Licensing Discussion

[Herb Lainchbury]

Hi All,

At the hackathon last week one of the main discussion areas was licensing.  

As more governments prepare to open up our data I think it would be useful for us to have an open discussion about licensing in general.  There is still plenty of time for us to influence this process and the sooner we provide input the better as it may be harder to change course as we get more and more momentum.  As the people who will end up using these licenses (or rejecting the data that uses them) it's important that we get our thoughts to the folks who are making these decisions.  

My preference would be to have the discussion in an open forum, like this one. Sometimes people are not comfortable with sharing their views publicly, which is fine.  I would encourage anyone who finds themselves in that situation to either share their thought directly with me and I can relay them to the group, or, share them directly with the folks who are making the decisions in the jurisdiction you're thinking about.

The main points that I heard people discussing were:

• use of standard licenses is preferred over custom licenses ( PDDL, ODC-by, ODbl, CCZero, CCby, )
• attribution should be optional
• revokability ( Kevin, can you flush this one out for us? )

Did I miss anything?  Please chime in if there is something I missed.

I would appreciate any thoughts anyone has on these points or any points they recall from the hackathon.

Thanks!

H

[Kevin McArthur]

Hey Herb, thanks for posting this thread.

Since the legal framework is so important I'll chime in with a few suggestions.

1) Encourage the use of context-aware licensing.

This means that licenses should be chosen specifically to meet goals of the data release. For example in open-source software development, we often chose between BSD licenses, the GPL or LGPL and theres strong reasons for supporting all of them in specific contexts. No one license is better than the others, they've just got different applicability.

So too should data providers pick the most appropriate license and while I like the OGL's introduction I'm fearful that it could become 'the license' for BC. As it relates to BC's open-data initiative, I think Ministries must have the ability to pick and choose from a list of appropriate licenses. For example, what works for Geographic Data has almost no relevance to Hansard information and has different privacy concerns than does health or vital statistics data. The OGL appears to be an attempt to create a one-size fits all approach, and as a starting point its a solid license that should help BC get more data out -- that said, I think there's room for improvement.

• use of standard licenses is preferred over custom licenses ( PDDL, ODC-by, ODbl, CCZero, CCby, )

2) The use of standard licenses should be thought of like the use of a logo program.

If you create your own program, no one knows or cares about it -- but if hundreds of data providers are using the same logos then they begin to create standard value. Adoption will be driven only by common understanding of the licensing choices made, and if you create a custom license, then developers will be expected to have their own council review your terms, and that's an unrealistic expectation of the marketplace and will hurt serious adoption.

• revokability ( Kevin, can you flush this one out for us? )

3) The revokability provisions of open licenses should be removed.

This point stems from discussion at the hackathon on the revokability of licenses and I'm happy to lend my $0.02.

I've reviewed the BC OGL license and found that its generally very good, but I believe it has an issue around revokability in its clause 7.

Clause 7 states:

• 7.
  Despite any rights granted herein, this license does not apply to the use of:
  • a.
    Personal Information;
  • b.
    information or Records that are not accessible pursuant to the Freedom of Information and Protection of Privacy Act (B.C.);
  • c.
    third-party rights the Information Provider is not authorized to license; and
  • d.
    Information subject to other intellectual property rights, including patents, trade-marks and official marks, and design rights.

From a developer's perspective this means that the data provider is shifting their rights clearing responsibilities onto the developer. Whether its PIA, or third-party rights, these things should be cleared up before publication of a data-set. It's easy to see why data providers would want to add a clause like this covering any mistakes made in the clearing process, however, at the end of the day the data publisher is the only party in a position to properly understand these considerations.

For example, say a dataset is released that accidentally contains PIA. The OGL is applied to it and citizens form a company and invest $1M to create some new revolutionary application. The application becomes popular and people start complaining that the data breaches privacy rights, maybe there's a court case and a judge decides the government never should have released the information because the PIA wasn't done properly. So the government pulls the dataset and notifies the company that their use is not considered licensed (because of clause 7). In this case, it is the government who made the mistake and released data they should not have, however it is the company and their investors who will be out of pocket. The government is not liable for damages because the data comes with no warranty and the developer has no legal recourse.

