own needs" - surely this is problematic?

We could also have an entry in the FAQ about attribution stacking.

cmungall commented Aug 11, 2016

I'm reopening as there was no reason given for its closure. While we are having the CC-0 discussion over on #285, there are various things that need to be acted on in this ticket. We could do a lot on the wording of fp-001 based on useful comments above. I myself am unclear on a lot of the wording in fp-001 - for example, the part about "Ontologies may use different wording appropriate to their own needs" - surely this is problematic?

We could also have an entry in the FAQ about attribution stacking.

@jamesaoverton

This comment has been minimized.

Show comment
Hide comment
@jamesaoverton
Member

jamesaoverton commented Apr 10, 2018

This issue was moved to OBOFoundry/OBOFoundry.github.io#594

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment