Choosing a license for open data. What should BC consider?
David Hume
Director, Citizen Engagement for the BC provincial government. A broad
team of us are thinking through what an open data strategy for the
Province could look like. I was wondering if you could help us answer
a question we’ve been working on.
In researching what other jurisdictions are doing around adopting
open data, one of the things that’s clear is the massive importance of
the licensing regime.
In talking and reading about open data licenses, the principles that
look most important are:
• that data be available for without cost
• that creators can repurpose the data, including for commercial
purposes
• that the risks of use are clear, in that data is provided with no
guarantees or warranty, and that everyone is responsible for the legal
upshots of their own use of the data
• that the license itself is easy to understand
So best practice is a broad license for data that is permissive,
simple to understand, and clear so that everyone can manage risks,
legal and otherwise.
When you canvass the options out there, a few rise to the top.
1. The Public Domain Waiver and Dedication License(http://
www.opendatacommons.org/licenses/pddl/)—these tools, created by
Creative Commons and Open Data Commons, make data public domain and
attempt to remove any rights the owner has over that data forever, and
where that is not possible, grant a broad license. As far as we know,
a public domain waiver/dedication like this one hasn’t been tested in
Canada. Given that other broad licensing options are available that
are better known and more consistent with the Province’s approach to
licensing, PDDL doesn’t look attractive.
2. Creative Commons licenses (http://creativecommons.org/licenses/)—
these can also be very broad licenses, with options for certain types
of restrictions (such as attribution, share alike and non-commercial
use). These licenses have the well-recognized Creative Commons branding
—a big advantage in helping people work with the license. The most
attractive version is Creative Commons 3.0 http://creativecommons.org/licenses/by/3.0/legalcode.
The summaries are concise and helpful, but the full versions of their
licenses are very wordy and quite hard to understand. These licenses
are geared towards a broad spectrum of copyright-protected works, not
just data.
3. Open Data Commons (http://www.opendatacommons.org/licenses/by/
1.0/)---this license is also well recognized and specifically focused
on data, so that’s a great help. But again, the full text of the ODC
license is legalistic and hard to understand. It also has a minor
restriction in terms of requiring attribution in the use of the data.
4. Draft our own license—Vancouver (http://data.vancouver.ca/
termsOfUse.htm) and the UK (http://www.nationalarchives.gov.uk/doc/
open-government-licence/) have drafted their own licenses. The
advantage of drafting our own license is that we could make the full
version of the license be as plain language as possible. The UK has
even included a Creative Commons like summary to help cut to the
chase. We could also make the license more permissive by waiving the
need for attribution, like Vancouver has done. The downside is that
our own license may not clearly harmonize with other licenses around
the world, which could hinder data users who want to bring data
together from different jurisdictions. In addition our license would
become harder to recognize since we wouldn not be using a well known
brand.
So, the Creative Commons licenses, Open Data Commons license and one
we might create ourselves are likely to be pretty close from a legal
and permission perspective. Each have their strengths and weaknesses.
The question is, what’s the best way to go? Would you rather see
something branded, but complicated (e.g. ODC and Creative Commons)? Or
something simple to understand, similar to the well-known open
licenses, but unrecognized? Ultimately, what license would you prefer
to work with?
We’d really appreciate any feedback you have, including options we
might have missed. Thanks!
Kevin McArthur
I'm guessing you're aware of the scoring system we came up with at the
last hackathon. Herb posted that to the list today and its worth
reading. The group came to the conclusion that Vancouver's custom
license would score poorly vs the PDDL/CC0 system -- take from that what
you will.
For my $0.02... developers are used to licensing, but we're not lawyers.
Our understanding of the license will largely be shaped based on group
consensus and evaluation. Eg can you do this/that, has anyone ever been
C&D' over it, how does it apply etc... I don't know many developers who
could really parse all the intricacies of any data license, fewer who
would have a lawyer review one before use, and in the end it doesn't matter.
In the open source world we've got lots of licenses, but we also have
trusted authorities. For example, the OSI (Open Source Initiative) has a
review process, found here http://www.opensource.org/approval , for a
license to be considered open source by their definition. It's not a
legal process per-se, but if a license gets stamped OSI approved -- then
I know what I need to know, and have no questions about using it. I
don't need to know the intricacies of how MIT differes from BSD or APL
etc, so long as I plan to conform to their definition of open source,
I'm ready to code.
Since we don't have an Open-Data OSI, the best we can do is work with
licenses that are well known -- as such, custom pieces like the
Vancouver license just wont work for a lot of us. Especially when they
start throwing around language about covering court costs for downstream
rights violators and such. That may be totally reasonable, but I'm not a
lawyer and I'm not going to use a license that I don't understand the
implications of.
I guess thats a long way of saying, it doesn't matter if the PDDL/CC0
are complex licenses themselves, if we know their attributes through
community discussion, the legalese can largely be left to the lawyers.
Any custom license, unless it gets the backing of a major entity and the
community is willing to parse its meaning, will therefore have a harder
time attracting developers -- which leaves us with the scoring values
from the last hackathon and Vancouver somewhat out in the cold.
Hope that helps....
--
Kevin McArthur
On 11-02-10 05:48 PM, David Hume wrote:
> For those of you who don�t know me, I�m David Hume, Executive
> Director, Citizen Engagement for the BC provincial government. A broad
> team of us are thinking through what an open data strategy for the
> Province could look like. I was wondering if you could help us answer
> a question we�ve been working on.
>
> In researching what other jurisdictions are doing around adopting
> open data, one of the things that�s clear is the massive importance of
> the licensing regime.
>
> In talking and reading about open data licenses, the principles that
> look most important are:
> � that data be available for without cost
> � that creators can repurpose the data, including for commercial
> purposes
> � that the risks of use are clear, in that data is provided with no
> guarantees or warranty, and that everyone is responsible for the legal
> upshots of their own use of the data
> � that the license itself is easy to understand
>
> So best practice is a broad license for data that is permissive,
> simple to understand, and clear so that everyone can manage risks,
> legal and otherwise.
>
> When you canvass the options out there, a few rise to the top.
>
> 1. The Public Domain Waiver and Dedication License(http://
> www.opendatacommons.org/licenses/pddl/)�these tools, created by
> Creative Commons and Open Data Commons, make data public domain and
> attempt to remove any rights the owner has over that data forever, and
> where that is not possible, grant a broad license. As far as we know,
> a public domain waiver/dedication like this one hasn�t been tested in
> Canada. Given that other broad licensing options are available that
> are better known and more consistent with the Province�s approach to
> licensing, PDDL doesn�t look attractive.
>
> 2. Creative Commons licenses (http://creativecommons.org/licenses/)�
> these can also be very broad licenses, with options for certain types
> of restrictions (such as attribution, share alike and non-commercial
> use). These licenses have the well-recognized Creative Commons branding
> �a big advantage in helping people work with the license. The most
> attractive version is Creative Commons 3.0 http://creativecommons.org/licenses/by/3.0/legalcode.
> The summaries are concise and helpful, but the full versions of their
> licenses are very wordy and quite hard to understand. These licenses
> are geared towards a broad spectrum of copyright-protected works, not
> just data.
>
> 3. Open Data Commons (http://www.opendatacommons.org/licenses/by/
> 1.0/)---this license is also well recognized and specifically focused
> on data, so that�s a great help. But again, the full text of the ODC
> license is legalistic and hard to understand. It also has a minor
> restriction in terms of requiring attribution in the use of the data.
>
> 4. Draft our own license�Vancouver (http://data.vancouver.ca/
> termsOfUse.htm) and the UK (http://www.nationalarchives.gov.uk/doc/
> open-government-licence/) have drafted their own licenses. The
> advantage of drafting our own license is that we could make the full
> version of the license be as plain language as possible. The UK has
> even included a Creative Commons like summary to help cut to the
> chase. We could also make the license more permissive by waiving the
> need for attribution, like Vancouver has done. The downside is that
> our own license may not clearly harmonize with other licenses around
> the world, which could hinder data users who want to bring data
> together from different jurisdictions. In addition our license would
> become harder to recognize since we wouldn not be using a well known
> brand.
>
> So, the Creative Commons licenses, Open Data Commons license and one
> we might create ourselves are likely to be pretty close from a legal
> and permission perspective. Each have their strengths and weaknesses.
>
> The question is, what�s the best way to go? Would you rather see
> something branded, but complicated (e.g. ODC and Creative Commons)? Or
> something simple to understand, similar to the well-known open
> licenses, but unrecognized? Ultimately, what license would you prefer
> to work with?
>
> We�d really appreciate any feedback you have, including options we
> might have missed. Thanks!
David Eaves
My own feeling is that the PDDL is actually the best license on offer.
It is, in fact, the only license that has been adopted by a Canadian
jurisdiction in regards to today so far (the city of Surrey).
The problem with CC-0 is that while as effective as the PDDL I think it
is a deeply problematic license from a citizen education perspective.
Although few people are aware of this, data cannot be copyrighted and,
thanks to the create work by creative commons, CC is seen as an
alternative to copyright, this applying a CC license to data helps
propagate the misinformation the data can be copyrighted, something that
I think sets the open data movement back.
I would encourage you not to go with a license the province creates
itself as this will be the one least understood by developers. More
importantly, by adopting a standard license like PDDL, you'll be
encouraging other jurisdictions to do the same and what I think
developers and data users would really like is a common license across
the country as this makes their lives much, much, much easier.
cheers,
dave
Herb Lainchbury
Herb
David Blades
Jason Birch
David Eaves
Completely agree with you on ODBL (have been meaning to write a blog post on licenses... this prompting me to get on with it). The viral nature of this license makes it a non-starter for most commerical use, which, in my opinion defeats a big chunk of the point of the open data movement.
dave
On 11-02-10 9:12 PM, Jason Birch wrote:
Hi David,
As Dave said, relying on Copyright (anything above CC-0 in the Creative Commons suite) for data is potentially on shaky legal ground. �Though there may be cases where you could argue that the data arrangement uses non-obvious techniques, I personally wouldn't want to try to make this argument for most datasets.
In my opinion, custom licenses are an absolute non-starter. �They may do a better job of covering the government's collective�derri�re, but at the cost of making the data unapproachable by any developer in their right mind (or without gobs of money to spend on legal advice). �I think Kevin said this best.
I personally don't like the ODB-BY and ODBL licenses because, apart from the viral nature of ODBL and some of the technical difficulties caused by requiring attribution (try showing datasets from six providers on the screen and leaving room for the data), �they carry some of the same "cost" of the custom licenses. �The interior content license is unpredictable; you can use the stock content license, or can just provide your own version, but in all cases the developer will need to read it to find out. �This creates a less-than-instant acceptance of the license. �I recognize that there is sometimes a requirement to identify external rights in the content published under open license (say, in documents containing graphics licensed from external sources) but these should be dealt with on a per-dataset basis.
If you haven't already, I would STRONGLY recommend reading a recent draft of CIPPIC's�Open Licensing and Risk Management: �A Comparison of the City of Ottawa Open License, the ODC-By License and the ODC-PDDL License. available�from�http://www.cippic.ca/open-licensing/��This document does an amazing job of outlining the legal risks associated with each of the licenses in understandable language, and provides a lot of background and discussion around data licensing in general. �This is the best treatment of the topic I have yet seen, and has replaced the GeoConnections geographic data�dissemination best practices document in my recommended reading.
In my mind, and PERSONAL opinion only, the PDDL is the only rational choice for government open data. �I understand the fears behind using PDDL, I truly do. �Once you release it, the data is absolutely free; you have no control over how it is used, and that's kinda scary. �If anyone does something stupid with the data that causes you harm, you will have to find some other means of dealing with it than revoking the license. �However, it's not like public domain has never been used for government data before. �In fact, in the United States EVERY work produced by a federal government employee, whether it's raw data or something potentially covered by copyright, is automatically in the public domain.
Obviously, "selling" a license to any organisation takes a lot of work. �Public servants are generally risk-averse, and the idea of putting something out there and potentially not getting credit for it may be a bit hard to swallow (especially when it comes time to justify budgets). However, we need to balance the inherent risks of the various licenses against the risk we are assuming every day by NOT providing our citizens with the best, most accurate data available. �We are risking poor decisions that could have economic and social costs. �We are risking placing smaller local businesses at a competitive disadvantage against companies that can afford to create their own data. �We are risking a reversal of the burgeoning goodwill that closer communications and community building around open data seem to be creating.
In short, making the right decision in this case will mean managing risk--rather than try to eliminate it--to achieve an optimal return from your investment in open data. I hope it is possible�for the BC Government to make this kind of decision; it would absolutely be easier for other levels of government to follow in your footsteps than to break trail on their own.
Jason
(These views are my own and written as an independent citizen; they may not be shared by my employer)
On 10 February 2011 17:48, David Hume wrote:
For those of you who don�t know me, I�m David Hume, Executive
Director, Citizen Engagement for the BC provincial government. A broad
team of us are thinking through what an open data strategy for the
Province could look like. I was wondering if you could help us answer
a question we�ve been working on.
In researching what other jurisdictions are doing around adopting
open data, one of the things that�s clear is the massive importance of
the licensing regime.
In talking and reading about open data licenses, the principles that
look most important are:
� � � � that data be available for without cost
� � � � that creators can repurpose the data, including for commercial
purposes
� � � � that the risks of use are clear, in that data is provided with no
guarantees or warranty, and that everyone is responsible for the legal
upshots of their own use of the data
� � � � that the license itself is easy to understand
So best practice is a broad license for data that is permissive,
simple to understand, and clear so that everyone can manage risks,
legal and otherwise.
When you canvass the options out there, a few rise to the top.
1. � � �The Public Domain Waiver and Dedication License(http://
www.opendatacommons.org/licenses/pddl/)�these tools, created by
Creative Commons and Open Data Commons, make data public domain and
attempt to remove any rights the owner has over that data forever, and
where that is not possible, grant a broad license. As far as we know,
a public domain waiver/dedication like this one hasn�t been tested in
Canada. Given that other broad licensing options are available that
are better known and more consistent with the Province�s approach to
licensing, PDDL doesn�t look attractive.
2. � � �Creative Commons licenses (http://creativecommons.org/licenses/)�
these can also be very broad licenses, with options for certain types
of restrictions (such as attribution, share alike and non-commercial
use). These licenses have the well-recognized Creative Commons branding
�a big advantage in helping people work with the license. The most
attractive version is Creative Commons 3.0 http://creativecommons.org/licenses/by/3.0/legalcode.
The summaries are concise and helpful, but the full versions of their
licenses are very wordy and quite hard to understand. �These licenses
are geared towards a broad spectrum of copyright-protected works, not
just data.
3. � � �Open Data Commons (http://www.opendatacommons.org/licenses/by/
1.0/)---this license is also well recognized and specifically focused
on data, so that�s a great help. But again, the full text of the ODC
license is legalistic and hard to understand. It also has a minor
restriction in terms of requiring attribution in the use of the data.
4. � � �Draft our own license�Vancouver (http://data.vancouver.ca/
termsOfUse.htm) and the UK (http://www.nationalarchives.gov.uk/doc/
open-government-licence/) have drafted their own licenses. The
advantage of drafting our own license is that we could make the full
version of the license be as plain language as possible. The UK has
even included a Creative Commons like summary to help cut to the
chase. We could also make the license more permissive by waiving the
need for attribution, like Vancouver has done. The downside is that
our own license may not clearly harmonize with other licenses around
the world, which could hinder data users who want to bring data
together from different jurisdictions. In addition our license would
become harder to recognize since we wouldn not be using a well known
brand.
So, the Creative Commons licenses, Open Data Commons license and one
we might create ourselves are likely to be pretty close from a legal
and permission perspective. Each have their strengths and weaknesses.
The question is, what�s the best way to go? Would you rather see
something branded, but complicated (e.g. ODC and Creative Commons)? Or
something simple to understand, similar to the well-known open
licenses, but unrecognized? Ultimately, what license would you prefer
to work with?
We�d really appreciate any feedback you have, including options we
might have missed. Thanks!
lukec
this data and building the Applications Of Tomorrow™.
My preference is the PDDL - I believe it will allow the most value to
be extracted from the data as possible.
On Feb 10, 6:21 pm, Kevin McArthur <ke...@stormtide.ca> wrote:
> Hey David,
>
> I'm guessing you're aware of the scoring system we came up with at the
> last hackathon. Herb posted that to the list today and its worth
> reading. The group came to the conclusion that Vancouver's custom
> license would score poorly vs the PDDL/CC0 system -- take from that what
> you will.
>
> For my $0.02... developers are used to licensing, but we're not lawyers.
> Our understanding of the license will largely be shaped based on group
> consensus and evaluation. Eg can you do this/that, has anyone ever been
> C&D' over it, how does it apply etc... I don't know many developers who
> could really parse all the intricacies of any data license, fewer who
> would have a lawyer review one before use, and in the end it doesn't matter.
>
> In the open source world we've got lots of licenses, but we also have
> trusted authorities. For example, the OSI (Open Source Initiative) has a
David Hume
Have a bunch of questions that I'm going think through a little
further over the weekend.
But one in particular I'd appreciate some clarity on.
Jason and Dave E--you both described the 'viral' nature of the ODC-By
license, and that this is a problem.
Can you say a little more about you mean?
Paul Ramsey
If you use the data and add to it, you have to share your additions under the same license. So, once open, always open, no matter who takes it. It lowers the likelihood of people taking the data and building value-added on it, because the added value will also have to be freely released.
In theory share-alike is all dancing fairies and sugar muffets, but in practice it freezes out a section of the community that can sometimes to very useful things with the data.
P.
David Eaves
It basically kills the business communities interest in the open data. It also has a negative impact for some non-profits that may want to create mashups but have sensitive data they don't want to(or can't legally) share.
Dave
--
www.eaves.ca
@daeaves
Sent from my iPhone
Jason Birch
This is a different license than ODC-By.
I'm not positive, but I don't believe ODC-By has the same viral
sharealike component. It does still have the technical hurdle of
attribution.
This might not seem like a big deal, but it's a royal pain to deal
with if you use more than one data source in your application or are
working ona a small-screen app. It also adds uncertainty/fear to the
use of the data. Developers have no way of knowing for sure if they
have chosen to attribute the source "properly".
Jason
Jason Birch
But, how do you anticipate all uses and acceptable forms of
attribution for all of them? Setting a community norm of attribution
and letting developers choose how to implement is far less
problematic.
Jason
David Eaves
1) where are people supposed to attribute the province? Especially on a mobile app. Worse still, what happens when I mash up data from all 10 provinces and say 10 cities, how to I attribute each of them? It is going to get really messy really fast.
2) It's not clear to me that the province is always going to want to be attributed. If I use the data to create a mashup that is embarrassing to the province do you really want to insist that I attribute?
David Hume
I’ve been giving some thought to how to respond to the preference the
group’s expressed for the PDDL. It’s also lead me to some questions
about the licensing game generally, and how to best meet everyone’s
interests over time.
Let me start by saying I appreciate the intent of why people like the
PDDL. It’s simple to understand and provides as much freedom as
possible to users of the data, which hopefully leads to more
engagement with the data.
But you may have noticed that, in my original note, I labelled the
PDDL ‘not attractive’. I reckon a more detailed explanation of why
might lead to some productive thinking and discussion from all of us.
Jason—your note that included the CIPPIC paper on open licensing and
risk management was incredibly helpful. Reading it, and in thinking
about the PDDL, one section on the bottom of page 5 stood out in
spades for me:
“The [Copyright] Act does not provide for assignment of a work into
the public domain or for waiving of rights in the way that the ODC-
PDDL license sets out. ..The ODC-PDDL license’s ‘dedication to the
public domain can be equated to an ‘assignment’ to the public domain
though there is caution here in that the Copyright Act does not
contemplate a ‘public domain’ and whether such an assignment is
possible has not been tested in the courts.”
Provided this is correct, I’d want to advance the argument that given
a choice between a conventional licensing approach (e.g. the ODC-By or
UK Open Government License) versus a license that hasn’t been tested
in court, like the PDDL, I think you’d want to go the conventional
route . It seems to me this would be especially good for users of the
data, otherwise everyone is operating under circumstances that could
radically change should a court challenge show the license doesn’t
hold up..
This leads me to one other point about PDDL. The CIPPIC paper points
out that if something bad happens with the data—such as a violation of
personal privacy or someone produces something that’s illegal—then the
data provider must use “other areas of law to terminate the user’s
access to the data” (page 13). Again, provided this is right, this
would seem to me to create a significant risk management burden in
that the data provider would now have to get specific about how it
would respond to various scenarios. The planning for that, involving a
lot of advice from lawyers, privacy professionals and others, would
get onerous and expensive quickly. If your data holdings are small,
this is not such a tough exercise. But if they’re large (as in the
thousands) and start touching on sensitive issues like health and
public safety, I think it becomes really difficult.
I recognize that the US is already putting its data into the public
domain. But they’ve also got a whole policy and legal infrastructure
that can help account for the risks. In Canada, we’d have to build it
from scratch. I know Surrey has adopted the PDDL, and I’d love to know
how they’ve accounted for these questions. I might be missing
something, after all. If someone has a contact there, I’d love to be
in touch with them.
This brings me to my question about the licensing game in general. It
seems to me that whatever license option we choose, there are
weaknesses. Either attribution ties up innovation, or custom licenses
cause confusion, or there is some other issue. The best strategy is
likely to pick one for the right reasons, and then work to evolve it
into a better place over time. I wonder, does anyone have any good
information or contacts about how licenses evolve? And what’s good
practice for helping that evolution happen? I know the G4/5 are
talking licenses right now. I’m certainly interested in the outcome of
those discussions. I’d also appreciate any advice you all have about
how a provincial government like BC could play a productive role in
addressing the data licensing challenge for Canada.
Cheers,
David H
@dbhume
On Feb 12, 3:14 pm, David Eaves <da...@eaves.ca> wrote:
> yeah, I think the whole attribution thing is not worth it.
>
> 1) where are people supposed to attribute the province? Especially on a
> mobile app. Worse still, what happens when I mash up data from all 10
> provinces and say 10 cities, how to I attribute each of them? It is
> going to get really messy really fast.