I would suggest instead that data release needs to be considered an act of publication and that the responsibilities for not breaching PIA or third-party rights can only be left to the data publisher. Once that decision is made to publish, there must be no going back. If the license is revokable, how do developers decide to invest their own money in something that may not be there tomorrow? How do developers make a call on PIA or third-party rights in the government information?

So if the OGL is to become a template license and join the list of re-usable off-the-shelf licenses, then really, this issue of clause 7 must be addressed. This could be addressed by striking the clause all-together or by providing some sort of economic/investment rights indemnity for scenarios when data is revoked.

Also, to be clear, this refers to the revocation of already published data, and not at all to the publishers ability to simply not provide updates to datasets going forward.

Did I miss anything?

I'd also like to see an official argument against the use of PDDL. There have been suggestions made that Canada does not have a public domain, however, I'm not sure such arguments have been defined clearly. Canada must have some form of a public domain, (eg where do works go after their copyright terms expire)... but we should have some sort of debate about the PDDL's specific use in Canada, as I believe it is the most easily applied and most open license we have available.

I'd also like to see discussion of a format for disclosure of the service-level-agreement around datasets.... eg this data has no updates guaranteed, or this data is guaranteed to be updated until date x... etc. This is a key risk assessment metric that if defined will allow for better business decisions to be made around opendata.

As always, just $0.02

--

Kevin McArthur

[James McKinney]

re: standard licenses: There are I think 10 cities now with more or less the Vancouver license (cosmetic changes in most, addition of No Waiver and Authority clauses in some), but as there is no acknowledgement of this I have to read each one and find out what's different. I'm not sure cities understand that data users care a lot about the license. They currently treat the license like a software EULA or web site TOS (in other words, as something most users of the software/site don't care to read).

re: revokability and rights-clearing, I worry that putting the onus fully on the provider and adding warranties and other options for legal recourse against the provider will mean less open data. If we are willing to accept that the provider may renege on its commitment to provide updates to datasets, how is that so different from accepting that the provider may remove datasets from its open data portal? In either case, users relying on the data will be in trouble.

I'd like clarification on whether it's OK to continue to use a dataset after its license has been changed. For example, if I CC a song and then remove the CC license, can people no longer legally share the song?

[Kevin McArthur]

Hey James,

Licenses cannot be removed once issued unless there is something in the license that allows it. A CC licensed work will always be CC.

I agree that properly clearing data could result in less data being released, however, it should be clearly stated through the selection of specific licenses whether or not that data is cleared properly and businesses can/should rely upon it. A PDDL release says the rights are fully cleared.... but what does an OGL release mean? "rights sort-of cleared"?

As for the distinction between SLA vs revoking licenses... I agree that it would be nice to have a SLA on certain datasets, but I think thats probably unrealistic to expect. What should happen is the SLA should be published, and if the data is not guaranteed for updates, then that should be a consideration in the use of the data. (Eg dont build your business around it if you require updates)... On the other hand, certain datasets have value even if not updated, physical geodata for example. Its these cases were revokability becomes really concerning as you cant really make a risk assessment on it from day one.

Consider business license datasets that are in geographic format. Even if not updated, they could serve as a basis for a crowd-sourced dataset for which the developer manages updates. However, if someone claimed a PIA issue after-the-fact in that dataset, the whole app/company could be shut down instantly under a license like the OGL but that couldn't happen under a license like the PDDL as the responsibility for publication PIA issues rests with the publisher and not the developer.

Newspapers and Book Publishers have been handling the legal concerns around publications for a very long time now, so the law around who is responsible for bad publication is relatively well established. I don't know of any of these cases where a newstand or bookstore is expected to pull the pages off the shelf or face their own liability -- however, that's precisely what we're talking about in the case of republished opendata under revokable licenses.

--

Kevin McArthur

[James McKinney]

Thanks for the clarifications. If there is a PIA issue, the app/company will have to respect it (by appropriately modifying/removing the data), whether the data at issue was licensed PDDL or OGL, no? It's just a question of liability then, right?

[David Eaves]

Hi everyone,

Am in the thick of some discussions related to this issue so this is helpful. 

A few questions/thoughts:

- my preference is not to give departments a choice around licenses as this will eat up lots of time and many will default to less open licenses.   My sense is that the UK OGL is actually good because it covers all materials well - I'd love to hear more about why a data specific license is required as there has been no use case or issue sites to subsantiate this request. 

- my sense is that very few governments, probably not our BC gov and definitely not the Feds (cities which are more flexible) will adopt a license whose governance they do not directly influence. This rules out the PDDL and others for them. So I'm in particularly interested in how to make the OGL a standard. 

- the Clause 7 comments are very noteworthy

--

www.eaves.ca

@daeaves

Sent from my iPhone

[Kevin McArthur]

Hey David

- my preference is not to give departments a choice around licenses as this will eat up lots of time and many will default to less open licenses.   My sense is that the UK OGL is actually good because it covers all materials well - I'd love to hear more about why a data specific license is required as there has been no use case or issue sites to subsantiate this request.

This road leads to regulatory capture. License diversity is a good idea for the same reason biodiversity is -- monocultures present a serious risk that someone makes a bad decision (either today or in the future) and that choice destroys the whole ecosystem. The federal portal is actually a pretty good example of where a monoculture license has already failed pretty spectacularly.

What would the linux ecosystem look like if GPL had been the only option? If BSD? If MIT? If Affero?

Its the classic Sauron's Ring problem... no one license can rule them all and we'll never all agree on how our datasets should be licensed or on what license terms we're willing to accept. That kind of stuff requires a marketplace, ideas need to succeed and to fail before a winner can be chosen. Call it license natural selection.

- my sense is that very few governments, probably not our BC gov and definitely not the Feds (cities which are more flexible) will adopt a license whose governance they do not directly influence. This rules out the PDDL and others for them. So I'm in particularly interested in how to make the OGL a standard.

Why is the government so interested in license development? If what you suggest is correct, what is the motivation for the government to be in 'data governance'... I suggest that opendata comes from the public's right to their own information and as such that data should be provided in a maximally open fashion. (Eg public domain)

- the Clause 7 comments are very noteworthy

:)

--

Kevin

[Kevin McArthur]

James,

No, if the data is published in PDDL then the only requirements on the developer are those found in PIPEDA/PIPA. If its OGL and a PIA (FIPPA) problem comes up, it could result in the entire dataset being withdrawn. There's very little in terms of regulation on the private sector on how they use information that is properly acquired.

In the PDDL case privacy law would reign, in the OGL case, copyright law and the language of the license would. The way I read it, clause 7's privacy stuff exports government FIPPA rules onto the private sector via copyright/license which is very different from the PIPEDA/PIPA requirements that would otherwise apply. However, "I am not a lawyer, this is not legal advice."

$0.02

--

Kevin

[Herb Lainchbury]

I would really appreciate a good argument against the PDDL, because I am currently advising governments to use it.  I think it resonates with software developers the most and will drive the most innovation and, I seriously cannot think of a good reason why citizens shouldn't just get full access to their data.  It's just hard for me to justify anything less with what I have heard so far.

I don't get the "public domain isn't recognized in Canada" argument.

First, it seems, that the federal government at least recognizes it as they use the term approximately 758,000 times on their web site according to this Google search: site:gc.ca "public domain"

And second, logically (but I'm no lawyer), it seems to me that if you take a set of all creative works (A) that have no protection at all, and then create a copyright law that takes a subset of those items (B), and say these ones are protected, then that doesn't make the remaining ones in A disappear.  They are still there, and they are called public domain.  

And, hopefully, since copyright protection is (supposed to be) time limited, some of the ones in B fall out of that subset and revert to not being in subset B, but still part of A.

What am I missing?

H

--
Herb Lainchbury
Dynamic Solutions Inc.
www.dynamic-solutions.com
http://twitter.com/herblainchbury

[David Eaves]

The Feds aren't going to hand over the terms of licensing their data to a group that have no control over. The municipalities will cause they don't have the resources (or care enough about it). The provinces care more but I suspect will inky jan it over to a body they trust and influence. For the Feds it is a non-starter. 

I also think it is a legitimate interest. 

D

--

www.eaves.ca

@daeaves

Sent from my iPhone

[Herb Lainchbury]

Dave, you know what happens when you tell hackers it can't be done.  :)

and who's inky jan?

[David Eaves]

Ugh, bad thumbs...

inky jan = hand it

My own feeling is that a lot of political capital can be spent trying to win this battle, and that it still won't be won. For me, it is better to deploy resources to other battles...

[David Eaves]

Kevin. Thank you for the response, I'm thinking this all over, but some notes below.

--

www.eaves.ca

@daeaves

Sent from my iPhone

On 2011-10-05, at 3:44 PM, Kevin McArthur <ke...@stormtide.ca> wrote:

Hey David

- my preference is not to give departments a choice around licenses as this will eat up lots of time and many will default to less open licenses.   My sense is that the UK OGL is actually good because it covers all materials well - I'd love to hear more about why a data specific license is required as there has been no use case or issue sites to subsantiate this request.

This road leads to regulatory capture. License diversity is a good idea for the same reason biodiversity is -- monocultures present a serious risk that someone makes a bad decision (either today or in the future) and that choice destroys the whole ecosystem. The federal portal is actually a pretty good example of where a monoculture license has already failed pretty spectacularly.

What would the linux ecosystem look like if GPL had been the only option? If BSD? If MIT? If Affero?

Its the classic Sauron's Ring problem... no one license can rule them all and we'll never all agree on how our datasets should be licensed or on what license terms we're willing to accept. That kind of stuff requires a marketplace, ideas need to succeed and to fail before a winner can be chosen. Call it license natural selection.

Kevin - I find this line of reasoning not at all uncompelling. Rather than talk in metaphors it would be helpful to hear about any data the uk ogl could not manage well at the moment. I could reply that one license creates a smooth surface that anyone can roam anywhere one... it's a nice metaphor, but doesn't advance the debate. There is a real legitimate interest in simplicity, it makes the job of managing the various actors and the unknowns much easier. I'm open to the possibility that diversity, but again I need to hear what the UK OGL does not allow us to do.

- my sense is that very few governments, probably not our BC gov and definitely not the Feds (cities which are more flexible) will adopt a license whose governance they do not directly influence. This rules out the PDDL and others for them. So I'm in particularly interested in how to make the OGL a standard.

Why is the government so interested in license development? If what you suggest is correct, what is the motivation for the government to be in 'data governance'... I suggest that opendata comes from the public's right to their own information and as such that data should be provided in a maximally open fashion. (Eg public domain)

If this is what you believe, why argue for any license? You should be advocating for public domain (Which I think we all agree would be ideal - and not going to happen in the next 6-12 months in BC and definitely not at the federal level). So if we are talking about license, the government is going to play a role in creating them, just as they have for the past 100+ years. This is a the political reality as I see it, so I'm interested in how we maximize impact within this constraint.

[Kevin McArthur]

Kevin - I find this line of reasoning not at all uncompelling. Rather than talk in metaphors it would be helpful to hear about any data the uk ogl could not manage well at the moment. I could reply that one license creates a smooth surface that anyone can roam anywhere one... it's a nice metaphor, but doesn't advance the debate. There is a real legitimate interest in simplicity, it makes the job of managing the various actors and the unknowns much easier. I'm open to the possibility that diversity, but again I need to hear what the UK OGL does not allow us to do.

Hansard is probably the most obvious. Issues of privilege and parody that we've gone back and forth about come to mind. Then there's data that has monetary value -- certain scenarios may allow for non-commercial use being free, commercial use being costly, and while I personally think charging for data is a bad idea, its better to have a CC-NC licensed dataset than no dataset at all. The government also makes software, like the 'Tables on Diskette' program from CRA, which could potentially be released under GPL or other F/LOSS licenses. Other datasets might need to be released under full crown copyright -- and in-fact we've already seen license diversity in BC with the difference between openinfo.gov.bc.ca datasets and data.gov.bc.ca ones. The former being fully crown copyright, and the latter fully OGL.

I can also forsee negative changes to the OGL in the future to patch policy framework issues, or as the result of lobbying to give a specific vendor an advantage. Lets not underestimate the power the large lobby organizations will eventually wield over this process, getting the ecosystem right from day one will help keep it free from the corruption issues that we've repeatedly seen creep in to government standardized processes. (Take the former requirements to use MS Frontpage for web development, or the lawsuits regarding f/loss software and the bidding process as examples of such politically failed standards)

I'm asking that we don't repeat a bad history, and instead learn from progressive policies like the awesome BC Social Media Guidelines which eschew centralized control over everything.

Its also very possible for data publishers to create a list of approved licenses (and this is probably even desirable) to ensure that any license chosen is appropriate, but I would hope such would be a long list like those that are approved as OSI OpenSource .... eg http://www.opensource.org/licenses/alphabetical

We really should have the best license for the circumstances, and a great diversity of standard licenses to choose from.

- my sense is that very few governments, probably not our BC gov and definitely not the Feds (cities which are more flexible) will adopt a license whose governance they do not directly influence. This rules out the PDDL and others for them. So I'm in particularly interested in how to make the OGL a standard.

Why is the government so interested in license development? If what you suggest is correct, what is the motivation for the government to be in 'data governance'... I suggest that opendata comes from the public's right to their own information and as such that data should be provided in a maximally open fashion. (Eg public domain)

If this is what you believe, why argue for any license? You should be advocating for public domain (Which I think we all agree would be ideal - and not going to happen in the next 6-12 months in BC and definitely not at the federal level). So if we are talking about license, the government is going to play a role in creating them, just as they have for the past 100+ years. This is a the political reality as I see it, so I'm interested in how we maximize impact within this constraint.

I am advocating for Public Domain... via the PDDL as the premier license choice. However, some agencies have suggested that there is an issue with the public domain concept specifically in Canada, but we've not been able to nail down exactly what that issue/concern is.

As for "license whose governance they do not directly influence" ... the PDDL isn't nebulous and its use would be by a specific version eg 1.0 of the license, so its terms would never change if applied. The purpose of the PDDL license is do undo the default copyright stance and disclaim all rights in a dataset, returning it to the public to whom it belongs. I don't believe BC, or any other jurisdiction needs their own license and creating a host of custom licenses just confuses data consumers. PDDL or CC0 or any of the other free dedication licenses are instantly recognizable, and approved by the marketplace as strong legal documents. The OGL on the other hand has some evolution to achieve before it could attain the same sort of reusable status -- and if you remove the offending clauses, it starts to look a lot like a public domain license, but with different wording.

I don't give the government much credit for creating a license, in fact, under the ODUI its a negative factor -- however, I do give them a lot of credit for using an open license on their data, even if that one is their own.

[David Eaves]

I think we are going to have to agree to disagree.

Hansard is not a good example since it is not government data. Parliament is a separate institution, governed by totally unique rules and its data isn't controlled by the government.

I think the second part of your example is still more problematic to me. I actually can't think of any data that should have a non-commerical license applied to it. It is should be public and open for all re-use, or it shouldn't (because of privacy or security implications). This is precisely why I want a default license. Of course, as in the UK rules you can not use the OGL and adopt a non-commerical license (so diversity exists) but you have to apply and explain why. Again, it sounds like we agree to disagree - I'm just not hearing any powerful arguments to the contrary. Certainly nothing that is a major alarm bell ringer.

There is also a radical difference between social media policy and IP management... but not going to dive into that.

There is also no conflict between crown copyright and the OGL, indeed the two are mutually dependent. When one releases something under the OGL you are also releasing it under crown copyright. Indeed, what I like about the OGL is it makes the parameters of crown copyright more clear (and constrained).

I agree as well that the OGL looks pretty much like public domain, especially if you can find a solution to clause 7. That's a good outcome, not a reason to dismiss it.

