Why do we really need so many different varieties of software licenses? NITRC lists over 42 different types of licenses for the (currently) 97 software resources listed. Shouldn’t something like the Creative Commons family of licenses provide a common framework to express virtually any of these individual licenses? http://creativecommons.org/about/license...

Just wondering...

These licenses do not seem to capture the essence of GPL (viral open source) or LGPL (limited viral open source). Perhaps there are other conceptual points that the creative commons license framework neglects.

David,

The Creative Commons licenses are not designed for software. Instead they target works of art.
They are well suited for papers, pictures and music, for example.


I agree with you however, in that a proliferation of licenses is detrimental for fostering an Free and Open Source Software community. They generate unnecessary confusion, and make difficult to combine packages that are distributed under different (potentially incompatible) licenses.



Regards,


Luis

David, I suspect you're pretending to be asking an innocent question, but really wanted to get a lively discussion started with a hot-button topic. That said, I can't resist the urge to spew out a few rambling thoughts.

First, the creative commons has a FAQ on their web site, in which they refer authors of software to the FSF and Open Source Initiative for guidance. The existing CC licenses are clearly inadequate for software, both from the standpoint of expressing the genuine preferences of software authors and from the standpoint of handling the many sticky situations that come up with software. There are lots of good reasons why it's a little harder with software, many of them historical, but many of them motivated by the fact that many people want to provide code in ways that aren't covered by the fair use provisions of copyright. This means that things have to be made explicit in licenses. We could live in a world in which nobody shared their code beyond fair use, which is basically the norm for writings, notwithstanding the CC.

Both the FSF and OSI maintain huge lists of licenses, each evaluated for its adherence to the standard of the respective organization (i.e., GPL-compatibility for the former). Why so many? Some are redundant, but my impression is that many of these licenses (a substantial minority) have things in them that were motivated by some deeply held belief of whomever wrote the license. Unless you can find someone else who shares your deeply held beliefs, it may be hard to find a license that expresses what you want to say. I'm aware of one license out there that forced a package I like to be removed from my favorite Linux distribution. The license was non-free because it prohibited use of the software in an industry the author didn't like. The GPLv3 was motivated in large part by an issue with something clever TiVO did. These kinds of issues don't come up with text so much, although that may just be a cultural thing.

Another issue is that authors of software don't always have control over what licenses they would like to use. They do in some abstract sense, but if you depend critically on a library that is licensed under GPLv3, your software will be too. If you're writing code at an academic institution, they may impose constraints. (I've seen a lot of boilerplate text amounting to “free for academic use,” which immediately rules out the GPL and BSD licenses.)

I think the real reason we have so many licenses for software and so few for other kinds of creative works is just that software people are into licensing, and into naming things. It's easy to imagine a multitude of licenses for writings, but until the creative commons came along, the popular licenses for writings distributed electronically were just copyright notices with a short list of rights granted (e.g., “I hold the copyright, but I give you license to do X, Y, and Z.”) Authors of text could certainly have named every slight variant of this, but they didn't. In reality that doesn't mean there are any fewer distinct licenses for text, it just means we don't have tidy names for all of them. If NITRC wanted to categorize bits of text according to license, it could have been even more difficult than with software. We're lucky in a sense that so many software authors do borrow existing licenses without changes. But beyond indexing a few of the common features of licenses (which are GPLv3 compatible? which permit modification in some form? which permit use of the software without restriction?), I don't think there's any great need to force the range of licenses to be any smaller and more systematic than it is.

I'm certainly sympathetic to Luis's comment about the perils of incompatible licenses (we were basically forced to adopt GPLv3), but authors do have the right to some say in how their code may be used, even if it's not always well thought-out.

Daniel,

I very much agree with the points you made.

and wanted to particularly emphasize one of them:

You mention:

"(I've seen a lot of boilerplate text amounting to “free for academic use,” which immediately rules out the GPL and BSD licenses.)"


Let's clarify that any license that includes restrictions like:

* "For academic use only" or
* "Free for research only"

can not be considered an Open Source license because it excludes a field of endeavor. Something that is contrary to item 6 of the Open Source Definition:

http://www.opensource.org/docs/osd


Luis

Hi,

I just wanted to confirm that Creative Commons licenses are not recommended for software.

For the official CC position on this, please refer to this FAQ entry:

http://wiki.creativecommons.org/FAQ#Can_...

License proliferation is a valid concern. For OSI's report on license proliferation, please refer to:

http://www.opensource.org/proliferation-...

The best way to avoid license proliferation is to work within your community to develop a consensus about appropriate licenses to use. License proliferation may reflect different preferences within the community, but often it also reflects lack of dialogue among providers and users of materials.


(These comments are provided for general reference, and are not intended to substitute for competent legal advice.)


Best Regards,

Thinh Nguyen
Counsel, Science Commons

Ironically,
when looking at the list of licenses from
packages available at NITRC, I see many
proprietary licenses, but couldn't find
any with the Apache 2 license.

Is there a particular reason why the Apache license is not listed ?

We are considering the Apache license for a
software that we are developing for the analysis
of microscopy images. Would the use of the Apache
2 license prevent a package from being listed at
NITRC ?


Thanks for any advice.


Luis

Licenses that aren't displayed in NITRC's list may be specified when you register the software and will then be added to the list you see. NITRC doesn't discriminate based on license and doesn't endorse any license over any other.