>
> to the province do you really want to insist that I attribute?
>
> On 11-02-12 3:01 PM, Jason Birch wrote:
>
> > If you require attribution in your license, then it's critical.
>
> > But, how do you anticipate all uses and acceptable forms of
> > attribution for all of them? Setting a community norm of attribution
> > and letting developers choose how to implement is far less
> > problematic.
>
> > Jason
>
>
> read more »
Kevin McArthur
I'll try to respond to a few items inline below.
As usual, I-am-not-a-lawyer-this-is-not-legal-advice and only good for
my $0.02 ;)
--
Kevin
On 11-02-16 07:56 PM, David Hume wrote:
> Hey all,
>
> I�ve been giving some thought to how to respond to the preference the
> group�s expressed for the PDDL. It�s also lead me to some questions
> about the licensing game generally, and how to best meet everyone�s
> interests over time.
>
> Let me start by saying I appreciate the intent of why people like the
> PDDL. It�s simple to understand and provides as much freedom as
> possible to users of the data, which hopefully leads to more
> engagement with the data.
>
> But you may have noticed that, in my original note, I labelled the
> PDDL �not attractive�. I reckon a more detailed explanation of why
> might lead to some productive thinking and discussion from all of us.
>
> Jason�your note that included the CIPPIC paper on open licensing and
> risk management was incredibly helpful. Reading it, and in thinking
> about the PDDL, one section on the bottom of page 5 stood out in
> spades for me:
>
> �The [Copyright] Act does not provide for assignment of a work into
> the public domain or for waiving of rights in the way that the ODC-
> PDDL license sets out. ..The ODC-PDDL license�s �dedication to the
> public domain can be equated to an �assignment� to the public domain
> though there is caution here in that the Copyright Act does not
> contemplate a �public domain� and whether such an assignment is
> possible has not been tested in the courts.�
Even should it be found that there's no legal assignment to the public
domain in Canadian copyright law, the PDDL, as I understand it, will act
as a copyright release. Certainly that is the intent, and the only
entity that could enforce their copyright should the concept of a public
domain be found wanting, is simply the government itself. Is your
concern that some time in the future that governments will try to reneg
on the PDDL and open lawsuits as if the public was never granted the data?
> Provided this is correct, I�d want to advance the argument that given
> a choice between a conventional licensing approach (e.g. the ODC-By or
> UK Open Government License) versus a license that hasn�t been tested
> in court, like the PDDL, I think you�d want to go the conventional
> route . It seems to me this would be especially good for users of the
> data, otherwise everyone is operating under circumstances that could
> radically change should a court challenge show the license doesn�t
> hold up..
I don't agree that ODB-By or a UK custom license is the 'conventional'
approach. From my perspective, 'conventional' would be BSD, MIT, GPL,
etc licenses... which aren't really great for pure data, but that are
often used anyway.
As for a court challenge... I'm not sure why the government would launch
such a challenge... this seems like an extremely fringe use case and not
one that would concern me. If an outside entity challenged the license
in some way, it stands to reason that like other licenses before it, the
license would simply be updated and those who wish to take advantage of
the new terms would simply relicense under the new version (see GPL V2
vs V3 which was largely rewritten to address software patents and
tivoization)
> This leads me to one other point about PDDL. The CIPPIC paper points
> out that if something bad happens with the data�such as a violation of
> personal privacy or someone produces something that�s illegal�then the
> data provider must use �other areas of law to terminate the user�s
> access to the data� (page 13). Again, provided this is right, this
> would seem to me to create a significant risk management burden in
> that the data provider would now have to get specific about how it
> would respond to various scenarios. The planning for that, involving a
> lot of advice from lawyers, privacy professionals and others, would
> get onerous and expensive quickly. If your data holdings are small,
> this is not such a tough exercise. But if they�re large (as in the
> thousands) and start touching on sensitive issues like health and
> public safety, I think it becomes really difficult.
This is an extremely important issue to not mess up. What you don't
handle on the government side, you are therefore expecting the developer
to handle on the implementation side. If your license tries to make me
responsible for the end-use for doing something as simple as
redistributing a data set or building a tool with your data, then your
license is severely broken. I don't agree that the government has to do
a big risk management task though -- as for 'do something illegal', last
I checked that was a matter for the justice system not the data agency.
For years the US govt was terrified that people would make guided
weapons out of the GPS system, and prevented its release, and after
that, restricted its accuracy for a long time. Now its arguably one of
our most important technologies. There's no license to use the GPS
system that says, you wont make a guided missile and use the GPS system.
Theres no license that says if I make a phone and someone uses it in a
missile, that I'm responsible for that downstream use. Free the data,
and take with that what innovation and consequences come of it.
> I recognize that the US is already putting its data into the public
> domain. But they�ve also got a whole policy and legal infrastructure
> that can help account for the risks. In Canada, we�d have to build it
> from scratch. I know Surrey has adopted the PDDL, and I�d love to know
> how they�ve accounted for these questions. I might be missing
> something, after all. If someone has a contact there, I�d love to be
> in touch with them.
Hold up there.... Canada very much has the concept of a public domain if
not a legal construct to which works can be assigned. Works will fall
out of copyright even under our ridiculously long copyright terms. Where
are they then? If not a legal concept, as a social construct the 'public
domain' around here simply translates to 'not owned' or 'no rights to
enforce' ... which is good enough for any data user. My understanding of
the PDDL is that it acts as a waiver to any such copy 'rights' the
government may otherwise have. What is the concern here?