Again, I'm curious about improvements people would like to make to the OGL - so if people have thoughts I'm very interested.

cheers,
dave

[Bruce Atherton]

On Thu, Oct 6, 2011 at 12:18 AM, Kevin McArthur <ke...@stormtide.ca> wrote:

Its also very possible for data publishers to create a list of approved licenses (and this is probably even desirable) to ensure that any license chosen is appropriate, but I would hope such would be a long list like those that are approved as OSI OpenSource .... eg http://www.opensource.org/licenses/alphabetical

But note how important it has been to have all of those licenses vetted to follow the OSI Definition so that anyone distributing code covered by one of them could ensure they had certain guarantees without getting lawyers involved. What is the Open Government Data Definition that could give the same certainty about licenses in that sphere?

Also note that in reality only a few licenses tend to be adopted by more than 1 project. That's why the CC license system has been such a great thing, because you can tweak it as you see fit while keeping the license analysis to a minimum.

I also really don't want to see the data equivalent of the GPL vs Apache license compatibility conflict. There is a reason that Apache projects ship no GPL code bundled with their own.
 

I am advocating for Public Domain... via the PDDL as the premier license choice. However, some agencies have suggested that there is an issue with the public domain concept specifically in Canada, but we've not been able to nail down exactly what that issue/concern is.

I don't know specifically what you are referring to, and IANAL. However my understanding about the lack of a real public domain, at least with code, is that it is a question of responsibility. If you write code that causes harm you cannot avoid responsibility by saying, "I put it in the public domain". You can, however, minimize your exposure by having a license with specific terms that state that you are not responsible for some types of harm. So the problem with the public domain is that it opens you up to liability.

How that applies to data I have no idea.

[Herb Lainchbury]

This license proliferation challenge occurs in FOSS as well, and there is a solution that I think is quite elegant, and Bruce eluded to it.  It's the conformance test.  It allows publishers to have control, and it allows consumers a level of assurance.

OSI and FSF do this in the software world.  In the open data world we have the Open Knowledge Foundation test based on OSI, here: http://opendefinition.org/okd/ and they provide a list of licenses with their conformity here: http://opendefinition.org/licenses/

( Interestingly, the OGL UK is not on any of the lists!!!  I have submitted a question to their mailing list and am awaiting response. )

Now, the OpenDefintion work has also been incorporated and enhanced based on practice and published as the Sunlight Foundation's principles:

 http://sunlightfoundation.com/policy/documents/ten-open-data-principles/

Unfortunately, this one isn't considered a conformance test as such, but perhaps it could be. 

If we could get some agreement on using a conformance test, then we could just test whatever licenses the governments come up with and developers would know where they stand without doing a ton of work and seeking their own legal council.  

In any case, I wonder if it might be a way to architect a solution that serves everyone's interests.

H

[Kevin McArthur]

David,

Hansard is not a good example since it is not government data. Parliament is a separate institution, governed by totally unique rules and its data isn't controlled by the government.

By that logic neither is Canada Post's data or the CNSC, Bank of Canada or any of the other independent but still very much government organisations. Our efforts are not focused around any one level of government or any one specific data provider. I don't want to see a different custom license for every crown corporation -- that would be a nightmare to deal with.

I think the second part of your example is still more problematic to me. I actually can't think of any data that should have a non-commerical license applied to it. It is should be public and open for all re-use, or it shouldn't (because of privacy or security implications). This is precisely why I want a default license. Of course, as in the UK rules you can not use the OGL and adopt a non-commerical license (so diversity exists) but you have to apply and explain why. Again, it sounds like we agree to disagree - I'm just not hearing any powerful arguments to the contrary. Certainly nothing that is a major alarm bell ringer.

I don't want data applied NC... but at the same time, we've been flat out told there's certain data that will not be released because they're currently charging for it and rely on the revenues to create the data. You want to talk about political capital, here's where we could actually get some data (like postal codes) without the organizations completely changing their funding models. (Though, yes I do think they should change their funding models too, however that process will undoubtedly take longer and I want the data in the mean time)

There is also a radical difference between social media policy and IP management... but not going to dive into that.

