NITRC Rosetta Bit Forum: rosettaqa

License - original source?

Matthew Brett — Tue, 18 Nov 2014 23:24:22 GMT

But the problem is, that for a clause with very little practical
benefit, it has a significant practical disadvantage, because it makes
it much more difficult to redistribute the data as part of other
packages (like software packages) where it is common for people to
only scan the overall license. For example, Python packages can be
downloaded automatically with ``pip install nibabel`` or similar, with
no chance for you to review a license, unless you want to. This is
perfectly sensible for something like a BSD license, where problems
only come up when the person wants to distribute a modified version,
but it doesn't fit the spirit of the no-id license, because it would
be easy for the person not to read the license for the data in
particular.

License - original source?

Bennett Landman — Tue, 18 Nov 2014 23:07:45 GMT

License - original source?

Matthew Brett — Tue, 18 Nov 2014 20:40:01 GMT

Do you mean that specific folks told you they would not release their
data without such a clause?

I had always imagined that the clause was effectively toothless, I
can't imagine that you could prevent people getting the data even once
you have demonstrated that they have identified it. In any case, by
that stage it is too late. Is the idea that the people using the data
will be frightened to de-identify it because of the license? Who
would prosecute them?

License - original source?

Bennett Landman — Tue, 18 Nov 2014 18:31:11 GMT

Folks are extraordinarily cautious about releasing data that could be legally used for something like:

Identifying Personal Genomes by Surname Inference
Science 18 January 2013:
Vol. 339 no. 6117 pp. 321-324
DOI: 10.1126/science.1229566

License - original source?

Matthew Brett — Tue, 18 Nov 2014 0:51:55 GMT

As a matter of interest - why do you need the no de-id clause?

License - original source?

Bennett Landman — Tue, 18 Nov 2014 0:49:13 GMT

I'm open to asking folks for a different license. We just need no de-id clause.

License - original source?

Matthew Brett — Tue, 18 Nov 2014 0:37:40 GMT

Another related problem is that the BIRN license is not attractive for
people with data they want to distribute under a more liberal license.

So for example, we may have some data that we want to release under
the PDDL or similar, but we would not want Rosetta to be the main
source for that data because we want to make it as easy as possible
for people to include the data in their own projects.

License - original source?

Matthew Brett — Mon, 17 Nov 2014 21:23:30 GMT

Being a bit of a license zealot, I found myself reading the NITRIC
copy of the license, and then wanting to find the original copy, to
check it.

Just in case there were some other nerds like me, maybe it would be
useful to either:

- update the link in the NITRIC copy of the license to point to the
archive.org copy or
- use the FBIRN license, that does appear to be available at :
https://www.birncommunity.org/wp-content/uploads/2009/09/FBIRN_Data_Use_Agreement.pdf

I've said this before I think, but the BIRN license is somewhat
unfriendly to open-source projects in the sense that you do have to
worry about people reading the license, and hence I would worry about
including any Rosetta data in our source repository. The OpenFMRI
PDDL is much more friendly in that respect, but hey, having the data
is already a big plus.

License - original source?

Bennett Landman — Mon, 17 Nov 2014 21:11:28 GMT

License - original source?

Bennett Landman — Mon, 17 Nov 2014 21:10:17 GMT