> This brings me to my question about the licensing game in general. It
> seems to me that whatever license option we choose, there are
> weaknesses. Either attribution ties up innovation, or custom licenses
> cause confusion, or there is some other issue. The best strategy is
> likely to pick one for the right reasons, and then work to evolve it
> into a better place over time. I wonder, does anyone have any good
> information or contacts about how licenses evolve? And what�s good
> practice for helping that evolution happen? I know the G4/5 are
> talking licenses right now. I�m certainly interested in the outcome of
> those discussions. I�d also appreciate any advice you all have about
> how a provincial government like BC could play a productive role in
> addressing the data licensing challenge for Canada.
David Eaves, yesterday posted on open data licensing....
http://eaves.ca/2011/02/16/the-state-of-open-data-in-canada-the-year-of-the-license/
... unsurprisingly his support is for the PDDL as well.
>> read more �
David Eaves
Know you are working hard to balance the interests of the province with those of maximizing innovation, re-use and the public interest. Thank you for consulting with us. A few thoughts in response:
First, it is unclear to me that any of the "conventional" licenses have been tested in a court either, so there are risks surrounding them as well.
Second, I'm not sure I get your point about the PDDL and copyright. The only party that can seek to exercise their copyright over material and pull it out of the public domain would be the government. What you are essentially saying is that the PDDL doesn't guarantee that something is in the public domain and that the courts could, in theory, limit the scope of the license.
But the risk here isn't from the courts. It is from the government. They, and only they, could initiate such a court case.
Under the other licenses, may not even need to do that. They could just cut a user off. I also see nothing in the other licenses that would preclude the negative outcome you attribute to the PDDL - the other licenses cannot guarantee something is in the public domain nor do they guarantee that, in the future, the goverment won't suddenly decide to exercise their right over the data. (Indeed, only by putting something in the public domain could one guarantee such an outcome, which is why it is worth endeavoring to do so). With all the other licenses, if government was really set on exercising this right it could simply change the license. So I'm not sure that the PDDL offers a weaker guarantee, indeed, the opposite, it offers an (legally untested) stronger guarantee that the government has actually put it in the public domain and cannot retract it.
In regards to the "we cannot revoke access" I find this interesting. Under any of these licenses the government can end access to everyone by simple retiring the data set. So there is always that method of managing risk. If someone is using the data to do something illegal, then let's prosecute them for the illegal activity, not for take away their access to the data. The reason is, I'm frankly not comfortable with letting the government decide who is and who isn't using the data appropriately.
In regards to evolving licenses, I'd be happy to put you in touch with the author(s) of the PDDL. I'm sure he'd be happy to talk with you. Maybe, at some point, there will be a standardizied Open Government License in Canada, like there is now in the UK (although, again, believe we are far better off using a standardized license). But, as far as I can tell, that day is at least 2 years (if not a lot, lot longer) out, so would hate for us to wait until it emerged.
Finally, it is worth noting that the CIPPIC report concludes, at the very beginning: "In general the three licenses pose very slight risks to licensors." So, part of my fear is that we can end up in heated (or pleasant academic!) discussions over the details (which we should explore) but at the end of the day, it is worth remembering that assessment, it would seem to me that that is an assessment for moving forward...
dave
On 11-02-16 7:56 PM, David Hume wrote:
Hey all, I�ve been giving some thought to how to respond to the preference the group�s expressed for the PDDL. It�s also lead me to some questions about the licensing game generally, and how to best meet everyone�s interests over time. Let me start by saying I appreciate the intent of why people like the PDDL. It�s simple to understand and provides as much freedom as possible to users of the data, which hopefully leads to more engagement with the data. But you may have noticed that, in my original note, I labelled the PDDL �not attractive�. I reckon a more detailed explanation of why might lead to some productive thinking and discussion from all of us. Jason�your note that included the CIPPIC paper on open licensing and risk management was incredibly helpful. Reading it, and in thinking about the PDDL, one section on the bottom of page 5 stood out in spades for me: �The [Copyright] Act does not provide for assignment of a work into the public domain or for waiving of rights in the way that the ODC- PDDL license sets out. ..The ODC-PDDL license�s �dedication to the public domain can be equated to an �assignment� to the public domain though there is caution here in that the Copyright Act does not contemplate a �public domain� and whether such an assignment is possible has not been tested in the courts.� Provided this is correct, I�d want to advance the argument that given a choice between a conventional licensing approach (e.g. the ODC-By or UK Open Government License) versus a license that hasn�t been tested in court, like the PDDL, I think you�d want to go the conventional route . It seems to me this would be especially good for users of the data, otherwise everyone is operating under circumstances that could radically change should a court challenge show the license doesn�t hold up.. This leads me to one other point about PDDL. The CIPPIC paper points out that if something bad happens with the data�such as a violation of personal privacy or someone produces something that�s illegal�then the data provider must use �other areas of law to terminate the user�s access to the data� (page 13). Again, provided this is right, this would seem to me to create a significant risk management burden in that the data provider would now have to get specific about how it would respond to various scenarios. The planning for that, involving a lot of advice from lawyers, privacy professionals and others, would get onerous and expensive quickly. If your data holdings are small, this is not such a tough exercise. But if they�re large (as in the thousands) and start touching on sensitive issues like health and public safety, I think it becomes really difficult. I recognize that the US is already putting its data into the public domain. But they�ve also got a whole policy and legal infrastructure that can help account for the risks. In Canada, we�d have to build it from scratch. I know Surrey has adopted the PDDL, and I�d love to know how they�ve accounted for these questions. I might be missing something, after all. If someone has a contact there, I�d love to be in touch with them. This brings me to my question about the licensing game in general. It seems to me that whatever license option we choose, there are weaknesses. Either attribution ties up innovation, or custom licenses cause confusion, or there is some other issue. The best strategy is likely to pick one for the right reasons, and then work to evolve it into a better place over time. I wonder, does anyone have any good information or contacts about how licenses evolve? And what�s good practice for helping that evolution happen? I know the G4/5 are talking licenses right now. I�m certainly interested in the outcome of those discussions. I�d also appreciate any advice you all have about how a provincial government like BC could play a productive role in addressing the data licensing challenge for Canada. Cheers, David H @dbhume On Feb 12, 3:14�pm, David Eaves <da...@eaves.ca> wrote:
yeah, I think the whole attribution thing is not worth it. 1) where are people supposed to attribute the province? Especially on a mobile app. Worse still, what happens when I mash up data from all 10 provinces and say 10 cities, how to I attribute each of them? It is going to get really messy really fast. 2) It's not clear to me that the province is always going to /want/ to be attributed. If I use the data to create a mashup that is embarrassing to the province do you really want to insist that I attribute? On 11-02-12 3:01 PM, Jason Birch wrote:If you require attribution in your license, then it's critical.