Not really. The same arguments for centralized communications control can be made -- speak with one voice, on-policy messaging, don't defame anyone, etc... but the BC guidelines recognise that BC's civil servants are intelligent people who can make these important decisions for themselves. They've won awards and actually allowed some of the open-data folk in the government to talk freely and frankly with the community. We've seen what forcing people through centralized IP or communications brokers looks like, and its not pretty. Its changing in social media, and it must change too in the government standards arena's around open data and software procurement. Diversity is a good thing.

There is also no conflict between crown copyright and the OGL, indeed the two are mutually dependent. When one releases something under the OGL you are also releasing it under crown copyright. Indeed, what I like about the OGL is it makes the parameters of crown copyright more clear (and constrained).

I refer to OGL vs fully enforced crown copyright (eg unlicensed use) ... openinfo falls under full, obstructed, copyright, whereas opendata.gov.bc.ca falls under the OGL. Thats a form of license diversity right there, and one the BC government clearly sees as important or they would have chosen to apply the OGL to openinfo.gov.bc.ca.

I agree as well that the OGL looks pretty much like public domain, especially if you can find a solution to clause 7. That's a good outcome, not a reason to dismiss it.

Point is we're spending tax dollars re-inventing wheels that are already well defined. Why create new when you can pick from a marketplace full of good licenses.

--


Kevin

[Kevin McArthur]

Bruce, thanks for the comments. Replies inline.

>
> But note how important it has been to have all of those licenses
> vetted to follow the OSI Definition so that anyone distributing code
> covered by one of them could ensure they had certain guarantees
> without getting lawyers involved. What is the Open Government Data
> Definition that could give the same certainty about licenses in that
> sphere?

Herb answers this in another post.

>
> Also note that in reality only a few licenses tend to be adopted by
> more than 1 project. That's why the CC license system has been such a
> great thing, because you can tweak it as you see fit while keeping the
> license analysis to a minimum.

+1

> I also really don't want to see the data equivalent of the GPL vs
> Apache license compatibility conflict. There is a reason that Apache
> projects ship no GPL code bundled with their own.

Here's where it gets complicated. I've developed software both under BSD
(apache-like) and under GPL and LGPL.... each time I pick a different
license for a very specific reason, and occasionally specifically
because it is incompatible.

Not to get too deep into a GPL vs Apache debate, but its a good example
of where the philosophies for open-source clash. The market allowed for
divergent thinking on this topic (copyleft vs bsd open)... but a
centralized government opendata license wont allow for that diversity.
If the govt picks copyleft, all is copyleft, if they pick BSD-style
openness, its all open and maybe no one contributes back.

I'm confident in saying there's simply no 'right way' to license data or
code... and that's why we need diversity that can recognize specific
circumstances and let publishers choose what is appropriate for them...
and all without resorting to custom licenses which confuse the subject.
Yes, some licenses will be incompatible with other licenses, so those
choices must be made with careful consideration.

>
> I don't know specifically what you are referring to, and IANAL.
> However my understanding about the lack of a real public domain, at
> least with code, is that it is a question of responsibility. If you
> write code that causes harm you cannot avoid responsibility by saying,
> "I put it in the public domain". You can, however, minimize your
> exposure by having a license with specific terms that state that you
> are not responsible for some types of harm. So the problem with the
> public domain is that it opens you up to liability.

Ya, the PDDL is very similar to an Apache/BSD-style license, providing
maximal rights while disclaiming responsibilities. ( See
http://opendefinition.org/licenses/odc-pddl/ )

--

Kevin

[Aaron Hall]

anyone coming to this council meeting? to hear marianne alto? i am front row...

>>>>>> www.eaves.ca <http://www.eaves.ca>

>>>>>> @daeaves
>>>>>> Sent from my iPhone
>>>>>>
>>>>>> On 2011-10-05, at 5:17 PM, James McKinney <oxford...@gmail.com

>>>>>>>>>>> * use of standard licenses is preferred over custom

>>>>>>>>>>> licenses ( PDDL, ODC-by, ODbl, CCZero, CCby, )
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> 2) The use of standard licenses should be thought of like the
>>>>>>>>>> use of a logo program.
>>>>>>>>>>
>>>>>>>>>> If you create your own program, no one knows or cares about it
>>>>>>>>>> -- but if hundreds of data providers are using the same logos
>>>>>>>>>> then they begin to create standard value. Adoption will be
>>>>>>>>>> driven only by common understanding of the licensing choices
>>>>>>>>>> made, and if you create a custom license, then developers will
>>>>>>>>>> be expected to have their own council review your terms, and
>>>>>>>>>> that's an unrealistic expectation of the marketplace and will
>>>>>>>>>> hurt serious adoption.
>>>>>>>>>>
>>>>>>>>>>

