[od-discuss] OKD questions from 2010

Tue Feb 12 02:59:46 UTC 2013

http://blog.okfn.org/2010/08/04/update-on-open-source-initiatives-adoption-of-the-open-knowledge-definition/
lists some question, with answers in
http://okfnpad.org/okd-questions (copied below).

I think Q3 is the only one pertinent to a new version of the OKD, but
feel free to suggest otherwise. I don't have a strong feeling as to
whether the suggested language is more clear or otherwise better.

Existing:

    The rights attached to the work must apply to all to whom it is
redistributed without the need for execution of an additional
*license* by those parties.

Suggested:

    The rights attached to the work must apply to all to whom it is
redistributed without the need for execution of an additional
*agreement* by those parties.

(one word change highlighted)

I do think reaching out to OSI again for feedback would make sense
before releasing a new OKD version, but I'll leave guidance on that to
Luis Villa.

Mike
...

Please add your name in bold to any comments/questions below -- and
ensure you have a different colour from others!

Question 1. What happens with data that's not copyrightable?
Question 1a. What about data that consists of facts about the world
and thus even a collection of it cannot be copyrighted, but the exact
file format can be copyrighted?  Many sub-federal-level governments in
the US have to publish facts on demand but claim a copyright on the
formatting.

Jordan Hatcher: These are both great questions -- I think it helps to
separate out a few areas when talking about open data and the law:

## 1.  the data / database distinction.

I prefer to use the term "contents of a databasea" because lots of
databases will have clearly copyrightable material in them, such as
music (ex: mp3s), images (ex: Flickr photos), or video (ex: Internet
Archive stuff). When people talk about "data" they start automatically
thinking  about 'facts" and then really starting to question the
available rights  (and rightly so, as it is different -- facts aren't
copyrightable on  their own).

## 2.  legal questions on the boundaries of how the law (in various
jurisdictions) interacts with databases and their contents

There's a bunch of tricky legal questions around how databases and
their contents, particularly when those contents contain factual
information, interact with the law.  The law varies between
jurisdictions and so there's no one set global answer and just like in
 most areas of the law, these aren't binary yes/no questions.

Just like with open source, I think you have to leave many of these
questions to the side when trying to define a standard in the area
such  as what "open data" means.  It doesn't matter in regards to the
Open  Source Definition what the outcome is on the "is it a license or
a  contract?" debate lawyers get into.

Question 2. What about data that's not accessible as a whole, but only
through an API?

Jordan Hatcher: As a concrete example to this question, what would
open data mean  in the context of the results ("data" or "contents")
returned from  querying an API? I think to be "open data" then you
should have the rights to do  what you wish (within the open
definition scope) with the results of  that query. I don't think that
opening up the database that sits behind that  API should be required
to be distributed in order to be open data.

However as a corollary to the above, any license restriction that
prevented you from being able to get all the database through running
multiple queries would prevent that from being open data. Technical
restrictions (number of API calls, for example) would be okay.

Incidentally, the Open Database License (ODbL) would view public  API
access as distribution and so would trigger access to the full
database -- see 4.6 of the text
<http://www.opendatacommons.org/licenses/odbl/1.0/>

Question 3. We're thinking that OKD #9 should read "execution of an
additional agreement" rather than "additional license".

Jordan Hatcher: Seems to make sense. Is OSI considering a similar
change to the Open Source Definition?

Question 4. Does OKD #4 apply to works distributed in a particular
file format? Is a movie not open data if it's encoded in a
patent-encumbered codec? Does it become open data if it's re-encoded?

Jordan Hatcher: Though in the end, we're nothing but data (GATC in
DNA) I think it might be stretching it a bit as a practical matter to
consider an MPEG4  file movie as a typical example for open data.This
does bring up the point though about file formats, and it can be an
issue. I think that you can still have open data in a proprietary
format, as long as users aren't restricted from converting the
database into an open format (or any other format), then it would be
open data.

Kirrily: here's a real use case, then. Is it open data if the govt
distributes it as files that need to be opened in a massively
expensive, non-free statistical software package?  What about geo data
in formats that are specific to non-free GIS platforms?  These are
real and common use cases.

Jordan: Yes, an expensive proprietary format can be a barrier, but I
think there is a line to be drawn on being open *data*. If you are
legally able to convert that data into an open format, then the data
is open. There are lots of things that can present a barrier in the
same way as a massively expensive software package to read a certain
format -- what about access to the internet, or access to a computer
at all, or skills to use a computer?  All can form a real barrier to
access and use/reuse of data, but to be a workable definition I think
it's best to focus on the legal rights around the ability to do these
things.

 Question 5. What constitutes onerous attribution in OKD #5?  If you
get open data from somebody, and they have an attribution page, is it
sufficient for you to comply with the attribution requirement if you
point to the attribution page?

Jordan Hatcher: I think this is one of those great questions that is
best left up to  users to define in practice, rather than to set down
in stone what is  onerous and what isn't as this will evolve over time
and across  technologies.  Asking attribution not to be onerous I
think is a good  starting place, particularly for open data as many
contributions could  get quite burdensome quite quickly depending on
how the license was  written.