But, how do you anticipate all uses and acceptable forms of attribution for all of them? �Setting a community norm of attribution and letting developers choose how to implement is far less problematic.Jason
On 2011-02-12, Wrate, David CITZ:EX<David.Wr...@gov.bc.ca> �wrote:Interesting point on attribution Jason. So would you say that clear instructions on providing attribution is a plus?Cheers, David W.-----Original Message----- From: opend...@googlegroups.com on behalf of Jason Birch Sent: Sat 2/12/2011 10:13 AM To: opend...@googlegroups.com Subject: Re: [OpenDataBC] Choosing a license for open data. What should BC consider?I probably threw the discussion off track by introducing ODC-ODBL. This is a different license than ODC-By.I'm not positive, but I don't believe ODC-By has the same viral sharealike component. It does still have the technical hurdle of attribution.This might not seem like a big deal, but it's a royal pain to deal with if you use more than one data source in your application or are working ona a small-screen app. It also adds uncertainty/fear to the use of the data. Developers have no way of knowing for sure if they have chosen to attribute the source "properly".Jason
On 2011-02-11, David Hume<dbh...@yahoo.com> �wrote:Hey all--wanted to say again thanks for the ideas and advice.Have a bunch of questions that I'm going think through a little further over the weekend.But one in particular I'd appreciate some clarity on.Jason and Dave E--you both described the 'viral' nature of the ODC-By license, and that this is a problem.Can you say a little more about you mean?
On Feb 11, 1:16 am, lukec<lukecl...@gmail.com> �wrote:
Kevin has a bang-on description of how I feel as a developer using this data and building the Applications Of TomorrowT.
My preference is the PDDL - I believe it will allow the most value to be extracted from the data as possible.
On Feb 10, 6:21 pm, Kevin McArthur<ke...@stormtide.ca> �wrote:
For those of you who don�t know me, I�m David Hume, Executive Director, Citizen Engagement for the BC provincial government. A broad team of us are thinking through what an open data strategy for the Province could look like. I was wondering if you could help us answer a question we�ve been working on. In researching what other jurisdictions are doing around adopting open data, one of the things that�s clear is the massive importance of the licensing regime. In talking and reading about open data licenses, the principles that look most important are: � � � that data be available for without cost � � � that creators can repurpose the data, including for commercial purposes � � � that the risks of use are clear, in that data is provided with no guarantees or warranty, and that everyone is responsible for the legal upshots of their own use of the data � � � that the license itself is easy to understand So best practice is a broad license for data that is permissive, simple to understand, and clear so that everyone can manage risks, legal and otherwise. When you canvass the options out there, a few rise to the top. 1. The Public Domain Waiver and Dedication License(http://
. The summaries are concise and helpful, but the full versions of their licenses are very wordy and quite hard to understand. �These licenses are geared towards a broad spectrum of copyright-protected works, not just data. 3. Open Data Commons (http://www.opendatacommons.org/licenses/by/
... read more �
David Hume
Dave -- I appreciate you recognizing us for doing our best to strike
the right balance. It's definitely what we're trying to do. And the
feedback from you all has been really helpful. The questions you've
raised are challenging and helpful. Getting your perspective is
helping educate us and push our thinking along.
Curious to know, Dave, why you think we're two years out from a
standardized license in Canada? Why would we wait so long? What could
make it move faster?
>
> >> 1) where are people supposed to attribute the province? Especially on a
> >> mobile app. Worse still, what happens when I mash up data from all 10
> >> provinces and say 10 cities, how to I attribute each of them? It is
> >> going to get really messy really fast.
>
> >> 2) It's not clear to me that the province is always going to /want/ to
> >> be attributed. If I use the data to create a mashup that is embarrassing
> >> to the province do you really want to insist that I attribute?
>
> >> On 11-02-12 3:01 PM, Jason Birch wrote:
>
> >>> If you require attribution in your license, then it's critical.
> >>> But, how do you anticipate all uses and acceptable forms of
> >>> problematic.
> >>> Jason
> >>>> Interesting point on attribution Jason. So would you say that clear
> >>>> instructions on providing attribution is a plus?
> >>>> Cheers,
> >>>> David W.
> >>>> -----Original Message-----
> >>>> From: opend...@googlegroups.com on behalf of Jason Birch
> >>>> Sent: Sat 2/12/2011 10:13 AM
> >>>> To: opend...@googlegroups.com
> >>>> Subject: Re: [OpenDataBC] Choosing a license for open data. What should BC
> >>>> consider?
> >>>> I probably threw the discussion off track by introducing ODC-ODBL.
> >>>> This is a different license than ODC-By.
> >>>> I'm not positive, but I don't believe ODC-By has the same viral
> >>>> sharealike component. It does still have the technical hurdle of
> >>>> attribution.