>>>>>>>>>>> * revokability ( Kevin, can you flush this one out for us? )

>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> 3) The revokability provisions of open licenses should be
>>>>>>>>>> removed.
>>>>>>>>>>
>>>>>>>>>> This point stems from discussion at the hackathon on the
>>>>>>>>>> revokability of licenses and I'm happy to lend my $0.02.
>>>>>>>>>>
>>>>>>>>>> I've reviewed the BC OGL license and found that its generally
>>>>>>>>>> very good, but I believe it has an issue around revokability
>>>>>>>>>> in its clause 7.
>>>>>>>>>>
>>>>>>>>>> Clause 7 states:
>>>>>>>>>>

>>>>>>>>>>> *

>>>>>>>>>>> 7.
>>>>>>>>>>> Despite any rights granted herein, this license does not
>>>>>>>>>>> apply to the use of:

>>>>>>>>>>> o
>>>>>>>>>>> a.
>>>>>>>>>>> Personal Information;
>>>>>>>>>>> o

>>>>>>>>>>> b.
>>>>>>>>>>> information or Records that are not accessible
>>>>>>>>>>> pursuant to the Freedom of Information and Protection
>>>>>>>>>>> of Privacy Act (B.C.)

>>>>>>>>>>> <http://www.bclaws.ca/EPLibraries/bclaws_new/document/ID/freeside/96165_00>;
>>>>>>>>>>> o

>>>>>>>>>>> c.
>>>>>>>>>>> third-party rights the Information Provider is not
>>>>>>>>>>> authorized to license; and

>>>>>>>>>>> o

[David Eaves]

Again, there is a misunderstanding of the nature of parliament. All those other organizations you mention while quasi independent do report into a minister and have an accountability to the government. Parliament does not. As a result the government can order requiring Canada post to use the ogl. It cannot order parliament to do so. The governance of parliament is totally independent of the government. 

But we are way off topic. I'm not hearing of any data sets the ogl could not apply to.

--

www.eaves.ca

@daeaves

Sent from my iPhone

[Kevin McArthur]

Yes. Im at the meeting. 

--

K

[Dan Pollock]

Me too. 

The meeting is running a little long, but they should get to the Open Data proposal soon. 

Dan

[Herb Lainchbury]

So, to summarize, I am hearing:

• we're all agreed public domain is ideal
• Dave says it's never going to happen because the federal government will never agree to a license they don't control or, in the case of provinces, that isn't controlled by a higher level of government.  
• we don't have agreement on whether one license is better than more than one approved license (sauron's ring vs simplicity argument)
• we are having trouble agreeing on what a less-than-ideal license would look like, but we have the BC OGL, which we all think is pretty good

The question then is, what can be done to improve the OGL so that it's as good as we can imagine, given the constraints that it's going to be defined by the government?

I have four suggestions:

1. Clarify the attribution clause.  Right now it seems ambiguous. I am not sure if I am required to attribute or not.  It looks like it's left up to my judgement, if so then it's not a requirement so why dress it up like a requirement?
   
   
2. Remove the revokability.  If the data is out there then it's out there.  it even says in clause 2. that it's "perpetual" so, that would also make it consistent with itself.  (not sure what's required to do this but we can have a separate discussion about that)
   
   
3. remove all references to BC or British Columbia or Acts of British Columbia and put them into the Definitions section so that the main body of the license is re-usable in any jurisdiction.
   
   
4. get the feds to adopt the new improved OGL 1.1 (not BC)

Thoughts on that?

[Kevin McArthur]

Thoughts on that?

As usual herb, you've summed up the issues very succinctly and better than the rest of it. I'd get behind an improved OGL if your four suggestions could be addressed.

--

Kevin