News

What good is a CC licensed specification?

Mike Linksvayer, March 29th, 2008

Proto-lawyer, GNOME hacker and CC friend Luis Villa’s brief “CC-licensed specification” rant is correct:

[I]mplementing a spec may require (among other things) licensing of “pending utility and design patent claims, copyrights, trade dress and trademark rights.” Putting a specification under a CC license gives you a copyright license to the text of the specification; it does not give license to the necessary trademarks, or to the patents, and depending on the license chosen, may not even give you the right to make a derivative work [...]

Fortunately all such specifications I’m aware of are published under free CC licenses (or placed in the public domain) so that derivative works and commercial use are legal.

However, liberal copyright terms on the text of a specification are not sufficient (and strictly speaking, perhaps not even necessary) for a protocol (or format or similar) to permit independent implementation, interoperability, and extension, including by free and open source software.

Software patents may be the main legal barrier to such use. This is why patent grants often get the most thorough public vetting of any non-technical aspect of a new specification and why (for example) the debate over the W3C’s patent policy several years ago was so important.

There are also non-legal items that will often be more important for a protocol being “open” in practice than the protocol specification’s copyright license. For example, the very existence and publication of an unambiguous specification, and the availability of a reference implementation and test suite, preferably under a free and open source software license.

So what good is putting a specification under a liberal copyright license? Is it just about signaling good intentions? As valuable as such signaling may be, it can be abused. I would argue that it is primarily useful for facilitating ongoing collaboration on the specification itself, extensions of the specification, and instructional materials and other non-software works around the specification — in other words, precisely the works and activities impacted by the copyright status of the text.

This is what I intended to highlight in a recent post on IE8 and removing copyright barriers to collaboration with technology communities (emphasis added):

It’s cool that Microsoft not only released the specifications under liberal terms, but followed the lead of the relevant communities, ensuring that there are no copyright barriers to collaboration with those communities.

As the post explains, Microsoft released their specifications under the same liberal terms (one under CC BY-SA, another dedicated to the public domain) as related specifications have been released under by others. This simply (but importantly) means that in terms of copyright anyway, the relevant communities are free to fold the Microsoft specifications into their wikis and other materials for ongoing collaboration (and so are you).

My post did not explain (as it should have) that as above, there’s much more to making a protocol usable than just placing its specification under liberal copyright terms. Microsoft obviously realizes this, as at the same time they also offered a (patent related) Open Specifications Promise for the specifications in question — though whether the promise is good enough may be subject to debate — see further quotation of Luis below.

I briefly raised another reason to place a specification under a well-known liberal copyright license in a post about the Sitemaps Protocol: Creating a space where Google, Microsoft, and Yahoo! can collaborate:

This is just a guess, but I imagine that agreeing to release the [Sitemaps protocol] under a CC license saved Google, Yahoo!, and Microsoft many hours of legal haggling over copyright in the protocol. It is not a guess that this decision allows anyone, e.g., non-incumbent search engines, to publish and extend the protocol, without asking for permission from the incumbents.

In other words, using a public license (CC BY-SA in this case) for the specification solves one — even if just one — of what must be a laundry list of issues that must face such a collaboration, and that’s valuable. However, I should have been more clear that this enables anyone to publish and extend the protocol specification, at least insofar as copyright is concerned.

The most recent development concerning a significant specification under a CC license is also the most interesting so far, in that the parties involved seem to have made a effort to address all of the known barriers to uses of a protocol (whether they’ve succeeded is presumably an open question). Tim Vollmer beat me to blogging about OpenSocial, so I’ll expand a bit here.

The OpenSocial Foundation Intent Agreement covers copyright in the specification, a patent non-assertion covenant, a patent right in joint development, and a license to the OpenSocial trademark. There’s also a reference implementation under the Apache 2.0 license. Of course this appears to be just a proposal, and it is not clear to me whether non-members of the proposed foundation would have any patent or trademark rights with regard to the specification or implementations thereof. But at least they’re looking at all of the elements.

Luis again:

So, creative commons folks: could you please, please scream for me? Or better yet, work with SFLC to create a good license for specifications (since they aren’t happy with the OSP), and then ask people who’ve ‘cc licensed’ specifications to use that instead?

I suspect free CC licenses are reasonable licenses for specifications — for specification text, as far as that goes. But I agree we should talk about barriers other than copyright in the specification text whenever we talk or are approached about CC licenses and specifications.

We have been approached over the years about developing a license for protocols, specifications, standards, and the like, and would doubtless be happy to work with the Software Freedom Law Center and other parties to develop such a thing, whether it would take the form of a license or a “protocol” that included using liberal copyright terms and other aspects. In such an effort it would be important to consider interoperability with CC licenses, so that specification texts could easily be used in contexts like wikis and instructional materials.

One Response to “What good is a CC licensed specification?”

  1. tubejay says:

    this is why cc licenses (which are very important, and i’m actually growing weary of any media that does not have a cc or similar license, especially with all this acta nonsense for an “alternative”) are completely wrong for software. cc licenses are perfect for music, film, text, and images.

    none of which are software. for software, either an x11-style license or (preferably) gpl license would be better. the latter i think covers the patent issue. obviously the other reason software licenses are better for software than cc licenses is that they exist also to keep sourcecode available.