> >>>> This might not seem like a big deal, but it's a royal pain to deal
> >>>> with if you use more than one data source in your application or are
> >>>> working ona a small-screen app. It also adds uncertainty/fear to the
> >>>> use of the data. Developers have no way of knowing for sure if they
> >>>> have chosen to attribute the source "properly".
> >>>> Jason
> >>>>> Hey all--wanted to say again thanks for the ideas and advice.
> >>>>> Have a bunch of questions that I'm going think through a little
> >>>>> further over the weekend.
> >>>>> But one in particular I'd appreciate some clarity on.
> >>>>> Jason and Dave E--you both described the 'viral' nature of the ODC-By
> >>>>> license, and that this is a problem.
> >>>>> Can you say a little more about
>
>
> read more »
Herb Lainchbury
David Eaves
I'm not optimistic of a national government standardized license as I'm
seeing very little leadership on this issue from the Feds and, in my
experience, trying to get governments to coordinate is an exceedingly
slow and painstaking process. On a file that involves open data and so
touches many departments, it feels like it would be even slower. This is
assuming, of course, that there is some negotiated process by which it
came about.
If however, there was a very good progressive license that emerged that
had a solid developer, non-profit and for profit community buy in, and
that different jurisdictions bandwagoned on... I think it could emerge
faster. But presently I don't see a license that looks like that. Pretty
much everyone agrees that the Vancouver needs to be abandoned for
something more effective so we're in a gray area.
This is why I'm excited that Surrey adopted the PDDL, I think it might
create some positive momentum in that direction. Adopting the UK Open
Government License would be an interesting possibility - and while not
as open as the PDDL - is still quite good. Would be interesting to know,
what, if any challenges, people might have with it.
Of course, if we were to adopt the UK Open Government License, it might
be interesting to immediately set up and organization to curate it. It
might be composed of majority government stakeholders but also with
representation from the developer, non-profit and for profit
stakeholders. That way the license wouldn't be exclusively in the hands
of one government, it might also help prevent it from being forked or
changed by different governments (Who would instead see joining this
body as a way to allow their interests to be heard) and it could also
engage the interests of non-government actors.
Food for thought.
Cheers,
dave
>>> I�ve been giving some thought to how to respond to the preference the
>>> group�s expressed for the PDDL. It�s also lead me to some questions
>>> about the licensing game generally, and how to best meet everyone�s
>>> interests over time.
>>> Let me start by saying I appreciate the intent of why people like the
>>> PDDL. It�s simple to understand and provides as much freedom as
>>> possible to users of the data, which hopefully leads to more
>>> engagement with the data.
>>> But you may have noticed that, in my original note, I labelled the
>>> PDDL �not attractive�. I reckon a more detailed explanation of why
>>> might lead to some productive thinking and discussion from all of us.
>>> Jason�your note that included the CIPPIC paper on open licensing and
>>> risk management was incredibly helpful. Reading it, and in thinking
>>> about the PDDL, one section on the bottom of page 5 stood out in
>>> spades for me:
>>> �The [Copyright] Act does not provide for assignment of a work into
>>> the public domain or for waiving of rights in the way that the ODC-
>>> PDDL license sets out. ..The ODC-PDDL license�s �dedication to the
>>> public domain can be equated to an �assignment� to the public domain
>>> though there is caution here in that the Copyright Act does not
>>> contemplate a �public domain� and whether such an assignment is
>>> possible has not been tested in the courts.�
>>> Provided this is correct, I�d want to advance the argument that given
>>> a choice between a conventional licensing approach (e.g. the ODC-By or
>>> UK Open Government License) versus a license that hasn�t been tested
>>> in court, like the PDDL, I think you�d want to go the conventional
>>> route . It seems to me this would be especially good for users of the
>>> data, otherwise everyone is operating under circumstances that could
>>> radically change should a court challenge show the license doesn�t
>>> hold up..
>>> This leads me to one other point about PDDL. The CIPPIC paper points
>>> out that if something bad happens with the data�such as a violation of
>>> personal privacy or someone produces something that�s illegal�then the
>>> data provider must use �other areas of law to terminate the user�s
>>> access to the data� (page 13). Again, provided this is right, this
>>> would seem to me to create a significant risk management burden in
>>> that the data provider would now have to get specific about how it
>>> would respond to various scenarios. The planning for that, involving a
>>> lot of advice from lawyers, privacy professionals and others, would
>>> get onerous and expensive quickly. If your data holdings are small,
>>> this is not such a tough exercise. But if they�re large (as in the
>>> thousands) and start touching on sensitive issues like health and
>>> public safety, I think it becomes really difficult.
>>> I recognize that the US is already putting its data into the public
>>> domain. But they�ve also got a whole policy and legal infrastructure
>>> that can help account for the risks. In Canada, we�d have to build it
>>> from scratch. I know Surrey has adopted the PDDL, and I�d love to know
>>> how they�ve accounted for these questions. I might be missing
>>> something, after all. If someone has a contact there, I�d love to be
>>> in touch with them.
>>> This brings me to my question about the licensing game in general. It
>>> seems to me that whatever license option we choose, there are
>>> weaknesses. Either attribution ties up innovation, or custom licenses
>>> cause confusion, or there is some other issue. The best strategy is
>>> likely to pick one for the right reasons, and then work to evolve it
>>> into a better place over time. I wonder, does anyone have any good
>>> information or contacts about how licenses evolve? And what�s good
>>> practice for helping that evolution happen? I know the G4/5 are
>>> talking licenses right now. I�m certainly interested in the outcome of
>>> those discussions. I�d also appreciate any advice you all have about
>> read more �